One of the primary scientific aims of the Open Insulin Foundation is to engineer a strain of microorganism that can produce sufficient quantities of insulin glargine for downstream purification and formulation. Toward this end, we took inspiration from a published protocol and came up with a strategy to engineer the budding yeast Pichia pastoris with the capacity to inducibly express glargine and secrete it into the extracellular growth media. P. pastoris presents numerous advantages for producing this particular protein over the bioengineer’s typical microbe of choice, Escherichia coli, including the ability to form disulfide bonds, grow in acidic media, and produce proteases necessary for downstream processing. However there is also a cost: while genetically manipulating E. coli generally only involves a simple plasmid transformation in which parameters like copy number and expression strength are encoded within the plasmid, the analogous procedure for P. pastoris requires genomic integration, which is more difficult to precisely control. Genomic integration relies on the cell’s machinery for attempting to repair catastrophic DNA damage, so there is a random element to exactly where the engineered gene gets inserted and how many copies make it in. Thus, we need a method to screen through many of these random insertion events and recover the ones that happen to be best for glargine production.
Following many experimental iterations, our volunteers have arrived at a working screening strategy that balances throughput, sensitivity, and cost. This comes from 3 steps of increasing stringency: 1) metabolic selection using a prototrophic marker, 2) antibiotic resistance screening, and 3) secreted glargine measurement using an enzyme-linked immunosorbent assay (ELISA).
First, plasmids containing the gene encoding glargine as well as the other selection markers are cut with a restriction enzyme, resulting in linear DNA fragments with homology to a specific location in the P. pastoris genome on both ends. We then insert these DNA fragments into P. pastoris cells with an electric current, and then the yeast homologous recombination machinery take over to splice the DNA fragments into the genome. We spread the cells on agar dishes with media lacking the amino acid Histidine. Only cells that successfully integrated the DNA should grow, since it includes a gene that allows the cell to synthesize its own Histidine. This is the basis for the first step of screening, metabolic selection.
Once the engineered cells have grown and reproduced enough to form visible colonies on the agar, we can pick individual colonies to transfer to small liquid cultures for the second stage of screening: antibiotic resistance. While the first step was an all-or-nothing selection, growing the colonies in different concentrations of antibiotic provides a more quantitative measurement. The integrated DNA contains a gene for antibiotic resistance, so if more copies happen to be inserted in a given cell, that should provide resistance to a higher concentration of antibiotic than if fewer copies were inserted. Thus we can use this second screening step to narrow down candidate strains to those with the highest insert DNA copy numbers, which will likely produce the most protein.
Though gene copy number measurements are often a good predictor of protein expression, they are not a direct measurement of a strain’s capacity to produce and secrete the desired protein. Therefore our third step is a direct measurement of secreted protein using ELISA. We grow the high-copy number colonies identified from the previous stage in another liquid culture, but this time add methanol to induce expression of the gene for glargine. After a few days of induction, we centrifuge the cultures to separate the cells from the media. Because we engineered the glargine gene to include a secretion signal, we take samples of the isolated media for glargine measurements. These samples incubate with antibodies that specifically bind to glargine. If antibody-glargine binding does occur in a sample, then it causes a chemical reaction that changes the color of the sample, which we can observe by eye or measure quantitatively with a spectrophotometer. Such direct detection of glargine is the final confirmation that the genetic engineering was successful for a given strain.
Currently, we observe very close agreement between each candidate strain’s degree of antibiotic resistance and the amount of signal it produces in the ELISA experiment. This is great news for a community laboratory, because antibodies for ELISA are by far the most expensive component of the whole procedure. With antibiotic resistance measurements serving as a reasonably accurate prediction for protein production, we can greatly increase our screening throughput at this stage in a cost-effective manner.