Iron from nanostructured ferric phosphate: absorption and biodistribution in mice and bioavailability in iron deficient anemic women

Nanoparticle production and characterization

For the production of nanostructured ferric phosphate (FePO₄-NP) by flame spray pyrolysis (FSP), iron nitrate nonahydrate (purity ≥ 97.0%, Sigma-Aldrich, Buchs, Switzerland) and tributyl phosphate (97%, Sigma-Aldrich) were dissolved in a 1:1 mixture by volume of ethanol (denat. 2% 2-butanone, Alcosuisse) and 2-ethylhexanoic acid (purity ≥ 99%, Sigma-Aldrich) at a total metal concentration of 0.4 mol L⁻¹ or 0.5 mol L⁻¹ for the two compounds. This precursor solution was fed at 2 or 7 mL min⁻¹ into the FSP spray nozzles by a syringe pump (Lambda, VIT-FIT) and atomized by co-flowing 5 or 7 L min^-1 of oxygen (purity ≥ 99.5%, Pangas) at 1.5 bar pressure drop. The spray was ignited by a methane/oxygen (2.5 L min⁻¹) ring-shaped flame⁴⁵. Using a vacuum pump (Busch, Mink MM1202 AV), product particles were collected on water-cooled Teflon membrane-filters (1TMTF700WHT, BHA Technologies AG) placed at least 70 cm above the burner. Pharmaceutical grade (German Pharmacopoeia DAB, Erg. B.6, no. 505033001, Lohmann) amorphous ferric orthophosphate with a SSA of 25 m² g⁻¹ served as reference compound for BET and XRD measurements. FePO₄-NPs for the human study were produced from stable isotope (⁵⁷Fe) enriched precursors, bulk FePO₄ and FeSO₄ from (⁵⁸Fe) enriched precursors (Chemgas). SSA was determined by N₂ adsorption (Micromeritics Tristar 3000, Micromeritics Instruments Corp) at 77 K in the relative pressure range p/p0 = 0.05–0.25 and calculated using Brunauer–Emmett–Teller (BET) theory. Assuming dense spherical particles, the particle diameter (d_BET) was calculated from the measured SSA according to d_BET = 6/(ρ·SSA), where ρ is the solid particle density (FePO₄^*2H₂O = 2.87 g cm⁻³)⁴⁶. For transmission electron microscopy (TEM) analysis, the powders were deposited on a parlodion foil supported on a copper grid and analyzed on a CM12 microscope (FEI, LaB6 cathode, operated at an acceleration voltage of 100 kV). The crystallinity of the powders was investigated by X-ray diffraction (XRD) on a AXS D8 Advance diffractometer (Bruker) operating with a Cu–Kα radiation. Hydrodynamic diameter was determined by dynamic light scattering using a Zetasizer Nano ZS (Malvern).

Animal studies

Intestine-specific DMT1 knockdown (DMT1^int/int) model

Intestine-specific DMT1 knockdown (DMT1^int/int) mice were bred by crossing floxed DMT1 (DMT^fl/fl) mice on a homogenous C57BL/6 strain (courtesy of Nancy Andrews, Duke University, USA²¹) with villin-Cre transgenic mice on the C57BL/S6J strain (ETH Zurich). The floxed DMT1 mice were back-crossed to a C57BL/6J background for six generations to establish the DMT1^fl/fl mouse breeding colony at ETH Zurich. The 6th generation was then shipped to the Vivarium at NWU to establish the intestine-specific DMT1 knockdown and villin-Cre mouse breeding colony for the experiments described in this article. To obtain DMT1^int/int mice, the animals needed to be homozygously floxed and have the villin-Cre transgene (Cre-positive). The homozygously floxed littermates (DMT^fl/fl) that did not have the villin-Cre transgene (Cre-negative) served as control in all mice experiments to account for a potential effect of floxing on iron absorption and biodistribution. To breed these mice, we mated heterozygously floxed males that have the villin-Cre (Cre-positive) transgene with homozygoulsy floxed females without the villin-Cre transgene (Cre-negative). Theoretically, this results in an approximate yield of 25% DMT1 KO and 25% controls per litter. In order to obtain the required number of mice for the different experiments, we continuously bred, genotyped and enrolled mice into the experiments. Thus, mice in the different experimental groups are from different litters.

Genotyping

The genotyping method to screen each mouse bred from the floxed DMT1 and villin-Cre mice colony was as follows. Briefly, a tissue sample (distal tail sample [≤ 2 mm]) was collected at PND 10–21⁴⁷. gDNA isolation was done using the GenElute™ Mammalian Genomic DNA (gDNA) Miniprep Kit (Sigma Aldrich) following the manufacturers protocol. The quality of DNA was assessed and quantified using the NanoDrop™ spectrophotometer (ND-1000, Wilmington, DE, USA). Gene specific polymerase chain reaction (PCR) was performed and amplicons were visualized using ethidium bromide stained gel electrophoresis. For PCR amplification of the DMT1 gene the forward primer 5’-atgggcgagttagaggcttt-3’ and the reverse primer 5’-cctgcatgtcagaaccaatg-3’ were used. For PCR amplification of the villin-Cre gene the forward primer 5’-gtgtgggacagagaacaaacc-3’ and reverse primer 5’-acatcttcaggttctgcggg-3’ were used together with an endogenous control (MyD88) primer pair 5’-agacaggctgagtgcaaacttgtgctg-3’ and 5’-ccggcaactagaacagacagactatcg-3’. Control gDNA with known genotype (ETH) as well as none-template control samples were included in each PCR run.

Housing and diets

Mice (for breeding and in experiments) were housed in polysulfone individually ventilated cages (391 × 199 × 160 mm [WxDxH]) (Tecniplast, UK) with Alpha-Dri® alpha cellulose bedding (Alpha-Dri, Shepherd Speciality Papers) (< 2.00 ppm iron) under a 12/12 h light/dark cycle (lights on at 06:00) at 22 ± 2 °C and 55 ± 10% relative humidity. The diets used in the experiments and for breeding of experimental mice were commercially obtained purified diets according to AIN93-G standard⁴⁸, with modifications in iron content and compound. Iron fortified diets contained 35 mg iron per kg (ppm) diet, while iron deficient reference diets contained 3 ppm iron (native iron only). The diets were produced by Dyets Inc. (2508 Easton Avenue, P. O. Box 3485, Bethlehem, PA 18017, USA). Iron content of diets was analyzed in spot samples from each batch by Covance Laboratory Services (Madison, WI, USA) before shipping. All mice had ad libitum access to food and to deionized water (18 mΩ). Mice were weighed three times per week (or more frequently during experiments) to monitor weight gain.

Ethics

All animal experiments were approved by the Animal Ethics Committee of the Faculty of Health Sciences of the North-West University (NWU-00050-16-A5 & NWU-00258-17-A5), Potchefstroom, South Africa, and were conducted following the 3R principles for animal research and the Animals in Research: Reporting In Vivo Experiments (ARRIVE) guidelines⁴⁹.

Characterization of intestine-specific DMT1 knockdown model

A total of 10 DMT1^int/int (n = 6 male; n = 4 female) and 21 DMT1^fl/fl (n = 10 male, n = 11 female) mice, born to dams that were kept on the purified AIN-93G diet containing 35 ppm iron (as ferrous citrate) ad libitum, were randomly allocated in pairs to receive an iron deficient (3 ppm native iron) or iron-sufficient (35 ppm ferrous citrate) diet ad libitum from PND 24 to PND 42. PND 42 was set as endpoint, as this was the time point when first mice reached an Hb < 4 g dL⁻¹, which we defined as humane endpoint²² (a humane endpoint is the earliest scientifically justified point at which pain or distress in an experimental animal can be prevented, terminated, or relieved, while meeting the scientific aims and objectives of the study⁵⁰). Hemoglobin (Hb) concentrations were measured in tail blood spots (20 µg dL⁻¹) at PND 24, 27, 30, 33, 36, 39 and 42 using a calibrated Hb 201 + HemoCue® system (HemoCue Angelholm, Sweden). At PND 42, mice were euthanized by decapitation, and liver, duodenum and colon tissue immediately removed, snap frozen in liquid nitrogen, and stored at −80 °C until analysis.

Expression of DMT1 mRNA harboring an iron-responsive element (IRE) in its 3’-terminal exon and the upstream 5’ exon1A in duodenum, colon and liver was analyzed with qPCR. PCR ready Syber green primers were synthesized by IDT (WhiteHead Scientific, South Africa). Primer pair sequences were adapted from Hubert and Hentze (2002)²³. Total RNA was isolated by using Trizol reagent following the standarized protocol. cDNA synthesis was performed using 10 mM oligo-dT_18mer and random hexamer primer mix with Superscript II reverse transcriptase (Qiagen), following the method prescribed by the manufacturer. A total of 50 ng cDNA was used as template in triplicate qPCR reactions with QuantiNova sybr green 2 × master mix together with 1 mM syber green ready primers. Reactions were prepared and amplification was performed at 94˚C for 20 s, 55 ˚C at 40 s and 72 ˚C at 30 s for 30 cycles. Gene expression was deduced using 18S and βActin as endogenous reference genes to calculate delta Ct values for further statistical analysis.

Eighteen-day FePO₄-NP feeding study in DMT1^int/int and DMT1^fl/fl mice

A total of 15 DMT1^int/int (n = 6 male; n = 9 female) and 13 DMT1^fl/fl (n = 6 male, n = 7 female) mice, born to dams that were kept on the standardized AIN-93 G diet containing 35 ppm iron (as ferrous sulphate [FeSO₄]) ad libitum, were randomly allocated in pairs to receive a diet fortified with 35 ppm FePO₄-NP (SSA 98 m²g⁻¹) or FeSO₄ (reference compound) added to an iron-free AIN93-G diet from PND 24 to PND 42 ad libitum. Hb concentrations were measured in tail blood spots at PND 24, 27, 30, 33, 36 and 42 using the Hb 201 + HemoCue® system (HemoCue Angelholm, Sweden). At PND 42, mice were euthanized by decapitation and liver tissue immediately removed, snap frozen in liquid nitrogen, and stored at −80 °C until analysis.

Liver tissue samples were homogenized and digested with nitric acid according to Erikson et al. (1997)⁵¹, and total iron concentrations were measured by using the hydrogen reaction mode on an Agilent 7900 quadrupole ICP-MS at the Central Analytical Facilities, Stellenbosch University, South Africa. Samples were introduced via a 0.4 ml/min micromist nebulizer into a peltier-cooled spray chamber at a temperature of 2 °C. The instrument was optimized for analysis in high matrix introduction (HMI) mode, and all samples and standards were diluted with argon gas to minimize matrix load to the analyzer. The instrument was calibrated using a National Institute of Standards and Technology (NIST) traceable standard (Inorganic Ventures, USA). NIST-traceable quality control standards at high and low concentration levels (De Bruyn Spectroscopic Solutions, Bryanston, South Africa) were analyzed to verify the accuracy of the calibration before sample analysis commenced and this was repeated for every 12 samples to monitor drift. A germanium (Ge) internal standard was introduced online to monitor instrument drift and correct for matrix differences between samples and standards. During the course of the analysis, internal standard recovery was between 90 and 110% for all samples, and recovery for drift monitor standards between 95 and 105%. Oxide formation was less than 0.3%. Three replicate measurements were completed for each sample.

Absorption and biodistribution of radiolabeled FePO₄-NP and FeSO₄ from an acute oral dose in iron deficient anemic DMT1^int/int and DMT1^fl/fl mice

A total of 13 DMT1^int/int (n = 7 male; n = 5 female) and 15 DMT1^fl/fl (n = 7 male, n = 8 female) mice, born to dams that were kept on the standardized AIN-93 G diet containing 35 ppm iron (as FeSO₄) ad libitum, were placed on an iron deficient diet (3 ppm native iron) from PND 21 and throughout the entire experiment. Mice were randomly allocated to receive radiolabeled FePO₄-NP (SSA 98 m²g⁻¹) or FeSO₄. At PND 24, mice were transported to the South African Nuclear Energy Corporation South Africa (NECSA) and left to acclimatize to the new environment until administration of radiolabeled FePO₄-NP or FeSO₄ by oral gavage at PND 30 (29–31). The day before compound administration, Hb was measured in a tail blood spot.

Mice were fasted for 2 h before administration of the oral gavage during which they were acclimatized to metabolic cages (3701M081; Tecniplast). Then, a single dose of ~ 50 µg iron in the form of the allocated compound labelled with ⁵⁹Fe (mean activity: 0.30 MBq) was orally gavaged (in 100 µL saline containing 0.1% bovine serum albumin) using disposable flexible gavage needles. Total dose administered was determined by measuring syringe activity before and after gavage using a dose calibrator (Capintec CRC-15R, Capintec Inc., Ramsey, NJ, USA). Mice were then placed into a clean metabolic cage and received ad libitum access to iron deficient diet (3 ppm iron) one hour after oral gavage. After 24 h, the mice were euthanized by decapitation and the following individual organs were dissected and analyzed for ⁵⁹Fe content and weighed using an automated Hidex®600 SL gamma-counter (Hidex Oy, Finland). All counts were adjusted for decay: whole blood, heart, lung, liver, spleen, stomach, duodenum, peyer’s patches, ileum, jejenum, colon, kidneys, femur, as well as feces and urine. Chyme was separated from the intestinal segments by washing with 1% phosphate-buffered saline.

A total of 15 mg of the nanostructured FePO₄ (SSA 98 m²g^-1) was irradiated in the SAFARI-1 20 MW research reactor in a hydraulic position at a neutron flux of 1 × 10¹⁴ n/cm²s for 16 days with a 3 day cooling period. Irradiation provided 512 MBq ⁵⁹Fe (t_½ = 44.5 days) per 1 mg of FePO₄-NPs⁵². The color of the FePO₄-NP (SSA 98 m² g⁻¹) remained yellowish during the irradiation given confidence that the particles retained their nano structure. This was further confirmed by measuring the surface area using a Tristar 3000 BET surface area and porosity analyzer (Micromeretrics, Norcross, USA). Using a special small volume adapter the SSA of the irradiated particles was determined to be 79.1 m² g⁻¹ which mimicked the SSA determined prior to irradiation. Special care was taken to remove the static nature of the NP after irradiation using a Antistatic Ionizer (RADAWG, Radom, Poland), before opening and during handling of the radioactive labelled NPs. FePO₄-NP were decayed a further 30 days prior to animal administration to allow for the co-activated ³²P (t_½ = 14 days) to decay. ³²P is a pure β emitter and hence did not interfere with the determination using gamma spectrometry at the > 1000 keV range⁵². ⁵⁹Fe-labelled FeSO₄ was purchased from American Radiolabeled Chemicals, Inc. (St. Louis, USA) with a specific activity > 185 GBq/g. In order to mimic the chemical administrated dose used for the NP’s (50 µg and 0.3 MBq), non radioactive FeSO₄ was added to a subset of the purchased stock solution. ⁵⁹Fe radioactivity (for administration) was detected by a dose calibrator (Capintec CRC-15R, Capintec Inc., Ramsey, NJ, USA), while samples were analyzed using an automated Hidex^®600 SL gamma-counter (Hidex Oy, Finland) making use of the 1099 keV 56.5% gamma emission. Cross calibration between the two instruments were obtained by using a ⁵⁹Fe standard curve after a series of dilutions. The irradiated particles were dispersed in in saline containing 0.1% bovine serum albumin by mixing for 30 s on a vortex followed by sonication for 10 min in a Sonorex Digitec waterbath (Bandelin Electronic) at 35 kHz and 80 W and were gavaged within 15 min after sonication. We acknowledge that we did not determine absorption and biodistribution of FePO₄-NP with an SSA of 188 m² g⁻¹ in our mouse experiments because we were not able to radio-label this compound without colour and structural changes (visual inspection) after irradiation in the reactor as opposed to the 98 m² g⁻¹ where this was not observed.

Statistical analysis

Data were analyzed using IBM SPSS Statistics software (version 24). Data were examined for normality of distribution (using q–q plots, histograms, and Shapiro–Wilk test) and the presence of outliers (using box plots). Homogeneity of variance was examined by the Levene’s test. Variables that significantly deviated from normality and/or variance of homogeneity were transformed prior to interferential statistical analysis. Differences in Hb trajectories over time (PND 24–42) by genotype (DMT1^int/int vs. DMT1^fl/fl) and by dietary iron content (35 ppm vs. 3 ppm) or iron compound (FePO₄-NT vs. FeSO₄) were determined using repeated measures ANOVA. Differences in tissue iron concentrations, gene expression and percentage initial ⁵⁹Fe dose/g tissue by genotype and by iron compound were determined by two-sided independent t-tests. The results were expressed as means ± SEM and differences were considered significant at p < 0.05.

Human study

Design and subjects

In a randomized cross-over study, iron deficient, mostly anemic Thai women (n = 18) aged 18 to 49 years consumed four test meals containing ⁵⁷Fe-labeled FePO₄-NP (SSA 98 m²g⁻¹), ⁵⁷Fe-labeled FePO₄-NP (SSA 188 m²g⁻¹), ⁵⁸Fe-labeled bulk FePO₄ (negative reference compound) and ⁵⁸Fe-labeled FeSO₄ (positive reference compound) in random order (using a computerized random number generator—Excel). Stable iron isotope incorporation in red blood cells was determined 14 days after test meal administration.

Women were eligible to participate in the study if: (1) female aged 18 to 49 years; (2) body mass index (BMI) < 23 kg/m² and body weight < 65 kg; (3) Hb ≥ 80 g/L and plasma ferritin < 25 µg/L; (4) not pregnant (confirmed by pregnancy test) or lactating; (5) healthy, no chronic diseases or medications (except oral contraceptives) and no inflammation (C-reactive protein (CRP) < 5 mg/L); (6) no blood donation or significant blood loss at least 4 months before study start; (7) no consumption of vitamin or mineral supplements at least 2 weeks before study start; (8) normal hemoglobin A (HbA) or HbE trait; and (9) nonsmokers.

Sample size calculations indicated that 16 women should be included based on 80% power to detect a 40% difference in iron bioavailability within subjects, an SD of 8.2% for log-transformed absorption data from previous absorption studies with the same meal and iron source/compound in a similar population of Thai women, and a type I error rate of 5%. We anticipated a dropout rate of 10% and therefore recruited 18 women.

Ethics

The study has been carried out according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human study participants were approved by the Ethics Committee of ETH Zurich (Zurich, Switzerland) and the Mahidol University Central Institutional Review Board (Salaya, Nakhon Pathom, Thailand). The study was registered at clinicaltrials.gov as NCT03660462 on 06/11/2018. Detailed oral and written information explaining the study purposes and potential risks and benefits were provided to the interested volunteers. Written informed consent was obtained from all participants. The oral administration of stable isotopes does not present any health risk. All data were coded and treated confidentially.

Study procedures

Women were recruited by screening at the Institute of Nutrition at Mahidol University and at the Khlong Yong Health Promoting Hospital in Nakhon Pathom, Thailand. All details of the study were explained to them, and if they were interested in participating in the study, they were asked to sign the written informed consent form. Then, weight and height were measured (to calculate BMI), and a venous blood sample collected to determine Hb, serum ferritin, and C-reactive protein concentrations. Women who fulfilled all the inclusion criteria were instructed to not eat red meat, fish or poultry four days prior to the scheduled test meals.

In a first phase, the women consumed two randomly assigned test meals (A, B, C or D) between 7.00 and 9.00 am after an overnight fast on study day 1 and 2. In the second phase, the women consumed the remaining randomly assigned test meals (A, B, C or D) between 7.00 and 9.00 am after an overnight fast on day 16 or 17. To distinguish between the absorption of the two forms of ⁵⁷Fe-labeled FePO₄-NP (SSA 98 m²g⁻¹ and SSA 188 m²g⁻¹) and between ⁵⁸Fe-labeled FeSO₄ and FePO₄, the subjects consumed one Fe compound labeled with ⁵⁷Fe and one with ⁵⁸Fe in each phase. The test meals (see details below) were fortified with 2 mg of the respective isotopically labelled iron compound and administered with a glass of deionized water (200 ml). Fourteen days after the last teast meal administration in the first phase (day 16) and again fourteen days after the last test meal administration in the second phase (day 31), a venous blood sample was taken to determine iron absorption.

Composition of the test meal

The test meal was composed of steamed white rice (50 g dry weight), which was served with a vegetable soup prepared from local vegetables (50 g white cabbage, 50 g Chinese cabbage, 30 g Thai mushrooms and 20 g steamed carrots) in 120 mL of water. All ingredients were purchased in bulk and used for the entire study. The food portions were kept frozen until use, and each portion was microwaved on the day of feeding.

Stable-isotope labels

⁵⁸FeSO₄ were prepared from ⁵⁸Fe-enriched elemental iron (> 99.8% isotopic enrichment; Chemgas) by dissolution in 0.1 mol/L sulfuric acid. ⁵⁸FePO₄ was prepared from ⁵⁸Fe-enriched elemental iron (> 99.8% isotopic enrichment; Chemgas). The two FePO₄-NP (SSA 98 m²g⁻¹ and 188 m²g⁻¹) were prepared from ⁵⁷Fe-enriched elemental iron (> 99.8% isotopic enrichment; Chemgas) by flame spray pyrolysis as described previously⁵. We analyzed the labeled iron compounds for iron isotopic composition and the tracer iron concentration via isotope-dilution mass spectrometry as described below.

Laboratory analyses

Venous blood samples were drawn into EDTA-treated tubes. We measured Hb immediately after blood draw with the use of hematology analyzer (Sysmex, Kobe, Japan) and quality controls provided by the manufacturer before each assessment. Hemoglobin typing for β-globin abnormality was done by using HPLC (Variant Hemoglobin Testing System; BioRad, Hercules. CA) with calibrators and controls provided by the manufacturer. DNA analysis for α-globin abnormalities was done by using a GeneAmp PCR System (Applied Biosystem, Foster City, CA) and a Gel Doc 2000 Gel Documentation System (BioRad, Hercules, CA). A 500 µL aliquot of whole blood was frozen for isotopic analysis (see below). The remaining blood was centrifuged and the plasma aliquoted and stored at − 20 °C. Whole blood and plasma were shipped frozen to the ETH Zurich, Switzerland, for analysis of iron (plasma ferritin and soluble transferrin receptor) and inflammation (C-reactive protein, alpha-1-acid glycoprotein) parameters using a multiplex immunoassay⁵³. Anemia was defined as Hb < 120 g/L⁵⁴. Iron deficiency was defined as plasma ferritin < 12 mg/L and/or soluble transferrin receptor > 8.3 µg/mL⁵³, and iron deficiency anemia was defined as Hb < 120 g/L⁵⁴ and plasma ferritin < 12 mg/L and/or soluble transferrin receptor concentration > 8.3 µg/ml⁵³. Normal C-reactive protein and alpha-1-acid glycoprotein concentrations for this assay in healthy adults are < 5 mg/L and < 1 g/L, respectively⁵³.

Calculation of iron absorption

Whole blood samples were mineralized in duplicate with the use of a nitric acid and microwave digestion followed by separation of the iron from the blood matrix via anion-exchange chromatography and a subsequent precipitation step with ammonium hydroxide. We measured iron isotope ratios by using an inductively coupled plasma mass spectrometer (Neptune, Thermo Finnigan, Germany) equipped with a multicollector system for simultaneous iron beam detection⁵⁵. We calculated the amounts of ⁵⁷Fe and ⁵⁸Fe isotopic labels in blood 14 days after administration of the test meals based on the shift in iron-isotopic ratios and the estimated amount of iron circulating in the body. We remeasured the baseline isotopic composition in blood at the start of the second phase of test meals. Circulating iron was calculated based on the blood volume that was estimated from body length and weight at endpoint measurement according to Linderkamp et al.⁵⁶ and measured Hb (mean Hb from baseline and endpoint). The calculations were based on the methods described by Turnlund et al.⁵⁷ and Cercamondi et al.⁵⁸, taking into account that iron isotopic labels are not monoisotopic.

Data and statistical analysis

Data were analyzed using SPSS (IBM SPSS statistics, version 22.0). Normally distributed data were presented as mean ± SD, and not normally distributed data as median (IQR). Repeated-measures ANOVA was used to assess the effect of iron compound on square root-transformed fractional iron absorption. Fractional iron absorption was the dependent variable and the iron compound was added to the model as the independent variable; pairwise comparisons were performed using two-sided paired t-tests with Bonferroni adjustment for multiple testing. Separate linear regressions were done to compare predictors of iron absorption from the nano-sized iron compounds and the bulk sized compounds. C-reactive protein and serum ferritin (iron status) were added as independent variables with the dependent variable being (1) iron absorption from bulk FePO₄ and FeSO₄ and (2) iron absorption from the two FePO₄-NPs. Significance was set at P < 0.05.