FASE DE EJECUCIÓN 4.1. Diagnóstico energético

Although the created split mDHFR variant can be regarded as our first switchable metabolic enzyme, it is quite apparent that it does not represent an optimal switch. For instance, we observed a high basal growth rate in both, the negative control strain as well as the FRB- mDHFR1,2 and FKBP12-DHFR3 expressing strain when grown in absence of rapamycin. This is probably due to an interaction of the mDHFR fragments independent of the fused regulatory proteins. Although we did not expect the observed extent of complementation as a result of random interactions, it could be explained by the structural properties of split proteins. Many proteins, including mDHFR, contain hydrophobic amino acids in their catalytic center. When split, these amino acids are in contact with the cytosol, which is energetically unfavored 116. In contrast, when reassembling as a consequence of an interaction of fused regulatory proteins or random interactions, the hydrophobic residues of the two fragments will interact as well, resulting in a stable, energetically more favored state. We can therefore assume that random interactions occur and that once reassembled split proteins will have certain robustness against a re-separation.

In addition, three other factors may have contributed to the observed high basal growth rate: First, it might be that the mDHFR variants were overabundant as a consequence of very high gene expression levels. The higher the copy numbers of split mDHFR variants are within the cell, the higher is the likelihood of random interactions of both fragments. A reduction of gene expression by reduced addition of IPTG might lead to a lower basal activity but would presumably also reduce the growth rates. As the growth rates and 7,8-dihydrofolic acid levels of D81 did not even reach the levels of the positive control in presence of IPTG and 10 µM rapamycin, a reduction of IPTG might only be advisable when the enzyme switch is further optimized.

In addition to that, it should be noted that the addition of casamino acids to the medium reduces the need for functional DHFR as folates are amongst others required for the biosynthesis of serine, methionine and glycine which can be taken up from the surrounding medium.

Finally, it might be that the linker sizes and compositions are not optimal. As mentioned above, the linker was primarily designed to allow both, the enzyme and the fused proteins, to fold correctly and has therefore been chosen to be long and flexible. It might be that the linkers are too long and flexible so that a random interaction of both mDHFR variants is not sufficiently prohibited. Shorter or rigid linkers could improve the switches by reducing unwanted random interactions between enzyme fragments. However, such linkers bear also the risk that folding of adjacent proteins or interaction of proteins with their counterparts might be impaired.

Both, optimal gene expression strengths and linker compositions are planned to be subjects in future rounds of re-design, creation, (possibly) screening and evaluation.

Similar to the assumed unfavored reversibility of assembly of the mDHFR fragments, it has been shown that the dimerization of FRB and FKBP12 is most likely irreversible as well 107_{. In}

Chapter 4 - Creation of switchable enzymes using the Split Protein approach

case a reversibly switching enzyme should be created, one therefore either has to use ligands such as FK506 which compete with rapamycin to bind to FKBP12 117 or use an alternative regulatory protein with a higher tendency to dissociate in absence of its effector.

In addition to these limitations, as rapamycin is a very expensive compound, its usage as inducer molecule in bioprocesses is not eligible, a usage of FRAP/FRB to control overproduction pathways is therefore not possible.

Despite the disadvantages of this existing switch, we would like to use it in the future as a platform to create new switches. For instance, we plan to replace the two mDHFR fragments with enzymes of more biotechnological relevance such as ArgA, LeuA or GPD1 of the glycerol production pathway of S. cerevisiae. It should be noted though that these enzymes are in contrast to mDHFR not active as monomers but instead form oligomers in order to be active. Instead of fragment reassembly we would therefore try to control in these cases the oligomerization and in that way the activity of the enzymes through the fused regulatory domains (Figure 20).

Figure 20: Control of oligomerization.

Many proteins that could potentially be used as regulatory proteins (red) are forming oligomers and only change their conformation upon binding of an effector (green). This conformational change can be used to control either reassembly of split proteins or to control the oligomerization of full length proteins that need to form oligomers in order to be active. Oligomer forming enzymes might form oligomers independent of the fused regulatory domain. In such cases, enzymatic activity is controlled through conformational changes upon binding of an effector.

As the linker composition might be important in these cases as well, we hope to profit from the experiences that we will hopefully gain from the experiments about the optimization of the linkers connecting the mDHFR fragments with FRB and FKBP12.

We will also try to replace FRB/FKBP12 with other regulatory proteins. In particular, we are interested in regulatory domains that sense and are active dependent on growth condition

Chapter 4 - Creation of switchable enzymes using the Split Protein approach

such as Cra usually form oligomers independent of the presence or absence of the effector. In contrast to FRB/FKBP1,2, split proteins with effector binding domains of transcription factors as regulatory domains might therefore interact constantly and enzyme activity will not be controlled by protein assembly but through a transmission of the conformational change the fused regulatory domains undergo upon binding of their effectors.

We already tried to construct of an oxygen-dependent mDHFR version by fusing both mDHFR fragments to the oxygen-binding domains of the transcription factor FNR. So far, with the chosen linkers and in the given growth conditions, no complementation of the DHFR knockdown phenotype could be observed.

Planned adaptions of the created split protein are depicted in Figure 21.

Figure 21: Planned adaptions and optimizations based on the created mDHFR- FRB/FKBP12 split protein switch.

Chapter 5 - Creation of switchable enzymes using the Domain Insertion approach

5 Creation of switchable enzymes using

the Domain Insertion approach

The second directed evolution-based method that we evaluated in this work to create metabolic enzymes with synthetic allosteric regulation is Domain Insertion in which an enzyme of interest is randomly inserted into a regulatory domain which changes its conformation upon binding of an allosteric effector 65,73_.

Proteins that are intended to be used as regulatory domains or metabolic enzymes can be differently suited for Domain Insertion. Our first goal was therefore to define guidelines for the choice of components.