This action might not be possible to undo. Are you sure you want to continue?
© Elsevier, Paris
Scanning transmission electron microscopy of biological structures
Christian Colliex, Claudie Mory
Laboratoire de Physique des Solides, B~timent 510, Universit~ Paris-Sud, 91405 Orsay, France
Summary - The design of the scanning transmission electron microscope (STEM) has been conceived to optimize its detection effi-
ciency of the different elastic and inelastic signals resulting from the interaction of the high energy primary electrons with the specimen. Its potential use to visualize and measure biological objects was recognized from the first studies by Crewe and coworkers in the seventies. Later the real applications have not followed the initial hopes. The purpose of the present paper is to describe how the instrument has practically evolved and recently begun to demonstrate all its potentialities for quantitative electron microscopy of a wide range of biological specimens, from freeze-dried isolated macromolecules to unstained cryosections. Emphasis will be put on the mass-mapping, multi-signal and elemental mapping modes which are unique features of the STEM instruments.
scanning transmission electron microscopy / quantitative electron microscopy / mass mapping / multi-signal mapping / elemental mapping
The scanning transmission electron microscope (STEM) conceived, designed and realized in the late sixties by Crewe and his coworkers, was deliberately aimed at biological applications. As a matter of fact, the first detailed papers describing the instrument, its working modes and the first results were published in 1970 [12, 13]. Many important potential capabilities were demonstrated rapidly and it was suggested that the STEM could be used for DNA sequencing. This was the state of things at the 1970 ICEM meeting in Grenoble. This idea has survived over the next decade as demonstrated by Cole et al  who published in 1977 an excellent assessment of DNA sequencing by imaging techniques. However, the initial impetus gradually faded away in a context of rapidly developing chemical methods. Very few STEMs were then maintained in operation around the world for biological research. At the same time, its intrinsic performance for achieving nanoanalysis in materials science was recognized. It then fostered regular progress in column design, in spectroscopy, in detector technology and in digital integration, leading to the very spectacular results commented as "the Ultimate Analysis" in one of the latest issues of Nature . The small community of STEM users who has remained active in biology until now, is concentrated in two national facilities in US (Brookhaven National Labs and NIH, Bethesda) and in a few European laboratories (Biozentrum in Basel, EMBL in Heidelberg, Molecular Genetics in Strasbourg, Solid State Physics in Orsay). They have gathered over the years a lot of expertise in STEM instrumentation, methodology and adaptation to the specificity of the biological specimen. The output of this longterm effort has come out during the last couple of years with substantial results on a wide range of molecular and supramolecular structures. For instance, 36 papers dealing with biological studies carried out with a STEM have been identified in the literature for the years 1991-1992.
More evidence of the vitality of the field is the simultaneous publication in 1993 of two review papers dedicated to the present state of STEM in biology [1, 17]. The purpose of the present contribution is to highlight, in the context of a special issue of Biology o f the Cell dedicated to a wide audience of biological electron microscopists, the specific aspects of the STEM which make it unique in terms of quantitative electron microscopy and information detection efficiency.
The STEM as an imaging instrument
The STEM is made of two major components (see fig I): on one side, the illumination system, including the field emission gun, which delivers a sub-nanometre probe of incident 100 kV electrons on the entrance surface of the specimen; on the other side, a family of detectors including an energy loss spectrometer for the discrimination in scattering angle and energy transfer of the transmitted electrons. Accessory detectors can be introduced to collect secondary electrons or X-rays. With this general design, the STEM can be regarded as a highly efficient instrument built and used by physicists to measure scattering events consecutive to the interaction of the well-defined primary beam with a given volume of material designed as the target. Consequently it is not surprising that it has become a performant instrument for point analysis in the hands of materials scientists. Nevertheless, such an instrument is also an excellent imaging device when the probe is rastered on the specimen and the ensuing signals displayed in accordance to the scanning grid. This sequential and controlled acquisition of information offers several advantages with respect to the conventional imaging modes (scattering contrast, phase contrast) used in the fixed-beam transmission electron microscope. It lends itself to quantitative analysis, it can handle the signals simultaneously delivered by the
using its linear relationship with the elastic dark field signal. Consequently this imaging mode can be used for visualizing and measuring different types of specimens. The technique independently introduced by Engel  and Wall . weak structures. This provides a nice opportunity to image beam-sensitive biological macromolecules at low dose. nucleoprotein particles and nucleic acids give rise to images with quite reproducible fine structures [27. Many papers dealing with mass per length of filaments. The contrast of single heavy atoms is so high (as demonstrated in the early images by Crewe et al ). A special use of the inelastic signal has been demonstrated by Colliex et al  for the imaging of thick specimens. clusters of a few atoms such as undecagold clusters. The major field of application of this ADF imaging mode is however the mass measurement of rapidly frozen. If the average atomic number is nearly constant. as long as it remains typically below 10. It corresponds to the thickness range where the 'mixed' type of scattering described in the table is dominant. the mixed signal becomes dominant for thicker specimens and the unscattered signal decreases as an exponential with thickness. fig 2a). Multisignal imaging The inelastic signal measured at the exit of the energy loss spectrometer. Similar experiments involving Europium-doped chelates attached on IgG molecules are presently undertaken in our laboratory in collaboration with Barbin and Delain. One (or several) annular detectors collect the electrons which have been scattered at large angles by an elastic scattering event. A positive contrast therefore corresponds to the presence of scattering centers with high Z (the total elastic cross-section ae varies roughly as Z3/2). which practically amounts to the total inelastic signal. the 'plasmon' peak) in the electron energy loss spectrum. The magnetic spectrometer followed by the PEELS detector is used to discriminate the different contributions in the energy loss spectrum. the ADF intensity varies linearly with the mass thickness. One can classify the interaction mechanisms. This method consists in selecting an energy loss window slightly below the maximum of the energy loss distribution. C Mory STEM Biological applications for the different imaging modes STEM annular dark field mode Magnetic spectrometer An annular dark field (ADF) detector collects with high efficiency all electrons scattered at large angles. is not very specific for the specimen composition. The lack of phase contrast. One can therefore imagine how a well chosen strategy of detection can be used to extract a selected type of information for a minimum dose. The beam issued from a field emission source is focused into a sub-nanometre probe on the specimen entrance surface.1) to supramolecular structures (such as the head of a mature bacteriophage T4 of mass 194 MDa. that sitespecific labeling techniques with clusters made of heavy atoms have emerged . In the above classification. has evolved into a routine tool that can be applied to a wide range of biological specimens from small macromolecules (such as DNA of mass 2 kDa n m . the sum of the elastic. the linearity and the strong signal-to-noise ratio produce images easy to interpret and adapted to image processing.5 g/cm 2.176 C Colliex. ie the intensity of the whole EELS spectrum excluding the zero-loss peak constitutes another approach to mass measurement. very sensitive to beam irradiation. Schematic diagram of a dedicated STEM with its illumination optics and detection system. ie with thickness of a few inelastic mean free paths. Negatively stained or freezedried metal coated s u p r a m o l e c u l a r assemblies. But. mass per area of sheet structures. it may be less sensitive to faint variations in average composition because of the Z 1/2 dependence of the total inelastic cross-section % In spite of its high intensity. Table I requires a few additional comments. The low loss contribution (generally a broad peak at about 25 eV. practically of the order of a few hundred nanometers. . The efficiency of detection is quite high because nearly all electrons having come through the specimen are collected by one or another of the detectors. Furthermore. on its low energy loss side. mixed and unscattered signals is equal to one. The elastic and inelastic signals behave linearly with thickness as long as it remains substantially below 100 nm. radial mass profile of cylindrical or spherical particles have been published over the last few years (see Engel and Colliex  for a review).) "o Eo. inelastic. / Large angle ADF detector Specimen ~ Objective apedure Objective lens Electron source Fig 1. freeze-dried macromolecules. are due to the molecular bonds. it is generally not used in this simple form because of its intrinsic poor spatial resolution due to the delocalization of the plasmon inelastic process. different channels of detection.AE LU LU Q_ Small angle ADF detector~. The interpretation of images relies on a few simple considerations in terms of scattering probability. To avoid the beam-induced movement of individual atoms. and has been Eo Quadrupoles / 0. thus providing access to multisignal-imaging. constitute stable specific chemical probes which can be visualized with a high degree of localization when covalently linked to antibody fragments . 28]. the type of information which is carried by the scattered electrons and the type of detector required to record them (table I).
bound elements can now be detected within macromolecular assemblies prepared by rapid freezing methods. The simple assumption is to assign the elastic contribution to the ADF signal and the inelastic one to the post spectrometer detector after exclusion of the zero loss.01-0. 25]. All these imaging techniques (annular dark field. The Z-contrast in particular is produced by the ratio o f the elastically and inelastically scattered electrons. structural. structural Scattering probability for a 50-nm organic layer = 25070 = 35°7o = 107o = 0. They provide many possibilities such as visualizing the ultrastructure. this is a critical value for the appearance of mass loss in beam-sensitive material. At room temperature. however. Using the classification in table I. In particular. Consequently one sees that at the opposite of ultrathin macromolecules deposited on carbon foils.I 07o = 10070 = 3007o Type of detector required Annular dark field Post spectrometer single detector Post spectrometer array of detectors Post spectrometer array of detectors Annular dark field Bright field before spectrometer (or post spectrometer single detector) shown to provide the maximum signal-to-noise for this kind of specimen. At low temperatures. The great success of the method has been the visualization o f single atoms o f high Z number by Crewe et al . of chromosomes for instance. Classification of scattering events and associated channels of information used in a STEM. such as those realized in the cryo-STEM in Heidelberg. which is unique to the STEM design. Then the current density is increased by several orders of magnitude for the elemental analysis of important elements such as phosphorus or calcium which have strong L23 edges at 150 and 350 eV [24. It means that their use for chemical analysis of biological objects requires very high doses.13th International Congress on Electron Microscopy Table I. The. Two configurations have then been practically realized to bring this suggestion into practice: the one in Orsay fulfills the above condition and measures the unscattered. the more as the physiological concentrations of interest are quite weak. suggesting that the limitation in spatial resolution imposed by statistical considerations could be partially raised. The reproducible observation of unstained embedded cellular material in thin sections by this technique has then been demonstrated through a collaborative study between the groups in the Biozentrum. in Heidelberg. The specimen is first visualized at low doses. this elementary ratio contrast between two simultaneous signals is not really thickness independent. the price to pay in terms of primary electrons to get access to such a specific information. it behaves roughly as the atomic number Z. and the Solid State Physics Laboratory. thick cellular sections. multisignal) can be used on any type of cryofixed specimens. Only in the case of very thin s~nples it depends more on the chemical composition than on thickness variations. Core losses Mixed Unscattered Nature of extracted information Topographical. 7. 177 Type of scattering Elastic Inelastic --.d show ADF. shows that as a consequence o f the non-negligible contribution of the multiple elastic-inelastic events into the ADF detector. On the other hand. structural. as the cross-sections for the involved signals are rather strong. exact formulae. projected mass Topographical. typically three orders of magnitude lower than those for elastic or inelastic scattering used in the above modes. Basel. eventually its molecular mass can be mapped using the ADF signal at typical doses of the order of 103 e / n m 2. have the. However. In addition to acquiring EELS data from a point on the . Furthermore. They show that three simultaneously recorded signals are required. The implementation of detector arrays that collect the energy-loss spectrum in parallel (hence the acronym PEELS for parallel (detection) EELS). Elemental mapping The occurrence of atomic specific core edges in the EELS spectra identifies the presence of the associated elements in the irradiated volume. the annular dark field and the inelastic signals . Figures 2 b . mapping the DNA/protein distribution and localizing labels. This latter configuration has been used to measure the concentration of chromatin in unstained cryofixed and freeze-substituted sections .~nbeen established to extract a thickness independent ratio contrast. one records a filtered elastic dark field and the inelastic sig- nals . can also be visualized in optimum conditiohs thanks to the versatility of the detection system. inelastic and ratio images of sectioned T4 phages adsorbed on E coli envelopes embedded in an organometallic methacrylate resin. it seems that much higher doses give no difference for the measurement of protein packing densities. 9]. we believe that the most pertinent field of application is to use the possibility of multisignal recording and mixing. The starting idea is to discriminate the weak changes in contrast due to the compositional differences between the embedded materials and the resin matrix from the roughness of the section surfaces produced by the cleavage process during sectioning. has greatly increased the efficiency of the technique. Their relevant cross-section (or relative probability) is very low. structural Topographical. Total Low losses --. However. Orsay [6. is very high. The detailed theory. low-dose imaging conditions with typically 100 e / n m 2 can be used. projected mass Molecular bonding Elemental composition Topographical.
+ . . the ratio image or the mapping of a more abundant element.178 ~++::i:~+ ~ . N and O edges. Although the total dose for acquiring these data is approximately 10 9 e/nm 2.d. using the possibility of acquiring entire spectra for each image pixel.~ ..~ + C Colliex.p B./"" E • P3 --¢+ z. . ~ ... Elements in unexpected locations may be found without a p r i o r i knowledge of their distribution.. elemental mapping can be realized at a rather quantitative level and with good sensitivity.. we think that these elements covalently integrated in the biological structure can be used as indicators for nucleic acids and phospholipids (phosphorus) and proteins (nitrogen). ". +~-E #P3 I A / "- t .. by fitting to reference edges the Ca signal summed over typically thousand spectra corresponding to all pixels within these compartments. . The elements to be imaged and the spectrum-processing parameters do not need to be known in advance.": ? P1 ~. which constitutes an extension of the multisignal approach to = 103-104 simultaneous images. . 21]. ~ Ap l ~' " . It increases the sensitivity of the analysis by a factor roughly equal to the square root of the number of summed spectra. contours can be drawn on the image and spectra summed over all pixels contained within these areas. C.' .~.2. Using interactive image analysis techniques. p/. As they remain within an area smaller than 10 nm in diameter under the dose required for their identification. A very elegant demonstration of the power of this approach has been recently published by Leapman et al . They have detected calcium concentrations at the level of a few millimoles per kilogram dry weight in specific cellular compartments on freeze-dried cryosections. Figure 2e shows the elemental maps for five elements extracted by numerical processing from a collection of 64 × 64 spectra encompassing the useful P. recorded on an air-dried T4 bacteriophage. ~" 4t "" ° + ... / 1/--."7" +. < . specimen. 11. C Mory .. This mode has been introduced as the spectrum-image technique by Jeanguillaume and Colliex  and practically used over the last 2 years [2.. such as the ADF image. no noticeable changes of the phosphorus and nitrogen signals have been observed and the measured atomic concentrations are in good agreement with the nominal values.-0 B.. offers many advantages. This spectrum-imaging technique.. ' + ' e GI ~' GI .. G1 ~1 E Pl E" @ Fig. I~ II~"* ~ " i ' I u ~ . The search for an element can be guided with the support of extra information... Ca.
13th International Congress on Electron Microscopy 179 Photodiode counts CK 400 200 P L23 V~on phaqe head 0 | I on carbon fo~l I | I 2OO 400 6O0 Energy loss ( e V ) e Fig 2. calcium (Ca). Dark field image mainly formed by elastically scattered electrons. respectively. Sectioned T4 phages adsorbed on E coli envelopes embedded in an organometallic (tin resin) methacrylate resin. c. Air-dried T4 bacteriophage elemental maps of phosphorus (P). The 4096 spectra were recorded with 4 nm pixel size.d . Different ways of seeing T4 bacteriophages with a STEM. Ratio of the two previous images with reversed contrast. There is no major impediment in applying all the quantitative data collection and processing techniques described in the present review. notice the poorer spatial resolution. a. Image formed by inelastically scattered electrons. filamentary bundles of DNA (G2) (courtesy of Carlemalm et al ). d. we are convinced that the expertise gained in all the laboratories which have continuously improved the instrumentation and the methodology. All our colleagues who have worked with us over the . Conclusion More than 20 years after the first impressive results b r o u g h t by the S T E M . used for mass measurement (capsid mass is 194 _+ 2 MDa from Baschong et al ). b . e. 0. nitrogen (N) and oxygen (O) extracted from the processing of the relevant core-losses in a 64 x 64 spectrum-image. to specimens prepared and maintained in their physiological state by using all the resources o f the cryotechniques. carbon (C). One can distinguish empty phages with contracted tail sheaths (P]). The two PEELS spectra correspond to the sum of six pixels on top of the phage head and on the supporting carbon foil.3 s dwell time per pixel and the recording dose has been 109 e/nm 2. Acknowledgments Special thanks are due to Andreas Engel whose collaboration in a critical evaluation of STEM potentialities in biology has been decisive. Annular dark field image of a freeze-dried specimen. can now be fruitfully used for biological studies. The psychological barriers due to the fact that the first instruments had been built and used by physicists should quickly vanish when more friendly machines appear on the market. b.
Wagner D. Jeanguillaume C. Tenc6 M (1992) New STEM multisignal imaging modes. Engel A. 137-175 . Ornberg RL (1988) Quantitative electron energy loss spectroscopy in biology.268 25 Leapman RD. 531-546 3 Baschong W. Buchanan RA. J Mol Biol 48. Williams DB (1991) Electron energy-loss spectrumimaging. Olins AL. 177-206 10 C. Beaman DR. 161-168 28 Trent JD. Ultramicroscopy 28. Leapman RD (1993) Biological scanning transmission electron microscopy. Christian Jeanguillaume. In: Scanning Electron Microscopy (Johari O. 205-217 24 Leapman RD. 93-101 4 Bohrmann B. 47-74 22 Jeanguillaume C. Kellenberger E (1982) The reproducible observation of unstained embedded cellular material in thin sections: visualisation of an integral membrane protein by a new mode of imaging for STEM. Colliex C. 225-238 26 Leapman RD. 240-247 19 Hainfeld JF (1987) A small gold-conjugated antibody label: improved resolution for electron microscopy. References 1 Andrews SB. Jeanguillaume C (1993) Electron energy loss spectrometry mapping. Haider H. Horwich AL (1991) A molecular chaperone from a thermophilic archaebacterium is related to the eukaryotic protein T-complex polypeptide-1. Aebi U (1991) Mass analysis of bacteriophage T4 proheads and mature heads by STEM and hydrodynamic measurements. Science 168.olliex C. Ultramicroscopy 45. Nature 366. In: Physical aspects of electron microscopy and microbeam analysis (Siegel BM. Lustig A. 252-257 23 Jeanguillaume C. Hunt JA. Science 236. 375-393 14 Crewe AV. Wall J. Wiggins JW. Kellenberger E (1985) Contrast formarion in electron microscopy of biological material. Michel J. Hartl FU. Microsc Microanal Microstruct 2. Langmore J (1970) Visibility of single atoms. Puchelle E. Ada and Dan Olins. C Mory 13 Crewe AV. eds) John Wiley and Sons. UItramicroscopy 28. Edouard Kellenberger. UItramicroscopy 24. 1-21 11 Colliex C. Zulauf M. Pinon JM (1991) Parallel EELS elemental mapping in STEM: use of the difference methods. Etienne Delain. Wall JS.180 C Colliex. Beer M (1977) Molecular microscopy of labeled polynucleotides: stability of osmium atoms. Q Rev Biophys 31. Andrews SB (1993) Measurement of low calcium concentrations in cryosectioned cells by parallel-EELS mapping. Gu H. in particular we wish to thank Eric Carlemalm. JMol Biol 117. Marcel Tenc6. 63-67 7 Carlemalm E. 387-400 9 Colliex C. Colliex C. In: Adv in Electronics and Electron Physics (Hawkes PW. Ultramicroscopy 3. Olins DE.281 17 Engel E. New York. Tenc6 M (1989) Energy filtered STEM imaging of thick biological sections. Eur Microsc Anal 24. Tenc6 M. Colliex C. 135-144 21 Hunt JA. Ploton D. EMBO J 1. 270-344 8 Cole MD. Nimmersgen E. Wall J (1970) A scanning microscope with 5 resolution. 490-493 29 Wall JS (1979) Mass measurements with the electron microscope. Reichelt R. J Ultrastr Res 88. Ultramicroscopy 38. Lef~vre E. Kellenberger E. Joanna and Romuald Wroblewski. Revet B. ed) Academic Press 63. ed) SEM Inc Chicago 2. in press 12 Crewe AV (1970) The current state of high-resolution scanning electron microscopy. 251 . Delain E (1981) Improved visualization of single and double stranded nucleic acid by STEM. Thomas X. Colliex C (1989) Spectrum-image: the next step in EELS digital acquisition and processing. 21-24 2 Balossier G. Nature 354. Ballongue P. Mikrochfln Acta 112. UItramicroscopy 7. J Microscopy 153. Ultramicroscopy 46. Bonhomme P. Langmore JP. 225 -234 27 Mory C. Isaacson MS (1975) Resolution and contrast in the STEM. Baschong-Prescianotto C. 291-303 10 years of active research on biological specimens with the Orsay STEM have to be acknowledged. Mory C. Bonhomme A. UItramicroscopy 49. 1338-1340 15 Crewe AV. 235-251 5 Brown LM (1993) The ultimate analysis. 721-721 6 Carlemalm E. Mory C (1984) Unconventional modes for STEM imaging of biological structures. 47-62 16 Engel A (1978) Molecular weight determination by STEM. made accessible through the evaluation of detection efficiencies. J Struct Biol 106. 403-411 18 Haider M (1989) Filtered dark-field and pure Z-contrast: two novel imaging modes in a STEM. Ultr~microscopy 49. Colliex C (1993) Application of STEM to the study of biological structure. Bouchet D. Mory C. J Microscopy 165. 273. Kellenberger E (1993) Concentration evaluation of chromatin in unstained resin-embedded sections by means of low dose ratio-contrast imaging in STEM. Andrews SB (1992) Characterization of biological macromolecules by combined mass mapping and electron energy loss spectroscopy. 450-453 20 Hainfeld JF (1992) Site-specific cluster labels. Curr Opin Biotechnol 4.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.