Portal de Periódicos da CAPES

Michael G. Rossmann (1930–2019): Leadership in structural biology for 60 years

2019; Wiley; Volume: 28; Issue: 9 Linguagem: Inglês

10.1002/pro.3671

1469-896X

John E. Johnson,

Genetics, Bioinformatics, and Biomedical Research

Michael Rossmann passed away on May 14, 2019 at nearly 89 years of age following a courageous 5-year fight with cancer. During that time he maintained the arduous schedule of laboratory leadership, preparing grant proposals (he had four R01 NIH grants at the time of his death), traveling and publishing that characterized his previous 55 years in structural biology. Although there were many medical challenges associated with his illness few people were aware of the monumental effort that was required to maintain the "Rossmann persona" that we saw at conferences and gatherings during that time. He would not have appreciated that descriptor because it implies a pretense. His "persona" came from doing the best science possible in his unique style and demanding the same from others. Many graduate students, post docs, and principal investigators recall the penetrating questions (putting it nicely) that Michael would raise during presentations at conferences, graduate student committee meetings and seminars. His passion was maintained to the end. Three weeks before he died he was working on manuscripts and grant proposals and there were already six publications in print for 2019. The first word that comes to mind when colleagues talk about Michael is energy. Throughout his life he exuded mental and physical energy and a passion for getting things done. The second word is impatience. Things rarely happened fast enough for Michael because he could not imagine others lacking the same gifts, focus, and discipline that he had. This could lead to edginess in interactions with lab members, collaborators, and administrators and Michael frequently experienced frustration with those around him, myself included. The upside for those on the receiving end was motivation, accomplishments, and exceptionally good science; the downside was sleepless nights when you were not living up to expectation. Meetings of lab alumni and collaborators often result in sharing "special moments" of dealing with these aspects of Michael G. Rossmann. The other side of Michael contrasts dramatically with the description earlier. He could be warm, considerate, and compassionate with colleagues, providing insightful, nonscientific, advice, as I experienced during difficult times in my personal and professional life. There is widespread respect from those that worked with Michael or knew him well and gratitude for how he shaped our professional lives. Forty years after leaving his lab his opinion of my choice and execution of projects was still very important to me. What follows is a commentary, mostly in chronological order, on some of Michael's contributions to structural biology most of which are easily found by going to PubMed and following the 446 (of over 600 total) publications listed there. I finish with some personal remembrances of times with Michael. Michael's contributions to structural biology span 60 years and started when he joined the Medical Research Council (MRC) laboratory of Max Perutz in 1958. He made pivotal contributions to the structure determination of hemoglobin1 and, with David Blow, developed the practical approaches that lead to the widespread use of molecular replacement (MR). The 1962 rotation function paper2 was prescient in describing the use of noncrystallographic symmetry (NCS) applied to virus structure determination, well before high resolution crystallographic data were collected from virus crystals. Rossmann moved from the MRC to the Department of Biological Sciences at Purdue University in 1964. Henry Koefler, the head of Biosciences at Purdue, was determined to build a molecular biology department and recruiting Rossmann was a high priority. The laboratory was running in a short time and produced the structure of dogfish Lactate Dehydrogenase (LDH) in 1970,3 one of the largest, early protein structures determined. Ironically the subunits formed a functional enzyme with 222 symmetry, but the symmetry site was on a crystallographic 222 site, so NCS could not be used in the structure determination. Three years later Rossmann's group determined the structure of lobster Glyceraldehyde-3-Phosphate Dehydrogenase (GPD),4 employing NCS 222 symmetry locate heavy atoms and to improve the multiple isomorphous replacement (MIR) phases. Rossmann immediately recognized the conserved nucleotide-binding site in GPD when compared to LDH (I recall arriving at the lab the morning after the team had spent the night tracing the GPD chain and seeing "GPD is LDH" on the blackboard) (Figure 1). The paper describing the Rossmann fold and its implications for molecular evolution was published in 1974.5 Inspired by this similarity he developed methods to superimpose atomic coordinates of different protein structures and, together with Pat Argos, contributed important examples that lead to the current view that a limited number of protein folds elaborate into the broad spectrum of protein function.6 Following his success with protein structure determination Rossmann turned to his true passion, virus structure. He spent a sabbatical in the laboratory of Bror Strandberg in Uppsala, Sweden in 1971 to work on the structure of satellite tobacco necrosis virus resulting in a controversial paper in the 1972 Cold Spring Harbor Proceedings.7 The rotation function indicated octahedral symmetry for the particle, but Caspar and Klug (inventors of quasi-equivalence theory) believed that it had to be icosahedral. In the post-meeting discussion printed as part of that paper it was shown by Klug and Caspar that the octahedral peaks were due to the packing of particles in the crystal and that the particles did have icosahedral symmetry. This was an important lesson because these "packing peaks" would arise again in the structural study of southern bean mosaic virus (SBMV), the virus studied by Rossmann from 1972 to publication in 1980. I joined the Rossmann lab in September 1972 and worked on the SBMV project from its beginning to the publication of the near-atomic resolution structure. I was a post doc from 1972 to 1977 and an assistant professor from 1978 to 1980, splitting my time between SBMV and projects in my own lab. SBMV crystallized in space group R32 (a = 318 å, α = 64 deg; 1 particle per unit cell positioned on a 32 symmetry site) with 10 of the 60-icosahedral asymmetric units in the crystallographic asymmetric unit. The rhombohedral crystal axes passed directly through the crystallographically aligned particle fivefold axes and precession photographs with the cell axis coincident with the X-ray beam had fivefold symmetry in the transform that was striking. The 10-fold NCS was critical for finding heavy atom positions in the four isomorphous derivatives prepared, refining their positions and improving the MIR phases by averaging the electron density over the NCS. The low resolution data (11 å) were collected with precession photography and processed with existing software,8 but the 2.8 å resolution data required oscillation photography and a major effort of Rossmann in this time period was developing programs for processing these images.9 Virtually every aspect of the SBMV project required the development of original software for NCS manipulations (the R32 space group had its own challenges) and data processing and adaptations of existing software for Fourier calculations. Reflecting on the sustained effort this project required over an 8-year period, it is clear to me that Rossmann's extraordinary focus and iron kept the project on track through some exceptionally difficult times. The final stages of the project were quite remarkable. Rossmann and Andrew Leslie were able to trace the chains of the three subunits in the T = 3 icosahedral asymmetric unit with the density plotted on Plexiglas sheets. As they followed the fold they realized that the topology (the viral jellyroll) of the subunit shell-forming domain of SBMV was identical to the topology of the shell-forming domain of tomato bushy stunt virus published by the Harrison group in 1978.10 This was totally unexpected at the time and was reminiscent of the discovery of the Rossmann-fold in 1974.11 After tracing the chain, a model of one subunit was built in a Richard's box. Celle Abad Zapatero and Rossmann were building the model with polypeptide sequences chemically determined by Mark Hermodson in the biochemistry department at Purdue. Rossmann would search for a region of density that had side chains corresponding to the chemical sequence. When he found it, he was so excited that he would accidentally crash his head into the top of the Richard's box in which he was working, with obvious consequences. This behavior required the purchase of a crash helmet to avoid further injury (Figure 2). Michael saw his next challenge as an animal virus and picornaviruses were an obvious choice because they had a size comparable to SBMV and an enormous amount of literature on their function. The problem was that it was much harder to produce milligram quantities of an animal virus in 1980 than for a plant virus. I met Roland Rueckert (University of Wisconsin) at a NATO meeting in 1978 and learned of his interest in the function and immunology of picornaviruses as well as his enthusiasm for virus structure. I introduced Michael and Roland at the International Virology congress in Strasbourg, France in 1981. They agreed to collaborate on the structural study of human rhinovirus 14 (HRV14). Later in the fall of 1981, Michael, John Erickson, and I drove to Madison and the details of the collaboration were worked out. John spent time in Roland's lab learning to make rhinovirus and before long the virus production was up and running at Purdue. Crystals of HRV14 were produced shortly after with a cubic space group P213 (a = 445 å) and four particles per unit cell. The particles were on a crystallographic threefold axis giving rise to 20-fold NCS that was critical for the structure determination. Eddy Arnold and Gerrit Vriend joined the team in 1982 (Figure 3) and the project moved with remarkable speed, resulting in a near-atomic resolution structure in 1985.12 In addition to the exceptionally good skills of the team, two technology factors accelerated progress. The first was synchrotron X-radiation and the second was a leap in computing performance. HRV14 data were collected primarily at the Cornell High Energy Synchrotron Sources (CHESS) on the beam line developed by Keith Moffat and colleagues. Patterns requiring 24 hr to record on a rotating anode X-ray generator were recorded in a few minutes at CHESS. Typical CHESS trips would be 4 days, 24 hr a day with at least 6 people to have 3 on a shift. All of the data were collected on film, so one person was in the darkroom, one mounting crystals and one at the hutch making the exposures. In four days, data previously requiring a year to collect were obtained. Because the HRV14 crystals were small, radiation damage was a problem, so Michael decided that aligning the crystals wasted too much time. He developed the American Method of data collection (Shoot first, ask questions later) that was important to the success of many projects in the lab.13 Michael was a strong advocate of centralized computing at Purdue and he was a critical factor in the decision of Purdue's leadership to purchase a CYBER 205 "super computer" in the early 80s at a cost of 2 million dollars. This computer was transformational. First it allowed efficient processing of the HRV14 diffraction patterns that were almost intractable to process using the previous hardware. Second, it had a huge impact on molecular averaging. It took 6 weeks to do a cycle of averaging and Fourier transformation (FT) on the Control Data Corporation 6500/6600 computers used for SBMV. After "vectorizing" all of the code for electron density averaging and FT we could do multiple cycles of MR in a day for HRV14 with the Cyber 205. It was particularly important because Michael felt that there was a good chance that "phase extension" would work and this required many cycles of MR. Indeed, it was possible to extend the resolution of the structure from 5 å to 3 å by taking steps of ∼1 reciprocal lattice unit (20 steps in all). This made it unnecessary to get derivatives that diffracted to high resolution. The density was spectacular! The structure had enormous impact. First the viral jelly roll was found again in each of the three different gene products in the icosahedral asymmetric unit and their quaternary organization was immediately recognizable as pseudo T = 3 quasi-symmetry. The likelihood of evolution from a capsid with T = 3 quasi-symmetry seen in SBMV to pseudo T = 3 symmetry formed by proteins with the same topology was compelling. The structure displayed a "canyon" around the pentamer axes that Michael proposed was the "protected site" for receptor binding. It also showed the likely binding site for a drug previously shown to inhibit picornavirus infections and that the site was occupied by a lipid-like molecule in native virus particles. Identification of this site would lead to a classic example of structure-based drug design carried out in collaboration with Sterling Winthrop Pharmaceuticals.14 Shortly after the structure was solved Roland Rueckert and Barbara Sherry came to Purdue with their monoclonal antibody escape mutant data. They had 35 mouse monoclonal antibodies and mapped 62 sites in the HRV14 subunit sequences that mutated to allow virus to grow in the presence of the antibody. When these sites were plotted on the HRV14 structure they clustered into four regions that defined the major antibody binding sites to HRV14 (and other picornaviruses).15 I recall seeing Michael taking a break in the hallway while they were plotting these mutant sites and he seemed uncharacteristically quiet. I asked if there was anything wrong and he sighed and said, "I am not sure I can handle this much pleasure" and then he returned to the work to find out! Later, with Tim Baker and his group, Michael would turn to electron cryomicroscopy (cryoEM) for the first time to visualize the ICAM1 receptor attached to HRV14, showing that it did bind in the canyon.16 There were numerous follow-up studies on picornaviruses that provided the structural understanding for a broad array of functions. Following the HRV14 structure determination, high resolution studies were performed on other picornavirues,17, 18 the ssDNA parvoviruses,19 and ssDNA bacteriophage Phi X 174 in collaboration with Nino Incardona.20 All of these investigations lead to a myriad of studies that demonstrated evolutionary relationships among viruses as well mechanistic functional studies that revealed unexpected modes of particle entry (Phi X 174) and immunological and receptor binding determinants (Parvoviruses). In the same time frame, the first work on alphaviruses was done in the Rossmann lab with the crystal structure determination of the purified core capsid protein that formed the nucleoprotein particle within a lipid bilayer of Sindbis virus. The protein showed a chymotrypsin-like fold21 that functioned as a "one cut" active enzyme because each subunit would undergo an autocleavage during maturation. While the size of membrane containing alphavirus particles was appropriate for crystallography (∼500 å), crystals could not be produced that diffracted X-rays to high resolution. This problem pushed Rossmann and Baker to blend the near-atomic resolution X-ray crystal structure of the sindbisvirus core protein with moderate resolution (∼25 å) cryoEM structures of the closely related Ross River virus particle.22 This required quantitative fitting of X-ray derived models to moderate resolution cryoEM density and Michael developed software to achieve this.23 These tools allowed an excellent fit of the core protein model to density inside the membrane and showed that the interior core had T = 4 quasi-symmetry as well as the outer glycoprotein distribution of the E1/E2 spikes on the surface. By this time Richard Kuhn (Figure 4) had joined the faculty at Purdue and provided critical virus resources for the structural work and a functional context for the alpha and later flavi virus studies. Eventually most of the near-atomic structure of sindbis virus was determined with an 11 å cryoEM map, the crystal structures of the core protein and E1 glycoprotein determined by Felix Rey and colleagues24 and the conclusion that the E2 glycoprotein probably had the same general structure as E1 based on difference cryoEM electron scattering maps with the crystal structures subtracted.25 The cryoEM structure of Dengue flavivirus was determined at 25 å resolution by cryoEM and the crystal structure of the E protein of Tick Borne Encephalitis, previously determined by Rey and Harrison,26 was successfully modeled into the density to reveal a novel distribution of subunits on the particle surface.27 This "herring bone" distribution of E proteins, with three in the icosahedral asymmetric unit, does not obey Caspar and Klug T = 3 quasi-symmetry and is the organization of all other flaviviruses studied to date. Rossmann, Kuhn and colleagues carried out many studies of alpha and flaviviruses relating to their assembly, maturation, and immunological properties with numerous publications in these areas continuing into the last years of Michael's work (Figure 5). Structures at near-atomic resolution of isolated alpha and flaviviruses, in complex with neutralizing antibodies and as asymmetric reconstructions were determined by cryoEM alone following the availability of the direct electron detector.28-30 Michael's interest in dsDNA bacteriophage started in the mid-90s, collaborating with Dwight Anderson (University of Minnesota) and Tim Baker on a series of studies of Phi 29 that continued for more than a decade. Their first paper described the entire prolate head and associated tail machine at moderate resolution, providing a description of the quasi-symmetry associated with the head as well as the symmetry mismatch at the point of interaction between the 12-fold symmetric portal and pentavalent site of insertion in the head.31 Methods were refined for dealing with symmetry mismatches allowing the interactions to be viewed with greater detail as well as providing a general approach to asymmetric reconstructions of icosahedral particles.32 They determined the structure of the portal by crystallography33 as well as a portion of the ATPase dsDNA packaging motor.34 They went on to describe structures of intermediates in assembly and maturation that Anderson had made with clever mutations in the virus.35-37 A second long-standing collaboration on dsDNA bacteriophage T4 was started around 2000 with Venigalla Rao at Catholic University in Washington DC. Like Anderson, Rao and colleagues had created a variety of intermediates or substructures of the virus that could be studied by crystallography and high resolution cryoEM although their first publication tackled the entire virus.38 In this project Rossmann had fully engaged cryoEM in his own group and was personally making use of the EM facilities at Purdue.39-42 To conclude I want to touch on some nonscientific aspects of Michael's life. He was exceptionally well-read and enjoyed engaging in conversations of current affairs. It always baffled me how he could have such broad knowledge when he seemed to be doing science all of the time. He loved the outdoors, hiking, camping, and skiing (which he did well into his 80s). His hiking pace was legendary at Gordon Conferences and other meetings where he would exhaust graduate students and post docs a small fraction of his age. About the time I came to Purdue, Michael was developing his interest in sailing. I had spent my summers in college as a paid hand on a 60-foot sailboat in Lake Michigan. Given my experience with sailing, Michael invited me to join him on a few occasions. These were decidedly unsuccessful outings. The first was to a reservoir near Peru, Indiana where we capsized the boat and lost the centerboard before the race started. I can still visualize Michael holding tight to the mainsheet shouting, "here we go" as the boat went over. The second time we arrived at the race site and we were the only ones there from the sailing club. Unbeknownst to us, the race had been canceled due to the gasoline shortage in the early 70s. We decided to sail for fun and things were going well with a broad reach when Michael suddenly wanted to sail closer to the wind. It was apparent that he had his eyes on a boat in front of us and we were closing relatively rapidly. Michael seemed very satisfied. I pointed out that they did not know they were racing and Michael snapped, "makes no difference." The third race was in Monticello, Indiana. The wind was nonexistent at the start time. There was some bumping of the boats and some accused us of sculling (Michael was at the tiller)! Finally after 30 min of absolutely no wind, they canceled the race. Michael looked me in the eye and said "this has never happened before" and I had the very definite impression that he felt it was my fault that we had no wind! I was not invited to join him anymore given my dismal record, however, Michael went on to hone his sailing skills and became a legend of M16-scow sailing in Indiana, winning the state championship in his class multiple times. Last, but certainly not least, Michael loved his family and they loved him. He and Audrey were married for 55 years when she died in 2009. They had three children (Martin, Alice, and Heather) and four grandchildren. Audrey was a renowned potter and artist in Indiana and Michael often accompanied her to art festivals, providing the muscle for moving things around. Audrey's plates and bowls were remarkably heavy! He thoroughly enjoyed the art-oriented crowd and they enjoyed his science. Audrey produced many items that incorporated icosahedra and these were often going away gifts when people left the lab. I have a banner in my office with a one-dimensional lattice of icosahedra from top to bottom screen-printed by Audrey. In 2015 Michael met Karen Bogan when he moved into a retirement center in West Lafayette in order to manage his cancer treatments. Karen was in charge of welcoming new guests and hospitality. She took her job seriously as they were married on May 1, 2018. They were soul mates almost from the time they met to the end and enjoyed their time together to the fullest. My wife Mary and I visited Michael and Karen in their apartment in February 2019 when we were in West Lafayette. It was clear that Michael was quite weak. More tumors had been discovered and he was contemplating some more surgery. At the end of our visit Michael and I had an exceptionally warm embrace and it was emotional to thank him for over 40 years of mentorship, collegiality, and friendship. His loss is more difficult than I expected.