Off-axis electron holography of magnetotactic bacteria: magnetic microstructure of strains MV-1 and MS-1

Off-axis electron holography in the transmission electron microscope is used to characterize the magnetic microstructure of magnetotactic bacteria. The practical details of the technique are illustrated through the examination of single cells of strains MV-1 and MS-1, which contain crystals of magnetite (Fe3O4) that are ~50 nm in size and are arranged in chains. Electron holography allows the magnetic domain structures within the nanocrystals to be visualized directly at close to the nanometer scale. The crystals are shown to be single magnetic domains. The magnetization directions of small crystals that would be superparamagnetic if they were isolated are found to be constrained by magnetic interactions with adjacent, larger crystals in the chains. Magnetization reversal processes are followed in situ, allowing a coercive field of between 30 and 45 mT to be measured for the MV-1 cell. To within experimental error, the remanent magnetizations of the chains are found to be equal to the saturation magnetization of magnetite (0.60T). A new approach is used to determine that the magnetic moments of the chains are 7 and 5×10-16Am2 for the 1600-nm long MV-1 and 1200-nm long MS-1 chains examined, respectively. The degree to which the observed magnetic domain structure is reproducible between successive measurements is also addressed. Key-words: magnetotactic bacteria, off-axis electron holography, magnetite nanocrystals, biologically controlled mineralization. 1.Introduction most important of these parameters is the phase shift, as it can be used to obtain quantitative inforThe transmission electron microscopy (TEM) techmation about the magnetic field in the sample to a nique of off-axis electron holography allows the spatial resolution that can approach the nanometer amplitude and the phase shift of an electron wave scale. In this paper, we show how the magnetic mithat has passed through a sample (rather than its incrostructure of intracellular, membrane-bounded tensity) to be recorded (Tonomura, 1992a; Lichte, ferrimagnetic crystals, which are synthesized by 1991). When studying magnetic materials, the magnetotactic bacteria and known as magnetoso­ mes, can be characterized using off-axis electron holography. The magnetic crystals in magnetotac­ tic bacteria are composed of magnetite (Fe3O4) and/ or greigite (Fe3S4) and are usually arranged in one or more linear chains within each cell (Blake­ more, 1975; Bazylinski & Moskowitz, 1997). The magnetic moment that they impart to the cell re­ sults in the alignment and subsequent migration of the cell along the Earth’s magnetic field lines (Frankel, 1984). This attribute is thought to in­ crease the efficiency with which the cell finds an optimal position in a vertical chemical or redox gradient in an aquatic environment (Frankel et al., 1997). Magnetotactic bacteria exercise a high degree of control over the morphologies of their constituent magnetosomes, which have a narrow size distribu­ tion, are specific to each cell type and usually have their magnetic easy axes aligned parallel to the chain axis (Bazylinski et al., 1994). The fact that the magnetic crystals are typically only 40-100 nm in size suggests that they should each contain a sin­ gle magnetic domain (Moskowitz, 1995). Howev­ er, magnetic force microscopy has insufficient res­ olution to characterize the magnetic microstructure of such crystals when they are still encapsulated within intact bacterial cells (Proksch et al., 1995). Differential phase contrast microscopy (Chapman et al., 1978; Daykin & Petford-Long, 1995) is also poorly suited to the characterization of magnetic microstructure in sub-100-nm-sized nanocrystals because of large ‘mean inner potential’ contribu­ tions to the contrast from their edges (to be dis­ cussed further below). Here, we show that electron holography can be used to provide high spatial res­ olution, quantitative information about the magnet­ ic microstructure of bacterial magnetosomes. The ability to obtain this information is crucial to under­ standing magnetic-field-sensing mechanisms in a wide range of organisms (Mann et al., 1988; Diebel et al., 2000), as well as providing magnetic bio­ markers that may be used to establish the occur­ rence of ancient life (McKay et al., 1996; Thom­ as-Keprta et al., 2000). Magnetosomes also pro­ vide a model system for studying interacting sin­ gle-domain magnetic crystals, which are of in­ creasing interest to the electronics, catalysis and magnetic recording industries (Mann, 1993). Pre­ vious examples of the application of electron ho­ lography to magnetic materials have included the characterization of recording tapes (Tonomura, 1992b), Co particles (de Graef et al., 1999) and lith­ ographically patterned nanostructures (Dunin-Bor­ kowski et al., 1998a and 2000; Smith et al., 2000). We begin by describing the basic experimental procedure required to obtain magnetic information from off-axis electron holograms of magnetic na­ nostructures. We then illustrate the specific precau­ tions that are required when applying the technique to magnetotactic bacteria. The characterization of single cells of two strains of bacteria, designated MV-1 and MS-1, is described in detail. These strains contain magnetite crystals that have radical­ ly different morphologies. Qualitative results that reveal the magnetic microstructures of the magne­ tosome chains in the two cells are initially present­ ed. Quantitative measurements of several of the cells’ magnetic properties, including their magnet­ ic moments, are then described. Finally, the results from the bacterial cells are compared with similar data obtained from lithographically patterned mag­ netic nanostructures. Some of the results that are described in this pa­ per have been presented in preliminary form else­ where (Dunin-Borkowski et al., 1998b). However, the present paper provides a more detailed descrip­ tion of the experimental procedure required to ap­ ply electron holography specifically to magnetotactic bacteria. A novel approach for measuring the magnetic moment of a bacterial cell is also intro­ duced. 2. Experimental details Cultured cells of Magnetospirillum magnetotacti­ cum strain MS-1 and the marine vibrioid strain MV-1 were prepared using methods that have been described by Frankel et al. (1997) and Bazylinski et al. (1988). Drops of water that had been enriched in bacteria were deposited directly onto 3-mm-diame­ ter carbon or Formvar-coated Ni grids for TEM ex­ amination. Off-axis electron holography was per­ formed at 200 kV in a Philips CM200 field-emis­ sion-gun (FEG) TEM. This microscope has a rotat­ able electrostatic biprism (a 0.6 μm quartz wire coated with gold) located in place of one of the con­ ventional selected area apertures. It also has a Lo­ rentz minilens, which is located in the bore of the objective lens pole-piece. The Lorentz minilens al­ lows the examination of magnetic materials in close-to-field-free conditions (with the conven­ tional microscope objective lens switched off) and has a line resolution of 1.2nm at 200kV. All of the holograms presented below were re­ corded directly onto a 1024×1024 pixel charge-cou­ pled-device (CCD) camera at a nominal microscope magnification of 30k×, corresponding to a field of view on the CCD camera of 630 nm. A positive voltage of 120Vwas applied to the biprism wire for holography. This biprism voltage was chosen be­ cause it provided an optimal holographic overlap width and fringe spacing (see below) of 640 and 3.9 nm, respectively. Data analysis was performed on a Silicon Graphics workstation using library programs written in the Semper image processing language (Saxton et al., 1979). A Philips EM430 TEM equipped with a Gatan post-column imaging spectrometer was used to confirm the chemical compositions of the magnetite crystals under inves­ tigation. The chemical characterization of the sam­ ples is not described further in this paper. 3. Background to off-axis electron holography The microscope geometry for off-axis electron ho­ lography is shown schematically in Fig. 1a. The sample is examined using defocused illumination from a FEG electron source, with the region of in­ terest positioned so that it covers approximately half the field of view. A positive voltage of between 50 and 200V is applied to an electrostatic biprism, which usually takes the form of either a Ptor a Au-coated quartz wire located in the selected area aperture plane of the microscope. This voltage causes the electron wave that has passed through the sample to overlap with a reference wave that has passed either through vacuum or through a thin region of the support film. The region of interest and the reference wave must be within ~ 1 μm of each other in order for the two waves to overlap co­ herently, and the orientation of the biprism wire may need to be optimized to satisfy this criterion. A low-magnification, bright-field image of the typi­ cal setup is shown in Fig. 1b. In this case, the bi­ prism has been rotated to achieve overlap between a hole in the support film and a magnetosome chain from an MV-1 cell. The bright band of intensity running diagonally across the image is the overlap region of the two waves. The coarse fringes at the edges of the overlap region are Fresnel fringes from the edge of the biprism wire. At higher magnifica­ tion, holographic interference fringes are visible within the overlap region, as shown for a hologram of a thin magnetic film in Fig. 1c. The spacing and the contrast of the fringes decrease with increasing biprism voltage. The amplitude and the phase shift of the electron wave that leaves the sample are recorded in the in­ tensity and the position of the holographic interfer­ ence fringes, respectively. Experimentally, these parameters are obtained by extracting one ‘side­ band’ from the Fourier transform of the hologram. This sideband is then inverse-Fourier-transformed, and the amplitude and phase of the resulting com­ plex image wave are calculated, as shown in the lower half of Fig.1c. Before further analysis the re­ constructed phase image is ‘unwrapped’ to remove phase discontinuities, which result from the fact that it is initially calculated modulo 2 . A reference hologram is always obtainedwith the sample out of the field of view in order to remove artifacts associ­ ated with the imaging and recording process (de Ruijter & Weiss, 1993). For magnetic materials, the use of a Lorentz lens allows samples to be exam­ ined in almost field-free conditions with the con­ ventional microscope objective lens, which would result in a large vertical magnetic field at the posi­ tion of the sample, switched off. When imaging us­ ing the Lorentz lens, the objective lens may also be excited slightly and the sample tilted in order to provide a component of the external field in the plane of the sample. In this way, magnetization pro­ cesses can be followed in situ (Smith & McCartney, 1999). The orientation of the sample may then need to be chosen to lie parallel to one of the tilt axes, and the applied magnetic field must be calibrated accu­ rately (Fig. 1d). Further details about the practical requirements for electron holography, including in­ formation about recording and processing holo­ grams, can be found elsewhere (Smith & McCart­ ney, 1999; Dunin-Borkowski et al., 1998c). The recorded phase shift is sensitive to both the mean inner potential (MIP) of the sample and the in-plane component of the magnetic induction B integrated in the incident beam direction. Neglect­ ing dynamical diffraction (i.e., assuming that no crystals are strongly diffracting), the phase shift can be expressed in the form