Dissecting and tuning primer editing by proofreading polymerases

Abstract

Proofreading polymerases have 3' to 5' exonuclease activity that allows the excision and correction of mis-incorporated bases during DNA replication. In a previous study, we demonstrated that in addition to correcting substitution errors and lowering the error rate of DNA amplification, proofreading polymerases can also edit PCR primers to match template sequences. Primer editing is a feature that can be advantageous in certain experimental contexts, such as amplicon-based microbiome profiling. Here we develop a set of synthetic DNA standards to report on primer editing activity and use these standards to dissect this phenomenon. The primer editing standards allow next-generation sequencing-based enzymological measurements, reveal the extent of editing, and allow the comparison of different polymerases and cycling conditions. We demonstrate that proofreading polymerases edit PCR primers in a concentration-dependent manner, and we examine whether primer editing exhibits any sequence specificity. In addition, we use these standards to show that primer editing is tunable through the incorporation of phosphorothioate linkages. Finally, we demonstrate the ability of primer editing to robustly rescue the drop-out of taxa with 16S rRNA gene-targeting primer mismatches using mock communities and human skin microbiome samples.

Figure 1.

Synthetic DNA standards for measuring primer editing activity. (A) Design of primer editing standard constructs. Each standard plasmid contains a copy of the E. coli 16S rRNA gene V4 region with a specific modification to the V4_515F primer binding region, as well as a REcount PCR-free barcode quantification construct for quantifying the abundance of each standard plasmid in the pooled mixture. (B) The synthetic standard pool contains every possible single base substitution in the last 10 bp of the V4_515F primer binding region. (C) Measured abundance of edited forward primer sequences when amplified with an E. coli-specific 515F/806R primer set and KAPA HiFi polymerase (solid line; n = 3, error bars = ± SEM)). Abundance of standard plasmids in the standard pool as assessed by REcount (dashed line). (D) Measured abundance of edited primer reverse sequences when amplified with an E. coli-specific 515F/806R primer set (n = 3, error bars = ± SEM).

Figure 2.

Assessing the effect of different polymerases, polymerase concentration, and sequence identity on primer editing. (A) Ratio of observed versus expected edits position in the forward primer sequence when amplified with an E. coli-specific 515F/806R primer set and the indicated polymerase. (B) Ratio of observed versus expected edits position in the forward primer sequence when amplified with an E. coli-specific 515F/806R primer set and KAPA HiFi polymerase at the indicated enzyme concentration. (C) Ratio of observed versus expected edits for each individual base of the forward primer sequence when amplified with an E. coli-specific 515F/806R primer set and 1× KAPA HiFi polymerase (n = 3, error bars = ± SEM).

Figure 3.

Reciprocal editing of mutant and wild type primers. Comparison of edited wild type E. coli-specific 515F primers used to amplify the primer editing standards and a collection of mutant primers which mirror the composition of the primer editing standards used to amplify an E. coli wild type template using either (A) KAPA HiFi polymerase (n = 3, error bars = ± SEM) or (B) Taq polymerase (n = 3, error bars = ± SEM).

Figure 4.

Tuning of primer editing using phosphorothioate protection. Effect of incorporating phosphorothioate bonds into E. coli-specific 515F primers on extent of primer editing observed when the primer editing standards are amplified using A) KAPA HiFi polymerase (n = 3, error bars = ± S.E.M.); B) NEB Q5 polymerase (n = 3, error bars = ± S.E.M.); C) Phusion polymerase (n = 3, error bars = ± S.E.M.).

Figure 5.

Improving performance of ITS1 primers. Fungal isolates and soil samples amplified with unprotected (left, ITS1) or phosphorothioate protected (right, ITS1*) primers. Primer dimer bands are indicated by the arrow.

Figure 6.

Primer editing recovers C. acnes across multiple variable regions. (A) Percent C. acnes observed in the HM-276D mock community when amplifying the V4 variable region using the indicated polymerase. Expected abundance of C. acnes is indicated by the dashed line. (B) Schematic of the 16S rRNA gene with variable regions indicated. The status of primer mismatches to the C. acnes template sequence for the primers used to amplify the V1V3, V4, V3V4 and V4V6 variable regions is indicated. (C) Percent C. acnes observed in the HM-276D mock community when amplifying different variable regions using either KAPA HiFi polymerase or Qiagen Taq polymerase (n = 3, error bars = ± S.E.M.). Expected abundance of C. acnes is indicated by the dashed line.

Figure 7.

Recovery of C. acnes in skin microbiome samples using primer editing. (A) Percent Cutibacterium observed in face microbiome samples (n = 8) amplified with the indicated conditions. (B) Percent Cutibacterium observed in forearm microbiome samples (n = 8) amplified with the indicated conditions. (C) Percent Cutibacterium observed in armpit microbiome samples (n = 7) amplified with the indicated conditions. (D) Percent Cutibacterium observed in scalp microbiome samples (n = 8) amplified with the indicated conditions. (E) Microbiome profiles for face samples from eight donors amplified with either V4 or V1V3 primers using KAPA HiFi polymerase. Cutibacterium genus is shown in purple. For full legend and visualization of all skin microbiome data see Supplemental File 4. For box plots, mean is indicated by ‘X’, median is indicated by horizontal line, data points are indicated by dots, boxes span the second and third quartiles, and the data range (min and max) is indicated by the whiskers or outlier data points.