Our major aim of error-prone PCR was to find out whether our in silico mutagenesis and library construction provides higher efficiency in evolving better IDP compared to epPCR. Our error-prone PCR mainly focus on using (1) higher-than-usual concentration of Mg2+ to stabilize the mismatched base pairs during replication, and (2) Mn2+ to replace Mg2+ in some Taq polymerase’s active site to reduce the discrimination between matched and mismatched base pairs and increase misincorporation rate. To be the opposite of the biased, in silico selected mutation in our degenerate IDP sequence library, the epPCR needs to mutate as random as possible. Therefore, we did not introduce an unbalanced ratio of dNTP because that would favor the formation of certain codons more than others.
In a previous disordered peptide directed evolution paper (https://doi.org/10.1038/s41589-025-02128-3) they used GeneMorph II Random Mutagenesis Kit (200550) and repeated 4 rounds of epPCR. 12 nucleotides per kb per epPCR round was documented for this kit, so their total template mutation was around 48 nt / kb. In two more previous research paper (https://patents.justia.com/patent/6803216, https://genome.cshlp.org/content/genome/2/1/28.full.pdf), 7 mM of MgCl2 and 0.5 mM of MnCl2 with some unbalanced dNTP reached a mutation rate of around 6 nucleotide per kb per epPCR round. Therefore, we expected to use 7 mM MgCl2 and 0.5 mM MnCl2 and repeat at least 8 rounds of epPCR to reach the same 48 nt / kb level mutation.
However, in practical, due to our special, unusually short template and nearly 10 ℃ difference in annealing temperature between forward and reverse primer (we have very limited choice in primer design due to significant mispriming), 7 mM MgCl2, 0.5 mM MnCl2 and annealing at 57.7 C give very smeared product with nearly unidentifiable band of correct amplicon. We subsequently did (1) testing annealing temperature from 57.7C to 62.7C, found optimal annealing temperature at 61.7C (2) testing Mg2+ concentration from 2.5mM (normal PCR concentration) to 7mM, found maximal tolerable Mg2+ concentration at 4mM (3) testing Mn2+ concentration from 0.0mM to 0.5mM, found maximal tolerable Mn2+ concentration at 0.15 mM.
However, even after these changes, our epPCR product was improved only to a limited degree. The correct amplicon seems to appear more, but there are cases of failure as well. To further understand what causes the smear, we send a typical sample containing the smearing for sequencing to understand the composition of the by-products. We discovered that although the original SSB IDP template and the expected amplicon was only 170 bp, the size of the by-product was surrounding 224 bp. We supposed the primer annealed correctly at first, but then as Taq replicates a highly GC-rich repeats region downstream of the sense primer, it slips and started detaching or anneal to foreign/chimeric template, replicates foreign ~24 bp extra sequence, and re-anneals to the start of the SSB template to replicate the complete sequence. This introduces the extra length and therefore creates the smear shown on agarose gel (we tried post-staining and running gel in the cold room and the smear did not disappear, so the smear is more likely due to these unwanted by-product). To avoid this type of smear, we attempted different concentration of GC enhancer (from Phusion Plus kit, likely the major component is betaine), DMSO, or both. The GC enhancer was expected to equalize the spontaneity of AT and GC base pairing, and DMSO might weaken the hydrogen bonding for incorrect annealing. The optimal result was using only GC enhancer, and at 1X concentration specified by Phusion™ Plus PCR Master Mixes (F631S). We also proposed to shorten the extension time so the replication for longer by-product would be incomplete while the correct amplicon is still amplified. This later appears to be not effective, but shortening the extension time to 10 seconds seems to still improve and optimized the yield of correct amplicon and reduce smearing.
Finally, we also attempted reducing the primer concentration because we realized that negative control without the SSB IDP template still produced some bright band area <300 bp on the gel. Therefore, lower concentration of primer might decrease this type of smear. Thus, the optimal primer concentration was adjusted to 0.3 mM instead of standard 0.5 mM.
With all our improvements, the epPCR maximally can only be repeated for 3-4 rounds. This was consistent with number of epPCR rounds done by the previous disordered peptide directed evolution paper (https://doi.org/10.1038/s41589-025-02128-3). Even with multiple implementations of strategies to persist epPCR for more cycles and increase the desired, mutated amplicon at correct size, the epPCR product yield still started decreasing very significantly in round 4 or round 5 for both SSB IDP and NEXT IDP.
We therefore concluded that we have explored nearly the best potential of epPCR in generating mutants for our SSB and NEXT IDP. The sequencing results of epPCR mutated NEXT IDP was discussed in NEXT library. It showed that epPCR to its best performance did not create many variants with properties similar to our winner NEXT variant with in silico design. Despite known caveats, we proposed that epPCR possibly perform less efficiently in generating better IDP variant (i.e. more solubility-enhancing IDP variant) than the method of in silico mutation-selection we designed.