Bacterial Production of Human Insulin

Human insulin derived from bacteria is the first clinically proven product of recombinant DNA technology. I say proven because successful clinical trials of bacterially produced insulin were reported in the August 23, 1980, issue of Lancet. This represents an exciting milestone and, although not generally recognized, it is interesting to note that Dr. Levine played a key role in initiating the human insulin project. I will explain this statement later on, but first I will review some of the essential technical aspects of the project. As illustrated in Figure 1, two separate bacterial strains were constructed. One strain carried a man-made gene for the A-chain of human insulin and the other strain carried a gene for the B-chain of human insulin. Inside the bacterial cells, the insulin chains are made as tails on a rather large precursor protein, the enzyme beta-galactosidase. The insulin peptide chains are efficiently clipped from the precursor protein by treatment, in vitro, with cyanogen bromide. After purification of the separate insulin chains they are joined to give complete, active insulin. Some additional details of each of the various aspects of the project will be discussed more fully. A common misconception is that we cloned the human insulin gene and used it to direct the production of human insulin in Escherichia coli. We did not. As illustrated in Figure 2, there are two approaches to molecular cloning of genes for expression in bacteria. The most common approach starts with messenger RNA (mRNA) and uses reverse transcriptase to make a DNA copy (cDNA), which is then cloned. Our work used an entirely different approach. The structure and amino acid sequence of insulin has been known for many years. Therefore, we were able to use the genetic code to design, on paper, two small genes that contained the proper sequence of nucleotides to command a bacterial cell to produce human insulin. These genes are shown in Figure 3. Because the genetic code is degenerate (that is, there are usually several codons corresponding to each amino acid), the genes we designed and made have a sequence of nucleotides different from the natural coding sequences of human insulin. Yet these man-designed and man-made genes function very well and the bacterial strains containing them produce large amounts of human insulin chains. We were not surprised that the insulin genes functioned, because we had earlier demonstrated the feasibility of this approach by producing somatostatin in E. coli.