Team:Stanford/QB DE In Vitro

Stanford iGEM Team Wiki

DiCE Cell-Free


Team Pic
Team Pic
DiCE is a promising candidate for accessible and inexpensive self-selecting directed evolution: A) This summer, we demonstrated the DiCE system’s functionality by first showing the individual components required for directed evolution, next showing that DiCE demonstrated integral stepping stones required for successful DE, and finally showing a successful proof-of-concept evolution. B) DiCE offers distinct advantages over other currently available self-selecting DE methods. C) We demonstrated that DiCE has the potential to be an accessible, easy-to-implement and inexpensive kit that could greatly increase accessibility to directed evolution across iGEM and beyond.

Current self-selective systems (SSS) such as CSR​ ​, CPR​​ ​, and PACE​ ​have been invented to evolve new proteins. These methods work by tying the desired function to be evolved to gene fitness through expression of a genetic “amplifier” that leads to proliferation of desired genotypes. Despite success evolving novel Bt toxins​ ​, polymerases with altered substrate specificities​, and antibodies with increased solubility, to name just a few, these SSS have limitations: they are organism-specific, usually require challenging phage-cloning, and, most centrally, require invention of a way to tie the function of the protein to evolve to expression of the amplifier protein. For instance, our team ran into numerous challenges when trying to set up and execute selection schema for PREDCEL+ and DiCE in vivo, including the need to clone multiple constructs before even beginning to undertake the work of implementing an evolution, which itself can take a significant amount of time and troubleshooting. An easy-to-use SSS would enable more groups to tackle the central problem of linking correct protein function to amplifier production.

Cell-free systems consisting of cell lysates that perform transcription and translation in vitro are valued as quick and easy prototyping tools for synthetic biologists, while remaining fairly inexpensive, at pennies per reaction. These benefits, along with the fact that cell extracts support production of products toxic to living cells, makes them an attractive basis for a directed evolution platform.


To design a directed evolution platform in vitro, we started from the general architecture for a SSS described below. There is a fundamental challenge with working with cell-free lysates, however - each protein in a cell-free extract is free to move throughout the solution and interact with any other component in the media, making it hard to selectively amplify “good” mutants. We solved this problem by compartmentalizing our cell-free into water-in oil droplets. Tawfik and Griffiths (1998) showed that by vortexing non-polar and polar solvents, emulsions would form with, on average, 1010 droplets produced per milliliter, and, within these emulsions, evolution can take place8.

Team Pic

Figure 1: General selection schema for in vitro evolution: Cell-free mixture with the gene of interest and amplifier genes on either linear or plasmid DNA are emulsified into water-in-oil droplets. The gene of interest is expressed and, if functional, drives the expression of the amplifier, which replicates the gene of interest. Droplets can be recombined and genetic material extracted for amplification and re-encapsulation for future rounds of evolution before sequencing.

Inspired by Ichihashi et al. (2013) and their ability to use cell-free to evolve Qß-replicase self-encoded by its RNA9, we hoped to harness Qß replicase, an RNA-dependent RNA polymerase, to conduct arbitrary evolutions in vitro. Because of Qß replicase’s simplicity and error rate of 1 mutation for every 700 base pairs, we believe that a Qß replicase can be implemented in a variety of in vitro environments with little dependence on which organism the cell-free lysate came from. We decided to call our envisioned system Directed Chassis-Agnostic Evolution (DiCE) with that goal in mind.

Team Pic
Figure 2: Diagram of a DiCE in vitro selection schema: Left - Cell-free mixtures are encapsulated into millions of small compartments within an oil mixture. Within each mixture, RNA sequences of the desired protein to evolve are flanked by midivariant regions (MDVs), sequences that enhance Qß replicability. These protein mutants are translated by ribosomes in the cell-free extract. A mutant with the desired properties will then upregulate the production of Qß, which is encoded on a linear or circular DNA strand. This Qß will replicate only the MDV-flanked RNAs within its respective droplet. Right – Envisioned workflow for DiCE. Selection and RNA Amplification can be alternated followed by reverse transcription, PCR, and sequencing, all standard protocols requiring minimal equipment and additional cost.

Experiment & Results

Exogenous RNA Amplification

Having conceived DiCE, we first needed to demonstrate that our system was capable of RNA amplification in cell-free systems via Qß replicase expression. We added GFP with MDVs flanking each end of the sequence and either added or withheld RNAs encoding Qß replicase.

Team Pic
Figure 3: Amplification of GFP RNAs by Qß replicase expression: 10 ng of GFP RNA was added to 16 µL of crude cell-free extract, 17.8 µL of energy premix, both prepared in accordance with reference 7, 1.0 µL of murine RNAse inhibitor, and either nuclease-free water, 5000 ng of Qß replicase, or 5000 ng of Qß replicase full fusion protein. Mixtures were split into three 10 µL aliquots and transferred to a well plate reader, covered in 10 µL of mineral oil, and incubated at 30ºC for 12 hours.

As shown in the figure above, we were able to observe significant increases in fluorescence when both Qß replicase RNAs were added compared to just GFP. We hypothesized that the larger Qß replicase fusion protein required more resources to make than the ß-subunit alone, which was free to complex with the existing EF-Tu and EF-Ts proteins in the cell extract, and contributed to its lower expression relative to just the ß-subunit.

Characterization of Midivariant Regions

To determine the replicability boost given by MDVs to the flanked RNA, we created a self-replicating RNA that produces the Qß ß-subunit and Spinach aptamer. We flanked this RNA with or without MDVs to determine the boost in replicability gained.

Team Pic
Figure 4: Self-replicability of RNAs is enhanced by midivariant regions: We made use of a self-replicating RNA with the Qß replicase ß-subunit and Spinach aptamer with or without midivariant (MDV) to test the effect of the MDVs on RNA replicability. 500 ng of each RNA was added to 9 µL of myTXTL Sigma 70 Master Mix, 1.2 µL of Murine RNAse Inhibitor, 1.2 µL of 10x DFHBI, and 0.43 µL of nuclease-free water. 10 µL was transferred to a well plate reader, covered in 10 µL of mineral oil, and incubated at 30ºC for 48 hours. Replicates are shown in fainter colors while solid lines indicate averages.

Our boost in replicability was modest, but observable, motivating the use of MDVs in our evolution schema.

Emulsification of cell-free lysate

As alluded to earlier, compartmentalization in water in oil prevents expressed Qß replicase from non-specifically amplifying mutant RNAs. To ensure that translation is not compromised during the emulsification process, we added GFP RNA to cell-free lysate and encapsulated the mixture within water in oil droplets. After overnight incubation, droplets were imaged under a GFP filter on a microscope.

Team Pic

Figure 5: Expression of GFP RNA within cell-free droplets: 794.3 ng of GFP RNA was added to 4.5 µL of myTXTL Sigma 70 Master Mix and 0.5 µL of Murine RNAse inhibitor. 100 µL of oil mix (Chimerx ePCR kit, prepared according to specifications) was added to the cell-free mixture, vortexed for 5 minutes at 4ºC and incubated overnight. 1 µL of water-in-oil mixture was imaged on a slide using filters for detecting GFP fluorescence.

Now that expression was confirmed within droplets, we could attempt a proof of concept evolution using the system. Because of the relative ease of designing a selection schema and desirable mutations being well-documented, we chose to evolve the T7 RNA polymerase to bind a T3 promoter.

We encapsulated the MDV-flanked T7 RNA into droplets containing our selection cassette, a linear strand of DNA containing the T3 promoter upstream of the Qß ß-subunit. After incubation, the resulting RNA was purified and re-encapsulated 4 more times before our assay for evolution. We then took the evolved RNA and added it to a bulk cell-free mixture with a linear DNA encoding the T3 promoter in front of an mRFP and compared it to an unevolved control.

Team Pic
Figure 6: Evolution of T7 Polymerase RNA using in vitro DiCE: Left – Selection schema for T7 evolution in DiCE. Five rounds of DiCE evolution were performed on T7 RNA transcribed from a T7 HiScribe kit. DNA was then removed using the protocol described in a ThermoFisher DNA-free kit. Typically, 0.8 µL of RNA from the previous round of evolution was added to 4.5 µL of myTXTL Sigma 70 Master Mix, 0.5 µL of 10 ng/µL T3-Qß linear construct DNA, and 0.2 µL of 150 µM GamS protein (to make a final concentration of 5 µM GamS). 300 µL of oil (Chimerx ePCR kit, prepared according to specifications) was added to the cell-free mix and vortexed for 5 minutes at 4ºC. After incubation at 30ºC overnight, RNA was harvested using the protocol in the kit provided by ChimerX for purifying ePCR product. RNA was reverse-transcribed and an amplicon PCR performed at each step to verify that the RNA was not lost at each step. If RNA was lost, as was after the fourth round of evolution, an amplification reaction was performed where 0.5 µL of RNA from the previous round (e.g. round 3) was added to 4.5 µL of myTXTL Sigma 70 Master Mix, 0.5 µL of Murine RNAse inhibitor, and 0.5 µL of transcribed Qß ß-subunit RNA and incubated overnight. The fifth round of evolution was performed after this amplification. Right – Assay to determine evolution of T7 Polymerase. 0.8 µL of purified round 5 RNA was added to 9.0 µL of myTXTL Sigma 70 Master Mix, 1.0 µL of Murine RNAse Inhibitor, 0.2 µL of 150 µM GamS (for a final concentration of 2.5 µM), and 1.0 µL of 7.3 ng/µL T3-mRFP linear construct. 10 µL of this mixture was added to a 384 well plate, covered with 10 µL of mineral oil and incubated at 30ºC for 7 hours.

Additionally, we were able to encapsulate our total extracted RNA with a linear DNAs encoding the T3 promoter in front of the Qß ß-subunit and mRFP. After overnight incubation, droplets were imaged under 40x magnification with bright field and RFP fluorescence filters. While our images could not prove that the mRFP fluorescence observed in our bulk cell-free expression assay was due to a few enriched binders, they suggest it - we observed only a few droplets with high mRFP expression. A characteristic image is shown below.

Droplet Picture
Figure 7: Evolved T7 Polymerase RNA Visualization.

While these initial results looked promising, indicating that a subset of the T7 polymerase mutant population had better binding ability to T3, we performed a Sanger sequencing read of a key region expected to contain mutations, but got the consensus T7 polymerase sequence. We did not have time in the competition to submit a NGS read of this data to observe if the population was enriched for the T3 binder phenotype and hope to do so after the competition in addition to attempting to repeat and refine the DiCE evolutionary method.