The Nobel Prize in Chemistry 2004
Discovery of the ubiquitin system – graduate studies
Towards the end of the military service, I had to make what I assume has been the most important decision in my career: to start a residency in clinical medicine, in surgery, which was my favorite choice, or to enroll into graduate school and start a career in scientific research. It was clear to me that I was heading for graduate school. My disillusionment from clinical medicine that diseases can be cured based on understanding their pathogenetic mechanisms, along with a magical and enchanting attraction to biochemistry made the decision easy. I received a strong support and encouragement from my wife Menucha, who started to realize she was married to a student in sciences with no clear future rather than to the physician with a bright career and broad financial horizons that she thought she had married. So in November of 1976, after my discharge from the national military service and a two-month driving trip across the USA, I started my graduate studies with Avram Hershko. At that time his group focused mostly on studying intracellular proteolysis, and I learnt from him that he had given up on trying to identify the mediator(s) and mechanism(s) involved in serum-induced “pleiotropic response”. The model system that was chosen to study proteolysis was degradation of abnormal hemoglobin in the reticulocyte which is the terminally differentiating red blood cell. The reason for the selection of the reticulocyte as a model system was that we were looking for a non-lysosomal and energyrequiring proteolytic system, as from many studies it had become clear that regulated proteolysis of intracellular proteins is non-lysosomal (see the accompanying Nobel Lecture), and the reticulocyte no longer contains lysosomes which are removed during the final stages of its maturation (see below) before its release into the circulation. Interestingly, in the summer of 1978, during a Gordon Conference on Lysosomes, I met Dr. Alex Novikoff from Yeshiva University School of Medicine in New York. Alex, along with Dr. Christian de Duve, was one of the pioneers of the lysosome research field. When I told him we were working on the reticulocyte because this cell does not have lysosomes, he angrily dismissed this argument, telling me that he characterized morphologically acid phosphatase-positive organelles in reticulocytes. He even gave me the relevant paper he published on the subject, though it was not clear that these are proteolytically functional organelles. Another reason for the choice of the reticulocyte as a model for studying intracellular proteolysis was that in its final stages of maturation in the bone marrow and prior to entering the peripheral circulation, a massive proteolytic burst destroys most of its machineries, making it clear that the cell is equipped with an efficient proteolytic system. Earlier studies by Rabinovitz and Fisher demonstrated that the reticulocyte degrades abnormal, amino acid analogue-containing hemoglobin, yet the mechanisms had remained elusive. We assumed that the same mechanism that is involved in differentiation and maturation of the reticulocyte is also involved in the removal of “naturally occurring” mutant abnormal hemoglobins that are synthesized in different hemoglobinopathies, such as thalassemias and sickle cell anemia, and also in the destruction of the amino acid analogs – containing abnormal hemoglobins. We wanted to believe and hoped that this mechanism would turn out to be “universal”, and involved in degradation of normal proteins in all cells. Years later this assumption turned out to be correct. Thus, this important piece of information – the existence of a non-lysosomal proteolytic system, made the choice of the reticulocyte an obvious one. It was still necessary to demonstrate that the process requires energy, and indeed, following our initial characterization of degradation of abnormal hemoglobin in the intact cell, we showed that the process required energy (was published in 1978 in the proceedings of a proteolysis meeting held in Buffalo, NY), and felt that the time was ripe to break the cell open and isolate and characterize the non-lysosomal and ATP-dependent proteolytic enzyme(s). Shortly before, in 1977, Dr. Alfred Goldberg and his post-doctoral fellow Dr. Joseph Etlinger at Harvard Medical School characterized, for the first time, a cell-free proteolytic system from reticulocyte, which was exactly the point where we wanted to start our own march, so we basically adopted their system.
I will not describe here the detailed history of the discovery of the ubiquitin system, but rather highlight two important points along the five years of my exciting graduate studies (1976-1981) with Avram and Irwin A. Rose (Ernie) that led to the discovery of the system. The more detailed history can be found in several review articles written on the system at that time (most notable is Hershko, A. and Ciechanover, A. (1982). Mechanisms of Intracellullar Protein Breakdown. Annual Review of Biochemistry, 51, 335-364.) and later, and in the accompanying Nobel Lecture.
(1) The first point relates to the multiplicity of enzymatic components of the system: Our first aim along the purification process of the ATP-dependent “protease” was to remove hemoglobin, the major protein in the crude extract. Towards that end, we resolved the extract on an anion exchange resin, where we encountered already the first exciting finding. The proteolytic activity could not be found neither in the non-adsorbed material which we denoted Fraction I, nor in the high salt eluted material (denoted Fraction II). Rather, we recovered the activity following reconstitution of the two Fractions. We learnt two important lessons from this experiment which was published in 1978 in Biochemical and Biophysical Research Communications (BBRC; in my opinion the first paper in the long historical trail of the ubiquitin system) and which I regard as one of two or three key publications in the field. We learnt two lessons from this experiment: (i) The first was that the protease we were after was not a “classical” single enzyme that degrades its substrate, but had at least two components. This was already a digression from the paradigm in the field at that time that proteolytic substrates, almost without exception, can be cleared, at least partially, by single proteases with limited, yet defined specificities. Now, following the unraveling of the human genome and the discovery of important common structural domains within several groups of enzymes of the system, we know that the number of components of the ubiquitin system exceeds one thousand, but the first hint was already there; once one is left without a paradigm, all possibilities are open. (ii) The second lesson was a methodological one. Each time we lost activity during purification of any of the components we were characterizing, we “returned” to the chromatographical column fractions and tried to reconstitute it via complementation: “classical” biochemistry at its best was on our side. Standing at a crossroad, we – luckily but thoughtfully – decided to start first with purification and characterization of the active component in Fraction I. We decided so because Fraction I was the hemoglobin-containing fraction that did not adsorb to the resin, and since many proteins do absorb, we thought that this fraction should not contain too many additional proteins beyond hemoglobin, and it would be easy to purify the active component. Ten months after I started my studies (summer of 1977), Avram started his sabbatical with Ernie at the Fox Chase Cancer Center in Philadelphia, PA, USA, and left me with the task to purify the active component from Fraction I. After many unsuccessful trials along with another graduate student of Avram, Yaacov Hod, my colleague Mickey Fry, who was appointed as my substitute thesis advisor for this year (1977-1978), came with the “crazy” idea to heat Fraction I and see if the active component is heat-stable, and indeed it was. He did so as all our attempts to resolve the activity – despite the large difference in the molecular mass between the active protein (~10 kDa) and hemoglobin, the other major protein in Fraction I (65 kDa) – failed: hemoglobin, that is so abundant, “contaminated” the entire resolution span of each column in every single resolution method we used. Following 5-10 min at 90°C, the hemoglobin in crude Fraction I was “cooked” and precipitated like mud, and the activity remained soluble in the supernatant. It was hard to believe it was a protein, but Mickey remembered several other heat-stable proteins. Immediately after, we showed directly that the activity in Fraction I was also a protein: it was sensitive to trypsin and precipitable with ammonium sulfate. Further characterization revealed that the protein had a molecular mass of ~8,500 Da, and we called it ATP-dependent Proteolysis Factor-1, APF-1, to denote that this was the first component in the system that we characterized. All along the way I corresponded with Avram, sent him my results, and during his sabbatical we wrote the BBRC paper.
(2) The second key finding was also discovered in Haifa during the winter of 1978-1979. We purified APF-1 to homogeneity and labeled it with radioactive iodine. When the radio-labeled protein was incubated in crude reticulocyte Fraction II in the presence of ATP, we observed a dramatic increase in its molecular weight: it now migrated as a sharp peak in the void volume of the gel filtration chromatographical column. For several months we tried to elucidate the mechanism that underlies the change in the molecular weight of APF-1, hypothesizing, for example, that APF-1 could be an activator of a protease that must generate a binary complex with the proteolytic enzyme in order to activate it, but to no avail. An important breakthrough occurred during our 1979 summer stay of several months in the laboratory of Ernie. Through a series of extremely elegant, yet simple, experiments, in the design of which the broad knowledge of Ernie in protein chemistry and enzymology played a critical role, we found that APF-1 is covalently attached to the substrate through a bond that had all the characteristics of a peptide bond. Furthermore, we found that multiple moieties of APF-1 are attached to each substrate molecule, and that the reaction is reversible: APF-1 can be removed from the substrate or its degradation products and recycled, though not via reversal of the conjugation reaction. Accordingly, we hypothesized that covalent attachment of multiple moieties of APF-1 to the target substrate is necessary to render it susceptible to degradation by a downstream protease that recognizes only tagged but not untagged proteins, followed by the release of free and reusable APF-1.
The APF-1 cycle predicted the existence of three, entirely novel activities: (i) APF-1 conjugating enzyme(s), (ii) a protease that recognizes specifically the tagged substrates and degrades them, and (iii) APF-1-recycling enzymes. All these activities were identified later by us (the three conjugating enzymes, E1, E2, and E3) and by others (the conjugates degrading protease known as the 26S proteasome complex, and the ubiquitin recycling enzymes, the isopeptidases; see the accompanying Nobel Lecture). The findings describing the covalent tagging of the target substrate by APF-1 as a degradation signal, along with the first model of the newly discovered proteolytic system, were published in 1980 in two manuscripts that appeared in the Proceedings of the National Academy of Sciences of the USA (PNAS).
Another important development also occurred during our stay in Ernie’s laboratory, and I am not sure whether it was sheer luck or serendipity, probably both. We were not aware of any other precedent of a modification of a protein by another protein. The neighboring laboratories of Martin Nemer, Alfred Zweidler, and Leonard Cohen studied dynamics of variants of different histones during sea urchin development. They drew our attention to a protein called A24 (uH2A) which was discovered earlier by Ira Goldknopf and Harris Busch, and that was a covalent conjugate between two proteins – a small, ~8.5 kDa protein called ubiquitin, and histone 2A (H2A). Goldknopf and Busch, and in parallel Margaret Dayhoff, identified the nature of the bond between the two protein moieties in the conjugate. They found that the ubiquitinhistone bond was an isopeptide/bifurcated bond between the C-terminal Gly76 residue in the ubiquitin moiety, and the -NH2 group of Lys119 in the histone moiety of the conjugate. The role of this conjugate was not clear at the time, though its level was found to be dynamic and change during differentiation, when the histone moiety is subjected to ubiquitination and de-ubiquitination. This information on the ubiquitin-histone adduct along with the striking similarities we found between APF-1 and ubiquitin in their general characteristics such as molecular mass and amino acid composition, led Keith Wilkinson and his colleague Arthur (Art) Haas who were post-doctoral fellows in the laboratory of Ernie, along with Michael Urban from Zweidler’s laboratory, to carry out a series of direct experiments, showing unequivocally that APF-1 is indeed ubiquitin. Our study on the characterization of APF-1 and its possible similarity to ubiquitin, and Wilkinson’s study (along with Urban and Haas) on the identification of APF-1 as ubiquitin, led to the convergence of two fields, that of histone research and of proteolysis. More important, they suggested that the bond between ubiquitin and the target proteolytic substrate might be identical to that between ubiquitin and histone, which turned out later to be true. The two studies were published in tandem in 1980 in the Journal of Biological Chemistry (JBC; see the accompanying Nobel Lecture). The identification of the nature and structure of the bond clearly paved the road to the later purification and characterization of the conjugating enzymes and their mode of action.
As for ubiquitin, the protein was identified in the 1970s by Gideon Goldstein (in the Memorial Sloan-Kettering Cancer Center in New York City) as a small, 76 residue thymic polypeptide hormone that stimulates T cell differentiation via activation of adenylate cyclase. Additional studies by Gideon Goldstein had suggested that it was universally distributed in both prokaryotes and eukaryotes, thus giving rise to its name (coined by Gideon Goldstein). Later studies by Allan Goldstein showed that the thymopoietic activity was due to an endotoxin contamination in the protein preparation, and not to ubiquitin. Using functional assays, it was found in my laboratory (and I believe that in several others as well) that ubiquitin was limited to eukaryotes, and its apparent presence in bacteria was due to the contamination of the bacterial extract with the yeast extract in which the bacteria were grown: growing the bacteria in a synthetic medium containing carbon (glucose) and nitrogen (ammonium chloride) sources and vitamins resulted in “disappearance” of ubiquitin from the preparation. The later unraveling of the bacterial genome demonstrated unequivocally that the ubiquitin tagging system does not exist in prokaryotes, though there is some similarity between the proteasome and certain bacterial proteolytic complexes. Thus, in a relatively short period of time, ubiquitin was converted from a ubiquitous thymopoietic hormone to a eukaryotic proteolytic marker. While it appeared that the term ubiquitin was not justified anymore, as it is clearly not ubiquitous, we stopped using the term APF-1 and adopted the term ubiquitin for the modifying protein in the newly discovered proteolytic system. At times habits and tradition are stronger than the scientific validity and/or the logic in nomenclature. Accordingly, we adopted a general policy to use in our terminology the name/term that was first coined by the discoverer of any novel protein.
From that point on, the road was relatively short to the identification and characterization of the conjugation mechanism and the three enzymes involved in this process. En route we followed partially, with great admiration, the footsteps of Dr. Fritz Lipmann, the great biochemist from the Rockefeller University (who was awarded the 1953 Nobel Prize in Physiology or Medicine for the discovery of Coenzyme A). Lipmann continued to contribute enormously to our understanding of basic biochemical processes. Among his many discoveries was the mechanism of non-ribosomal (and hence non-genetically encoded) peptide bond formation that is involved in the biosynthesis of bacterial oligopeptides such as Gramicidin S. We learnt that the basic biochemical principles, such as generation of high-energy intermediates involved in peptide bond formation, were preserved along evolution regardless of whether the bond is encoded genetically or not, or whether it links two amino acids, or two proteins, or an amino acid to the elongoting polypeptide chain. Initially, we identified the general mechanism of activation of ubiquitin in crude extract. Later, using “covalent” affinity chromatography over immobilized ubiquitin and a stepwise elution (that was based on the general mechanism we deciphered earlier), we purified the three conjugating enzymes that act successively, in a cascade-like mechanism, and catalyze this unique process: (1) the ubiquitin-activating enzyme, E1, the first enzyme in the ubiquitin system cascade, (2) the ubiquitin-carrier protein, E2, to which the activated ubiquitin is transferred from E1, and (3) the ubiquitin-protein ligase, E3, the last and critical component in the three step conjugation mechanism that specifically recognizes the target substrate and conjugates it with ubiquitin. The E3 was also adsorbed to the immobilized ubiquitin, although via a yet unknown mechanism, distinct from that of E1 and E2: the binding of these two enzymes was mediated by the activation mechanism. Later studies by Avram in the late 1980s revealed that the E3 adsorbed by the column was E3 that recognizes substrates via their N-terminal residue. At this point, however, unknowingly and unintentionally, we were extremely lucky when we used as model substrates commercial proteins such as BSA, lysozyme and RNase, that were all recognized (as we learnt later) by this ligase and via a similar targeting motif – their N-terminal residue. Had we used other substrates, such as globin, the protein we used in our initial experiments, the E3 adsorbed to the column would have probably escaped our attention, as E3s do not typically adsorb to ubiquitin. Independently, and in parallel to the later characterization of the enzyme by Avram, I also used this enzyme in order to characterize a distinct subset of proteins recognized via this signal (see below). Lastly, using antibodies that we raised against ubiquitin with the help of Arthur Haas, we found that the ubiquitin system is involved in degradation of abnormal, short-lived proteins in hepatoma cells, demonstrating that the system is not limited to the terminally differentiating reticulocyte, but is probably distributed “universally” in nucleated mammalian cells, playing an important role in maintaining the cell’s quality control, by removing abnormal proteins. During my graduate studies at Avram’s laboratory, I collaborated with Hannah Heller, an extremely talented and knowledgeable research associate (who also joined us for some of our summer stays in the laboratory of Ernie in Philadelphia), and with Yaacov Hod, who was also a graduate student with Avram at that time. Other colleagues in the laboratory provided me with a lot of help during this period, including Dvorah Ganoth, Sarah Elias, and Esther Eythan who were research associates with Avram, and Clara Segal and Bruria Rosenberg, two dedicated technicians.