Benefits from Unearthing “a Biochemical Rosetta Stone”

Administration of ade-REFLECTIONS:

cally active form and that it might be phosphoenol-acetoacetic acid. I began a project to synthesize this compound to test its potential as a precursor of cholesterol, and Gurin went to England for a six-month sabbatical. Needless to say, I did not succeed in making it, nor was it ever shown to be an intermediate in cholesterol formation. When Gurin returned, we began to examine the conversion of acetate and other short-chain fatty acids and acetone to long-chain fatty acids and cholesterol in slices of rat liver tissue. We learned that the conversion of 14 C-labeled acetate and other short-chain precursors to long-chain fatty acids had an absolute requirement for bicarbonate in the incubation medium (2). Moreover, our findings indicated that the conversion of acetone to long-chain fatty acids occurred by its conversion to a metabolically active 2-carbon fragment that was not acetate but might be secondarily converted to acetate (1). Additional studies at that time included investigations on the impairment of fatty acid synthesis in diabetes (3) and the demonstration that insulin stimulated fatty acid synthesis, whereas growth hormone and cortisone inhibited this process (4).
Gurin and I realized that to gain insight into the biochemical reactions involved in the biosynthesis of long-chain fatty acids, we needed to find an enzyme system that catalyzed this process. Pigeon liver was known to be metabolically quite active. We decided to prime fatty acid synthesis in that organ by feeding pigeons glucose over a four-hour period before preparing homogenates of liver tissue. Using a buffer supplemented with potassium bicarbonate, we discovered that such preparations could convert 14 C-labeled acetate to long-chain fatty acids (5). Eventually, a completely water-soluble enzyme system was developed (6). A marked stimulation of fatty acid for-mation was produced by the addition of citrate, even though carbon atoms from that tricarboxylic acid were not incorporated into fatty acids in that system. The role of acetyl-coenzyme A was examined but not conclusively demonstrated in those initial experiments (6,7).
Shortly after the discovery of the water-soluble enzyme system, a serious interruption of the investigations occurred. I was called to active duty in the Medical Corps of the United States Navy from 1952 to 1954. I was put in charge of the clinical chemistry laboratory at the National Naval Medical Center in Bethesda, Maryland. It was very difficult to carry out research under those circumstances, although several publications emanated from studies carried out there (8,9). I attempted to expand on the pigeon liver fatty acid-synthesizing enzyme system after hours and on the weekends as a guest worker in Earl Stadtman's laboratory in the then-National Heart Institute at the National Institutes of Health (NIH). This effort was disappointingly unproductive. Stadtman suggested that I undertake a project on enzymatic thioltransacetylation. That investigation involved the use of acetyl-coenzyme A and resulted in the demonstration of three distinct thioltransacetylases in extracts of pigeon liver and Clostridium kluyveri (10).
My tour of active duty in the Navy ended in September 1954. I was hired by Seymour Kety, director of intramural research in the combined basic research laboratories of the then-National Institute of Neurological Diseases and Blindness and the National Institute of Mental Health, where I resumed my investigations of the biosynthesis of long-chain fatty acids. At the 1958 Gordon Research Conference on Lipide [sic] Metabolism, Salih Wakil of the University of Wisconsin reported on studies he was carrying out with the pigeon liver fatty acid-synthesizing enzyme preparation that Gurin and I had discovered. In the presence of the 135 attendees, Wakil said that there was an absolute requirement for bicarbonate for the synthesis of long-chain fatty acids, similar to our findings using slices of rat liver tissue (2). Norman Radin of the University of Michigan stood up and asked Wakil if he had ever considered malonyl-coenzyme A as an intermediate. Wakil replied that he had not. The night before Wakil's talk, Feodor Lynen and I had a conversation in my room and discussed the enzymatic synthesis of long-chain fatty acids. The discussion included the possibility of the participation of malonyl-coenzyme A in this process. After Radin's comment, that possibility was common knowledge. I felt that anyone at that presentation was at liberty to do experiments with malonyl-CoA. I probably had a slight advantage. I knew that heart muscle could metabolize malonic acid, and I presumed that it did by making malonyl-CoA. I was right. I synthesized malonyl-CoA using an extract of an acetone powder of pig heart. When I added malonyl-CoA to the pigeon liver enzyme in an incubation system that contained acetyl-CoA and reduced triphosphopyridine nucleotide, there was a remarkably rapid formation of long-chain fatty acids (see Fig. 2 in Ref. 11). This was the first demonstration of the involvement of malonyl-CoA in the biosynthesis of long-chain fatty acids. I erroneously deduced that the condensation between acetyl-CoA and malonyl-CoA was a Knoevenagel type of reaction. I later demonstrated that the reaction was actually a Claisen-type condensation (12).
Because of my appointment in the National Institute of Neurological Disorders and Stroke, I gradually turned my attention to lipids that were prominent in the central and peripheral nervous systems. It had been known for nearly three-quarters of a century that the major lipid on a weight basis in the brain and in myelinated nerves was galactocerebroside, which consisted of the long-chain amino alcohol sphingosine (Fig. 1A), a long-chain fatty acid, and one molecule of galactose (13). I initiated my studies in this area with an investigation of the enzymatic synthesis of sphingosine (14). The reaction is catalyzed by an enzyme complex in microsomes. Palmitoyl-CoA condenses with carbon 2 of serine in the presence of pyridoxal phosphate and Mg 2ϩ or Mn 2ϩ ions, forming a Schiff base-metal complex. A reducing material, such as TPNH, is required. Carbons 1 and 2 of sphingosine arise from carbons 3 and 2 of serine. The carboxyl carbon of serine is lost in this condensation. It must be present initially because ethanolamine does not participate in the formation of sphingosine.
I began to investigate the formation of glycosphingolipids and found an enzyme in microsomes of young rat brain tissue that catalyzed the incorporation of radioactive glucose and galactose into cerebrosides using uridine diphosphate galactose as the hexose donor (15). The formation of cerebrosides was of particular interest because it had been known since 1924 that Gaucher disease, the most prevalent hereditary metabolic storage disorder of humans, was characterized by an accumulation of cerebroside. The German physician H. Lieb thought that it was galactocerebroside (16). However, the optical rotation of the aqueous cleavage product was incompatible with that deduction. In 1934, the French chemist A. Aghion correctly identified the accumulating glycolipid as glucocerebroside (Fig. 1B) (17).
Patients with Gaucher disease experience a number of problems that can lead to death. Among them are massive enlargement of the spleen and liver; severe anemia; low blood platelets, leading to easy bruising and frequent hemorrhages; reduced white blood cell count; skeletal damage, including Erlenmeyer flask deformity of the distal femur; multiple fractures of the hips, spine, and elsewhere; and a predisposition to multiple myeloma. From 1934 until the early 1960s, there was considerable speculation about the metabolic basis of Gaucher disease (18). I undertook studies to try to resolve this dilemma. Because of the possibility of an error of carbohydrate metabolism that resulted in substitution of glucose for galactose in the cerebrosides of patients with Gaucher disease, the first investigation was an examination of galactose tolerance in a patient with Type 1 Gaucher disease. It was anticipated that, if patients could not carry out the intermediary steps required for galactose metabolism, including the formation of UDPgalactose, their cerebrosides would contain only glucose. This appeared not be the case because galactose tolerance was found to be normal in a patient with Gaucher disease.
The next experiments to identify the metabolic basis of Gaucher disease were performed with surviving slices of spleen tissue obtained from patients with the disorder who underwent splenectomy because of severe hematological difficulties. Investigations with [ 14 C]glucose and [ 14 C]galactose revealed that both of the hexoses were efficiently utilized as precursors of both glucocerebroside and galactocerebroside, thereby eliminating an abnormality in the pathway of cerebroside synthesis in Gaucher disease. The rate of glucocerebroside formation was then compared in spleen slices obtained from patients with Gaucher disease with the rate in similar preparations from two patients with Niemann-Pick disease and one patient with idiopathic thrombocytopenic purpura. That investigation revealed that there was no increase in the rate of glucocerebroside formation in the tissues derived from the patients with Gaucher disease. Those observations led to the prediction that a defect in glucocerebroside catabolism was the basis of Gaucher disease (19).
At that time, no information was available concerning the catabolism of cerebrosides. Experiments were therefore undertaken to examine the enzymatic breakdown of glucocerebroside. The first investigation was an attempt to detect the release of free glucose from unlabeled glucocerebroside that had been isolated from the spleens of patients with Gaucher disease. No increment of free glucose that might have originated from glucocerebroside could be detected in surviving animal or human spleen tissue preparations. The reason for this inability to detect cerebroside catabolism in this manner was that glucose is extensively metabolized and largely converted to CO 2 under those conditions. A second attempt to examine glucocerebroside metabolism was carried out by radiolabeling glucocerebroside throughout the molecule with tritium. That approach was accomplished by exposing unlabeled glucocerebroside to 3 H in a sealed vessel for several days: the Wilzbach technique. Under those conditions, covalently bound hydrogen atoms were replaced with 3 H. However, the background radioactivity from [ 3 H]glucocerebroside prepared by that procedure was too great to permit metabolic investigations in vitro.
To overcome these obstacles to investigate glucocerebroside catabolism, I decided that the chemical synthesis of glucocerebroside should be undertaken. I read an article by David Shapiro and H. M. Flowers at the Weizmann Institute of Science in Rehovot, Israel, describing the chemical synthesis of sphingomyelin (20). I wrote to Shapiro telling him that I would like to go to Israel and work with him on the preparation of radioactive sphingomyelin so that we might use it to identify the metabolic defect in Niemann-Pick disease, in which excessive quantities of that lipid accumulate. He wrote back saying that he did not have access to the use of radioactive materials, but, if I could find support, he would come to the United States and help me prepare radioactively labeled glucocerebroside to look at its metabolism in Gaucher disease. I approached Richard Masland, my institute's director, and he provided Shapiro with modest support for the project. Shapiro came to the NIH and, working with Julian Kanfer, synthesized two preparations of [ 14 C]glucocerebroside. Radioactivity was introduced into the fatty acid moiety in one preparation and into the glucose portion in the other. Experiments with these labeled lipids revealed that all mammalian tissues contain an enzyme called glucocerebrosidase (glucosylceramide ␤-glucosidase) that catalyzes the hydrolytic cleavage of glucose from glucocerebroside (21).
Experiments performed in 1964 revealed that the metabolic defect in Gaucher disease is a deficiency of glucoce-rebrosidase (22). Those initial observations were substantiated in a more extensive study carried out the next year (23). Patients with Type 1 (non-neuronopathic) Gaucher disease exhibited glucocerebrosidase activity that averaged ϳ11% of the normal activity. Patients with Type 2 (acute neuronopathic) Gaucher disease had extremely low glucocerebrosidase activity (1-3% of the normal activity), and patients with Type 3 (subacute neuronopathic) Gaucher disease had glucocerebrosidase activity that was generally between that in Types 1 and 2. The deficiency of glucocerebrosidase in Gaucher disease has been universally confirmed. In time, it was shown that glucocerebrosidase is primarily a lysosomal enzyme (24). Because the highest concentration of the glucocerebrosidase activity is in lysosomes, Gaucher disease is properly classified as a lysosomal storage disorder. Maximal catalytic activity of this enzyme occurs at pH 5.5-5.9.
This discovery was of primary importance concerning the underlying cause of lysosomal storage disorders. In fact, the American Chemical Society had a cover story depicting it with me in tennis clothes on the cover (25). Once the etiology of Gaucher disease was established, the enzymatic defects in other hereditary metabolic storage disorders were rapidly elucidated. These included Niemann-Pick disease in 1966 (26); Fabry disease in 1967 (27), and Tay-Sachs disease in 1969 (28,29). Practical benefit soon followed. Facile tests were developed to identify patients based on enzyme assays in white blood cells (30) and cultured skin fibroblasts (31), the detection of heterozygous carriers of these disorders (32), and the prenatal diagnosis of these conditions (33)(34)(35).
The year after the discovery of the enzymatic defect in Gaucher disease, I proposed that enzyme replacement therapy should be investigated to treat patients with metabolic storage disorders (36). I wanted to use a human source of the missing enzymes, and one evening I thought of the placenta. The next day, I went into the laboratory, homogenized some fresh placental tissue, and found that it did contain several sphingolipid hydrolases. The quantities were not great, but I began to work with this source. The first enzyme that a colleague, William Johnson, and I obtained in a reasonable quantity was ceramide trihexosidase (␣-galactosidase A) (37). The enzyme is lacking in patients with Fabry disease, the second most prevalent metabolic storage disorder of humans (27). When ceramidetrihexosidase was injected into patients with Fabry disease, there were significant reductions of ceramidetrihexoside (globotriaosylceramide, Gb3) in the blood (38). We were not permitted to perform organ biopsies in that investigation. We were a bit discouraged by the fact that ceramide trihexoside in the blood retuned to pre-infusion values by 48 h. Further studies in patients with Fabry disease were considerably delayed because of difficulties in purifying larger quantities of ceramidetrihexosidase.
My associates and I eventually succeeded in purifying small amounts of glucocerebrosidase from human placenta (39). When the enzyme was injected into two patients with Gaucher disease, there were significant reductions of glucocerebroside in the livers and blood of the recipients (40). A particularly gratifying finding in the investigation was the slow return of glucocerebroside to pre-infusion levels in the blood of the recipients (41). The finding raised the hope that, if sufficient enzyme were available, a therapeutic benefit might be obtained. Although encouraged by those observations, two major impediments to successful enzyme replacement therapy for patients were immediately encountered. My colleagues and I worked for one year to obtain 9 mg of purified placental glucocerebrosidase. When the enzyme was injected into a third patient with Gaucher disease, we did not observe a significant reduction of glucocerebroside in the liver, and there was no decrease of glucocerebroside in the circulation. She had accumulated 20 times more glucocerebroside in the liver than the first patient and 10 times more than the second. It was apparent that we would need much larger quantities of the enzyme to expect consistent benefit. A purification procedure based on hydrophobic column chromatography was developed (42). Although the enzyme was obtained in good yield and was perfectly active with [ 14 C]glucocerebroside as substrate, another potentially formidable problem was encountered. Glucocerebrosidase purified in this fashion was taken up primarily by hepatocytes in the liver, in which there is no accumulation of glucocerebroside. The lipid accumulates in macrophages, such as the Kupffer cells in the liver. We needed to devise a strategy to divert glucocerebrosidase from hepatocytes to macrophages. Three important lessons were learned. 1) Glucocerebrosidase and other lysosomal enzymes are glycoproteins. Glucocerebrosidase has four oligosaccharide side chains. 2) Macrophages have a lectin on the plasma cell membrane with high affinity for mannose-terminal glycoproteins.
3) The molecules of mannose on placental glucocerebrosidase are shielded from interacting with the lectin by three monosaccharides: N-acetylneuraminic acid, galactose, and N-acetylglucosamine. To expose the mannose residues, glucocerebrosidase was treated sequentially with three exoglycosidases: neuraminidase, ␤-galactosidase, and hexosaminidase (43). Glucocerebrosidase modified in that manner is endocytosed by macrophages 50 times more effectively than unmodified glucocerebrosidase. Administration of ade-REFLECTIONS: Unearthing a Biochemical Rosetta Stone DECEMBER 31, 2010 • VOLUME 285 • NUMBER 53 quate quantities of mannose-terminal glucocerebrosidase (Fig. 2) produced remarkable benefit in patients with Gaucher disease (44,45). The size of the enlarged spleen and liver was reduced, usually returning to their normal sizes. Hemoglobin, white blood cells, and blood platelets increased. They also usually returned to normal values. Skeletal abnormalities improved, and they, too, often returned to normal. I hoped that we might be able to stop the progression of the disease, and I was quite surprised by the virtually complete reversal of all of the manifestations of the condition after this treatment (Fig. 3).
Based on these findings, enzyme replacement therapy for patients with Gaucher disease was approved by the United States Food and Drug Administration (FDA) in 1991. Glucocerebrosidase was subsequently produced recombinantly in Chinese hamster ovary cells. That product also was treated with the three exoglycosidases cited previously to target it to macrophages. Recombinant glucocerebrosidase was found to be just as effective as the placental enzyme (46). It was approved by the FDA for use in patients with Gaucher disease in 1994. At the time of this writing, more than 5,600 patients with Gaucher disease throughout the world are benefiting from enzyme replacement therapy. Enzyme replacement therapy has been approved also for patients with Fabry disease and four additional lysosomal storage disorders. I am certain that readers of these reflections will have acquired a clear concept of how arduous and time-consuming these investigations were. I feel very fortunate to have had the support of my co-workers, my institute directors at the NIH, and the members of many scientific review boards during the lengthy period required to bring enzyme replacement therapy to fruition.
Editor's Note-The work reported in this Reflections article is of dramatic importance in indicating how carefully performed basic biochemical studies can lead to dramatic therapeutic successes.