The Three Stages of My Experience in Discovering the Mode of Action of Penicillin
Once penicillin came into common use, it became imperative to figure out how it works. This knowledge might be expected to lead to the development of better antibiotics or a more effective use of their therapeutic power. In actuality, the research effort directed to understanding the mode of action of penicillin led to major insights into basic biological phenomena. We are delighted to be able to share this personal perspective by one who made many of the key discoveries in this field.
by Ted Park,
aka James T. Park
Ted Park is Professor Emeritus, Department of Molecular Biology and Microbiology, Tufts University.
Act 1: Precursors of an unknown, and presumably essential, metabolic process accumulate in penicillin-treated Staphylococcus aureus cells.
The scene opens in 1943. I was a graduate student in fermentation biochemistry at the University of Wisconsin in Madison. As this was at the height of World War II, most research projects were war-related. My first assignment was no different: to identify mutants of Aerobacter aerogenes that produced improved yields of 2,3-butylene glycol (2,3-butanediol), a potential candidate for conversion to synthetic rubber. It was not a terribly exciting project.
My fortune changed when a new assignment came to the lab. Penicillin had recently been shown to be a miracle drug. Production had started in England by growing the mold, Penicillium notatum, as mycelial mats on the surface of media in individual bottles—hardly an efficient method for mass production. The new project was to produce penicillin by growing the mold in submerged culture in large vats, and I was able to sign on to the team. My job was to develop a turbidometric assay for penicillin. It was easy enough to produce a beautiful standard curve of final turbidity versus penicillin concentration using Staphylococcus aureus 209P. However, the fermentation broths varied in their effects—because, as we learned much later, they contained multiple penicillins with different fatty acid side chains with different potencies. Somewhere along the way I managed to pick up an S. aureus 209P infection in my eye. It was treated successfully with a mercury salt cream, penicillin not yet being available for the general public.
The next year, I volunteered to become an officer in the Navy, which seemed preferable to being drafted as a private into the Army. I soon found myself one of 900 at the 'Ship of the Desert' in Tucson, Arizona. Two months later, as a freshly minted ensign, I was almost the last of the 900 to receive orders! I was to report to Camp Detrick, Maryland, a place I'd never heard of. It turned out that a fair number of Navy personnel with scientific backgrounds, myself included, were assigned to this Army research unit studying biological warfare. I spent the remainder of my 2 years in the Navy trying to develop a vaccine to protect against one of the organisms under study.
During this time, I came across an interesting article by Chain and Duthie about the mode of action of penicillin. Their results indicated that, contrary to prevailing thought, penicillin did not interfere with respiration and the production of ATP. In the back of my mind I stored the naïve—and at that time unlikely—notion that maybe penicillin interfered with some essential biosynthetic process. Towards the end of my tour of duty, I learned that money was available from the US Army to support education and research. I wrote to my former thesis adviser, Marvin Johnson, and told him that I would like to return to study under him, but that I wanted to investigate the mode of action of penicillin. Would he agree to this if I supported myself with a small grant from the Army?
He did, and I returned to Wisconsin and began my search for a penicillin-sensitive biosynthetic process. Protein synthesis and DNA synthesis, as measured by the color tests of the day, were not affected, RNA synthesis only slightly. I did find, however, a dramatic accumulation of soluble, acid-labile, phosphate-containing compounds in the penicillin-treated S. aureus H cells. I turned my attention to identifying them. At an early stage of purification, I determined that the compound may have one acid-labile and one stable phosphate, that it had an adsorption spectrum similar to uracil, and that it contained a reducing sugar that was readily released by heating with dilute acid. Paper chromatography in a variety of solvents revealed that there were actually three compounds that could be detected by UV and phosphate content, and were positive in the Morgan-Elson test for N-acetylamino sugars. But to make further progress, I needed more material.
To this purpose, I grew 60 liters of S. aureus culture in four large carboys. The cultures were treated with penicillin at mid-log phase and harvested 30-40 min later in a Sharples centrifuge. The centrifuge produced aerosols of S. aureus H and I harbored staphylococci in my nose for the next 15 years (until treatment of a chest infection with penicillin cured me). Sharples went out of business a few years later, to the detriment of bacterial physiology studies in general (but possibly also to a decrease in aerosols!).
A culture of a penicillin-producing Penicillium and
the structure of penicillin G. Source.
I also needed a method to separate the three compounds on a larger scale. Paper chromatography would hardly do. I learned from Marv Johnson that the same separation could be achieved using column chromatography. Hence I used phenol-saturated sulfuric acid buffer at pH 2 supported in diatomaceous earth (celite 545) as the stationary phase and buffer-saturated phenol as the mobile phase. This approach provided me with milligram quantities of somewhat purer material, using which I was able to determine that one compound consisted to uridine diphosphate attached to an N-acetyl-amino sugar. An exciting moment came when titration of the purified amino sugar revealed that it contained a carboxyl group. The second compound also contained L-alanine; the third, and by far the most abundant, had additional nitrogen-containing material later identified as L-lysine, D-glutamic acid and 2 additional alanine molecules. Another dramatic moment occurred when I saw that the glutamic acid isolated from that third compound rotated polarized light in the opposite direction to that of L-glutamic acid. It was subsequently shown that this compound contained L-ala-γ-D-glu-L-lys-D-ala-D-ala. This work was published in three papers in 1952 (click here, here, and here). (Ed. Note: These compounds became widely known as "Park nucleotides.")
We surmised that penicillin interfered with a hypothetical reaction that resulted in the accumulation of this novel compound that contained D-amino acids as well as an amino sugar with a carboxyl group. But what was the biological role of this mysterious—but key—product that accumulated due to the action of penicillin? Or was it even a natural compound? After all, it contained two unnatural amino acids and an unknown amino sugar.
Act II: Identification of cell wall peptidoglycan as the product of the penicillin-sensitive reaction.
At this time (1949-1952), knowledge of the cell wall of bacteria was scant. In fact, one review at that time postulated that the wall might consist of chitin, since only glucosamine could be identified in hydrolysates. Enter the pioneering work of Milton Salton. He purified walls from Gram-positive bacteria ruptured by shaking with small glass beads in a "Mickle disintegrator" (a fancy shaker, really) and showed that only a few amino acids were present, including alanine, glutamic acid, and lysine. Also, a 1952 paper by Mitchell and Moyle reported that S. aureus cell walls contained alanine, lysine, glutamic acid, and glycine, but no amino sugars. It should have been obvious to me from these results that the major compound accumulating in penicillin-treated S. aureus was a cell wall precursor.
But no. I didn't pursue this at the time, partly because the walls were said to lack amino sugars, and partly because I did not have a Mickle disintegrator available. It wasn’t until four years later that Jack Strominger and I independently made the connection and reported at an ACS meeting that penicillin interfered with the synthesis of the cell wall, thus causing the accumulation of the uridine compounds that must be cell wall precursors. I suggested that we publish together, which we did in Science. Since cell walls are essential for the survival of these bacteria but are lacking in mammals, penicillin kills growing bacteria but is non-toxic to mammals. But what is the specific reaction that is inhibited by penicillin?
Act III: Identification of a penicillin-sensitive transpeptidation reaction required to produce an intact, cross-linked cell wall.
Another six years passed. More detailed and quantitative data on the composition of the cell walls of Gram-positive bacteria accumulated. It finally dawned on me one Saturday evening that, since Gram-positive cell walls contained only a little over two alanine molecules per repeating unit and the precursor contained three, perhaps the energy of the D-ala-D-ala bond was used to drive a hypothetical transpeptidation reaction that cross-linked the glycan chains in the cell wall. I actually went into the lab the next morning and did an experiment that revealed that a penicillin-sensitive transpeptidation reaction was involved in cell wall synthesis. Thus, I achieved the third step in elucidating the mode of action of penicillin: the identification of a specific enzymatic reaction that was sensitive to penicillin. Ed Wise and I published our results in 1965. Tipper and Strominger submitted a paper three months later confirming our conclusions and reporting the additional observation that penicillin probably competes with D-ala-D-ala for the active site of the transpeptidase.
In the 44 years since this discovery, much has been learned about the effects of various penicillins and related β-lactam antibiotics. Among the most surprising developments was the revelation that bacteria possess multiple proteins—often 5 to 10—that are sensitive to β-lactam antibiotics: the penicillin-binding proteins or PBPs. These typically include several transpeptidases and D-ala-D-ala carboxypeptidases. Penicillin binds to the active site, thus inactivating the enzyme. There is no argument about that. At its minimal inhibitory concentration (MIC), any given β-lactam antibiotic binds to several different PBPs. However, a caveat: it is not generally appreciated that just because a particularβ-lactam at its MIC does not bind to a specific PBP does not mean that the enzyme in question is not inhibited. As Guiseppi Botta and I discovered in 1980, the penicillin analog furazlocillin, at 10 μg/ml, binds only to PBP3 of E. coli. However, at this concentration, the activity of the D-ala-D-ala carboxypeptidases (PBP5and PBP6) is inhibited by 98%. Since there is not a convenient assay for the activity of the other PBPs, the question remains open whether a particular β-lactam at its MIC concentration is inhibiting other PBPs without binding to them.
Looking back from today's vantage point to our first sally into the field of antibiotics, it is certainly satisfying to note that the basic mode of action of penicillin that we uncovered remains indisputable: inhibition of a transpeptidation reaction required for production of a functional cell wall during bacterial growth and division.