Blue Colonies, White Colonies and an Insulin That Did Not Work
A Pune laboratory is asked to make recombinant human insulin in Escherichia coli. The team PCR-amplifies the insulin gene from a human genomic DNA library using primers flanking the coding sequence, ligates the product into pUC19 cut with EcoRI, and transforms DH5-alpha. The plating medium contains ampicillin, IPTG and X-gal. Of 100 colonies, 62 are blue and 38 are white. Ten white colonies are checked by colony PCR with primers that anneal 150 bp inside the insert; no product appears in any of the ten, although the positive-control plasmid gives a strong band. Separately, a second team used a cDNA clone in pET-28a, induced with IPTG, and saw a strong band near 6 kDa on SDS-PAGE; the purified protein did not lower blood glucose when injected into mice. The laboratory has reagents and time for one more round of experiments and a budget of about 1.5 lakh rupees.
1. Account for the 62 blue and 38 white colonies, and explain what the colony PCR result means for this cloning strategy.
Read the worked answer
The colour screen
pUC19 carries lacZ-alpha; DH5-alpha supplies the omega fragment, so alpha-complementation gives active beta-galactosidase. Active enzyme cleaves X-gal and the colony turns blue. An insert at the multiple cloning site, which lies inside lacZ-alpha, disrupts the reading frame and the colony stays white. Blue therefore means vector with no insert; white means a putative recombinant.
Sixty-two per cent blue is a poor ratio, pointing to incomplete digestion, failure to dephosphorylate the cut vector, or a low insert-to-vector ratio, so that vector recircularises on its own and gives blue colonies.
The conflict and its resolution
Colony PCR with primers 150 bp inside the insert should amplify from any true recombinant regardless of orientation. Ten out of ten white colonies gave no product while the control gave a strong band, so the white phenotype is not being produced by the intended insert. The colour screen is only a surrogate for the desired genotype, and the direct test overrides it. The usual explanation is that EcoRI digestion was followed by nuclease nibbling of the AATT overhangs, after which the vector recircularised with a small deletion; the lacZ-alpha frame is still broken, so the colony is white, but there is no insert to amplify.
What to change
Dephosphorylate the vector, gel-purify the insert, use a high insert-to-vector ratio, use two different enzymes for directional cloning, include a no-insert ligation control, and screen by PCR rather than trusting colour.
2. Why is a genomic library the wrong starting material for expression in E. coli, and why did the 6 kDa band fail to lower blood glucose?
Read the worked answer
Wrong starting material
The human insulin gene contains two introns. E. coli has no spliceosome and cannot remove introns, so a genomic clone cannot yield a correct mRNA-like coding sequence, and the protein made is aberrant or truncated. A cDNA clone, made by reverse transcription of mature beta-cell mRNA, is the correct starting material. This alone explains why the first approach was never going to give insulin.
Why the 6 kDa protein was inactive
Insulin is made as preproinsulin (110 amino acids). The 24-residue signal peptide is removed in the endoplasmic reticulum; the remainder folds as proinsulin with three disulphide bonds (A6-A11 within the A chain, and A7-B7 and A20-B19 between chains), all formed in the oxidising ER with protein disulphide isomerase. In the Golgi, PC1/PC2 remove the 31-residue C-peptide, leaving A chain (21 residues) and B chain (30 residues), about 51 residues or 5.8 kDa.
A band at 6 kDa is roughly the right size, so the chains were probably made, but E. coli cytoplasm is reducing, lacks PC1/PC2 and signal peptidase, and cannot form the interchain disulphides. Insulin with free thiols or wrong disulphides cannot bind the receptor, so the mouse assay would fail even though SDS-PAGE looked correct. Note also that SDS-PAGE under reducing conditions would not reveal the missing disulphides.
The fix
Express A and B as separate fusions, or as a single chain with a cleavable linker; secrete to the periplasm with a pelB signal, or use an oxidising-cytoplasm strain; refold and oxidise in vitro; or switch to a yeast host that secretes correctly processed insulin.
3. With only one experimental round left, state the single most informative experiment and the expected result.
Read the worked answer
The decisive experiment
The most informative and least expensive test is to link the two failures into one experiment: reclone the cDNA (not the genomic PCR product) into an expression vector that directs the protein to the periplasm, or into an oxidising-cytoplasm strain, and analyse the product under both reducing and non-reducing conditions.
Practical design
- Ligate the cDNA into pET-28a with a pelB leader and a His tag, using two different restriction sites so orientation is fixed; transform an expression strain (or a trxB/gor oxidising strain).
- Plate on the selecting antibiotic and confirm the insert by colony PCR with insert-specific primers; this time the PCR is the primary screen, not the colour.
- Induce with IPTG at 16–20 degrees Celsius overnight. Run induced and uninduced samples side by side on SDS-PAGE, with and without dithiothreitol.
- Purify by nickel affinity and test for glucose-lowering in mice, alongside a commercial insulin standard.
Expected result and interpretation
If disulphide formation was the problem, the non-reduced sample will show a compact band that shifts on reduction, and the purified protein will lower glucose. If the band appears but activity is still absent, the remaining explanation is failure of C-peptide processing, and the next step is a yeast host, not more E. coli work. This single experiment distinguishes the two hypotheses and also confirms the cloning has finally worked.