Magnetosome-inspired synthesis of soft ferrimagnetic nanoparticles for magnetic tumor targeting

Significance

Magnetic targeted delivery of nanoparticle drugs has become one of the most promising means of tumor imaging and drug therapy. Inspired by magnetosome biomineralization in magnetotactic bacteria (MTB), in this study, we construct a biomimetic nanoreactor similar to that of the magnetosome by integrating Mms6 protein into a reverse micelle system. The magnetosome-like magnetic nanoparticles (MNPs) with a single domain were synthesized in this magnetosome-inspired nanoscale chamber. Their morphology and magnetic property were subsequently characterized and compared with the natural magnetosomes produced by AMB-1 MTB. The small size of magnetosome-like MNPs and their strong magnetic targeting ability produced by soft ferromagnetism improved the tumor penetration by an order of magnitude, showing a positive contrast in the tumor area.

Abstract

Magnetic targeting is one of the most promising approaches for improving the targeting efficiency by which magnetic drug carriers are directed using external magnetic fields to reach their targets. As a natural magnetic nanoparticle (MNP) of biological origin, the magnetosome is a special “organelle” formed by biomineralization in magnetotactic bacteria (MTB) and is essential for MTB magnetic navigation to respond to geomagnetic fields. The magnetic targeting of magnetosomes, however, can be hindered by the aggregation and precipitation of magnetosomes in water and biological fluid environments due to the strong magnetic attraction between particles. In this study, we constructed a magnetosome-like nanoreactor by introducing MTB Mms6 protein into a reverse micelle system. MNPs synthesized by thermal decomposition exhibit the same crystal morphology and magnetism (high saturation magnetization and low coercivity) as natural magnetosomes but have a smaller particle size. The DSPE-mPEG–coated magnetosome-like MNPs exhibit good monodispersion, penetrating the lesion area of a tumor mouse model to achieve magnetic enrichment by an order of magnitude more than in the control groups, demonstrating great prospects for biomedical magnetic targeting applications.

With the ongoing development of nanotechnology, the targeted delivery of nanoparticle drugs has become one of the most promising means of tumor imaging and drug therapy. In recent years, many reports have shown that although nanomedicines can enter targeted tumor regions, either through the passive targeting enhanced permeability and retention effect or tag-modified active targeting, the average tumor targeting efficiency, however, is less than 1% (1, 2). Consequently, how to enrich nanomedicines in tumor tissues, specifically to improve the effective penetration efficiency of tumor tissues, remains a challenge (2, 3). A magnetic targeting strategy applies an external magnetic field to a specific area after intravenous injection, enriching and capturing magnetic nanoparticles (MNPs) as they flow through the blood vessels in the region, so as to reduce the near-miss effect of systemic drug administration (4, 5). Like other nanomedicines, MNPs have low targeting and tissue penetration in solid tumor tissues (1) due to the increased interstitial fluid pressure and the dense extracellular matrix (ECM) in tumor tissues (4, 6). To overcome these biological barriers, an efficient magnetic targeting system requires MNPs to have the following characteristics: (1) a sufficient magnetic moment and saturation magnetization (M_s) to realize an effective and rapid response to an external magnetic field; (2) be superparamagnetic to avoid agglomeration of MNPs; (3) be a small size to improve penetration ability (7, 8).

As a natural MNP of biological origin, the magnetosome is a special “organelle” formed by biomineralization in magnetotactic bacteria (MTB) and is essential for MTB magnetic navigation response to geomagnetic fields (9–11). For AMB-1 MTB, magnetic nanocrystals of a cubo-octahedral shape can be produced in the magnetosome, a nanoscale mineralization chamber (9, 10). The unique magnetic properties of the AMB-1 magnetosome enables AMB-1 MTB to promptly respond to external magnetic fields (12–14). Its small size (35 to 50 nm), high M_s, good stability, and low toxicity make the magnetosome an excellent candidate for magnetic targeted nanocarriers and magnetic resonance imaging (MRI) applications (14–16). However, the strong magnetic interaction between natural magnetosome particles causes aggregation and precipitation in water and biological fluid environments, seriously undermining their ability to penetrate diseased tissues (16–18). Consequently, it can be an attractive research challenge to synthesize magnetosome-like MNPs that retain the advantages but not the drawbacks of natural magnetosome MNPs.

The size, morphology, and magnetism of magnetosomes can be determined by the space-limited mineralization chamber using a variety of regulating proteins. Among many magnetosome proteins, Mms6 has been found to bind tightly to magnetite particles in AMB-1 and is a key protein used to control mineralization kinetics (19, 20). Fig. 1A shows a simplified schematic diagram of the magnetosome, with the Mms6 protein residing within the magnetosome membrane lumen via its hydrophobic N terminus and interacting with the magnetite crystal directly via its C-terminal crystal-binding motif (21, 22).

Fig. 1.

Nanoscale reaction chambers for biomineralization. (A) Magnetosome from MTB, where the Mms6 protein assembles into a mineralization template on the magnetosome membrane inner surface. The hydrophilic acid–rich region of Mms6 is highlighted in light orange and the N-terminal hydrophobic membrane-associating region in green, respectively; the red dotted frame highlights the lipid bilayer of the magnetosome membrane. (B) Schematic illustration of the synthesis process of magnetosome-like MNPs within the nanoreactor of an Mms6-containing reverse micelle. The red dotted frame highlights the surfactant monolayer in the magnetosome-like MNPs.

Recently, many advances have been made in the design and construction of nanoreactors as mineralized templates. These biotemplates can be built from individual biological components, either naturally occurring, such as ferritin-based protein cages (23, 24), or synthetic, that is, self-assembled peptide cages (25). Among these efforts is the use of reverse micelles, which are aqueous droplets of small size and separated from the bulk organic phase by a surfactant layer (26–28). The constrained aqueous “chamber” acts as a nanoreactor in which the mineralization reactions occur. The size of the chamber can be adjusted by varying the water:surfactant molar ratio (26), ultimately determining the sizes of the synthesized microcrystals. The system allows good control over the size and shape of synthesized particles, including gold and magnetite nanoparticles (27–29).

Inspired by magnetosome biomineralization in MTB, in this study we integrated Mms6 protein into a reverse micelle system to construct a biomimetic nanoreactor (Fig. 1B). The morphology, magnetic properties, and MR relaxation properties of the obtained MNPs were characterized and compared with the natural magnetosomes produced by AMB-1 MTB. The small size of magnetosome-like MNPs and their strong magnetic targeting ability produced by soft ferromagnetism improves the tumor penetration by an order of magnitude, demonstrating a positive contrast in the tumor area. This combination of magnetic targeting, tumor penetration, and MR imaging make magnetosome-like MNPs very promising for potential nanomedicine applications.

Results

The Construction and Characterization of an Mms6-Containing Reverse Micelle Nanoreactor.

We constructed a magnetosome-like nanoreactor and synthesized Fe₃O₄ MNPs through thermal decomposition, as shown in Fig. 1B. The Mms6-containing reverse micelle system was prepared by mixing Mms6 protein powder and Fe(acac)₃ with oleylamine in a benzyl ether solution containing 0.1% water. The solution became clear after being heated at 60 °C for 3 h. Dynamic light scattering (DLS) analysis shows that reverse micelles formed with a hydrodynamic size of ∼9 nm (Fig. 2A). As the temperature increases to 200 °C, the reactant iron (Fe) acetylacetonate decomposes, and Fe ions enter the aqueous pools of the reverse micelles, eventually forming ferroferric oxide. The postreaction reverse micelles are of a similar size (∼9 nm), demonstrating the reverse micelle system to be fairly stable during the assembly and thermo-decomposition reaction.

Fig. 2.

Construction of the Mms6-containing reverse micelle nanoreactor and characterization of Mms6 during different reaction stages. (A) DLS profiles of the reaction buffer without oleylamine surfactant (gray line), reverse micelle reaction system after reaction at 60 °C for 3 h (red line), and reverse micelle reaction system after reaction at 200 °C for 8 h (blue line). (B) SDS-PAGE analysis of Mms6 after reaction at 60 °C and 200 °C. Lanes 1 and 4, purified Mms6 protein; lanes 2 and 5, supernatant of control reaction without Mms6 participation; lanes 3 and 6, supernatant of reaction at 60 °C (lane 3) and 200 °C (lane 6) with Mms6 participation. Red arrow: Mms6 tetramers; orange arrow: Mms6 dimers; blue arrow: fragments of Mms6 degradation. (C) The ¹H,¹⁵N-HSQC spectra of Mms6 in the reverse micelle nanoreactor after reaction at 60 °C. Red dotted box: glycine or tryptophan. (D) The ¹H,¹⁵N-HSQC spectra of Mms6 in the reverse micelle nanoreactor after reaction at 200 °C. Blue dotted box: glycine. (E) Mms 6 sequence with the hydrophilic acid–rich region in blue and the hydrophobic membrane region in pink. The scissors represent the degradation site.

To further investigate the state of Mms6 protein during the two reaction stages, that is, 60 °C and 200 °C, we used ¹⁵N isotope–labeled Mms6 protein for ¹H,¹⁵N-heteronuclear single quantum correlation (HSQC) experiments (Fig. 2 C and D). As shown, Mms6 maintains its protein integrity at 60 °C, but it degrades into two peptide fragments after reaction at 200 °C (Fig. 2B and SI Appendix, Fig. S1). The nuclear magnetic resonance (NMR) spectrum of the hydrophilic fragment after treatment at 200 °C (Fig. 2D) is similar to that of the Mms6C25 fragment as reported in a previous study (30). Mass analysis of the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) band in Fig. 2B (lane 6) indicates that degradation may occur at Pro-27 (Fig. 2E and SI Appendix, Fig. S2), an amino acid that is not resistant to high temperatures (31, 32). In addition, we conducted Fourier transform infrared spectroscopy (FTIR) analysis of magnetosome-like MNPs, control MNPs (in the absence of Mms6), natural magnetosome MNPs, and Mms6 proteins (SI Appendix, Fig. S3). The appearance of protein-related FTIR signals confirms that the Mms6 protein is attached on the surface of the magnetosome-like MNPs. These data indicate that Mms6 protein is involved in the assembly of the reverse micelle system and possibly the mineralization process.

Crystallographic Analysis of the Magnetosome-Like MNPs.

High-resolution transmission electron microscopy (HRTEM) was performed to analyze the morphology and crystallography of the MNPs. As shown in Fig. 3, nanoparticles of uniform size can be obtained in reverse micelle nanoreactors with or without Mms6 protein. HRTEM crystallographic analysis was performed on crystals synthesized in the presence (Fig. 3, b1–b3, and SI Appendix, Fig. S4) and absence of Mms6 (Fig. 3, c1–c3), as well as on bacterial magnetite from strain AMB-1 for comparison (Fig. 3, a1–a3). The images clearly show lattice fringe patterns, indicating that single-crystal structures are formed in all three cases. X-ray diffraction patterns of the synthesized MNPs are shown in SI Appendix, Fig. S5, which can be readily indexed to magnetite Fe₃O₄ (Joint Committee on Powder Diffraction Standards no. 74-0748).

Fig. 3.

HRTEM characterization of natural magnetosomes, magnetosome-like MNPs, and magnetic nanocrystals from the control reaction. (A1) and (A2) High-resolution electron micrographs and (A3) three-dimensional morphology of natural magnetosome crystals from AMB-1. (B1) and (B2) High-resolution electron micrographs and (B3) three-dimensional morphology of magnetosome-like MNPs. (C1) and (C2) High-resolution electron micrographs and (C3) three-dimensional morphology of MNPs from the control reaction. The insets in (A2), (B2), and (C2) show the corresponding fast Fourier transform patterns of the crystal structure.

The magnetite particles synthesized in the presence of Mms6 have a cubo-octahedral shape, identical to that of the magnetite crystals isolated from the magnetospirillum AMB-1. Fig. 3, b2, shows the hexagonal plane view of magnetosome-like MNPs from zone axis [110] with interior angles of the (110) and (111) planes. The interlayer distances of 4.83 and 2.97 Å agree well with the separation between the (111) and (220) lattice planes of Fe₃O₄ with a face-centered cubic structure. Additional projection images from the zone axis of [100], [111], [−101] were observed and collected (SI Appendix, Fig. S4). The shapes of the top view and the positions of each crystal plane are completely consistent with those of the natural magnetosome (Fig. 3 A and B). Consequently, the decomposition synthesis in the Mms6-containing reverse micelle system successfully reproduces the formation of the cubo-octahedron nanocrystals of the magnetosome, except that the average size (9.1 ± 1.2 nm; SI Appendix, Fig. S6) is smaller than that of natural magnetosome MNPs (39 nm; SI Appendix, Fig. S7).

The magnetite particles synthesized in the absence of Mms6 protein reveal octahedral or tetrahedral morphologies and consist mainly of (111) crystal faces (Fig. 3, c2 and SI Appendix, Fig. S8). This is consistent with previous work (33). The average size of nanoparticles synthesized in the absence of Mms6 is 13.7 ± 3.9 nm (SI Appendix, Fig. S8). The homogeneity in particle sizes again demonstrates the superiority of the reverse micelle nanochamber in size control compared to other synthesis strategies, where the surfactant-only template favors the formation of the (111) surface. Furthermore, the formation of the magnetite (100) face in cuboidal-shaped particles from both magnetosome and magnetosome-like MNPs is likely caused by the face-specific interaction of Mms6 protein. The introduction of Mms6 protein into reverse micelles, therefore, endows an additional crystal growth regulation in the nanoreactor. By associating with a (100) surface, Mms6 selectively blocks and regulates crystal growth, directing the crystal shape to be cubo-octahedron.

Magnetic and MR Relaxation Properties of the Magnetosome-Like MNPs.

The magnetic properties of the magnetosome-like nanoparticles were evaluated and compared to native magnetosome nanoparticles, with the synthesized control samples as shown in Fig. 4A. The magnetic hysteresis (M-H) curves were assessed by measuring the field-dependent magnetization of the magnetosome-like MNPs and control MNPs at 300 K, the low field regions being highlighted as insets. The magnetosome-like MNPs exhibit a soft ferromagnetic behavior with a high M_s (95 emu [electromagnetic unit]/g Fe) and low magnetic coercivity (H_c: 20 Oe) (Fig. 4A). The magnetic properties of the natural magnetosomes exhibit behavior similar to that of the synthesized magnetosome-like MNPs. The M-H results of the control group MNPs, however, show weak magnetic interaction, such as a lower M_s and a higher H_c, compared to the magnetosome-like MNPs. Meanwhile, at low temperatures (5 K), the magnetosome-like MNPs exhibit stable soft-ferromagnetic behavior, while the coercivity of the natural magnetosomes increases, indicating their weak soft-magnetic behavior at low temperatures (SI Appendix, Fig. S9). The soft magnetism of the magnetosome-like MNPs enables them to respond to external magnetic fields quickly (SI Appendix, Fig. S10; Movie S1).

Fig. 4.

Magnetic characterization and hydrophilic coating of the magnetosome-like MNPs. (A) Field-dependent magnetization curves (M-H) showing the H_c and M_r (see magnified inset) at 300 K. (B) TEM image of the DSPE-mPEG–coated magnetosome-like MNPs. The inset shows the aqueous solution of the DSPE-mPEG–coated magnetosome-like MNPs and DLS profile.

To improve their water solubility and reduce aggregation, the synthesized magnetosome-like MNPs were coated with a layer of hydrophilic polyethylene glycol (PEG)–modified phospholipid, the hydrodynamic diameter (HD) after 1,2-distearoyl-sn-glycero-3-phosphoethanolamine with conjugated methoxyl PEG (DSPE-mPEG) coating being 35 nm as shown by the DLS measurements. The TEM image also shows good monodispersion (Fig. 4B). Meanwhile, the DSPE-mPEG–modified magnetosome-like MNPs still exhibit soft ferromagnetism with high saturation magnetization (SI Appendix, Fig. S11). The unique features of the DSPE-mPEG–coated magnetosome-like MNPs, such as their small and uniform size, monodispersion, good biocompatibility, and special magnetic properties, also enable them to become potential MRI contrast agents. The mean r₁ and r₂ values of the magnetosome-like MNPs at 3 T are 3.69 and 51.75 mM⁻¹ s⁻¹, respectively, while the corresponding values at 14.1 T are 1.21 and 79.13 mM⁻¹ s⁻¹, respectively (SI Appendix, Fig. S12).

The cellular-level biocompatibility was assessed next. As shown in SI Appendix, Fig. S13, the cell viability was greater than 90% even at a concentration of up to 100 μg Fe/mL. The nontoxicity of magnetosome-like MNPs could be attributed to the negative charge caused by the DSPE-mPEG coating (SI Appendix, Fig. S14), generating charge repulsion with the cell membrane and preventing the internalization of the nanoparticles. In addition, the HD of the DSPE-mPEG–coated magnetosome-like MNPs is virtually unchanged after incubation in phosphate-buffered saline solution and serum at 37 °C for different periods, showing good stability (SI Appendix, Fig. S15). The above results demonstrate the good biocompatibility of magnetosome-like MNPs for further in vivo magnetic targeting studies.

In Vivo Magnetic Tumor Targeting Efficiency Evaluation.

The magnetosome-like MNPs (0.15 mmol Fe/kg mice body weight) were intravenously injected into 6-wk-old male breast tumor model mice (average weight 20 g, subcutaneous xenografted tumor in the right leg) via the tail vein. The tumor area of the mice was covered by a magnet (0.5 T) (Fig. 5B) for magnetically targeted MNP enrichment for 30, 60, and 120 min. For comparison, two control groups were introduced. In the first group, the same volume of magnetosome-like MNPs were injected intravenously, but not enriched by magnetic targeting; in the second group, the synthetic control MNP samples were injected via the caudal vein and magnetically treated under the same conditions.

Fig. 5.

In vivo 14.1 T tumor mice MRI with or without external magnetic targeting treatment (0.5 T). (A) Rapid response of the DSPE-mPEG–coated magnetosome-like MNPs to an external magnetic field (0.1 T). (B) 14.1 T MRI mouse holding device and magnet treatment setup. Magnet: 0.5 T; red dotted circle: tumor area. (C) Image of the 14.1 T Bruker NMR spectrometer for MRI. (D) Dynamic T₁-weighted (T_1w) MRI (at 14.1 T) of tumor-bearing mice after intravenous injection of the DSPE-mPEG–coated magnetosome-like MNPs or control MNPs. Time point above each image represents the cumulative treatment time of the magnet (0.5 T). White dotted circle highlights the tumor area of mice. (E and F) Relative CNR change (ΔCNR) of (E) tumor rim and (F) tumor interior at different time points. ΔCNR = (CNR_post − CNR_pre)/CNR_pre, where post/pre refers to postmagnetic/premagnetic treatment.

Since tumor tissues are abundant in small blood vessels, T₁-weighted MR angiography can accurately reflect the magnetic tumor-targeting efficiency of nanoparticles administrated by vascular injection (34, 35). Fig. 5D and SI Appendix, Fig. S16 show the MR images of the tumor areas in the mouse models acquired on a 14.1-T MRI scanner (Bruker Biospec, Fig. 5C), for which the strong magnetic field effect at 14.1 T provides additional enhancement in both sensitivity and resolution.

Compared to the two control groups, the magnetosome-like MNPs exhibit a considerably enhanced contrast in the tumor area after exposure to a magnetic field for 30 min. The T₁ signal is strongest 60 min after the magnetic field treatment, with the contrast-to-noise ratio (CNR) change in the tumor rim increasing by 132% (nearly 6 times that of the two control groups) and the CNR change in the tumor interior increasing by 110% (nearly 8 to 9 times that of the two control groups) (Fig. 5 E and F). The results clearly demonstrate that magnetic field treatment can effectively accumulate magnetosome-like MNPs injected via the mouse tail vein into the tumor tissues, and considerably improve the contrast of the MR signal in the tumor area.

The tumor-targeting efficiency was also evaluated quantitatively using the Fe concentration distribution in the tissue and organs. As shown in Fig. 6A, the Fe concentration in the magnetosome-like group reaches 110 μg Fe/g tumor tissues (more than twice that of the two control groups) after 30 min of magnetic field treatment, further increasing to 177 μg Fe/g tumor tissues after 60 min (3 times that of the two control groups). After 60 min, 3.1% of the magnetosome-like MNPs gather in the tumor area, which is 7.6 times that of the nonmagnet control group and 10.8 times that of the normal MNP control group, respectively (Fig. 6B). Correspondingly, the accumulation of magnetosome-like MNPs in the liver is 69.8%, which is considerably lower than that in the two control groups (i.e., 74.8% and 77.8%, respectively) (Fig. 6C). Similarly, the Fe accumulation in the spleen (9.9%) is also the lowest among the three groups, it being 13.3% and 11.9% in the control groups, respectively (Fig. 6D). In addition, we calculated the tumor:liver(/spleen) Fe quantity ratio. As shown in Fig. 6 E and F, the ratio of Fe quantities of the magnetosome-like MNPs in the tumor relative to those in both liver and spleen tissue are an order of magnitude larger than those in the control groups. Compared with the liver and spleen, which account for 90% of FeO nanoparticles, FeO nanoparticles are much less distributed in the heart, lung, and kidney. As shown in SI Appendix, Fig. S17, the distribution in heart, lung, and kidney is hardly affected by magnetic targeting treatment. We also did a half-life analysis of magnetosome-like MNPs. The content of magnetosome-like MNPs in blood decreased significantly within 2 h and nearly cleared after 6 h (SI Appendix, Fig. S18). The half-life of magnetosome-like MNPs is 25 min, which is extended to 32 min after magnet treatment. It is clear that the soft ferromagnetism endows magnetosome-like MNPs with a particularly good local enrichment effect under an external magnetic field.

Fig. 6.

The magnetic targeting efficiency of the MNPs in mice. (A) Fe distribution in the tumor tissues by magnet treatment at different times after intravenous MNP injection. (B–D) The enrichment efficiency of the MNPs in (B) tumor, (C) liver, and (D) spleen tissues after 60 min of magnetic treatment. % ID means the percentage of the injected dose. (E and F) The ratio of Fe enriched in (E) the tumor:liver tissues and (F) the tumor:spleen tissues. (G) The Fe distribution analyzed by Prussian blue staining in the tumor tissues after 60 min of magnetic treatment. # P < 0.05, *P < 0.01 compared to the magnetosome-like MNPs treated with a 0.5 T magnetic field. Scale bar: 100 μm.

The tumor tissues were collected and further analyzed by Prussian blue staining. As shown in Fig. 6G, the magnetosome-like MNPs could pass through the loose outer region of the tumor tissues and enter the dense inner region after magnetic field targeting treatment. There is no obvious accumulation of MNPs in the tumor region of the two control groups. Moreover, major organs, including the heart, liver, spleen, lung, and kidney, do not show substantial histological variation (SI Appendix, Fig. S19), indicating that there is no tissue injury in vivo after the injection of magnetosome-like MNPs.

Discussion

Magnetic targeting uses a magnetic field to regulate the distribution of magnetically responsive nanoparticles or drug carriers, so as to reduce the off-target effects of systemic administration (4, 5). As important magnetic targeting carriers, FeO nanoparticles have good biocompatibility and provide a feasible drug delivery approach for cancer treatment (4, 7). In addition, their intrinsic photoelectric and magnetic properties and high surface area enable MNPs to integrate multiple diagnostic functions, such as MRI and computed tomography (35, 36). It has been reported that their small size is beneficial to the spread of the nanomaterials into the tumor tissues (37, 38). Nevertheless, most small-sized MNPs (less than 10 nm) are superparamagnetic, making it difficult to achieve effective magnetic targeting due to their weak magnetic response. Given that their special magnetic properties result from good crystallinity and well-defined crystal surfaces, natural magnetosomes are considered to be good candidates for magnetic targeting nanomedicine applications (39). However, they can agglomerate easily in aqueous solutions and have poor monodispersity, limiting their penetration of the lesion tissue, especially in the dense tissue around the tumor, thus affecting the magnetic targeting and MRI performance. Consequently, it is of great significance for magnetic targeting to develop an appropriate synthesis method to construct small-sized MNPs with the magnetic response capabilities of magnetosomes.

Given its essential activity in formatting magnetite crystals in MTB, Mms6 has been expressed from Escherichia coli bacteria, purified, and introduced into in vitro magnetite synthesis (40). In particular, it has been reported that magnetite nanoparticle arrays can been fabricated on Mms6-modified templates (13). Although it has been reported that MNPs with cubo-octahedron morphology can be obtained, the size of the synthesized MNPs is not uniform and tends to form multidomain rather than single-domain particles (41, 42). As distinct from the mineralization process occurring on extracellular matrixes such as bones, the natural mineralization process of magnetosomes does not occur in an open reaction space, but in a confined reaction space of magnetosome vesicles. This explains why the biomimetic synthesis using Mms6 protein in an open space tends to produce nanoparticles with wide size distributions and polycrystalline morphology (17, 21, 43). Conversely, using a protein cage as the biomineralization template—a typical example being ferritin—it is possible to synthesize ultra-small (2 to 6 nm) MNPs of uniform size (44, 45). However, compared with natural magnetosomes, the MNPs obtained normally have poor crystallinity (with lattice defects) without well-defined crystal surfaces, resulting in low saturation magnetization. In this study, by introducing the Mms6 protein into the reverse micelle system, we could realize the two basic conditions of biomimetic mineralization of magnetosomes, namely Mms6 and a nanoreaction chamber. Moreover, combined with the feasible controllability of the decomposition reaction, our method provides another way of producing MNPs of high quality and yield.

The reaction processes of the biomimetic synthesis can be divided into two stages. The first stage is the assembly of nanoreactors. At 60 °C, the self-assembly of Mms6 protein and surfactant forms a reverse micelle nanoreactor similar to magnetosome vesicles. The second stage is the production of MNP nanoparticles at 200 °C. The SDS-PAGE analysis and respective NMR spectrum of Mms6 extracts from a sample of the first stage show that Mms6 protein maintains its protein integrity under reaction conditions at 60 °C. During the second stage, the high temperature of 200 °C does not completely degrade the Mms6, and the hydrophilic C-terminal of Mms6 is well retained to regulate the morphology of the MNPs.

In terms of amino acid composition, Mms6 is composed mainly of high-temperature–resistant amino acids, enabling Mms6 to regulate mineralization under harsh conditions. Moreover, the hydrophobic environment provided by the reverse micelles further enhances the ability of the protein to withstand the denature conditions (46). Our preliminary results suggest that the protein chain breaks at the position of proline under high temperatures, and further study is needed to determine whether other amino acids undergo chemical changes and modifications under similar high-temperature conditions. However, it is clear that proteins such as Mms6 can be used as a biological template or catalyst for high-temperature chemical reactions, given that the heat resistance of proteins can be improved using sequence optimization and protein modification.

HRTEM analysis of the crystal form and surface shows that the magnetosome-like MNPs share the same cubo-octahedral morphology as natural magnetosomes (comprising six (100) crystal faces and eight (111) crystal faces). From an energy point of view, the (111) surface of the Fe₃O₄ crystal is more stable. The appearance of the energy-unfavored (100) crystal surface in the presence of Mms6 indicates that the activation energy of the (100) crystal surface can be efficiently reduced by the specific interaction between it and the spatial arrangement of the Mms6 surface formed by self-assembly (43, 47).

To verify the role of Mms6 protein in the biomimetic synthesis system, we mutated DEEVE—the key regulatory site of Mms6 protein (22, 30)—into AAAVA to obtain a mutant protein (i.e., Mms6^MT; SI Appendix, Fig. S20). Both Mms6 and Mms6^MT are N-terminal hydrophobic and C-terminal hydrophilic. This amphiphilic property is essential for them to participate in the reverse micelle system. TEM characterization shows that the MNPs that are synthesized in the presence of Mms6^MT share the same morphology as the reverse micelle–only system (SI Appendix, Fig. S21), while the Mms6 reaction system produces MNPs with unique cubo-octahedron morphology. Consequently, the difference in crystal morphology of the Mms6 reaction system from those of Mms6^MT and the reverse micelle–only system, is due to the specific regulation abilities of the C-terminal sequence of Mms6.

The synthesized magnetosome-like MNPs have similar magnetic properties to natural magnetosomes, both of which have high M_s, and low H_c and magnetic remanence (M_r). At present, the M_ss of small MNPs—that is, less than 10 nm—obtained by chemical synthesis are usually in the range of 65 to 80 emu/g Fe (48). However, the magnetosome-like MNPs synthesized in this work have high saturation magnetization (95 emu/g Fe), which is rarely seen in small-sized MNPs. This can be attributed to their higher crystallinity and crystal morphology. In addition, the magnetosome-like MNPs also have a small coercive force—that is, 20 and 50 Oe at 300 and 5 K, respectively.

According to Zheng et al., the magnetic coercive force of superparamagnetic materials is directly proportional to the magnetic anisotropy of nanomaterials (49). Moreover, the [100] crystal direction is the easy direction of magnetization, and hence the permeability is larger. The octahedron and tetrahedron structures of the control group contain only (111) crystal faces. By contrast, there are six (100) crystal faces in the cubo-octahedron structure of the natural magnetosomes and magnetosome-like MNPs. The conservancy effect of Mms6 protein on the (100) crystal surface leads to smaller magnetic anisotropy of the cubo-octahedron structure, and eventually a smaller magnetic coercive force. Consequently, the same crystal conformation and structure as natural magnetosomes endow magnetosome-like MNPs with soft ferromagnetic properties—that is, a higher M_s and lower H_c.

The tumor-targeting data shown in Fig. 6 indicates that magnetosome-like nanoparticles can penetrate the internal tumor area owing to their soft ferromagnetic properties and small sizes. The tumor-targeting efficiency was significantly improved by approximately 3% compared with the control. In contrast, magnetic targeting reduces the distribution of magnetosome-like nanoparticles in the liver and spleen by approximately 10% (SI Appendix, Fig. S22). The remaining 7% changes of FeO nanoparticles biodistribution caused by magnetic treatment we believe is due to the accumulation of MNPs in the blood vessels around the tumor, which is supported by the MRI results in Fig. 5D. The soft ferromagnetic properties of magnetosomes enable them to be effectively attracted by the external magnets near the tumor site, thereby reducing the biological distribution in the liver and spleen. The above results showed that magnetosome-like MNP significantly improved the efficiency of magnetic targeted drug delivery, and the excellent magnetic targeting performance makes it an outstanding candidate for various biomedical applications.

The biomedical application of magnetosome-like MNP still faces many challenges. In particular, once the external magnetic field is removed, the possibility of nanoparticles escaping from the target organ increases, leading to off-target effects and reducing the delivery efficiency (4, 50). Chemically modifying the surface of magnetosome-like MNPs with specific small molecules and proteins to recognize specific cells and tissues and improve the retention time will help to further improve the accuracy and efficiency of magnetic targeting delivery.

Conclusions

In this study, we constructed a magnetosome-like nanoreactor, recapitulating the two principal factors required for MTB biomineralization—magnetosome vesicles and magnetosome regulatory proteins—by introducing the amphiphilic Mms6 protein into self-assembled reverse micelles. Magnetosome-like MNPs exhibit the same crystalline form as natural magnetosomes and high-performance magnetic properties that respond rapidly to an external magnetic field. Moreover, magnetosome-like nanoparticles coated with DSPE-mPEG, to negate the disadvantages of natural magnetosome MNPs, possess excellent monodispersity, good water solubility, and small hydrodynamic dimensions. It could be considered that if highly water-soluble magnetosome-like MNPs were to be modified with suitable target molecules and drugs, they could provide many more application prospects for magnetic hyperthermia, MR imaging, and magnetically guided drug release. We believe that this work provides an unprecedented opportunity for high-quality magnetic nanomaterials, affording us insights into bio-inspired mineralization and further understanding of the thermal decomposition mechanism.

Data, Materials, and Software Availability

All of the study data are included in the article and/or supporting information.