Key Symposium

|
doi: 10.1111/j.1365-2796.2010.02319.x

The value of failure: the discovery of TNF and its natural inhibitor erythropoietin
A. Cerami
Leiden University Medical Center, Leiden, The Netherlands

Abstract. Cerami A (Leiden University Medical Center, Leiden, The Netherlands). The value of failure: the discovery of TNF and its natural inhibitor erythropoietin (Key Symposium). J Intern Med 2011; 269: 815. The hurtful feelings associated with failing can be devastating especially if the failure occurs after the investment of a considerable effort. The reection of a lifetime of work in translational medicine has revealed that the study of failures can give birth to new insights that can be explored with important consequences. This article discusses the analysis of two

failures that have led to remarkable discoveries. The rst led to the discovery of TNF as an important mediator of inammation that can, if unchecked, cause severe damage in mammals. The second is the identication of erythropoietin as the natural inhibitor of the production and biological activity of TNF. I hope that this paper will help give students the courage to persist in looking for the insights that are the by-products of failure, and to understand the long time lines in the path of discoveries. Keywords: biological therapy, erythropoietin, immunomodulation, inammation, tissue protection, TNF.

Introduction Recently, in the defence of the validity of patents that I have been an author of, I reviewed forty years of my discovery work, which consisted of a truck load of le boxes, laboratory note books, scientic papers, correspondence, personal journals and the occasional paper napkin that contained scribbles of a discussion over lunch. The process was overwhelming, but revealed to me one very important truth: many of the biggest discoveries of my career were the results of failure of another research project. Failure strikes a negative tone, but it appeared in my personal history that it was an essential experience on the path to important discoveries. Failures happen all the time, especially in experimental biology, and they can be very disheartening. In this paper, Im going to embrace the concept that failure has a value, and review with the details of the two failures, which eventually led to discoveries that have turned out to be quite remarkable. I hope that students reading this paper will use the lessons of these examples to help alleviate the pain of failing and give them courage. The discovery of TNF In the mid-1970s, I received a grant from Ken Warren of the Rockefeller Foundation as part of the Great

Neglected Disease Network to support new research initiatives in the eld of parasitology. This was a unique 10-year grant that required us to spend a considerable effort to advance the eld. One of the projects I choose to study was trypanosome infections in cattle. This came about because of the interesting work that had been done at the beginning of the 20th century by the famous pharmacologist, Paul Ehrlich. He discovered the rst anti-trypanosomal drug for the treatment of cattle in Africa. Of historical interest, when the spirochete causing syphilis was rst discovered, Ehrlich thought that this organism was a small trypanosome and began to try the compounds that he had developed for trypanosomes on this spirochete. We now know that these organisms are not related, but it did lead to the development of the magic bullet 606, a highly potent drug for the treatment of syphilis. Over a 2-year period, we developed several compounds that showed promise as anti-trypanosomal drugs. A single injection could completely cure mice and rats of the parasites. With this information in hand, we travelled to Kenya to evaluate one of these compounds in cattle infected with trypanosomes. One of the main problems with trypanosome infections is that the animals lose rapidly a considerable amount of muscle and fat literally wasting away until their death. When we administered our new

2010 The Association for the Publication of the Journal of Internal Medicine

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

compound to the infected cows, they proceeded within minutes to go into shock and die of some unknown reason. If we gave the compound to noninfected animals there was no visible effect. These results were, needless to say, quite discouraging and embarrassing. The joke around the facility was that Have you heard that those guys from New York City have developed a great diagnostic to identify trypanosome infected animals the problem was that the readout was their death. Although we spent a considerable amount of effort trying to understand the basis of this phenomenon, we never came up with an explanation and eventually abandoned this series of compounds. One day, when I was in the corral examining the cattle, I developed severe abdominal pains from some unknown but common gastrointestinal bug. I decided to sit in the corral among the cows and wait until the pain stopped. As I sat there I decided to take my mind off the pain by analysing the events of the past few weeks. The cutting remark that this compound would be useful for identifying infected animals was actually an interesting idea, because in contrast to mice and rats who had vast numbers of organisms, the infected cows had very few parasites in the blood or for that matter in the rest of the body. Then, the question arose in my mind how could so few parasites cause this tremendous loss of weight and death? A common explanation for cachexia or the wasting associated with infectious diseases or cancer was the usurping of the energy stores of the host for its own growth. In some instances, such as human Leishmania donovani infections, the parasite load in the liver and spleen is immense. In the case of the infected cows, it seemed hard to believe that so few organisms could use that much energy. I then remembered speaking with an investigator who was studying trypanosomes in wild antelopes. He found that they developed a level of parasitaemia similar to that seen in cattle, but they did not develop cachexia. It then occurred to me that there was a good possibility that the European cattle were overreacting to the presence of a minute number of parasites and producing a mediator that was responsible for the wasting. The question was what was this mediator and how could we prevent its action. I thought that if we knew the answers to these questions we would have insight into the pathogenesis of many diseases. Over the next 30 min, I not only laid out the scientic plan that

we would follow for the next 10 years but was also relieved of my abdominal pains. Upon returning to New York, Carol Rouzer, a graduate student, began to study rabbits infected with trypanosomes that like the cow had a very low parasitaemia, but severe wasting. We began by studying the anaemia associated with the cachexia of this infection. We found that one of the main reasons for the anaemia was the inability to make new red cells. She then made the observation, that during the course of the infection, the serum of the animals became extremely lipemic [1]. Further studies revealed that this was due to the loss of the enzyme lipoprotein lipase (LPL) (Fig. 1). It was clear that study of the loss of this enzyme in response to invasion would be a good biomarker to identify and isolate the mediator. The project was then turned over to Masanobu Kawakami, a dedicated and talented post-doctoral fellow. We decided to switch from trypanosome-infected rabbits to mice that were challenged with endotoxin or lipopolysaccharide (LPS). It had been known that when some strains of mice were injected with endotoxin, they developed severe organ damage that led to death, whereas, other strains of mice were resistant. We performed a very simple experiment in which we showed that the administration of LPS to the sensitive mouse led to a decrease in LPL, whereas the administration of LPS to the resistant mouse did

Fig. 1 Trypanosome-infected rabbits have increased triglycerides and decreased LPL (LPL). During the course of the infection, the triglyceride levels go up dramatically, whereas the LPL levels fall [Ref. 1].

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

not. We were then able to show that if we took sera from the sensitive mouse that had been injected with LPS 60 min previously and gave it to a resistant mouse, there was a suppression of LPL signalling the presence of a mediator (Fig. 2). We continued to simplify our animal models by showing that macrophages from sensitive mice when incubated with LPS also produced the mediator that could suppress LPL in the resistant mice [2]. Subsequently, we were able to simplify our models even more by developing an in vitro assay with Phil Pekala and Dan Lane of Johns Hopkins utilizing 3T3 L1 cells that had been induced to form fat cells. The addition of the mediator to these fat cells led to a specic turning off of LPL and other enzymes responsible for fatty acid synthesis and uptake [3, 4]. This LPL assay formed the basis for our isolation of the mediator. In a series of papers, we showed that the supernatants of macrophages that had been exposed to LPS or other molecules from infective organisms had the ability to induce a mediator with many activities that had been associated with cachexia and anaemia [5 7]. Although we believed that there was a single mediator, it was only after we had recombinant cachectin TNF that we could say conclusively that there was only one mediator that had these many diverse biological effects. In 1981, we submitted an application to the US patent ofce [8] describing the mediator as a protein, with an apparent molecular weight of 70 000 daltons

as measured by gel ltration chromatography, that was produced by macrophages in response to various agents mimicking invasion. A number of the biological activities of the mediator were described as well as the principle of inhibiting its activities with poly- and mono-clonal antibodies in a number of human maladies including shock, cachexia, rheumatoid arthritis and other conditions that were the result of elevated levels of this mediator. At this point, we named the mediator cachectin because many of its biological activities were associated with the phenomena of cachexia. At this point in time, micro-sequencing of small quantities of protein was just beginning. Most batches of our puried material were lost attempting to obtain the amino terminal sequence. In desperation, I reached out to Dr. Y. Pan of Roche who had just obtained the latest sequencing apparatus. On his rst attempt, he was able to obtain the amino-terminal sequence of mouse cachectin that had been isolated by Bruce Beutler, a post-doctoral fellow [9]. Examination of the available databases for other proteins did not reveal any similarity in sequence. A few days later, Bruce noticed that the sequence we had obtained for mouse cachectin was very similar to that of the recently cloned protein, human TNF. This came as a big surprise to us because several years before, we had exchanged material with Lloyd Olds group at SloanKettering to see if TNF had LPL suppressive activities or cachectin had anti-tumour activity. Both assays were negative. We were subsequently able to show that mouse cachectin and mouse TNF were identical [10]. Why neither laboratory was able to nd the other activity remains a mystery. The fact that cachectin and TNF were in fact the same caused a considerable stir in the biotechnology eld. It had been hoped that TNF would be a specic antitumour agent that could be administered safely to patients [11]. The activities associated with cachectin, on the other hand, would be detrimental if not lethal to the patient. At one point, I attended an early biotechnology meeting and was accused by one of the participants of wanting to destroy the young biotechnology industry. Implicit in his statement was that our observations on the biological activities of cachectin TNF were faulty. I assured him that we were condent of all of the biological activities that we described were associated with TNF and that we understood that these activities would be dose-limiting or preclude its use as an anti-tumour agent in patients. I then proposed that the prevention of the activity of TNF with monoclonal antibodies would be a much

Fig. 2 Presence of Mediator in Serum of lipopolysaccharide (LPS)-treated LPS-sensitive Mice. Injection of LPS into LPSsensitive mice results in a decrease of lipoprotein lipase (LPL). Injection of LPS into LPS-resistant mice does not diminish LPL activity. When the serum is obtained from sensitive mice that were injected previously with LPS and injected into resistant mice, there is a fall in LPL denoting the presence of a mediator that can decrease LPL [Ref. 2].

10

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

larger market as so many diseases are associated with elevated cachectin TNF levels. Unfortunately, except in certain instances, the clinical use of TNF as an anti-tumour agent has not been possible [12]. Clinical studies of TNF conrmed all of the animal studies that we had seen with mouse cachectin [13, 14]. Fortunately for the biotech industry, anti-TNF therapies did succeed and are one of the largest selling biotechnology products sold in the world today. Over the years, it has become clear that in response to invasion (including bacteria, viruses, parasites, tumours) or tissue damage, there is the local and systemic production of TNF (Fig. 3). At the local level, it can induce apoptosis, production of other cytokines, nitric oxide and other mediators that enlarge the amount of damage and in the process kill the invader [15]. If the invasion (either real or perceived) is signicant, then TNF will spill over into the blood stream and on a systemic level cause fever, anorexia, malaise, a generalized catabolic state, shock and death [16, 17]. Armed with recombinant TNF, it was possible to show all of these biological activities in vivo and in vitro. The administration of recombinant h-TNF led to a rapid rise in the respiratory rate, due to an increased production of lactic acid, a fall in blood pressure and the subsequent death of the animal (Fig. 4). We also believed that cachectin TNF would play an important role in inammatory diseases such as rheumatoid arthritis. In collaboration with Jean-Michele Dayer, we were able to show in 1985 that small concentrations of cachectin TNF could induce human synovial cells to produce PGE2 and collagenase [18] two mediators that were believed to be involved in the damage of joints in rheumatoid arthritis patients.

Fig. 4 Vital Signs after TNF Infusion. The infusion of rhTNF into rats prompts a rise in respiratory rate, RR, due to TNF induced lactic acidosis, a fall in blood pressure, BP, followed by a cessation of heart rate, HR [Ref. 17].

Utilizing recombinant human-cachectin TNF, we were able to produce specic monoclonal antibodies. One of these antibodies was able to cross-react and neutralize the bioactivity of baboon TNF. We were able to evaluate its potential in preventing septic shock in animals that were given a lethal amount of live Escherichia coli (Fig. 5) [19]. When the antibody was administered 4 h before the infusion of the bacteria, there was a complete prevention of the pathological sequelae and death that occurred in the saline treated animals. All of the control animals died within the rst 7 h. If the animals were given the antibody at the time of the infusion of bacteria, the animals also died. The timing of the administration of the antibody was obviously critical and probably accounts for the lack of clear-cut clinical efcacy in human trials of anti-TNF antibodies in septic shock. The chronic administration of TNF to experimental animals can induce all of the changes that are seen in cachexia. This was best demonstrated by Alan Oliff and his colleagues who showed that CHO cells secreting TNF could reproduce all of the sequelae of cachexia [20]. The administration of anti-TNF antibodies to mice bearing tumours revealed a signicant alleviation of the sequelae of cachexia [21]. Surprisingly, there have been very few clinical studies evaluating monoclonal antibodies to TNF in cancer cachexia. The anti-TNF monoclonal antibodies have proven to be very successful in the treatment of patients with chronic inammation including Crohns disease, rheumatoid arthritis and psoriasis. These pioneering clinical studies were carried out by Sander van Deventer [22], Ravinder Maini and Mark Feldmann [23] utilizing a monoclonal produced by Centocor. An

Fig. 3 Effects of TNF in vivo. Many pathogens, tissue damage and tumours have been described to elicit the production of TNF by macrophages. The TNF can have local as well as systemic effects in vivo [Ref. 10].

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

11

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

Fig. 5 Anti-TNF antibodies prevent septic shock in lethal bacteraemia. An LD100 bolus of live Escherichia coli was injected into baboons that had been given 4 h before either saline or a neutralizing monoclonal antibody to human TNF that cross reacted with baboon TNF. The animals receiving the antibody were protected from the lethality and when sacriced 7 days later, they did not display the usual organ damage associated with sepsis [Ref. 19].

unsung hero in these early endeavours is Jim Woody who, as the head of research at Centocor, promoted and protected these studies. The discovery of erythropoietin (EPO) as an endogenous TNF antagonist TNF can cause a signicant amount of local collateral damage as a result of the body believing that there is an invasion in progress. We concluded that a natural scheme must have evolved to counter the activity of this potent cytokine, otherwise small injuries could lead to devastating damage. For a several year period in the end of the 1980s we began a programme to nd this mechanism. For several years, we searched for other cytokines made by macrophages that had either pro-inammatory or anti-inammatory action. During this period, we identied a number of new pro-inammatory cytokines, but completely failed in identifying any anti-inammatory molecules [24 27]. After a few years of searching, in frustration, the project was abandoned. In 1998, Mike Brines, Carla Hand and I received a grant from Johnson and Johnson to study why EPO administration to cancer patients improved their feeling of well-being. This phenomenon accounts for the widespread use of this cytokine in oncology. As we

knew from animal studies that TNF causes malaise that can be demonstrated by a slower rate of learning in a water maze, we thought that in this model, we could evaluate whether EPO could interfere with TNF activity. As a control, we administered EPO intraperitoneally to normal young animals and found to our astonishment that these animals were able to learn the maze faster. After conrming this experiment many times, we hypothesized that as EPO had the opposite activity of TNF in this model, perhaps it could counteract TNF in other situations. In a series of experiments, we were able to show that EPO indeed had the opposite prole to that seen with TNF, e.g. decreasing apoptosis and inammation while improving the rate of healing and regeneration of tissue after damage. Another interesting observation we have found is that EPO and TNF can regulate the production and the activity of the other. Thus, elevated TNF levels can suppress the production of EPO that accounts, in part, for the anaemia associated with cachexia. The mistake we made in our earlier programme to nd the mechanism that countered TNF-associated damage was to assume that the inhibitor of TNF would be produced by macrophages. It never occurred to us that the surrounding cells would be the source. The signal for local production of EPO is the activation of HIF 1 by hypoxia that is the caused by TNF impairing blood ow. The administration of EPO to experimental animals, having been subjected to many kinds of injury, has been found to reduce the amount of damage greatly. These include stroke, brain and spinal cord trauma, ischaemia reperfusion models of the retina, heart, kidney and bowel, to name a few [2831] (Fig. 6). Interestingly, a clinical study of EPO administration (40 000 units) once a week for 4 weeks to trauma patients admitted to the ICU showed a 50% decrease in death rate at 28 days compared with the saline controls [32]. Despite this signicant decrease in mortality, the patients receiving EPO also had a 40% increased risk of thrombosis. This side effect of EPO is well known and is one of the reasons why EPO has a black label warning by the FDA. It was readily apparent that the potential of EPO as a tissue-protective molecule was greatly diminished by these other activities. The question was whether we could engineer the molecule to remove these other activities. In pharmacology, this can successfully be achieved in situations where there are two different receptors. Early in our studies it became clear that the amounts of EPO that we needed for tissue protection were signicantly higher than that needed

12

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

Fig. 6 Tissue-protective molecules (erythropoietin and Carbamylated-erythropoietin, a nonerythropoietic molecule) can limit injury in many tissues. In this Figure are shown the protective effects from brain trauma, retinal ischaemia, myocardial infarction and kidney ischaemia reperfusion injury [Ref. 28 and unpublished work].

for erythropoiesis, suggesting a tissue-protective receptor with a lower afnity. We then studied two EPO derivatives that were previously reported to not be erythropoietic in vivo. Asialo-EPO can sustain erythropoiesis in vitro, but is completely inactive in vivo because of its short half-life of a few minutes. Carbamylated-erythropoietin (CEPO which is made by the reaction of cyanate with the amino groups of EPO) is not erythropoietic in vitro or in vivo. When these compounds were evaluated in vivo, both were found to retain completely the tissue-protective activities [33, 34]. Utilizing a CEPO-column, we were able to isolate the tissue-protective receptor [35]. It was found to be composed of the EPO-receptor disulde linked to CD131, commonly known as the beta-common

receptor because of its participation with other specic alpha chains for the cytokines IL3, GM-CSF and IL5 (Fig. 7). In the early history of EPO, this receptor had been isolated and was found to be phosphorylated after exposure to EPO [36]. Subsequent studies showed that the homodimer of EPO-R was the haematopoietic receptor [37]. In addition, the beta-common knock out mice for the beta-common gene did not have any haematopoietic decits [38]. When these mice were damaged in vivo or their cells in vitro, EPO was no longer able to protect against collateral damage. The tissue protective receptor has a lower binding constant, 16 nM, compared to 0.2 nM for the homodimer of EPO-R. Another important difference is that homodimer requires low constant levels to maintain haematopoietic effects, whereas the tissue-protective receptor requires high, but infrequent levels. As the x-ray structure is known, we have been able to design EPO molecules that have lost their haematopoietic activities by substituting critical amino acids of EPO needed for binding to the EPO-R homodimer. There are a number of problems associated with the production of large protein molecules by mammalian cells. The rst is the cost of the regulatory barrier (estimated to be $2025 million) that needs to be overcome to do even proof of concept clinical studies and the second is the cost of goods when the product is eventually sold to patients. These were challenges that we could not meet.

Fig. 7 Erythropoietin Signals via Two Receptors. (a) The classical receptor is a homodimer of the erythropoietin (EPO)R that is found primarily on haematopoietic cells has a high afnity for erythropoietin that requires the ligand to be present at constant levels. (b) The tissue-protective receptor is a hetero-receptor of the EPO-R disulde linked to the Beta-common receptor (CD131). This receptor has a much lower afnity, but once it interacts with its ligand, the presence of the ligand is no longer required for bioactivity.

The fact that asialo-EPO, which has an extremely short half-life in vivo, has all of the tissue-protective activity of EPO, encouraged us to evaluate peptides as tissue-protective molecules [39]. We have identied a number of peptides that are tissue protective in vivo. Some of these are pieces of EPO, while one is a linear peptide of 11 amino acids that reect the amino acids in space along a helix that, we believe, interacts

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

13

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

Fig. 8 Developmental Timeline of important points in the discovery and development of TNF and its natural inhibitor, erythropoietin.

with the tissue-protective receptor. This peptide, ARA290 that is not a piece of EPO, is as effective at tissue protection on a molar basis in a number of animal models as the parent molecule EPO. Thus, we were able to obtain a chemically synthesized tissueprotective molecule, reduce the molecular weight by nearly thirty fold and remove the biological activities (erythropoiesis, producing hyperactive platelets, promoting coagulation and increasing blood pressure) that were safety issues associated with EPO. Pre-clinical animal studies and clinical studies in human volunteers have not revealed any untoward effects. Phase 2 clinical trials in patients with critical limb ischaemia, painful diabetic neuropathy and retinopathy and rheumatoid arthritis are underway or in the planning stages. Whether this peptide will be as effective in clinical medicine as anti-TNF therapies will have to await the results of these trials. Although I believe that examination of failure can lead to new discoveries, I hope that these clinical trials will be successful as I would not like to have to worry about another failure. Final reections The other observation I made as I reviewed my lifes work in translational medicine was that considerable patience was needed. Figure 8 shows a time line for the development of the programmes I have discussed above. In both instances, there was about a 10-year period between the basic discovery until clinical studies could be carried out to determine whether the

approach would be successful. During this period, there were many discouraging moments of conict and uncertainty, but in looking back, I am glad that I chose this difcult path. To be able to discover diagnostics or therapies that can relieve pain and suffering is very rewarding. I hope that young people reading this paper will be encouraged to choose to follow this long and arduous path. Acknowledgements I thank Ann Dunne, Ed Gray and Ulf Andersson for helpful discussions. Conict of interest statement Anthony Cerami is a share holder of Araim Pharmaceuticals that owns the patent on peptides that mimic the tissue protective activities of erythropoietin. References
1 Rouzer CA, Cerami A. Hypertriglyceridemia associated with trypanosoma brucei brucei infection in rabbits: role of defective triglyceride removal. Mol Biochem Parasitol 1980; 2: 318. 2 Kawakami M, Cerami A. Studies of endotoxin-induced decrease in lipoprotein lipase activity. J Exp Med 1981; 154: 6319. 3 Kawakami M, Pekala PH, Lane MD, Cerami A. Lipoprotein lipase suppression of 3T3-L1 cells by an endotoxin induced mediator from exudate cells. Proc Natl Acad Sci USA 1982; 79: 9126. 4 Pekala PH, Kawakami M, Angus CW, Lane MD, Cerami A. Selective inhibition of synthesis of enzymes for de novo fatty acid biosynthesis by an endotoxin-induced mediator from exudate cells. Proc Natl Acad Sci USA 1983; 80: 27437.

14

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

A. Cerami

Key Symposium: The discovery of TNF and its natural inhibitor erythropoietin

5 Pekala P, Kawakami M, Vine W, Lane MD, Cerami A. Studies of insulin resistance in adipocytes induced by macrophage mediator. J Exp Med 1983; 157: 13605. 6 Hotez PJ, Le Trang N, Fairlamb A, Cerami A. Lipoprotein lipase suppression in 3T3-L1 cells by a haematoprotozoan induced mediator from peritoneal excudate cells. Parasite Immunol 1984; 6: 2039. 7 Sassa S, Kawakami M, Cerami A. Inhibition of the growth and differentiation of erythroid precursor cells by an endotoxin-induced mediator from peritoneal macrophages. Proc Natl Acad Sci USA 1983; 80: 171720. 8 Cerami A, Kawakami M. Method for reducing adverse effects of a human 70 KDa mediator which results from endotoxin stimulation of macrophages. U.S. Patent #6,419,927 B1. 9 Beutler B, Mahoney J, Le Trang N, Pekala P, Cerami A. Purication of cachectin, a lipoprotein lipase-suppressing hormone secreted by endotoxin-induced RAW 264.7 cells. J Exp Med 1985; 161: 98495. 10 Beutler B, Greenwald D, Hulmes JD et al. Identity of tumor necrosis factor and the macrophage-secreted factor cachectin. Nature 1985; 316: 5524. 11 Carswell EA, Old LJ, Kassel RL, Green S, Fiore N, Williamson B. An endotoxin-induced serum factor that causes necrosis of tumors. Proc Nat Acad Sci USA 1975; 72: 366670. 12 Grunhagen DJ, de Wilt JHW, Graveland WJ, Verhoef C, van Geel AN, Eggermont AMM. Outcome and Prognostic Factor Analysis of 217 Consecutive Isolated Limb Perfusions with Tumor Necrosis Factor-a and Melphalan for Limb-Threatening Soft Tissue Sarcoma. Cancer 2006; 106: 177684. 13 Blick M, Sherwin SA, Rosenblum M, Gutterman J. Phase I Study of Recombinant Tumor Necrosis Factor in Cancer Patients. Cancer Res 1987; 47: 29869. 14 Kimura T, Taguchi T, Urushizaki I et al. Phase I study of recombinant human tumor necrosis factor. Cancer Chemother Pharmacol 1987; 20: 2239. 15 Dinarello CA, Cannon JG, Wolff SM et al. Tumor necrosis factor (cachectin) is an endogenous pyrogen and induces production of interleukin-1. J Exp Med 1986; 163: 143350. 16 Beutler B, Cerami A. Cachectin and tumour necrosis factor as two sides of the same biological coin. Nature 1986; 320: 5848. 17 Tracey KJ, Beutler B, Lowry SF et al. Shock and tissue injury induced by recombinant human cachectin. Science 1986; 234: 4704. 18 Dayer JM, Beutler B, Cerami A. Cachectin tumor necrosis factor stimulates collagenase and prostaglandin E2 production by human synovial cells and dermal broblasts. J Exp Med 1985; 162: 21638. 19 Tracey KJ, Fong Y, Hesse DG et al. Anti-cachectin TNF monoclonal antibodies prevent septic shock during lethal bacteremia in baboons. Nature 1987; 330: 6624. 20 Oliff A. The role of tumor necrosis factor (cachectin) in cachexia. Cell 1988; 54: 1412. 21 Sherry BA, Gehn J, Fong Y et al. Anticachectin tumor necrosis factor-alpha antibodies attenuate the development of cachexia in two murine transplantable tumor models. FASEB J 1989; 3: 195662. 22 Derkx B, Taminiau J, Radema S et al. Tumour-necrosis-factor antibody treatment in Crohns disease. Lancet 1993; 342: 1734. 23 Elliott MJ, Maini RN, Feldmann M et al. Treatment of rheumatoid arthritis with chimeric monoclonal antibodies to tumor necrosis factor alpha. Arthritis Rheum 1993; 36: 168190.

24 Wolpe SD, Davatelis G, Sherry B et al. Macrophages secrete a novel heparin-binding protein with inammatory and neutrophil chemokinetic properties. J Exp Med 1988; 167: 57081. 25 Wolpe S, Sherry B, Juers D, Davatelis G, Yurt RW, Cerami A. Identication and characterization of macrophage inammatory protein 2. Proc Natl Acad Sci USA 1989; 86: 6126. 26 Sherry B, Yarlett N, Strupp A, Cerami A. Identication of cyclophilin as a pro-inammatory secretory product of LPS-activated macrophages. Proc Natl Acad Sci USA 1992; 89: 35115. 27 Bernhagen J, Calandra T, Mitchell RA et al. MIF is a pituitaryderived cytokine that potentiates lethal endotoxaemia. Nature 1993; 365: 7569. 28 Brines ML, Ghezzi P, Keenan S et al. Erythropoietin crosses the blood-brain barrier to protect against experimental brain injury. Proc Natl Acad Sci USA 2000; 97: 1052631. 29 Celik M, Gokmen N, Erbayraktar S et al. Erythropoietin prevents motor neuron apoptosis and neurologic disability in experimental spinal cord ischemic injury. Proc Natl Acad Sci USA 2002; 99: 225863. 30 Gorio A, Gokmen N, Erbayraktar S et al. Recombinant human erythropoietein counteracts secondary injury and markedly enchances neurological recovery from experimental spinal cord trauma. Proc Natl Acad Sci USA 2002; 99: 94505. 31 Calvillo L, Latina R, Kajstura J et al. Recombinant human erythropoietin protects the myocardium from ischemia-reperfusion injury and promotes benecial remodeling. Proc Natl Acad Sci USA 2003; 100: 48026. 32 Corwin HL, Gettinger A, Fabian TC et al., EPO Critical Care Trials Group. Efcacy and safety of epoetin alfa in critically ill patients. N Engl J Med 2007; 357: 96576. 33 Erbayraktar S, Grasso G, Sfacteria A et al. Asialoerythropoietin is a non-erythropoietincytokine with broad neuroprotective activity in vivo. Proc Natl Acad Sci USA 2003; 100: 67416. 34 Leist M, Ghezzi P, Grasso G et al. Derivatives of erythropoietin that are tissue protective but not erythropoietic. Science 2004; 305: 23942. 35 Brines M, Grasso G, Fiordaliso F et al. Erythropoietin mediates tissue protection through through an erythropoietin and common beta-subunit heteroreceptor. Proc Natl Acad Sci USA 2004; 101: 1490712. 36 Masuda S, Nagao M, Takahata K et al. Functional erythropoietin receptor of the cells with neural characteristics. Comparison with receptor properties of erythroid cells. J Biol Chem 1993; 268: 1120816. 37 Livnah O, Stura EA, Middleton SA, Johnson DL, Jolliffe LK, Wilson IA. Crystallographic evidence for preformed dimers of erythropoietin receptor before ligand activation. Science 1999; 283: 98790. 38 Scott CL, Robb L, Papaevangeliou B, Manseld R, Nicola NA, Begley CG. Reassessment of interactions between hematopoietic receptors using common beta-chain and interleukin-3-specic receptor beta-chain-null cells: no evidence of functional interactions with receptors for erythropoietin, granulocyte colonystimulating factor, or stem cell factor. Blood 2000; 96: 158890. 39 Brines M, Patel NS, Villa P et al. Nonerythropoietic, tissue-protective peptides derived from the tertiary structure of erythropoietin. Proc Natl Acad Sci USA 2008; 105: 1902530. Correspondence: Anthony Cerami, Leiden University Medical Center, Leiden, The Netherlands and Araim Pharmaceuticals, Inc, 712 Kitchawan Rd. Ossining, NY 10562 (Town of Yorktown). (fax: +914-762-7292; e-mail: acerami@araimpharma.com).

2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 815

15