Historical Background of the BNL STEM The STEM (Scanning Transmission Electron Microscope) concept was pioneered by von Ardenne at the same time as the early development of the TEM. However his efforts were limited by the low brightness of available sources, lack of suitable electronics and poor recording media. Development of the field emission electron gun by Crewe, et al. solved the first problem. Commercially available electronics components provided most of the electronics needs. The annular detector dark field geometry developed by Crewe & Wall made possible imaging of unstained DNA1 and single heavy atoms2. Initial biological results by Wall3 indicated that specimen preparation and radiation damage were the main limiting factors in STEM imaging of biological specimens, but that with suitable precautions quantitative measurements could answer important questions. Even at that early stage it was possible to show that filamentous viruses were single cylinders with M/L of ~1.6 kDa/A and "split" structures were actually side by side aggregates of two particles rather than separate strands of a single virus. The early progress with the STEM at Chicago attracted the attention of John Blewett and Elliott Shaw at Brookhaven who obtained funding from NSF, NIH and DOE in 1971 to develop a STEM Facility along the lines of already-functioning user facilities at BNL. The initial plan was to copy one of the Chicago designs and M. Isaacson was recruited to oversee the construction. When Isaacson returned to Chicago in 1973, J. Wall was recruited with the understanding that design modifications would be considered (not permitted in the original plan). By then it was clear that specimen preparation was severely limiting the quality of results. Therefore a new design of the lens, specimen changer and detectors was developed, emphasizing optimum conditions for the specimen. The present external freeze dryer and vacuum transfer cartridges were part of that complete package. Optics of the STEM were optimized to provide detector acceptance angles and magnification independent of specimen position in the lens gap. All of these features have proven their importance since the official opening of the Facility in Oct. 1977. The STEM remains in the forefront of biological electron microscopes in reliability, productivity and performance. Work using the STEM at Brookhaven began with a careful study of the first premise of the Facility, namely that single heavy atoms could serve as specific labels for biological specimens. It was determined that these predictions were overly optimistic and that a dose of at least 1000 el/Å2 was needed for adequate S/N (Signal to Noise Ratio)4, whereas dose-response measurements for mass loss and loss of resolution on freeze dried specimens set a useful imaging limit at 10-100 el/A2 with the specimen cooled to -140?C. Furthermore, the heavy atoms were found to move significantly from one scan to the next.5,6 Early experiments with S. Lippard and J. Barton (then at Columbia) confirmed the low label S/N for filamentous virus fd which had each of the 5,000 coat protein subunits each labeled with platinum. A promising alternative was TAMM (tetra-mercuri methane) which had 4 mercury atoms within a 3A radius. This behaved well biochemically, giving 100% labeling, but decomposed in the electron beam.7 A student of Lippard's, J. Lipka, came to Brookhaven to work on a variety of promising tungsten clusters. At the same time D. Safer (U. Penn.) made some undecagold according to the published methods of Bartlett and Singer. This gold cluster was shown to be sufficiently stable for easy visualization in the STEM, although a dose of 100 el/Å2 was required.8 The fact that the gold cluster had 21 functional groups on its surface made it interesting to work with. Biotinylating the surface groups and reacting with avidin gave spectacular ladder images which gave this effort momentum.9 Safer's interest was in labeling actin filaments and this was much more difficult, since it required a monofunctional gold cluster. Interesting images were obtained, but not of publishable quality. However cryoEM studies of these same samples did give useful results.10 At the same time as the actin work, P. Frey (U. Wis.) developed his own method for preparing monofunctional undecagold and used it to label PDH (pyruvate dehydrogenase). The 96 SH groups on the ends of the flexible arms of this multi-enzyme complex were labeled and gave impressive images, which were, however, difficult to interpret.11,12 At this point J. Hainfeld decided that the gold and/or the tungsten clusters were important enough for the future of STEM to justify a full-time effort. The Advisory Committee supported this decision and Hainfeld obtained separate NIH funding for his work, particularly development of larger clusters and biological projects. Applications of radioactive gold clusters for cancer therapy13 attracted DOE funding. The first external biological project was a study of native and reconstituted nucleosomes by C. Woodcock (U. Mass.) which confirmed most of the previous EM observations.14,15 A subsequent study of di-nucleosomes permitted image averaging.16 Woodcock also measured mass per unit length of chromatin fibers to test various nucleosome packing models.17 J. Langmore (U. Mich.) performed similar measurements to aid in his interpretation of x-ray patterns and Venki Ramakrishnan (then at BNL) also carried out a STEM study in conjunction with his neutron scattering work.18 None of these studies was definitive in proving one of the competing models for the packing of nucleosomes into chromatin fibers due to the lack of uniformity of the specimens and the question remains open for further study, with most investigators favoring Ramakrishnan's model. It may be that the observed variability is biologically important in the context of the replication and transcription "machinery". The first definitive STEM biological project was on fibrinogen and set much of the tone for future collaborations. Trinodular TEM fibrinogen images were interpreted either as aggregates or as the circulating species. E. Shaw, a noted protein chemist at BNL and co-founder of the STEM project, was able to persuade leading advocates of opposing views to agree on a joint STEM experiment to interpret conflicting results. M. Mosesson (then at Downstate Medical Center and a proponent of the aggregate interpretation) and R. Haschemeyer (Cornell U. Medical College) each provided specimens which were freeze dried on our carbon films. The resulting images were so clearcut that they have become a classic example of STEM specimen preparation.19 Trinodular objects 430A long and 75A wide with a mass of 340 kDa were found isolated on a perfectly clean background. The central domain was less intense and somewhat variable in mass but the outer nodules were reproducibly 111 ± 17 kDa, consistent with their assignment as D domains. Negative staining with uranyl acetate gave much greater internal detail. Interestingly, compact molecules of 340 kDa were observed also and a deletion was localized to the central domain. This study effectively answered any uncertainties about the trinodular nature of the fibrinogen molecule. Mosesson became a long-term collaborator and recruited a network of blood biochemists for follow-up studies on clotting factors20,21,22 and lipid vesicle-bound clotting complexes. Mosesson's own work concentrated on the initial steps of fibrin polymerization after thrombin activation and he has produced a controversial new model based on STEM mass mapping and gold labeling 23. Much of this has now been confirmed.72, 73, 75, 88 An early focus of STEM activity was specific DNA-protein interactions. Initial results showed problems in maintaining the complex on the grid, so P. Hough and I. Mastrangelo (BNL) began an extensive investigation of the requirements for preserving significant structures for STEM imaging.24 Their first result to attract the attention of the biochemists was the demonstration that SV40 T antigen bound at the adenovirus origin of replication as a double hexamer.25,26,27 This is widely accepted now but was totally unexpected and had important implications for the mechanism of opening the DNA to start replication. Other studies elucidated the mechanism of action of long range interactions.28,29, 63 Non-specific nucleic acid-protein interactions were also studied by Egelman and his collaborators.30-33, 62, 90 Our first project with A. Steven (NIH, STEM Advisory Committee) concerned the mass per unit length of intermediate filaments (IF), and the demonstration that reconstituted filaments had many of the same characteristics as native.34-36 However, considerable controversy arose from the observation that these filaments were 15 nm wide instead of the accepted 10 nm diameter. Steven and Steinert were able to rationalize this by observing that a portion of the IF protein was very similar in all types of filament, while the remainder was quite variable. This led to the now-accepted concept that the IF structure is somewhat like myosin with the tails wound together in a tight filament while the head groups protrude from the surface.37,38 The structures observed in thin sections were the central shaft but the less dense head groups were not visible. In order to test this, radial mass density profiling was developed39, which found wide application in our studies of T4 tail tubes, bacterial flagella, pili and filamentous viruses. Work on coated vesicles extended the radial density method to spherical particles and viruses. Many of the same techniques have been applied to study of Alzheimer’s filaments,57, 84 f ilamentous hemagglutinin,58 and flagellar microtubules.74 One of the key achievements of the STEM program was elucidation of the dynein structure in collaboration with K. Johnson (Penn State, STEM Advisory Committee). This motor protein for microtubules consists of three head groups connected by thin threads to a baseplate and resembles a "bouquet of flowers".40 This success attracted a number of Johnson's colleagues to replicate this work on their systems,41-43, 71 which have all been shown to be variations on the same theme. The ribosome projects of M. Boublik are among the most comprehensive STEM studies to date. The goal was to image the reassembly process for both 30S 44 and 50S 45,65 subunits, then repeat with labeled subunits44,45 +20 more, not listed The early retirement and untimely death of Boublik prevented us from completing the labeling work, which still interests us deeply. However, several important labeling results were obtained. We also examined 16S and 23S RNA from E. coli ribosomes in distilled water and various buffers, including reconstitution buffer, demonstrating an interesting progression of structures, but none with the degree of compactness of the reconstituted ribosome. In the case of specimens in distilled water, the RNA's were still compacted by a factor of 4 relative to a fully extended molecule and side branches seen in these images correlated well with predicted RNA secondary structures. Two quantitative measures were essential elements of the entire reconstitution study. First, mass and M/L mapping demonstrated that molecules were intact and not decorated with salt (giving an objective criterion for exclusion of contaminated or defective molecules). They also demonstrated that backbone and side branches tended to be either 2 or 4 stranded. A second useful measure was the apparent radius of gyration defined by ?(M(x,y)*(x-x0)2*(y-y0)2) where (x0,y0) are the coordinates of the center of mass of the selected particle and the summation is taken over all pixels within the particle image. This is not strictly comparable to the small-angle scattering Rg because it is not averaged over all possible orientations, but agrees surprisingly well in the cases tested. Its main use is in comparing the degree of compactness of particles prepared under similar or different conditions. The aims of this study were to define intermediates in the assembly pathway and to label individual proteins and trace their progress through the entire reassembly process. The first aim was accomplished but unfortunately the second aim has not yet been fully realized due to retirement, illness and death of Boublik. However several interesting labeling results were obtained with tRNA. The reassembly experiments followed protocols worked out biochemically and were highly successful. 30S 44 and 50S 45, 65 subunits were each subjects of major publications detailing initial folding steps, stable intermediates, major rearrangements and final assembly. As mentioned, the key element in both these studies was the quantitation of both mass and Rg which: 1) demonstrated that added subunits were incorporated in correct amounts, 2) allowed critical size comparison and 3) screened out flawed experiments. Multi-enzyme complexes have a compact core with binding sites at the edges and faces for other enzymes and a flexible arm involved in transport of substrate along the enzyme pathway. Three projects have given significant results with the BNL STEM: pyruvate dehydrogenase (PDH, with P. Frey et al., U. Wis.)11,12, 67, ?-ketoglutarate dehydrogenase (T. Wagenknecht, Albany, N. Francis & D. De Rosier, Brandeis) and branchedchain ?-keto acid dehydrogenase (M. Hackert, L. Reed et al., U. Texas). In each case the stoichiometry was established as well as the overall arrangement of subunits. In the case of PDH, extensive gold labeling studies were also carried out. One of the most difficult complexes to date which has given useful STEM results is the basal body studies by G. Sosinsky (then at Brandeis, now at UCSD), N. Francis and D. DeRosier (Brandeis). This structure serves as the motor to drive the flagellum, providing motive power for the bacterium. The hook, shaft and various rings making up this structure had been defined by cryoEM, but the relative masses of different structures were not determined. In this case several deletion mutants were available to determine incremental mass with and without the selected unit. These specimens are difficult to prepare entirely free of detergent and the internal TMV standard was degraded by bound detergent. Nevertheless, good differential mass measurements were obtained using the hook portion of well-preserved basal bodies as a local standard. This is another example of a case where general improvement in specimen preparation could render difficult experiments straightforward and extend the range of applications. Many viral projects, such as the assembly/disassembly projects of Bill Newcomb, Jay Brown & Alasdair Steven on herpes simplex, Sherwood Casjens on P22 and others could be considered examples of complex systems also, but for the purpose of this discussion they are not so relevant since these routinely give good quality specimens. The problem we wish to address is the variability in mass, stoichiometry and 3-D structure attributable to problems in specimen preparation (prior to freeze drying). This is distinct from the variability in specimens such as earthworm hemoglobin (EHb) which we believe to be intrinsic, or the variability due to artifacts introduced in freeze drying. A significant technical advance was adaptation of R. Williams' polylysine film treatment which increased DNA attachment by an order of magnitude but did not degrade background quality.(20) However this did not remove the need for crosslinking in the case of protein specifically bound to DNA. Another approach was to search for an improved substrate with a surface more like denatured protein. The requirements for a STEM substrate are: minimum thickness (scattering), high strength, flatness & smoothness, electrical conductivity and controlled attachment properties. Evaporated carbon is close to ideal down to liquid N2 temperature in all but the last property. At liquid He temperature carbon becomes nearly an insulator, so we were attracted by the description of conducting polymer films by T. Skotheim (then at BNL). These hydrocarbon-pyrrole films could be formed in monolayers on an air-water interface and could be crosslinked by an oxidizing agent. We evaluated them as STEM substrates and found them to be nearly ideal: 27A thick, good conductors, flat over extended regions and very radiation resistant.(21) The only problem was that they shrank during drying and invariably broke when stretched over holes larger than 1000A. This could be overcome by longer polymerization, but only at the expense of increased thickness and severe lumpiness. TMV deposited on such films behaved as on carbon. When it became clear that development of these polymer films to a useful point was a full-time project, we opted to defer it until STEM3 was available. The filamentous virus projects of L. Day (NY Public Health Research Lab) have been among the most controversial. The collaboration between Wall and Day dates back to the beginning of the STEM project. Based on biochemical, biophysical and STEM studies, as well as extensive modeling, Day has proposed a model for the structure of Pf1 which has the DNA phosphates along the axis and the bases pointed outward.46,47 The untimely death of S. Reisberg, Day's graduate student for much of this work, limited the number of STEM publications, but many of the concepts were based on Reisberg's STEM data. Other types of filamentous viruses have more normal structures and almost all have been characterized with the STEM as to length, M/L and radial density profile. Many other virus projects have been active over the years54. One interesting result not presented there is the observation that Sindbis nucleocapsid has T = 4 symmetry, rather than T = 3 listed in text books.48 Similarly, both T749 and T481 tail fibers were found to be trimers instead of the dimers quoted in texts. Many of the virus projects concerned assembly intermediates and mechanisms.77, 79, 80,86 Several large invertebrate protein complexes have considerable historical significance with many interesting questions remaining. These are particularly useful as controls for specimen preparation, STEM imaging and image analysis. S. Vinogradov (Wayne State) and collaborators have studied a wide variety of giant hemoglobins50, 85, 87, 89, 91 and M. Hamilton (Fordham) has investigated a wide variety of hemocyanin structures.51 Chaperone proteins and heat shock proteins became a lively STEM topic several years ago when A. Horwich (Yale) began his studies on GroEL and similar proteins. The success of J. Trent (Yale, now ANL) on TF5552 led to a series of collaborations to establish the relationship between most of the known chaperonins. The Horwich lab showed that the protein being folded was located in the central cavity of GroEL53 and went on to obtain the 3-D structure of GroEL by x-ray crystallography. The Lindquist lab studied Hsp 10461. John Flanagan’s work on ClpP68 led to his solving the x-ray structure shortly thereafter. A few projects have had success with membrane proteins in the STEM. Gina Sosinsky was able to show that different types of connexins could be distinguished by mass in gap junction membranes.76 Collier’s group was able to study pore formation with anthrax toxin.59 In all these projects we have demonstrated a level of impartiality and dedication which has attracted projects and collaborations from all quarters. We have worked with high school teachers and leaders in many fields with equal mutual respect and motivation to find answers. In answer to the complaint that we do not have any strong in-group biological projects, we point to our track record of finding the best available collaborators in a timely fashion. Our approach eliminates any aura of competition or feeling that anyone's ideas or results may be diluted or even co-opted. Our successes are their successes.