Adventures in Clinical Chemistry and Proteomics: A Personal Account

Abstract

My 90 years have witnessed a basic transformation in the understanding of disease in terms of molecules, largely through the application of new instruments and technologies. The ultimate distillation of what really works at this level—the quantitative measurements that generate clinical insight from specimens like blood—is clinical chemistry. This field has fascinated me for a long time, partly because of my interest in inventing or improving analytical instruments, and partly as an anchor to real-world biology that is frequently missing in academic research. A second thread of interest to me is how successful research gets done, and how to know when a solitary inventor is needed and when it takes an army. Here I recount some personal experiences relevant to these interests, ranging across several fields and in organizations of widely varying scale, all ultimately linked to clinical chemistry and the human proteome. Interdisciplinary R&D has always fascinated me, and my introduction to it occurred in unusual times, during World War II. I was on active duty in the US Navy before Pearl Harbor as a Photographer’s Mate 2nd Class, and was discharged at the war’s end as a Lieutenant (jg) line officer, with zero instruction in between on how to be a naval officer. Despite (or because of) this fortuitous absence of formal tuition, I found that much of the fun and adventure in life lies in the cracks between disciplines, and that these cracks can be wider in large organizations (like a Navy in wartime) than smaller ones. Flying in blimps off the Carolina coast during the height of antisubmarine warfare, it occurred to me that maybe, lacking a bombsight, we couldn’t actually sink a German submarine if we found it. After developing proper instrumentation, I found experimentally this was largely true, and a proper bombsight was developed. This was the start of a series of projects that put together all sorts of technologies, raised interesting questions, and whose results were usually translated into immediate action. Transferred to the Pacific and the submarine service, I worked as a movie photographer on a project to be called “The Silent Service.” This was authorized by a personal letter from Franklin Delano Roosevelt, which proved to be a magical passport to getting things done far from home. As I was shooting background footage of 2 submarines I had arranged to do the required postrefit maneuvers, a radioman came topside to say that Truman had announced use of the atomic bomb. This ended the war and with it my introduction to interdisciplinary work with effectively unlimited resources. Suddenly I found myself at Duke University immersed in the culture of Little Science. I was taught (by a future president of the National Academy of Sciences) that proteins and nucleic acids were too complex to ever be sequenced, that chromatography, while interesting, could never be quantitative, and that no one knew for certain where and how genetic information was stored. The general attitude was very different from the “win at all costs” approach adopted in war—it was painstaking and slow, but it was biology. I began to realize I had been contaminated by the notion of Big Science, but felt I should learn to be comfortable at both ends of the Big Science–Little Science spectrum (1). This pendulum has swung back and forth for me several times, and is an invigorating oscillation. Returning to the Big end, I obtained an Atomic Energy Commission (AEC)1 postdoctoral fellowship in the Biology Division of the Oak Ridge National Laboratory (ORNL). My PhD thesis had concerned subcellular components isolated using very simple centrifuges, and my hope at Oak Ridge was to extend this work to proteins in different subcellular particles using some new type of centrifuges, yet to be conceived. ORNL’s unprecedented facilities, with staffs running into the tens of thousands, included almost all disciplines of science and engineering. Almost anything one could reasonably imagine was either available or could be designed and built quickly, even if it happened to involve nonstandard laboratory supplies like large titanium forgings. The saying, “Why use lead when gold will do?” reflects a little of the flavor. Separation, either physical (as in the case of uranium isotopes) or chemical (as was the case for plutonium), and accurate analysis were the key technologies at most of the Manhattan Project facilities. My initial laboratories were in the same valley that housed more than a thousand giant Calutrons (preparative mass spectrometers) used to enrich kilograms of U-235. After World War II, this facility was used to go straight through the atomic table, isolating and characterizing all the stable isotopes. I wondered if the same sort of effort and philosophy could be adapted to the comfortable field of biology? Could one ever separate the components of living cells into a “parts list” for man? If so, it should provide a powerful way to study and ultimately understand disease. As it happened, the major nuclear weapons laboratories needed new missions after the success of the Manhattan Project. I suggested one in the winter of 1959–60 entitled “The Cell Fractionation Project,” an effort to separate and characterize all the molecules in cells, which much later became the Molecular Anatomy Program. It appealed to nearly everyone at ORNL except my fellow biologists, who did not like big projects (unless it was mouse genetics). We had thought about sequencing DNA but were assured by biochemists that, while RNA could in theory be sequenced, DNA simply could not be for purely chemical reasons (this was before the discovery of restriction enzymes or dideoxy sequencing). So the thinking focused on proteins. Protein fractionation had been advancing on multiple fronts during the preceding decades. In the 1930s and ’40s, Svedberg had developed the analytical ultracentrifuge which showed, unexpectedly, that proteins had well-defined masses, and Tiselius, who once described to me how he had inadvertently left his gardening shoes on when he went to hand out Nobel Prizes, had developed electrophoresis by which plasma proteins could be classified into 4 discrete groups (albumin and the famous α, β, and γ globulins). By the mid-1950s, Sober and Peterson had begun to fractionate proteins on cellulose columns, and Waldo Cohn, who had pioneered separating fission products on ion-exchange columns at Oak Ridge, began to work on nucleic acids, convincing Moore and Stein to use ion exchange in place of starch columns for amino acid analysis. Precipitation was explored in parallel by Gerhard Schwick at the Behring Institute in Germany. He isolated dozens of human plasma proteins, made antibodies to them, and distributed these worldwide. This approach with distributable reagents allowed specific protein assays to be performed on clinical samples, thus starting immunodiagnostics on the present road to broad coverage of the human proteome. While largely forgotten in the field of proteomics, this effort has survived through multiple commercial marriages with Hoechst, then Dade Behring, and finally Siemens Diagnostics. My own work really began with the invention of the zonal centrifuge (2) to fractionate subcellular particles. In this device, the volume limitation inherent in swinging bucket gradient separations was surmounted by using large, hollow, bowl-shaped (zonal) rotors. In these, gradients and samples were caused to flow through rotating seals into a rotor spinning at low speed and then accelerated to maximum speed to effect a separation based on either sedimentation rate or isopycnic banding density (or, in later designs, both). This was followed by deceleration to a low speed and recovery of the gradient as isolated fractions by displacement from either the center or the edge. I had designed and built a slow and crude proof-of-principle zonal rotor and then had arranged to have one built commercially, which was unfortunately unstable at high speed. Instability of a large rotor at 40 000 rpm, especially if it leads to catastrophic self-disassembly (a phrase we adopted from Los Alamos, which knew about such things) is undesirable. We needed real engineering expertise in rotating systems, an unusual discipline but one that was by chance very popular at Oak Ridge. Gas centrifugation for uranium enrichment had been tried and abandoned in 1943 because of its high cost. Subsequently it was discovered that a captured German Luftwaffe engineer named Guernot Zippe had designed for the Russians a remarkably simple centrifuge that used very little power and was surprisingly efficient. The need to catch up with this development accounted for the presence of an engineering staff working at top speed (in all meanings of the phrase) in Oak Ridge. The resulting urgency, money, and minimal administration helped as usual to eliminate the curse of delayed gratification, chief destroyer of creativity. We built (and sometimes blew up) a lot of centrifuges, and they became progressively better at separating biological materials. In the early ’60s, Robert Huebner of the National Institute for Allergy and Infectious Diseases and others found that many animal cancers were caused by viruses, especially if the viruses were given to newborns. Numerous groups were set up across the US to attempt to isolate cancer viruses, grow them in culture, test them in primates, and see if a cancer vaccine was possible. When these efforts failed to find culturable human cancer viruses, I suggested to Huebner that we try to isolate them by physical means, using density gradient centrifugation, instead of relying on growth in culture. If this were successful, then similar physical methods could be used for large-scale purification of virus for a vaccine. The US Food and Drug Administration (FDA) was insistent that any killed virus vaccine should contain no (or at least very little) cancer cell DNA to be sure that the vaccine itself did not cause cancer. To make a pure virus vaccine for large-scale human use by physical means would require a liquid centrifuge of a size never before built. Testing these systems required large quantities of virus, and neither Sabin nor Salk, who were very cooperative, had poliovirus in the quantities we needed (milligrams rather than infectious doses). Initially we settled on seawater obtained from the Woods Hole laboratory and discovered to our surprise that the ocean has about the same viral load as a viremic human’s blood (3). For more realistic development, though, we obtained a batch of human viral vaccine that did not meet FDA standards and thus could not be sold. To avoid risk of viral contamination to ORNL’s enormous mouse genetics facility, we relocated the centrifuge development program to the most distant site available on the Oak Ridge reservation, which was, fortunately, right next to the giant Oak Ridge Gaseous Diffusion Plant, locus of the gas centrifuge project. Our “lab” was a mothballed power plant, whose Manhattan Project pedigree was visible on the wall as a framed single-page purchase order for “One coal-fired steam-driven electrical generating plant, 237 megawatt.” It had railroad tracks coming in one end of the 100-yard long main floor and a 30-ton overhead crane for moving large equipment, among other conveniences. We needed a general theory on which to base our search for viruses in tissue homogenates. To see the possibilities of such a separation, I plotted the sedimentation coefficient S against the banding density ρ for viruses and for the major subcellular particles and discovered that viruses generally are found in the middle of this plot in an otherwise thinly populated area away from nuclei, mitochondria, proteins, etc. (4). This plot was key to the whole project, and it suggested that we combine sequentially rate and banding techniques into one 2-dimensional (2D) S–ρ separation. This theoretical plot was converted into a real one in which bacteriophage were recovered from rat liver and other tissue homogenates (5), perhaps the first integrated high-resolution 2D separation in biology. As it became clear that no cancer viruses were being found around which to design a vaccine purification system, I decided that we should work on an existing vaccine that required better purification. We would thus be ready if a human cancer virus was actually found. At that time, egg-grown influenza vaccines contained appreciable amounts of egg proteins, resulting in many deaths from anaphylactic shock each year and the requirement that they be given under close medical supervision. We approached Eli Lilly about designing a centrifugal system specifically to purify influenza vaccine. Their batch size was 100 L, and the purification run had to be completed in an 8-h day. Knowing these parameters and both the sedimentation coefficient and banding density of influenza, it was possible to design a rotor system that used continuous flow to band the virus from 100-L batches in a narrow gradient that could be recovered at the end of a run. The result was the K-II continuous-sample-flow-with-banding ultracentrifuge (6). Use of this centrifuge essentially eliminated vaccination deaths from anaphylactic shock and allowed vaccination in supermarkets under minimal supervision. Almost 40 years later, it is still in use around the world with minimal modifications for vaccine manufacture, and we have recently proposed its use to isolate the viral load from 100-L batches of pooled diagnostic serum discarded in clinical reference laboratories each week (7). The viral DNA and RNA, concentrated and free of host nucleic acids, could then be shotgun-sequenced to screen for new viruses, while providing a running index of the known viruses “going around.” separations of cell components many To specific across these we used from clinical chemistry. I once to had most of the that be some other way to clinical chemistry. He that this was not possible. I thought about this a It was my introduction to clinical and clinical chemistry. The was to a system for between samples and reagents in parallel rather than It out that centrifugal is an way to and liquid while at the of a rotor spinning a provide measurements that very accurate The rate was in that we needed to a like the to it. these in the early was given ORNL’s but we that a a caused no The resulting was named the Energy centrifugal It was a commercial success for and and in many it did to the and it still be the system for very accurate The the rotor of an early centrifugal system At right is a of the system used to and measurements from at the of the spinning in during one of the Despite the success of zonal and the centrifugal the National that human cancer was to viral with interest in me to to the University of I was to be in by my who had completed a PhD at the University of under Nobel and done a with had famous on 2D electrophoresis and had the we set up a laboratory and a research to the most we ever had worked out a which was, of a In a system and the major plasma proteins by with the whole of Behring The of plasma proteins, called was many and Protein on were clear We found the 2D of plasma proteins to be and to an but it was more to a than a clinical 2D plasma and serum from the same The and the and on with This was in of plasma proteins, The plasma and genetic 2nd It became clear that to 2D we needed once the of a National and this with an to the biology at For several we worked in during the week and at on designing together what we called the system for and running large of 2D in parallel Our initial analysis system was an designed to 2 by between them, a used by to the This was by an and large for these we explored the protein called of human as as rat liver and many other We were to host the first 2 major on 2D first at and the second at the both as of (in and The results at these 2D are in some surprisingly similar to the of with the of protein using mass perhaps was before DNA we felt that of all the human proteins by cell fractionation and 2D electrophoresis was the way to in biology the effect that the had in chemistry. This was as the Protein at providing a for and ultimately what is systems biology. an effort would require large and so, with several we suggested of a Protein the general we had in Oak Ridge, to and this who was the of the US at the time, was in research with on and on in his in the on these a Protein was and in a was out the and size of a on the human and a new of much to and more to the the National Laboratory it that study of proteins was to a in the that time, the in biology to the Big Science approach of the National We left in and set up Biology to 2D and protein index and the the years a with in protein and finally a successful initial in the year an 2D electrophoresis system running 100 we explored in rat the of and to the of human a approach we had developed at Oak Ridge we the first columns that the plasma proteins used as the of the system columns with fractionation the of 2D from the most plasma proteins to more than The of mass for protein allowed finally to all the or we had in of and analysis quantitative of in specific protein this I to a in centrifugal systems, developing a large-scale centrifugal for and a centrifuge for viruses from clinical samples, banding them or them in to a plasma This has to and concentrated viruses from serum in about 2 in quantities that the of thus the way and sequencing of human viral it is a little for my own the broad of a approach to understanding the human and it for has recently begun to To start the of a human has the means to the proteins, and perhaps most that are really about of them rather than the 100 000 we were once to the and to on a protein of each this like a to at a or several large-scale are with an effort to the of all the proteins. because is really this is being done at Big Science like the at (in a large The resulting should provide a for understanding and thus the of cell and In and are of a project antibodies to each human and then to see where these proteins are in and success in this a broad for major clinical in will be the to be protein real with clinical to be into A new of mass for is that can ultimately in terms of and while and this it possible to specific assays for proteins starting from a and project to quantitative, and specific assays for all human proteins a of assays in the present protein and even into the clinical laboratory mass is for better of and It to me that this of project, up basic clinical research and clinical chemistry at the same time, is even more than the human and for a of Big Science thinking in the protein If all this to it will a in clinical it at the of biological and at the of clinical would be Atomic Energy Oak Ridge National US Food and Drug Energy Protein Biology initial and they have to the of this and have the to the and of or analysis and of or the for and of the of of any of of The organizations no in the design of of and of or or of I in to the of I have not and I my many and for at Oak Ridge, and through the especially of which extend through his