9th Greek Australian Legal and Medical Conference
PURITY AND THE PURSUIT OF EXCELLENCE
Professor Donald Metcalf, AC
In medical research, as in athletics, excellence is an accolade that is awarded by others: it cannot be self-conferred. One may strive, as do Olympic athletes, but it is others who must decide whether efforts made have achieved excellence - although improvement of patient health can be an alternative reward if the accolade of excellence has escaped us or is inappropriate.
Excellence in Medical Research
The public view of medical research workers is hopefully that they are intelligent, hard-working, sober and responsible people. While many may have these qualities, to be judged excellent implies something special.
At the start of their careers, most medical research workers are confident that they will tackle and solve a major disease problem. Most quickly learn the unreality of such aspirations and modify their goals to ones that are less ambitious and more attainable. These become the small advances that lead to progress in a field - typically progress made on the shoulders of predecessors. Because of this, excellence in medical research is defined by the multiple parameters of success in making repeated substantial discoveries that surpass the "shoulder-progress" level; success in exploiting these, perhaps with clinical application; and, finally, by peer recognition of superiority as evidenced by election to learned academies, the award of prizes or honours. Here excellence is being defined perhaps as being the best in a particular international field or being among the best 50-100 scientists in all fields in the country. However, these are everyday parameters of superiority that are still far short of the Mozart class of excellence. If the name of a particular research worker becomes retained for posterity, this might suggest a higher order of excellence, but such an event is quite arbitrary and is now quite unlikely to occur because of the necessity for teamwork in current medical research.
What is known by very few outsiders is that the pursuit of excellence in medical research is not an activity for the faint-hearted. Medical research is not a benign, contemplative, exercise carried out in ivory towers. It is arguably the most competitive of all occupations, and certainly one in which few prisoners are taken. No research worker can afford to be the second to discover something. If this becomes a pattern for you, quite quickly you will have no research funds and shortly afterwards no job. Success, and evidence of this from peer-reviewed publications and other peer acknowledgement, is necessary for obtaining grant money or promotion and indeed for a salaried position. Medical research must be one of the few occupations where you are paid a salary to work but must then go looking for money to allow you to work. Being human, the result of this is fairly predictable. Forget collegiality. You do not discuss work in progress with others. You may well find that the publication of your data is held up or blocked by anonymous referees who then make use of your data. Similar anonymous referees may well block a competitor's promotion or grant funds. Paranoia? Not really. One becomes resigned to human frailties. The situation is more complex in that we cheerfully and carefully train our future competitors and even assist competing colleagues with reagents, at the same time as behaving in a ruthless competitive manner. Honesty is assumed always but is never absolute: self-deception and data selection or modifications are universal although actual data fabrication is rare.
Top-level medical research differs little from an Olympic event. There are winners and losers, ruthless determination is required and there are few friends when pounding down the track for the tape.
Is there a better way to achieve excellence in medical research without the competitive 'publish or perish' dimension? Alternatives have been tried in which scientists are appointed for life, are adequately funded and are not expected to publish until a study is complete. The Max Planck Institutes in Germany were a successful example of this idyllic operation. For a brief period, so was the John Curtin School of Medical Research in Canberra and the Medical Research Council Units in the UK. Most have fallen by the way or have had to alter. In most cases, comfort and absence of pressure did not lead to sustained productivity or excellence. Remarkably, the successful and excellent have always emerged out of slightly inadequate, highly competitive and rather stressful circumstances.
Purity and Excellence
Moving on to the main theme of my presentation, I put forward the proposition that purity is a form of excellence and that to strive to achieve purity is to strive for one type of excellence. Whether or not this proposition is accepted, this particular form of striving is what has occupied me for the past forty years - not the pursuit of personal purity but the pursuit of purifying the hormones controlling blood cells. The pursuit has required the recognition of some very strange aspects of the way in which the body is built and arranges its affairs. Purity has been revealed to be an unattainable goal but our pursuit of purity was nonetheless successful. Has this been a story of excellence or merely the pursuit of excellence? As always, the judgement is in the hands of others.
The Pursuit of Purity: Speculation
When I was young, I seemed to spend a lot of time wondering whether, with sufficient effort, the apparently impossible could actually be achieved. Could you run a hundred metres in five seconds if you really, really, tried and were superlatively fit? Could you play a round of golf with a succession of 18 hole-in-ones? And could six monkeys really type by random chance the complete works of Shakespeare without a single typographical error - the homely example once used by a distinguished astronomer to convey the notion of infinite time?
I think what set me off on this dubious pathway of compulsive speculation was a throwaway line (or quotation) in my high school chemistry textbook that, in every glass of water we drink, there are five (or was it 500?) molecules of water from the draft of hemlock taken by Socrates. Was it really possible that such a complete mixture of the water on earth could occur in two and a half thousand years of tides, storms and diffusion? Could the ultimate perfection of mixing be achieved, given time? Weighty matters for a no doubt atypical teenager to be speculating about.
It does not need a rocket scientist to recognise a rather direct connection between such compulsive speculations and my behaviour in the laboratory 30 years later in attempting, year after year, to achieve complete purity of the hormones controlling white blood cells. Compulsive thought patterns and behaviour, strangely enough, seem able to coexist with otherwise pragmatic and commonsense behaviour. The average brain seems able to achieve a working compromise between neuronal circuits concerned with the compulsive pursuit of the impossible and other circuits likely to contain, amongst other things, information that such pursuits are not achievable for a variety of cogent theoretical reasons.
I once had a colleague who was sorely tried by an inability to achieve a compromise between such neuronal circuits, to the point of becoming quite unproductive. He became determined to perform the perfect experiment; nothing short of perfection in the behaviour of all cells and reagents in the experimental and control groups was to be acceptable. Until then, not one further word was going to be published by him on the subject, in which he was actually a current leading worker. Needless to say, the perfect experiment was not achieved to his standards and his publication record over this period was dismal. I could never quite decide whether or not this was a source of satisfaction to him. He remained cheerful and otherwise apparently normal, although shortly thereafter he left research for a teaching career, perhaps confirming his essentially normal state.
The Pursuit of Purity: Reality
In the mid 1960's two groups, one in Rehovot, one in Melbourne, by accident discovered how to grow colonies of blood cells in culture dishes. This was a remarkable technical advance but one aspect in particular of these colonies intrigued me. The cell division necessary to form colonies of progeny cells did not occur spontaneously. Cell extracts had to be added to the cultures to stimulate cell division to occur and we gave the name colony-stimulating factor (CSF) to this unknown substance. We speculated that CSF might be the hormone controlling the behaviour of these blood cells in the body. If so, it was of importance to purify CSF and understand how CSF worked. We now know that CSF is a large glycoprotein molecule - a mixture of carbohydrate with a core of protein in the form of a large twisted chain of amino acid building blocks. When, in the late 1960's, we set out to purify from human urine what we then believed to be a single colony stimulating factor (CSF), using cultures of mouse bone marrow cells as the assay system to monitor purification, we were quite naive in matters of protein chemistry. At the time, there were few examples in the literature of the successful purification of a protein of this type that probably only existed in very low amounts and thus no real guidelines about the possible amounts of starting material that would be needed. Certainly, there was no appreciation that, with existing techniques, such a project was quite impossible in the 1960's. These matters were slowly to become apparent to us and others in the following years.
The English language can be delightfully ambiguous. Nowhere is this more evident than with the word "purification." The word can be used equally correctly to describe the most trivial of procedures such as removing cigarette butts from buckets of CSF-containing urine, or to describe the quintessence of purification to a stage at or beyond sequence grade purity. I am as guilty as most of making full use of this ambiguity in the language, with a succession of publications on the purification of CSFs. These papers were, in essence, progress reports on attempted purification, but all carried in the title the somewhat mischievous word "purification." It is of course understood that I am referring to events of thirty years ago. It is far less common now to see such publication titles for an incomplete story. The unfortunate author of today will be hard-pressed to find a journal willing to publish a description of a bona fide purification of a new factor, even with detailed information on the amino acid sequence of the molecule. Often the editorial response will be, not only to go back and clone the gene responsible, but also to establish the biological function of the gene product before again darkening the door of the journal.
The dilemma that was raised recurrently during our initially amateurish, then more professional, efforts to purify CSF from human urine was how to decide when we had successfully purified CSF? Without this endpoint, the project could not be declared completed. Moreover, as events were to prove to the discomfort of others, without the certainty that essentially all contaminating molecules had been excluded, one could not establish with confidence the apparent biological properties of the CSF. So a chronic uncertainty persisted in our laboratory regarding what was acceptable proof of purity, to the vexation of all.
How does one decide that something is pure? What were people using at the time as a definition of purity? I do not recall that any of my colleagues ever pointed out to me that, on theoretical grounds, absolute purity was impossible to attain. Perhaps I lacked colleagues able to convince me of what must have been deducible from available knowledge at the time. Few would now harbour the delusion that anything can be obtained in absolutely pure form, although I have yet to see this statement in black and white in relation to any biological material. Maybe there remain people who believe absolute purity is attainable.
As the technology for protein purification advanced in the 70's, so did acceptable criteria of purity. Nowadays, a sharply-defined active peak in fractions from the last of a sequence of purification steps would in practice probably be able to be shown to contain a largely homogeneous preparation of protein. The purity of the protein would be assessed by whether or not it was found to have a dominant unique sequence of amino acids as determined by amino acid sequencing or, today, by the Nobel prize-winning technique of mass spectroscopy.
Such material is now usually referred to as having been purified to "homogeneity." This material usually is not entirely pure because minor contaminants remain present. However, few would now persist with purification attempts to reduce or exclude such contaminants. Operationally, the purification project has achieved its intended objective of sufficient purification to obtain an unambiguous amino acid sequence and the action moves then to molecular biology, to isolate the gene encoding this amino acid sequence and the remainder of the polypeptide. Proof that the original bioactive material had in fact been adequately purified comes retrospectively with the documentation that the recombinant protein, generated by the cloned gene after being placed in bacteria or yeast again after "purification," has essentially the same biological properties as the original native material in its most purified form.
It should be noted in this sequence that at no stage were the active native molecules genuinely purified to the absolute exclusion of all other polypeptides and the same is true also of the recombinant material. The latter material can be produced in vastly larger amounts than the precious original native material and is technically much easier to purify. Despite shortcuts facilitating purification, nothing fundamentally has changed. Recombinant proteins are never purified to the absolute exclusion of all other macromolecules, nor can they ever be. Separative procedures ideally result in a peak of active material. Even with ideal methods for separating various proteins, although contaminating material will be segregated to adjacent fractions, the tails of such fractions extend infinitely to adjacent fractions. No matter how many separative procedures are used and how often they are repeated, some contamination from material with related properties will always persist in material under purification.
Does crystallisation of such a "purified" protein ensure ultimate purity? No, because contaminating proteins can be trapped in the crystal lattice or remain attached to the crystal surface. Furthermore, the water used in all solutions itself may contain some protein contaminants, or contaminating proteins may arrive in a preparation from reuse of an affinity column or from handling of glassware or plastic or even arrive as airborne dust or droplets. While many of these may be rather trivial, they also serve to make the absolute purification of a protein quite unattainable.
We also realised that, within milliseconds of adding purified CSF to target bone marrow cells in cultures, the marrow cells themselves would be secreting or shedding all manner of uncharacterized proteins. Furthermore, ten or twenty percent of bovine serum must be added to cultures of bone marrow cells to allow them to grow. This serum contains thousands of uncharacterized proteins. If the presence of other proteins, or possibly other regulatory molecules, was going to confound the interpretation of what was being observed, then the reductionist dream of a pure regulator acting on a pure population of target cells was nonsensical: there was no conceivable situation in which such a combination could be attained or maintained.
What Did Purification Achieve?
We were actually successful by the early 1980's in purifying CSF to homogeneity, or to be more accurate "CSFs", because eventually we recognised that there were four different, but related, CSFs. The project took 15 years to complete using various tissue sources for the CSFs and three generations of biochemists. Adequate purification of CSFs could not be achieved until the invention of high performance liquid chromatography in the late 1970's because the final purification required was up to 1 million-fold - only one CSF molecule being present for every one million contaminating molecules.
Success with purification gave us minute amounts of highly purified material, which we could add to bone marrow cultures to establish the action of each CSF. Much more importantly, amino acid sequence data could be obtained from the purified CSF that would then make it possible to isolate the gene encoding the CSF and then hopefully enable the mass-production of recombinant CSF in bacteria or yeast. By the early 1980's, a few examples of successful mass production of recombinant protein had been achieved, so mass production of CSFs might be possible. Certainly, the need for such a measure had become appallingly apparent. For example, our team of six working on the purification of murine G-CSF had finally been successful in 1983 after five years of work, the use of 250,000 mice to donate lung tissue and 250,000 bone marrow cultures for the required bioassays on various fractions from the purification procedures. We could calculate from our capacity to purify native G-CSF, using a hard-won efficient purification sequence, that it would probably take 250 years of purification by the team to produce enough G-CSF for injection into one patient for two weeks. Even with our capacity for unremitting effort, the calculations were appalling and the prospect not to be contemplated.
It says much for the giant strides made by molecular biology in the 1980's that we and others succeeded in finding (cloning) the genes for all four CSFs in the mouse and man in a three year period from 1983-86. Mass production of recombinant CSF was then achieved, not without some headaches, by inserting these cloned genes into suitable bacteria or yeast. All subsequent work in the laboratory and clinic has used purified recombinant CSFs.
Moving Beyond Purification
The use of simple bone marrow cultures stimulated by purified CSF are necessary to establish basic facts but of course these are no longer representative of the actual complex situation in the body. Bone marrow cells do not exist in isolation in the body and their behaviour is likely to be significantly influenced by the behaviour of their cellular neighbours and possibly by the behaviour of cells in remote locations in the body. Furthermore, the fluid surrounding the cell will contain hundreds of potentially biologically-active molecules many of which might have contrary or cooperative actions on the responding cells. What reductionist biology and purification can achieve is simply to identify one player in the orchestra of signalling molecules. The player may not actually be of much importance in real life and contribute little to the complete orchestral piece. To identify the player is certainly not to identify the composition being performed.
There were many scientists who felt quite convinced in the mid 1980's that to inject one purified regulator into a body which contained a large mixture of potentially interacting or inhibitory regulators would fail to recapitulate in any significant manner the effects observable in simple tissue cultures. The injections would fail to induce any change in blood cell production. These sceptics were proved wrong in the case of the CSFs and the red cell-controlling hormone, erythropoietin, but perhaps were not so wrong for some subsequently discovered hematopoietic regulatory molecules. Subsequent evidence from millions of patients injected with CSF showed that predictable cellular responses are obtainable. Nevertheless, it was still possible for sceptics to doubt whether this evidence proved that the CSFs were genuine regulators of basal granulocyte and macrophage formation in the body. Perhaps they were merely emergency molecules called upon when increased cellularity was required.
To counter such criticisms, much subsequent work was required on gene inactivation studies where whole mice were designer-built in which every cell in the body lacked a particular CSF gene. The changes in these mice were then able to verify that each CSF really does perform certain unique functions in the biology of normal granulocyte and macrophage populations. Sophisticated as these most recent studies are, it is curious that this has involved a retreat from the elegant simplicity of purified molecules and clonal cultures to the black box confusion of the intact animal. There is certainly a lesson to be learned here for those with a compulsion for purity and exactness in their biology.
The Inexactitude of Our Body
One of the things that became very apparent during our decade or two of purification was that the body's cells are extremely imprecise in the manner in which they produce biologically-active macromolecules. We still assume that individual polypeptide molecules are synthesized with some precision by various cells, although this has not been stringently tested. However, what became evident, at least from work with recombinant proteins, is that there are different ways in which these polypeptides chains can fold to achieve a three dimensional structure. By no means all such molecules become folded correctly, resulting in inactive molecules or ones with reduced specific activity or stability. There is no reason to suppose that similar errors do not occur in the body during the production of native glycoproteins.
More striking is the highly variable manner in which carbohydrate is added to the polypeptides to produce the finished glycoprotein. This variable content of carbohydrates is the basis for much of the difficulty in purifying glycoproteins. It is unclear whether cells in the body are simply careless about how much carbohydrate they add while producing CSF molecules or whether the differing versions of CSF each have particular properties of special value in certain tissues or circumstances.
It is highly disconcerting to realize that the body's cells can produce a quite specific regulator like a CSF in such heterogeneous forms. This seems an extraordinarily untidy situation in the execution of a highly specific piece of biological signalling. It also makes it somewhat of a nonsense to set out to purify such a regulatory molecule. Which of the many variants is to be chosen as the "correct" molecule and on what rationale? Why purify GM-CSF from lung tissue, as we did, rather than differing forms of GM-CSF that can be produced by heart or muscle tissue?
It is an uncomfortable thought that a single gene product may never actually be able to be mass-produced to achieve a product that is truly representative of the heterogeneous versions that are present in different tissues in the body.
If the pursuit of purity for regulator molecules in fact is a Quixotic tilting at windmills that do not actually exist, another continuing passion of mine may well be a similar example of compulsive thinking and action. I have long had the desire eventually to be able to account for the origin and fate of every molecule of a CSF produced. Which exact cells in the body produced the CSF? How many molecules did each make? Why did each cell undertake this exercise? What actually happened to each of these molecules? What proportion were defective? What proportion, although varying in form, were adequate-enough functionally? Did most find a target cell, bind to specific membrane receptors, influence the behaviour of the cells then become internalized and degraded? How many were lost in transit by degradation? How many became degraded by some non-specific mechanism and what cells were scavenging these important molecules? What proportion were simply lost from the body in the urine or the gut? Leaving aside the nasty reality that many of these molecules were defective or less than ideally constructed, for those that were structurally adequate, are we talking about a highly economical signalling system or one with an enormous built-in wastage? Are hematopoietic regulators like radio waves being disseminated from a transmitter in a wasteful manner in the hope that someone has a radio tuned in at the correct wave-length or are they more akin to a landline telephone conversation where most signals reach the intended recipient?
I must admit that these quirky questions regarding the overall economy of what is a highly specific signalling system do not seem to be of much interest to other investigators. Certainly I would hate to prepare a research grant application on the subject because I am well aware of the puzzled indifference such a proposed study would arouse. Despite this, I believe that the answers would provide a quite revealing insight into the manner in which a multicellular organism functions.
In our lectures, we portray the CSFs and similar regulators as precisely designed molecules able to operate with exquisite specificity on equally specific membrane receptors to achieve desired responses. The reality is frighteningly otherwise - imperfectly synthesized molecules produced in vast numbers to achieve the concentrations necessary for the chemistry of regulator-receptor interactions. For every molecule entering into an effective binding interaction with its receptor, much greater numbers of molecules suffer all manner of unintended and purposeless fates of degradation and clearance. Far from being an exquisitely selective and efficient process, the system appears to be grossly wasteful and inefficient when considered from the viewpoint of the cells called upon to synthesize the molecules.
The situation is not improved by recognition that the intended target cells are themselves highly variable populations that are not always able or ready to respond to received signals. Hematopoietic cells are in general highly responsive to regulator stimulation, responding to picogram per ml concentrations. This they do even though they commonly express at any one time only a few hundred membrane receptors, only a fraction of which need to be activated to elicit a cellular response. However, these responding cells are in fact highly heterogeneous with respect to their intrinsic capacity for proliferative or other responses to signalling. Receptor numbers on cells within individual maturation stages vary widely, as judged from autoradiographs of cells binding radiolabeled regulators, and certainly such cells can exhibit an almost 100-fold variability in quantitative responsiveness to stimulation. It is unknown just how variable the structure of membrane receptors can be but, because they also are glycoproteins, some intermolecular variability must occur comparable with the heterogeneity of secreted regulatory molecules. For those whose purification ambitions extend to obtaining purified target cells, the extreme heterogeneity within even a single hematopoietic lineage is to say the least disconcerting.
So we have a doubly imperfect system - variable regulator molecules and variable target cells with an almost endless list of possibilities for inefficiency and wastefulness in the system. What a situation for a compulsive purifier and simplifier to find himself in, even with the encouraging example of the six monkeys with their near-completed Shakespeare manuscripts!
Final Reflections on the Pursuit of Purity
Was it an idiotic venture to set out to purify macromolecules of biological importance when absolutely uniform populations of such molecules probably do not in fact exist? Obviously the answer is yes, if purification was to be the only goal. However events have proved otherwise. Adequately-enough purified material can provide the key to identifying the relevant gene and thus a solution to the problem of mass-producing these interesting and clinically valuable molecules. Purity may be a quixotic goal and the windmills may not exist, but the quest for purity can actually achieve a hoped-for goal with the helpful intervention of molecular biology.
The fifteen-year pursuit of purity for the colony-stimulating factors did in the end allow their mass production by molecular biology and the clinical use of two of the CSFs to the benefit of many millions of patients. Fifteen years is a large fraction of a lifetime to spend on a project in which progress was so painfully slow as to be almost imperceptible. However, a new field of biology was being opened up and this in itself takes time and brings its own rewards.
With our present knowledge and experience of the past thirty years would we ever again attempt the feat of purification of a new protein? I believe we would under favourable circumstances if obviously high quality starting material was available. There are alternative methods now for cloning relevant genes such as expression cloning and, if we believe all we are told, we now have available at least parts of the sequence of the thirty thousand genes in the entire human genome. Even if some of these data contain errors, this information now allows rapid shortcuts for cloning genes when some rationale presents itself for paying attention to a particular sequence. In a very real sense, the new gene databases have supplanted much of the need for protein purification. Conversely, transgenic mice or mice with selective inactivations of particular genes, have met many of the investigational needs which might have been envisaged as requiring the production of a purified protein.
Regardless of whether newly discovered proteins in future will be purified in native form using conventional separative protein chemistry, or will be produced in equivalent purified recombinant form by expression of genes from the databases, further progress then requires purification of these for subsequent characterization. The need for purification and simple reductionist assay systems is still as great as it was thirty years ago.
Purity may not exist in real life and absolute purity may be an unattainable goal but near-purity remains essential on many occasions in experimental biology. Those who pursue excellence through purity are not always foolish dreamers.
Copyright 2003. Greek/Australian International Legal and Medical Conference.