Results

Compartmentalisation

Wiffleball

The synthetic BMC compartment originates from the HO-shell, which is built from five different proteins: H, T_1, T_2, T_3, P. The synthetic full wiffleball is lacking the P-protein, which opens the shell and enables the molecules to diffuse in and out. The minimal wiffleball is the simplest form of shell, consisting only of H and T₁-proteins (Figure 1.1).       
Most publications focused on in vitro experiments with bacterial microcompartments. We, iGEM Freiburg 2022, want to establish the wiffleball systems in vivo and generate an easy-to-use platform for utilizing compartments with click chemistry protein loading and applications in bioproduction.

Figure 1.1: Protein structure and schematic architecture of the synthetic BMC architectures compared to the native HO-shell, PDB structure 6MZX; schematics drawn with powerpoint; adapted from Kirst et al. 2022 [1.1]

Table 1.1: Acronyms of HO-shell plasmid with their genes and their shell architecture

Targeting Proteins into the BMCs

Aim: Testing whether the full and minimal wiffleballs form in E. coli and proteins are recruited inside of them.

Experimental design: We used fluorescent microscopy to monitor the localisation of fluorescent proteins (FPs) fused to the Spy and Snoop (Snp)-Catchers when co-expressed with the proteins required to form the full and the minimal wiffleballs, with the T1 protein being fused to the Spy and SnpTags. Specifically, the SpyCatcher is fused to mVenus2 and the SnpCatcher to mTurquoise2 (Figure 1.2). We expected that the uptake of the FPs into the compartments should alter the fluorescence distribution in the cells: instead of cytoplasmic fluorescent signal, we would expect the formation of fluorescent foci.

                     
To confirm the formation of the peptide bond linking the FP to the T1 protein, we performed Western Blotting. If the peptide bond is formed, the bands corresponding to the FP and the T1 protein are expected to run higher in the gel because of the higher molecular weight. An anti-His antibody was used against the His-tag of the T1 protein and an antibody against the beta-subunit of the RNA Polymerase was used a loading control.             

We used the same induced BL21 cells for the microscopy and the western blot. After an induction test, we decided to use 100µM IPTG for wiffleball induction and 50ng/ml doxycycline for the fluorescent protein expression (mVenus2 or mTurquoise2). The bacteria were grown in overnight cultures shaken at 30°C, 200rpm, induced at OD₆₀₀= 0.6-0.7. Samples were harvested after 24h of incubation at 200 rpm at 18°C. Conditions were based on literature research and the initial paper from Kirst. et al. 2022 [1.1]. In general, culture, induction and expression conditions are highly sensitive for microcompartments since they tend to form insoluble aggregates. We also fractioned the cell lysate into supernatant and pellet to observe the solubility of the wiffleballs.  
All experiments were repeated a total of three times, except for the Snp-catching experiments. These were just performed two times. 

Figure 1.2: Principle of catching proteins via Spy/SnpCatcher by T1 and their incorporation into the BMCs, illustrated as an example of incorporating mVenus2 in the minimal wiffleball.

Results:

All our raw data can be found on the labbook page. In this section, we only show the most salient results.

We expressed tagged mVenus2 either alone or co-expressed with the wiffleballs, which were either tagged or not tagged with the Spy/SnpTag at the T1 protein. The results gave us important insights: samples with untagged wiffleballs show a homogeneous distribution of the fluorescence within the cells (Figure 1.3, B). When the T1 protein had the tags, we observed the appearance of fluorescent foci in some cells (Figure 1.3 B, red arrows). The foci in the full wiffleball were always found at one or both poles of the cells. The minimal wiffleball had fewer and smaller foci, which were also localised in other areas of the bacteria. The foci were easier detectable in less fluorescent cells; therefore, more foci could be hidden in cells with brighter fluorescence. BL21(DE3) showed in some induced cells an elongated phenotype (Figure 1.3 B). 

Figure 1.3: Fluorescent microscopy of T1 catching the mVenus2, when the minimal or full wiffleball construct is expressed; A: Controls for the induction; B: T1 with and without the Spy/Snp tags; scale bar, 5µm.

We could detect the T1 protein (29 kDa w/o tags or 37 kDa with tags) on the Western Blot (Figure 1.4 and 1.5). The absence of the Spy/Snp-tag resulted in one single band. When the tag was present, a second band corresponding to circa 80 kDa was observable (Figure 1.4). The molecular weight of mVenus2-SpyCatcher is the same as that of the T protein with the tags (37 kDa). We observed the formation of the peptide bond both, with the full and the minimal wiffleball (Figure 1.4).

Always a small fraction of the unbound T1 was found in the insoluble fraction of the cells, but not when fused to mVenus2, which excludes the possibility that the nature of foci results from insoluble T1-mVenus2-aggregates. Most insoluble T1 was found in the pellet of cells expressing the minimal wiffleball, when mVenus2 was also expressed.

Figure 1.4: Western Blot showing the formation of the peptide bond between the T1 protein and mVenus2 in cells expressing the minimal wiffleball with and w/o tags. The detection of the T1 protein was performed with an anti-his antibody.

Figure 1.5: Western Blot showing the formation of the peptide bond between the T1 protein and mVenus2 in cells expressing the full wiffleball with and w/o tags. The detection of the T1 protein was performed with an anti-his antibody.

When testing SnpTag/Catcher system, we found the same kind of foci, visible under the fluorescence microscope (Figure 1.6). Therefore, we assume that we can already see the encapsulation in the BMCs and the formation of the compartment. To confirm these results, further experimental evaluation will be needed. The covalent catching of mTurquoise2 by the T1 is proven by western blots (Figure 1.7). Surprisingly, the full wiffleball showed insoluble T1 and mTurquoise2-bound T1, although only for some induction conditions. The expression of the minimal and full wiffleball does not seem to influence the formation of insoluble aggerates of the proteins and seems to happen in a more random fashion.  

Figure 1.6: Representative fluorescence microscopy images of BL21 cells transformed with the indicated constructs. A: Controls; B: T1 with and without the Spy/Snoop tags. Scale bar, 5 µm.

Figure 1.7: Western Blot showing the formation of the peptide bond between the T1 protein and mTurquoise2 in cells expressing the full and the minimal wiffleball with and w/o tags. The detection of the T1 protein was performed with an anti-his antibody.

Conclusion:

Taken together, we believe that the FPs are being encapsulated into the wiffleballs, which form in the E. coli cells we used. A more direct proof of the formation of the compartments and the localisation of the FPs into them is, however, needed.     

Strain comparison for testing the orthogonality of wiffleball expression

Aim: Testing different E. coli strains for their ability to support the formation of the compartments monitoring their respective burden.

Experimental design: We tested the full and minimal wiffleball, besides in BL21(DE3), in MG1655, the standard K-12 E. coli strain and in C321∆exp, a strain engineered for more efficient incorporation of non-canonical amino acids into proteins of interest. In this strain all the instances of the amber STOP codons have been replaced with other STOP codons and Release Factor 1 has been deleted. The experimental conditions stayed the same as in the previous experiment for the Spy/SnpTag/Catcher testing.

Results:

We first tested the BMCs with the SpyTag/Catcher system catching mVenus2 in MG1655. Under the fluorescent microscope we were able to observe the same foci like in BL21(DE3), but in lower quantities (Figure 1.8).
Unfortunately, the microscopy results could not be reproduced with the western blot, as there was no detection of T1 in the full wiffleball construct. Not all cells contained the foci, therefore it could be that not all cells express the compartments. In microscopy, we examined the individual cells, while western blot is a population-based assay.
In comparison, all the T1 of the minimal wiffleball was fused to the mVenus2 and showed only the combined band of the mVenus2-T1 complex at 80 kDa (Figure 1.9). Furthermore, the full wiffleball needs to express two more proteins, T2 and T3, and therefore has more metabolic burden compared to the minimal wiffleball; also, more of detectable T1 should be produced in the minimal wiffleball. 

Figure 1.8: SpyTag/Catcher system in MG1655, when the minimal or full BMC construct is expressed; A: Controls for the induction; B: T1 with and without the Spy/Snp tags; scale bar, 5 µm.

Figure 1.9: Western Blot of the BMC minimal wiffleball (pT1spysnp) + mVenus2 in MG1655.

Next, we repeated the same experiments under the same conditions in C321∆exp. We did not see any signal on the Western Blot corresponding to the T1 protein when using the anti-His antibody in cells expressing both, the full and the minimal wiffleball. The loading control was still observable (data found in the supplementary). Cells expressing the full wiffleball showed one fluorescent focus per 100 cells when observed under the microscope (data found in the supplementary).

Conclusion:

The expression of the wiffleball proteins in MG1655 and in C321∆exp is not easily achievable. A problem could be the protein synthesis at 18°C. On the other hand, the use of higher temperature is known to result in unproper compartment formation. The expression and correct formation of BMCs is an often-claimed challenge in publications.     
As the next step we try to enhance the BMC expression by testing different inducer concentration in the used strains.

Induction experiments for improvement of wiffleball expression

Aim: Enhancing the protein expression level to an optimal balance between high enough expression and low metabolic burden. 

Experimental design: We tested different inducer concentration and validated the expression with fluorescence microscopy for the foci formation, western blot for the protein expression level and growth assays for the metabolic burden. The experimental conditions for the microscopy and the western blotting stayed the same. For the growth assay, the bacteria were induced at OD₆₀₀= 0.1 in a 96-well plate with all the different inducer concentrations and incubated at 30°C, shaken in a plate reader for 20h. These conditions were changed due to the lack of a cooling function to 18°C of the device, limited time of the availability of the plate reader and a restricted maximum growth to OD₆₀₀= 1.4 due to the 200µl well volume. For comparison between the strains, both strains were measured on the same 96-well plate.

We tested the full and minimal wiffleball in BL21(DE3) and MG1655 with different concentration of IPTG reaching from 100µM (best induction condition from the first induction test) up to 1mM.

Results:

BL21(DE3)

In the microscopy, the number of foci increased with the amount of the inducer. Expressing the minimal wiffleball with a concentration of 400µM IPTG and higher, resulted in many small foci across the whole cell. This can be easier observed in less fluorescent cells, where the foci can be better distinguished from the fluorescent background (Figure 1.10). An exact quantification for this is not possible due to the bright, overlaying fluorescent signal. Again, the longer phenotype occurred in the minimal wiffleball. For the full wiffleball we quantified the bigger foci at the poles per number of bacteria. Compared to the minimal wiffleball, they can be still observed in presence of high fluorescent background (Figure 1.11). The mean percentage of foci per cell number was merely constant around 52%. An induction with 1mM effected in a drop of 8%.            

Figure 1.10: Fluorescent microscopy of different induction concentrations for the BMC expression with IPTG in BL21(DE3), mVenus2 was always induced with 50ng/ml Doxycycline; scale bar 5µm.

Figure 1.11: Quantification of the foci of the full wiffleball visible under the microscope with different induction concentrations of IPTG for the BMC expression in BL21(DE3); n=2

The western blots showed similar results like the microscopy data (Figure 1.12). For the induction with 100µM and 400µM IPTG we could detect an insoluble part of the T1 and T1-mVenus2 in the pellet of the full wiffleball. We quantified the signal for the bands at the expected molecular weight for T1 and T1-mVenus2 and normalised it to the signal of the loading control (RNA polymerase b subunit) and have been quantified by the area under peak method. The quantification showed that the best inducer concentration was 400µM for the full wiffleball and 700µM for the minimal wiffleball (Figure 1.13). The Western Blots were repeated three times, but not all membranes were able to be quantified due to overlapping bands or curved SDS-PAGE runs.

Figure 1.12: Western Blot used to quantify the expression levels of the T1 protein and the amount of T1-mVenus2 ligation product for the indicated IPTG concentrations in BL21(DE3) cells expressing the minimal and the full wiffleball. The detection of the T1 protein was performed with an anti-his antibody.

Figure 1.13: Quantification of the Western Blots shown in Figure 1.12.

For the growth assay in BL21(DE3) we could find different, but not constant trends. First, BL21(DE3) grew the best with empty plasmids, followed up by the minimal and the full wiffleball, which seemed to affect the growth the most (Figure 1.14). We repeated the growth assay experiment three times, but the results seemed to be influenced randomly. We could not determine a major trend in growth by the expression of our synthetic constructs. Overall, BL21(DE3) seems not be highly burdened by expressing the wiffleball.

Figure 1.14: Growth comparison between BL21(DE3) and MG1655 with induction of 100µM IPTG 25ng/ml Doxycycline, n=1

In MG1655, there were remarkably fewer foci in cells expressing both, the full and the minimal wiffleball (Figure 1.14). Quantification of the number of foci in cells expressing the full wiffleball showed a range from 18% to 30% of foci per number of cells, with an increase at 400µM IPTG to 26% (Figure 1.15).  We did not quantify the cells expressing the minimal wiffleball because of the too high background fluorescence.

Figure 1.15: Fluorescent microscopy of different induction concentrations for the BMC expression with IPTG in MG1655 mVenus2 was always induced with 50ng/ml Doxycycline; scale bar, 5µm

Figure 1.16: Quantification of the foci of the full wiffleball visible under the microscope with different induction concentrations of IPTG for the BMC expression under the microscope in MG1655; n=2.

The western blot of the same microscopy samples showed, as in the prior data, no bands of the T1 proteins.  After increasing the sensitivity, slight band of bound and unbound T1 could be detected in the full wiffleball construct (Figure 1.17). The protein is still expressed, which correlates with the microscopy data. It could be assumed, the detection range of the western blot of the first test of the SpyTag/Catcher in MG1655 was just too low. The foci seem to already be formed at low BMC protein concentrations. An insoluble fraction of the protein wasn’t found. The western blots were repeated 3 times.

The minimal wiffleball showed distinct bands compared to the full wiffleball. The protein expression seems to be overall higher compared to the full construct. Less diverse proteins results in less metabolic burden. Surprisingly, in Figure 18 the 700µM induction had the most bound T1-mVenus2 fraction. In the repetition of the Western Blot, the bound T1 appeared stronger than the unbound compound. These conditions should be treated carefully, because overexpressing the wiffleball can result in not proper forming of the compartment.

Figure 1.17: Western Blots of different induction concentrations for the BMC expression with IPTG in MG1655. The 100µM IPTG 0ng/ml DOX sample of the pellet and supernatant had been switched, when loading the SDS-gel.

Figure 1.18: Quantification of the Western Blots shown in Figure 1.16 only for cells expressing the minimal wiffleball.

The growth assay highlighted that the induction conditions of 100µM IPTG + 25ng/ml doxycycline did not affect the growth of MG1655 (Figure 1.19). As we expected, the full wiffleball had more negative impact on the growth than the minimal wiffleball, which consists of less diverse proteins.

Figure 1.19: MG1655 growth in plate reader at 30°C with different induction conditions. In legend: Wiffleball (IPTG (µM)/doxycylcine (ng/ml)).

Comparing the growth of both strains in the same plate reader, MG1655 is growing faster and does not die after 10h like BL21(DE3) (Figure 1.20). However, the protein expression of MG1655 cannot be related to the comparable high expression level of BL21(DE3), which affects the metabolic burden, therefore resulting in less growth.

For more data of the growth check our supplementary data.

Figure 1.20: Growth comparison in plate reader at 30°C between BL21(DE3) and MG1655 with induction of 100µM IPTG 25ng/ml doxycycline.

Conclusion:

Overall, the metabolic burden of the cell influenced than less then estimated. With an inducer concentration of 100µM IPTG, the cells are hardly affected in growth. The minimal wiffleball shows in general better expression and foci forming at higher inducer concentration and should be therefore induced with higher IPTG concentrations than the full wiffleball.

Late protein expression and degradation

Aim: Does MG1655 not express the compartment at all, or could it be possible that it gets degraded before we can even detect it? Does this strain just need more time to produce the proteins instead?

Experimental design: The already used induction conditions for the wiffleball induction at OD₆₀₀= 0.6-0.7 with incubation at 18°C, 200rpm, stayed the same. We took samples from the induced cultures from different time points, after 19h, 24h and 48h from MG1655. We examined the foci formation and protein expression by fluorescent microscopy and western blotting as in the wiffleball experiments before.

Furthermore, to investigate whether harvesting the samples earlier than the prior 24h induction has a remarkable effect on our protein expression and BMC formation, the 19h and 24h time points of induced BL21(DE3) were tested by western blotting.

Results:

MG1655 showed a slight decrease in foci formation after 48h compared to the 24h incubation (Figure 1.21). The incubation between 19h and 24h barely had any influence, detectable by qualitative fluorescence microscopy observation.

After 24h of incubation, BL21(DE3) had a higher T1 expression than the 19h sample. 

Figure 1.21: Fluorescent microscopy after different incubation times for the wiffleball expression in MG1655 with 100µM IPTG for BMCs and with 50ng/ml Doxycycline for; scale bar, 5µm.

Conclusion:

Longer incubation times had nearly no effect on the overall protein expression of MG1655.
After the induction of BL21, the T1 expression increased from 19h to 24h, as the 24h induction has a higher yield of protein expression.

Is the wiffleball only expressed at the poles or in the whole cell?

Immunostaining for BMC distribution:

Aim: We still don’t know if the wiffleball is formed properly. We are certain that the fluorescent spots do not represent inclusion bodies but rather compartment structures, judging by the investigation of the insoluble cell fraction. To determine their distribution in the cell and the right formation of the compartments, electron microscopy would be the best option. However, due to the high effort and limited use of this advanced technique, we first want to detect the overall distribution and formation of the wiffleball. We chose detection via immunofluorescence microscopy, even if this method is less established in bacteria. 

Experimental design: For this experiment, BL21(DE3) bacteria were induced at OD₆₀₀ = 0.6-0.7 with 100µM and 400µM IPTG, additionally with and without 50ng/ml doxycycline for mVenus2 expression, and subsequently incubated at 18°C for 18h, shaking at 200 rpm (the same way as in the previous experiments). After fixation with paraformaldehyde, the cells were treated with 70% ethanol to reduce the cell shrinking caused by the fixation. Next, the cells were immobilised in poly-L-Lysine coated chamber slides and treated with lysozyme and DNAse, for optimal diffusion of the antibodies through the cells. For the staining, our previously applied anti-His-antibody, which binds the T1, was used. The secondary anti-mouse antibody was linked to fluorescent stain DyeLight 650 (Invitrogen sponsoring, Goat anti-Mouse IgG (H+L) Secondary Antibody, DyeLight 650), which fluoresces in the far-red spectrum and does not interfere with the mVenus2 emission.                        
Detection took place with a Nikon Ti-E microscope with N-SIM module for high resolution of the internal structures.  Due to the advanced SIM technique, Kai Stobe, from the AG Römer, helped us using the microscope.

Results:         

We could prove that the full wiffleball is still forming foci at the poles in absence of mVenus2 (Figure 1.22 A, Full wiffleball +IPTG). Additionally, more round structures were highlighted by the anti-His DyeLight650. A fluorescent background in the bacteria was also observable, which could be interpreted as unbound T1. In the minimal wiffleball these round structures could be observed as well, mostly distributed over the whole cytoplasm. The specificity of the anti-His antibody was shown in the negative control, when only mVenus2 was expressed. No fluorescent signal could be detected in the far-red channel. (Figure 1.22 A +DOX)           
When expressing the full or the minimal wiffleball in presence of mVenus2, the foci at the poles of the anti-his DyeLight 650 mostly correlated with the mVenus2 foci (Figure 1.22 B). A possible explanation for only mVenus2 foci without the anti-His-Dylight 650 signal could be for example a hindered diffusion of the antibody through the cell, caused by lacked DNA degradation from the DNAse. Even more foci of DyeLigth650 than foci of mVenus2 were detectable. Due to the induction of the wiffleball and mVenus2 at the same time, some wiffleballs might be formed without the incorporation of mVenus2.  

Figure 1.22: Immunofluorescence staining with anti-His-antimouse-DyeLight650 conjugate, full wiffleball induced by 100µM IPTG, minimal wiffleball induced by 400µM lPTG, mVenus2 induced by 50ng/mFl doxycycline: A: controls and wiffleball expression in absence of mVenus2; B: anti-His staining in presence of mVenus2,; scale bar, 5µm.

Conclusion:

In comparison to the previously estimated fluorescence microscopy experiments, the immunostaining showed that T1 is spread all over the cell and a lot more structures that seemed to be assembled wiffleballs could be observed. Ultra-resolution microscopy has determined the roundish structures caused by the wiffleballs, with two distinct types depending on the size. The bigger foci often polarised in the full wiffleball, and the smaller, more distributed structures, in the minimal wiffleball.

Visualizing the wiffleball formation at highest scale:

Aim: To push the structural resolution to the next level, electron microscopy was used to gain insights into the wiffleball structure and distribution in the cells in the highest resolution possible. 

Experimental design: We induced BL21(DE3) with four conditions: empty backbone as control, full wiffleball w/o any incorporated cargo (100µM IPTG), full wiffleball with mVenus2 (100µM IPTG + 50ng/ml doxycycline) and the full wiffleball with our incorporated enzymes (150µM IPTG; the wiffleball and the indigo enzymes were induced by IPTG). After 24h of incubation at 18°C, 200rpm, we delivered the samples to Marta Rodriguez, the head of the TEM facility of the university of Freiburg. With her help, we could observe the four conditions of the full wiffleball under the TEM. The samples were prepared by negative staining with osmium tetraoxide and uranylacetate.

 Results:

After a thoroughly inspection of all the pictures taken by the TEM, we noticed that in all the samples where the microcompartment has been induced, darker regions at the cell poles (or sometimes in the middle of the cell) are forming (Figure 1.23, light blue arrows). This observation correlates to the dots we detected under the fluorescence microscope. The dark regions are probably too big to directly be microcompartments, however we suppose they are colocalisations of microcompartments, which would mean that the wiffleballs gather at the cell poles. To be able to certainly affirm this hypothesis, we would need an immunogold labeling to mark the exact position of the constructs. Previous research already showed the localisation of bacterial microcompartments in darker spots at the cell poles, even with the immunogold labeling technique [1.2], which represents a great hint for the veracity of our work.

By analyzing the poles and the darker spots more closely, we discovered some other, smaller, interesting structures (Figure 1.23 red arrows, Figure 1.24). In the bacteria containing an empty backbone we saw some circular texture that looked to us like air bubbles, maybe generated due to the preparation of the sample. In the bacteria with the expressed wiffleball, we found some circular structures that might indicate the formation of a microcompartment. We noticed something that looks like a multi-layered element, containing what seems to be particles on its inside. A similar observation was already described by other research groups in the past [1.2][1.3] which makes everything even more intriguing.    
A further type of texture was encountered in the bacteria expressing the wiffleball, especially where the enzyme production was induced as well. Here we found even smaller and darker spots, usually on the inside of an already darker region. Once again, it would be interesting to investigate through immunogold labeling, whether these spots are actually microcompartments or other mysterious compounds.

Figure 1.23: Transmission electron microscopy. Darker regions at the cell poles can be observed where the wiffleball is induced.

Figure 1.24: Transmission electron microscopy. closer look at the cell structure showing possible formations of the full wiffleball; in the empty backbone, a structure is shown which can be differentiated from the others, but was highlighted for comparison.

Conclusion:

The results look promising, but for a definite proof the immunogold labeling method should be applied. We still have upcoming experiments with the electron microscopy facilities to investigate the wiffleballs even further.

Minimal wiffleball affecting the cell morphology

Along the course of our experiments with BL21(DE3), we noticed that in the samples where the minimal construct of the wiffleball was induced, the cells were very often elongated, sometimes even “spaghetti-like”, reaching the length of over 100µm, as you can see in Figure 1.25. This leads to the conclusion that the expression of this microcompartment probably hinders the cell division, letting the bacteria grow more than usual. It would be interesting to see, through a DNA staining/screening, whether the DNA is evenly distributed along the whole cell, at only one or two poles of the cell, or aggregated in many parts of the organism, maybe indicating the attempt of the bacterium to divide.

Figure 1.25: BL21(DE3) alterated phenotype by expressing the minimal wiffleball. Microcompartment constructs induced with 400µM IPTG, mVenus2 with 50ng/ml doxycycline. Scale bar: 20µm.

Improving the interaction between XiaI and TnA and the T1 protein:

Aim: To adjust and better control the wiffleball and indigo enzyme interaction, we decided to switch the promotor of all wiffleball constructs from the IPTG-induction to the Tet-induction system from the mVenus2 promotor. In this way, the concentration of the pathway and compartment protein can be optimised, next to the temporal expression, to ensure that more enzymes are caught by T1, and less empty compartments are present.

Experimental design: We switched the promotor by Gibson assembly, where we amplified the wiffleball backbone of every construct by itself and used base pair extension on the Tet-operator and Tet-promotor sites for an 15bp overlap to the backbone. The success of the cloning was proved by colony PCR and a followed sequencing of the Tet-insert.                     
For the induction test with doxycycline, 10ng/ml, 20ng/ml, 50ng/ml, 75ng/ml, 100ng/ml were tested in BL21(DE3) and the protein expression validated per Western Blot. Induction conditions remained at OD₆₀₀= 0.6-0.7 with an incubation time of 24h, 200rpm, at 18°C.      

Results:

The sequencing result confirmed the correct cloning. After the induction of transformed BL21(DE3), the cells showed a reduction in growth. With the brightfield microscopy, it became clear that the cells were unhealthy or dying when the plasmid was induced and not affected by the presence of doxycycline itself, as seen in our control. The western blot showed a slight protein band in the full wiffleball and thin bands in the minimal wiffleball. No correlation between the inducer concentration and the band thickness was found.  
Another lab of our university also did a Tet-promotor switch out of the same mVenus2 plasmid and had the same results of dying bacteria. Due to the limited time, we could not find the reason for this issue. In Figure 26 the promotor switch from lacI to TetR is visualised.

Figure 1.26: Upper plasmind: plasmid with exchanged Tet promotor and operator. Bottom plasmid: previous expression under lacI regulated operator.

Encapsulins

Aim: our goal was to use the encapsulin protein EncA from M. xanthus, express it heterogeneously to form nanocompartments in E.coli . We want to use these encapsulins nanocompartments as chassis for enzymatic reactions to colocalise individual enzymes of a given pathway. Proteins are targeted to the Encapsulins due to a targeting peptide (TP), which is present C-terminal on the cargo protein. Therefore, we cloned two versions of the sfGFP, one without any targeting peptide and one with the targeting peptide C-terminally. In this way we can easily distinguish whether the sfGFP is incorporated into the Encapsulins with fluorescent microscopy.

Figure 2.1: The Encapsulin gene was cloned into pBAD33 and the sequence of sfGFP was cloned into a pET plasmid which are induced with arabinose and IPTG, respectively.

A plasmid containing a sfGFP with the same sequence as the biobrik # was used as a template PCR to extend the sequence with a linker and a targeting peptide (TP) on the C-terminus. This targeting peptide is responsible for the targeting of the native cargo into the Encapsulin of m. Xanthus. In this case we clone it onto sfGFP for direct detection of encapsulation with fluorescence microscopy. The sequence of the targeting peptide has been codon optimised for E. coli expression.

Sequence of GS-linker: GGCGGCGGCGGCAGC

Sequence of targeting peptide: CCGGAAAAACGTCTGACCGTGGGCAGCCTGCGTCGTTAA

The Encapsulin protein EncA, was first codon-optimised for E.coli from the sequence from the NCBI with the entry: MXAN_3556. We ordered the sequence optimised version from IDT containing a linker and a his6-tag on the C-terminus.

Sequence of the GS-linker and his6-tag: AGTGGCAGCGGCCACCACCACCACCACCAC

Experimental setup: BL21 were transformed with pBAD_encA and either pET_sfGFP or pET_sfGFP_TP. A larger culture was grown until OD: 0.6-0.8 and induced with arabinose for the Encapsulins and IPTG for the sfGFP. Then cells were incubating for 24 h at 18°C until samples were analyzed with fluorescent microscopy.

Figure 2.2: Fluorescence microscopy of Encapsulin expression. E. coli (BL21) was co-transformed with EncA and either A) sfGFP or A) sfGFP C-terminally fused to the targeting peptide for the Encapsulin. Arabinose and IPTG induces the expression of EncA and sfGFP respectively. Panels presents GFP fluorescence. sfGFP localises mostly throughout the cytoplasm in both conditions. Single dots in cells are formed when IPTG is only expressed due to leaky expression. Representive pictures of three replicates.

Fluorescent microscopy of the Encapsulins with target peptide showed for all conditions with 100 µM IPTG induction of sfGFP occasionally fluorescent “foci”. When sfGFP is only present due to leaky expression, a larger population of the cells expressing the sfGFP with targeting peptide have visible fluorescent dots. The results were surprising, as we expected the encapsulins to be formed when sfGFP was induced and not before. The absence of fluorescent dots when sfGFP_TP with targeting peptide we hypothesise is related to an overexpression of sfGFP_TP in comparison to encapsulins. The amount of sfGFP is likely too high and not all proteins gets targeted to the encapsulins and as a result distribute through the cytoplasm. In theory, every EncA protein could bind a cargo protein, however, the actual number of cargo proteins are found to be 24% [2.1].

Because some fluorescent dots were observed even in the absence of encapsulins for both conditions, we performed a control experiment with the two types sfGFP and no encapsulins.

Aim: The aim is to observe whether the observed fluorescent dots for the experiments with Encapsulins are due to leaky expression the Encapsulins.

Experimental setup: BL21 was transformed with pBAD33, same backbone as for the Encapsulins, and a plasmid containing either sfGFP or sfGFP with the encapsulation tag C-terminally.  The induction and incubation follow the same procedure as for the previous experiment.

Figure 2.3: Fluorescence microscopy of Encapsulin expression. E. coli (BL21) was co-transformed with pBAD33 and either A) sfGFP or A) sfGFP C-terminally fused to the targeting peptide for the Encapsulin. Arabinose and IPTG induces the expression of pBAD33 and sfGFP respectively. Panels presents GFP fluorescence. sfGFP localises mostly throughout the cytoplasm in both conditions. Single dots are formed occasionally. 

Figure 2.3 shows that expression of both the sfGFP alone and sfGFP fused C-terminally to the encapsulation tag mostly leads to localisation throughout the cytoplasm. However, for both strains, occasional dots with higher fluorescent intensity are formed. These fluorescent dots have similarities to those observed when expressed together with Encapsulins. As these dots are more commonly seen when sfGFP is expressed in the presence of the Encapsulins, this could be due to the generally higher protein production causing formation of inclusion bodies. With this experiment, however, we can also conclude that both the normal sfGFP as well as sfGFP fused to targeting peptide can form fluorescent dots, which may be difficult to distinguish from the encapsulated sfGFP. Furthermore, the sfGFP containing the targeting peptide show similar characteristics as the normal sfGFP.

As the interaction between the targeting peptide and the encapsulin protein is non-covalent, it can not be detected as a shift on a western-blot. However, we decided to blot the encapsulins with an antibody against the his6-tag.

Figure 2.4: Western blot of Encapsulin expression. E. coli (BL21) was co-transformed with EncA and  sfGFP C-terminally fused to the targeting peptide for the Encapsulin. Arabinose and IPTG induces the expression of pBAD33 and sfGFP respectively. Panels presents GFP fluorescence. EncA expression is at very low levels and is detected in supernatant as well as pellet. Encapsulins are detected using anti-his antibody, loading control is the RNA polymerase ß-subunit. 

Figure 2.4 shows a western blot for the expression of EncA. The signal from EncA was almost not detectable and so the brightness has been changed to show the presence of the bands. The marker is added seperately from the same membrane. There is a faint band for the EncA at about 35 kDa which is close to the expected size (37 kDa). The western blot supports the results seen in the microscope, that EncA is relatively low expressed and this could be the reason why we only see encapsulation at low concentrations of sfGFP.

Figure 2.5: Plasmids of encapsulin and sfGFP in changed backbone.

We cloned our EncA into the same as the microcompartment and the SPD-5 (pBbE6a). We also changed the backbone of the sfGFP containing the C-terminally fused target peptide to the same as the mVEN (pBbA2c). We repeated the microscopy results seen for the encapsulins in the pBAD33 backbone. 

Experimental setup: BL21 was transformed with pBAD33, same backbone as for the Encapsulins, and a plasmid containing either sfGFP or sfGFP with the encapsulation tag C-terminally.  The induction and incubation follow the same procedure as for the previous experiment.

Figure 2.6: Fluorescence microscopy of Encapsulin expression. E. coli (BL21) was co-transformed with sfGFP C-terminally fused to the targeting peptide for the Encapsulin. and either A)  or B) EncA  IPTG and Doxycyclin induces the expression of the EncA/ and sfGFP_TP respectively. Panels presents GFP fluorescence. sfGFP localises to fluorescent dots when expressed with encapsulins (scale bar 5 µm). 

Figure 2.6 shows that the expression of sfGFP was generally very low and mostly not detectable. However, when expressed in the presence of the encapsulins the formation of fluorescent dots could be observed. These results are similar to the previous where even leaky expression of sfGFP led to the formation of fluorescent dots.

From these results, it is clear that the formation of fluorecent “foci” is related to the expression of the encapsulins, but also that similar structures occasionally can occur even in the absence of encapsulins. The encapsulation via the targeting peptide is, as mentioned, non-covalent and as a result would be possible to be detected on native PAGE.

Aim: The encapsulins in their native state are expected to run as >5 MDa and wouldn’t be able to enter far into a native PAGE gel due to its size [2.1]. Successful encapsulation of sfGFP containing the targeting peptide would be detected in these bands.

Experimental setup: BL21 were transformed with pBAD_encA and either pET_sfGFP or pET_sfGFP_TP. A larger culture was grown until OD: 0.6-0.8 and induced with arabinose for the Encapsulins and IPTG for the sfGFP. Then cells were incubating for 24 h at 18°C until samples were analyzed with fluorescent microscopy. Prior to microscopy, 3 ml liquid culture is pelleted and lysed for the native PAGE.

Figure 2.8 shows the native PAGE gel fluorescent signal. Figure 2.7 shows the microscopy pictures of the same samples for comparison. It is important to note that while 3 ml of each sample was used for the preparation for the native page it is not possible to fully compare the fluorescent signal betwween the samples, as there is no loading control. For all samples there is some signal detected, corresponding to the expression of sGFP. This is also the case, even in the absence of induction. However, leaky expression of sfGFP is also detected in fluorescent microscopy. In the supernatant, fluorescent signal could be detected for conditions where sfGFP or sfGFP_TP was induced, even in the absence of encapsulin induction. Based on the microscopy results, the fluorescent “foci” were only detected for the sfGFP_TP strain and so this is an unexpected result. This would suggest that the presence of the targeting peptide does not influence the occurence of these structures. Furthermore, if these bands represent the encapsulins we would only expect them to be formed when encapsulins are induced. Therefore, it is difficult to conclude on the occurrence of fluorescent “foci” in the microscopy pictures and the native page results. For further proof of the correct formation of encapsulins methods such as Electron Microscopy (EM) would be nessesary. However, due to time constraint, this was not possible.

Figure 2.7: Fluorescence microscopy of Encapsulin expression. E. coli (BL21) was co-transformed with EncA and either A) sfGFP or A) sfGFP C-terminally fused to the targeting peptide for the Encapsulin. Arabinose and IPTG induces the expression of EncA and sfGFP respectively. Panels presents GFP fluorescence. sfGFP localises mostly throughout the cytoplasm in both conditions. Representive pictures of Native PAGE samples.

Figure 2.8: Native PAGE of encapsulins. E. coli (BL21) was co-transformed with either sfGFP or sfGFP C-terminally fused to the targeting peptide for the Encapsulin, and EncA. The plasmids were induced as described in the table. Fluorescence signal of GFP channel is shown in dark. Signal from sfGFP is present for all samples, however only in the supernatant is a fluorescent signal in the bottom of the wells detected.

Conclusion

Through this project, it was possible to express and detect the formation of the encapsulin-forming protein, EncA. Furthermore, results of fluorescent microscopy suggest the formation of fluorescent “foci” is related to the expression of EncA together with the targeting peptide containing sfGFP. However, occasionally a few similar structures could be found for other conditions. Native PAGE results suggest the formation of a larger structure in the supernatant, however, in contrast to the microscopy, this is also present without EncA expression and also for sfGFP with no targeting peptide. As mentioned, further repetitions and experiments would be needed to confirm these findings.

Liquid droplets: SPD-5 

The sequence of the SPD-5 from C. Elegans was codon-optimised for expression in E. coli and ordered in two parts from IDT. Gibson assembly was performed for the cloning into a pBAD33 backbone already containing the sequence for sfGFP. The fusion of SPD-5 to sfGFP allows for the visualisation with fluorescent microscopy.

Figure 3.1: Plasmids schemes of SPD-5 construct and sfGFP in pBAD

Aim: the aim of the first experiment was to test the SPD5::sfGFP construct with microscopy, with our newly cloned construct.  As it is a fusion construct the fusion of SPD-5 to another protein at the C-terminus could cause the proteins to misfold.  

Experimental setup: Strains are prepared in BL21 and are transformed with either pBAD33 containing sfGFP or SPD5::sfGFP and induced with either 0, 0.005 or 0.1% arabinose at OD: 0.6-0.8, then incubated for 24h at 18°C. 

Figure 3.2 shows that the fluorescent signal is dispersed throughout the cytoplasm when sfGFP is expressed alone and the occurrence of "foci" with strong intensity when it is expressed as a fusion protein with SPD-5. However, the negative control of the fusion protein appears to equally express the protein of interest and form similar dots. The plasmids were later sequenced for mutations in the promoter region, however, no differences were found between the plasmid containing sfGFP alone and the SPD-5::sfGFP fusion protein.   

Figure 3.2: Fluorescent microscopy of SPD-5::sfGFP expression. BL21 was transformed with either. A: pBAD33 containing sfGFP or B: SPD-5::sfGFP fusion protein. Increasing concentrations of inducer results in stronger fluorescence. Bacteria expressing SPD5::sfGFP form dots with strong fluorescent intensity. The images present representable regions of two experiments (scale bar 5 µm). 

Phase-separation droplets are expected to be dynamic, however, the dots observed did not move over time or fuse with other droplets. The experiment was repeated twice with similar results. Previous experiments have described how liquid droplet forming proteins transition from a dynamic state, where the droplets can move inside the cell to a more static gel-like phase [3.1]. There is a clear difference in the structures formed between the induction concentrations with higher concentrations leading to undefined large areas of fluorescence. Similar results have been shown for constructs where different arrangements of hydrophilic regions were tested [3.1]. The observations could be due to protein overexpression leading to inclusion bodies formed at the poles or that SPD-5 droplets are in their gel-like state. As BL21 is a strain optimised to produce large amounts of heterologous protein we decide to test our constructs in MG1655, a reference K-12 strain.  

MG1655 

Aim: the aim of the first experiment was to replicate the SPD5::sfGFP microscopy results, in MG1655 a standard K-12 E.Coli strain.  

Experimental setup: Strains are prepared with MG1655 and are transformed with either pBAD33 containing sfGFP or SPD5::sfGFP and induced with either 0, 0.005 or 0.1% arabinose at OD: 0.6-0.8. Then incubated for 24h at 18°C. 

Figure 3.3 shows the expression of the cloned constructs in MG1655 with fluorescent signal dispersed over the cytoplasm when sfGFP is expressed alone and the occurrence of areas with strong intensity when it is expressed as a fusion protein with SPD-5. Again, the uninduced sample of SPD5::sfGFP shows a fluorescent signal. For this experiment a lower concentration of Arabinose was tested, 0.001%, which also showed the ability to form fluorescent "foci".   

Figure 3.3: Fluorescent microscopy of SPD-5::sfGFP expression. MG1655 was transformed with either A: pBAD33 containing sfGFP or B: SPD-5::sfGFP fusion protein. Panels show GFP fluorescence. Bacteria expressing SPD5::sfGFP form dots with strong fluorescent intensity. The images present representable regions of two experiments (scale bar 5 µm). 

In contrast to the previous results, this time there was obvious movement of the fluorescent dots. Meaning, they changed position inside the cells independently of cell movement. This type of movement was recorded through multiple pictures of the same cell. Figure 3.4 shows a timeseries of one cell over the course of 1 min. This type of movement was not observed for all dots, but only for a smaller population. This could support the hypothesis that most dots are in the gel-like state, while some still having a dynamic character. 

Figure 3.4: Time series of dynamic liquid droplets. MG1655 was transformed with pBAD33 containing SPD-5::sfGFP fusion protein and induced with 0.005% Arabinose. Panel show GFP fluorescence. SPD5::sfGFP form dynamic dots. Pictures were acquired every 5 sec for 1 min (scale bar 5 µm). 

 Fluorescent recovery after photobleaching (FRAP) was perfomed on SPD-5::sfGFP fusion construct expressed in MG1655 as this showed some signs of dynamic droplets (Fig. 3.5). However, even after 30 min fluorescence is not recovered. It is worth noting that none of the photobleached dots showed dynamic characteristics. 

Figure 3.5: Time series of photobleaching of liquid droplets. MG1655 was transformed with pBAD33 containing SPD-5::sfGFP fusion protein. Panel show GFP fluorescence. SPD5::sfGFP fluorescent dots do not recover. Pictures were acquired every 30 sec for 30 min (scale bar 5 µm). 

Aim: To observe wether fluorescent "foci" are formed even after the fusion N- and C-terminally to the tags. To observe wether only successful catching of catcher-containing fluorescent proteins leads to fluoresence "foci".

Experimental setup: Strains are prepared in BL21 and are co-transformed with either pBbE6a containing SPD-5 or SPD-5 with spy- and snoop-tag on N- and C-term, respectively and pBbA2c containing either mVenus or mTurqouise. Strains were induced at OD: 0.6-0.8 and then incubated for 24h at 18°C. IPTG and Doxycyclin induces the expression of SPD-5 and mVenus/mTurquoise, respectively. Samples were induced with 100 µM IPTG and 25 ng/ml Doxycycline for mVenus and mTurquoise. Visualizing via fluorescent microscopy will allow us to observe SPD-5 based droplets only in the case where mVenus or mTurquise is catched and therefor colocalise inside the cell. 

Figure 3.6: Fluorescent microscopy of cells expressing SPD-5 and mVenus. BL21 was co-transformed with either A: a plasmid with SPD-5 as well as a plasmid expressing mVenus or B: a plasmid with SPD-5 with spy- and snoop-tag on N- and C-term as well as a plasmid expressing mVenus. Bacteria expressing SPD-5 and mVenus form dots with strong fluorescent intensity for both SPD-5 constructs (scale bar 5 µm).Representive pictures of a single replicate.   

Figure 3.7: Fluorescent microscopy of cells expressing SPD-5 and mTurquoise. BL21 was co-transformed with either A: a plasmid with SPD-5 as well as a plasmid expressing mTurquoise or B: a plasmid with SPD-5 with spy- and snoop-tag on N- and C-term as well as a plasmid expressing mTurquoise. Bacteria expressing SPD-5 and mTurquoise form dots with strong fluorescent intensity for both SPD-5 constructs (scale bar 5 µm).Representive pictures of a single replicate. 

Figure 3.6 shows the expression of the cloned constructs in BL21 with fluorescent signal dispersed over the cytoplasm when mVenus is expressed alone and the occurrence of areas with strong intensity when it is expressed in the presence of SPD-5. In this case, we observe no visible difference for whether SPD-5 contains the spy- and snoop-tag or not. This suggests that the dots we are observing are either inclusion bodies or that the fluorescent protein is interacting with SPD-5 in a noncovalent manner. However, if these fluorescent dots were only due to aggregated proteins we would expect similar result when expressing another protein together with mVenus. The experiments done with mVenus and the T1 protein from the microcompartments (see Figure) does not show the same tendency and this suggests that these structures are related to the expression of SPD-5. Figure 3.7 shows very similar results for the co-expression with mTurquoise. In contrast to mVenus, the expression of mTurqoise is stronger and the "foci" are larger and occasionally form large structures covering the half of the cell. The high intensity dots are not moving or seemingly dynamic.  

MG1655 

Aim:As BL21 did not show same type of dynamic droplets observed with the fusion construct in MG1655, we also decided to use MG1655 for the same experimental setup to see if this would change the characteristics of the "foci". Besides, lower concentrations of IPTG were choosen to lower the risk of forming inclusion bodies.

Experimental setup: Strains are prepared in MG1655 and are co-transformed with either pBbE6a containing SPD-5 or SPD-5 with spy- and snoop-tag on N- and C-term, respectively and pBbA2c containing either mVenus or mTurqouise. Strains were induced at OD: 0.6-0.8 and then incubated for 24h at 18°C. IPTG and Doxycyclin induces the expression of SPD-5 and mVenus/mTurquoise, respectively. Samples were induced with 10 µM IPTG and 25 ng/ml Doxycycline for mVenus and 10 ng/ml for mTurquoise.

Figure 3.8: Fluorescent microscopy of cells expressing SPD-5 and mVenus. MG1655 was co-transformed with either A: a plasmid with SPD-5 as well as a plasmid expressing mVenus or B: a plasmid with SPD-5 with spy- and snoop-tag on N- and C-term as well as a plasmid expressing mVenus. Bacteria expressing SPD-5 and mVenus form dots with strong fluorescent intensity for both SPD-5 constructs (scale bar 5µm).  

Figure 3.9 shows the western blot for the catching of mVenus. The expected size of mVenus with the spy-catcher is 37 kDa and the expected size when bound to SPD-5 is 167 kDa. In the case where a covalent bond between catcher and tag is created, a shift of mVenus band would be expected. The western blot shows a band at the right size of mVenus for the induced samples (lane 3, 5, 8, 10). In lane 10 a shift in the band for mVenus is detected at the expected size. This suggests that SPD-5 is capable of catching constructs with a spy-catcher.

Figure 3.9: Western blot of catching between SPD-5 and mVenus. MG1655 was co-transformed with either a plasmid with SPD-5 as well as a plasmid expressing mVenus (lane 2, 3, 4, 5) or a plasmid with SPD-5 with spy- and snoop-tag on N- and C-term as well as a plasmid expressing mVenus (lane 7, 8, 9, 10). There is a weak shift for the mVenus band when SPD-5 with spy- and snoop-tag is expressed. 

Figure 3.10 shows the westernblot of the same samples as above, for SPD-5 and mTurquoise. The figure shows similar results to the previous western blot. In agreement with the observations for mVenus the transfer of the large catched construct (lane 10) shows a clearer band. It also shows that there are a band at about 60 kDa when mTurquoise is expressed in the strain containing SPD-5 with tags. These bands are not observed when mTurquise is expressed with SPD-5 without tags. These results show that the SPD-5 with spy-and snoop-tag is capable of catching the respective cargo proteins, which the SPD-5 alone is not capable of. 

Figure 3.10: Western blot of catching between SPD-5 and mTurquoise. MG1655 was co-transformed with either a plasmid with SPD-5 as well as a plasmid expressing mTurquoise (lane 2, 3, 4, 5) or a plasmid with SPD-5 with spy- and snoop-tag on N- and C-term as well as a plasmid expressing mTurquoise (lane 7, 8, 9, 10). There is a shift for the mTurquoise band when SPD-5 with spy- and snoop-tag is expressed. 

Conclusion

Based on these results we were able to show that SPD-5 fused to GFP formed fluorescent dots in MG1655 and in BL21 which we identify as some state of liquid droplets. We could show that after 24 h at 18°C it was possible to find single dynamic droplets changing position in the cell. However, as described previously [3.1] the droplets transition over time to a gel-like state where the droplets localise to the poles and have little to no movement. In agreement, it was not possible to observe recovery of the fluorescent dots after photobleaching possibly due to the long induction time.  

We also created a construct with SPD-5 containing spy and snoop-tag allowing catching via spy or snoop catcher. We showed that these strains still form fluorescents dots when expressed in BL21 and MG1655. However, in agreement with the fusion construct these droplets do not seem to be in the dynamic state but rather in a gel-like state. We could confirm via western blot that even though we observe fluorescent dots when SPD-5 is expressed without tags together with catcher containing mVenus or mTurquoise, they do not catch either mVenus or mTurquoise.  

Future perspectives 

Our project seeks to compare types of compartents for co-localizing proteins. From this view, there would be several considerations to make when choosing to work with Liquid droplets. First of all, as is generally known, the formation of liquid droplets is sensitive to protein concentration and induction time [3.4]. As the gel-like state is not as dynamic it might be less suitable for increasing product yield due to less exchange with the cytoplasma. Testing different induction conditions for a liquid droplet forming protein of interest would be necessary. Along the same lines, having the possibility to further regulate the formation of droplets and possibly avoid them turning to the gel-like state could be beneficial. 

Optimisation with ncAAs

The different mutations of the BMC-T1-spyt-snpt-6xHis protein were not randomly selected. We selected potential non-conserved amino acids using the consurf database. We followed the generally accepted view that the less conserved an amino acid is, the less relevant it is for the correct folding of the protein. The remaining non-conserved amino acids and their positions were considered and only amino acids that had a similar size and shape compared to the ncAAs were selected for site directed mutagenesis at their position.

This resulted in the mutations: BMC-T1-spyt-snpt-6xHis F8, T35, R78 and Y96. 

The same applies to the mutations introduced into sfGFP.

ncAAs in GFP

Detecting the incorporation of non-canonical amino acids by flow cytometry

Besides the well-known 20 canonical amino acids, there is a variety of non-canonical amino acids (ncAAs) which can be used to further modify proteins. One way of incorporating ncAAs is via the amber stop codon suppression technology. 

An aminoacyl-tRNA synthetase (aaRS) and a tRNA (orthogonal) derived from other species must be introduced into the organism. The tRNA must not be recognised by the aaRSs of the organism, the same applies to the introduced aaRS and tRNAs of the cell. In addition, the new tRNA must introduce the ncAA via a codon that does not encode any of the 20 natural amino acids, and the ncAA must be taken up by the cell. The requirements described above are fulfilled by E. coli in combination with the evolved tyrosyl-tRNA/synthetase pair of Methanococcus jannaschii. Amber stop codon (UAG) suppression technology is used for this purpose, whereby the codon is recognised by the tRNA, and the loaded ncAA is incorporated. 

Testing the incorporation of 4-Azido-L-phenylalanine (p-AzF) into sfGFP 

Aim:  Indirectly test the incorporation of p-AzF into sfGFP measuring fluorescence using flow cytometry under different induction levels 

Experimental design: E. coli BL21 (DE3) cells were co-transformed with the plasmid coding for the pAzF synthetase and the corresponding tRNA, and the plasmid coding for the sfGFP Y74 mutant (pTrc99a_sfGFP Y74; http://parts.igem.org/Part:BBa_K4229023).

Colonies from the co-transformation were  grown overnight at 37°C in LB medium supplemented with ampicillin and chloramphenicol. Overnight cultures were diluted and grown to OD600 of 0.4 at 30°C before induction. Samples were induced with different arabinose concentrations as indicated in Figure 4.1 1 mM p-AzF was added into the medium, unless differently indicated (for instance, for the control, in which no ncAA was included). One hour after arabinose induction (which triggers expression of the tRNA synthetase), the samples were induced with 500 µM IPTG to induce sfGFP and growth was continued for additional 3 h at 30°C. 

The incorporation of the ncAA was indirectly measured by fluorescence, which was quantified by flow cytometry with the CyAn ADP Analyzer from Beckman Coulter. Samples were prepared by mixing 10 µL of the liquid bacterial cultures with 990 µL Dulbecco's Balanced Salt Solution 

(DPBS).  

Results: 

As showcased in Figure 4.1, different levels of arabinose do not seem to significantly affect the median fluorescence intensity (MFI) of the cells expressing the sfGFP mutant or wild-type. This is of course expected for WT sfGFP given that arabinose is required for the expression of the orthogonal tRNA synthetase, which is not needed for the WT protein. The MFI of the cells expressing the mutant is consistently lower than that of the cells expressing wild-type sfGFP. We were not surprised by this result, considering that we do not expect 100% efficiency of ncAA incorporation. Nonetheless, even at very low arabinose levels, incorporation of p-AzF is occurring. Interestingly, fluorescence can be detected even without induction with arabinose, but in the presence of p-AzF. We speculate this is due to the leakiness of the promoter, which leads to the expression of the synthetase even without arabinose. It moreover suggests that very low amounts of synthetase are sufficient, which is in line with the results obtained with increasing amounts of arabinose (Fig. 4.1). 

Figure 4.1: Bar graph showing the Median Fluorescence Intensity (MFI) of E.coli BL21 (DE3) cells transformed with pEVOL-pAzF and pTrc99a coding for the indicated GFP variant induced with the indicated arabinose concentrations. GFP fluorescence was measured by flow cytometry. 500 µM IPTG were used to induce sfGFP expression. 1 mM p-AzF were added to all samples except for the sample labelled “N.C.”, which did not receive p-AzF. This sample was also not induced at all (no arabinose, no IPTG).

Influence of ncAAs and the synthetases on wild-type sfGFP fluorescence 

Aim:  Explore the impact of the ncAA and/or the expression of the aminoacyl tRNA synthetase on the median fluorescence intensity of wild-type sfGFP. 

Experimental setup: E. coli BL21(DE3) were co-transformed with the plasmid coding for either the pAzF or the pBpF synthetase, and the plasmid coding for wild-type sfGFP (pTrc99a-sfGFP).

A single colony from each co-transformation was grown overnight at 37°C in LB medium supplemented with ampicillin and chloramphenicol. Overnight cultures were diluted and grown to OD600 of 0.4 at 37°C before induction. Samples were induced with 1mM arabinose (to induce the expression of the synthetase). 1 mM p-AzF or 0.2 mM p-BpF were added to all samples except in the controls, where the ncAA were omitted. One hour after arabinose induction the samples were induced with 200 µM IPTG (to allow expression of sfGFP) and growth was continued for additional 2.5 h at 37°C  

The incorporation of the ncAA was indirectly measured by fluorescence, which was quantified by flow cytometry with the CyAn ADP Analyzer from Beckman Coulter. Samples were prepared by mixing 10 µL of the liquid bacterial cultures with 990 µL DPBS.  

Results: 

We expected no difference in the GFP fluorescence across all the samples, given that WT sfGFP does not rely on the orthogonal tRNA/tRNA synthetase pair and does not incorporate the ncAA. We maximally expected to see lower fluorescence in cells expressing the synthetase, because of the burden of producing an additional heterologous protein. The results we obtained were quite surprising and puzzling (Figure 4.2). Since we performed this experiment only once, we tend to be cautious in interpreting the data. While doing the experiments, we indeed realised that the GFP fluorescence varied a lot from day to day and we could not always pin down the reason for it. 

Figure 4.2: Bar graph showing the Median Fluorescence Intensity (MFI) of BL21(DE3) cells transformed with pEVOL-pAzF (pAzFRS), pEVOL-pBpF (pBpf RS) or empty pBAD33 (pBAD33). GFP fluorescence was measured by flow cytometry. 500 µM IPTG were used to induce sfGFP expression. 1 mM pAzF or 0.2 mM p-BpF were added to all samples labelled with the “+”

Testing different positions within sfGFP for the incorporation of ncAAs

Aim: Design a toolbox for the easy quantification of the efficiency of ncAA incorporation via the amber stop codon suppression technology. 

Experimental setup:

We introduced the amber stop codon at positions G4, F8, W57, Y74, Y151 and D216 in sfGFP by site directed mutagenesis.  E. coli BL21(DE3) were co-transformed with the following combinations of plasmids:

• pEVOL-pBpF and pTrc99a-sfGFP WT
• pEVOL-pBpF and pTrc99a-sfGFP G4
• pEVOL-pBpF and pTrc99a-sfGFP F8
• pEVOL-pBpF and pTrc99a-sfGFP W57
• pEVOL-pBpF and pTrc99a-sfGFP Y74

• pEVOL-pBpF and pTrc99a-sfGFP Y151
• pEVOL-pBpF and pTrc99a-sfGFP D216
• pBAD33 and pTrc99a-sfGFP WT
• pBAD33 and pTrc99a-sfGFP G4
• pBAD33 and pTrc99a-sfGFP F8

• pBAD33 and pTrc99a-sfGFP W57
• pBAD33 and pTrc99a-sfGFP Y74
• pBAD33 and pTrc99a-sfGFP Y151
• pBAD33 and pTrc99a-sfGFP D216
• pBAD33 and pTrc99a

A single colony from each co-transformation was grown overnight at 37°C in LB medium supplemented with ampicillin and chloramphenicol. Overnight cultures were diluted and grown to OD600 of 0.4 at 37°C before induction. Samples were induced with 1mM arabinose and 0.2 mM of p-BpF were added in all samples except in the control. One hour after arabinose induction the samples were induced with 200 µM IPTG and growth was continued for additional 2.5 h at 37°C 

The incorporation of the ncAA was indirectly measured by fluorescence, which was quantified by flow cytometry with the CyAn ADP Analyzer from Beckman Coulter. Samples were prepared by mixing 10 µL of the liquid bacterial cultures with 990 µL DPBS. 

Results:

Interestingly, not all mutations exhibited fluorescence in the presence of p-BpF and the corresponding synthetase (Figure 4.3). This suggests that the amino acid context plays a role in facilitating or inhibiting the incorporation of a ncAA.

Figure 4.3:Bar graph showing the Median Fluorescence Intensity (MFI) of BL21(DE3) cells transformed with pEVOL-pBpF and the indicated sfGFP variants. GFP fluorescence was measured by flow cytometry. 200 µM IPTG were used to induce sfGFP expression. 0.2 mM p-BpF were added to all samples labelled with the “+”. Blue indicates that no synthetase was present (-) but the medium contained 0.2 mM p-BpF (+), orange stands for the presence of both the synthetase (+) and 0.2 mM p-BpF (+), and grey indicates the presence of the synthetase (+) and the absence of p-BpF (-). Data are means ± standard deviation of two independent experiments.

Photo-crosslinking of bacterial microcompartment subunits with p-benzoyl-l-phenylalanine

P-benzoyl-l-phenylalanine (p-BpA or p-Bpf) is a non-canonical amino acid with a benzophenon residue and commonly used for photochemical crosslinking [4.1]. When irradiated with UV light it can form covalent bonds with nearby C-H bonds. The reaction mechanism for photo-crosslinking with p-Bpf is shown in the figure below.

Figure 4.4: Irradiation of p-Bpf at 365 nm: If p-Bpf is irradiated with a wavelength of 365 nm, a diradical species is generated. If another molecule, with an activated C-H group, approaches, a radical also occurs at this molecule, since the H binds to one of the two previous radicals. Subsequently, a radical recombination occurs and the two molecules are linked, adapted from [4.2].

For the following experiments p-Bpf was incorporated via amber stop codon suppression technology with an orthogonal translation system consisting of an aminoacyl tRNA synthetase (from Addgene, pEVOL-pBpF (plasmid #31190)) and the corresponding tRNA.

The amber stop codons were introduced with the help of site directed mutagenesis at the positions F8, T35, R78 and Y96. 

Aim: Covalently crosslinking the building blocks of microcompartments can significantly impact their stability. With this experiment we want to show that in vivo crosslinking of the subunits of our bacterial microcompartments is possible by utilizing non-canonical amino acids (ncAAs). 

Experimental Setup:  E.coli BL21 (DE3) were co-transformed with the following combinations of plasmids: 

• pEVOL-pBpF and pHT1_spyt_snptT2T3 F8
• pEVOL-pBpF and pHT1_spyt_snptT2T3 T35
• pEVOL-pBpF and pHT1_spyt_snptT2T3 R78 
• pEVOL-pBpF and pHT1_spyt_snptT2T3 Y96
• pBAD33 and pHT1_spyt_snptT2T3 WT
• pBAD33 and pHT1_spyt_snptT2T3 T35

A single colony from each co-transformation was grown overnight at 37°C in LB medium supplemented with ampicillin and chloramphenicol. The first five overnight cultures were transferred into fresh Lb medium supplemented with 0.2 mM unnatural amino acid, p-Bpf and grown to OD600 of 0.4 at 37°C before induction with 1mM arabinose. One and a half hours after arabinose induction the samples were induced with 200 µM IPTG and growth was continued 28 h at 18°C. The last construct (pBAD33 and pHT1_spyt_snptT2T3 T35) was not induced and no ncAA was added, it is the negative control.

To crosslink the proteins, 6 ml of the medium per batch was centrifuged at 5000 g, and the supernatant was discarded. 3.6 ml of PBS was used to resuspend the cells, then these were centrifuged at 3000 g. The supernatant was discarded again, and the pellet was resuspended in 3 ml PBS. Samples were placed in 47 mm petri dishes. The samples were irradiated with a UVP crosslinker from analytikjena and 100 μl per sample were collected after each irradiation step. Samples were then centrifuged and resuspended with Lämmli buffer and heated to 95 °C for 5 min. Then they were separated using an SDS gel and a Western blot was made.

Firgure 4.5: The different mutations are represented by red spheres in the T1 protein of the HO shell for spatial viewing. The three subunits of the pseudohexamer were coloured different. (A) Mutation F8 in each of the three subunits of the pseudohexamer. (B) Mutation T35. (C) Mutation R78. (D) Mutation Y96. Images were generated using PyMOL [4.3].

In Figure 4.5, it can be seen that no cross-linking is expected for F8 and R78 because the mutations are located in the centers of the three subunits, respectively, and thus p-Bpf cannot come into spatial proximity with other proteins and cannot form a covalent bond. For the T35 mutation, based on the figure, it can be assumed that a crosslink to a neighboring T1 protein, is possible. The same is true for Y96, but here the formation of a bond also to other proteins such as the H-Protein is likely, since p-Bpf is located on the outside.

Figure 4.6: Western blot of BMC-T1-spyt-snpt-6xHis Y96 expression. E. coli (BL21) was co-transformed with pEVOL-pBpF and pHT1_spyt_snptT2T3 Y96. Arabinose and IPTG induced the expression of the aminoacyl-tRNA synthetase and the proteins of the plasmid pHT1_spyt_snptT2T3, p-Bpf was added. BMC-T1-spyt-snpt-6xHis are detected using anti-his antibody, loading control is the RNA polymerase ß-subunit. 

In the unirradiated sample (Fig. 4.6, left), it can be clearly seen that only the bands of the BMC-T1-spyt-snpt-6xHis protein and the RNA polymerase ß-subunit are visible. For the other samples, one can clearly see bands at other heights. Especially the bands between 72 and 100 kDa stand out. However, this does not indicate a connection to H-proteins as initially assumed but suggests connections to another BMC-T1-spyt-snpt-6xHis or a protein of similar size. 

Due to the particularly strong bands in rows three, four and five, the parameters used were classified as suitable for crosslinking. Therefore, for the following two Western blots, these samples, among others, were selected for the other mutations.

Figure 4.7: Western blot of BMC-T1-spyt-snpt-6xHis F8, T35 and controls pHT1_spyt_snptT2T3 WT and pHT1_spyt_snptT2T3 T35 uninduced. The table shows which samples were loaded and with which intensity they were irradiated for how long. Arabinose and IPTG induced the expression of aminoacyl-tRNA synthetase and proteins of plasmid pHT1_spyt_snptT2T3, p-Bpf was added. This applies to all samples except the one in row two, which was not induced and no ncAA was added. BMC-T1-spyt-snpt-6xHis are detected using anti-his antibody, loading control is the RNA polymerase ß-subunit. 

Figure 4.7 shows that BMC-T1-spyt-snpt-6xHis F8 does not show crosslinking to other proteins as expected, however this is also the case for T35, previous experiments came to the same conclusion.

Figure 4.8: Western blot of BMC-T1-spyt-snpt-6xHis R78, Y96 and controls pHT1_spyt_snptT2T3 WT and pHT1_spyt_snptT2T3 T35 uninduced. The table shows which samples were loaded and with which intensity they were irradiated for how long. Arabinose and IPTG induced the expression of aminoacyl-tRNA synthetase and proteins of plasmid pHT1_spyt_snptT2T3, p-Bpf was added. This applies to all samples except the one in row two, which was not induced and no ncAA was added. BMC-T1-spyt-snpt-6xHis are detected using anti-his antibody, loading control is the RNA polymerase ß-subunit. 

Figure 4.8 shows that BMC-T1-spyt-snpt-6xHis R78 is not cross-linked with other proteins as expected. For BMC-T1-spyt-snpt-6xHis Y96, the same samples as in Figure 4.7 were loaded and cross-linking could again be detected between 72 and 100 kDa. 

The first Western blot (Fig. 4.6) was performed on the same day as the cross linking. Interestingly, this one shows stronger bands than the western blots two and three (Fig. 4.7 and 4.8),especially for cross linking. For these, the exact same amounts were loaded, but the samples were frozen overnight and thawed the next day for the blots. The fact that the bands produced less intense signals after freezing, under otherwise exactly the same conditions, was observed several times throughout the project.

NcAA in wiffleballs

Short Abstract about achievements

Our goal is to introduce non-canonical amino acids (ncAAs) at specific positions into wiffleballs (bacterial microcompartment shells derived from Haliangium ochraceum also referred to as BMCs) to allow cross-linking with other proteins at these positions (pBpF, p-benzoyl-l-phenylalanine) or to introduce amino acids with azido groups as side chains for possible click chemistry reactions (pAzF, p-azido-l-phenylalanine). We used rational design by first looking at the crystal structure with PyMOL while colouring the amino acids based on their conservation using the ConSurf database and looking for positions that were not highly conserved but close to other compartment components. We decided to look at the BMC-T1-F8TAG, BMC-T1-T35TAG, BMC-T1-R78TAG and BMC-T1-Y96TAG mutations. To detect the incorporation of the amino acid, we tested different conditions and detected them with Western blots using an α-HisTag antibody. Finally, to establish that these mutant wiffleball proteins can still be caught by the SpyCatcher and SnoopCatcher systems associated with our pathway enzymes, we tested all systems together under different conditions and detected them with an α-HisTag as well as α-Catcher antibody.

Results: In order for the incorporation of the ncAA to succeed, an amber stop codon (TAG) was inserted at amino acid position 35 (Fig. 4.5A) in the DNA sequence, where a threonine is normally located. This construct (BMC-T1-T35TAG) was then transformed with aminoacyl tRNA synthetase (aaRS) and the complementary tRNA.
Aim: We looked at the BMC-T1-T35TAG mutant in E. coli BL21 (DE3) and performed various tests with it to characterise the other mutations more quickly. We want to show the incorporation of the ncAA into the protein.
Experimental design: Matching controls were also transformed so that the incorporation can be attributed to the addition of the ncAA, the expression of the aaRS/tRNA and the expression of the BMC-T1-T35TAG mutant. As controls, we transformed "pBAD + pHT1spysnpT2T3" as the pHT1spysnpT2T3 plasmid contains all the constructs required to produce the complete wiffleball structure (BMC C). The BMC-T1-T35TAG mutant was created through site directed mutagenesis of the pHT1spysnpT2T3 construct to still be able to benefit from the structure of a whole wiffleball instead of the minimal wiffleball.
Transformed controls in E. coli BL21 (DE3):

• pBAD + pTrc99a
• pBAD + pHT1spysnpT2T3
• pBAD + BMC-T1-T35TAG
• pEVOL-pAzFRS + pHT1spysnpT2T3
• pEVOL-pAzFRS + BMC-T1-T35TAG

For cells treated with pAzF, 1 mM pAzF was added into the medium before growing the cells. The cells were grown until an OD of 0.4, were then treated with 1 mM arabinose and grew for two more hours at 37 °C. After, the cells were treated with 100 µM of IPTG, grew for 5 more hours at 30 °C and were harvested by centrifugation. The samples were diluted with Laemmli buffer depending on the OD600 of the cell culture, and then heated for 5 minutes at 95°C. After, the samples were loaded on an SDS-PAGE, running for 2.5 hours with constant 0.01 mA per gel. Then, the samples were wet blotted for 2 hours at 0.38 mA. An anti-6xHisTag antibody (1:1000) as well as an anti RNA polymerase antibody (1:1000) in combination with an anti-mouse-HRP secondary antibody (1:10000) were used for detection. The anti-6xHisTag antibody should only detect the BMC-T1 protein due to the N-terminally added 6xHis-Tag.

Results: The Western blot (Fig. 4.10) shows a band at the level of slightly above 35 kDa for "pBAD + pHT1spysnpT2T3" induced with IPTG, "pEVOL-pAzFRS + pHT1spysnpT2T3" induced with IPTG only and IPTG plus arabinose, and for "pEVOL-pAzFRS BMC-T1-T35TAG", upon addition of 1 mM pAzF and induction by IPTG and addition of 1 mM pAzF and induction by IPTG + arabinose.

Figure 4.10: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 35 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 5 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 180 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, Ara – arabinose, pAzF – p-azido-l-phenylalanine.

The bands seen at 35 kDa is assumed to be our proteins of interest, as samples possessing a plasmid with the BMC-T1 gene and being induced with IPTG should induce our BMC-T1 protein. The BMC-T1 was detected by ECL due to its N-terminal 6xHisTag antibody.
ConclusionInterestingly, we thought that BMC-T1 should be found at a size of 29 kDa, but the band runs at 35 kDa. However, there is no gene on the BMC plasmid that has a 6xHisTag and is 35 kDa in size. The same effect also occurs within other membranes.
The aaRS for pAzF does not appear to affect the expression of the BMC-T1 protein. For our sample, which has the aaRS, tRNA as well as the BMC-T1-T35TAG plasmid, a clear band is seen but not in the absence of pAzF despite induction by IPTG and arabinose. Therefore, we can assume that specifically pAzF is incorporated into this protein. Interestingly, a band is seen in the "pEVOL-pAzF BMC-T1-T35TAG" sample where 1 mM pAzF was added to the medium and induced with only 100 µM IPTG. However, this band can be explained due to synthetase having two variants of the synthetase, one with a mid-strength glnS promoter next to the synthetase and the other with a lac promoter induced by IPTG.
After the expression of BMC-T1-T35TAG worked well with the incorporation of pAzF, we also tested the three other mutations (Fig. 4.5 B-C) with pAzF (Figure 4.11 + 4.12) and all four mutations with pBpF (Figure 4.13 + 4.14) as well. As we figured out by previous experiments with the synthetic full wiffleball construct (BMC C, pHT1spysnpT2T3), we increased the growth time after ITPG induction to 24 hours at 18°C to maximise the production of the protein.
Experimental design: The following transformations of E. coli BL21 (DE3) cells were used:

• BAD + pTrc99a
• pBAD + pHT1spysnpT2T3
• pBAD + BMC-T1-T35TAG
• pEVOL-pAzFRS + pHT1spysnpT2T3
• pEVOL-pAzFRS + BMC-T1-F8TAG
• pEVOL-pAzFRS + BMC-T1-T35TAG
• pEVOL-pAzFRS + BMC-T1-R78TAG
• pEVOL-pAzFRS + BMC-T1-F96TAG
• pEVOL-pBpFRS + BMC-T1-F8TAG
• pEVOL-pBpFRS + BMC-T1-T35TAG
• pEVOL-pBpFRS + BMC-T1-R78TAG
• pEVOL-pBpFRS + BMC-T1-F96TAG

For cells treated with pAzF, 1 mM pAzF was added into the medium before growing the cells. The cells were grown until an OD of 0.4, were then treated with 1 mM arabinose and grew for two more hours at 37 °C. After, the cells were treated with 100 µM of IPTG, grew for 5 more hours at 30 °C and were harvested by centrifugation. The samples were diluted with Laemmli buffer depending on the OD600 of the cell culture, and then heated for 5 minutes at 95°C. After, the samples were loaded on an SDS-PAGE, running for 2.5 hours with constant 0.01 mA per gel. Then, the samples were wet blotted for 2 hours at 0.38 mA. An anti-6xHisTag antibody (1:1000) as well as an anti RNA polymerase antibody (1:1000) in combination with an anti-mouse-HRP secondary antibody (1:10000) were used for detection. The anti-6xHisTag antibody should only detect the BMC-T1 protein due to the N-terminally added 6xHis-Tag.
Results: We get the same results for BMC-T1-T35TAG (Fig. 4.11), -R78TAG, and -Y96TAG (Fig. 4.12) when incorporating pAzF, as well as for BMC-T1-F8TAG, -T35TAG (Fig. 4.13), -R78TAG, and -Y96TAG (Fig. 4.14) when incorporating pBpF.

Figure 4.11: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 8 and 35 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 250 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) antibody were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.12: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 78 and 96 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 250 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) antibody were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.13: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 8 and 35 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 250 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) antibody were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

Figure 4.13: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 78 and 96 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 250 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) antibody were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

Figure 4.14: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 78 and 96 of the BMC-T1 protein. The cells grew until OD 0.4, following the addition of arabinose and a growth period of 2 h. After the 2 h the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. To purify the protein, the His-Spin Protein Miniprep from Zymo Research was used.

Discussion: As can be seen in figure 3, there were problems detecting the production of BMC-T1-F8TAG with pAzF, due to an unnoticed air bubble remaining in the SDS gel during polymerisation. However, we can provide further evidence of this protein in figure 7, as the mutants with incorporated pAzF were purified using a HisTag column and loaded onto an SDS-PAGE, to confirm the working purification for further analysation via mass spectrometry.

Figure 4.15: SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 8, 35, 78 and 96 of the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. To purify the protein, the His-Spin Protein Miniprep from Zymo Research was used.

The primary band (induced pHT1spysnpT2T3) in figure 4.15 was loaded as a positive control to see whether the band where the protein has been overexpressed corresponds to the band of the purified samples. The purified proteins run at ~29 kDa, despite being detected at 35 kDa on the western blots. There is no other construct containing an HisTag.
Aim: After we found out that the incorporation of the non-canonical amino acids we use works with every mutation, it was of course interesting to transform the plasmid containing the indigo enzymes or an empty plasmid with these constructs.
Before looking at the mutants, we expressed the non-mutant full wiffleball along with the indigo enzymes (Fig. 4.16). When only the BMC is transformed and expressed, a band at 35 kDa is evident, as has been detected in previous Western blots.

• pTrc99a + pCDFDUET-1
• pTrc99a + sTAF sXTB
• pHT1spysnpT2T3 + pCDFDUET-1
• pHT1spysnpT2T3 + sTAF sXTB

Figure 4.16: Western blot of an SDS-PAGE with whole cell lysis confirming the catching of Indigo enzymes by the BMC-T1 protein. The cells grew until OD600 0.4, following the addition of 200 µM IPTG and grew 24 hours before harvesting by centrifugation. The PageRuler™ Prestained Protein Ladder, 10 to 250 kDa from ThermoFisher Scientific was used. For detection, an anti 6xHisTag (1:1000) as well as an anti RNA polymerase (1:1000) antibody were used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside.

Aim: We then decided to focus on the BMC-T1-T35TAG mutant and therefore have four different samples to test: mutant with pAzFRS or pBpFRS and empty indigo backbone (pCDFDUET-1) as well as mutant with pAzFRS or pBpFRS and the genes for the indigo enzymes (sTAF sXTB). We want to show the incorporation of the ncAAs as well as the catching of the enzymes simultaneously.
Experimental design: The following transformations were inoculated:

• pTrc99a + pBAD + pCDFDUET-1
• pTrc99a + pBAD + sTAF sXTB
• pHT1spysnpT2T3 + pBAD + pCDFDUET-1
• pHT1spysnpT2T3 + pBAD + sTAF sXTB
• BMC-T1-T35TAG + pBAD + sTAF sXTB
• BMC-T1-T35TAG + pEVOL-pAzFRS + pCDFDUET-1
• BMC-T1-T35TAG + pEVOL-pAzFRS + sTAF sXTB
• BMC-T1-T35TAG + pEVOL-pBpFRS + pCDFDUET-1
• BMC-T1-T35TAG + pEVOL-pBpFRS + sTAF sXTB

For cells treated with pAzF, 1 mM pAzF was added into the medium before growing the cells. The cells were grown until an OD of 0.4, were then treated with 1 mM arabinose and grew for two more hours at 37 °C. After, the cells were treated with 200 µM of IPTG, grew for 24 more hours at 18 °C and were harvested by centrifugation. The samples were diluted with Laemmli buffer depending on the OD600 of the cell culture, and then heated for 5 minutes at 95°C. After, the samples were loaded on an SDS-PAGE, running for 2.5 hours with constant 0.01 mA per gel. Then, the samples were wet blotted for 2 hours at 0.38 mA. An anti-6xHisTag antibody (1:1000) as well as an anti RNA polymerase antibody (1:1000) in combination with an anti-mouse-HRP secondary antibody (1:10000) were used for detection. The anti-6xHisTag antibody should only detect the BMC-T1 protein due to the N-terminally added 6xHis-Tag.
Results: First, we decided to detect the 6xHisTag to see if the band would shift if the indigo enzymes were caught. We loaded controls for each blot to better prove our assumptions (Fig. 4.17-4.20). We get a band at the level of 35 kDa when only the full wiffleball is expressed. However, if the wiffleball is expressed together with the indigo enzymes, we still see a band at 35 kDa, but also at 100 kDa. If only the indigo enzymes are expressed, no band is seen, therefore, the 6xHisTag antibody has no interaction with the indigo enzymes.

Figure 4.17: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pAzFRS, and pCDFDUET-1. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-6xHisTag antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.18: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pAzFRS, and sTAF sXTB. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-6xHisTag antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.19: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pBpFRS, and pCDFDUET-1. TThe cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-6xHisTag antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

Figure 4.20: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pBpFRS, and sTAF sXTB. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-6xHisTag antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

If the mutant BMC-T1 proteins are formed with pAzF or pBpF and simultaneously transformed with the pCDFDUET-1 plasmid (Fig. 4.17 + 4.19), no shift is seen, and it looks like the previous Western blots. However, when the mutants are transformed and expressed with sTAF sXTB, a clear band is seen at 100 kDa, which can also be seen for the positive control when the unmutated BMC and sTAF are transformed and expressed (Fig. 4.19 + 4.20).
Conclusion
The incorporation of the ncAAs as well as the catching was successful. In both variants (pAzF and pBpF) it can be seen that two bands exist (35 kDa and 100 kDa). Accordingly, more BMC-T1 could be produced than is necessary to capture all enzymes. Aim: After detecting the membranes with the 6xHisTag antibody, we wanted to detect the same membranes with a catcher antibody, which should detect the catchers of the indigo enzymes and not BMC-T1.
Experimental design: For this purpose, it is important to inactivate horse radish peroxidase (HRP) after the first detection so that it does not give a background signal during the next detection and thus distort the results. Sodium azide is also mixed into our primary antibodies so that the HRP can be inactivated in precisely such cases. After incubation with the catcher antibody and before incubation with the secondary antibody, the membranes were detected again via ECL to measure any remaining signal (Fig. 4.21). The membranes were detected for 2 minutes each and no signal is measured, therefore the catcher antibody can be detected reliably.

Figure 4.21: Western blot of an SDS-PAGE with whole cell lysis after decorating the membranes with an α-6xHisTag antibody and α-mouse secondary antibody via ECL and then incubating with an α-Catcher antibody and detecting via ECL After the membranes have been decorated and detected with anti-6xHisTag, the same membranes should also be detected and decorated with an anti-Catcher antibody. After incubation with the anti-Catcher antibody spiked with sodium azide, the membranes were detected using ECL to see if there was still functional HRP from the detection of the first antibody. The sodium azide inactivates the remaining HRP. HPR - horse radish peroxidase.

Results: After re-incubation with a secondary mouse antibody, bands at the level of ~55 kDa are now detectable as well as bands at 100 kDa (Fig. 4.22-2.25).

Figure 4.22: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pAzFRS, and pCDFDUET-1. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-Catcher antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.23: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pAzF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pAzFRS, and sTAF sXTB. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-Catcher antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pAzF – p-azido-l-phenylalanine.

Figure 4.24: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pBpFRS, and sTAF sXTB. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-Catcher antibody was used (1:1000) in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

Figure 4.25: Western blot of an SDS-PAGE with whole cell lysis confirming the incorporation of pBpF at position 35 in the BMC-T1 protein with the transformation of BMC-T1-T35TAG, pEVOL-pBpFRS, and sTAF sXTB. The cells grew until OD600 0.4, following the addition of arabinose and a growth period of 2 hours. After 2 hours, the cells were treated with IPTG and grew 24 hours before harvesting by centrifugation. The BlueClassic Prestained Protein Marker, 10-180 kDa from Jena Bioscience was used as ladder. For detection, an α-Catcher antibody (1:1000) was used in combination with an anti-mouse-HRP secondary antibody (1:10000). IPTG - Isopropyl β-D-1-thiogalactopyranoside, pBpF – p-benzoyl-l-phenylalanine.

Conclusion

The bands at 55 kDa fit well with the expected size of the band, as XiaI (Indigo enzyme) has a size of 42.5 kDa and the spyCatcher has a size of 12.3 kDa, which should be detectable at 55 kDa. If the empty backbone of the indigo enzymes is transformed, no band is visible. However, if you transform the plasmid with the enzymes but do not express it, you see a weak band at 55 kDa, which suggests leakiness. If the enzymes are expressed, strong bands are seen at 55 kDa, but also below, which indicates degradation. When the full wiffleball is expressed together with the indigo enzymes, a band at ~100 kDa can be seen, which was also detected by the 6xHisTag antibody. This should be the BMCs that captured the indigo enzymes. However, you can see that there is still a strong band at 55 kDa, which suggests that not every enzyme is captured. Since we saw the same for the 6xHisTag antibody, there are both BMCs and indigo enzymes in the cell that are not catching each other.
For the BMC mutants that were not transformed with sTAF sXTB (Fig. 4.22 + 4.24), light bands are seen for the non-induced samples, but not for the induced samples. Here it looks like a little of the other sample has leaked into the pocket before running the SDS-PAGE. We have also seen previously that it can bind antibodies non-specifically, and we are mainly interested in the samples where both the indigo enzymes and the mutant BMC were expressed.
In the samples in which both the mutant BMC and the indigo enzymes were expressed (Fig. 4.23 + 4.25), a band at the level of 100 kDa is also seen. However, the same phenomenon is seen as with the unmutated BMCs and indigo enzymes: not all enzymes seem to be captured. Accordingly, the stoichiometry of the proteins would have to be refined, but we can prove that our enzymes are captured despite the mutation.

Pathways

---

Indigo/Indirubin pathway

Key findings:

• With the help of three different techniques, we were able to prove that it is indeed indigo and indirubin that we produce.
• We have successfully incorporated our enzymes into the full wiffleball.
• We have successfully created constructs that include our pathway, mutated wiffleball (with the T1 protein bearing the amber stop codon) and the incorporation of ncAAs using the pAzF synthetases.
• We have successfully produced indigo on a large scale by using a bioreactor.

Design of the expression plasmid

The first step to work on the production of indigo was to design the plasmid that we wanted to use. We have decided to use the genes that were used by Yin et al. [5.1] for the production of indigo and indirubin in E. coli.

Figure 5.1: Schematic representation of the indigo/indirubin pathway.L-Tryptophan is imported by the membrane protein TnaB (low affinity tryptophan permease). L-tryptophan is cleaved into indole, NH4+ and pyruvate by the Tryptophanase TnaA. The reaction continues by the hydroxylation of indole through XiaI. To enhance the effectivity of this enzyme, the NAD(P)H-flavin reductase provides XiaI with FADH2 by adding Hydrogen to FAD. Finally, indole is transformed to either 3-Hydroxyindole or 2-Hydroxyindole. These two substances spontaneously react to 3-Oxindole and 2-Oxindole through the secession of hydrogen from the OH-group. Through spontaneous dimerization indigo and indirubin are formed. Graphic adapted from [1.2].

These genes are TnaA, Fre, XiaI and TnaB (protein sequence information available on UniProt [2][3][4][5]), which were codon optimized for expression in E. coli using the IDT-codon-optimization tool. We decided to use pCDFDUET-1 as our backbone as it is designed for the coexpression of two target genes. The vector encodes two multiple cloning sites (MCS) each of which is preceded by a T7 promoter, lac operator, and ribosome binding site (rbs). Thereby only one plasmid is needed for the full pathway, which can be later be combined with the compartment systems (wiffleballs, EncA, SPD5) and the ncAA incorporation system.
For the design of the expression construct, we paired two genes together as follows: The first gene block consisted of TnaA and Fre (referred to as TAF) and was included into the MCS1. The second block consists of XiaI and TnaB (referred to as XTB) and was included into the MCS2. A ribosome binding site (RBS) was added between the two genes of each MCS.
To target the pathway to the compartments we added the SpyCatcher sequence to XiaI and the SnoopCatcher sequence to TnaA as these enzymes are the ones supposed to be incorporated into our compartment systems (wiffleball or SPD5). Two catcher constructs were designed, having attached the Catcher sequences either attached to the C-terminus or N-terminus of XiaI and TnaA . We added GS-Linkers between enzyme and catcher sequences. (Figure 5.2)

Figure 5.2: Schematic representation of the designed pathway constructs.Via restriction enzymes, gene fragments were cloned in the pCDFDuet-1 backbone. A ribosomal binding site was inserted between the two target genes of one MCS.

The following constructs were generated:

Plasmid name	Description
pCDFDuet-TAF-XTB	TnaA and Fre inserted in MCS1 XiaI and TnaB inserted in MCS2
pCDFDuet-sTAF-sXTB	TnaA with N-terminal SpyCatcher and Fre inserted in MCS1 XiaI with N-terminal SnoopCatcher and TnaB inserted in MCS2
pCDFDuet-TAFs-XTBs	TnaA with C-terminal SpyCatcher and Fre inserted in MCS1 XiaI with C-terminal SnoopCatcher and TnaB inserted in MCS2
pCDFDuet-sTAF-XTBs	TnaA with N-terminal SpyCatcher and Fre inserted in MCS1 XiaI with C-terminal SnoopCatcher and TnaB inserted in MCS2
pCDFDuet-TAFs-sXTB	TnaA with C-terminal SpyCatcher and Fre inserted in MCS1 XiaI with N-terminal SnoopCatcher and TnaB inserted in MCS2

Due to time constrains, we could not compare all the different constructs to each other to see which one works best. Because of that, we decided to use pCDF-DUET 1-snoopTnaA-Fre-spyXiaI-TnaB (Figure 1 B) for all of our assays, as this construct was the first one to be finished. We have used this construct for each assay.

Experimental design

Throughout our project, we have done several assays testing the production of indigo and indirubin under different conditions. To learn more about our procedures click here, to learn about the protocols that we have used or click here to get to our LabBook, where you can retrace our experiments.
All of our assays were performed in triplicates. This allowed us to create a mean value for each condition that we have tested. Furthermore, this significantly lowers the risk of human errors that could falsify our data and we can recognize outliers with much more precision. By making triplicates of all conditions, we can ascribe our final data with more weight.
Additionally, we have designed our assays as self-contained experiments. We planned our experiments to test an assumption, and the evidence to evaluate this assumption always emerged from one single assay. That way we could make sure, that effectivity differences stem from the tested conditions, and not because of a slightly different experimental environment.
By combining the use of triplicates together with a closed, self-contained design for each experiment testing an assumption, as underlying concepts for every single assay, we produced data that is as meaningful as possible.

Proof of the production of Indigo and Indirubin

Aim: Verify that our bacteria can produce indigo and indirubin by recombinant overexpression of XiaI, TnaA, TnaB and Fre in E. coli BL21(DE3).

Experimental setup: According to our Indigo/ Indirubin production protocol, we collected samples of our E. coli BL21 transformed with the pCDFDuet-1-sTAF-sXTB plasmid and tested for indigo and indirubin production via photometric measurements, NMR and mass spectrometry. A detailed protocol for each step of sample preparation can be found here.

Results: We successfully produced indigo and indirubin with our designed constructs, pCDFDuet-1 sTAF-sXTB, what we could clearly see by a colour change of liquid cultures 24 hours after induction (see Fig. 5.3). Following sample preparation, we measured the absorbance/fluorescence of both compounds at their respective absorption peaks (see fig. 5.6) for all our production assays.

Figure 5.3: Liquid cultures of E. coli BL21(DE3) transformed with pCDFDuet-sTAF-sXTB, showing the production of indigo and indirubin. A: Left erlenmeyer flasks with 200 mL culture of indigo producing bacteria, right erlenmeyer flasks with 200 mL culture of indirubin producing bacteria; B: Pellet of 1.5 mL bacterial culture after indigo production; C: Pellet of 1.5 mL bacterial culture after indirubin production. To shift the reaction from indigo to indirubin production cultures were supplemented with L-Cysteine.

Measuring proton NMR spectra of indirubin from our bioproduction shows small peaks at a shift of 7 – 7.6 ppm (see Fig. 5.4), which roughly fits the expected spectrum of indirubin from the literature [5.1]. However, the peaks are too weak to reveal any further information. This might be due to too high contamination with other cell substances, as we did not have the equipment to purify further our samples. The most prominent peak at ~4 ppm is most likely from water, and the peak at ~2.5 ppm from DMSO. To prevent the massive water contamination, we freeze-dried the indigo sample indigo before measuring. It might also be that the shift to indirubin production, by adding L-cysteine, was not efficient enough to produce detectable amounts. However, NMR clearly shows successful production of indirubin.

Figure 5.4: NMR proton-spectrum of indirubin from bioproduction using pCDFDuet-sTAF-sXTB E. coli BL21(DE3). Indirubin-producing bacteria were lysed, the non-soluble pellet was, after washing with KPI buffer, resuspended in deuterated DMSO for measurement. B shows a zoomed-in section of A.

Additionally, we have performed HLPC separation of indirubin from our samples and detected the molecular mass using mass spectrometry (Fig. 5.5).

Fig. 5.4: HPLC-followed mass spectrometry of Indirubin from own bioproduction using E. coli BL21 pCDFDuet-1 sTAF-sXTB.Indirubin-producing bacterial culture was spun down; the pellet was resuspended in KPI buffer, sonicated, spun down again and resuspended in deuterated DMSO for measurement. A: HPLC diagram, B: mass spectrometry diagram.

Calibration Curves using synthetic Indigo and Indirubin

Aim: To calculate the concentration of indigo and indirubin in our bioproduction photometrically, we needed standard curves to compare our measurements.

Experimental setup: Using synthetic indigo and indirubin (commercially available from Carl Roth), we set up different calibration curves according to the calibration-curves protocol. Absorbance of indigo and indirubin was measured at 540 nm. Additionally, fluorescence of indigo was measured at 612 nm excitation and 670 nm emission, using a plate reader. These experiments were performed in triplicates. For Indigo measurements, the experiments were repeated additionally with lower concentrations to obtain a better resolution for a possibly low indigo production.

Results:

Fig. 5.6: Calibration curves with synthetic Indigo and Indirubin. Absorbance at 540 nm was measured for indigo (A) and indirubin (C), as well as fluorescence of indigo with excitation at 612 nm and emission at 670 nm (B).

The calibration curves (see Fig. 5.6) were performed successfully, showing a linear increase of the relation of absorbance/fluorescence to the concentration for all tested conditions. using excel linear regression calculator allowed us to calculate the concentration of indigo/indirubin, by using the following formulas (with x = concentration (µM), y = absorbance (a.u.):

• Indirubin absorbance 540nm: 0.0059 * x = y
• Indigo absorbance 540nm: 0.0048 * x = y
• Indigo fluorescence: 758.95 * x = y

With this, we calculated the produced indigo and indirubin concentration for our samples by replacing the measured absorbance/fluorescence for each of the following production assays.
However, as the absorbance spectra of indigo and indirubin interfere with each other, only the indigo concentration could be calculated directly from the fluorescence with the corresponding formula. With the estimated indigo concentration, we could calculate the fitting absorbance that indigo must have at 540 nm, subtracting the calculated absorbance from the absorbance measurement at 540 nm and getting only the absorbance from indirubin. We could then enter the corresponding formula for this value to calculate the totalindirubin concentration in the samples.
We also considered that the IPTG concentration might influence our production assays and eventually falsify the measured of absorbance/fluorescence. Therefore, we tested different concentrations of IPTG (up to 2 mM) showing no effect on the photometric output.
One interesting observation we did was that for indigo concentrations of >30 µM, the fluorescence was not going higher in a linear matter but stagnated first and then went down again (see in the labbook in our raw data). At concentration of more than 30 µM, the solution of indigo was so dark assuming that too much light is absorbed and thereby less fluorescence can be measured. With this, we knew that samples need to be diluted before fluorescent measurement.

Optimisation of IPTG and L-Trp concentration for the pathway

Aim: Test different IPTG and L-Tryptophan (L-Trp) concentrations to find the most suitable one for the highest production of indigo/indirubin. IPTG is needed to induce the pathway promoter (T7-promoter), and L-Tryptophan is the substrate of the pathway.

General experimental setup: We used E. coli BL21 with pCDFDuet-1 sTAF-sXTB (exclusively containing the pathway) for the following experiments. All conditions were tested in triplicates, following the Indigo/Indirubin-production protocol and the Indigo/Indirubin-quantification protocol.

Experimental setup IPTG optimisation: We first induced samples with varying IPTG concentrations (from 0 to 0.15 mM final concentration). Following induction at OD=0.8 5 mM L-Trp was added temperature was set to 25°C for further culture. The tested IPTG concentration we used based on the experimental data published by Yin et al. [5.1]. Moreover, O₂ supply was also highlighted by Yin et al. to be an essential factor for yield increase. By using 500 mL Erlenmeyer flasks for the 200 mL bacterial cultures, we could enhance the aeration of the cultures and guarantee sufficient O₂ supply.

Results IPTG optimisation: Several experiments were done to test for optimal IPTG concentration (see labbook). In this preselected area, we found out that there is a tendency for 0.1 mM of IPTG in medium to work best to produce high yields. Nevertheless, 0.2 mM IPTG concentration seems to be nearly similarly effective. A more significant difference occurs 24h after induction (Figure 5.7). Production of indigo/indirubin was detected following induction with IPTG. Already 0.02 mM ITPG significantly leads to the production of about 2.5 µM indigo, 30 hours after induction. There were no major increase in indigo detected with longer expression time (3 µM after 45 hours).
A significant productionof indirubin of about 20 µM was determined 30 hours after induction. The highest concentration of indirubin (~50 µM) was obtained after 36 hours with no further increase with culture time. There was no major difference for the tested IPTG concentrations.
An interesting observation was that we already produced higher Indirubin amounts than Indigo while not switching the reaction by adding cysteine to the culture (see the comparison between A and B under the same conditions in fig.).
In comparison to our initial trial (see lab book fig.) in this experiment (see fig.), we could also see the positive effect of better O2 supply, like Yin et. al described [1]. In our initial experiment we only used 250 mL erlenmeyer flasks for culturing 200 mL bacteria, while in this assay we used 500 mL erlenmeyer flasks. Thereby the bacterial cultures had more contact surface with the air while shaking and should be able to take up more O2.
(see Fig. 5.7).

Fig. 5.7: Optimisation of IPTG-concentration for the Indigo/Indirubin pathway using E. coli BL21 pCDFDuet-1 sTAF-sXTB (pathway-enzymes with catcher). A: Indigo-concentration measured via fluorescence (excitation 612 nm, emission 670 nm), B: Indirubin-concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 25 °C after IPTG-induction at OD = 0.8, with L-Tryptophan-concentration of 5 mM. Photometric readout using a plate reader.

From this assay, we concluded to use 0.02 mM IPTG to induce the expression of pathway genes. This concentration is enough to get a reasonable amount of product and not too high to stress the cells. Moreover, higher IPTG concentration seem to have to further beneficial effect on the production of indigo/indirubin.

Experimental procedure L-Trp optimisation: For the L-Trp optimisation, we tested concentration between 0 and 10 mM L-Trp. The IPTG concentration of 0.02 mM was fixed for this experiment. The conditions (25 °C after L-Trp induction at OD = 0.8, 500 mL Erlenmeyer flasks for the 200 mL bacterial cultures) stayed the same as in the experiment before.

Results L-Trp: A concentration of L-Trp of 2.5 mM worked the best to produce high yields (~70 µM) of indirubin (see fig. 5.8 B). Higher concentrations (>7.5 mM) of L-Trp have no further positive effect; production of indirubin is comparable to that obtained without L-Trp supplementation (Fig…B, 0 vs 10 mM). Indigo production also increased following L-Trp supplementation. The highest difference was obtained compared to non-supplemented condition was obtained 18 to 30 hours after induction (Fig. 5.8 A).
Interestingly, cells already produced Indigo and Indirubin without L-Trp addition, probably due to cell’s own L-Trp consumption. However, with the addition of additional L-Trp, the cells produced more indigo/indirubin.
As observed in the IPTG optimisation assay, more Indirubin was produced than Indigo, ~4.5 fold (see the comparison between A and B under the same conditions in fig. 5.8). The highest production yield could also be improved compared to the IPTG optimisation assay, most likely because now most optimal tested IPTG and L-Trp concentrations worked together in at least one condition (see the best condition with 2.5 mM L.Trp and 0.02 mM IPTG concentration in fig. 5.8).

Fig. 5.8: Optimisation of L-Tryptophan-concentration for the Indigo/Indirubin pathway using E.coli BL21 pCDFDuet-1 sTAF-sXTB (pathway-enzymes with catcher). A: Indigo-concentration measured via fluorescence (extinction 612 nm, emission 670 nm), B: Indirubin-concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 25 °C after IPTG-induction at OD = 0.8, with IPTG-concentration of 0.02 mM. Photometric readout using a plate reader.

Conclusion & Outlook

We could verify and measure the production of Indigo and Indirubin, and calculate with the help of calibration curves, how much Indigo and Indirubin was produced. We could further test, which IPTG and L-Trp concentrations work best for our construct to reach optimal levels of Indigo of ~16 µM, and ~73 µM of Indirubin.
As we were more interested in the effects of compartment systems, we focused more on the combination of both together instead of further optimizing the pathway. Here still a lot could be done. E.g. a comparison of different Spy-Snoop-catcher positions could be tested and importantly also with a no-Catcher construct, to see the influence of the catchers on the pathway itself. Furthermore, as we saw hints in our first experiments and from the literature of the importance of O₂ supply for higher yield, different O2 concentrations could be tested. And many more screws could be turned to increase the yield. One last to mention, which would be very interesting for shifting the reaction to Indirubin, is, how much L-Cysteine is needed or optimal for this.

Trehalose pathway

Key findings:

• We successfully designed and cloned plasmids containing the two key enzymes of the Trehalose pathway
• We have shown in vivo trehalose production in E.coli

Plasmid Design

Aim: Our goal was to clone the genes otsA and otsB from the E. coli genome and insert them into pCDF-Duet-1. Later, we switched to the pET-Duet-1 backbone. 

Figure 6.1: Those are the plasmids we wanted to make. pCDF-Duet-1 was chosen as we already used it for the Indigo pathway. Both plasmids have two MCS (multiple cloning sites) and are regulated by a T7 promotor. 

Experimental setup: We first started by amplifying otsA and otsB out of the genome using olG22_001 (ATTACCATGGatgAGTCGTTTAGTCGTAGTATCTAACCGG) and olG22_002 (ATTAGCGGCCGCctaCGCAAGCTTTGGAA) for otsA and olG22_003 (ATTACATATGatgacagaaccgttaaccgaaacc) and olG22_004 (ATTACTCGAGttagatactacgactaaacgact) for otsB. 

Figure 6.2: PCR for amplifying otsA and otsB out of the genome of E. coli using primer olG22_003 and olG22_004 for otsB and olG22_001 and olG22_002 for otsA and different temperatures.

otsA and otsB were then subcloned into the subcloning vector pJET1.2 with the provided kit. 

Figure 6.3: Vector maps of the subcloned otsA and otsB. 

otsB was successfully subcloned and sequenced. otsA initially caused us difficulties but was also successfully subcloned. This was shown by colony PCR (Fig. 6.4) and subsequent sequencing. 

Figure 6.4: Colony PCR made from subcloning otsA into pJet1.2. with a length of 1566 between the sequencing primers.   

After the successful subcloning we tried to clone otsB into pCDF-Duet-1. This did not work, so we decided to use pET-Duet-1. We did this by digesting pJet1.2+ots and pET-Duet-1 with the restriction. enzymes XhoI and NdeI.  

Figure 6.5: An enzyme digest for pJet1.2 + otsB and pCDF-Duet-1 to prepare it to be ligated into the MCS2.  Both constructs were digested with NdeI and XhoI.

Which was then ligated by a T4 ligase. That was also successful as seen in the following colony PCR. 

Figure 6.6: Colony PCR made from the ligation made on 09.07.2022. Here are the colonies from pET-Duet-1 + otsB. Colonies with a band running at 800bp were positive. 

After also sequencing it, we would start to clone otsA into pET-Duet. That was also made by restriction enzyme digest. This time the restriction enzymes NcoI and NotI were used. 

Figure 6.7: Two gels with pJet1.2 + otsA and pET-Duet-1 + otsB were digested with NcoI and NotI to ligate otsA into the MCS2.

We ligated them by using a T4 ligase. And after a heat shock traffold into chemically competent TOP10 E. coli we also successfully cloned otsA into pET-Duet-1 +otsB. The sequencing showed also a positive result.

Figure 6.8: Colony PCR with pET-Duet-1 + otsA + otsB.  Colonies with a band running at 1500bp were positive. 

Results:
We have successfully cloned the plasmid pET-Duet-otsAB. For the later detection of trehalose, two glycerol stocks were created. First, E. coli BL21 transformed with pET-Duet. And secondly, E. coli BL21 transformed with pET-Duet-otsAB.

Quantity / Amount of Trehalose

Aim: To quantify the production of trehalose, a calibration curve was first performed using different trehalose concentrations.

Experimental setup: The calibration curve was performed with the Trehalose Assay Kit from Megazyme. The protocol for the microplate assay procedure was followed. Lysed E. coli BL21 served as sample. Different amounts of trehalose were added (0 µg, 1 µg, 2 µg, 4 µg, 6 µg, 8 µg and 10 µg).

Results:
The trehalose present is hydrolysed by the enzyme trehalase to 2 D-glucose. In the presence of ATP, the added hexokinase phosphorylates the D-glucose to glucose-6-phosphate. The enzyme glucose-6-phosphate dehydrogenase oxidises this to gluconate-6-phosphate in the presence of NADP+. This produces NADPH + H+. The presence of NADPH can be seen in the increase in absorbance at 340 nm and thus provides a detection for trehalose. The calibration curve was plotted using Excel (Microsoft Office 365 Pro Plus). The indicated error corresponds to the standard error of the mean (SEM). The calibration curve is shown in Figure 6.9.

Figure 6.9: Calibration curve for trehalose. Different amounts of trehalose were measured (0 µg, 1 µg, 2 µg, 4 µg, 6 µg, 8 µg and 10 µg). The error bars correspond to the standard error of the mean. Plotted using Excel 365.

Best growth conditions

Aim: We want to find out the best conditions for the growth of E. coli BL21.

Experimental setup: In order to find out the best growth conditions for E. coli BL21, we performed a growth curve in LB medium with different concentrations of glucose. Since trehalose is endogenously synthesised under osmotic stress, we decided to additionally test different salt concentrations. For this purpose, a 96-well plate was used and measured in the plate reader for over 17 hours.

Results:
Figure 6.10 shows the growth curve. The absorbance was plotted against time. The sample without glucose reached a plateau earliest after 180 minutes and continued to grow slowly thereafter. The samples with glucose reached their maximum OD after 330 to 360 minutes. The sample with 4% glucose reached the highest OD with 0.793. However, since a sample with 3% reached a lower OD with 0.762 than those with even less glucose, i.e., 1 and 2%, we decided against the sample with 4% glucose. Since the sample with 2% glucose showed slower growth with comparable maximum (0.79 for 1% glucose, 0.78 for 2% glucose), we decided to use 2% glucose for the production assays.
The samples with added salt showed less growth. Furthermore, they showed a partial fluctuation in their course. As expected, the samples with a higher salt concentration showed a lower OD.

Figure 6.10: Growth curve of E. coli BL21. Growth conditions in LB medium with different glucose concentrations (0-4%) were investigated. Different concentrations of NaCl (200-600 mM) were added to three samples with 2% glucose. The error corresponds to the standard deviation.

Improve of Trehalose

Aim: We want to detect the production of trehalose in the transformed E. coli BL21.

Experimental setup: The growth and trehalose production of the transformed E. coli BL21 were investigated in different media. We tested bacteria with the empty backbone pET-Duet and our plasmid pET-Duet-otsAB. We decided to use LB Medium with 2% Glocose as described. The bacteria were incubated at 37 degree and induced with 0.5 mM IPTG from an OD of 0.6. After that, a sample was taken every two hours. The last sample was taken after 12 hours. The samples were centrifuged, washed with 1 x PBS and resuspended in ddH2O. Lysis was carried out by sonication. The production assay was performed with the Trehalose Assay Kit from Megazyme. The protocol for the microplate assay procedure was followed.

Results:
Using the kit, trehalose production can be detected as described. The result is shown in Figure 6.11. The absorbance was plotted against time. The induced sample containing the plasmid pET-Duet-otsAB shows a maximum of 0.654 after 4 hours of incubation. After reaching the maximum, the absorbance subsequently decreases until after 12 hours it is only slightly above the other samples. The uninduced plasmid, as well as only the backbone (induced and uninduced) shows a significantly lower absorbance of about 0.32. This allowed us to show the production of trehalose. And thus, also confirm the functionality of our plasmid. We were able to detect a mass of 2 to 4 µg Trehalose.

Figure 6.11: Trehalose detection after productionassay. E. coli BL21 with the plasmids pET-Duet and pET-Duet-otsAB induced and uninduced were tested for the production of trehalose within a 12-hour incubation. Only the IPTG-induced bacteria with the plasmid pET-Duet-otsAB reach a maximum of 0.654 after 4 hours. The indicated error corresponds to the SEM.

Outlook

Our next step is to introduce catchers into our plasmid. This involves adding a spy catcher to otsA and a snoop catcher to otsB. After that, it is possible for the trehalose pathway to take place in our Wiffleball. After fine-tuning, we can hopefully achieve a higher efficiency for our production.

Genome Reduced Bacteria

---

Key findings:

• The growth of genome reduced strains is heavily limited at low temperatures, e.g. 18°C
• We can reliably incorporate non-canonical amino acids into sfGFP in different E. coli strains

Wiffleball expression in genome reduced strains

Aim: Testing the wiffleball system in different genome-reduced E. coli strains to check their specific behavior, their protein expression and metabolic burden. 

 

Experimental design: We tested the full and minimal wiffleball, besides the already tested BL21(DE3), MG1655 and C321∆exp, in ME5119, a derivate of MG1655, which has 15.8% of its genome deleted, and in MDS69 LowMut T7, a commercial genome reduced strain von Scarab Genomics with 20.32% deleted genome, with the same experimental conditions as in the previous induction experiment. The cells were treated with different inducer concentration reaching from 100µM up to 1mM IPTG and validated the expression with fluorescence microscopy for the foci formation, western blot for the protein expression level and growth assays for the metabolic burden. For the growth assay, the bacteria were induced at OD₆₀₀= 0.1 in a 96-well plate with all different inducer concentrations and incubated at 30°C, shaken in a plate reader for 20h. These conditions were also changed due to the lack of a cooling function to 18°C of the device, limited use of the plate reader and a restricted maximum growth to OD₆₀₀= 1.4 due to the 200µl well volume. For comparison between the strains, they are shown in the same plots and were measured on the same 96-well plate.

Results: For our goal of using genome reduced bacteria in bioproduction, we tested ME5119 and MDS69 LowMutT7. Both strains showed at 18°C a slower growth compared to MG1655 and BL21(DE3) Furthermore, the fluorescent expression of mVenus2 of ME5119 and LowMutT7 was also less than in all other strains. When the BMCs were expressed, we could not observe any foci in both strains and no bands of T1 could be detected on the western blots. Cell division and induced protein expression at 18°C seem to be a large burden for these strains. The experiments were repeated twice for ME5119.

 

The growth assay of ME5119 showed an overall trend: the higher the inducer concentration, the less the growth (Figure 8.1). In the growth comparison in the plate reader at 30°C, ME5119 grew remarkably slower than MG1655 and BL21. (Figure 8.2)

Figure 8.1: ME5119 growth in plate reader at 30°C with different inducer conditions. In legend: Wiffleball (IPTG (µM)/Doxycylcine (ng/ml))

Figure 8.2: Growth comparison in plate reader at 30°C between BL21(DE3), MG1655 and ME5119 with induction of 100µM IPTG 25ng/ml doxycycline

LowMut T7 was hardly influenced by IPTG concentrations lower than 400µM (Figure 1.29). For the minimal wiffleball induction, IPTG concentrations higher than 400µM had only slight impact on the growth (Figure 8.3). 

Figure 8.3: LowMut T7 growth in plate reader at 30°C with different inducer conditions,

By observing the growth curve of both genome reduced strains, we noticed a further difference between the two: ME5119 never reaches the OD₆₀₀ of 1, while LowMutT7 goes a bit further. At this point, we were curious to compare this strain to our well working BL21(DE3), so we repeated the same experiment under the same conditions, inducing both BL21(DE3) and LowMutT7 on the same 96-well plate. Surprisingly, as shown in Figure 1.30, no big difference in the growth course of the compared strains could be detected.

Figure 8.4: Growth comparison in plate reader at 30°C between BL21(DE3) and LowMut T7 with induction of 100µM IPTG and 25ng/ml doxycycline

For more data of the growth assays check our supplementary data in the labbook.

 

Testing longer induction times

Aim:In previous we saw that the growth of the genome reduced strain ME5119 was strongly reduced compared to MG1655 or BL21(DE3). This was accompanied by lower expression of Wiffleballs, we therefore test longer induction time trying to enhance the Wiffleball expression.                      

 

Experimental design:The already used induction conditions for the wiffleball induction at OD₆₀₀= 0.6-0.7 with incubation at 18°C, 200rpm, stayed the same. We took samples from the induced cultures from different time points, after 24h and 48h from ME5119. We examined the foci formation and protein expression by fluorescent microscopy and western blotting as in the wiffleball experiments before.

 

Results:

ME5119 showed low fluorescence under the microscope after 24h of incubation. No foci were observable. Even after 48 hours, the results stayed the same. We could not detect any T1 protein in the samples by western blot (data can be found in the supplementary).

Conclusion:

Longer incubation times had no effect on the overall protein expression on ME5119.

 

Adjusting incubation temperature

Aim: The expression and formation of the wiffleball under high temperatures results in larger amount of unfunctional inclusion bodies. Therefore, we always expressed the wiffleball at 18°C. Through an interview with Erik, from the Kerfeld Lab, we received the valuable information of incubating our cells after induction for 3h at 37°C, followed by the typical incubation temperature at 18°C for 16-21h. This should boost our protein expression, together with properly formed compartments.

 

Experimental setup: Following this new advice, we induced our BL21(DE3) and the genome reduced LowMut T7 at OD₆₀₀ = 0.6-0.7 with 100µM IPTG for the full and 400µM IPTG for the minimal wiffleball with additional 50ng/ml Doxycycline for mVenus2. The first incubation step for 3h incubation at 37°C, followed by 21h at 18°C, all shaken for 200rpm.

Samples were harvested after 24h of whole incubation time and then put back in the incubator.
In past experiments, the growth of LowMut T7 at 18°C was heavily decreased, so we enlarged the incubation time to a total of 48h to additionally investigate late protein expression and protein degradation.

The observation of the samples took place by the already used fluorescence microscopy method and western blot.

Unfortunately, the western blot will be detected after the wiki freeze due to the limited time.

 

Results:

Change the incubation conditions resulted in a remarkable change visible under the microscope. Almost every BL21(DE3) cell with the full wiffleball construct contained foci. Quantification for this will follow the wiki freeze, but by eye the estimation is a nearly two-fold increase of the number of foci. Some cells even showed a second polarised focus, pushing right toward the pole against the first focus. After 48h of induction, the number of secondary foci increased (Figure 8.5 A, red arrows). However, these results must be treated with care since we first need to exclude the possibility of inclusion bodies by western blot.                     
The minimal wiffleball also showed a positive effect. Bigger foci were observed at the poles next to an increased number of small foci distributed all over the cytoplasm. The longer induction time of 48h only resulted in a slight increase of foci in this construct.

Figure 8.5 : adjusted incubation temperature at 37°C for 3h followed by 21h at 18°C; full wiffleball induced with 100µM IPTG, minimal wiffleball induced with 400µM lPTG, mVenus2 induced with 50ng/ml; A BL21(DE3); B LowMut T7

In the previous experiment, LowMut T7 had big troubles growing at 18°C and expressing our non-native proteins. The new expression conditions changed their behavior drastically. The bacterial cultures were much denser than the previous induced LowMut T7 cultures. Also, under the microscope the change was drastic. With the full wiffleball, a lot of bacteria were showing the foci structures toward one cell pole. After 48h, the number of foci increased and were sometimes additionally localised in the middle of the cell.

The results seen in the minimal wiffleball construct were like the ones of BL21(DE3) with the same construct, after the 48h incubation (Figure 8.5 B).

 

Conclusion:

The adjusted incubation time seems to be a game changer for BL21(DE3) as well as for the genome reduced LowMut T7. Before approving this induction protocol, we need to exclude the nature of inclusion bodies by western blot. Due to the limited time, this experiment will be done after the wiki freeze and the data will be included in our presentation at the grand jamboree.

Testing our systems in genome-reduced strain (MDS69 LowMutT7)

Aim: Analysing the influence of using genome-reduced strains (MDS69 LowMutT7) as a chassis for our constructs in terms of product yield.

Experimental setup: We have done a production assay following protocol #023 with different conditions of genome-reduced strains as well as BL21. We have checked different comparable conditions between our two strains to see how much the reduced genome influences the yield of our pathway. We have used the constructs BL21 + pathway, BL21 + pathway + BMC, MDS69 LowMutT7 + pathway, MDS69 LowMutT7 + pathway + BMC.

Results: Unfortunately, we have seen, that transforming our plasmids into genome reduced strains results in lower yields compared to BL21 (see Fig. 8.8). Additionally, we did not see any mentionable production of indigo in the construct MDS69 LowMutT7 + pathway + BMC. We suspect that this is due to the low temperature (18°C) under which we perform protein expression. In different experiments regarding compartmentalization in these genome-reduced strains we have found out that they grow significantly better under higher temperatures. As a result, we would do follow-up experiments comparing different temperature conditions.

Figure 8.8: Indigo and indirubin concentration in the comparison between BL21 and genome reduced strain MDS69 LowMutT7

Incorporating p-Azf and p-Bpf into different E. coli strains

Incorporating p-AzF and p-BpF into different E. coli strains

So far, we used BL21 cells, which are optimal for protein expression and purification. Nonetheless, in the past years, special strains have been engineered with the specific purpose to improve the efficiency of ncAA incorporation. A notable example is the C321.ΔA.exp strain, in which all instances of the amber stop codon have been mutated into other stop codons, which allowed for the deletion of Release Factor 1. We, therefore, wanted to see if C321.ΔA.exp is better suited for ncAA incorporation. Additionally, we were curious to see if genome reduction would improve ncAA incorporation efficiency, given that such cells would have more resources to produce our heterologous proteins. For this reason, we selected a genome-reduced version of E. coli called ME5119 derived from ME5000 [7.2][7.3].

Aim: Test if ncAA incorporation is possible and/or works better in different E. coli strains. 

Experimental setup:We obtained C321.ΔA.exp from Addgene (Bacterial strain #49018) and ME5119 from Tokyo Metropolitan University Group. We used BL21(DE3) as control, since we had already proven incorporation of the ncAA pAzF in this strain (Figure 8.6). All strains were co-transformed with the following combinations of plasmids:

• pEVOL-pAzF and pTrc99a-sfGFP WT
• pEVOL-pBpF and pTrc99a-sfGFP WT
• pEVOL-pAzF and pTrc99a-sfGFP Y74
• pEVOL-pBpF and pTrc99a-sfGFP Y74

• pBAD33 and pTrc99a-sfGFP Y74
• pBAD33 and pTrc99a

A single colony from each co-transformation was grown overnight at 37°C in LB medium supplemented with ampicillin and chloramphenicol. Overnight cultures were diluted and grown to OD600 of 0.4 at 37°C before induction. Samples were induced with 1mM arabinose and 1 mM of the unnatural amino acid (p-AzF or p-BpF) were added, except in the control, where the ncAA was omitted. One hour after arabinose induction (needed for the orthogonal tRNA synthetase) the samples were induced with 200 µM IPTG (for sfGFP expression) and growth was continued for additional 2.5 h at 37°C.

The incorporation of the ncAA was indirectly measured by fluorescence, which was quantified by flow cytometry with the CyAn ADP Analyzer from Beckman Coulter. Samples were prepared by mixing 10 µL of the liquid bacterial cultures with 990 µL DPBS. 

Results:

All three E. coli strains showed incorporation of the two ncAAs into sfGFP Y74 when the corresponding synthetase was present. As expected, in all cases the fluorescence intensity was lower than that of wild-type sfGFP (Figure 2). 

For the incorporation of p-AzF, C321.ΔA.exp shows the highest efficiency; however, in this strain GFP fluorescence is detectable also in samples where either the synthetase or p-AzF are not present, suggesting that the amber stop codon is being suppressed with another canonical amino acid (Figure 8.7). Interestingly, the genome-reduced ME5119 strain showed similar fluorescence to BL21 when expressing mutated sfGFP Y74, but much lower fluorescence when expressing wild-type sfGFP (Figure 8.7).

Figure 8.6: Bar graph showing the Median Fluorescence Intensity (MFI) of the indicated E. coli strains transformed with pEVOL-pAzF and pTrc99a coding for the indicated GFP variant induced with 1 mM arabinose. GFP fluorescence was measured by flow cytometry. 500 µM IPTG were used to induce sfGFP expression. 1 mM p-AzF were added to all samples labelled with the “+”.

E. coli C321.ΔA.exp also showed the highest efficiency of incorporation of p-BpF, closely followed by BL21 (Figure 3). Again C321.ΔA.exp exhibited some fluorescence where none was expected, suggesting suppression of the amber stop codon with a canonical amino acid (Figure 5).

In contrast to the results shown Figure 8.7, in this experiment BL21 cells showed higher fluorescence when expressing wild-type sfGFP than C321.ΔA.exp (Figure 8.8).

Figure 8.7:Bar graph showing the Median Fluorescence Intensity (MFI) of the indicated E. coli strains transformed with pEVOL-pBpF and pTrc99a coding for the indicated GFP variant induced with 1 mM arabinose. GFP fluorescence was measured by flow cytometry. 500 µM IPTG were used to induce sfGFP expression. 1 mM p-BpF were added to all samples labelled with the “+”.

Full system - combining Compartments, ncAAs and pathways

---

Incorporation of the Indigo/Indirubin-pathway in the full Wiffleball

1st run: IPTG-optimisation of pathway-BMC construct
Aim: We wanted to determine if our key hypothesis is correct: “Combining a metabolic pathway with a compartment system and thereby bringing the enzymes in closer proximity enhances the yield of that pathway”. Therefore, we tested our Indigo/Indirubin pathway along with a construct that contains pathway + bacterial microcompartments (Wiffleball) under the same conditions to compare both production yields. We did not know yet under which conditions the pathway + BMC construct would work best, so we tested this one for different IPTG concentrations (both pathway and BMC are IPTG-induced).
By measuring the growth of these samples parallel to the production assay, we wanted to see how much of a metabolic burden the BMC expression is on top of the pathway.

Experimental setup: To test the different conditions, we followed the Indigo/Indirubin-production protocol and the Indigo/Indirubin-quantification protocol. All conditions were tested in triplicates. We used E. coli BL21 with pCDFDuet-1 sTAF-sXTB (contains exclusively the pathway), and E. coli BL21 with pCDFDuet-1 sTAF-sXTB + Wiffleball (contains pathway and Wiffleball).
All samples were kept under the same conditions: 200 mL bacterial cultures in 500 mL Erlenmeyer flasks at 18 °C after induction at OD 0.8, with L-Tryptophan concentration of 2.5 mM (regarding results from L-Trp optimisation assay) and L-Cysteine concentration of 3 mM. IPTG concentration was the only parameter changing between the conditions; we tested 0, 0.04, 0.1 and 0.2 mM in the medium.
The temperature was switched from earlier experiments from 25°C to 18°C because Wiffleball seemed to form better at lower temperatures (see here). L-Cysteine was added to shift the reaction towards enhanced Indirubin production, as it was published by Han et al. []. This shift was not only proven to work, but also made OD measurements easier when making growth curves, as Indigo absorbs strongly at OD₆₀₀.
For the growth curve, 200 µL samples were collected from the culture flasks after L-Trp-supply and diluted to start below OD 0.1. They were incubated in a plate reader for ~ 14h at 30°C with an L-tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM and IPTG-concentrations according to the legend. We measured the OD₆₀₀ in the plate reader every 10 minutes. The plate lid was treated with a triton-ethanol mixture to prevent it from fogging. This experiment was performed using triplicates.

Results: Our first run indeed showed what we stated in our hypothesis. The pathway + BMC-construct induced with 0.04 or 0.1 mM of IPTG generated higher production yields than the pathway-only construct induced with 0.1 mM IPTG (see Fig.7.1). The Indigo-concentration was more than double the size, and the one for Indirubin concentration was even around one-third more in the pathway + BMC construct than in the pathway-only construct.

Figure 7.1 : Comparison between BL21 pCDFDuet-1 sTAF-sXTB with and without Wiffleball, testing different IPTG-concentrations for the strain with both plasmids. The pathway was shifted towards Indirubin production through L-Cysteine addition. A: Schematics of the used plasmids; B: Indigo-concentration measured via fluorescence (extinction 612 nm, emission 670 nm); C: Indirubin concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 18 °C after IPTG-induction at OD = 0.8, with L-Tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM, and with IPTG concentration shown in the legend. Photometric readout using a plate reader.

Nevertheless, the general production of the pathway-only construct was way lower (3 - 4fold) than in the L-Trp production assay. These assays are comparable because the same conditions were applied except for the additional L-Cysteine-supply in this assay. Maybe something went wrong in this experiment with the pathway-only construct. Alternatively, L-Cysteine is the reason for the negative effect because of its inhibiting effect on bacterial growth, as it was shown by Han et al. []. However, the Wiffleball must reverse this effect, for which we have no proof.
To really get a meaningful statement about whether our hypothesis is proven right, we repeated the assay two times more (see here and here).
The condition with 0.2 mM IPTG induced stagnated at half the end-concentration of that one with 0.04 mM, so it worked poorly. Probably, the cells are too stressed with such a high number of enzymes produced. The results from the according growth curve (see Fig. 7.2) also indicated that the pathway-BMC constructs with 0 mM and 0.2 mM IPTG-induction grew less fast and reached earlier saturation than the constructs induced with 0.04 or 0.1 mM IPTG. That speaks for cell stress through metabolic burden. Unfortunately, we could not compare the growth and, thereby, the metabolic burden of the pathway BMC construct with only the pathway condition. This one did not grow at all due to the addition of the wrong antibiotics in these samples.

Figure 7.2: Growth-curve of relevant Indigo/Indirubin production-strains, with inducer. Using 200 µL samples of 200 mL cultures in 500 mL Erlenmeyer-flasks (took them after L-Trp supply), at 30 °C, with L-tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM and IPTG-concentrations according to legend. OD₆₀₀ was measured in the plate reader.

We also saw that the L-Cysteine worked as expected and shifted the production nearly completely to Indirubin, while the Indigo concentration stayed near zero over time. As in earlier experiments, nearly no Indigo or Indirubin was produced without induction with IPTG, which affirms that the T7-promoter shows nearly no leakiness.
An interesting observation we made during this assay was that the production starts later compared to earlier assays with only the pathway tested and does not stagnate 48h after induction. This might be because we reduced the temperature from 25°C to 18°C, and thereby the bacteria grew less, and the metabolism was slower.

2nd run: with addition of an empty plasmid control and extended measurement

Aim: Repeating the comparison between the pathway-only construct and the pathway + BMC construct for the first time, we wanted to get validation for our results from the first comparison assay (see here), as that one was not fully meaningful due to the general lower production of Indigo and Indirubin. We also repeated the growth-curves to see the effect of the conditions on cell metabolic burden.

Experimental setup: All conditions were kept the same for the production assay as in the initial comparison (see here). However, this time, we tested an additional control containing the pathway as well as pTrc99A, which also contains ampicillin resistance like the BMC plasmid. This way, we could see potential (probably negative) impacts the second antibiotic to select has. We also wanted to see how much higher the Indigo and Indirubin concentrations can go and when they stagnate. Therefore, we expanded the measurement time from 48 h to 72 h after induction, measuring every six hours between 24 h –72 h after induction.
This time, one growth curve was obtained using a plate reader again, but another via OD₆₀₀ measurements manually at the Nanodrop 2000. For that one, we took 1 mL samples from the 200 mL bacterial cultures every half an hour, then every hour for ~17 h, and then measured at specific key points again until the production assay ended. That way, our growth-curve conditions were the same as the culture conditions.
We tried to simulate these conditions as close as possible in the plate reader. After L-Trp addition to the culture conditions, 200 µL samples were taken from there and incubated at room temperature (~20 °C), with L-tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM and IPTG-concentrations varying between conditions. Every 5 min OD₆₀₀ measurement was done for 25 h. The temperature was not strictly at 18 °C because that plate reader could not cool down so fast.

Results: Unfortunately, we could not reproduce the data from the first comparison between the pathway-only construct and the pathway + BMC construct. Other than the first assay suggested, the pathway-only condition performed better or equal to any of the pathway + BMC conditions (see Fig. 7.3).

Figure 7.3: Comparison between BL21 pCDFDuet-1 sTAF-sXTB with and without Wiffleball and pTrc99A plasmid-control. The pathway was shifted towards Indirubin production through L-Cysteine addition. A: Schematics of the used plasmids; B: Indigo-concentration measured via fluorescence (extinction 612 nm, emission 670 nm); C: Indirubin-concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 18 °C after IPTG-induction at OD = 0.8, with L-Tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM, and with IPTG concentration shown in the legend. Photometric readout using a plate reader.

As this time the pathway-alone production yield was in the same range as it was for the L-Trp production assay under the same conditions, we trust more in the results of this assay more than the one before. Still, a second repetition will be made to get clarity.
However, an interesting aspect is that the pathway + BMC construct produced higher yields than the newly added pathway + pTrc99A control. Except for one outlier at 72 h, where that one jumped up in concentration very far, probably due to a measurement error regarding the huge standard deviation.
This could already be an improvement because the addition of the Wiffleball compensated for the negative effect of the empty plasmid on the production of Indigo and Indirubin. So it could be, without the second antibiotic, that the BMC constructs would perform even better because that probably is why the control with the empty plasmid performed worst. This will be tested in the last repetition of this comparison.
For this observation, the growth curve results (Figure 7.4) were also quite interesting.

Figure 7.4: Growth curve of relevant Indigo/Indirubin production strains, with an inducer, starting measuring after induction. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at ~18 °C after induction at OD = 0.8, with L-Trp concentration of 2.5 mM, L-Cysteine concentration of 3 mM and IPTG concentrations according to legend. A: OD600 was manually measured with the Nanodrop2000; B: Measured OD₆₀₀ nm in plate reader.

They showed that bacteria grew less with the pTrc99A plasmid on top of the pathway plasmid, most likely due to the additional antibiotic. However, the BMC-containing constructs grew even worse the more target proteins they produced (when more IPTG was added; see Fig. 7.4). Still, they produced more Indigo and Indirubin than the construct with the pTrc99A plasmid.
As all constructs saturated at the same OD, in the end, it could be that the constructs with both plasmids need more time to produce in a good manner, but the potential is there that they still reach the same maximum.
Like in the first of these comparisons, where it was also tested for the optimal IPTG concentration, 0.04 and 0.1 mM of IPTG in the medium were the best for the highest product yield. Between these two, no clear tendencies could be shown.
Due to fluctuations, it is hard to say, but stagnation could be reached after 42 – 48 h. The subsequent measurement will be tested another time until 72 h after induction.

3rd run: including antibiotics control + and incoorporating the ncAAs

Aim: Repeating the comparison between the pathway-only construct and the pathway + BMC construct for the second time, we wanted to get validation for our results from the first comparison assays (see here and here), as they showed different outcomes.
We furthermore wanted to test, if we could express our pathway (sTAF-sXTB) together with mutated Wiffleball and noncanonical amino acids (ncAA), in this case, the pAzF and still produce Indigo/Indirubin then. Without the pAzF, which will be introduced into the mutated sites in the Wiffleball, they would not assemble.

Experimental setup: All conditions were kept the same for the production assay as in the other two comparisons. However, on top of this time, we also tested the triple-construct. Therefore, we added for that condition the noncanonical amino acid pAzF into that culture at OD 0.4 (41.24 mg) and, at the same time, induced the plasmid for the tRNA-synthetase (for the ncAA to integrate) with 150.38 µL Arabinose.

Results: With the second repetition of the comparison between pathway-only and pathway with Wiffleball, we can relatively surely say that the pathway-only construct under optimal conditions worked better than with Wiffleball, falsifying the results from the first comparison assay (see Fig. 7.5 ).

Figure 7.5: Comparison between BL21 pCDFDuet-1 sTAF-sXTB with and without Wiffleball, pTrc99A plasmid-control and ncAAs. The pathway was shifted towards Indirubin production through L-Cysteine addition. A: Schematics of the used plasmids; B: Indigo concentration measured via fluorescence (extinction 612 nm, emission 670 nm); C: Indirubin concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 18 °C after IPTG-induction at OD = 0.8, with L-Tryptophan concentration of 2.5 mM, L-Cysteine concentration of 3 mM, ncAA-amount of 41.24 mg pAzF and with IPTG concentration shown in legend. Photometric readout using a plate reader.

The pathway-only construct with 0.1 mM IPTG-concentration produced ~ 2x as much Indigo/Indirubin as the best pathway + BMC construct. Two out of three assays with identical conditions speak for that, while in the one showing higher yield with Wiffleball, the yield of the pathway-only construct was very low. So most likely, something went wrong in that assay, inhibiting the pathway-only construct.
With this assay, we could also show that the triple-construct containing pathway, Wiffleball and ncAAs, produced still Indigo and Indirubin, and interestingly not even less than the constructs containing pathway + BMC or pathway induced with 0.04 mM (Figure 7.5B). Unfortunately, not enough time were left to perform a control to check, if the ncAA was integrated into the Wiffleball or if the Wiffleball form without it. But western blots from the triple-construct under similar conditions have shown, that without the ncAA the Wiffleball did not form, while with ncAAs, they do (see here). So most likely in this assay, the ncAAs are also integrated.
Interestingly, this time more Indigo and a little less Indirubin were produced compared to other assays with similar conditions. There is also a discrepancy between the relative Indigo and Indirubin yield of the “pathway + BMC with only Spectinomycin at 0.04 mM” condition compared to the pathway-only construct with 0.04 mM IPTG. Because for Indigo, it seemed like the “pathway + BMC with only Spectinomycin” construct performed better than the pathway-only when same IPTG-concentration is applied. For Indirubin, that was not truly the case.
It was shown again that when the pTrc99A plasmid is added to the pathway construct, it performs the worst, most likely due to the second antibiotic applied.

Conclusion & Outlook

In total the assay was performed three times, the last two times showing quite similar results except that in the last one the shift of the reaction towards Indirubin was not that strong, therefore more Indigo and less Indirubin was produced in comparison to the second trial. We neglected the first trial in the further evaluation of the assay, because there is too much to suggest that the pathway-only construct grew badly there.
Evaluating the 2nd and last trial only, they showed, that the pathway + BMC construct initially produced worse than the pathway-only construct. But in the condition when both constructs had to face both antibiotics and thereby the same metabolic burden, the pathway + BMC construct suddenly performed better than the pathway-only construct (containing empty pTrc99A with Amp-resistance).
This indicated to us, that the Wiffleball have a positive impact on the production yield but hindered by the enlarged metabolic burden due to a second plasmid and antibiotic. The further investigation of this phenomena by leaving away the second antibiotic in the pathway + BMC condition truly enhanced the yield of the production, pushing it on about the same (for Indirubin-production) or even clearly higher levels than the pathway-only construct (without second plasmid!). One could say that the second plasmid without the antibiotic selection could be kicked out by the bacteria already during the measurement. But this would be a very fast deletion, and it would not explain, why the yield went higher than without a second plasmid. So, we can be relatively sure that the second plasmid stayed in the cells during the measurement.
To expand on these discoveries, one could try to get rid of the need of a second antibiotic, e.g. by using the technology of split enzymes enabling to use one antibiotic for selecting multiple plasmids [7.1]. In any case, it should also be checked via cryo-electron microscopy or similar devise, if the Wiffleball are actually forming. We couldn’t show that yet (TEM-results were not yet available). What we could identify with western blots was, that the enzymes were catched to the T1 proteins of the Wiffleball by the SpySnoop Catcher system we used.
The assay also showed us, that the triple construct, including pathway, mutated Wiffleball and ncAAs is able to produce Indigo and Indirubin, and that the addition of the ncAAs and their corresponding tRNA on an extra plasmid does not influence the production yield negatively compared to the pathway + BMC construct. We have to say, that in this case we didn’t test all needed controls (without ncAA addition, without induction, etc.), but as we performed western blots another time with similar conditions, which showed, that the triple system worked (see here), we could suspect, that in this case also all three systems worked together. To verify this assumption, another production assay could be done in the future testing all necessary controls for the triple construct.

Optimisation of inducing conditions through promoter change

Aim: To get more control over how much of pathway enzymes in comparison to Wiffleball are produced and to optimise the ratio and thereby potentially the production yield, we changed the BMC-promoter from IPTG-inducible PLac to a doxy-inducible tet-promoter. The pathway was still induced by IPTG (T7 promoter). This would allow us to induce the two systems indivually.

Experimental setup: We tested four different doxy-concentrations (10, 20, 50 and 70 ng/ mL) while the IPTG-concentration stayed at 0.04 mM, which proved to be the optimal value from the experiments before. This we did using the sTAF-sXTB pathway together with the Wiffleball with the new tet-promoter. As controls for comparison, we also used the pathway-only construct (sTAF-sXTB) with 0.1 mM IPTG and the pathway + BMC (old promoter) construct, both induced with 0.04 mM IPTG. The other production conditions stayed the same as in the assays before. Both controls were performed in triplicates, the doxy-induced constructs only one sample per conditions because the other ones grew too bad.

Results: The conditions with the new promoter didn’t produce much at all, no matter how much doxy was given (Figure 7.8).

Figure 7.8: Comparison between pathway + BMC constructs, either with both IPTG-inducible promoter or both different promoters. Wiffleball have either pLac promoter (IPTG-inducible) or tet-promoter (doxy-inducible), different doxycyclin concentrations are tested with consistent IPTG concentration of 0.04 mM. The pathway was shifted towards Indirubin production through L-Cysteine addition. A: Schematics of the used plasmids; B: Indigo-concentration measured via fluorescence (extinction 612 nm, emission 670 nm); C: Indirubin-concentration measured via absorbance at 540 nm. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, at 18 °C after IPTG-induction at OD = 0.8, with L-Tryptophan concentration of 2.5 mM and L-Cysteine concentration of 3 mM. Photometric readout using a plate reader.

As we found out, most likely there was something designed wrongly in the plasmid and because of that, the tet-resistance was not working correctly in these conditions. Doxy supply then damaged the bacteria, they suffered and because of that grew and produced not much. So, no conclusion can be drawn yet, if a promoter change enhances the production of pathway + BMC constructs or not. We will repeat this experiment, therefore.

Aim:

Test of the metabolic burden of the used plasmids on the bacteria via growth curves

Aim: We tested all constructs, that were used in production assays on their metabolic burden, which already occurs through the plasmid alone, uninduced. These constructs contain one or more of the following:

• The pathway (pCDFDuet-1 sTAF-sXTB, spectinomycin)
• Wiffleball (full whiffleball with T1, T2 and T3, ampicillin)
• ncAAs (pAzF and the corresponding tRNA+tRNA synthetase, encoded on pEVOL-pAzF, chloramphenicol)

Experimental setup: The constructs were tested in growth curves by measuring OD₆₀₀ manually with Nanodrop2000 and the other time with a plate reader. All conditions were tested in triplicates.
For measurement with Nanodrop 2000, 200 mL cultures of all tested conditions were grown in 500 mL Erlenmeyer flasks at 37 °C. Every half an hour, three 1 mL samples were taken out for OD₆₀₀ measurement, for a total time of 9 h.
As preparation for the plate reader measurement, 200 mL cultures of all tested conditions were grown in 500 mL Erlenmeyer flasks at 37 °C. However, at OD 0.8, small samples of all conditions were diluted to reach OD 0.1 in the final 200 µL samples, which were then taken into a 96-well plate. The samples were incubated at 20 °C for ~ 16.5 h with OD₆₀₀ measurements every 10 min.

Results: Both manually and automated growth readouts showed that the pathway as well as the pathway + BMC plasmid, both uninduced, do not affect cell growth, as they do not differ from each other and also not from the control without the plasmids (this control only in manually growth curve tested) (see fig. and fig. ).

Figure 7.9: Growth curve of relevant Indigo/Indirubin production strains, without inducer. Production with 200 mL cultures in 500 mL Erlenmeyer flasks at 37 °C. OD₆₀₀ was manually measured with the Nanodrop2000.

Figure 7.10 : Growth-curve of relevant Indigo/Indirubin production-strains, without inducer. “sTAF-sXTB + BMC with only spec” received spectinomycin as antibiotics alone and no additional for the second plasmid. Production with 200 mL cultures in 500 mL Erlenmeyer-flasks, then transfers of 200 µL in the plate for measurement at 20°C when synchronised cultures. OD₆₀₀ was measured in the plate reader.

For the plate reader growth curve, the condition with pathway + BMC plasmid grew better, but not very much, which could also be due to the slightly higher starting OD of that condition. So, it seems like these plasmids themselves are no metabolic burden to the cells, which is interesting, as the condition containing both plasmids must face two antibiotics instead of one.
Nevertheless, one apparent deviation from the growth of the other conditions has been seen. The triple construct containing all three plasmids, for pathway, Wiffleball and ncAAs, grew less fast and reached a significantly earlier plateau. That could be because, in this condition, the bacteria had to face another antibiotic on top, this time a rather harsh one, chloramphenicol. The combination of all antibiotics together, or simply the massive amount of additional DNA, could trigger the growth decline.

Upscaling the Indigo/Indirubin-BMC system

---

Fig. 5.20: Bacterial pellet after harvesting and centrifuging samples from the bioreactor. We can see three different subcultures; one forms a very dark blue surface on the inside of the flask, the second one is located towards the wall of the flask and has a lighter but still blue colour, the last subculture is between the other two cultures and shows only very little indigo.

Aim: To check whether our constructs are able to produce our target molecules on a greater scale, we made first efforts towards upscaling by performing production assays with the construct “BL21 pathway + BMC” in a bioreactor as well as in a fed batch experiment. An enormous problem in the upscaling of biological concepts is the high amount of division cycles that bacteria have to go through. With these experiments, we wanted to find out, how our constructs behave under these conditions.

Experimental setup: (L-Cysteine will be used to shift the production of our samples towards Indirubin)

We started by inoculating an overnight culture of 4 mL with BL21 containing the pathway plasmid as well as the wiffleball. On the next day, the culture was transferred into a flask filled with 1L LB incl. The antibiotics spectinomycin and ampicillin. While this sample grew to an OD₆₀₀ of 1.0, our supervisor Dr. Pavel Salavei prepared the bioreactor (Eppendorf BioFlo 415 bioprocess control station 14 L stainless steel vessel). All steps regarding the bioreactor were performed by following protocol #052.

We took a sample of 3 mL that will be required for our fed batch before induction. We then preceded sample collection by following our protocol for our ordinary production assays. Additionally, we measured the OD600 of our samples every six hours (see fig. 5.19), and we only took 100 µl samples for our later measurements due to the increased cell density in the bioreactor.

Fed batch:

Our fed batch was set up using the 3 mL sample that we have taken from the bioreactor before induction. The sample was centrifuged and then washed with PBS to remove the medium of the bioreactor from the cells. Afterwards, the pellet was resuspended with LB. Finally, we have diluted the sample until it reached an OD₆₀₀ of 0.8. Then, the sample was induced using IPTG. To quantify this assay, we followed our protocol for production assays. This setup of the fed batch was deliberately chosen to make it comparable to our previous assays as medium, amount of inducer, substrate etc. stay the same here, only the amount of division cycles has been adjusted.

Results:

Our assay has shown very positive results, as the increased amount of division cycles did not appear to have a significant effect on the yield of our product. During our fed batch assay, we have reached concentration of around 15 µM Indirubin (see figure 5.18), which is almost identical to the yield of the same construct that we have tested in a previous assay (see figure 5.15) (Optimisation of BMC-pathway interplay by BMC promoter switch to tet-promoter with test of different induction-concentration).

Figure. 5.18: Concentrations of indigo and indirubin in bioreactor and fed batch assays. The high concentration of the bioreactor in figure B at 18 hours is likely to be a measurement inaccuracy.

Fig. 5.17: Graphs resembling the conditions within the bioreactor

Fig. 5.19: Growth curve in bioreactor

Unfortunately, the shift towards Indirubin was not successful for our sample in the bioreactor. Although we added L-Cysteine, mostly Indigo as produced. We suspect that we did not use the correct amount of L-Cysteine to achieve this shift. Our approach to solve this issue would be to test our construct again, in a multitude of small bioreactors, experimenting with different L-Cysteine concentrations until achieving the expected shift. However, this would be beyond the scope of this research. Nevertheless, we were able to produce usable amounts of Indigo in this assay, which can also be seen as a great success! During our measurement, we have reached concentrations of around 10 µM Indigo. By scaling this up, we could produce around 0,262g of Indigo in our 10L bioreactor! Although this is not an industrially relevant amount, there is still much space for improvement. We can make adjustments to our constructs, as well as to the setup of the bioreactor (see fig. 5.17) to achieve higher yields or to produce significant amounts of Indirubin. After harvesting, we have realised, that our bacteria have apparently formed three subcultures (see fig.5.20) which all appear to produce Indigo with different efficiency. This is another problem that would need further revision and analysis. Especially Indirubin might be very promising in a bioreactor, as its usage as a drug does not require incredibly high concentrations. Once we are able to achieve a strict shift towards Indirubin, our production might be of relevance on a greater scale.

References

---

[1.1] Kirst, H., Ferlez, B. H., Lindner, S. N., Cotton, C. A. R., Bar-Even, A., & Kerfeld, C. A. (2022). Toward a glycyl radical enzyme containing synthetic bacterial microcompartment to produce pyruvate from formate and acetate. https://doi.org/10.1073/pnas.2116871119/-/DCSupplemental

[1.2] Parsons et al, „Synthesis of Empty Bacterial Microcompartments, Directed Organelle Protein Incorporation, and Evidence of Filament-Associated Organelle Movement“, Molecular Cell, VOLUME 38, ISSUE 2, P305-315, APRIL 2010

[1.3] Cesleet al, “Variety of size and form of GRM2 bacterial microcompartment particles”, Protein Science, 30: 1035– 1043, march 2021

[2.1] au, Y. H., Giessen, T. W., Altenburg, W. J., & Silver, P. A. (2018). Prokaryotic nanocompartments form synthetic organelles in a eukaryote. Nature Communications, 9(1). https://doi.org/10.1038/s41467-018-03768-x

[3.1] M. C. Huber et al., “Designer amphiphilic proteins as building blocks for the intracellular formation of organelle-like compartments,” Nat Mater, vol. 14, no. 1, pp. 125–132, Jan. 2015, doi: 10.1038/nmat4118.

[3.2] S. Alberti, “Phase separation in biology,” 2017, doi: 10.1111/pbi.1280.

[4.1] Yeh, FL., Tung, L., Chang, TH. (2016). Detection of Protein–Protein Interaction Within an RNA–Protein Complex Via Unnatural-Amino-Acid-Mediated Photochemical Crosslinking. In: Lin, RJ. (eds) RNA-Protein Complexes and Interactions. Methods in Molecular Biology, vol 1421. Humana Press, New York, NY. https://doi.org/10.1007/978-1-4939-3591-8_15

[4.2] C. M. Joiner, M. E. Breen, and A. K. Mapp, “Electron-deficient p-benzoyl-l-phenylalanine derivatives increase covalent chemical capture yields for protein–protein interactions,” Protein Sci., vol. 28, no. 6, pp. 1163–1170, 2019, doi: 10.1002/pro.3621.

[4.3] CPyMOL Molecular Graphics System, Version 2.5.3, Schrödinger, LLC.

[4.4] https://shigen.nig.ac.jp/ecoli/strain/resource/longDeletion/detail/20120

[4.5] https://resourcedb.nbrp.jp/searchResourceDetailAction.do?resourceId=28467239&organismId=18&categoryId=6

[5.1] PubChem, https://pubchem.ncbi.nlm.nih.gov/compound/10177, Computed by PubChem 2.1 (PubChem release 2021.05.07).

[5.2] Yin H. et.al., "Efficient Bioproduction of Indigo and Indirubin by Optimizing a Novel Terpenoid Cyclase XiaI in Escherichia coli", ACS OMEGA, vol.6, 20569−20576, 2021, https://doi.org/10.1021/acsomega.1c02679.

[7.1] Palanisamy N. et.al., "Split intein-mediated selection of cells containing two plasmids using a single antibiotic", Nature Communications, vol.10, 4967, 2019, https://doi.org/10.1038/s41467-019-12911-1.

[7.2] https://shigen.nig.ac.jp/ecoli/strain/resource/longDeletion/detail/20120 (opened 10/10/2022)

[7.3] https://resourcedb.nbrp.jp/searchResourceDetailAction.do?resourceId=28467239&organismId=18&ca tegoryId=6 (opened 10/10/2022)