MICROSCOPY RESEARCH AND TECHNIQUE 46:319–324 (1999) Imaging Faces of Shadowed Magnetite (Fe3O4) Crystals From Magnetotactic Bacteria With Energy-Filtering Transmission Electron Microscopy ULYSSES LINS,1 BECHARA KACHAR,2 AND MARCOS FARINA3* 1Setor de Microscopia Eletrônica e Departamento de Microbiologia Geral, Instituto de Microbiologia Professor Paulo de Góes,Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brasil 2Laboratory of Cellular Biology, National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Bethesda, Maryland 20892 3Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, CCS, Bl. F, 21941–590 Rio de Janeiro, Brasil KEY WORDS magnetotactic bacteria; bacterial magnetite; energy filtering transmission electron microscopy; biomineralization ABSTRACT We used energy-filtering transmission electron microscopy to image magnetite crystals isolated from uncultured magnetotactic bacteria. These magnetite crystals were shadowed in high vacuum with platinum at 45°. The shadowed crystals were observed in a Zeiss (Thornwood, NY) CEM902 transmission electron microscope. Imaging shadowed crystals with inelastically scattered electrons provided information of the decoration pattern of small platinum particles over crystal surfaces, and thus information on surface characteristics of crystals. Results were comparable to those obtained from scanning electron microscopy using a field emitter gun. Electron energy loss spectra of the crystals as well as of the supporting film were recorded to evaluate variations of image contrast with energy losses. Results indicated that the contrast is attenuated with inelastic imaging and that the effect of contrast tuning caused a contrast inversion at a given point between 100 and 150 eV. We believe this approach can be useful for studying multilayered materials by transmission electron microscopy. Microsc. Res. Tech. 46:319–324, 1999. r 1999 Wiley-Liss, Inc. INTRODUCTION Magnetotactic bacteria navigate along magnetic field lines because of the presence of magnetosomes (Lins de Barros et al., 1990). Magnetosomes contain a magnetic crystal enveloped by a membrane (Gorby et al., 1988), and usually form chains. Magnetite (Fe3O4) or iron sulfides form the magnetic crystals of magnetosomes (Stolz, 1993; Pósfai et al., 1998), which are usually 50 to 100 nm in length. The membrane that envelops the crystals may influence this size restriction and morphology. Because of the presence of a magnetosome membrane, it is thought that bacteria control the deposition of crystals within magnetosomes and, as a consequence, singular morphologies, usually different from inorganic magnetite, are formed. It has been considered that magnetosomes are species-specific, which reflects a genetic control on their precipitation (Vali and Kirschvink, 1990). Thus, the study of the morphology and structure of magnetosomes may contribute to understanding the process involved in magnetosome formation. High-resolution transmission electron microscopy (Matsuda et al., 1983; Mann et al., 1990), electron diffraction, X-ray microanalysis, and recently scanned probe microscopy (Farina et al., 1994; Proksch et al., 1995) have been applied in the study of bacterial magnetite. With these approaches, several morphologies of bacterial magnetite crystals were reported: octahedral with truncated faces, pseudo-hexagonal prismatic, and bullet-shaped (Matsuda et al., 1983; Thornr 1999 WILEY-LISS, INC. hill et al., 1994). In samples of uncultured magnetotactic bacteria, it is desirable to directly visualize the faces of the magnetic crystals or to distinguish crystals of different morphologies when mixed. In this work, we used an energy-filtering transmission electron microscope to image bacterial magnetite crystal faces with low-loss inelastic electrons. The objective was to directly observe faces of magnetite crystals from bacterial origin. The results obtained using this approach are comparable to those obtained by high-resolution scanning electron microscopy (this work) and atomic force microscopy (Farina et al., 1994). MATERIALS AND METHODS Sampling and Magnetic Enrichment of Bacteria Sediments were collected in Itaipu lagoon (43° 04’ W ⫻ 22° 57’ S), near Rio de Janeiro city, placed in bottles, and stored under dim light for several weeks. Periodically, a drop with sediment was exposed to a properly aligned magnetic field and observed under a light microscope. If the presence of a relatively large number of magnetotactic bacteria was confirmed, the bottle was considered suitable for enrichment. In this Contract grant sponsor: PRONEX; Contract grant sponsor: FINEP; Contract grant sponsor: TWAS; Contract grant sponsor: NIDCD-NIH. Correspondence to: Dr. Marcos Farina, Laboratório de Biomineralização, Departamento de Anatomia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, CCS, Bl. F, 21941–590, Rio de Janeiro, RJ Brasil. E-mail: email@example.com Received 4 December 1998; accepted in revised form 6 April 1999 320 U. LINS ET AL. Fig. 1. a: Isolated magnetite crystals deposited over a formvar-coated grid observed by TEM in bright field mode. The crystals appear dark in most of the cases. b: Image obtained by inelastically scattered electrons (⌬E ⫽ 50 eV). Crystals present different optical densities though faces are not well defined. Bar ⫽ 400 nm. case, a specially designed glass chamber was filled with sediment and water in a proportion of approximately 1:3, and exposed to the magnetic field for 15 minutes. Afterwards, drops with enriched bacteria were withdrawn with a capillary tube and placed in an eppendorf tube for isolation. Isolation of Magnetic Crystals Triton X-100 or SDS detergents were added to an Eppendorf tube containing enriched magnetotactic bacteria and agitated vigorously using a Pasteur pipette. The tube was then washed several times with distilled water to remove the detergent. Afterwards, the crystals were treated with a solution of 40% NaOH in water at 60°C for 1 hour to remove the organic matter, particularly nucleic acids. To avoid losing crystals during the isolation procedures, a strong magnet was placed at the lower end of the Eppendorf tube during the whole procedure and after each step the tube was left still for 30 minutes. The isolated crystals were then washed extensively with distilled water and spread over formvar-coated copper grids. Some samples were treated with NaOCl in place of the detergent, however, all other procedures are identical. Scanning Electron Microscopy For scanning electron microscopy observation, isolated crystals were deposited on silicon chips specimen supports (Pelco International, Redding, CA, USA) and air-dried. Samples were mounted on aluminum stubs and observed uncoated at 5 kV in a Hitachi S-4500 scanning electron microscope equipped with a cold cathode field emission gun. Shadowing of Isolated Crystals and Observation Grids with isolated crystals were shadowed unidirectionally with platinum at 45° in a Balzers freezefracture apparatus and observed in a Zeiss CEM902 transmission electron microscope operating at an accelerating voltage of 80 kV. This microscope has a spectrometer that permits the selection and imaging of inelastically scattered electrons of defined energy losses (Reimer, 1991). The diameter of the objective aperture was 60 µm (collection angle of approximately12 mrad) and the energy-selecting slit aperture was approximately 20 eV. All negatives from shadowed crystals were photographically reversed before printing. This is a common procedure in quick-freeze, deep-etch preparation of biological samples for 3D electron microscopy (Heuser, 1989). For energy loss spectroscopy, the microscope was operated at 80 kV and an energy-selecting slit aperture of approximately 2 eV was used. Energy loss intensities from 0 to 200 eV were registered at 2-eV intervals by a photomultiplier attached to the microscope and connected to a digital multimeter that loaded the output voltage into the computer memory of a Kontron Zeiss image analysis system. RESULTS There was a mixture of morphologically different crystals in all samples observed. Most of the crystals presented the already described pseudo-hexagonal morphology (Farina et al., 1994), showing long faces 51016 and small oblique faces of the type 51006 and 51106 and 51116 end faces. These crystals were probably isolated from some of the two most abundant morphotypes present in the samples (Spring et al., 1998). Observation of isolated crystals was carried out in different imaging modes of TEM. Crystals without platinum shadowing were imaged with zero-loss electrons (Fig. 1a) and inelastic imaging mode (Fig. 1b). Little surface information could be obtained from these images no matter which diffraction and photographic exposure conditions were employed during imaging of the crystals. Magnetosomes shadowed with platinum were imaged in both in zero-loss mode (Fig. 2a; see also Fig. IMAGING FACES OF Fe3O4 CRYSTALS 321 Fig. 2. a: Zero loss filtered transmission electron microscopy image of shadowed magnetosomes. Very little surface information about the crystals can be assessed particularly in the region of large crystals. b: Inelastically filtered image (⌬E ⫽ 77 eV) of the same shadowed crystals. Due to contrast tuning, the surface of the crystals can now be observed. Arrow indicates a 51016 crystal face. Bar ⫽ 120 nm. Fig. 3. Inelastically filtered image of shadowed crystal. The effect of the decoration pattern of the platinum microcrystals can be observed and the faces of crystals in different orientations can be observed. ⌬E ⫽ 68 eV; Bar ⫽ 100 nm. Inset: Idealized faces for some of the shadowed crystals. (䊏) Faces of the type 51006 or 51106; (ⴱ) Faces of the type 51016. 4a) and with inelastically scattered electrons (Figs. 2b, 3, and 4b). In zero-loss, the contrast between magnetosome and support film was very high. However, because of the superposition of the platinum layer on the whole magnetite crystal, very little surface information could be obtained from images formed primarily by the 322 U. LINS ET AL. Fig. 4. Shadowed isolated octahedral crystals. a: Zero loss filtered image. Note that no information on crystal faces can be observed. b: Inelastically filtered image of the same area in the previous figure (⌬E ⫽ 104 eV). Now the characteristic 51116 crystal faces can easily be evaluated. Bar ⫽ 90 nm. Fig. 5. Electron energy loss spectra of the supporting film (I) and of a shadowed crystal (II). The intensities of the spectra obtained from the supporting film alone and over the crystal are closer in the inelastic part of the spectra. This means that the contrast is much reduced in the inelastic images. At a certain point (⬇110 eV) there is an inversion of the intensities. scattering contrast of the elastically scattered electrons, especially in the case of the larger magnetosomes (Fig. 2a). A much better imaging condition was obtained when shadowed crystals were imaged with inelastically scattered electrons (Figs. 2b, 3, and 4b). There was a contrast attenuation in the images, associated to the shadowing of the platinum layer covering the crystals, and the surface information could be retrieved from the crystals. It was possible to visualize the decoration pattern of the platinum crystals on the surface of the underlying magnetite crystals. Faces of the crystals could be directly observed and the relative disposition of one face to another could be established in some cases. In the inset of Figure 3, elongated faces 51016, as well as smaller faces of the crystals of the type 51006 or 51106, can be seen. Besides pseudo-hexagonal prismatic crystals, we also observed a few octahedral crystals presenting the representative well-defined 51116 faces (Fig. 4a and b). Electron energy loss spectra from regions of the crystal and the supporting film were obtained (Fig. 5). In these spectra, it can be noted that the differences in intensities between the crystal and the supporting film were much closer in the inelastic part of the spectra (note the vertical log scale which means that the ratio between the elastically scattered electron signal of both curves is very high). The effect of contrast tuning observed for energy losses caused an inversion of the contrast in the images in a given point between 100 and 150 eV (Fig. 5). Scanning electron microscopy of similar preparations imaged using a field emission microscope produced IMAGING FACES OF Fe3O4 CRYSTALS 323 Fig. 6. Scanning electron microscopy image of isolated magnetite crystals from magnetotactic bacteria. Crystals are disposed in chains and arranged in different orientations. Several faces can be directly visualized. Bar ⫽ 150 nm. results of comparable image quality (Fig. 6). Magnetic crystals appeared in linear chains and were disposed in different orientations within a chain. The faces could be directly visualized. DISCUSSION Magnetosomes have been studied with a structural approach particularly in aspects of biomineralization and crystallography (Stolz, 1993; Bazylinski, 1995). With atomic force microscopy (Farina et al., 1994) and the methodology described here, it is possible to obtain data from different faces of previously known crystals, and to discriminate different morphologies when a mixed population of crystals is studied. We used an ‘‘in column’’ magnetic spectrometer to directly observe surfaces of magnetic crystals decorated with metal aggregates of platinum. The replication at 45° produced a decoration pattern of platinum particles of probably 2.5 to 3.0 nm. The use of vertical deposition of a thin (1 nm) platinum-carbon metal layer (Aksiyote-Benbasat et al., 1990) would have produced smaller platinum particles with a better resolution at normal (global) imaging conditions. In these normal imaging conditions, quantification of particles sizes is possible by calibrating shadow widths with gold ball standards (Ruben, 1995). However, when using inelastic electrons to image crystals in an energy-filtering electron microscope, the resolution is limited not only by the size of the platinum crystals but also by the delocalized nature of the inelastic collisions (Reimer, 1991). This could reduce the possibility of improvement of final image resolution in inelastic mode even with a thinner metal layer of smaller platinum particles. This improvement will be checked in the future. Indexation of planes is not possible with contrast tuning of metal-shadowed crystals, which is different from the results obtained by high-resolution transmission electron microscopy and electron diffraction. On the other hand, our approach is less time-consuming, and does not require tilting of crystals to determine correct faces, being a useful tool for qualitative work. Inelastic scattering results from the interaction of incident electron with atomic electrons. The total kinetic energy of the incident electron is reduced by a certain amount of energy necessary to excite an electron of the specimen atom (Reimer, 1991). The effect of the scattering events can be observed in the electron energy loss spectrum and many of its features are dependent on the nature of the elements present in the samples. Differences observed between the electron energy loss spectra of the crystal region and the supporting film can be explained by differences in mass thickness between these two regions. The electrons diffracted by the whole magnetite crystal, which is a crystallographic single domain, do not generate contrast necessary for the visualization of surfaces because the whole crystal is in the same Bragg condition. However, different regions of the crystal present different mass thickness of the decoration pattern of platinum particles, which result in differential scattering of the inelastic electrons. Therefore, shadowed faces of the underlying crystal can be selectively visualized in one micrograph due to differences in scattering properties of inelastically scattered electrons from the bulk and from the film on the surfaces of the magnetite crystals. This process of image formation is the same regardless of the size of the platinum particles. As no interference fringes were observed on carbon holey films by defocusing (not shown) in the energy losses used in the present work, it was concluded that the contrast observed for the platinum film has no contribution from the phase contrast effect. We recently reported unusually large magnetosomes, twice as large as usual magnetosomes, in Southseeking magnetotactic bacteria from a lagoon near Rio de Janeiro city (Farina et al., 1994). These magnetosomes were considered unusual because they could fall outside the narrow size range for single domains of magnetite according to theoretical calculations raising questions about the phylogenetic relationship among bacteria that produced them (Spring et al., 1998). In this work, we observed the elongated faces and also the smaller oblique faces of the large crystals. The pseudohexagonal crystals are capped by 51116 faces that are larger than similar crystals already described. Probably, the crystals remain in chains because these faces are quite large as a consequence of the exceptional width of the crystals. The precise characterization of the faces, however, requires high-resolution transmis- 324 U. LINS ET AL. sion electron microscopy studies. The oblique faces 51006 and 51106 present in these larger crystals have a relatively smaller area if compared to other pseudohexagonal crystals already described for magnetotactic bacteria. Sometimes these faces are even absent. This probably means that the velocity of growth in the lateral faces of the crystals with respect to their 51116 capping faces is higher in these larger crystals than in the smaller ones usually described in the literature. Control by the organisms over the biominerals formed is exerted by a series of possible mechanisms involving controlled ionic environments and chemical processes (Mann, 1986). We conjecture that the present technique, despite its relatively low resolution compared to other shadowing techniques (Ruben, 1998), represents an easy method for the discrimination of morphologies of previously known small particles when mixed and may contribute as complementary data for studying multilayered objects under transmission electron microscopy. ACKNOWLEDGMENTS We thank Carolina N. Keim and Dr. Angela Hampshire for discussion and Marianne Parakkal for technical help. REFERENCES Aksiyote-Benbasat J, Ruben G C, Marx K A. 1990. Solution structure of bacteriophage T4D and icosahedral capsid geometry visualized in freeze-fractured, deep-etched replicas. J Biol Struct Dyn 7: 773–794. Bazylinski DA. 1995. Structure and function of the bacterial magnetosome. ASM News 61:337–343. Farina M, Kachar B, Lins U, Broderick R, Lins de Barros HGP. 1994. The observation of large magnetite (Fe3O4) crystals from magnetotactic bacteria by electron and atomic force microscopy J. Microsc 173:1–8. Gorby YA, Beveridge TJ, Blakemore RP. 1988. Characterization of the bacterial magnetosome membrane. J Bacteriol 170: 834–841. Heuser J. 1989. Protocol for 3-D visualization of molecules on mica via the quick-freeze, deep-etch technique. J Electron Microsc Tech 13: 244–263. Lins de Barros HGP, Esquivel DMS, and Farina M. 1990. Magnetotaxis. Sci Progr 74:347–359. Mann S. 1986. Biomineralization in lower plants and animals: chemical perspectives. In: Leadbeater BSC, Riding R, editors. Biomineralization in lower plants and animals. New York: Clarendon Press. p 39–54. Mann S, Sparks NHC, Wade VJ. 1990. Crystallochemical control of iron oxide biomineralization. In: Frankel RB, Blakemore RP, editors. Iron biominerals. New York: Plenum Press. p 21–49. Matsuda T, Endo J, Osakabe N, Tonomura A. 1983. Morphology and structure of biogenic magnetite particles. Nature 302:411–412. Pósfai M, Buseck P R, Bazylinski D A, Frankel RB. 1998. Reaction sequence of iron sulfide minerals in bacteria and their use as biomarkers. Science 280:880–883. Proksch R, Moskowitz BM, Dan Dahlberg E, Schaeffer T, Bazylinski DA, Frankel RB. 1995. Magnetic force microscopy of the submicron magnetic assembly in a magnetotactic bacterium. Appl Phys Lett 66: 2582–2584. Reimer L. 1991. Energy-filtering transmission electron microscopy. Adv Electr Electr Phys 81:43–126. Ruben GC. 1995. Quantification of particle sizes with metal replication under standard freeze-etching conditions: a gold ball standard for calibrating shadow widths was used to measure freeze-etched globular proteins. Microsc Res Tech 32:312–329. Ruben GC. 1998 Vertical Pt-C replication, a revolution in imaging non-periodic macromolecules, biological gels and low-density polymer networks. Micron 29:359–396. Spring S, Lins U, Amann R, Schleifer K, Ferreira L, Esquivel D, Farina M. 1998. Phylogenetic affiliation and ultrastructure of uncultured magnetic bacteria with unusually large magnetosomes. Arch Microbiol 169:136–147. Stolz JF. 1993. Magnetosomes. J Gen Microbiol 139:1663–1670. Thornhill RH, Burgess JG, Sakaguchi T, Matsunaga T. 1994. A morphological classification of bacteria containing bullet-shaped magnetic particles. FEMS Microbiol Lett 115:169–176. Vali H, Kirschvink JL. 1990 Observations of magnetosome organization, surface structure and iron biomineralization of undescribed magnetic bacteria: evolutionary speculations. In: Frankel RB, Blakemore RP, editors. Iron biominerals. New York: Plenum Press. p 97–115.