H-Ferritin is essential for macrophages’ capacity to store or detoxify exogenously added iron

Macrophages are central cells both in the immune response and in iron homeostasis. Iron is both essential and potentially toxic. Therefore, iron acquisition, transport, storage, and release are tightly regulated, by several important proteins. Cytosolic ferritin is an iron storage protein composed of 24 subunits of either the L- or the H-type chains. H-ferritin differs from L-ferritin in the capacity to oxidize Fe2+ to Fe3+. In this work, we investigated the role played by H-ferritin in the macrophages’ ability to respond to immune stimuli and to deal with exogenously added iron. We used mice with a conditional deletion of the H-ferritin gene in the myeloid lineage to obtain bone marrow-derived macrophages. These macrophages had normal viability and gene expression under basal culture conditions. However, when treated with interferon-gamma and lipopolysaccharide they had a lower activation of Nitric Oxide Synthase 2. Furthermore, H-ferritin-deficient macrophages had a higher sensitivity to iron-induced toxicity. This sensitivity was associated with a lower intracellular iron accumulation but a higher production of reactive oxygen species. These data indicate that H-ferritin modulates macrophage response to immune stimuli and that it plays an essential role in protection against iron-induced oxidative stress and cell death.

Iron is essential for almost all living organisms, due to its key roles in biological processes such as DNA and RNA synthesis, mitochondrial respiratory chain, cell proliferation, and differentiation, among others 1 . On the other hand, iron needs to be strictly controlled and stored, due to its potential to cause cell damage 2 . Proteins like ferritin, transferrin, and ferroportin are essential to control the intracellular and intercellular iron fluxes and imbalances in their expression or function lead to severe health problems [3][4][5] . In particular, ferritin is essential to safely store iron inside cells, avoiding its harmful effects. This protein is composed of a 24 subunits complex that forms a nanocage, inside which iron is trapped 6 . Ferritin subunits can be of the H, L (cytosol) or M (mitochondria) types 7 . The cytosolic subunits have distinct functions and are present in different proportions in different cell types 8 .
H-ferritin (FTH1) is essential for mammalians' development, as the full knock-out of its gene leads to embryonic death in mice 9 . FTH1 differs from L-ferritin (FTL) in its ferroxidase activity, which is responsible for the oxidation of iron Fe 2+ to Fe 3+ , preventing the oxidative damage that could be caused by Fenton reactions involving Fe 2+ 10 . These reactions can generate oxygen reactive species that damage proteins, lipids, and other cell components.
Macrophages are central players in iron metabolism. They recycle senescent red blood cells and nurse the formation of new ones [11][12][13] , thus participating in the major systemic iron fluxes occurring in higher vertebrates. Macrophages are also a first line of defence against pathogens, and they modulate iron availability as part of host protective mechanisms 12,[14][15][16] .

Results
H-ferritin is not necessary for in vitro differentiation of bone marrow-derived macrophages. Since H-ferritin is essential for mouse development 9 , we started by evaluating whether it was also necessary for in vitro differentiation of macrophages from their bone marrow (BM) precursors. Bone marrow-derived macrophages (BMDM) were obtained from Fth1 Fl/Fl ; Lyz2 +/+ (Fth1 +/+ , H-ferritin-sufficient) and Fth1 Fl/ Fl ; Lyz2 cre/+ (Fth1 −/− , H-ferritin-deficient) mice. The absence of H-ferritin in Fth1 −/− BMDM was confirmed by western-blot (Fig. 1a) and also by gene expression (Table 1). Although it tended to be increased, the level of L-ferritin was not significantly different between Fth1 −/− and Fth1 +/+ macrophages, either at the protein (Fig. 1a) or at the RNA level (Table 1). Throughout the culture, cells with either genotype were indistinguishable in terms of morphology, as well as in terms of viability (measured by a resazurin reduction assay) and in the acquisition of macrophage differentiation markers (evaluated by flow cytometry) (Fig. 1b-d). These data also showed that for both genotypes, at the 7 th and 10 th day of culture, nearly 90% of the cells had acquired the macrophage phenotypic markers (Fig. 1d). Thus, subsequent experiments were performed 7 days after the beginning of the culture.
In addition, we asked whether the absence of H-ferritin expression would impact the basal iron and inflammatory status of the macrophages. For that, we measured the expression of genes involved in the cellular handling of iron: transferrin receptor (Tfrc, iron uptake), ferroportin (Slc40a1, iron export), heme oxygenase-1 (Hmox1, heme degradation), inducible nitric oxide synthase (Nos2a), arginase (Arg1) and tumor necrosis factor (Tnf) (macrophage activation), and we did not find any significant differences between the two genotypes ( Table 1).
H-ferritin modulates macrophage response to immune activation. In order to investigate the potential role of H-ferritin in macrophage activation, interferon gamma and lipopolysaccharide (IFNG + LPS) were added to the BMDM on the 7 th day of culture, as described in Materials and Methods. No differences in cell viability were observed during the treatment with IFNG + LPS in cells of either genotype (data not shown). The effects of treatment on macrophage gene expression were evaluated 12, 24 and 72 hours later ( Fig. 2 and Table S1). In wild-type cells, Fth1 gene expression increased upon treatment with IFNG + LPS, especially at 24 h (Fig. 2a), in accordance with previous reports 14 . Ftl expression slightly increases with IFNG + LPS treatment at 24 h, however, no differences were observed between the two genotypes (Fig. 2b). Interestingly, the levels of Slc40a1, which codes for the iron exporter ferroportin, increased during the first 24 h of treatment and decreased at 72 h (Fig. 2c). Of note, at 24 h post-treatment, the expression of Slc40a1 was significantly higher in Fth1 −/− macrophages than in wild-type cells. The expression of Hmox1 also increased with treatment, with Fth1 −/− cells having a significantly higher expression than wild-type cells at 24 h post-treatment (Fig. 2d). The expression of Tfrc in Fth1 +/+ cells increased with time, whereas in Fth1 −/− cells its expression was significantly lower at all the evaluated time-points (Fig. 2e). The gene showing the highest induction with treatment was Nos2a, a hallmark of macrophage activation 18 (Fig. 2f). However, the induction of Nos2a upon treatment with IFNG + LPS was significantly lower in Fth1 −/− macrophages when compared to Fth1 +/+ cells (Fig. 2f).
Macrophage activation was also evaluated by measuring the release of nitrites and tumor necrosis factor alfa (TNFa) to the culture supernatants. In agreement with the Nos2a gene expression data, significantly lower levels of nitrites were found in the supernatants of Fth1 −/− than in Fth1 +/+ macrophage cultures treated with IFNG + LPS (Fig. 3a). However, the release of TNFa was the same, independently of the expression of Fth1 (Fig. 3b).

H-ferritin supports macrophage survival in the presence of exogenous iron.
Although Fth1 −/− macrophages had normal basal iron-related gene expression (Table 1), we hypothesized that these cells could have a hampered capacity to deal with exogenously added iron. To assess that, cells were treated with different concentrations of ferric ammonium citrate (FAC) or hemin. Cellular viability was evaluated by resazurin reduction 3 days after treatment. As expected, FAC and hemin caused a concentration-dependent reduction of cell viability, but more interestingly, Fth1 −/− macrophages were more susceptible to iron toxicity than Fth1 +/+ macrophages (Fig. 4a,c and Table 2). Moreover, FAC was more toxic than hemin, for both cell types (Table 2).
In order to confirm these results, we evaluated the impact of the same compounds in macrophage cell death using the membrane-impermeable dye SYTOX Green TM . Both FAC and hemin caused a dose-dependent increase in cell death, which was again more marked in Fth1 −/− cells compared to Fth1 +/+ cells (Fig. 4b,d). The damage induced by exogenous iron was also evident in morphological alterations detected by Transmission Electron Microscopy (TEM) (Fig. 4e). In cells treated with FAC, the main alteration observed was an increase in the number of vesicles in the cytoplasm. Some cells also had nuclear alterations suggestive of necrosis. With the hemin treatment, cells were larger and there were signs of membrane damage (Fig. 4e, arrows). These alterations were more frequent and marked in Fth1 −/− cells (Fig. 4e).
The data presented here indicate that Fth1 −/− cells are less capable of handling exogenous iron, being more sensitive to iron-induced cell death.
H-ferritin-deficient macrophages retain less exogenously added iron. In order to evaluate how H-ferritin deficiency impacted iron internalization and retention by macrophages, we used TEM equipped with treatment with FAC (Fig. 5b). Overall, these results indicate that the lack of H-ferritin hampers macrophages' capacity to retain iron when this is exogenously added.
In order to understand the mechanisms underlying these differences in cellular iron dynamics, we evaluated the alterations in gene expression occurring upon iron treatments (Tables 3 and S2). In general, Fth1 −/− macrophages exhibited similar trends as Fth1 +/+ macrophages in the response to the iron treatments, with upregulation of genes involved in iron storage (Ftl) and iron export (Slc40a1) as well as with a decrease in the expression of the iron internalization-related gene Tfrc in response to heme (Table 3). However, a tendency was observed for an increased expression of Slc40a1 in Fth1 −/− in comparison with wild-type cells. This tendency was also observed at the protein level, although it only reached statistical significance in the case of the hemin treatment (Fig. 6b). We did not observe statistically significant differences between Fth1 +/+ and Fth1 −/− macrophages concerning the protein expression levels of FTL ( Fig. 6d) or HMOX1 (Fig. 6c). Overall, these data indicate that the absence of H-ferritin in macrophages leads to an increase in iron export through ferroportin.

FTH1 protects macrophages against iron-induced oxidative stress.
Knowing that Fth1-deficient macrophages are more prone to iron-induced death despite lower intracellular iron levels, we investigated whether their increased cell death was caused by oxidative stress. For that, macrophages were incubated with iron together with the antioxidant N-acetyl-cysteine (NAC) and cell death was quantified by incorporation of SYTOX Green. The toxic effect of exogenous iron was partially prevented by the presence of NAC (Fig. 7a,b), indicating that it is, at least partially, mediated by reactive oxygen species. To evaluate whether Fth1 −/− macrophages had an intrinsically higher susceptibility to oxidative stress, we treated them with either hydrogen peroxide or tert-butyl-hydrogen peroxide (tBHP). Cell death increased with both treatments in a dose-dependent manner, but no differences were observed between Fth1 +/+ and Fth1 −/− macrophages (Fig. 7c,d). Altogether, these results ( Fig. 7a-d) indicate that both genotypes are able to withstand similar levels of oxidative stress. However, when faced with equal concentrations of iron, the oxidative damage in Fth1 −/− cells appears to be higher than in Fth1 +/+ cells. To confirm that, we measured the levels of ROS, lipid peroxidation and protein carbonylation. FAC induced a clearly higher level of ROS formation on Fth1 −/− cells compared to Fth1 +/+ cells, especially 1 and 3 hours after treatment (Fig. 7e). Fth1 −/− macrophages also had significantly higher levels of lipid peroxidation at 6 and 12 hours after treatment with FAC ( Fig. 7f). In the same line, protein carbonylation was higher in iron-treated macrophages of the Fth1 −/− than Fth1 +/+ genotype (Fig. 7g,h). Interestingly, basal levels of protein carbonylation tended to be higher in non-treated Fth1 deficient cells, although the difference relative to wild-type was not statistically significant.
In order to understand whether Fth1 −/− macrophages had more oxidative stress due to some defect in the induction of cell-protective mechanisms, we measured the expression of several genes involved in these mechanisms. Fth1 −/− macrophages upregulated their expression of Hmox1, superoxide dismutase (Sod2), catalase (Cat), thioredoxin reductase (Trxr) and glutamate-cysteine ligase catalytic subunit (Gclc) similarly to Fth1 +/+ cells (Table 3).
Overall, the data obtained in this work, clearly indicates that Fth1 has an essential, non-redundant role in cell protection against iron-induced toxicity.

Discussion
H-ferritin is crucial for mouse development and survival, with the total knockout being embryonically lethal 9 . In this work, we show that H-ferritin is not necessary for in vitro differentiation of murine BMDM nor has a significant impact in these macrophages' basal state, but it influences macrophage response to immune activation or iron treatments. In particular, H-ferritin-deficient BMDM produce less nitric oxide in response to IFNG + LPS treatment and are more prone to oxidative stress and cell death induced by exogenously added iron.
H-ferritin-deficient BMDM were indistinguishable from wild-type BMDM regarding the kinetics of differentiation, morphology, viability, and the expression of several iron-and activation-related genes. In particular, no significant compensatory increase in L-ferritin expression was found in H-ferritin-deficient macrophages. This is in contrast with the results obtained by Bolisetty et al. using the same type of cells 19 . The explanation could be related to differences in culture conditions. Of note, in previous studies, it was observed that transient silencing of Fth1 expression resulted in upregulation of Ftl 20,21 , but stable H-ferritin-deficient HeLa cell clones had normal levels of L-ferritin 21 . Together with the results presented here, this indicates that when cells undergo a long-term www.nature.com/scientificreports www.nature.com/scientificreports/ adaption to the lack of H-ferritin the expression of L-ferritin is not increased. This adaptation may include a decrease in intracellular iron accumulation, as indicated by our results.
H-ferritin is known to be involved in macrophage activation. When we treated BMDM with IFNG and LPS, the expression of Fth1 doubled, in agreement with previous results obtained with several TLR agonists 14,22 . An increase in expression of Fth1 upon IFNG + LPS treatment is apparent in Fth1 −/− cells (Supplementary Table 2), which is probably related to a minor leakage of Fth1 expression due to Cre failure, but this remained at residual levels compared to wild-type cells (Fig. 2a). In contrast, Ftl expression was not significantly altered by IFNG + LPS treatment, irrespective of Fth1 expression. In agreement with previous reports 22,23 , the ferroportin-coding gene Slc40a1 was down-regulated by IFNG + LPS treatment in wild-type macrophages. However, in Fth1 −/− cells, a significant up-regulation of this gene was observed at 24 hours post-treatment, indicating that the expression of this important iron-export gene is regulated by cellular H-ferritin levels. In agreement to what has been described www.nature.com/scientificreports www.nature.com/scientificreports/ previously by Ludwiczek et al. in THP-1 monocytes 24 , wild-type BMDM had a significantly higher expression of the Tfrc gene (coding for the transferrin receptor 1, involved in cellular iron uptake) from 24 hours onward after IFNG + LPS treatment. In contrast, in Fth1 −/− macrophages this gene was down-regulated by treatment at all the time-points analysed. Together, the differential regulation of both ferroportin and transferrin receptor genes in Fth1 −/− macrophages emphasize the inability of these cells to store iron. Additionally, Fth1 −/− BMDM had a significantly higher expression of Hmox1 24 hours post-treatment, in comparison to Fth1 +/+ macrophages. Hmox1 gene expression is known to increase in response to ROS generating agents and to be up-regulated in an inflammatory environment, such as mycobacterial infections 25,26 . This increase in Hmox1 expression indicates that, besides an impairment on iron-retaining capacity, Fth1 −/− cells exhibit a higher degree of oxidative stress in response to IFNG + LPS treatment. Intriguingly, this increased expression of Hmox1 occurred concomitantly with a decrease in Nos2a expression and nitrite release (Figs. 2 and 3), which is in contrast with a previous study that suggested that Hmox1 upregulation upon IFNG + LPS treatment was due to increased Nos2a expression and nitrite production 25 . Conversely, our data suggest that overall ROS level, rather than NOS2 activity, underlies Hmox1 induction. Of note, although we did not thoroughly studied the level of ROS in Fth1 −/− versus Fth1 +/+ BMDM after IFNG + LPS treatment, there was a tendency to have higher levels of carbonylated proteins in Fth1 −/− cells under basal conditions (Fig. 7g). The clear dampening of Nos2a gene induction in the absence of H-ferritin and in a situation of low intracellular iron level, was also an intriguing observation per se and corroborates a previous observation by Bolisetty et al. 19 . Several previous studies showed that treating macrophages with excess iron decreased Nos2a expression and nitrite production upon treatment with IFNG and/or LPS [27][28][29] . We now postulate that these effects were due to increased intracellular oxidative stress rather than to a direct effect of iron. Different signalling pathways have been implicated in the relationship between iron and Nos2a expression, including signal transducer and activator of transcription 1 (STAT1) 27 and hypoxia-inducible factor 1-alpha (HIF1A) 30 . The intracellular iron levels, labile iron pool (LIP) and Iron Responsive Proteins (IRP) have also been implicated 31 . The exact mechanisms linking the lack of H-ferritin to the modulation of Nos2a expression are beyond the scope of this work but will be the subject of future studies.
H-ferritin is an essential player in iron metabolism since it has iron storage and detoxifier activities. To understand the importance of H-ferritin in this scenario, we treated macrophages with exogenous iron in the forms of FAC and hemin. In agreement with previous reports 25,28 , both forms of iron caused a dose-dependent reduction in macrophages' viability (and an increase in cell death). However, there was a clearly higher susceptibility to exogenous iron in the absence of Fth1. We reasoned that this occurred because without Fth1, cells are unable to oxidize Fe 2+ , and this will accumulate in a labile reactive pool. Previous studies have indicated that in the absence of H-ferritin cells have an increased LIP 20,32,33 . In order to have an insight on the alterations in intracellular iron distribution caused by the lack of H-ferritin in our model, we used electron microscopy coupled with EDS. To our surprise, Fth1 −/− BMDM had significantly less intracellular iron than normal BMDM, making it technically impossible to quantify iron in different intracellular compartments. This decrease in intracellular iron may be caused by an inability to store iron in a normal ferritin nanocage in the absence of FTH1 34 , as well as a tendency to have increased levels of ferroportin, the only known protein exporter. In accordance with this finding, studies on the tissue iron distribution in Fth lox/lox ; MxCre mice revealed that the lack of FTH1 results in decreased iron storage in the liver and spleen 17 .
Despite the decrease in the amount of intracellular iron, Fth1-deficient BMDM had increased levels of oxidative stress when exposed to exogenous iron, as evaluated by the levels of ROS, lipid peroxidation and protein carbonylation. Our data indicate that iron-induced oxidative stress is the main cause of increased cell death in the Fth1 −/− macrophages upon iron treatment since in the presence of NAC, Fth1 −/− cells were as protected from cell death as www.nature.com/scientificreports www.nature.com/scientificreports/ Fth1 sufficient cells. As mentioned before, this effect is probably due to the vulnerability of Fth1 −/− cells to Fe 2+ as they lack the ferroxidase activity conferred by H-ferritin. Importantly, the lack of H-ferritin did not increase the susceptibility of BMDM to different peroxide sources, indicating that H-ferritin is not directly involved in peroxide detoxification, but it is involved in decreasing peroxide formation from iron. Furthermore, the data show that macrophages do not have an alternative mechanism to deal with iron-mediated oxidative stress, even though they can upregulate the transcription of several genes involved in the response to oxidative stress. Overall, the alterations seen in Fth1 −/− macrophages are consistent with the recently described type of programmed cell death, "ferroptosis" [35][36][37][38] , although the exact mechanisms of cell death were not thoroughly characterized here. www.nature.com/scientificreports www.nature.com/scientificreports/ In conclusion, this work shows that although H-ferritin is not essential for the in vitro differentiation of bone marrow-derived macrophages, it modulates macrophage activation by immune and microbial stimuli and, more importantly, it has a non-redundant role in cell protection against iron toxicity.     Macrophage cultures. Bone marrow-derived macrophages (BMDM) were obtained as described previously 39 from the bone marrow of Fth1 +/+ or Fth1 −/− mice.
Mice referred to as Fth1 −/− mice are conditional Fth1 deficient (Fth1 Fl/Fl ; Lyz2 cre/+ ) mice, obtained by crossing Fth1 Fl/Fl mice 17 with Lyz2 cre/+ mice. An initial breeding pair was kindly provided by Prof. Lukas Kuhn (Swiss Institute for Experimental Cancer Research, Lausanne, Switzerland). In these mice, Cre recombinase deletes the Fth1 gene in cells expressing Lyz2 (cells of the myeloid lineage). Fth1 Fl/Fl ; Lyz2 +/+ Cre-negative littermate mice were used as controls and will be referred to as Fth1 +/+ mice.
All experimental animal procedures described in this work were approved by the Local Animal Ethics Committee of IBMC/i3S and licensed by the Portuguese Authority "General Directory of Agriculture and Veterinary" (DGAV), on July 6th, 2016 with reference 0421/000/000/2016. All animals were handled in strict accordance with good animal practice as defined by national authorities (DGAV, Decreto-Lei 113/2013, august 7th) and European Directive 2010/63/EU. Cell cultures were visualized and imaged every day during the experiment in an inverted optical microscope (Leica DMI6000 Timelapse).
For macrophage activation experiments, recombinant mouse IFNG (Gibco, MD, USA) was added at 16 ng/ml, at the 7th day of culture. For kinetic studies, IFNG was added two additional times thereafter with 24 h intervals. Lipopolysaccharide (LPS) was given once at 10 ng/ml.   (Table 4). All reactions were performed in a 20 μL total volume with iTaq SYBR green PCR master mix (Bio-Rad, Hercules, CA, USA). assay. The results represent the mean ± SD of three independent cultures per condition and are expressed as the ratio between peroxidized lipids (intracellular fluorescence intensity at 510 nm) relative to the total lipids (intracellular fluorescence intensity at 510 nm plus at 590 nm); (two-way ANOVA with Tukey multiple comparisons test; *p < 0.05). (g,h) Fth1 +/+ (black) and Fth1 −/− (grey) BMDM were treated with 10 µM FAC or 100 µM hemin for 24 hours. (g) Detection of carbonylated proteins by slot blot. The image corresponds to one representative experiment with BMDM obtained from 1 animal per genotype. Cell extracts were blotted in duplicate. The whole membrane is shown. (h) Densitometry analysis with ImageLab TM software of carbonylated proteins. The bars represent the mean + SD from 4 independent cultures, of the relative density of the bands obtained from Fth1 +/+ (black) or Fth1 −/− (grey) BMDM in each condition, as compared to non-treated Fth1 +/+ cells.
Evaluation of cell viability. Cell metabolic activity was measured by resazurin reduction. Briefly, 10% (v/v) of resazurin (0.3 mg/ml) was added to the cells and incubated for 20 h. The fluorescence of resorufin, resulting from the reduction of resazurin by metabolically active cells, was measured at λ ex = 530 nm and λ em = 590 nm in Synergy TM Mx (BioTek, Winooski, VT, USA).
For the evaluation of membrane permeability, cells were incubated with a mixture of a cell-permeable nuclear dye Hoechst 33342 (1:12000, Invitrogen, Eugene, OR, USA), and a membrane-impermeable dye SYTOX TM Green (1:45000, Thermo Fisher Scientific, Waltham, MA, USA), for 20 minutes at 37 °C and 7% CO 2 . Cells were visualized in a 37 °C, CO 2 atmosphere with a 20× Nikon objective in a high-throughput automated fluorescence wide-field microscope (IN Cell Analyzer 2000, GE Healthcare, Little Chalfont, UK). The acquired images were then analyzed with Developer Toolbox 1.9.2 (GE Healthcare).

Quantification of TNFa by ELISA.
Tumor Necrosis Factor alpha (TNFa) was quantified in macrophage culture supernatant 72 h after treatments, using ELISA Ready-SET-Go (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions.
Lipid peroxidation was measured with BODIPY ® , following the manufacturer's instructions (Thermo Fisher Scientific). Upon oxidation by lipid hydroperoxides, BODIPY ® changes its maximum fluorescence emission wavelength from 590 nm to 510 nm. Briefly, cells were incubated with BODIPY ® (10 µM) for 30 minutes, washed with PBS, and then visualized in a 37 °C, CO 2 atmosphere with a 20× Nikon objective in a high-throughput automated fluorescence wide-field microscope (IN Cell Analyzer 2000, GE Healthcare). The acquired images were analyzed with CellProfiler TM . Iron quantification. Iron levels were measured through transmission electron microscopy (TEM)energy-dispersive X-ray spectroscopy (EDS) and through atomic absorption spectrometry. Sample processing for TEM was performed as described previously 41 . For EDS analysis, the sections were mounted on nickel grids and a beryllium holder (EM-21150, Jeol Ltd.) was used. An X-Max 80 mm 2 (Oxford Instruments, Bucks, England) operated at 120 kV was coupled to the microscope, at the HEMS/i3S of the Universidade do Porto, Portugal.
For iron quantification by atomic absorption spectrometry, cells were collected 24 hours after iron treatments. Pellets were dried at 65 °C overnight and digested with 69% nitric acid. The measurements were performed on an AAnalyst 200 (PerkinElmer, Waltham, MA) flame atomic absorption spectrometer, with a hollow cathode lamp (SCP Science, Quebec, Canada) as the radiation source. The absorbance was read at 248.33 nm in an air (10 L/ min) -acetylene (2.5 L/min) flame. Data were normalized using the phosphorous content in each sample.
Phosphorus ( 31 P) determination was performed through inductively coupled plasma-mass spectrometry (ICP-MS) using an iCAP ™ Q (Thermo Fisher Scientific, Bremen, Germany) instrument, equipped with a Meinhard ® (Golden, Colorado) TQ + high sensitivity nebulizer, a baffled cyclonic spray chamber (Peltier-cooled), a standard quartz torch, and a two-cone design (nickel sample and skimmer cones). High purity (99.9997%) argon (Gasin, Matosinhos, Portugal) was used as a nebulizer and as a plasma gas source. The ICP-MS instrument operational parameters were as follows: RF power (1550 W); plasma gas flow (14 L/min); auxiliary gas flow (0.8 L/min); nebulizer flow rate (1.02 L/min). The equipment control and data acquisition were made through the Qtegra ™ software (Thermo Fisher Scientific). The elemental isotope 45 Sc was used as an internal standard.
Statistical analysis. Data analysis was performed using GraphPad Prism 7.0 program (GraphPad Software, Inc., La Jolla, CA, USA), and data were expressed as means plus standard deviations of the number of samples indicated in the legend of each figure. Multiple comparisons were performed by using two-way analysis of variance (ANOVA) followed by a Tukey multiple-comparison post hoc test. Comparisons between two groups were performed by using Student's t test. In both cases, differences were considered significant when the P value was <0.05.