Compartmentalized partnered replication for the directed evolution of genetic parts and circuits

Abstract

Compartmentalized partnered replication (CPR) is an emulsion-based directed evolution method based on a robust and modular phenotype-genotype linkage. In contrast to other in vivo directed evolution approaches, CPR largely mitigates host fitness effects due to a relatively short expression time of the gene of interest. CPR is based on gene circuits in which the selection of a 'partner' function from a library leads to the production of a thermostable polymerase. After library preparation, bacteria produce partner proteins that can potentially lead to enhancement of transcription, translation, gene regulation, and other aspects of cellular metabolism that reinforce thermostable polymerase production. Individual cells are then trapped in water-in-oil emulsion droplets in the presence of primers and dNTPs, followed by the recovery of the partner genes via emulsion PCR. In this step, droplets with cells expressing partner proteins that promote polymerase production will produce higher copy numbers of the improved partner gene. The resulting partner genes can subsequently be recloned for the next round of selection. Here, we present a step-by-step guideline for the procedure by providing examples of (i) selection of T7 RNA polymerases that recognize orthogonal promoters and (ii) selection of tRNA for enhanced amber codon suppression. A single round of CPR should take ∼3-5 d, whereas a whole directed evolution can be performed in 3-10 rounds, depending on selection efficiency.

Figure 1 |

General CPR concept, (a) Schematic of CPR principle. A gene circuit is generated in which a partner-gene activity allows the expression of a DNA polymerase in bacterial cells. Inactive gene variants lead to no expression of the DNA polymerase. The genetic circuit containing the diversified partner-gene pool is expressed in vivo, allowing DNA polymerase production only in cells with active partner-gene variants. The live cells are subsequently emulsified to produce no more than a single cell per emulsion droplet. The initial boiling step of the ePCR lyses the cells, releasing the produced DNA polymerase protein as well as the partner-gene-encoding plasmid into the aqueous solution of the emulsion droplet. Ensuing thermal cycling amplifies only the active partner-gene variant, which is recovered and used in the next round of CPR selection, (b) Examples of genetic circuits for CPR. Partner-gene function can be linked to expression of a DNA polymerase in a number of ways, as is demonstrated by the example of T7 RNAP, pol, polymerase; tRNA synthetase, and tRNA engineering.

Figure 2 |

Overview and time line of experiments. A CPR procedure encompasses bacterial expression of the DNA polymerase and partner genes for several hours, emulsification of the bacterial cells in water-oil droplets, in vitro amplification of active partner genes via ePCR, breaking of the emulsions, purification of amplified DNA from cell debris and PCR primers, purification of amplified DNA from background plasmid DNA by Dpnl digestion, and final recovery and bulk amplification of the partner gene using an additional recovery PCR. A single round of CPR typically takes 3–5 d, and the procedure can be repeated 3–10 times until no more enrichment is observed.

Figure 3 |

Schematic of the two Recovery strategies, (a) Recovery of biotinylated ePCR product. 5’-biotinylated PCR primers are used for amplification by ePCR. The biotinylated ePCR product is later recovered from contaminating plasmid DNA using streptavidin-coated magnetic beads. The ePCR product is further amplified via nested PCR to obtain enough product to continue with the next round of selection, (b) Recovery of tagged ePCR product. PCR primers with unique 5’-sequence tags are used for amplification by ePCR. The tagged ePCR product is later recovered from contaminating plasmid DNA by amplification using primers that bind to the unique tag. Finally, internal primers can be used in a final nested PCR to obtain enough product to continue with the next round of CPR selection.

Figure 4 |

Anticipated results, (a-f) Appearance of mixtures in the process of emulsification/de-emulsification, (a) A 2 ml-tube containing cells suspended in the ePCR buffer, oil mixture, and the rubber stopper from a 1-ml syringe before agitation on TissueLyser. The mixture appears clear and is separated into two phases, (b) Same mixture as in a after agitation on TissueLyser. The emulsion is viscous, and appears homogeneous and of milky white color, (c) The appearance of the mixture after being aliquoted in 12 × 100-μl PCR samples, subjected to thermal cycling, and pooled together in a 1.5 ml tube, (d) The appearance of the mixture in c after 10-min centrifugation. The mixture separated into two visible layers, with a top cloudy oil phase and a bottom remaining emulsion layer. The top oil phase is to be discarded. The remaining bottom emulsion layer appears as an amorphous white solid, (e) The appearance of a broken emulsion after phenol/chloroform/isoamyl alcohol addition and vortexing. The mixture is still cloudy but exhibits a greatly reduced viscosity. The bottom amorphous solid-like layer is no longer present, (f) The same mixture as in e after 2-min centrifugation. The mixture separated into two clear phases: the top aqueous phase (to be transferred to a new tube) and the bottom organic phase (to be discarded), (g) Example of a gradient of emulsion stability that can be generated under different emulsification conditions. After 30 rounds of PCR thermal cycles, the emulsions were visually analyzed for stability. A gradient of emulsion stabilities is observed, in which unstable emulsions separated into two phases (left), while stable emulsions remained opaque, with minimal phase separation (right). Green squares, intact emulsion; red squares, oil phase separated from disrupted emulsion droplets, (h) Phase-contrast microscopy image of 50x-diluted emulsion. Scale bar, 10 μlη. (i) Superimposed GFP fluorescence/phase-contrast microscopy image of emulsified GFP-expressing DH10B(DE3) E. coli cells under 40× magnification. Scale bar, 10 μlη. (j,k) Example of mock selection data. E. coli expressing either wild-type tyrosil-tRNA synthetase from Methanocatdcococcus jannaschii (MjYRS) or its nonfunctional variant (containing a stop codon and a Notl restriction site) were mixed at the indicated ratios and subjected to a single round of ePCR. (j) After the mock selection samples were amplified by re-amp PCR and equal amounts of DNA were restriction-digested by Notl, the DNA fragments were analyzed by gel electrophoresis to distinguish active (uncut) from inactive (cut) variants of MjYRS. Several thousandfold enrichment of active enzyme variant is observed. Star, active variant fragment size; arrows, inactive variant fragment sizes. Adapted with permission from ref. 29, American Chemical Society, (k) Gel-electrophoresis image of recovery PCR. 1: pure active MjYRS amplicon; 0: pure inactive MjYRS amplicon; 10^_1-10⁻⁴: amplicons of activeiinactive MjYRS dilutions. (1) Monitoring enrichment progress by GFP assay. BL21 E. coli cells carrying pACYC-GFPmut2 plasmid (in which P_T7 drives GFP expression) and plasmid ligations from the initial T7 RNAP selection rounds were assayed in a microplate reader for GFP fluorescence. XX, negative control T7 RNAP with two premature stop codons; WT, parental T7-RSS plasmid reported; R0, naive library; R1–R12, the output for each subsequent round during the selections for use of PT7; CGG-R7–8, a single clone from round 7 (this mutant was subject to error-prone PCR, yielding CGG-R7 epPCR); CGG-R12-KI, a single clone from R12; other CGG-R12 variants are selected combinations of mutations seen in the round 12 population. Data represent averages of three independently grown samples. Error bars represent 1 s.d. Adapted with permission from ref. 28, Nature Publishing Group.