Engineering of Pex5 and PTS1

Designing our receptors

To achieve the engineering of an orthogonal import pathway, we followed two approaches regarding Pex5. The first is based on a targeted mutagenesis based on educated guesses which is first verified by molecular dynamics and later experimentally in the laboratory. The second approach is based on a recently published paper: During our research we came across a paper by Cross et al., published in 2017, in which they present a synthetic construct of the Pex5 protein, partly consisting of Arabidopsis thaliana and partly of Physcomitrella patens. Compared to the wild type Pex5, this construct shows different binding affinities since it interacts with a PTS1* variant that does not interact with the wild type Pex5 Cross et al. . Since the protein sequences of the Pex5 of yeast and plants differ significantly, we aligned both sequences to understand the specific mutations.

The alignment shows three red marked amino acids we changed in our receptor sequence. Interestingly, these mutations are located within the TPR motifs of our Pex5 protein and this persuaded us to try out this receptor, we call it R19. Due to lack of time we tested this Pex5 variant in silico and in vivo simultaneously. We started molecular dynamics simulations with a couple of PTS variants that we already tested with our previous designs − one of them was actually the variant they used in the paper (YQSYY). The details and results of our structural modeling can be found in the modeling section.
Furthermore, we synthesized this variant and together with two receptors we designed based on educated guesses we got three receptors for our experimental work.

Experimental design

Verification of peroxisomal protein import was performed by tagging the fluorescent protein mTurquoise with our designed PTS variants. Additionally, a peroxisomal membrane protein was used to ensure peroxisomal localization. For that reason, we chose the transmembrane domain of Pex13 tagged with the fluorescent protein mRuby.

Pex13−mRuby

We used the peroxisomal membrane protein Pex13 as a fluorescent marker − by just using the transmembrane domain of Pex13 with a short linker, we make sure that it has no influence on the peroxisomal features. To obtain a higher differentiation from mTurquoise, which we use for another construct, we chose to work with mRuby. Literature research revealed that such constructs have been tested before − Erdmann et al. (2004) described a construct containing only Pex13_200-310 with a C-terminal GFP.

Pex5 variant

In order to achieve an orthogonal peroxisomal protein import machinery we used a Pex5 knockout yeast strain in which we transformed our artificial Pex5 variant containing a modified PTS1 binding pocket. Our variation facilitates the detection of a non native PTS1 variant instead of the wild type PTS1. The construct contains a medium strength promoter, the Pex5 gene and a terminator. The remaining plasmid parts can be seen in the plasmid map below.

mTurquoise−PTS

Our approach for import verification is based on the fluorescent protein mTurquoise tagged with our modeled PTS variants. After several promoter tests with different strengths, we decided to express this construct only in low amounts, since this was the most suitable possibility to detect potential mTurquoise localization.
Our construct is depicted in the figure below.

Combination of our constructs

To combine our constructs, we cloned our Pex5 and mTurquoise constructs into a level 2 plasmid portrayed below.

Subsequently, co-transformation of the Pex13−mRuby plasmid and the level 2 plasmid was performed in order to verify peroxisomal colocalization.

PTS screening

Trusting on our targeted approach alone seemed risky − that is why we planned a PTS screening to find the most favorable PTS for our three receptors. Dueber et al. (2016) used the Violacein assay for a similar purpose. They screened for the best PTS for the wild type receptor and were successful. Hence another subproject of our team is the integration of the Violacein pathway into the peroxisome (Violacein), we were already supplied with all necessary enzymes − VioA, VioB and VioE.

The figure above shows the principles of the assay. VioA and VioB are localized in the cytosol and lead to the production of the IPA imine dimer while VioE is tagged with a PTS1 variant. Successful import leads to white colonies whereas missing import results in green colonies due to the cytosolic production of prodeoxyviolacein.
Our rests upon the following two plasmids which are co-transformed into yeast.

As shown above, we created one plasmid containing VioA, VioB and one of our Pex5 variants while the other plasmid only contained VioE. We then designed primers which bind to the VioE plasmid to amplify the whole plasmid except the terminator − random PTS1 variants were attached to VioE with the help of a random primer library. Following up, we did the ligation with the corresponding terminator and obtained a mix of several different VioE-PTS1 plasmids.
After plasmid amplification in Escherichia coli we then co-transformed yeast with the two constructs and waited for the colonies to grow. With the yeast growing, prodeoxyviolacein should be produced in yeast cells with absent import (green color) and the IPA imine dimer (white color) should be produced in those with functional import.
Plasmid preparation of those with white color and subsequent sequencing leads to the identification of functional PTS1 variants. Afterwards, we repeat the cloning steps described before to obtain a mTurquoise−PTS1* construct and co-transform it with the corresponding Pex5 variant. Eventually, the correct localization of mTurquoise tagged with these PTS1 variants provides proof for its function.

Mutagenesis of PTS2

Imagine you need different protein concentrations in your artificial compartment. What to do? Take our modified PTS2 sequences with varying import efficiencies.

To characterize the import efficiency for the site-directed PTS2 firefly luciferase was used. Luciferase is a luminescent protein which can be split in a C- and a N-terminal part. Only when combined, luminescence can be detected. To measure the import efficiency the two parts will be expressed and imported into the peroxisome in a separated way. The smaller part (Split2) of the split luciferase will be brought into the peroxisome first via the PTS1 dependent pathway. The other part is imported via the respective modified PTS2 sequence. The better this sequence is recognized by Pex7, the stronger the luminescence of the assembled luciferase can be detected in the peroxisome. There is a chance of split parts of the luciferase assembling in the cytosol if the import is too slow. To avoid wrong conclusions of the luciferase localisation, we designed a negative control experiment. It includes a split luciferase similar to the one used in the initial experiment, but without the peroxisomal targeting sequence. Consequently there will be no import into the peroxisome. If we subtract the luminescence of the negative control experiment from the luminescence of the main experiment we can define the degree of import.

In addition to a directed approach according to Kunze and colleagues we also want to perform a random mutagenesis experiment to alter the five variable amino acids of the core region of the PTS2 sequence in an unbiased manner. The aim is to generate a library of different peroxisomal PTS2. The 15 nucleotides are assembled by chance. In the DNA synthesis this sequence will either be described as [NNN]₅ or [DNK]₅. N stands for all four nucleotides mixed, K for either G or T and D for A,G or T. The “DNK” composition prohibits two out of three termination codons. Additionally with this library the amino acid frequency is improved towards a balanced ratio in between the different kinds (Dueber).

Each approach could generate up to 415 DNA sequences, which is roughly 1,07 billion. On the level of the amino acid sequence there are 3,2 million possibilities, since each residue can be taken by 20 different amino acids. For the assay we therefore need a high throughput method.

We adapted work of DeLoache, Russ and Dueber using the violacein pathway to measure the import effectiveness of tripeptides. The pathway consists of Violacein A (VioA), Violacein B (VioB) and Violacein E (VioE). It converts tryptophan into the green product prodeoxyviolacein (PDV). The first two enzymes, VioA and VioB, are expressed in the cytosol, and the third one, VioE, is targeted to the peroxisome with a PTS1 sequence. The degree of import can be measured by the intensity of green colour of the colonies. An efficient import signal leads to a strong import of the VioE into the peroxisome and subsequently to white colonies, because the intermediates cannot diffuse into the peroxisome to its respective enzyme. DeLoache et al. showed that there is a proportional correlation between the concentration of the green product PDV and a red fluorescent substance. The concentration of this product displays the import efficiency of the respective sequence.

This assay has been used for the evaluation of the generated PTS2 sequences. The VioE-PTS2 plasmids are harvested and cotransformed with a VioA-VioB plasmid. Each plasmid contains a specific auxotrophy marker. Consequently every growing colony contains both plasmids. To evaluate the respective sequence the concentration of the red fluorescence is measured. The more fluorescence is detected the more VioE is in the cytosol. Therefore the respective PTS2 is not that efficient. The other way around a low concentration of the fluorescent substance correlates with an efficient import via the respective PTS2.

Scientific background

Peroxisomal membrane proteins are synthesized on free polysomes in the cytosol and afterwards integrated into the membrane via two major pathways: one dependent on the endoplasmatic reticulum and one dependent on PEX19 and PEX3.

PEX19-dependent

ER-dependent

Experimental Work/Design

In order to test our hypothesis we fused the last 59 amino acids of the C-terminus of human PEX26 (AA 246-305) to a red fluorescent protein, to further elucidate the Pex3/Pex19-dependent import. mRuby is generally used as a marker in combination with a fluorescent microscope to visualize the localization of the fusion protein. The C-terminus of PEX26 contains a helical signal-anchor, which serves as both, a mPTS and transmembrane domain. We designed our construct with mRuby2 fused to the N-terminal side of the PEX26-C-terminus, this way the mRuby should face the cytosolic side of the peroxisomal membrane. Quite similar to our mRuby-PEX26 approach, we designed a construct for the ER-dependent import. Therefore, we fused the mRuby2 fluorescent protein to the N-terminus of Pex3 (AA 1-39). This construct should be N-terminally anchored in the peroxisomal membrane, with mRuby2 again facing the cellular lumen.

Our main goal is to introduce a rather complex membrane protein to the peroxisome that can alter specific traits. For that we fused the Pex3 N-terminus (AA 1-39) to a Halobacterium salinarum bacteriorhodopsin protein (AA 16-262), replacing the first 16 amino acids (Pex3-BacR). The original archaeal bacteriorhodopsin acts as a proton pump by capturing light energy to move protons across the membrane out of the cell. The resulting proton gradient is subsequently converted into chemical energy. Our assumption is that the first transmembrane segment determines the orientation of the following protein and that therefore due to the N-terminal anchoring signal the bacteriorhodopsin will be inserted in reverse orientation, pumping the protons into the peroxisome. This way the pH of the peroxisomal lumen could actively be controlled and adjusted.

Finally, our project involved combining the work of other subteams to verify the localization of our constructs in the peroxisome and analyze the effects they have on the import. Therefore, we are using the superfolded-GFP protein, another fluorescent marker, which is in our case fused to the peroxisomal import sequence PTS1, and a version of Pex11 that is fused to the fluorescent marker Venus. Both markers emit light in the green light spectrum, were as mRuby2 emits light in the red part of the spectrum, giving us a strong contrast and an easy way of differentiating between the two under the fluorescent microscope.

To physically create our constructs, we researched the DNA sequences of bacteriorhodopsin, Pex3 and PEX26 via UniProt and pre-designed our fusion constructs with the software tool „Geneious“. We ordered the synthesis of three separate parts (Pex3, PEX26 and Pex3-BacR) from IDT. To ease out the process of assembling our plasmids, we used the „Dueber Toolbox", containing various parts such as promoters and terminators, to tailor the plasmids specific to your needs. Finally, to combine all the selected parts, we used the „Golden Gate” assembly method.

Experimental Design

We will adapt the system of Sagt and colleagues to secrete the content of our modified compartments (Sagt et al, 2009) .
For the application of this system in S. cerevisiae we use a truncated version of the v-SNARE Snc1 to decorate our compartments (Figure 3.1) (Gerst et al, 1997) .

To decorate the compartments with the SNARE we use a peroxisomal transmembrane protein . In our case we use the proteins Pex15 or Pex26, which were further investigated in another sub project, and fuse Snc1 to the N-terminus. We expressed these constructs of membrane anchor and Snc1 constitutively under control of the RPL18B promotor. In case of Pex15 we used a truncated version, lacking a large part of the N-terminus, only consisting of the transmembrane domain (315-383) (Figure 3.1). For Pex26 we use the truncated version published in Halbach et al. (Halbach et al, 2006) .

We verified our secretion using beta-glucuronidase (GUS) as a reporter protein. In 2012 Stock and colleagues described the GUS reporter assay for unconventional secretion (Stock et al, 2012) . With it, it is possible to determine whether a protein is secreted conventional and is N-glycosylated or secreted unconventional and not N-glycosylated. GUS is a bacterial protein with an N-glycosylation-site, which is active only if the protein is not N-glycosylated. The GUS-activity can be measured with different reagents in plate or liquid assays. Liquid assays can be applied qualitatively as well as quantitatively to measure differences in activity. If GUS is secreted by the conventional pathway the N-glycosylation leads to inactivation of the enzyme (Stock et al, 2012) (Fig 3.2).

GUS will be imported to the peroxisome with the PTS1 sequence and measured quantitatively in the supernatant. We will use a coexpression of GUS-PTS1 and Snc1-Pex15 or Snc1-Pex26 to identify the secretion of the compounds. Furthermore, we will use GUS-PTS1 expressed in S. cerevisiae without Snc1 fused to a membrane anchor for a control. We will measure the active GUS in the supernatant with a liquid assay based on the turnover of 4-methylumbelliferyl-beta-D-glucuronide to 4-methyl umbelliferone (4-MU) (Blázquez et al, 2007) . Here we expect a higher activity of GUS in the supernatant of cultures with Snc1 decorated peroxisomes.
To increase the variability of our constructs we also designed vectors with and without a GS-Linker connecting the Snc1 with the Pex15. Additionally we tested our constructs in strains with a deletion of Pex11 . This deletion leads to formation of larger peroxisomes and may increase the efficiency of our secretion mechanism.

Size&Number

Pex11

For our first approach we took to regulate size and number of our compartments, we chose to work with Pex11, which is naturally involved in the constriction of peroxisomes. Thereby, an overexpression of Pex11 leads to a large amount of smaller peroxisomes, whereas a knock-out provides a small amount of enlarged peroxisomes.
To achieve full control of peroxisome size and number with Pex11, we first aimed at showing the expression of Pex11 in the cell and its integration into the peroxisomal membrane. Therefore, we used parts of the Dueber Toolbox.

This way we were able control the strength of the gene expression. To reveal the successful expression we designed Pex11 as a Toolbox part and fused the fluorescence protein mVenus to the n-terminal end of the peroxin. This way, we could both detect the expression and the localization in the peroxisomal membrane. To measure the change in size and number of the peroxisome, the yeast cells had to be observed under the microscope. For the examination of the effect of different Pex11 concentrations in the cell, from total deletion to overexpression, we transformed all constructs into the wild type as well as in the PEX11$\Delta$ strain. For comparison with the normal peroxisome phenotype in both strains we use sfGFP fused to a PTS1 sequence which is transported into the peroxisomal matrix. The deviation in peroxisome morphology indicates the functionality of our constructs.

Pex34

Similar to Pex11 we aimed at examining the impact of different concentrations of Pex34 on the peroxisome morphology. In this case we investigated a concentration range between the wild type phenotype and different levels of over expression, achieved by different constitutive promoters and inducible promoters.

ue to the fact that Pex34 is fully inserted into the membrane, a fusion with a large fluorescence protein could lead to the degradation of this peroxin. Therefore we worked with two different approaches.For the first approach, we fused an mTurquoise protein to the C - terminal end of Pex34, assuming that it doesn't affect the native structure and the import into the membrane.

As a second approach we decided to express Pex34 without any tag in order to conserve its natural function.

To show the variability of the peroxisome morphology in the second approach we used sfGFP-PTS1, which is imported into the matrix to visualize the change in their size. To analyze the success of our constructs we used fluorescence microscopy, either searching for a membrane localization of mTurquoise or enhanced and amplified peroxisomes visualized by sfGFP.

Sensors

To enrich our toolbox we decided to measure four essential physiological factors: ATP, NADPH, Glutathione and pH.
ATP/ADP conversion is used as an energy currency in many cellular processes like translocation of proteins and metabolites or anabolic and catabolic turnovers. It is assumed that ANT1P, an ATP/AMP antiporter, is located in the peroxisomal membrane. Furthermore, the shuttle mechanism of the peroxisomal protein import machinery is ATP dependent (Palmieri L. et al., 2001). NADPH plays an important role in anabolic pathways and is also indispensable to our desired pathways of nootkatone and violacein. Glutathione is an antioxidant and redox buffer which is also found in yeast peroxisomes. It is used as cofactor by at least two types of peroxisomal proteins, the glutathione peroxidases and glutathione transferases, which reduce lipid- and hydrogen peroxides or transfer glutathione to lipid peroxides for the purpose of detoxification (Horiguchi H. et al., 2001). Furthermore, it partly represents the redox state of the peroxisome. Knowing the pH of a compartment is important for predicting the activities of almost all enzymatic processes inside and to follow up acidification and basification upon conversion of metabolites.

We finally chose two ratiometric sensors to perform measurements with. The pH sensitive and glutathione redox state reporting green fluorescent proteins pHLuorin2 (Mahon M. J. et al., 2011) and roGFP2 (Schwarzländer M. et al., 2016). Our aim was to target them either to the peroxisomal lumen or the cytosol. To achieve peroxisomal targeting we attached the peroxisomal targeting signal 1 via Golden Gate cloning.

Using promoters with different expression strength in order to find optimal measurement conditions is of high interest. There is a trade-off between a high signal-to-noise ratio and self-induced effects which are both dependent on expression levels (Schwarzländer M. et al., 2016) . This is not generally the case for each sensor. For example, pHlourin2 has only a minor influence on the existing pH because of the buffer effect of proteins.
Validation of the peroxisomal localization can be achieved via fluorescence overlap of the sensor and a peroxisomal marker, in our case pex13-mRuby (import mechanism).

It can also be achieved by transforming the sensors attached to the PTS1 sequence into a $\Delta$Pex5 yeast strain. The sensor is expected to show no specific localisation due to the missing import sequence. We calibrated the sensors in living yeast cells and physiological ranges so that we can not only perform relative but also quantitative measurements. We aim to confirm our hypothesis of a more oxidized redox state of roGFP2 in peroxisomes with violacein pathway activity and want to measure differences in pH within yeast strains with peroxisomal membrane anchored pex3-bacteriorhodopsin protein (membrane proteins) .

Once expression and localisation of the sensor is verified by microscopy, measurements with a plate reader or a fluorometer are acceptable. This allows a high number of replicates to be measured accurately in reasonable time. Microscopy was performed with a filter based Nikon Eclipse TI fluorescence microscope at 100-fold magnification and plate reader measurements were performed with a Tecan infinite plate reader (Mahon M. J. et al., 2011) . For pHLuorin2 emission intensity is measured at 535 nm upon excitation at 405 nm and 485 nm. Same settings were used for roGFP2 (Schwarzländer M. et al., 2016) . Evaluation of the reported signals was done via analysis of the intensity ratio of the corresponding excitation wavelength.

The initial step is to find a reliable source to prove the abundance of our precursor farnesyl pyrophosphate (FPP) in yeast peroxisomes. So far FPP has not been proven to exist inside yeast peroxisomes yet. However it is predicted to be present, as it has been detected in mammalian and plant peroxisomes Olivier et al. (2000).

FPP is converted into valencene by a valencene synthase (ValS). We chose the version derived from Callitropsis nootkatensis because of its comparably high efficiency in microorganisms. It achieves greater yields in yeast than the citrus valencene synthase. Furthermore, the product specificity is relatively high, while production of byproducts is low Beekwilder et al. (2014). The valencene synthase was also chosen because of its robustness towards pH and temperature changes Beekwilder et al. (2014). Our modelling approach revealed that for optimal yields an overexpression of valencene synthase is necessary because of its slow conversion rate (Model). This is why we chose the strongest promoter supplied by the yeast toolbox (Dueber Toolbox) for this attempt.

The intermediate valencene is then converted into nootkatol by a P450 monooxygenase as well as into small amounts of our desired product nootkatone. The P450 monooxygenase we chose for this project was taken from the bacterium Bacillus megaterium. In this case it is not only a simple P450 monooxygenase, but an entire P450 system, consisting of a soluble P450 fused to a cytochrome P450 reductase (CPR) enzyme, making an additional reductase obsolete De Mot et al. (2002). Unlike eukaryotic P450s, which are mostly membrane bound, this prokaryotic BM3-P450 is located in the cytosol, thereby facilitating an easier transport into the peroxisome. Membrane integration of proteins is a more difficult task to achieve than import of cytosolic proteins Girvan et al. (2006). BM3 normally catalyzes the hydroxylation of long chain fatty acids Narhi et al. (1986), which in our case could inhibit the conversion of valencene. For that reason we used a mutated version of BM3, which is called AIPLF. This variant is an enhanced version of the BM3 AIP version, which has a ten times better substrate oxidation rate for valencene than the wildtype BM3 and produces less byproduct when valencene concentration is saturated. Additionally to previous named benefits the AIPLF variant with 5 point mutations in the active side has a significantly lower binding affinity towards long chain fatty acids and therefore increase the transposition rate of valencene. Schulz et al. (2015), Lehmann (2016).

The alcohol dehydrogenase from Pichia pastoris subsequently converts nootkatol into nootkatone by oxidation. It uses NAD⁺ as a cofactor which is reduced in the reaction. The regeneration of this cofactor is facilitated by the BM3, which oxidizes NADH Schulz et al. (2015).

The first milestone to achieve our goal is the separate integration of each of our three enzymes ValS, BM3 and ADH into level 1 vectors (Dueber Toolbox) in yeast and to verify their expression by Western Blot analysis. Therefore, a 3xFlag/6xHis-tag was added to the N-terminus of each of the proteins. It enables us to use an anti-His antibody followed by an anti-mouse-antibody to make protein abundances visible. Subsequently, the two enzymes ADH and ValS were combined in a level 2 cassette plasmid. BM3 is designed as a level 1 plasmid and will be co-expressed with the level 2 plasmid to achieve a nootkatone production. The expression of the enzymes in the cytoplasm is again verified by Western Blot analysis. The further approach aims to provide the enzymes with C-terminal peroxisomal targeting signals type one, which finally converts the nootkatone pathway into our artificial compartment.

Since the production of nootkatone does not lead to a change of colour we need to apply different methods for verifying substrates. Once we succeed with the qualitative validation via Western Blot analysis, we can verify the presence of nootkatone by using high performance liquid chromatography and mass spectrometry, respectively.

Model influence on Nootkatone expression

We modeled the nootkatone biosynthesis pathway using ordinary differential equations in order to optimize nootkatone production. We found two hard and an easy problem, all of which we could find a solution for. The easy problem is optimization of the enzyme concentrations of the biosynthesis pathway. According to our model of the Nootkatone pathway we found that overexpression of Valencene Synthase is necessary to maximize the Nootkatone yield, while both alcohol dehydrogenase and p450-BM3 have only minor effects on the yield.

One of the hard problems, as shown in our penalty model is the toxicity of nootkatone and nootkatol. Since the toxicity most likely stems from both nootkatone and nootkatol clogging up the cell wall we present our peroxisome as a solution for this problem. When comparing the cytosolic model to our peroxisomal model we found that, if our assumption that neither Nootkatone nor Nootkatol are able to pass the peroxisomal membrane holds up, we can greatly increase Nootkatone production.

A further problem is the influx of the pathway precursor farnesyl pyrophosphate (FPP). We used OptKnock analysis to design yeast strains with optimized FPP production. With this analysis we got hints that growing the yeast cells on a fatty acid medium might be a simple alternative to knocking out the desired genes.

Based on described of advantages of violacein this pathway was chosen. Violacein is naturally produced in numerous bacterial strains, most popular in the gram-negative Chromobacterium violaceum . It is related to biofilm production and shows typical activities of a secondary metabolite (Seong Yeol Choi et al., 2015).

Relocating the pathway into the peroxisome enables proximity of the enzymes and substrates. Furthermore the yeast cell is protected from the toxic substance hydrogen peroxide. Yeast peroxisomes have no problem with this as their main function is the beta-oxidation of fatty acids and the detoxification of the thereby produced H₂O₂ (Erdmann R. et al., 2007) . Because VioC and VioD are FAD-dependent, it is additionally an evidence for FAD availability inside of the peroxisome, if the synthesis of Violacein works. Otherwise the two enzymes would not be able to catalyze the reaction.
The genes for VioA, VioB and VioE were amplificated via PCR with Golden Gate compatible overhangs from the biobrick VioABCE (Part: BBa_K274004).

By Golden Gate cloning the peroxisomal targeting sequence (PTS1) was attached to the C-terminus of every pathway protein. Combined with the other necessary parts of the toolbox they represent the level 1 plasmids.

The PTS-tag marks the proteins for the import into peroxisomes. This should first of all point out the functionality of the yeast’s natural import mechanism and also be the basis for demonstrating our own modeled PTS*, proving our designed orthogonal import mechanism. Furthermore we also aim to optimize the working conditions for the enzymes inside of the reaction room - the peroxisomes. For example to vary the pH with new membrane proteins such as bacteriorhodopsin. To secure this change, we can also check the current conditions by our designed sensors.

There are several methods to verify the pathway’s enzymes. First of all, violacein and several intermediates (prodeoxyviolacein, deoxyviolacein, proviolacein) are colorful and the production in yeast can be visualized easily. Furthermore we added a His-/Flag-tag to the N-terminus of every protein (see geneious plasmid cards) to confirm their expression via SDS page and western blot. After verifying the presence of the enzymes the next step is to test their functionality. Before performing in vivo experiments in yeast an in vitro assay was implemented. For this the three enzyme pathway leading to PDV was reconstructed, testing VioA, VioB and VioE. To enable the best conditions for the enzymes, the pathway was studied intensively and all needed cofactors were calculated and added to the in vitro reaction (see protocol prodeoxyviolacein in vitro assay). This included FAD, MgCl₂, catalase for decomposition of hydrogen peroxide, and the substrate L-tryptophan. The in vitro reaction was followed by qualitative analysis via HPLC and mass spectrometry.

Introduction

Our optogenetic toolbox enhancements can be divided into three subgroups: controllable protein import via Pex5, controllable protein import via Pex7 and controllable gene expression. Each of these sub-projects are of different design which will be illustrated in the following.

Pex5 import with LOV2

LOV2 is an optogenetic protein derived from Avena sativa’s Phototropin 1. In its dark state the J$_{\alpha}$-helix located at the C-terminus is bound to the core of the protein. Upon irradiation with blue light (~460 nm), a covalent bond between a cysteine residue on the LOV2 protein and a flavin mononucleotide chromophore causes the J$_{\alpha}$-helix to unfold, which in turn exposes the C-terminus (Spiltoir et al, 2016) . This property is very useful, as short amino acid sequences can be attached to this end of the LOV-protein, for example a PTS1. The idea for this project was to attach PTS1 to the C-terminus and the protein of interest to the N-terminus. Upon irradiation with blue light, the fusion protein would be imported into our compartment. We used a mutated version of LOV2 whose C-terminus has an increased dark-state binding affinity. This is caused by the substitutions and T406A and T407A. These mutations greatly reduce the possibility of the J$_{\alpha}$-helix being exposed in the dark state. Our PTS1 sequence consists of the amino acids LQSKL. As a proof of concept for this construct we fused sfGFP to its N-terminus:

This was done in order to visualize our experiment’s results during microscopy: upon successful import of the fusion protein, one would observe GFP fluorescence localized to our compartment. Otherwise, the whole cell would be illuminated.

Pex7 import

The idea behind this project is to initially block the protein of interest’s PTS2 with a fluorescent protein which can be removed by an optogenetically activated TEV-protease. For this project we use the protein Phytochrome-B from Arabidopsis thaliana and its interaction partner PIF6. These two proteins, also derived from Arabidopsis thaliana, bind together upon irradiation by red light (660 nm) and separate upon irradiation with far-red light (780 nm). We used this property to activate a split version of a TEV-protease whose split halves were each fused to one of the two optogenetic proteins.

The TEV-protease was obtained from the Biobrick BBa_K1319004. This variant contains the anti self-cleavage mutation S219V. Using overhang-PCR we created a split version of the Biobrick protease based on work done by Wehr et al, 2016 and the iGEM team Munich 2013. The split was made between amino acid 118 and 119.
Our construct for attaching proteins N-and C-terminally is highly variable: it consists only of a TEV-cleavage site, the PTS2 sequence and a short linker and was designed as a 3b-part for the yeast-toolbox. This means that we can attach any protein to its N- or C-terminus we desire. We planned on attaching different fluorescent proteins to each sides of the 3b-part in a level 1 ligation. For our experiment we planned on using the pairs mTurquoise-mVenus and GFP-mRuby.

The two other constructs were planned as follows: Phytochrome B was fused to the C-terminal TEV-half, PIF6 was fused to the N-terminal TEV-half. The PhyB-TEV2 part and the TEV1-PIF6 part were supposed to be inserted into a shared plasmid via a level 2 Golden Gate ligation. Finally, the level 2 plasmid and the remaining level 1 construct were to be co transformed into S. cerevisiae. Our experiment consisted of illuminating one sample with red light (660 nm) while keeping another sample in the dark. Fluorescence microscopy would then be used to check whether the import was successful [3]. If cleavage and subsequent protein import was successful, fluorescence of one protein would be localized to the compartment while that of the other would be observed throughout the cell.

Optogenetically controlled gene expression

This project is based on work done by Weber et al, 2016 . Using the interaction between Phytochrome B and PIF6 they designed an optogenetic switch for enabling and disabling transcription of a chosen gene. It is based on the tetracycline operon and the transcription factor VP16. The tetO operator is located upstream of a minimal promoter which in turn is located upstream of the gene of interest. The tetR repressor binds to the tetO sequence. Fused to it is PIF6. Phytochrome B is fused to the transcription factor VP16. Upon illumination with red light, Phytochrome binds to the tetR-PIF6 complex. VP16 is now located in close proximity to the minimal promoter, which enables the RNA-polymerase-2 to start transcription of the gene of interest.
We designed a promoter part for the Dueber toolbox which consists of tetO and the minimal promoter region. This can be used as a promoter in a level 1 ligation for any desired gene of interest (a GFP-tagged Pex11 in our example).

Transformation into S. cerevisiae is accompanied by co-transformation of a level 2 plasmid containing both the tetR-PIF6 and PhyB-VP16 constructs:

Verification methods depend on which gene is expressed. See size and number and secretion for details.

For the future one could think of much more radical strategies for peroxisomal engineering with a final goal of a “minimal peroxisome” by redirecting metabolic pathways through changing the protein-localization-signal in the yeast genome. Additionally, endogenous metabolic pathways could be redirected to our novel artificial compartment for establishing a customized metabolism specifically tailored for the user's application.

Introduction

In order to achieve a fully controllable artificial compartment, the first step was to design a completely orthogonal import system. Next was the knockout of endogenous import systems. However, a few proteins are imported neither by the Pex5 nor the Pex7 import machinery. Therefore, specific genome engineering designs, such as knockouts, deleting or redirecting the protein localization could be utilized for the ultimate goal of creating a synthetic organelle.

Additionally, knockouts or genome integrations enable customization of the peroxisomal properties, such as membrane permeability, size/number, decoupling of peroxisomes from cytoskeleton and the peroxisomal metabolism.

All these strategies allow a rational design of an artificial compartment, which is fully engineerable regarding the proteome, metabolome and the entire peroxisomal environment.

Design of yeast multi -knockout strains

The Crispr Cas9 System

The demands on yeast engineering have significantly increased with the design of more complex systems and extensive metabolic pathways. Genetic techniques that have historically relied on marker recycling are unable to keep up with the ambitions of synthetic biologists. In recent years the Crispr Cas9 system has been used for several strain-engineering purposes, including:

• Markerless integration of multiple genetic cassettes into selected genomic loci
• Multiplexed and iterative gene knockouts without the need to recycle a marker
• Precise genome editing – nucleotide substitutions, etc.

We utilized the Cas9 system as a tool for peroxisomal engineering and have adopted the existing toolbox from (Lee et al. 2015) and the complete cloning system which also provides the possibilities for genome integration and gene editing by Cas9. For this, two oligonucleotides have to be designed for targeting the Cas protein to the gene of interest.

Several gRNA vectors can subsequently be assembled into one vector with a Cas9 expression cassette and then be transformed into yeast. The expression of Cas9 together with gene specific gRNA´s leads to double strand breakage followed by non-homologous end joining repair or homologous recombination, in case of added repair DNA (figure 3).

The combination of the Cas9 system with DNA repair sequences enables not only knockouts of peroxisomal proteins, but also allows redirecting protein localization by changing protein targeting signals or integration of linear DNA into yeast chromosomes. Genome engineering facilitates yeast strain development for customized peroxisomes.

The peroxisomal proteome of yeast (saccharomyces cerevisiae)

The peroxisomal proteome is studied extensively for saccharomyces cerevisiae and contains exactly 67 proteins (Kohlwein et al. 2013). The function is characterized for the most of those proteins and it is known, that yeast peroxisomes are expendable under optimal growth conditions. Nevertheless, some knockouts are lethal under oleate or stress conditions.

Knockout designs in our project

Pex 9

Pex9 is a recently discovered import receptor for PTS1 proteins, which is induced by oleate and is an import receptor for both malate synthase isoenzymes Mls1p and Mls2p. In order to get a completely empty reaction room, a Pex9 knockout was designed to prevent unintended protein import.

Pex 31 & Pex 32

It has been shown that knockouts of Pex31 and Pex32 leads to an increased Peroxisomal size, but additionally the membrane permeability was affected (Zhou et al. 2016). This effect can be used as a tool for engineering membrane permeability by knocking out or overexpress both genes. A knockout would lead to an increased permeability and one could think of an opposite effect in case of overexpression, but this has not been shown yet.

INP1

The INP1 knockout was designed after the skype call of Prof Dueber, who recommended us, decoupling of peroxisomes from cytoskeleton in order to improve the secretion efficiency. INP1 is responsible for the tethering of peroxisome, which would inhibit the secretion of peroxisomes.

POT1

The only protein, which is imported by the Pex7 import machinery in saccharomyces is the 3-ketoacyl-CoA thiolase (POT1). A knockout of POT1 would enable utilizing the Pex7 import for proteins of interest, which cannot be tagged at the C-terminus with pts1, without having unintended import of other enzymes.

Genomic integration of our novel Pex5 import receptor

After testing our new Pex5 import systems, which is completely orthogonal to the natural import, the next step would be to replace the endogenous system with our artificial import system. Therefore, an integration plasmid was designed with help of the previously described yeast toolbox, containing HO locus homologies and a hygromycin resistance (Figure 4). Afterwards the plasmid was transformed into the yeast strain which was created by our collaboration partner Aachen (double knockout strain Pex5 & Pex7).

The resulting yeast strain allows full control over the peroxisomal matrix proteome, by replacing the whole protein import machinery, which is the first step for creating our artificial compartment.

Outlook

Besides the genome engineering approaches, which were performed in our project one could think of more radical strategies for peroxisomal engineering. A final goal could be a “minimal peroxisome”, which contains only the proteins that are required for the biogenesis of the peroxisome and import of proteins and metabolites. On the one hand peroxisomal pathways could be redirected to cytosol or other organelles and one the other hand endogenous metabolic pathways could be redirected to our novel artificial compartment by changing the protein localization signal in the yeast genome with help of the Cas9 system. All these strategies would allow tremendous improvements for metabolic engineering applications by creating an artificial compartment, which can be rational designed and customized for specific metabolic pathways.