Introduction

Ave Caesar

Materials and Methods

Starting PACE

Before starting a PACE run, several prerequisites have to be fulfiled. Most of the preparations for the different PACE runs are the same in terms of tests for MP activity, F-Pilus plasmids and contamination. Additionally, several pre-tests are recommended to test the APs activity and the general functionality of the implemented genetic circuit.

1. Every part of the PACE device including all tubings and connectors have to be autoclaved. All open ends should be wrapped in aluminium foil. It is important to check all ends and tubings to be closed before start the dry autoclavation (Be aware of autoclaving only autoclavable parts of the PACE device).
2. After autoclavation, the PACE device should be treated with highest carefulness to prevent phage contamination in the turbidostat. To make this possible, the use of 10% H2O2 or incidin as well as the usage of gloves is advised.
3. Rebuild the PACE apparatus carefully using incidin to desinfect all for the autoclavation wrapped and thereby closed ends. Connect all necessary parts of the tubings.
4. The medium should be prepared slightly different to the medium used in literature [Esvelt et al., 2011] by mixing 140g dikaliumhydrogenphosphate with 40g kaliumdihydrogenphosphate, 20g ammoniumsulfate and 20 ml tween-80 in 20l dH20. The medium should be autoclaved as well before using it.
5. the autoclaved medium should be mixed with medium supplements, which should be prepared during autoclavation. 20g glucose, as well as 10g sodium citrate, 0.5g L-leucin, 0.5g and 100g casamino acids or trypton from casein have to be solved in at last 500 ml dH20. If the chemicals cannot be dissolved in this volume, water can be added until it can be solved. The resulting solution have to be steril filtrated.
6. the appropiate volume of the prepared supplements can now be added to the autoclaved medium. This should be implemented in as steril conditions as possible, using incidin to sterilize the used pipette. In addition to the supplements, the appropiate antibiotics have to be added into the medium. Final concentrations should be choosen according the stock concentrations proposed by addgene. A blank for the OD600 measurements should be taken before connecting the medium to the tubings.
7. After connecting the media line of the turbidostat to the medium container, the turbidostat should be filled with medium until a volume of 1.5l is reached, by starting the media pump.
8. 50 ml bacterial culture resulting from the *MP testing* should be used for inoculation. Therefore draw up the culture into a syringe and inoculate the turbidostat using a cannula through the septa in the turbidostat. Reduce the flow rate to a minimum to ensure an efficient growth of the culture in the turbidostat.
9. lagoon pump can be started when the turbidostat reaches an OD600 = 0.6 - 1.0. The lagoon volume can be adjusted at a range of 100 - 150 ml lagoon volume.
10. induce mutagenesis by start adding 10% w/v arabinose to the lagoon. Arabinose should be added at last one hour before infection with bacteriophages to secure the induction of the MPs
11. When the lagoon is ready, arabinose is added and the cells are on a constant optical density, the lagoon can be infected with bacteriophages. Add 1 ml of 10 ^{10>sup> PFU/ml to the lagoon and start the existing PACE run}
12. during the PACE run, samples should be taken every four hours for the first 24 hours and every eight hours from the second day on until the run is finished. During a PACE run, phage detection PCRs and plaque assays should be implemented, proving the presence of the phage of interest and a contamination free turbidostat. Positive and negative control always have to be included into the detection PCR as well as the plaque assays. We recommend

Results

Dicksinson-PACE

Based on the high-potential concept of Phage-assisted continous evolution the group of Bryan Dickinson evolved a T7 RNA polymerase in the "Evolution of a split RNA polymerase as a versatile biosensor platform"-paper, publicated in early 2017(RN158). We choose this example of PACE as our role model for testing of this highly complex method, needing the set-up of a continous culture, as well as a flow setup, demanding costly and highly sensitive hardware for its application. In the paper, Pu et al. grasp at evolving a small N-terminal domain of the T7RNAP for efficiently binding its N-terminal domain by help of two leucine zipper, strongly binding each other. Depending on their proximity, the two domains of the RNAP interact and can activate geneIII transcription. Taking this highly challenging experiment, we wanted build our own PACE device and investigate the general handling of the apparatus. We were first started with building up our PACE apparatus using the instructions of Kevin Esvelt back in 2011 (RN44´). For an accurate description of our PACE device and our procedure before starting PACE, please have a look at the Materials&Methods section. Thankfully, Bryan Dickinson´s lab was very helpful, so that they were sending us their constructs, strains and the phages they used for their evolution. Based on this, we could design our own constructs. Beyond that, we could quickly test our turbidostat, which we used instead of a chemostat, Dickinson and colleagues used for there PACE device. First, we aimed at cultivating our bacterial strains at an nearly constant optical density. This turbidostat-test was also used for evaluating the media consumption of the apparatus during an estimated PACE run suitable for iGEM conditions. Since the original PACE run has to include 29 days of continous evolution as well as ten different strains with different positive and negative selection stringencies, we were targeting at a shorter variant of this PACE experiment. By analyzing most of the data for the PACE run, we detected 3 of 7 essential mutations of the final variant of the N-terminal domain to be present in phages after only seven days of evolution. Outlining a more realistic example of a PACE run during a project implemented in only five wet lab month, we aimed for reproducing some of the mutations detected in the first seven days. In this context, we especially wanted to learn to handle the most difficult aspects of PACE:

• phage washout
• turbidostat contamination and working without contamination
• flow rates
• induction of mutagenesis
• evaluation of successfully performed PACE
  During several short PACE runs we already faced two of these challenging problems. First we had to adjust the flow rates in comparison to the flow rates in Dickinson-PACE based on the fact that we (i) used a turbidostat instead of a chemostat and (ii) investigated a low growth rate of the bacterial culture and thereby a doubling time of more than the expected 30 minutes. Second, we faced problems with a phage contamination in the turbidostat due to an extendable design of our PACE device and the waste tubings. Based on this observations, we slightly adjusted the tubings on our apparatus, yielding the actually used design. In the next steps we further tried to continously cultivate the N-terminal T7 RNAP phages with the first strain, used for PACE in the original paper. Since you cannot detect the phages simultanously during running PACE, we wanted to establish a quick detection method, which reveals a result in a shorter time than the usually used detection and quantification method - plaque assays. Several methods like the dot blot with M13 antibodies or the qPCR did not showed reliable and reproducible results. In contrast to these methods, a simple phage detection PCR using three characteristic PCR products in lengths of 200, 400 and 750 bp. The results of this PCR can be quickly analyzed on a agarose gel and yield an final result for the presence of your phages in the lagoon in only two hours. This detection method enables a more efficient and direct way of controlling the turbidostat.
  
  Figure 2: Agarose gel of phage detection PCR
  This image shows the results of an phage detection PCR of samples from the final Dickinson-PACE run. NEB 2-log ladder was loaded on the first lane. All lanes are
  Following the phage propagation experiment further, we not only wanted to propagate the phages but trying to start our directed evolution in analogy to the first seven days of the original publicated PACE run. During this seven days, three different strains with different negative APs were used to integrate various selection stringencies. To implement this, the required APs were transformed with MP1 in S1030 cells following the *transformation protocol*. By performing seven days of PACE, it was necessary to change strains for the first time. This presents another challenging step, since a contamination during the build-up procedure pose a risk for contamination again. On this account, it should be avoided if possible. On this basis, first seven days of PACE were implemented. For efficient organisation, the expiry was documented before the run (siehe figure PI-PACE). This enables several team members to carry the experiment forward, without decrease in quality. The PACE run was successfully performed for 137 hours, displaying an acceptable phage titer in comparison to the literature (Esvelt et al., Nature 2011; Pu et al., Nature 2017). Sadly, we were running into another problem, which was highly reported by Kevin Esvelt - phage washout. The results of our plaque assays showed no infective phages in our lagoons after the 23. timepoint, corresponding to 137 hours of PACE. In addition, we could not reveal any mutations in our sequencings in a number of eight plaques which were picked for detection of potential mutations in the N-terminal domain of the RNAP.
  
  Figure 2: Design of the Accessory Plasmids for the Evolution of Cas9
  The AP consists of five subparts that are devided by homology regions for Gibson assembly (numbers). It carries an expression cassette for the transcription of a gRNA (between 1 and 5). GeneVI (2-3) is under control of a that can be activated by the Cas9-rpoZ in context with the respective gRNA. luxAB accounts as a reporter for fluorescent readout of geneIII activation (3-4). The whole plasmid can be produced with different origins of replication (4-5) to modulate the copy number and by exchanging the geneVI part with the RBS.
  Facing these problems, the general experimental design of our PACE device, as well as the bacterial strains and the inducer concentration were investigated for potential mistakes. Based on the observation, that we can detect phages even after almost six days, we focused on the induction of mutagenesis during our PACE run. As the mutagenesis is induced by arabinose and inhibted by glucose, we primarily concentrated on the glucose and arabinose concentrations in our PACE apparatus. Therefore, our modeling hypthesized a too high glucose concentration in our medium, as well as a low arabinose concentration, which could be a reason for neither mutations in our picked plaques. Together with this theoretical assumption, we developed an alternative PACE setup for testing of the induction of mutagenesis during PACE.

  Random-mutagenesis PACE

  Following our first PACE tests based on the the split T7 RNAP paper, we designed a new PACE test, using our modeling to estimate the glucose concentration in the lagoons as well as in the turbidostat. First tested in a pre-test using PREDCEL, we tried to transfer equivalent conditions yielding mutations in PREDCEL on the PACE apparatus. In order to do that, an alternative bacterial strain which contains the pJC175e plasmid, a plasmid usually used for plaque assays, providing geneIII under a psp-promotor for all phages was used. In addition to this propagation plasmid, MP1 and MP4 were transformed into one strain respectively. These strains were used for propagation of phages, used for the Dickinson-PACE testing too. Since the Dickinson-phage propagates in previously performed pre-experiments very well, the effect of the concurrent performed propagation and mutagenesis should be at its maximum. This is further underlined by the fact that in this case, geneIII is provided for free. Thus, there should be no selection pressure except from pressure on the best propagation regarding codon optimisation. Beyond that, a lot of random mutations should be observed when the MPs work reliable. To reach this point, three days of mutations using the MP1 and the MP4 strain simultanously in two turbidostats with one lagoon for each turbidostat were implemented. Sequencings were performed by picking eight plaques of each lagoon from the last sample included into the plaque assay. In this case, phage washout was not observed, displaying a phage titer which were settled between 10^4 and 10^7 in both lagoons.
  
  Figure 2: Design of the Accessory Plasmids for the Evolution of Cas9
  The AP consists of five subparts that are devided by homology regions for Gibson assembly (numbers). It carries an expression cassette for the transcription of a gRNA (between 1 and 5). GeneVI (2-3) is under control of a that can be activated by the Cas9-rpoZ in context with the respective gRNA. luxAB accounts as a reporter for fluorescent readout of geneIII activation (3-4). The whole plasmid can be produced with different origins of replication (4-5) to modulate the copy number and by exchanging the geneVI part with the RBS.
  Sanger sequencing results from these plaques showed a high number of different mutations in the 16 picked plaques. Thereby, the number of mutations resulting as a consequence of MP4 usage was slightly higher than the number of mutations in phages cultivated with the MP1 strain. In total, the mutations were widely spread over the N-terminal domain of the RNAP. Certainly, one mutation was recurrent, even the fact that there were no selection pressure. Overall, 15 of 22 mutations acquired during this three day PACE run entailed an amino acid exchange in the domain, while one mutation which occured in four clones resulted in no amino acid exchange.
  
  Figue 2: Random mutations in N-terminal domain of T7 RNAP
  The figure is presenting the nucleotide sequence of the N-terminal domain of the T7 RNAP in light grey, a linker sequence in black and the leucine zipper in dark greyRN158. At the top of the figure a consensus like sequence is shown, summarizing all mutations of the sequenced plaques, which are displayed as single bars under the summary. All mutations annotated are color coded. Recurrent mutations are divided into mutations leading to an amino acid exchange (red) and the mutations which do not have an amino acid exchange as a consequence (orange). In addition, mutations with amino acid exchanges, which are not recurrent are displayed in yellow, while the mutations without amino acid exchange are shown in blue. In total, only plaques showing mutations are shown. Sequences which did not contain mutations were omitted due to clarity reasons. Overall, sequencing was performed on 24 plaques.
  By this, we demonstrated that the we can induce the mutagenesis plasmids successfully and sufficient manner. The evaluated conditions, hypothesized by our modeling, confirmed by our sequencing results from the real experiment should finally describe conditions in which a good induction of the MPs should be ensured.

  PI-PACE

  Coming to our final experiment we finally had the knowledge which is required to perform our own PACE run. This PACE run is just like Dickinson´s PACE approach based on protein-interaction of a split T7 RNAP. In contrast to the run before, both split sites are located on the selection phage. In principle, the split T7 RNAP is evolved on a better and faster reassembly of both fragments, yielding in a higher transcription of geneIII which is encoded under control of a T7 promotor. For further information on the principle of protein-interaction PACE, please visit our special site (hier Link einfügen). Building up on the equivalent conditions to the random-mutagenesis PACE run, we used the same amount of glucose and arabinose for induction of the mutagenesis plasmids. In this case, we only used a strain with MP4 due to the observation of slightly more mutations in the random mutagenesis experiment. Since this PACE experiment was performed with selection pressure, we estimated for difficulties in the phage propagation during PACE, which is why the flow rate was decreased for enabling better phage propagation. Nevertheless, our plaque assays showed phage washout after only 38 hours. Regardless of these findings, plaque PCRs and sequencings were performed, using plaques from the last available time point. The sequencing results showed one mutation in each of the split sites, of which one mutation could have a functional input on the reassembly of both sites.

  

  Figure 2: Agarose gel of phage detection PCR

  This image shows the results of an phage detection PCR of samples from the final Dickinson-PACE run. NEB 2-log ladder was loaded on the first lane. All lanes are

  {Heidelberg/templateus/Imagesection