The magnetosome ultrastructure: Pure magnetite

It was recognized early on that magnetotactic bacteria form magnetite as a mineral inclusion in their magnetosome organelles.^{Reference Mann, Frankel and Blakemore16,Reference Frankel, Papaefthymiou, Blakemore and O’Brien17,Reference Matsuda, Endo, Osakabe, Tonomura and Arii66} However, the presence of maghemite (α−Fe₂O₃), an oxidized spinel phase analog to magnetite where vacancies replace Fe(II), has been the subject of debate.^{Reference Frankel, Papaefthymiou, Blakemore and O’Brien17,Reference Faivre, Böttger, Matzanke and Schüler18,Reference Kopp and Kirschvink67} The magnetic properties of iron oxide nanoparticles typically depend on the mineral phase, in addition to numerous other factors. In particular, maghemite has different magnetic properties when compared to magnetite.^{Reference Dunlop and Özdemir68}

The reduced precision of electron diffraction for lattice parameter determination has long prevented quantitative characterization of the magnetosomes mineral phase and its differences with respect to synthetic particles of similar sizes. However, with the development of synchrotron sources with high brilliance, high-resolution x-ray diffraction (XRD) has developed into a powerful technique for such experiments.

Using high-resolution XRD, significant differences in the lattice parameter were found between intracellular magnetosomes from cultured magnetotactic bacteria when measured in whole cells, which exhibited the perfect expected value of magnetite (8.3969 Å),^{Reference Fleet69,Reference Fleet70} and those measured after isolation from the bacteria, which exhibited slightly lower values.^{Reference Fischer, Schmitz, Aichmayer, Fratzl and Faivre71} Synthetic materials of similar size used as control and for comparison showed that this difference is related to different oxidation states, and the biogenic nanomagnetite found in the whole cells is stoichiometric, or structurally pure. Isolated magnetosomes and synthetic particles are slightly oxidized (an oxidation parameter of 21% was calculated when using the formalism of Dunlop and Özdemir,^{Reference Dunlop and Özdemir68} meaning that such particles are a mixture of about .79/.21 magnetite/maghemite). This stoichiometric magnetite is ferrimagnetic with a magnetic moment (4 µ_B) larger than that for maghemite (3 µ_B).^{Reference Dunlop and Özdemir68} By fitting the relationship between the magnetic moment and the oxidation parameter of Dunlop and Özdemir (68), Fischer et al.^{Reference Fischer, Schmitz, Aichmayer, Fratzl and Faivre71} showed that a loss of about 4% of the saturation moment was expected for the oxidized synthetic particles as compared to the genuine magnetosomes; therefore, a reduced efficacy can be expected when they are used for actuation. Since iron is very insoluble in a natural aqueous environment, it might be important for the bacteria to carefully control the fate of the iron in the biomineralization process. Besides the role of the bacterial membrane, it is currently not clear how such control is achieved at the molecular level.

The magnetosome dimension: Stable single domain

Magnetotactic bacteria of a given strain always form magnetosomes of the same dimensions of the mature crystals of a given strain, but a variety particle sizes are observed for different strains.^{Reference Devouard, Posfai, Hua, Bazylinski, Frankel and Buseck26,Reference Thomas-Keprta, Bazylinski, Kirschvink, Clemett, McKay, Wentworth, Vali, Gibson and Romanek72,Reference Faivre and Ukmar Godec73} The magnetic properties of magnetite nanoparticles typically depend, besides numerous other factors, on their size. Particles smaller than ∼25–30 nm are superparamagnetic, that is, they only have a measurable magnetic signal if placed in an external magnetic field (Figure 3). Larger nanoparticles have intrinsic and permanent magnetic moments. Particles with isometric shapes from 25–30 nm to ∼80–100 nm fall within the stable single-domain category. Finally, particles larger than this threshold of 80–100 nm are called multi-domain, in which several magnetic domains coexist, reducing the magnetostatic energy.^{Reference Dunlop and Özdemir68} The magnetosome size usually falls within the stable single-domain region, where the volume magnetization is the highest.^{Reference Muxworthy and Williams74,Reference Muxworthy and Williams75}

Figure 3.

Magnetic property (e.g., coercivity) as a function of the particle dimension (here, diameter). Magnetosomes are found in the stable single-domain region.

Although the chemical mechanism leading to magnetite has now been unraveled with some detail for the magnetospirilla and includes an intermediate ferrihydrite phase that is poorly crystalline,^{Reference Fdez-Gubieda, Muela, Alonso, García-Prieto, Olivi, Fernández-Pacheco and Barandiarán19,Reference Baumgartner, Morin, Menguy, Perez Gonzalez, Widdrat, Cosmidis and Faivre20} it is still not fully clear what factor controls the particle size. By deleting specific genes and then analyzing the bacterial phenotype, the role of several proteins in the formation of magnetosomes has been exposed. This is a somewhat indirect methodology, since other genes present in the genome of the bacteria might compensate for the loss of function upon deletion of a particular gene, and the level of expression of other related genes and proteins might also be altered. Such genetic compensation has already been observed in the case of the magnetotactic bacteria.^{Reference Rioux, Philippe, Pereira, Pignol, Wu and Ginet76,Reference Draper, Byrne, Li, Keyhani, Barrozo, Jensen and Komeili77} Still, this is currently the best available technique and has permitted the discovery of several genes and proteins of interest, even if the mechanism by which those biomolecules act remains unclear.

For example, the mamGFDC (the genes were called from the letter of the alphabet) cluster that encodes for membrane proteins was one of the first to be discovered.^{Reference Scheffel, Gärdes, Grünberg, Wanner and Schüler45} When mamGFDC are absent, the magnetosomes formed by the bacteria are smaller, in turn, when the proteins are overexpressed in the cells, the magnetosomes are larger than typical wild-type magnetosomes.^{Reference Scheffel, Gärdes, Grünberg, Wanner and Schüler45} The size of the empty magnetosome vesicle is also changed accordingly. But as the study points out, even for mature magnetosomes, apparently the magnetosome organelle is never fully filled by the magnetosome crystals, ruling out the possibility of simple size control by the organelle size.^{Reference Scheffel, Gärdes, Grünberg, Wanner and Schüler45}

Another extensively studied protein is mms6,^{Reference Arakaki, Webbs and Matsunaga78} an acidic protein that is tightly attached to the magnetosome particles and is thought to help in the control of the magnetosome size elucidated in vitro and in vivo.^{Reference Tanaka, Mazuyama, Arakaki and Matsunaga44} However, in the in vivo studies, since the deletion was done in a way in which genes in proximity to mms6 were also altered, it is not fully clear if the change in magnetosome size is due to the mms6 alone. A more recent study without this limitation points toward the role of an alternative gene in size control, mmsF, which is present and expressed in the neighboring region of mms6.^{Reference Murat, Falahati, Bertinetti, Csencsits, Körnig, Downing, Faivre and Komeili40} Advancements in structural biology may help unravel this mechanism parallel to that of magnetosome formation more generally.^{Reference Nudelman and Zarivach79–Reference Siponen, Legrand, Widdrat, Jones, Zhang, Chang, Faivre, Arnoux and Pignol81} While the main molecular players are summarized here, a more detailed analysis is available of the molecular players involved in the biomineralization of magnetite in Magnetospirillum gryphiswaldense ^{Reference Lohße, Borg, Raschdorf, Kolinko, Tompa, Posfai, Faivre, Baumgartner and Schüler42,Reference Lohße, Ullrich, Katzmann, Borg, Wanner, Richter, Voigt, Schweder and Schüler43} and Magnetospirillum magneticum. ^{Reference Murat, Falahati, Bertinetti, Csencsits, Körnig, Downing, Faivre and Komeili40,Reference Murat, Quinlan, Vali and Komeili41}

The crystal orientation: Favor the easy axis

Magnetic anisotropy in magnetite nanoparticles is expressed by a squarer hysteresis loop for a magnetic field applied parallel to the direction of the alignment than for a magnetic field applied perpendicular to that direction. In the case of the magnetosome chain, this magnetic anisotropy is the combination of two parameters: the preferred orientation of the magnetosomes along their easy axis of magnetization (〈111〉) with respect to that of the hard axis (〈100〉) (Figure 4), and from dipolar interactions due to the assembly in the chain.^{Reference Alphandéry, Ding, Ngo, Wang, Wu and Pileni82}

Figure 4.

Schematic of magnetic moment as a function of the applied field. In a small region of the field, the magnetic moment is larger when measured along the easy axis than when measured along the hard axis of magnetite.

Early transmission electron microscopy (TEM) studies have shown that the cubooctahedral crystals of Magnetospirillum magnetotacticum are preferentially aligned with the [111] direction parallel to the chain axis of the cell, which is one of the 〈111〉 directions corresponding to the easy axis of magnetization. This result was confirmed later by electron holography^{Reference Dunin-Borkowski, McCartney, Frankel, Bazylinski, Pósfai and Buseck83} and the palette of bacterial strains has expanded with further high-resolution TEM studies.^{Reference Alphandéry, Ding, Ngo, Wang, Wu and Pileni82} In turn, particles from strains producing magnetosomes elongated along the [100] direction such as Desulfovibrio magneticus or strains MYR-1, LO-1, and AV-1 have been suggested to have their easy axis switched to that direction,^{Reference Pósfai, Moskowitz, Arato, Schüler, Flies, Bazylinski and Frankel84–Reference Lefèvre, Pósfai, Abreu, Lins, Frankel and Bazylinski86} possibly as a result of the shape anisotropy and magnetostatic interactions. Even particle elongation along [110] for strain HSMV-1 has been reported,^{Reference Lefèvre, Pósfai, Abreu, Lins, Frankel and Bazylinski86} which represents the intermediate direction of magnetization.^{Reference Dunlop and Özdemir68} All of these results, however, are based on a limited number of samples used in microscopy studies.

To gain further insight into the organization of magnetosome nanomaterials, two-dimensional synchrotron XRD was recently applied to cells of magnetotactic bacteria that were pre-aligned with an externally applied magnetic field to determine the crystallographic orientation of bulk magnetosomes relative to the chain axis.^{Reference Körnig, Winklhofer, Baumgartner, Gonzalez, Fratzl and Faivre87} This analysis revealed a [111] fiber texture along the chain direction for Magnetospirillum gryphiswaldense and Magnetospirillum magneticum, and a [100] fiber texture for Desulfovibrio magneticus. These bulk results confirm those obtained earlier for individual particles. The [100] axis is primarily energetically unfavorable because it represents a magnetic hard axis in magnetite, but can be turned into an effective easy axis by particle elongation along [100] for aspect ratios higher than 1.25, consistent with aspect ratios in RS-1 magnetosomes.^{Reference Pósfai, Moskowitz, Arato, Schüler, Flies, Bazylinski and Frankel84} The pronounced fiber textures can be explained possibly by strain-specific biological control of crystal orientation at the chain level, as already observed for the strain-specific size and organization of the magnetosome and of the magnetosome chain respectively. Even if this is the case, it remains questionable how such an accomplishment can be achieved biologically. Another possibility is that the physical alignment results from intra-chain magnetic interactions. In this case, biological control of the axis of elongation would be sufficient to influence the crystallographic texture of the magnetosome chain.

The magnetosome alignment: Dipoles sum up

The final level of hierarchy observed in the magnetosome chain is the formation of the chain itself. The magnetic anisotropy was shown to arise mainly from the dipolar interactions between the particles. However, free magnetic particles would have the tendency to collapse and form a ring^{Reference Philipse and Maas88,Reference Kiani, Faivre and Klumpp89} (Figure 5). Therefore, biological determinants probably play a critical role in the formation and stabilization of magnetosome chains.^{Reference Körnig, Dong, Bennet, Widdrat, Andert, Müller, Schüler, Klumpp and Faivre27} The molecular players involved in the process have been extensively studied and mainly involve the proteins MamJ and MamK.^{Reference Komeili, Li, Newman and Jensen28,Reference Scheffel, Gruska, Faivre, Linaroudis, Plitzko and Schüler29,Reference Rioux, Philippe, Pereira, Pignol, Wu and Ginet76,Reference Draper, Byrne, Li, Keyhani, Barrozo, Jensen and Komeili77,Reference Abreu, Mannoubi, Ozyamak, Pignol, Ginet and Kimeili90–Reference Katzmann, Scheffel, Gruska, Plitzko and Schüler98} Briefly, MamK is involved in the formation of the magnetosome filament^{Reference Komeili, Li, Newman and Jensen28,Reference Katzmann, Scheffel, Gruska, Plitzko and Schüler98} to which magnetosomes are attached via MamJ.^{Reference Scheffel, Gruska, Faivre, Linaroudis, Plitzko and Schüler29} However, MamJ is only present in magnetospirilla, so that the formation of the magnetosome chain in other strains is still unclear.

Figure 5.

Sketch of the magnetic property of an assembly of individual dipoles. (a) Magnetosomes arranged in chains as in wild-type cells. The cellular dipole is the sum of the individual dipoles of each magnetosome. (b) Shorter chains can be formed as for example, in the mamK deletion mutant. The magnetic property is unchanged when compared to (a). However, the dipole of one of the shorter chains can be switched to the opposite direction (as in (c), red arrow versus blue arrows) and reduces the resulting cellular dipole (black arrow). This was also observed in Magnetovibrio blakemorei.^{Reference Kalirai, Bazylinski and Hitchcock99} (d) Finally, in the absence of a stabilizing biological template, isolated magnetosomes form a closed ring structure.

The formation of a magnetic dipole large enough to orient the cell in the Earth’s magnetic field is a combination of biological and magnetic forces.^{Reference Klumpp and Faivre100,Reference Bennet, Bertinetti, Neely, Schertel, Kornig, Flors, Muller, Schuler, Klumpp and Faivre101} First-order-reversal-curve diagrams showed that nucleation and growth of magnetosome crystals were a continuous process in Magnetospirillum gryphiswaldense,^{Reference Carvallo, Hickey, Faivre and Menguy102} meaning that crystals in the superparamagnetic size range were cohabiting with particles in the stable single-domain size range along the time of the chain formation. The role of the first nucleating magnetite crystal has been proposed to be critical, as it is thought to be employed as the docking station for the further growth of neighboring particles.^{Reference Faivre, Fischer, Garcia-Rubio, Mastrogiacomo and Gehring103} Therefore, if biological forces play a major role in the formation of a stable chain from non-magnetic bacteria,^{Reference Klumpp and Faivre100} when a magnetic bacterium divides, its chain is split into two identical parts, which are inherited by the daughter cells.^{Reference Katzmann, Müller, Lang, Messerer, Winklhofer, Plitzko and Schüler104,Reference Lefèvre, Bennet, Klumpp and Faivre105} In this case, the biological apparatus is already present, and the newly formed magnetosome particles are strongly influenced by their neighbors,^{Reference Körnig, Dong, Bennet, Widdrat, Andert, Müller, Schüler, Klumpp and Faivre27,Reference Körnig, Hartmann, Teichert, Fratzl and Faivre106,Reference Kirschvink107} having no other choice than to align with the right orientation as the rest of the particles forming the magnetic dipole. This will occur as soon as they are close enough.