Similitudes del lenguaje en la ascesis espiritual y en la imagen de la poesía

Though the state-based latch was recombined efficiently under the operation of two different signals (set reaction: arabinose; reset reaction: arabinose and aTc) in strain DS941Z1, the device did not perform as anticipated, switching between BP and RL state under the control of a single signal in DS941 (section 4.4.1). The switch-on reaction was very inefficient in DS941 even with a very short pulse of arabinose. It was hypothesized that the inefficient switch-on reaction was caused by the expression of Gp3 as soon as the device started switching, inhibiting further switch-on reaction. To extend the interval between degradation of Int and expression of Gp3, a delay for Gp3 expression was added by inserting the tetR gene under the control of sequence state, resulting in the binary counting module. The expected performance of the binary counting module is described as below (Fig. 4.22).

In this module (with tetR), the expression of Gp3 was delayed and the switch-on reaction efficiency (with 30 min arabinose pulse) was improved to about 90% compared to the device without tetR (section 4.4.1).

It was proved that using a slightly longer preculture length (two hour) could promote quicker recombination reactions. The explanation might be that the induction of Int was more efficient later in the exponential growth phase. It was also thought that there would be more Tet repressor accumulated in cells at this time point, leading to tighter repression of Gp3 and improved switch-on reaction. Furthermore, a longer preculture length might also beneficial to accumulate Gp3 when the substrate was in RL state, improving switch-off reaction efficiency.

Such binary counting module proved the ability of changing state in response to a single repeating input signal. After encountering an odd number of input signal, the device switched

Figure 4. 22 Expected performance of the binary counting module with single induction pulse. This device

can be switched between state 0 (BP) and state 1 (RL) by the Int pulse induced using a single input signal. The change of protein expression level is represented using symbols (↑, increase; ↓,decrease).

to an opposite state to the initial one. Whereas, after encountering an even number of input signal, device switched to same state as initial. By tuning the RBS strength of gp3 and using optimal induction conditions (two hour preculture length and 15 min length of input pulse), the efficiencies of the binary counting module maintained a range of 80%-95% for switch- on reaction and 60%-90% for switch-off reaction during a muticycle operation (Fig. 4.20). There was no obvious degradation in efficiency after multiple induction operation, but the fluctuations of efficiencies were obvious. These fluctuations might be caused by the stochastic factors during the culture and induction (such as temperature, inducer, and time). More repeated experiments are needed to confirm these performances and to find out an optimal operation condition.

It was noticed that starting the recombination from the pure substrate plasmid (BP state or RL state) showed better recombination efficiency than starting from a mixture with the same induction pulse (e.g. 94% of lane 2 in Fig. 20A compared to 87% of lane 3 in Fig. 4.20B; 88% of lane 2 in Fig.4.20B compared to 60% of lane 3 in Fig. 4.20A). The explanation might be that the module state had great influence on the recombination efficiency since the expressions of both Gp3 and TetR were governed by the state of the substrate sequence. Starting from pure BP substrate (TetR expression on, Gp3 expression off), all of the cells expressed sufficient amount of Tet repressor to delay Gp3 expression, resulting in more efficient switch-on reaction; starting from pure RL substrate (TetR expression off, Gp3 expression on), none of the cells expressed Tet repressor to repress Gp3 expression, resulting in more efficient switch-off reaction.

A limitation with the binary counting module is that the reaction efficiencies are influenced by the length of the induction pulse, and only a small range of pulse lengths can fulfill the requirement of high recombination efficiency. The previously constructed recombinase- based counter (Friedland et al., 2009) and push-on push-off switch (Lou et al., 2010) lost function with a long time input signal. To avoid this problem, Subsoontorn and Endy (2012) proposed the design principle of an activation-inhibition toggle flip-flop by connecting two independent recombinase based set-reset latches to store state during and between input pulses (Subsoontorn and Endy, 2012). However, this system is complex and two integrases are involved in recording one bit information which limits the information storage capacity. Thus, the future direction of developing the binary counting module constructed in this study is to extend its functional pulse width.

Though the plasmid-borne memory devices gave promising recombination efficiency when viewing the overall DNA states, the individual cell states were not consistent because mixed populations of multi-copy substrate plasmids were generated in each cell. To solve this problem, the invertible substrate sequence could be integrated into the genome, which would decrease the copy number to one. This change was expected to divide cells into either a completely OFF or completely ON state, without any other intermediate states. The next chapter will discuss how to transfer the substrate sequence into the chromosome of E. coli strain and the performances of different chromosomal memory devices.

5 The design and engineering of chromosomal

memory devices

5.1 Introduction

In chapter 4, three different plasmid-borne memory devices, the set-reset latch (section 4.2), the state-based latch (section 4.3), and the binary counting module (section 4.4), were constructed. All three devices showed efficient recombination performances as measured by the overall substrate DNA state (Fig. 4.3, Fig. 4.13, and Fig. 4.20). The fluorescent performance of individual cells which contained the binary counting module was checked by flow cytometry. These cells showed variant fluorescent intensities after recombination, especially after the attR×attL recombination (Fig. 4.21). The mixed cell populations were thought to be generated by recombination on a multi-copy substrate plasmid, so that each individual cell can contain a mixture of plasmids in different states (Brophy and Voigt, 2014; Moon et al., 2011). In order to mitigate the inconsistency of cell states and increase the stability of the genetic device, the target sequence was integrated into the chromosome of

E.coli strain DS941 or DS941Z1 as a single copy. It was expected that this change would

also improve the recombination efficiency, since the initial data collected in chapter 3 showed more efficient recombination on a low copy-number substrate plasmid than that on a high copy one (section 3.3.2 and section 3.3.3).

In the first part of this chapter, the method used to construct the chromosomal memory devices is introduced. To integrate a single copy substrate sequence into the chromosome, the invertible substrate sequence (from the state-based latch or the binary counting module) was placed on a suicide ISY100 transposon delivery plasmid and mobilized by RP4 conjugative machinery (Demarre et al., 2005; Ferrières et al., 2010; Urasaki et al., 2002).

Later in this chapter, methods used to check the recombination efficiencies of the chromosomal devices are discussed. These methods include fluorescence microscopy, flow cytometry, and colony PCR. Flow cytometry was found to be an efficient method and was therefore used to check the behaviours of the chromosomal devices in all the subsequent experiments. Based on this initial characterisation, tests of the chromosomal devices are presented and ways to improve their efficiencies are discussed.

5.2 Chromosomal delivery of memory devices using