Magnetic analyses of human tissues have revealed the presence of nanosized ferrimagnetic magnetite/maghemite particles (MNPs) in the brain, meninges, heart, liver, and spleen (1–7). The two distinct types of magnetic crystals identified in human brain are euhedral crystals possibly of biogenic origin (1, 5, 8, 9) and rounded crystals of combustion-derived origin (5). Many organisms are able to biochemically precipitate magnetite (10), and one theory is that biogenic magnetic nanoparticles in human tissues crystallize in situ, with the possible precursor being ferritin (11, 12).

The magnetic properties of magnetite depend on magnetic relaxation time (13), determined by particle size, morphology, the microscopic coercive force of the particle, saturation magnetization of the material, and thermal energy (14–17). Particles with ultrashort relaxation times are referred to as superparamagnetic (SPM). For magnetite, particles under 30 nm are considered SPM, whereas those of 30–80 nm are single-domain (SD) particles (14, 17, 18). At room temperature, SD particles are able to carry magnetic remanence, while SPM are not (18). Generally, static magnetic fields can reduce mitochondrial membrane potentials, generate oxidative stress, and induce apoptotic processes (19, 20). Magnetic nanoparticles can translate possible effects to tissues. In human cells, magnetic nanoparticles have induced loss of mitochondrial membrane potential as an early sign of apoptosis (21). The magnetic iron oxide, magnetite (Fe²⁺Fe23+O₄), and its oxidized equivalent maghemite (γ-Fe23+O₃), may each possibly play a role in the pathogenesis of neurodegenerative diseases (3, 22). Iron and MNPs, by also containing ferric iron (Fe³⁺), can promote alpha-synuclein protein (α-Syn) aggregation (23, 24). Importantly, the α-Syn aggregates in the central and peripheral nerves are hallmarks of Parkinson's disease (PD) (25). Phosphorylated α-Syn aggregates have been detectable in cervical skin samples of PD patients but not in control samples (26). Exploring the occurrence of MNPs in skin tissue may thus reveal important information on PD pathogenesis.

The primary aim of the present study was to examine the presence and potential role of MNPs in the cervical (neck) skin of subjects with and without PD. This site was targeted because it has most consistently harbored alpha-synuclein protein pathology in PD skin samples (26). We studied the mineralogy, morphology, and composition of MNPs in human cervical skin samples by magnetometry, transmission electron microscopy (TEM), and the associated spectroscopic techniques, electron energy loss spectroscopy (EELS) and energy-dispersive X-ray spectroscopy (EDS).

Alternating field (af) demagnetization of SIRM of the samples supports the presence of ferrimagnetic remanence carriers (Figures 1A,B). IRM acquisition curves showed that the skin tissue samples contained low-coercivity magnetic particles which were magnetically saturated by 300 mT, indicating ferrimagnetic magnetite/maghemite (Figures 1C–F). Wohlfarth's ratios (28, 29) between 0.19 and 0.24 for both PDs and controls were obtained (Figure 1). The MDF of SIRM for PD samples showed a larger range (7.9–20.5 mT) than did those for control samples (10–18 mT) (Figures 1, 2), indicating a wider range of domain size of remanence carriers for PD samples than for controls. The limited number of subjects did not, however, provide sufficient statistical power to detect a significant difference. The Lowrie-Fuller (30) test, where MDF values of ARM are higher than those of IRM, indicates that the SD size MNPs dominate over MD size in all of the samples.

The concentration of magnetic material in the samples was calculated, assuming that the material was magnetite, having saturation magnetization 92 Am²/kg (1). The highest observed values of magnetite content were in the PD group, which displayed a higher spread of the distribution, most clearly in the dermal (D) samples (median 2228.1 ng/g, range 904-27842 ng/g; in controls, median 1884.2 ng/g, range 943–5,700 ng/g) (Figure 2 and Table 2). Despite the trend, no significant difference appeared between the subpopulations (p = 0.34, Mann-Whitney U-test). Values of the dermal samples showed no gender difference (p = 0.12, Mann-Whitney U-test) nor did they correlate with age or body mass index of study participants (R = −0.01, p = 0.97, and R = −0.32, p = 0.17, respectively, Spearman Rank Correlation).

Induced magnetization, after both low-temperature experiments, where the sample was cooled in a zero field (ZFC) and cooled in an applied field of 50 mT (FC), showed a rapid loss in magnetization on rewarming from 5 to 25 K. This range includes the temperature range for a paramagnetic signal from blood in tissue (2–8 K) (7, 31). With further warming, the ZFC showed an anomaly for the majority (six of ten) of the D samples of PD patients, and for one D sample and one E sample of the controls, with the peak ranging from 41 to 54 K (Tables 2, 3 and Figure 3). As to dermal samples, Fisher's exact test showed a trend-like difference in the occurrence of this anomaly between patients and controls (p = 0.06). The anomaly was absent for the FC curve. With continued warming, ZFC curves indicated Verwey transition for D samples of six PD patients at a temperature range of 118–123 K and for five D samples of controls at a temperature range of 115–125 K (Figure 3 and Table 2).

We measured induced magnetization as a function of the field for PD patient 18D near the low temperature anomaly peak at 48 K and at 300 K (Figure 4). After the diamagnetic correction, nearly saturated magnetization emerged at both temperatures, indicating ferrimagnetic material. A slightly higher coercivity value of 20 mT appeared at 48 K, compared to 15 mT at 300 K. At 48 K, in the descending and ascending limbs of the curve in both positive and negative fields showed a bifurcation. At 300 K, this behavior was absent.

TEM and EELS analyses of the extracted particles confirmed the occurrence of magnetite (Figure 5). The size of these particles varied in the range of 50–100 nm, close to SD size, exceeding the size of nanoparticles formed within an 8-nm diameter of the ferritin cors (10). The nanoparticles displayed a faceted morphology, in contrast to the rounded or spherical morphologies found in brain tissue and attributed to combustion-derived sources (5). Energy loss near ege structure (ELNES) analysis of the Fe L2,3-edge clearly shows a magnetite profile, both in patients and controls (Figure 5). The L3–edge presents a single maximum located at 709.4 (±0.1) eV, and L2–edge at 721–723 eV. The energy between the two lines being 13.2 eV and the relative intensity 4.0, are in agreement with earlier values for magnetite particles (32). The O K-edge shows a pre-peak around 530 eV that presents an asymmetry, with a shoulder located at 530 eV on the high-energy side, and a principal feature at 540 eV that is common for all iron-oxide phases.

Room-temperature IRM and SIRM measurements identified ferrimagnetic particles in all of the samples. The broad range of Verwey transition temperatures (PD: 118–123 K; controls: 115–125 K) indicates both pure magnetite and partial oxidation of magnetite to maghemite or the presence of impurities (33), such as Ti⁴⁺, Zn²⁺, Al³⁺, in magnetite. TEM observations demonstrated that some of the particles were pure magnetite (Figure 5). Lowrie-Fuller test (30) refutes the presence of MD grains and Wohlfarth's ratios < 0.5 for both PDs and controls indicate that the grains capable of carrying remanent magnetization at room temperature are magnetically interacting (28, 29). The median MNP concentrations for the dry skin-tissue samples were in the same range as reported for dry brain-tissue samples (5). Although the highest concentrations of MNPs came from the PD patients, no statistical difference between PD patients and controls emerged, possibly due to the small number of samples studied.

The majority of the D samples from PD patients showed anomalous behavior in the direct current magnetization at around 50 K. Earlier low temperature measurements have indicated that heme-iron and ferritin do not contribute to the obtained anomaly around 50 K (7, 31). The only iron phase in human tissue, one that has shown anomalous behavior in its magnetic properties around 50 K, is magnetite/maghemite (34, 35). An anomaly at ca. 50 K in measurement of magnetic after-effect (MAE) or AC susceptibility has been associated with the onset of electron hopping upon warming and ionic ordering within domain walls, which indicate wall movement (36). However, the size of MNPs in this study indicates SD or SPM grain size, contradicting the presence of domain walls, being in line with the size range of MNPs in human brain tissue (7).

In 1998, Moskowitz et al. associated a low-temperature anomaly during ZFC analyses with SPM- to SD-sized maghemite particles (partly aggregated clusters) in magnetoferritin (37). Accordingly, we propose that the ~50 K anomaly indicates magnetically interacting clusters of possible SPM- and SD- sized ferrimagnetic MNPs. The peak for this perturbation occurred only on the ZFC curve, since, due to increased temperature, initially SD sized particles progressively unblock, align their moments with the applied field, and eventually randomize due to thermal energy. This anomaly did not occur during the FC analyses, since magnetization was already aligned with the field during the cooling through blocking temperature and this experiment only shows the demagnetization of the blocked particles.

Additional support for SPM grains comes from low temperature hysteresis measurements, which show higher coercivity at 48 K than at room temperature (Figure 4B), indicating the presence of SPM grains as being able to carry remanence at 48 K (4). The bifurcations similar to those obtained in the hysteresis loop have been evident in studies on human tissue (7, 36), with a suggestion that the switching of the a-b crystalline axes of magnetite around 45 K could possibly explain this bifurcation (36). In the initial field, the easy axis of magnetization would lie along one of the axes. As the field is reduced and then raised to the starting value again, what may occur is a switch in preferred axis, leading to the bifurcation observed.

Another explanation for the anomaly at ~50 K may be the magnetic ordering of molecular oxygen, which undergoes an antiferromagnetic transition at about 43 K being strongly paramagnetic above this temperature (38). We took care to prevent any oxygen leak into the system. Furthermore, the magnetic intensity of the ~50 K signal was for most of the samples an order of magnitude larger (10⁻⁴ Am²/kg) than known effects related to the magnetic ordering of oxygen at 43 K (< 2 × 10⁻⁵ Am²/kg) (7), providing additional evidence against the oxygen as a cause for the ~50 K anomaly.

Our magnetometric studies revealed MNPs with SD characteristics in all skin samples and interacting SPM characteristics for the majority of the D samples from the PD patients. The shapes of the SD-size magnetic nanoparticles studied by TEM did not resemble the cubo-octahedral crystals of possible biogenic magnetite (1, 9), but merely resembled non-biogenic magnetite. No rounded or spherical morphologies were detectable. EEL spectra of four samples demonstrated that their particles were composed of magnetite (Figure 5); on the basis of their size, we propose a source different from that of ferritin cores.

The origin of MNPs in human tissues has remained speculative (biogenic, non-biogenic, or anthropogenic/combustion). Polluted air (combustion-derived MNPs) via the respiratory route has been proposed as one source of these MNPs (5). The olfactory route was considered to be the passageway to the brain for these round and spherical particles (5), but was recently contradicted (6). Here, no particles were detectable with such morphology. However, we cannot fully exclude that airborne MNPs, via a respiratory route or even through skin structures, may in part explain the results. Low-temperature magnetic measurements showed the presence of apparent SPM particles mainly in the dermal samples of PD patients (Tables 2, 3), hinting strongly of a source other than the skin surface. As to intrinsic sources, the ferrihydrite core of the iron-storage protein ferritin has been proposed to be a precursor of MNPs (11, 12). Considerable argumentation goes against this view, as the size of single-domain MNPs is not compatible with the 6- to 8-nm diameter of the inner cavity of the ferritin protein (10). Further, nuclear magnetic resonance relaxometry studies have shown that the content of magnetite in the ferrihydrite core is < 1% (39). At least, what is not ruled out is that ferritin could by some mechanism produce small superparamagnetic MNPs, which can then aggregate into larger MNPs with SD-like characteristics. Experimens have shown that nucleation and growth of magnetite proceeds through rapid agglomeration of nanosized primary particles (40). Recently, it has been reported that human stem cells have an ability to synthesize magnetic nanoparticles from nanodegeneration products (41).

A potential source of MNPs may be the gastrointestinal tract. Having been present in many organisms (10), MNPs may be expected to be found in various foodstuffs, which, when ingested, are probably absorbed into the bloodstream from the gastrointestinal tract. In addition, drinking-water can contain MNPs (42). Ferrihydrite in soils deserves attention, since in its amorphous form, it can percolate into groundwaters and further into drinking-water (43), and magnetite crystal formation may occur by an abiotic mechanism (44). As to bacteria sporadically residing in the human gastrointestinal tract, sulfate-reducing Desulfovibrio species are able to synthesize magnetite (45–47). As these bacteria also produce hydrogen sulfide, known to cause olfactory dysfunction (48), a prevalent phenomenon in PD (49), the potential occurrence of these bacteria in PD is relevant. The uptake of nanoparticles from the intestinal lumen into the blood circulation may take place mainly through an endocytic transcellular transport, with even 100-nm-sized particles transported (50). Iron nanoparticles can evidently cross plasma membranes by a non-endocytic pathway, gaining direct access to the cytoplasm (51). The gut epithelial enteroendocrine cells, known to contain α-Syn, may therefore be structures to display α-Syn aggregation promoted by magnetite nanoparticles (24, 52, 53). In the blood, nanoparticles become coated by proteins, small molecules and ions (50). Tissue macrophages are considered the primary cell types that phagocytose nanoparticles which have ended up in the blood circulation (54). As the macrophage density in normal human dermis is considerable (55), skin may play an important role in removing MNPs from the blood circulation. This view is supported by an animal study showing intravenously administered nanoparticles to accumulate in the dermis, first localizing in dermal macrophages (56). The substantial magnetite content of the skin samples detected in this study may reflect such a process.

MNPs can evidently interfere with cell functions. Interactions of SPM particles with prolonged external magnetic fields may lead to biological impacts, because SPM particles have a strong magnetic susceptibility; relatively weak static magnetic fields can induce loss of membrane potentials in mitochondria (19, 20), which may potentially lead to mitophagy (56). Magnetite nanoparticles per se may induce mitochondrial dysfunction (21), and promote, in an uncoated form, α-Syn aggregation (24).

In the present pilot study, the number of samples was too small to allow definitive conclusions as to the true differences in MNP characteristics and quantities between the study groups. As an additional factor, PD is a disease with diverse characteristics in which various pathophysiological mechanisms may come into play (57), and the overall accuracy of clinical diagnosis of PD is far from satisfactory particularly in its early stages (58, 59). Notably, relatively high amounts of single-domain MNPs were observable in all cervical skin samples. A possible clustering of such particles with superparamagnetic particles needs further exploration in Parkinson's disease. Our study raises questions as to the origin and role of MNPs in physiologic and pathologic states in human beings as well as to the importance of the skin as an internalizer of these particles.

The datasets generated for this study are available on request to the corresponding author JS.

Ethics Committee of Helsinki and Uusimaa Health District Area of Finland.

KM and JS contributed equally to the work in this research. KM, JS, LP, DF, and EM-S designed the research. KM conceived the original idea. KM, JS, JW, JV, DF, EM-S, LS, MP, and PS performed research. JS, KM, JW, LS, LP, DF, and EM-S analyzed the data. KM, JS, and JW wrote the manuscript, including input from EM-S, DF, LS, and LP.

Human internal organs have been reported to harbor magnetic nanoparticles. In 1992, Kirschvink and his colleagues were the first to report magnetite nanoparticles in human brain tissue. Later, apparent magnetite and maghemite particles were found in the human heart, spleen, and liver. With their magnetic properties, these iron oxide particles may have the propensity to induce deleterious effects in cells and play a role in the pathogenesis of neurodegenerative diseases. The present study, using magnetometric and electron microcopy methodology, revealed ferrimagnetic single-domain-sized magnetite/maghemite nanoparticles in all skin samples of patients with Parkinson's disease as well as in those of healthy controls, and revealed superparamagnetic material in several samples. These findings complement studies focused on magnetic material in human tissues. The particles are proposed to be airborne or gut-mediated, and the skin may play a role in internalizing them from the blood circulation. The present findings open a new perspective on the occurrence of magnetic nanoparticles in human tissues and underline the need to clarify not only the origin of these particles but also their role in physiologic and pathologic states. To this end, the skin offers a convenient target.