Team:KU Leuven/Experiments
BLADEN
Experiments
Our research was divided into three main phases: experiments in E. coli, cell-free, and BY-2 cells. Additionally, we also implemented other specific experiments such as the molecular ruler experiment and riboswitch experiment to improve on an existing part. On this page, we describe in detail our experimental design and procedures.
EvolvR in E.coli
Rationale
To validate that our EvolvR constructs are functional, we decided to test in Escherichia coli before introducing them into Tobacco BY-2 cells. We received two EvolvR plasmids of the original researchers from Addgene that we would use in our experiment: pEvolvR-enCas9-PolI3M-TBD and pEvolvR-enCas9-PolI5M. We wanted to use superfolder green fluorescent protein (sfGFP) as our reporter by introducing a nonsense mutation in its sequence to make it no longer fluorescent. If the EvolvR construct is functional, it should be able to revert the nonsense mutation back to wild type if given a guide RNA (gRNA) that targets the sfGFP region of interest, serving as an easy readout.
Preparation Workflow
Construction of pEvolvR plasmids with gRNA
The EvolvR complex consists of an enhanced Cas9 nickase fused to an error-prone DNA polymerase I (PolI). In the original paper by Halperin et al. (2018), the two most efficient variants of PolI DNA polymerase were found to be PolI3M-TBD with three mutations providing lower fidelity and a thioredoxin binding domain (TBD), and PolI5M with two additional mutations conferring increased targeted mutation rate. We received these variants from Addgene, but, additionally, we decided to also take out the "TBD" domain from the PolI3M-TBD variant to test it as a control against the more efficient variants.
These original pEvolvR plasmids already contained an enCas9 gRNA scaffold with an sfGFP as a placeholder for a 20-nucleotide protospacer sequence, as shown on Figure 1. We designed the protospacer targeting ΔsfGFP(Y93X) according to the design in the original paper by choosing a protospacer adjacent motif (PAM) site within a few nucleotides from the target region. Additionally, we designed a protospacer that does not bind the target plasmid or E. coli genome (“unspecific” protospacer) to use as a negative control. To introduce the gRNA protospacers into pEvolvR plasmids, we PCR-amplified the plasmids around the protospacer sfGFP placeholder. This amplification was performed in two separate PCRs as the pEvolvR plasmids are large (> 10 kb). We took advantage of this by using these PCRs to also take out the TBD domain from pEvolvR-enCas9-PolI3MTBD to get pEvolvR-enCas9-PolI3M. We then assembled the plasmids back together using Gibson Assembly. All assembled plasmids were confirmed by Sanger sequencing.
Generation of a mutant sfGFP
We chose to use the sfGFP protein as a reporter to easily monitor targeted mutations introduced by EvolvR via green fluorescence. We introduced a nonsense mutation into the target that EvolvR would then restore back to functional form so that green fluorescence would be a readout for successfully targeted mutation. The nonsense mutation was introduced at position 93 of sfGFP to replace the TAT codon coding for tyrosine with the stop codon TAG via site-directed mutagenesis. We relied on E. coli recombination machinery to ligate the plasmid back together after transformation. The non-fluorescent sfGFP variant with a nonsense mutation is noted as ΔsfGFP(Y93X) and was sequence-confirmed via Sanger sequencing.
Plasmid Compatibility Issue
The pEvolvR plasmid contains a pBR322 origin of replication (ori), and the pET29 vector containing our target ΔsfGFP(Y93X) has an F1 ori. Unfortunately, those origins are not compatible and the plasmids cannot be maintained together in the same bacterial cell. We would therefore not be able to use them together in our experiment. To solve this issue, we decided to clone previously made ΔsfGFP into a different plasmid. For that we used a pBAD plasmid carrying an ori that is compatible with pBR322. This new plasmid already contains an arabinose-regulated araBAD promoter and a rrnB terminator, so we amplified only the coding sequence of ΔsfGFP and inserted it into the pBAD plasmid via Gibson Assembly.
E.coli T7 Express strain
To perform targeted mutagenesis experiments with EvolvR, we chose the T7 express lysY/Iq strain of E. coli (T7) from NEB[1] . The T7 express strain comes from BL21 origin, which allows high-level protein expression, however, certain modifications in relation to original BL21 strain also allow efficient DNA recovery for further sequencing of targets after the directed mutagenesis. However, our chosen T7 express stain carries chloramphenicol resistance gene, which was used as selection marker for our target pBAD-ΔsfGFP(Y93X) plasmid, so we needed to switch the antibiotic resistance gene in that plasmid before proceeding with the continuous directed evolution experiment.
Antibiotic Resistance Casette Switch
To change the selection cassette in our plasmid we needed to change the gene providing antibiotic resistance. To do that, we amplified the pBAD-ΔsfGFP(Y93X) plasmid around the chloramphenicol resistance gene as well as an ampicillin resistance gene from another plasmid we had, then inserted the ampicillin resistance gene into the target plasmid by Gibson Assembly.
Directed mutagenesis experiments in E. coli
Transformation of three EvolvR variants
We sequentially transformed T7 express cells with our target pBAD-ΔsfGFP(Y93X) and pEvolvR plasmids, shown in Figure 2.
The different conditions in our experiment can be found in the table below. Overall, we have three variants of pEvolvR , and for each variant we have a sample with gRNA targeting the ΔsfGFP mutation site, a sample with gRNA that is not targeting anything in the system with a sfGFP instead of a gRNA protospacer (unfunctional gRNA). All those samples contained target palsmid as well, and we also had samples containing only EvolvR plasmids with taget gRNA, but no target plasmid, samples containing only the target plasmid but no EvolvR plasmid as negative controls.
| Number of plasmids | Name |
|---|---|
| 2 | pEvolvR-enCas9-PolI5M -targeted gRNA + pBAD-ΔsfGFP(Y93X) (A1) |
| 2 | pEvolvR-enCas9-PolI3M-TBD -targeted gRNA+ pBAD-ΔsfGFP(Y93X) (A2) |
| 2 | pEvolvR-enCas9-PolI3M -targeted gRNA+ pBAD-ΔsfGFP(Y93X) (A3) |
| 2 | pEvolvR-enCas9-PolI5M -unspecific gRNA+ pBAD-ΔsfGFP(Y93X) (B1) |
| 2 | pEvolvR-enCas9-PolI3M-TBD -unspecific gRNA+ pBAD-ΔsfGFP(Y93X) (B2) |
| 2 | pEvolvR-enCas9-PolI3M -unspecific gRNA+ pBAD-ΔsfGFP(Y93X) (B3) |
| 2 | pEvolvR-enCas9-PolI5M -sfGFP+ pBAD-ΔsfGFP(Y93X) (C1) |
| 2 | pEvolvR-enCas9-PolI3M-TBD -sfGFP+ pBAD-ΔsfGFP(Y93X) (C2) |
| 1 | pBAD-ΔsfGFP(Y93X) (D1) |
| 1 | pEvolvR-enCas9-PolI5M -targeted gRNA (E1) |
| 1 | pEvolvR-enCas9-PolI3M-TBD -targeted gRNA (E2) |
| 1 | pEvolvR-enCas9-PolI3M -targeted gRNA (E3) |
| 0 | No plasmid - WT T7 express (F1) |
All samples were grown in liquid media and expression of EvolvR was induced by adding anhydrotetracycline (ATC) because EvolvR is under the regulation of the TetA/TetR system. Cells were first grown for 4 hours, then a part of the liquid culture was plated on antibiotic plates to screen for green colonies, which would indicate that a mutation reversing ΔsfGFP back to its fluorescent form was made by EvolvR. The remaining culture was left to grow overnight to allow EvolvR variants more time to function and introduce mutations.
First results
After growing the cultures for 4 hours and subsequent induction of EvolvR gene expression, we observed a drastic drop in optical density in two samples carrying the target plasmid and enCas9-PolI5M and enCas9-PolI3M-TBD with targeted guide RNA (A1 and A2). After leaving these cultures in the incubator overnight, we discovered that all the bacterial cells in those samples have died. However, all the other samples with pEvolvR carrying unspecific gRNA or no gRNA at all, or targeted gRNA but no target plasmid survived. These results suggest that EvolvR may be interfering with the maintenance of the target plasmid and either destroying it or drastically reducing its copy number in the cell, resulting in the ampicillin resistance gene not providing the necessary level of resistance for survival and bacteria were left to die.
We also plated samples containing the original pEvolvR plasmids with sfGFP in place of the protospacer (C1 and C2) on selective and non-selective plates. We discovered that on non-selective plates the fraction of green colonies is approximately 10 times lower, leading us to believe that EvolvR by itself is relatively toxic to the bacteria and they may be trying to get rid of the plasmid as soon as they are released from selection pressure.
Adjustments of experimental conditions
To test whether lethality caused by EvolvR is affecting the maintenance of ampicillin resistance in the target plasmid, we decided to repeat the experiment with three other protospacers at varying distances from the sfGFP nonsense mutation in the pBAD-ΔsfGFP(Y93X) plasmid that were initially designed for the "molecular ruler" experiment (see below). To reduce the potential toxicity of EvolvR, we also lowered the incubation temperature from 37°C to 25°C after induction of EvolvR expression, slowing down protein expression and reducing the burden that EvolvR expression imposes on the bacterial cell.
New results
We managed to repeat the experiment with the old protospacer under new temperature conditions. We observed that samples carrying the target plasmid and variants of EvolvR that were deadly at 37°C have survived in lower temperature, suggesting that slowing down protein expression can reduce toxicity of EvolvR. We did not observe any green or partially green colonies on plates, and no GFP signal was detected by flow cytometry. These results can suggest that the protospacer we chose is not optimal for our case or that EvolvR efficiency is not high enough to be observed without additional selective pressure. More details can be found on the Results page.
Cell Free experiments
Rationale
In addition to testing our EvolvR constructs in a bacterial system, we wanted to do so in a system that more closely resembles plants at the cellular level. We, therefore, opted to test our EvolvR constructs in a cell-free system. Using such a system will allow us to rapidly identify whether EvolvR is functional in a eukaryotic plant-derived environment. Many cell-free systems have been developed over the years ranging from E. coli extracts to wheat germ extracts [3]. Because our plant model is a Tobacco Bright Yellow 2 (BY-2) liquid culture, we chose Leniobio’s ALiCE cell-free protein expression kit which is based on a BY-2 cell lysate [4]. This kit comes with reaction vials that contain all the transcriptional and translational machinery required for protein expression without problems one would encounter at the whole-cell level such as low transformation efficiency, toxicity or interactions with the other proteins in the cell. Since we are interested in seeing whether EvolvR can mutate a user-defined target region rather than alter protein expression or yield, we adjusted the ALiCE® protocol for our purposes by adding multiple plasmids to our reactions (Figure 3).
Experimental Procedure
1. We constructed pALiCE01-ΔsfGFP(Y93X)-gRNA by PCR amplifying both the pALiCE01 vector and the gRNA scaffold with BsaI overhangs for Golden Gate Assembly. The gRNA scaffold comes from the original EvolvR plasmids [5]. The scaffold was modified by introducing a 20-nucleotide protospacer that targets the nonsense mutation that was previously introduced in sfGFP at position 93. Both vector and insert were digested with BsaI and ligated following our Golden Gate Assembly protocol. This assembled plasmid would then be linearized using a single restriction cutter (EcoRI) downstream of the gRNA scaffold so as to ensure the gRNA is transcribed appropriately. We do not anticipate running into stability issues since the gRNA is transcribed under the T7 promoter.
2. Construct pALiCE01-enCas9-PolI5M plasmid following the kit’s instructions. The kit’s protocol instructs us to do so via restriction-ligation using NcoI and KpnI to replace eYFP with enCas9-PolI5M while preserving the 5’ and 3’ untranslated regions of the vector. Because enCas9-PolI5M contains an NcoI restriction site, we opted to design a BsaI restriction site that would leave a NcoI-like sticky end after digestion. Our plan was to PCR amplify enCas9-PolI5M from the original pEvolvR-enCas9-PolI5M plasmid with primers creating BsaI and KpnI restriction sites in the overhangs. Unfortunately, at this stage, we faced many issues with our PCRs. Originally, we noticed that the amplified fragments were too large (around 10kb instead of 7kb) despite our bands being strong and sharp on the agarose gel. Despite many troubleshooting attempts, we were unsuccessful in cloning this plasmid.
3. Purify all the above plasmids and make two mixtures of them following the table below. We opted to have each component of interest expressed individually in parallel. We wanted to use separate plasmids instead of inserting all the components on one plasmid to reduce complexity in the system. We were also concerned that having the enCas9 protein expressed from the same plasmid as its target could potentially hinder its own expression. pALiCE01 is used as a positive control in each mixture to confirm that the reaction worked. One of the mixtures serves as a negative control with no added gRNA in order to control for potential off-target effects of EvolvR. The target plasmid (pBAD-ΔsfGFP(Y93X)) is added in excess because it is not expressed in the cell-free system, lacking a T7 promoter that is needed for expression. Therefore, the addition of this plasmid is not a burden on the system and we would just sequence it after running the reaction to confirm that EvolvR is functional.
| pALiCE01 | pALiCE01-ΔsfGFP(Y93X)-gRNA | pALiCE01-enCas9-PolI5M | pBAD-ΔsfGFP(Y93X) | |
| Ratios | 1 | 9 | 10 | 50 |
| Ratios for gRNA(-) control | 1 | 0 | 10 | 50 |
4. The plasmid DNA mixture would then be concentrated via lyophilization and resuspended to the appropriate volume. Additional deoxynucleotides (dNTPs) and magnesium sulfate (MgSO4) are added to the resuspension to allow for the DNA polymerase in EvolvR to induce mutations in the ΔsfGFP(Y93X) target region.
5. Each mixture would then be added to the cell-free reaction mix and we would let them run for 48h following the ALiCE® protocol.
6. We would extract the DNA using phenol-chloroform extraction followed by isopropanol precipitation.
7. Finally, we would send our sample for Next-Generation Sequencing to confirm that EvolvR has induced mutations in the sfGFP gene of the target plasmid.
Construction of Plant Plasmids
Rationale
The ultimate goal of our project was to develop a toolkit to perform continuous directed evolution (CDE) in plant cell culture using the EvolvR complex. To introduce EvolvR into plant cells, we opted for Agrobacterium-mediated transformation. This method is based on the ability of the plant pathogen, Agrobacterium tumefaciens, to insert a piece of DNA called T-DNA from the Ti plasmid into the plant genome during infection. This T-DNA is flanked by recognition sequences called “borders”. Since this discovery by Zaenen and colleagues (1974) [6], plasmids have been designed that contain these borders along with multiple selection markers. When using such plasmids, any DNA that is cloned between the left and right borders will be inserted into the plant genome in a successful transformation. For our purposes, we received the plasmids we needed to build this toolkit from the VIB gateway vectors collection.
Plasmid Design
To make a vector that will carry the EvolvR gene into plant cells, we used a Golden Gate Cloning method adapted by the VIB-UGent Center for Plant Systems Biology. This method is modular in that it uses standard genetic building blocks called entry vectors to build a destination vector via Golden Gate Assembly. There are 6 standard entry vectors shown in Figure 4 that are available with different overhangs that allow for all cassettes from the entry vectors to be combined in a specific order into a destination vector.
Our entry vectors:
(1) pGG-A-35SP-B – entry vector with A-B overhangs: We chose a standard constitutive CamV35S promoter.
(2) pGG-B-NLS_N7-C – entry vector with B-C overhangs: A N7 nuclear localization signal (NLS) was placed at the N-terminal of EvolvR to ensure that it is transported into the cell nucleus after translation in order to have access to the genomic DNA.
(3) pGG-C-linker-D – entry vector with C-D overhangs: This empty entry vector was cloned with three different variants of EvolvR: enCas9-PolI3M, enCas9-PolI3M-TBD, and enCas9-PolI5M. We amplified the EvolvR coding region from the pEvolvR plasmids of Halperin and colleagues (2018) [5] using PCR primers containing BsaI cut sites and specific C-D overhangs. The EvolvR cassettes were then individually cloned into the empty entry vector for further Golden Gate assembly.
(4) pGG-D-P2A-mCherry-NLS-E – entry vector with D-E overhangs: We placed a P2A-mCherry cassette as a C-terminal tag. This construct consists of a self-cleaving P2A ribosome skipping peptide that allows for the separation of the peptides at the N and C-terminals of P2A during translation. As a result, the mCherry will not be fused to EvolvR, but mCherry (a red fluorescent protein) has an NLS allowing for the localization of red fluorescence in the nucleus, serving as a marker for successful transformation into the plant cell.
(5) pGG-E-35ST-F – entry vector with E-F overhangs: This entry vector contains a 35ST terminator.
(6) Entry vector with F-G overhangs: This is a variable block in the Gateway destination vector that generally serves as a placeholder for the user to insert any cassette of interest. We have used and designed three versions of such an entry vector. One version contains a short linker containing two AarI restriction sites. Such a linker adds no additional metabolic burden on the cell and can thus be used as a negative control for our CDE experiments. A second F-G entry vector contains a SacB gene flanked by AarI and BsaI sites allowing us to use negative selection with sucrose against a self-ligated plasmid. Self-ligation is a problem we would otherwise face with the F-G entry vector without SacB as we would have no way to select for the plasmid containing an inserted cassette. Therefore, the third version contains the gRNA scaffold corresponding to enCas9, originally from Halperin and colleagues (2018) [3] and a protospacer that targets the BY-2 genomic region of our interest. We PCR amplified this gRNA scaffold with BsaI overhangs and constructed the third F-G entry vector via Golden Gate assembly.
The protospacer that was selected targets the glutamine synthetase gene of BY-2 that is responsible for the susceptibility to BASTA herbicide. We examined the structure of the enzyme, identified ligand binding sites, and designed two protospacers to be a few nucleotides downstream from the targeted regions that we believed to be the most likely to confer resistance to BASTA.
The final destination vector containing the building blocks from each of the entry vectors is shown in Figure 5A. In total, 15 destination-vectors were constructed as shown in figure 4: 3 variants of EvolvR and 4 different final blocks – short linker as negative control, SacB linker to be able to easily clone another gRNA cassette, 2 gRNA scaffolds with protospacers specific for glutamine synthetase and 1 with a non-specific protospacer as another negative control (Figure 5B).
One of these plant destination-vectors (pGGK_35SP_NLS-N7_enCas9-PolI5M_P2A-mCherry-NLS_35ST) was successfully transfected into BY-2 cell culture, confirming that our method to deliver EvolvR into plant cells works.
BY-2 experiments
Rationale
Ultimately, we strived to test all our plant EvolvR constructs in Tobacco BY-2 cells. Firstly, we wanted to confirm that we can successfully transform BY-2 cells with a plant destination vector carrying EvolvR by detecting the fluorescence of the mCherry that should be localized in the nucleus as it is associated with an NLS. Secondly, we want to validate whether EvolvR is mutating the user-defined target region of the BY-2 genome, which we chose to be the glutamine synthetase gene encoding for the target of BASTA herbicide. The activity of EvolvR and the targeted mutation rate can be measured by sequencing this target gene. Finally, by applying selection pressure with a sublethal concentration of BASTA in a CDE experiment, we can evolve BY-2 cells to become resistant to BASTA, proving that EvolvR is indeed functional in plant cells.
The road to functional EvolvR in plant cells
BY-2 Cell culture maintenance
We started our experiments from a BY-2 liquid cell suspension culture. Tobacco BY-2 cells are a plant cell line established from Nicotiana tabacum cultivar Bright Yellow-2 [6]. This cell line exhibits homogeneity and rapid growth rates and is therefore often used as a model organism for plant research [7]. The cells form small clusters in the liquid culture, but each cell is still exposed to changes in the culture medium. We also confirmed this in our own culture under a microscope. Since it has even been referred to as the “HeLa” system of higher plants, we chose this plant cell line to demonstrate the activity of EvolvR in plant cells [7].
The culture was maintained in its suspension form throughout the project. To ensure reproducibility, we maintained a constant growth environment and sterile conditions. Due to a limited amount of incubator space, we also attempted to optimize growth conditions by using different suspension culture containers.
First, we grew all the cells in the same incubator which was isolated in the lab as to avoid possible contamination from experiments happening in the environment. The erlenmeyers were taped with surgical tape to allow for oxygenation while preventing contamination. We had multiple isolated cell lines running in parallel which got diluted on separate days. Since the incubator had limited space, we tried different adjustments to the suspension culture containers: with aluminum foil instead of metal cap [6], growing in 50 mL falcon tubes and in deep tissue culture petri dishes.
BY-2 Genomic work
The BY-2 genome is not fully characterized. Therefore, in order for us to be able to target the BY-2 glutamine synthetase gene and design the correct gRNA protospacers for EvolvR, we needed to sequence this region of the BY-2 genome. From our own BY-2 liquid cell suspension culture we isolated the genomic DNA by combining the Edwards genomic DNA extraction method with a TissueLyser (Qiagen). Next, we performed multiple PCR reactions with different primer pairs to maximize our chances of amplifying the BASTA target gene in the incompletely characterized BY-2 genome. BASTA is a herbicide that inhibits glutamine synthetase, an essential plant gene involved in the production of glutamine and vital for ammonia detoxification. We wanted to target this gene in our EvolvR CDE experiments. The sequencing results helped us choose protospacers for the EvolvR experiments in BY-2.
Agrobacterium work
Agrobacterium-mediated transformation (AMT) is a common plant transformation method where foreign genes are randomly inserted into the plant genome. We first made our strain of Agrobacterium tumefaciens (C58C1 Rif (pMP90)) electrocompetent so that they can effectively take up our construct. We then transformed the Golden Gate-assembled EvolvR plant destination vectors into Agrobacterium by electroporation. For each electrocompetent Agrobacterium batch, we tested the transformation efficiency and made sure no contamination was present.
BY-2 Transformation work
We transformed BY-2 cell suspension cultures using ATM. The optimal transformation conditions were selected by experimenting with different antibiotics and herbicide concentrations on wildtype (WT) cells. We tested the effects of different antibiotics (Timentin, Carbenicillin, Vancomycin, Kanamycin and Hygromycin) on WT BY-2 cells and Agrobacterium carrying our EvolvR constructs to determine what combinations and concentrations best killed the Agrobacteria, but not the BY-2 cells. Next, for our proof-of-concept experiment we guided EvolvR to mutate the BASTA target gene in the BY-2 genome. For this the best sublethal (and lethal) BASTA concentrations were determined both on plates and in liquid to optimize the selection pressure applied during CDE experiments with BY-2 cells.
For transformation, we first performed two trial transformations, one with a GFP-hygromycin construct and one with the pGGK_A-G destination vector. During these experiments we tested different transformation conditions involving BY-2 and Agrobacteria concentrations and volumes to determine the best transformation conditions specific to our project. The GFP-hygromycin BY-2 cells were imaged via fluorescence confocal microscopy. Next, we tested the transformation protocol both on plates and in liquid. The common procedure during transformation of a BY-2 suspension cell culture is to incubate BY-2 cells with Agrobacterium for up to two days and then plate them on solid medium to let them grow into calli, which can take up to three weeks to become visible. This method is typically used as it allows one to isolate a single mutant and start a new transgenic culture.
However, waiting for sufficient callus growth is time-consuming and negates the advantage of speed in a continuous directed evolution experiment with EvolvR. Therefore, we also tested BY-2 transformed cell growth in liquid. Finally, we transformed BY-2 cells with an enCas9-PolI5M-unspecific construct (pGGK_35SP_NLS-N7_enCas9-PolI5M_P2A-mCherry-NLS_35ST_unspecific-gRNA) which would be used as a control in CDE experiments and does not target the BASTA target gene. This construct also includes mCherry, which can be visualized using a fluorescent confocal microscope to determine that these transformed BY-2 cells contain mCherry, and therefore EvolvR.
Optical density measurements of BY-2 cell suspension cultures
Rationale
The plant cells used for the experiments in our project are Tobacco BY-2 cell culture, sourced from VIB-Ghent Centre for Plant Systems Biology. These cultures were diluted at a ratio of 1/40 (i.e., 1 mL of cells for 39 mL of Murashige and Skoog media) once every seven days. Growth rate was an interesting variable for the model and we wanted to confirm that the BY-2 cell suspension culture is healthy and representative of documented BY-2 cultures. Thus, it was important to understand the rate of reproduction of BY-2 cells by making growth rate measurements. While measuring the growth rate of cell culture is quantitatively feasible, it may not be fully reliable. Some associate growth rate with an increase in the number of cells, while others associate it with either an increase in mass or volume. In our studies, we have assumed that an increase in the number of cells or cell population constitutes the growth rate of cells.
Methods to measure Growth Rate
One method to measure growth rate is based on mass-based measurements, including fresh weight (FW) and dry weight (DW). Cells sampled from suspension culture are vacuum filtered, followed by FW measurement. The same cells are then dried in an oven overnight, and the DW is measured. Volume-based measurements like Packed Cell Volume (PCV) are obtained by centrifuging a sample of cells from suspension culture. This results in a pellet whose volume is measured. Alternatively, cells are also counted using a hemocytometer [9].
Optical Density (OD) measurements record the turbidity of a sample of cells in suspension culture. The turbidity of the sample increases as the cell population increases. For our experiments, we chose to perform OD measurements using a spectrophotometer. Compared to mass or volume change, we chose to quantify the growth rate of cells as an increase in the population of cells in culture. The advantage of using OD measurements is that the process is simple and fast compared to the methods listed above [10].
Relating OD to growth rates
The time in which the population of cells in a culture double is called the doubling time td. Doubling time is related to growth factor µ (min-1, hr-1, day-1) as is described in [12]:
$$\frac{ln(2)}{t_{d}}=μ $$
In the exponential phase of the OD curve, a difference of two points is taken, yielding equations that relate OD, time and growth rate as:
$$ ln(OD_{2}) - ln(OD_{0}) = μ t_{2} $$
with the logarithmic growth rate:
$$ μ=\frac{2.303(ln(OD_{2})-ln(OD_{1}))}{t_{2}-t_{1}} $$Our Experiments
A singular line of five-day-old BY-2 cells (named A1 for convenience), was used for the experiments. Three cultures were diluted from the BY-2 cell line in Erlenmeyers with a dilution ratio of 1/40 (i.e., 1 mL cells to 39 mL MS media). An Eppendorf BioPhotometer® was used to obtain OD600 measurements. The cell cultures were placed in a shaking incubator at 28°C and measurements were taken every 12 hours over 7 days with different dilutions to ensure OD600 values remained below 1 to maintain accuracy.
Molecular Ruler
Rationale
As mentioned above in the E. coli experiments, we tested three EvolvR variants (enCas9-PolI3M, enCas9-PolI3M-TBD, and enCas9-PolI5M). However, by doing so we realized that it would also be useful for us to measure the optimal editing window of our EvolvR variants. Therefore, we have chosen 10 evenly spaced gRNA protospacers ranging from 30 to 500 bp from the mutation site in the sfGFP gene present in the target plasmid (pBAD-ΔsfGFP(Y93X)). If EvolvR guided by a specific protospacer is able to mutate the target region, we would expect to see a reversion of the nonsense mutation in ΔsfGFP(Y93X) and (partially) fluorescent colonies. Based on what protospacers show the highest fraction of green colonies we can rate their efficiency and therefore measure the optimal editing window of that EvolvR variant. A basic scheme of its workings is shown in Figure 6.
Experimental Procedure
1. Construct 10 individual plasmids following the same PCR and Gibson Assembly procedure as described above under "E. coli proof of concept" experiments.
2. Use competent T7 express cells previously transformed with pBAD-ΔsfGFP(Y93X) to transform all 10 new pEvolvR plasmids containing one of the molecular ruler protospacers.
3. Follow the E. coli experiment workflow with all molecular ruler pEvolvR plasmids in parallel.
4. Check for fluorescence on plates to see a reversion of the nonsense mutation and quantify the reversion rate for each protospacer.
5. Measure the optimal editing window based on our results.
Improving on existing parts
Rationale
Riboswitches are cis-acting regulatory mRNA elements, comprised of an aptamer domain, most often found in the 5’ untranslated region of the RNA. Like aptamers, the riboswitch folds very specifically according to its sequence [11]. Such specific folding creates a binding pocket that can discriminate very precisely between different ligands. Upon ligand binding, the riboswitch changes conformation, modulating protein expression. Riboswitches can either be ON-switches, i.e. binding of the ligand leads to increased expression, or OFF-switches where binding leads to decreased expression. Different mechanisms of expression control by riboswitches are knows, but a common example of a control mechanism is as follows: the conformation of a riboswitch in the absence of its ligand shields the ribosome binding site (RBS) but as soon as the ligand binds, the conformational change opens up the RBS, activating translation. The high specificity and modular nature of riboswitches have increased interest in the field of synthetic biology for new and better routes of expression-regulation. They are also being explored as a possible new route for drug development, (such as novel antibiotics) since they have not yet been identified in mammals and seem to bind differently to the same ligand in comparison with mammalian proteins, which is expected to give fewer side effects [11] [14].
Since the function of the riboswitch is very much sequence-dependent, changing the sequence through the introduction of mutations, for example, can also greatly change its properties. Our experiment aims to take a riboswitch previously described in the iGEM registry and improve its characteristics such as leakage and dynamic range by creating a library using EvolvR and selecting the variants with the desired characteristics.
Parts
For this experiment, we chose a theophylline-responsive riboswitch, with the goal of having it be responsive to caffeine in the presence of the CYP1A2 enzyme. CYP1A2 is able to metabolize caffeine into theophylline which would allow a theophylline-responsive riboswitch to be indirectly activated by caffeine as well. By using what normally would be treated as waste, caffeine could be purified from this waste and repurposed. The specific part that was used in this experiment was part BBa_K598005. This riboswitch is an ON-switch, exposing the RBS when binding to theophylline. Since theophylline has a rather low toxicity, a constant working concentration of 1mM theophylline would be used in all the conditions that required the addition of a ligand.
Theophylline response will be linked to cell survival through the expression of a TetA-sfGFP fusion. TetA provides the cells with resistance against tetracycline and GFP gives us a fluorescent reporter that can be detected in both spectrophotometer and flow cytometry platforms. TetA and GFP are linked by a short flexible linker (encoded by GGATCCGCTGGCTCCGCTGCTGGTTCTGGCGAATTC) to minimize interaction between the two proteins. Positive selection can be applied by increasing tetracycline concentration in the presence of theophylline so that higher TetA expression (and therefore higher GFP signals) can give the cell a survival advantage in the experimental conditions in the presence of EvolvR. For a more detailed description see “selection details” below.
We selected pBAD/AraC (part BBa_I0500) as the promoter for the riboswitch/reporter construct. This inducible promoter can be tightly regulated and it works especially well for low-copy number plasmids (cfr. next paragraph). The optimal concentration of (+)-arabinose would be determined by testing different concentrations while measuring the response in terms of fluorescence.
Finally, the experiment also makes use of a second plasmid containing the EvolvR itself (enCas9-PolI5M) with a gRNA scaffold containing a protospacer that targets the 3’ end of the riboswitch with the EvolvR mutating upstream of the protospacer continuously create new variants. This plasmid contains a KanR cassette and EvolvR is under the regulation of a TetA promoter and is induced by tetracycline or ATC (anhydrotetracycline). The gRNA is under a constitutive promoter.
Both plasmids were cloned into T7 Express lysY/LacIq Competent E. coli (High Efficiency) from NEB. This strain is well suited for our experiments since they have a high transformation efficiency and are optimized for protein expression. The overall scheme of the plasmid is shown in figure 7.
All parts, except for the backbone, were ordered as gene fragments from IDT. Due to technical constraints on the length of synthesized DNA, this was split into two different fragments: one containing the promotor and riboswitch (with the BioBrick prefix at the 5’ end, and overlap with TetA at the 3’ end) and one containing the TetA-sfGFP (with a 5’ end overlap with the riboswitch and a 3’ end containing the BioBrick suffix). The overhangs allowed for easy Gibson assembly with the pBAD30 backbone that had the BioBrick parts introduced through PCR.
Selection Process Details
As described above, this experiment used a TetA gene fused by a flexible linker to sfGFP for the selection process. The strength of this mechanism is that it contains a dual-selection function, taking away the need for plasmid isolation steps and decreasing the chance of getting false positives [12]. In this case – with an ON-switch – the presence of tetracycline (+theophylline) selects for the cells that express the resistance gene, and thus have an activating riboswitch. Conversely, the addition of NiCl2 (in absence of theophylline) will eliminate false positives as TetA increases the uptake op Ni2+, which is toxic to the cells. A schematic of this is shown in Figure 8.
If one were to work with an OFF-riboswitch, one could use this mechanism as well by using Ni2+ with the ligand for positive selection and tetracycline without ligand for negative selection instead.
Selection conditions should be first optimized, since the appropriate concentrations are dependent on the experimental conditions. The best concentrations for either selection compound (theophylline and NiCl2) can be determined by maximizing the difference in survival between positive and negative controls [13]. Due to time constraints, however, an initial working concentration of 30 µg/mL theophylline and 0.3 mM NiCl2 were chosen – based on frequently reported concentrations in similar experiments [12] [13] [14] [18].
This process also allows for working in liquid culture with high-density genetic selection. However, time must be allowed between positive and negative selection cycles to allow selection reporters to be expressed (in positive selection cycles) or degraded (in negative selection cycles).
Ideally, one would go through multiple consecutive rounds of positive and negative selection. One could then also increase the selection pressure each round by increasing the concentration of the selection agents, allowing the most resistance variants to continue growing leading to their enrichment. To improve upon this process in liquid culture, one could also directly use FACS after each round to select for the cells that give the best signal (i.e. the highest signal for positive selection) [12] [14].
However, to make optimal use of the continuous directed evolution (CDE) aspect of EvolvR, we propose an alternative selection regime. It would start with an initial negative selection, without using EvolvR. Secondly, a continuous positive selection would take place with continuous expression of EvolvR where one would start with an initial concentration of tetracycline, measure the OD600 every hour (repeating throughout one day) and adjust the tetracycline concentrations to the relative growth. More information on the predicted optimal selection conditions can be found on our modelling page. Finally, one last negative selection round would take place to remove EvolvR-derived variants that disrupt the riboswitch regulation by theophylline.
Measurements
The aim of this experiment was to improve the riboswitch by reducing its background activity and increasing its dynamic range, compared to the original sequence. A narrow dynamic range is the most common problem with synthetic riboswitches [15]. Additionally, the proposed measurement set-up permits easy measurement of the riboswitch background activity with the same data.
Dynamic range (η) is defined as the difference between, or ratio of, high and low expressed protein levels, whereas the background or leakage is the basal protein level when no ligand is added [16]. The level of expression would in this case be the level of green fluorescent signal after induction and binding of the ligand to the riboswitch. To test (the improvement) of the dynamic range and leakage, the cells (ones containing the original riboswitch and versions that have gone through selection) will be induced with different concentrations of arabinose while adding increasing amounts of theophylline to create a full 2D matrix of the parameter space. Separate variants will be selected for further characterization. The responses of the separate variants after about two hours are measured in a 96-well plate using a spectrophotometer and compared between to the original riboswitch.
An important aspect to keep in mind is the possible toxicity of the TetA(C) gene when overexpressed. Quantitative responses of the fused gene can then vary in function of the amount of riboswitch ligand added. Optionally, if necessary, one could use PCR mutagenesis to remove most of the TetA coding sequence to remove the TetA gene before starting the measurement procedure, leaving only the sfGFP as a functional reporter.
It could happen that certain colonies when characterizing do not show a significant increase in dynamic range or decrease in leakage, despite having gone through the selection process. One possible explanation for scenario, is that by putting the cells through antibiotic stress the system not only selects for higher TetA expression, but also for better tetracycline-destructing enzymes. It is always advisable to pick multiple colonies for characterization to maximize the likelihood of finding a variant with the improved desired gene.
Summary of experimental Procedures
1. Assembly of the riboswitch device. First introduction of the BioBrick prefix and suffix into a pBAD30 plasmid through PCR. Followed by Gibson Assembly with the two IDT inserts.
2. Add the riboswitch-specific protospacer to the gRNA scaffold present in pEvolvR-enCas9-PolI5M following the PCR and Gibson Assembly protocols as described under “EvolvR in E. Coli”
3. Transformation of the riboswitch device into competent T7 Express (NEB) cells.
4. Inoculate liquid medium containing appropriate antibiotics with a single isolated colony. Perform initial testing of the construct, using different concentrations of arabinose to determine the best concentration for GFP induction. Simultaneously, test the construct under different conditions for correct functioning: (a) induction + ligand + tetracycline, (b) induction + tetracycline, (c) ligand + tetracycline and (d) only tetracycline.
5. Make the T7 Express cells chemically competent again and transform with the pEvolvR-enCas9-PolI5M-riboswitch spacer device.
6. Inoculate liquid medium containing the appropriate antibiotics with a single isolated colony. Perform negative selection by adding NiCl2 in the growth medium.
7. Spin down the cells and resuspend in fresh medium. Induce, then perform positive selection by adding tetracycline to the growth medium, as well as theophylline. Measure OD600 hourly and adjust tetracycline concentrations according to the relative growth of the E. coli.
8. After allowing time for the degradation of TetA in fresh medium, perform one last negative selection by adding NiCl2 in the growth medium.
9. Plate the cells on plates with the appropriate antibiotics. Select a couple of colonies for detailed characterization.
10. Measure the green fluorescence in a 96-well plate with a spectrophotometer, using increasing concentrations of arabinose and theophylline. Compare these measurements to the fluorescent signal under the same conditions by the original riboswitch.
References
[1] Guide to differences between ER2566 and BL21(DE3). Brian Anton, William Jack, James Samuelson, Deborah Dila, Julie Menin and Elisabeth A. Raleigh Protein Expression with T7 Express Strains, competent cells application note, New England Biolabs.
[2] https://www.leniobio.com/about-alice/
[3] Gregorio, N. E., Levine, M. Z., & Oza, J. P. (2019). A user’s guide to cell-free protein synthesis. Methods and Protocols, 2(1), 24. https://doi.org/10.3390/mps2010024
[4] Buntru, M., Vogel, S., Spiegel, H., & Schillberg, S. (2014). Tobacco by-2 cell-free lysate: An alternative and highly-productive plant-based in vitro translation system. BMC Biotechnology, 14(1). https://doi.org/10.1186/1472-6750-14-37
[5] Halperin, S. O., Tou, C. J., Wong, E. B., Modavi, C., Schaffer, D. V., & Dueber, J. E. (2018). CRISPR-guided DNA polymerases enable diversification of all nucleotides in a tunable window. Nature, 560(7717), 248–252. https://doi.org/10.1038/s41586-018-0384-8
[6] I. Zaenen, N. van Larebeke, H. Teuchy, M. van Montagu, J. Schell,
Supercoiled circular DNA in crown-gall inducing Agrobacterium strains,
Journal of Molecular Biology,
Volume 86, Issue 1,
1974. https://doi.org/10.1016/S0022-2836(74)80011-2.
[7] Experimental Plant Division, R. B. R. C. (2021, September 7). RPC00001: Nicotiana tabacum by-2 cell suspension culture. rpc00001: Nicotiana tabacum BY-2 cell suspension culture - BRC plant cell line documentation. Retrieved from https://plant.rtc.riken.jp/resource/cell_line/web_documents/cell_lines/rpc00001.html.
[8] Nagata, T., Inzé, D., & Matsuoka, K. (2006). Tobacco BY-2 Cells: From Cellular Dynamics to Omics (1. Aufl., Vol. 58). Springer-Verlag.
https://link.springer.com/content/pdf/10.1007%2F3-540-32674-X.pdf
[9] Mustafa, Natali R, et al. “Initiation, Growth and Cryopreservation of Plant Cell Suspension Cultures.” Nature Protocols, vol. 6, no. 6, 2011, pp. 715–742.
[10] Widdel, F. (2007) Theory and Measurement of Bacterial Growth. Universiteit Bremen, Germany
[11] Riboswitches, RNA Regulation | Learn Science at Scitable [Internet]. [cited 2021 Oct 18]. Available from: https://www.nature.com/scitable/topicpage/riboswitches-a-common-rna-regulatory-element-14262702/
[12] Muranaka N, Sharma V, Nomura Y, Yokobayashi Y. An efficient platform for genetic selection and screening of gene switches in Escherichia coli. Nucleic Acids Res [Internet]. 2009 Apr 1 [cited 2021 Oct 18];37(5):e39–e39. Available from: https://academic-oup-com.kuleuven.e-bronnen.be/nar/article/37/5/e39/2410494
[13] Nomura Y, Yokobayashi Y. Dual Genetic Selection of Synthetic Riboswitches in
[14] Page K, Shaffer J, Lin S, Zhang M, Liu JM. Engineering Riboswitches in Vivo Using Dual Genetic Selection and Fluorescence-Activated Cell Sorting. ACS Synth Biol [Internet]. 2018 Sep 21 [cited 2021 Oct 18];7(9):2000–6. Available from: https://pubs-acs-org.kuleuven.e-bronnen.be/doi/full/10.1021/acssynbio.8b00099
[15] Dwidar M, Yokobayashi Y. Riboswitch Signal Amplification by Controlling Plasmid Copy Number. ACS Synth Biol [Internet]. 2019 [cited 2021 Oct 18]; Available from: https://pubs-acs-org.kuleuven.e-bronnen.be/doi/full/10.1021/acssynbio.8b00454
[16] Beisel CL, Smolke CD. Design Principles for Riboswitch Function. PLoS Comput Biol [Internet]. 2009 Apr [cited 2021 Oct 18];5(4). Available from: /pmc/articles/PMC2666153/
[17] Waldo GS, Standish BM, Berendzen J, Terwilliger TC. Rapid protein-folding assay using green fluorescent protein. Nat Biotechnol 1999 177 [Internet]. 1999 [cited 2021 Oct 19];17(7):691–5. Available from: https://www.nature.com/articles/nbt0799_691
[18] Zhou L-B, Zeng A-P. Engineering a Lysine-ON Riboswitch for Metabolic Control of Lysine Production in Corynebacterium glutamicum. ACS Synth Biol [Internet]. 2015 Dec 18 [cited 2021 Oct 20];4(12):1335–40. Available from: https://pubs.acs.org/doi/full/10.1021/acssynbio.5b00075