|Home | About | Journals | Submit | Contact Us | Français|
Iron deficiency is a major global public health problem despite decades of efforts with iron supplementation and fortification. The issue lies on the poor tolerability of the standard of care soluble iron salts, leading to non-compliance and ineffective correction of iron-deficiency anaemia. Iron nanoformulations have been proposed to fortify food and feed to address these issues. Since it was just postulated that some nanoparticles (NPs) might cross the plasma membrane also by a non-endocytotic pathway gaining direct access to the cytoplasm, we have studied iron NP uptake under this perspective. To this aim, we have used a recently tested protocol that has proven to be capable of following the cytoplasmic changes of iron concentration dynamics and we have demonstrated that iron oxide NPs, but not zerovalent iron NPs nor iron oxide NPs that were surrounded by a protein corona, can cross plasma membranes. By electrophysiology, we have also shown that a small and transient increase of membrane conductance parallels NP crossing of plasma membrane.
Iron deficiency is a major global public health problem, common in adolescence, in menstruating women and in elderly, and particularly evident in developing countries. In addition, several diseases cause iron deficiency, including intestinal parasite infections, malaria, gastric and duodenal ulcers, gastrointestinal cancer and rare mutations in genes encoding ion transporters1, 2. Iron deficiency can be efficaciously controlled by iron supplementation or food fortification. Unfortunately, the water-soluble and bioavailable ferrous sulphate, fumarate or gluconate, that are currently used therapeutically, affect the gastrointestinal tract with significant side effects such as constipation, diarrhoea and nausea3; moreover, iron salts cause unacceptable changes in the colour and taste of foods4. In contrast, poorly water-soluble compounds cause fewer sensory changes, but present limited bioavailability. Recently, the use of NPs in iron supplementation and fortification has been suggested since their use may combine high bioavailability, good product stability, limited side effects and absence of changes of taste and colour of the fortified foods5–8. Moreover, in vitro and in vivo experiments have shown that iron NPs can be considered safe9, 10. Nanoformulations have also found their way into the fortification of animal feeds11–13. Ferritin, which is well absorbed14, is itself composed of an iron oxide nanocore surrounded by a protein shell and, recently, a nanoparticulate mimetic of the ferritin core was proposed as a potentially side effect-free form of cheap supplemental iron15.
In the gut lumen, dietary iron is present as heme and non-heme (inorganic) iron; inorganic iron can be found in two oxidation states: ferrous (Fe2+) iron and, predominantly, as ferric (Fe3+) iron. Ferric iron is reduced to ferrous iron by duodenal cytochrome b, an enzyme present on the plasma membrane of the enterocyte brush border; Fe2+ is eventually transported into the cytoplasm by means of the divalent metal transporter DMT11, 16 Nanoparticulate iron might follow a different route to access enterocyte cytoplasm. Powell and colleagues15, 17–19 have suggested that nanoparticulate iron, like ferritin20, enters cells by endocytosis reaching the cytoplasm by lysosomal escape. Endocytosis is, indeed, the canonical mechanism by which NPs are known to be up-taken21, but it might not be the only one. The work on ferritin core mimetics focused on highly disperse NPs with an outer layer of charged short-chain carboxylates15, however, poorly charged or ‘naked’ (i.e. lacking an outer coating) NPs may behave differently. In effect, it has been recently postulated that some metal NPs might also cross the plasma membrane by a non-endocytotic pathway22–28 gaining a direct access to the cytoplasm. Indeed, we have recently confirmed this pathway for cobalt oxide NPs. This pathway, which involves perforation of the lipid bilayer, is usually poorly considered and challenges the idea of non-permeability of membranes to hydrophilic molecules or supramolecular structures.
To verify this possibility, we have studied commercially available iron based NPs that do not comprise a dispersing outer layer. We have characterized their size and surface charge and, through a recently developed protocol, we have followed the cytoplasmic concentration dynamics of iron, demonstrating that iron oxide NPs, but not zerovalent iron NPs, were up-taken. We have also shown that the NP crossing of the plasma membrane is accompanied by a transient increase of membrane conductance, but this crossing did not occur when the iron oxide NPs were aggregated or surrounded by a protein corona.
As a preliminary step, we have tested the ability of Calcein to detect intracellular iron concentration increases in fully-grown Xenopus laevis oocytes. Fully-grown oocytes are large cells arrested at prophase of meiosis I, which provide a simple system for the study of membrane transport. Calcein is a divalent metal ion chelating fluorochrome whose fluorescence, at physiological pH, is strongly quenched by cobalt, nickel, copper, iron and manganese divalent cations.
To this aim, we have prepared transfected Xenopus oocytes expressing the Divalent Metal ion Transporter 1 from rat (rDMT1). This membrane protein is a proton dependent transporter of divalent metal ions such as Fe2+, Mn2+, Co2+, Ni2+ and Cd2+ 29–32. We have performed voltage-clamp experiments in transfected (i.e., injected with DMT1 cRNA) and in non-transfected (control) oocytes exposed to FeCl2, both at pH 5.5 and 7.6. We have calculated transport currents as the difference between the currents recorded in the presence and in the absence of 100µM FeCl2 in the external solution. As shown by the current-voltage (I-V) relationship (Fig. 1A), no transport currents were recorded in non-transfected (control) oocytes when exposed to iron ions confirming the absence of endogenous electrogenic Fe2+ transporters in the oocyte plasma membrane. Conversely, in rDMT1 transfected oocytes, at pH 5.5, but not at pH 7.6, the presence of FeCl2 in the extracellular solution elicited large inward currents at negative potentials. This indicates that, at pH 5.5, a transmembrane transport of iron ions occurs in rDMT1 transfected oocytes.
We repeated the same experimental plan in oocytes previously injected with Calcein (25µM) to demonstrate that Calcein was able to detect an intracellular increase of Fe2+ concentration. As shown in Fig. 1B, we monitored the reduction of Calcein fluorescence for 30min in transfected and in non-transfected oocytes exposed to 100µM FeCl2 at pH 5.5 and 7.6. The results clearly show that a significant quenching of the fluorescence occurred only in DMT1 expressing oocytes at pH 5.5. It has to be noted that in DMT1 expressing oocytes at pH 7.6 no current (Fig. 1A) and no Calcein quenching (Fig. 1B) was recorded, according to the transporter pH dependence. These experiments, taken together, demonstrate that Calcein can detect an increase in the concentration of intracellular iron ions, and that oocytes are devoid of endogenous iron transporters.
To define the sensitivity of the method, we have injected oocytes with a fixed amount of Calcein together with increasing amounts of FeCl2. We have then measured the residual fluorescence for each iron concentration and we have recorded a dose dependent reduction of the emission values (Fig. 1C) that we have fitted with a logistic equation (Fig. 1D). Given an oocyte volume of about 1µl, the K0,5 resulted 3.97±2.09 (SE) µM. By means of this curve, we can estimate the concentrations of iron ions in the cytoplasm of Xenopus oocytes. Indeed, Fig. 1B shows that DMT1 transfected oocytes reach, 30min after exposure to FeCl2, quenching values corresponding to concentrations of about the 5µM.
After 30min of exposure of Xenopus oocytes to radioactive iron, Marciani and colleagues33 reported concentrations slightly higher. Their experimental approach based on radioactive 55Fe2+, however, considered the total amount of iron that entered the cytoplasm through DMT1. Since a large amount of iron might become protein-bound, they estimated the total concentration of the iron ions.
After having verified that we were able to detect an increase of iron ions in the cytoplasm of Xenopus oocytes that were filled with Calcein, we have used them to reveal the possible permeation of iron NPs inside the cell. To this aim, we have chosen iron NPs in two different forms, zerovalent (Fe) and oxide (Fe3O4). Similarly to other metal NPs10, 21, 34, iron35 and iron oxide36, 37 NPs undergo dissolution releasing iron ions that, as we have here demonstrated, can be detected by Calcein quenching. As such, we exposed Calcein-filled oocytes to iron NPs. In oocytes obtained from different batches, iron oxide NPs constantly induced a quenching of Calcein fluorescence (Fig. 2, blue column). This fluorescence decrease, although lower than that occurring in rDMT1 expressing oocytes exposed to 100µM FeCl2 (Fig. 2, dark blue column), was significantly greater than that occurring in non-transfected oocytes either exposed (Fig. 1B, black squares) or not exposed (Fig. 1B, white diamond and Fig. 2, white bar) to FeCl2. These results suggest that Fe3O4 NPs interact with the plasma membrane of the oocyte, cross it and, once in the cytoplasm, dissolve causing the observed Calcein quenching. Zerovalent iron NPs, instead, did not cause a reduction of fluorescence (Fig. 2, dark cyan column) suggesting that, in this case, NPs do not cross the plasma membrane of the oocyte.
In a previous paper, we attributed the different biological behaviour (capacity to cross cell membranes) between metal oxide and zerovalent metal NPs to the chemical and physical characteristics of their surfaces. To better understand the causes of these differences, in this paper, we have measured z-potentials and described the aggregation dynamics of both NPs. The z-potentials, as determined by Laser Doppler Micro-electrophoresis (Fig. 3, top), were −5.0±0.2mV and −2.2±0.1mV for zerovalent iron and iron oxide NPs, respectively. Thus, both types of NPs are poorly charged when suspended in external control solution and this does not explain the difference in capacity to cross cell membranes, but explains why both NPs rapidly aggregate into micronized assemblies when added to media (Fig. 3, bottom) despite prior ultrasonication. Indeed, z-potentials above 30mV or below −30mV are generally required to maintain NPs disperse in solution through electrostatic repulsion. However, the rate of aggregation, as determined by Static Light Scattering (Fig. 3, bottom), differed between zerovalent NPs, which aggregated very rapidly, and iron oxide NPs, which retained a sub-micron population for at least 5min (Fig. 3, insets of bottom graphs). We think that the NPs that constitute this sub-micron population are those that can cross the plasma membrane possibly because these remained below a not yet determined upper particle size limit for plasma permeation. Physical penetration of lipid bilayer membranes by NPs has been demonstrated to possibly occur by dissipative particle dynamics simulations for single NPs and for very small clusters. Aggregated NPs, instead, are likely too large to penetrate the lipid bilayer28. Indeed, when the oocytes were exposed to aggregated NPs, we did not record Calcein quenching (Fig. 2, blue barred column).
To learn more about the interactions of NPs with the plasma membrane of the oocytes, we have followed possible changes of its integrity due to exposure to Fe3O4 NPs by Two Electrodes Voltage Clamp (TEVC). With a similar approach, Bernareggi and colleagues38 have recently evaluated the effects of asbestos fibres on Xenopus oocyte membrane.
We have measured membrane capacitance, resting potential and resistance during a 30min period of exposure to Fe3O4 NPs. Since we have noticed that the alterations of membrane parameters occur within the first 5min, we have decided to perform the electrophysiological measurements after 5 and 20min of exposure; we have also exposed oocytes to NPs that, after sonication, were left undisturbed for 30min. A statistically significant decrease of membrane resistance, as shown in Fig. 4, occurs only in oocytes exposed for 5min and disappears completely in oocytes exposed for 20min. A slight decrease of resting potential, statistically significant, accompanied the decrease of membrane resistance. Membrane capacitance, instead, remained unaffected by Fe3O4 NPs treatment.
The decrease of membrane resistance can be attributed to the opening of a small transient conductance. On the other hand, the constancy of membrane capacitance is an indication that no large endocytosis phenomena occur after exposure to iron oxide NPs39, while the small reduction of the resting potential proves that the membrane has maintained its integrity40, notwithstanding a small increase of ion permeability.
To investigate further this transient conductance, we have performed a protocol to obtain I-V curves. Figure 5A shows representative current traces recorded after imposing voltage steps in a control oocyte and in an oocyte exposed to Fe3O4 NPs for 5min. We have calculated the average currents at each potential and reported these data in the I-V curves of Fig. 5D. At potentials more negative than −40mV and more positive than +30mV, a slight increase of the currents obtained in exposed oocytes, compared to those in the control condition, is evident. This current increase is statistically significant only in oocytes tested within 5min of the NPs treatment and it tends to disappear in oocytes exposed for more than 20min and is not present in oocytes exposed to NPs that, after sonication, were left undisturbed for 30min. The I-V curves confirm the onset of a conductance caused by the exposure to Fe3O4 NPs for 5min and its transient nature.
To verify our hypothesis concerning the role of the interactions between NPs and plasma membrane, we tested uptake of Fe3O4 NPs in bovine serum albumin (BSA) containing media. These molecules create a coating around the NPs, referred to as protein corona. This corona is highly dynamic and its formation and stability depends on NP characteristics and environmental variables41. BSA is negatively charged in water, but not in this media (Supp Figure S1) presumably due to the presence of positively charged calcium and magnesium ions. Consequently, the addition of BSA did not alter the surface charge of the iron oxide NPs (zeta potential was −1.3±0.2mV; Fig. 6 top). However, the addition of BSA did increase dispersibility as determined by the presence of the sub-micron fraction for the duration of the assay (Fig. 6). As shown above, plasma membrane crossing occurs in the presence of disperse colloids; however, exposure to BSA-coated Fe3O4 NPs did not cause fluorescence quenching (Fig. 2, cyan column). Therefore, protein corona prevents or significantly reduces the interactions between NPs and the oocyte plasma membrane blocking or limiting the passage of NPs into the cytoplasm. These findings are also in agreement with experiments where the interaction of cationic polystyrene NPs with artificial lipid bilayers were eliminated with serum proteins42.
To confirm this hypothesis, we have performed TEVC experiments to obtain I-V curves (Fig. 7). After having verified that BSA per se does not affect membrane conductance, we have compared oocytes exposed to BSA-modified Fe3O4 NPs with unexposed oocytes, and to oocytes exposed to Fe3O4 NPs. These experiments suggest that, when NPs are surrounded by a protein corona, they become unable to modify membrane resistance.
In conclusion, we have added further evidence to the possibility of cytomembranes being permeable to uncoated NPs. Endocytosis is the typical mechanism cells use to uptake NPs that, in this way, gain access to the endosomal compartment. Conversely, by directly crossing lipid bilayers, NPs enter cytoplasm and other cellular compartments. Moreover, we have shown that the penetration by NPs of the lipid bilayer opens a transient conductance that does not impair plasma membrane integrity.
Whether we should favour NP presence in the cytoplasm or avoid it, we have to understand the mechanisms that regulate their uptake and consider how physicochemical properties of NPs impact on their fate. In this work, we have also learned that, unlike iron oxide NPs, zerovalent iron NPs do not cross plasma membranes. We have explained this different behaviour with the very fast aggregation rate of zerovalent iron NPs. In addition, we have shown that the crossing of metal oxide NPs can be impeded by a protein corona. Indeed, proteins are abundant in the tissue fluids, but NPs could be functionalized to become corona-free43, thus potentially enabling the non-endocytic pathway. For example, superparamagnetic iron oxide NPs can be induced to rotate around their axes by a remote magnetic field44 and have been used to kill cancer cells through mechanical rupture45. Understanding the mechanisms that regulate NP uptake or their interactions with the cellular membrane could help in potentiating the anticancer properties of such NPs.
We also understand that the conditions iron NPs find in the gastrointestinal tract greatly differ from those of the simple model system we have used. In particular, the intestinal millieu comprises a diverse group of soluble species, ranging from small carboxylates throught bile acids to high molecular weight mucins46–48, many of which may interact with the surface of uncoated NPs and, in doing so, may alter their behaviour. This now merits assessment in future studies using conditions that are representative of the complex gastrointestinal environment.
ND96 solution had the following composition (in mM): NaCl 96, KCl 2, CaCl2 1.8, MgCl2 1, HEPES 5, pH 7.6; NDE solution was composed of ND96 plus 2.5mM pyruvate and 50µg/mL Gentamycin sulphate; external control solution contained (in mM): NaCl 98, MgCl2 1, CaCl2 1.8, HEPES or MES 5, pH 7.6 or 5.5; intracellular solution contained (in mM): KCl 130, NaCl 4, MgCl2 1.6, EGTA 5, HEPES 10, glucose 5, pH 7.6. The final pH values of 5.5 or 7.6 were adjusted with HCl and NaOH.
Oocytes were obtained from adult Xenopus laevis females. Animals were anaesthetised in 0.1% (w/v) MS222 (tricaine methansulfonate) solution in tap water; after carefully cleaning the frog abdomen with an antiseptic agent (Povidone-iodine 10%)49, laparotomy was performed and portions of the ovary were collected. The oocytes were treated with 0.5mg/mL collagenase (Sigma Type IA) in ND96 calcium free for at least 30min at 18°C. Healthy and fully grown oocytes were selected and stored at 18°C in NDE solution50. The oocytes to be transfected with the cRNA coding for rDMT1 were injected with 25 ng of cRNA in 50 nL of water, the day after the removal, using a manual microinjection system (Drummond Scientific Company, Broomall, PA) and incubated at 18°C for 3–4 days before electrophysiological or fluorescence experiments. The experimental protocol was approved locally by the Committee of the “Organismo Preposto al Benessere degli Animali” of the University of Insubria (OPBA-permit #02_15) and nationally by Ministero della Salute (permit nr. 1011/2015).
Zerovalent (Fe, 25nm, IOLITEC, Salzstrasse 184, D-74076 Heilbronn) and oxide (Fe3O4, <50nm TEM determined, Sigma-Aldrich) iron NPs were prepared as 10mg/mL stock suspensions in deionised water and sonicated before addition of 10μL to the test chamber (4 wells plate) containing 990μL of external control solution (pH 7.6). Oocytes were then added to the chamber.
For electrophysiological experiments, the oocytes were transferred to the TEVC recording chamber within 5min or after 20min from the beginning of the exposure. For the fluorescence assay, recordings start immediately after oocytes addition to the test chamber. Treatments with the aggregated NPs were performed adding oocytes to the suspension of NPs that were let to aggregate in the test chamber for 30min after sonication.
All experiments were carried out at room temperature.
Oocytes were injected with a 50 nL drop of intracellular solution containing 25µM Calcein. In dose-response evaluation experiments (Fig. 1C and D), solutions containing Calcein and increasing amounts of FeCl2 were co-injected in oocytes. Effective intracellular metal concentrations were calculated using the nominal oocyte volume of 1µL51. Images were acquired for every oocyte at fixed acquisition parameters, calculated on control oocytes.
In experiments with NPs, images of single oocytes were acquired every 2min for 30min with a fluorescence microscope (AxioVert 200, Carl Zeiss with a 4x objective, COLIBRI fluorescence filters, 470nm excitation - 515 to 565nm emission) equipped with CCD camera (Axiocam ICM1, Carl Zeiss).
The two-electrode voltage clamp was performed with an Oocyte Clamp OC-725B (Warner Instruments, Hamden, CT, USA) that was controlled by Clampex 10.2 (Molecular Devices, Sunnyvale, CA, USA, www.moleculardevices.com). Intracellular glass microelectrodes, filled with 3M KCl, had tip resistances in the 0.5–4MΩ range. Agar bridges (3% agar in 3M KCl) connected the bath electrodes to the experimental chamber. The holding potential applied (Vh) was −40mV for all the experiments performed. The I-V curves of Fig. 1A were obtained applying 20mV steps of 200 ms from −140 to +40mV.
Oocytes were transferred in the recording chamber (Warner- RC-1Z) (warneronline.com) and impaled with microelectrodes; to recover from possible damage they were left for 2min in a continuous solution flux. Only oocytes with resting potential equal or lower than −20mV were used for the experiments. The number of discarded oocytes was not significantly different between treated and controls.
The capacitance and resistance values that were reported in Fig. 4 were obtained applying 10mV steps of 20 ms every 200 ms in voltage clamp conditions. The protocol for the I-V curves of Fig. 1 was previously described52 and for Figs 5 and and77 is shown in Fig. 5A (10mV steps of 750 ms from −80 to +40mV).
Data were analysed using Clampfit 10.2 software (Molecular Devices, Sunnyvale, CA, USA, www.moleculardevices.com) while OriginPro 8.0 (OriginLab Corp., Northampton, MA, USA, www.originlab.com) was used for statistics and figure preparation. For the Voltage Step protocol, current values were measured for every voltage step at the steady state condition; mean values at every voltage were calculated and plotted.
Fluorescence decay images were analysed with ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, http://imagej.nih.gov/ij/, 1997–2015). For F30/F0 quantification, the fluorescence intensity at time 0 (F0) and at time 30min (Ft) was calculated in the entire area of the oocyte. In dose-response experiments, images were analysed calculating fluorescence intensity on the entire area of oocytes and normalized to the control oocytes. Mean values were calculated for every condition and used to determine through a non-linear fitting (Logistic Fitting, OriginPro8) the K0.5, relating residual fluorescence to metal concentration.
SLS was performed on a Mastersizer 2000 with a Hydro 2000µP Micro Precision sample dispersion unit (Malvern Instruments Limited). Baseline correction was carried out with fresh external control solution (prepared as described above). Next, 200µL of 10mg/mL stock suspension was added to the 20mL dispersion cell to achieve a final concentration of 0.1mg/mL (as per Oocyte assays). The dispersion unit was run at 1000rpm and care was taken to prevent bubble formation. The stock suspension was carefully sonicated prior to introduction to the dispersion cell, after which data acquisition was immediately initiated. Size measurements then were carried out over a period of 60min (refractive index: 2.42; absorption 1.0; dispersant refractive index: 1.33).
The zeta potential of suspensions of iron NPs was determined by Laser Doppler Micro-electrophoresis (Zetasizer NanoZS, Malvern Instruments Ltd) using disposable folded capillary cells (DTS1070). Measurements (N=3) were carried out using the diffusion barrier as per instrument manufacturer instructions. Briefly, the zeta cell was filled with 100mM NaCl solution and then 100µL sample (0.1mg/mL suspension of NPs in external control solution) was gently injected to the bottom of the cell with a gel-loading tip. Electrophoretic mobility of particles was converted into zeta potentials by Dispersion Technology Software 7.11 using the Smoluchowski approximation, and a viscosity of 0.8872cP and dielectric constant of 78.5 for the dispersant.
This work was supported by a grant of Fondazione Cariplo (2013-1052) to R.G. We thank Dr. Raffella Cinquetti for her help in performing experiments. This work was supported by the Medical Research Council (programme grant number: MC_U105960399).
All authors designed the experiments and reviewed the manuscript. D.Z., E.B. and C.B. performed the experiments and prepared the figures, E.B., N.F., R.G. and G.B. wrote the manuscript.
The authors declare that they have no competing interests.
Electronic supplementary material
Supplementary information accompanies this paper at doi:10.1038/s41598-017-11535-z
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.