Results

Introduction

To bring our biosensing tattoo to life, we needed to develop and integrate three essential components: Nanocages for localized and controlled melanin synthesis, engineered tyrosinases for efficient pigment production, and a MESA receptor that selectively detects a target biomarker to trigger downstream signaling. We performed a series of experiments to evaluate each of these systems individually, identifying key challenges, and to guide further engineering toward a functional biosensing platform designed to fit into everyday life.

MESA

A central goal of our project was to engineer a synthetic receptor system that could sense an external protein ligand and translate that binding event into a precise cellular response. For this, we utilized the Modular Extracellular Sensor Architecture (MESA), which provides a powerful framework for engineering custom cellular receptors. A typical MESA receptor consists of two distinct transmembrane proteins, each containing three core functional domains: a Ligand-Binding Domain (LBD), a Transmembrane Domain (TMD), and an Intracellular Effector Domain (IED). The system’s activation is initiated when a specific ligand binds to the LBDs of both MESA chains, inducing receptor dimerization.

For our MESA receptor, we evaluated two different variants. The first one being a membrane-bound receptor that can be activated by extracellular ligands, such as mCherry which we used as a proof-of-concept. The second one is a soluble receptor that detects ligands inside of the cell. To be able to visualize receptor function in mammalian cells, we used fluorescent signals, such as Blue Fluorescent Protein (BFP) and Yellow Fluorescent Protein (YFP). We transfected HEK293T cells with our constructs 6-8 h post seeding, induced with the respective ligand approximately 12 h later and measured consequently within 24 h.

Membrane-Bound MESA

This section details the assembly and testing of a membrane-bound MESA system. We present the results of our initial characterization, which aimed to confirm the central hypothesis: that the addition of the mCherry ligand would trigger a visible translocation of a BFP-tagged tTA, eventually activating YFP reporter expression. The experiments herein evaluate the functionality of different receptor variants and reveal the specific engineering challenges associated with achieving tight, ligand-dependent control in a complex cellular environment.

Fig. 1.1 / Overview membrane-bound MESA. Illustrated signal cascade of mCherry ligand binding to the membrane-bound MESA receptor, thus triggering the reconstruction of split TEV, which subsequently cleaves of a BFP-tagged tTA that activated reporter gene (YFP) expression in nucleus.

Translocation of BFP Signal

The goal of this experiment was to demonstrate ligand-dependent activation by visualizing the translocation of a BFP signal from the membrane to the cytosol as a proof-of-concept. For this, we constructed a membrane-anchored MESA system where the receptor’s external domain is designed to bind an mCherry ligand. The internal signaling mechanism relies on the reconstitution of split TEV protease, where we tested two different variants of the N-terminal TEV fragment, Ntev (75S) and Ntev (75E). For qulitative readout, the C-terminal fragment was fused to BFP, preceeded by a TEV cleavage site, and purified mCherry protein was used as a ligand to trigger receptor activation.

In the absence of the mCherry ligand (- mCherry), the MESA system should remain in its default “off” state. In this configuration, the Ctev-BFP fusion protein is expected to be stably anchored to the plasma membrane. Microscopy analysis should reveal a BFP signal exclusively outlining the cell periphery, with no significant fluorescence detected within the cytosol.

Upon addition of the mCherry ligand (+ mCherry), the system should undergo activation. The binding of mCherry is designed to induce specific dimerization of the receptor chains. This dimerization, in turn, reconstitutes the split TEV protease, triggering a proteolytic cleavage event at the cutting site (ENLYFQ*M) that releases the TEV-cleavable receptor cargo domain from its membrane tether. Consequently, a successful activation is defined by a distinct translocation of the BFP signal from the membranous outline of the cell to a concentrated signal within the cytosol. This visible shift provides direct visual confirmation of the entire signaling cascade, from ligand binding to transcription factor release.

This experiment provides a direct visual assay for the functionality of our membrane-bound MESA receptors. A ligand-dependent shift in BFP localization would confirm successful design, from ligand binding and receptor dimerization to the final step of translocation of the BFP reporter from the plasma membrane to the cytosol.

Fig. 1.2 / Testing the MESA receptor with and without 200 mg/mL mCherry induction via BFP localisation in HEK293T cells, 24 h post induction, 40x. Functional validation of the membrane-bound MESA reporter construct localisation with a fluorescence microscopy-based analysis in comparison to controls.

As expected, transfection with the untagged Ntev variants (75E and 75S) alone resulted in no fluorescence, regardless if the mCherry ligand was present in the culture media. Transfection with Ctev-BFP yielded the expected blue signal after 24 h, confirming the construct location in the cell. However, contrary to our design, the Ctev-BFP signal was localized in the cytosol rather than the membrane. This suggests, that the construct is either not properly anchored to the membrane or is cleaved and translocated independently of the MESA system.

When we co-transfected cells with either Ntev variant and the Ctev-BFP construct, we observed a significantly brighter BFP signal compared to the Ctev-BFP alone control. This indicates a functional interaction between the Ntev and Ctev fragments, leading to the reconstitution of the TEV protease and the cleavage and release of Ctev-BFP. A comparison of the two Ntev variants, 75S and 75E, revealed no substantial difference in the intensity or localization of the BFP signal, with both exhibiting similarly high constitutive activity. Although it should be mentioned here, that these results are primary qualitative, so for further hypotheses a more quantitative method would be needed. This indicates that the single amino acid mutation in the N-terminus does not have the pronounced impact on regulating dimerization that we had anticipated. Nevertheless, critically, the addition of the mCherry ligand did not result in a discernible difference in the location of the BFP signal when compared to the non-induced (- mCherry) wells. In both conditions, the signal appeared within the cytosol, not showing the ligand-dependent shift we anticipated.

The absence of a ligand-induced response indicates fundamental shortcomings in the current receptor design. The constitutive localization of Ctev-BFP in the cytosol, observed regardless of mCherry addition, demonstrates that the system exhibits significant baseline signaling. This high background activity can be attributed to several potential, non-mutually exclusive mechanisms. A primary explanation is non-specific dimerization, where the receptor domains spontaneously associate without ligand binding, leading to unintended TEV protease reconstitution. The mere physical proximity of both tethered fragments within the same membrane compartment may be sufficient for spontaneous reconstitution, effectively bypassing the requirement for ligand-induced dimerization. This hypothesis is directly supported by the bright cytosolic BFP signal in non-induced controls. Furthermore, faulty membrane anchoring could be responsible, as the Ctev-BFP construct may not be reliably retained at the plasma membrane, leading to its default localization within the cell cytosol. The notably intensified BFP signal in the Ntev and Ctev-BFP co-transfections confirms that protease-mediated cleavage is functional, yet it is dominated by a high background signal.

Signalling Cascade Validation

To test a more integrated system, we combined the membrane-bound MESA receptors with a Ctev-tTA-BFP construct and the tTA-inducible YFP reporter. The goal was to achieve mCherry-dependent YFP expression.

Fig. 1.3 / Testing the MESA receptor with and without 200 mg/mL mCherry induction via BFP localisation in HEK293T cells, 24 h post induction, 40x. Functional validation of the membrane-bound MESA reporter construct transcription activation with a fluorescence microscopy-based analysis in comparison to controls.

We observed a clear BFP signal in cells transfected with our Ctev-tTA-BFP construct, confirming its successful expression. Furthermore, in the + mCherry condition for this sample, we observed a corresponding signal in the YFP channel. Given the confirmed leakiness of the YFP reporter and the phenomenon of spectral bleed-through observed in our controls, we interpret this YFP signal as a crosstalk from the intense BFP signal.

The YFP-only control showed signal in the YFP channel, confirming the intrinsic leakiness of the reporter. The slight signal in the BFP channel for this sample confirms the presence of spectral bleed-through from the bright YFP into the BFP filter set, a critical technical artifact for interpreting all other results.

We detected BFP signal in the Ctev-tTA-BFP with each Ntev variant, confirming the constitutive expression of the constructs with some slight crosstalk to the YFP channel.

In the samples containing YFP and either Ntev variant alone, we observed no significant YFP expression, which was unexpected in regards to the previously visible background leakiness. The YFP reporter’s expression is designed to be strictly dependent on the tTA transcription factor, which is absent in these transfection conditions. Theoretically, a clean outcome from these controls would successfully verify that the Ntev constructs themselves do not non-specifically activate gene expression, and the YFP reporter remains silent without its specific transactivator.

Next, the CTev-tTA-BFP + YFP showed signal in both BFP and YFP channels even without mCherry. This indicates that the Ctev-tTA-BFP construct is constitutively active, likely through basal cleavage by endogenous proteases, releasing tTA-BFP to drive YFP expression. The slight increase with mCherry suggests a possible minor ligand-induced effect, but it is superimposed on a high background.

Cells transfected with Ntev + Ctev-tTA-BFP + YFP produced a strong signal in all channels. This indicates that the addition of the functional Ntev fragment leads to very efficient, but ligand-independent, reconstitution and cleavage. The system is constitutively “on,” and the signal is too saturated to discern any specific mCherry induction. If the Ntev (75S) was used in this combination, it showed no signal. This demonstrates that this specific Ntev mutation may have abolishes the spontaneous dimerization and reconstitution with Ctev that plagued the original Ntev construct, which is in contrast to existing literature (Dolberg et al., 2021) (Dolberg et al., 2021).

The next step would be to screen for a new Ntev mutant that finds a middle ground: one that has low enough affinity to prevent spontaneous dimerization (unlike the original) but high enough affinity to successfully dimerize in the presence of the mCherry ligand (unlike the current mutant). The consistent observation of BFP and YFP signals in the absence of mCherry across multiple combinations (e.g., Ctev-tTA-BFP + YFP, Ntev + Ctev-tTA-BFP + YFP) confirms a high level of basal signaling. This suggests that the tTA transcription factor is being unspecifically released, likely through a combination of endogenous protease activity and spontaneous receptor dimerization.

Soluble MESA

To establish a robust and rapid proof-of-concept for our signalling system, we prioritized the development of a soluble, rapamycin-inducible MESA platform, based on the feedback we received from Prof. Leonnard. Since some biomarkers, like progesterone, are soluble, we also focused on a highly controlable and reproducible system to validate the core MESA mechanism: ligand-induced dimerization and subsequent release of a transcription factor. Therefore, the goal of this workstream was to demonstrate that our soluble MESA design successfully dimerizes upon rapamycin addition, leading to the cleavage and nuclear translocation of a membrane-tethered transcription factor, ultimately activating gene expression.

Our system is built upon a modular, three-component design centered on the inducible reconstitution of a TEV protease. We engineered two soluble receptor halves: one fused to the H75E variant of the N-terminal fragment of TEV protease (Ntev) and the other to the C-terminal fragment (Ctev). Each fragment should be inactive on its own. The Ntev fragment is fused to the FRB domain, while the Ctev fragment is fused to the FKBP domain. The addition of rapamycin induces the heterodimerization of FRB (FKBP-rapamycin binding domain of mTOR) and FKBP (FK506-binding protein), which brings the Ntev and Ctev fragments into proximity, reconstituting a functional TEV protease. To monitor the successful dimerization and activation of the soluble receptors, we designed a membrane-anchored transcription factor, which comprises - among others - a TEV protease cleavage site and a tTA fused to BFP to allow for fluorescent tracking of its localization.

The system’s functional output is driven by the Tet-Off mechanism. The released tTA-BFP translocates to the nucleus and binds to a tetracycline-responsive element (TRE), activating the expression of a YFP reporter gene.

Fig. 1.4 / Overview soluble MESA Illustrated signal cascade of rapamycin ligand binding to the soluble MESA receptor, thus triggering the reconstruction of split TEV, which subsequently cleaves of a membrane bound BFP-tagged tTA by the TEV cleavage site and activating the reporter gene (YFP) expression in nucleus.

Fig. 1.5 / Testing the soluble MESA receptor with and without 0.1 µM rapamycin induction via YFP and BFP induction in HEK293T cells, 12 h post induction, 10x. Functional validation of the MESA reporter construct with a fluorescence microscopy-based analysis in comparison to controls.

Fig. 1.6 / Testing the soluble MESA receptor with and without 0.1 µM rapamycin induction via YFP and BFP induction in HEK293T cells, 24 h post induction, 40x. Functional validation of the MESA reporter construct with a fluorescence microscopy-based analysis in comparison to controls.

As anticipated, cells transfected with an empty filler plasmid as a negative control and with the untagged Ntev-Ctev construct alone showed no fluorescence in the BFP or YFP channels under any condition. This confirmed that the background signal is low and that the fluorescent proteins themselves are not expressed without the appropriate genetic components.

The tTA-BFP construct variant showed no detectable BFP signal. We consulted with Prof. Leonnard, the inventor of the MESA system, regarding our constitutive activity. Looking at the tTA-BFP + YFP construct, he suggested that a high plasmid dose of the “dangling tTA” could lead to saturation of the Endoplasmic Reticulum (ER) and potentially causing mislocalization that could contribute to a constitutive signal. However, this specific model of ER saturation appears inconsistent with our key control experiment. If high-level expression and ER saturation were the primary issue, we would expect to see a strong, mislocalized BFP signal in the tTA-BFP only transfection, where the protein is expressed without any other components. Instead, we observe no BFP signal in this condition. As a second alternative, Prof. Leonnard suggested that either the expression level of the single construct is very low, or more likely, the BFP signal is quenched, improperly folded when the protein is anchored to the membrane or proteolytically cleaved in a non-specific manner by endogenous cellular proteases, bypassing our designed control mechanism.

Cells transfected with the YFP reporter plasmid alone exhibited a faint signal at 12 h in comparison to 24 h post transfection, as visible in Fig. 1.5 and 1.6. This indicates that the tetracycline-responsive promoter driving the YFP gene is leaky. In a perfectly orthogonal system, YFP expression should be completely dependent on the presence of the tTA transcription factor. This observed leakiness means that the reporter plasmid itself is capable of high basal expression, which could mask weaker, induced signals and complicate the interpretation of future experiments. Consequently, addressing this promoter leakiness, either by sourcing a more stable Tet-Off reporter plasmid or by fine-tuning its regulation, needs to be a prerequisite for any subsequent quantitative studies.

Building on the characterization of individual components, we tested the functionality of our system by transfecting key combinations of plasmids.

Consistent with our earlier transfection, the combination tTA-BFP + NTev-CTev showed no detectable fluorescence in either the BFP or YFP channels. However, the continued absence of BFP signal suggests a persistent issue with the detection or stability of the membrane-tethered tTA-BFP protein in our experimental setup.
Cells transfected with the YFP reporter and the soluble NTev-CTev showed a time-dependent increase in YFP fluorescence, starting as a faint signal after 12 h and becoming strong by 24 h. This aligns with the data from the YFP-only control, confirming that the observed YFP expression is due entirely to the inherent leakiness of the reporter plasmid and is not influenced by the presence of the Ntev and Ctev proteins.

We observed apparently strong BFP and YFP expression in both induced and non-induced wells of the YFP + tTA-BFP combination. The intensity of the fluorescence was significantly higher than in the YFP transfection alone. This might demonstrate that the tTA-BFP protein is being expressed and is functionally active in driving YFP expression, adding on the basal expression of the reporter gene. However, its activity is constitutive and not dependent on rapamycin-induced dimerization of Ntev-Ctev. On top of that, the localisation of the BFP and YFP signal appeared to be identical, even when measured at the respective different wavelengths (BFP: 381-445, YFP: 514-529). This might be due to the leakiness of YFP, bleeding through into the BFP channel, drowning out the weaker BFP signal by this crosstalk.

Analysis of the complete, rapamycin-inducible system (YFP + tTA-BFP + Ntev-Ctev) revealed a fluorescence output identical in both intensity and pattern to the constitutively active YFP + tTA-BFP combination. The presence of the Ntev and Ctev constructs did not alter the output, and no rapamycin-dependence was observed. This indicates that the system’s activity is driven by a dominant, ligand-independent mechanism.

A critical finding was the precise spatial overlap between the BFP and YFP signals in these samples. Instead of the expected pattern, where BFP from the released nuclear tTA would be distinct from cytosolic YFP, the two signals were colocalized with each other. This identical pattern is not consistent with the intended biological distribution and points to two concurrent issues.

Firstly, the high-level of YFP expression in the YFP-only control demonstrates the leaky reporter.

Secondly, the identical localisation of the BFP and YFP signals points to a technical phenomen known as fluorescence bleed-through (crosstalk). The exceptionally bright YFP signal, amplified by the constitutively active tTA, is so intense that its emission bleeds into the BFP detection channel. This creates a false-positive BFP signal that perfectly mirrors the YFP pattern, thereby masking the true localization of the tTA-BFP protein and complicating the biological interpretation. One solution might be recloning one of the proteins to a non-overlapping variant.

In conclusion, the system is constitutively active due to non-specific tTA cleavage, and the resulting intense fluorescence creates spectral artifacts that prevent an accurate assessment of protein localization and system function. For subsequent tests, its possible to replace the tTA transcription factor with an alternative output module. A promising candidate are our engineered tyrosinases, which produce visible melanin. This system is mechanistically distinct, potentially more robust to leaky expression, and provides a direct, easily quantifiable output (pigment accumulation) that is not based on fluorescence, thereby avoiding the spectral bleed-through issues encountered in this study.

Future MESA Experiments

The experimental work conducted has provided a clear roadmap for the next stages of development. To advance the MESA system towards a robust and reliable platform, future work should focus on the following key strategies:

• Replace the tTA/Tet-Off System: Substitute the problematic tTA transcription factor and the leaky YFP reporter with a more robust output module. A primary candidate is our engineered tyrosinase, which produces a visible melanin pigment. This provides a direct, quantifiable, and non-fluorescent readout that avoids issues of spectral bleed-through and transcriptional leakiness.
• Optimize protein stability: Address the potential misfolding and constitutive cleavage by systematically optimizing the transmembrane domains and linker sequences.
• Test with therapeutically relevant ligands: Once a stable, low-background system is established with mCherry, the LBDs should be replaced with domains that bind to therapeutically relevant targets (e.g., disease-specific cell surface antigens) to demonstrate the platform’s translational potential.

Encapsulins

For our tattoo to change contrast, we needed to localize melanin synthesis within controllable nanocages, since high melanin concentrations in the cytoplasm are toxic for cells. Therefore, we designed and tested several encapsulin variants to assess their ability to assemble and load tyrosinases. We investigated three different encapsulin systems:

1. “Naked” encapsulins that should self-assemble spontaneously into nanocages
2. Pro-encapsulins with LBT-15 domains that sterically prevent nanocage assembly until cleaved, allowing regulated nanocage formation
3. Tyrosinase-encapsulin fusions, where nanocage assembly results in encapsulated tyrosinase complexes.
   
   Additionally, we tested the efficiency of tyrosinase loading into the nanocages formed by “naked” and pro-encapsulins via a cargo signal.
   The basic experimental setup for assembly validation was the same for all encapsulin constructs. We transfected HEK293T cells with the respective plasmids, lysed the cells 24 hours post transfection, and analyzed the lysates using a Native PAGE. First, we imaged the gels using fluorescence and then stained them with Coomassie Blue to visualize protein content. Since assembled nanocages are substantially bigger than endogenous proteins found in HEK cells, we expected them to appear as distinct, high-molecular-weight bands near the top of the gel, even though the assembled nanocages have a higher molecular weight than the range of the ladder (Sigmund et al., 2023). To confirm that these bands represent our encapsulins, we checked for the presence of a corresponding fluorescent signal, since all of our encapsulin constructs carry the fluorescent tag eUnaG. In addition to experimental validation, we also used computational modeling to predict the assembly behavior of our encapsulin constructs.

”Naked” encapsulins

For our “naked” encapsulins, we tested each variant, Mx, Qt, and Tm, individually to prove their ability to spontaneously self-assemble into nanocages. To confirm that their assembly does not depend on proteolytic cleavage, we also tested co-transfections with a plasmid encoding a TEV protease. Identical bands for “naked” encapsulins and their co-transfected samples on the Native PAGE at the expected molecular weight, accompanied by a fluorescent signal, indicate an identical assembly behaviour, prooving that the “naked” encapsulins indeed self-assemble autonomously. Lastly, we also tested co-transfections of the Mx and Tm encapsulins with their respective cargo plasmids with and without TEV to evaluate successful loading of the nanocages. The Qt cargo construct could not be included in this analysis, since cloning for this plasmid was not successful.
Fig. 2.1 shows the Native PAGE containing the mentioned “naked” encapsulin combinations in the following order:

• “naked” encapsulin alone
• “naked” encapsulin co-transfected with TEV protease
• “naked” encapsulin co-transfected with its respective tyrosinase cargo plasmid for Mx and Tm
  
  On the left, the fluorescent imaging of the gel is shown, on the right is the same gel after Coomassie Blue staining.
  
  
  Fig. 2.2 / Fluorescent imgaging (left) and Coomassie Blue staining (right) of Native PAGE for “naked” encapsulins. Assembly was tested dependent on the presence of TEV protease as well as tyrosinase cargo. The white arrow highlights the expected height of the assembled nanocages.

As expected, the “naked” Mx encapsulin shows distinct bands in the upper region of the gel, togehter with a strong fluorescent signal, independent of the presence of TEV protease. The identical migration patterns with (lane 2) and without (lane 1) the presence of TEV protease indicates that these constructs self-assemble autonomously. Interestingly, co-expression with the cargo plasmid (lane 3) resulted in no distinct band on the gel, neither for fluorescence imaging nor Coomassie Blue staining. However, the fluorescent image shows clear fluorescence signals in the loading pockets of the gel for all three Mx combinations, suggesting that the assembled nanocages might have been to big to fully migrate into the gel. Therefore, it is possible that loading the nanocages increased their molecular weight, leading to them being remained trapped in the loading pockets. In contrast, non-assembled Mx encapsulin monomers would be expected to appear as a distinct fluorescent band at the bottom of the gel at a molecular weight of 51 kDa, which was not observed, meaning the nanocages likely assembled.

Surprisingly, the Qt encapsulin combinations showed different results. The “naked” Qt encapsulin (lane 4) showed a strong fluorescent signal in the loading pocket, whereas co-transfection with TEV protease (lane 5) resulted in a much weaker fluorescent signal in the loading pocket. Additionally, we could not detect distinct bands in the expected area after Comassie Blue staining for Qt alone or the co-transfection sample with TEV protease. Since nanocages formed by Qt encapsulins are bigger than Mx nanocages Sigmund et al. (2023b), it’s not surprising that the assembled nanocages remained in the loading pockets and did not migrate into the gel. However, the reduced fluorescence intensity for the Qt encapsulin in presence of TEV protease contradicts our hypothesis that the “naked” encapsulins are not being affected by TEV protease. Importantly, we could not detect any fluorescent signal at the bottom of the gel, where we would expect non-assembled Qt monomers at 51 kDa. This suggests that assembly still occured, and that the difference in fluorescence intensity is being caused by a reduced expression of the “naked” Qt encapsulin during co-transfection with TEV protease, as we did not quantify expression levels.

Lastly, the “naked” Tm encapsulins showed strong fluorescent bands across all three conditions, namely encapsulin alone (lane 6), co-transfection with TEV protease (lane 7), and co-transfection with cargo (lane 8) as well as distinct bands visible after Comassie Blue staining. These results indicate that the “naked” Tm encapsulin spontaneously assembled independently of TEV protease, and that the addition of cargo did not interfere with nanocage formation. However, we can’t confirm that the nanocages were successfully loaded with tyrosinases based solely on this data. Overall, the results further support the hypothesis that bigger nanocages do not migrate into the gel, while Tm forms the smallest nanocages among our three encapsulin variants and was therefore able to migrate more efficiently into the gel.

Pro-encapsulins

For our pro-encapsulins, we also tested each variant individually, namely Mx, Qt, and Tm, to confirm that they do not spontaneously self-assemble into nanocages in the absence of TEV protease. To verify that their assembly is dependent on proteolytic cleavage, we again performed co-transfections with a plasmid encoding a TEV protease. Distinct bands in the expected area, accompanied by a fluorescent signal for samples containing pro-encapsulins and TEV, but not pro-encapsulin samples alone, demonstrate that nanocage assembly only occurs in the presence of TEV, indicating that the LBT-15 domains effectively prevent assembly and function as intended. Just like for the “naked” encasulins, we additionally tested co-transfections of the Mx and Tm pro-encapsulins with their respective cargo plasmids, both with and without an additional plasmid encoding TEV protease. That way we can evaluate successful loading of the nanocages dependent on TEV protease, the presence of tyrosinase cargo, as well as the presence of both. Again, Qt cargo construct could not be included.
Image 2.2 shows the Native PAGE for our pro-encapsulins in the following order:

• pro-encapsulin alone
• pro-encapsulin co-transfected with TEV protease
• pro-encapsulin co-transfected with its respective cargo plasmid for Mx and Tm
  pro-encapsulin co-transfected with both its respective cargo plasmid and TEV protease Mx and Tm
  
  On the left, the fluorescent imaging of the gel is shown, on the right the same gel after Coomassie Blue staining.
  
  
  Fig. 2.2 / Fluorescent imgaging (left) and Coomassie Blue staining (right) of Native PAGE for pro-encapsulins. Assembly was tested dependent on the presence of TEV protease and tyrosinase cargo. The white arrow highlights the expected height of the assembled nanocages.

As expected, the Mx pro-encapsulin alone (Lane 1) showed no distinct band or fluorescent signal at the top of the gel, but instead displayed signals at the bottom, which is likely caused by non-assembled monomers. This indicates that assembly did not occur, and that the LBT-15 domains function as intended. In contrast, co-transfection with TEV protease (lane 2) resulted in a faint band near the top of the gel accompanied by a fluorescent signal at the same height and in the loading pocket. Therefore, the TEV protease was probably able to cleave off the LBT-15 domains, resulting in naocage assembly. Interestingly, co-transfection of the Mx pro-encapsulin with the cargo plasmid (lane 3) produced fluorescent signals both at the bottom of the gel, indicating non-assembled monomers at 60 kDa, and in the loading pocket. This could indicate that some nanocages assembled without LBT-15 cleavage. However, since we could not detect assembly in the sample without cargo, it is possible that the uncleaved pro-encapsulins formed non-specific aggregates with the tyrosianse cargo, that did not migrate into the gel. Another possibility is that cargo loading might enhance nanocage assembly, which we would have to examine in further experiments. Finally, the co-transfection of Mx pro-encapsulin with both TEV protease and cargo (lane 4) also resulted in a fluorescent signal in the loading pocket, but without any additional signal at the bottom of the gel, mirroring the results we observed for the “naked” Mx encapsulin with cargo. This suggests that the nanocages were succesfully loaded and assembled, but were unable to migrate into the gel due to their increased molecular weight.

Interestingly, the Tm pro-encapsulins behaved diffferently under the same conditions. Without TEV or cargo (lane 5), the Tm pro-encapsulin did not show the expected fluorescent signal at the bottom of the gel at 60 kDa, which would indicate non-assembled monomers. Instead, we observed fluorescence in the loading pocket. While this might suggest that the pro-encapsulins assembled without proteolytic cleavage, the signal differs from the distinct fluorescent band we saw for “naked” Tm encapsulin. However, since these two samples were not run on the same gel, this might be due to minor differences in gel properties. Importantly, the combination of the Tm pro-encapsulin with TEV protease (lane 6) produced a clear fluorescent band at the expected height, resembling the band we observed for “naked” Tm, indicating that the presence of TEV protease affects pro-encapsulin assemby. Therefore, it could be that Tm pro-encapsulins with uncleaved LBT-15 domains form non-specific aggregates that remain in the loading pocket, while proteolytic cleavage leads to nanocage assembly. This hypothesis is supported by the fluorescent smear visible in lane 5, which could be the result of various non-specific aggregates. Co-transfection of the Tm pro-encapsulin with the respective cargo plasmid (lane 7) led to a faint fluorescent signal in the loading pocket, implying again nanocage assembly or possible aggregate formation. When co-transfected with both TEV protease and cargo plasmids (lane 8), we could observe a faint fluorescent signal in the loading pocket, along with a blurred fluorescent band at the expected height. This result is similar to the one observed for “naked” Tm and Tm pro-encapsulin with TEV (lane 6). However, the fluorescent signal is much weaker, indicating lower expression levels. While this could suggest nanocage assembly with successful loading, further experimental validation is needed. Additionally, the Tm samples without TEV protease (lanes 5 and 7) also showed a faint fluorescent band in the middle of the gel. Since we neither expect non-assembled monomers nor fully formed nanocages at this molecular weight, it further supports the possibility of pro-encapsulins forming non-specific structures when LBT-15 domains are uncleaved.

Unfortunately, the sample containing only the Qt pro-encapsulin (lane 1) did not show any fluorescent band, indicating no expression. This is very likely, since the cells in the corresponding well showed reduced viability and changes in morphology under the microscope, and had reduced protein expression levels, as seen on the Coomassie Blue stained gel. In comparison, the co-transfection with TEV protease (lane 2) resulted in a clear fluorescent signal located in the loading pocket, similar to the one we observed for “naked” Qt encapsulin without TEV. Therefore, we can assume that cleaved Qt pro-encapsulins assemble into nanocages that are too big to migrate into the gel, although we cannot evaluate if this depends on the presence of TEV protease.

Tyrosinase-encapsulin fusions

For our tyrosinase-encapsulin fusion constructs, we only tested the Mx, Qt, and Tm variants individually, as they do not require separate loading via a cargo construct. Distinct bands at the expected molecular weight, accompanied by a corresponding fluorescent signal, similar to those observed for our “naked” encapsulins, indicate that our fusion constructs successfully assembled into nanocages, and that the N-terminal tyrosinase fusion does not interfere with encapsulin assembly.

In image 2.3, a Native PAGE containing our tyrosinase-encapsulin fusions, as well as controls, is shown under fluorescent imaging, as well as after Coomassie Blue staining.

Fig. 2.3 / Fluorescent imgaging (left) and Coomassie Blue staining (right) of Native PAGE for tyrosinase-encapsulin fusions and controls. Assembly was tested without the addition of TEV protease.

For our tyrosinase-fusion encapsulins, we detected fluorescent signals in the loading pockets of the gel for Mx (lane 3) and Qt (lane 4), indicating successful nanocage assembly with the N-terminal tyrosinase fusion. Since the fusion encapsulins are likely to have an increased molecular weight, it’s not surprising that they remained in the loading pockets of the gel. Still, this result was unexpected, since our modeled fusion constructs predicted unsuccessful assembly. In contrast, our fusion with Tm (lane 5) showed no fluorescent signal, indicating both unsuccessful nanocage assembly and failed expression of the fusion construct. Lastly, we also included three control samples. Firstly, we tested the lysate of non-transfected HEK cells (lane 6). As expected, we did not detect any fluorescent signal, which indicates our cells did not express any fluorescent tags that could interfere with our results. The second negative control (lane 7) consisted of lysate from cells transfected with our backbone only. As expected, we did not detect any fluorscent signal. Finally, we also included lysate of cells transfected with a plasmid containing the Mx encapsulin without any fluorescent tag (lane 8), which also showed no fluorescent signal and therefore confirmed the specificity of our fluorescence detection.

Future encapsulin experiments

Even though our results already highlight some promising encapsulin variants, further experiments are required to identify the most efficent system for localized melanin sysnthesis and controllable nanocage assembly. Therefore, we have already planned additional experiments, which could not be performed yet due to time limitations.

Co-transfection of Mx pro-encapsulins with rapamycin-inducible MESA and tyrosinase cargo

For our tattoo to function as a true biosensor, we need to link nanocage assembly to the detection of a specific biomarker. Therefore, we plan to co-transfect HEK cells with Mx pro-encapsulins, their corresponding tyrosinase cargo, and our rapamycin-inducible MESA receptor. Upon addition of rapamycin, we expect the split TEV protease subunits, which are fused to the MESA domains, to reconstitute into a functional enzyme, which then cleaves off the LBT-15 domains of the pro-encapsulins, triggering nanocage assembly. If our system works as intended, the newly assembled nanocages will be loaded with tyrosinases, thereby localizing and controlling melanin synthesis. To verify nanocage assembly, we will analyze cell lysates using Native PAGE, as already established in our previous experiments. Upon confirming assembly, we will proceed with testing tyrosinase activity assay to evaluate enzymatic function within the nanocages, demonstrating controlled melanin synthesis in response to the target signal.

Tyrosinase assay to prove loading and functionality of tyrosinases in encapsulins

To verify that our encapsulated tyrosinases retain enzymatic activity after loading, as well as fused to encapsulins, we plan to perform in-gel activity assays. Specifically, we will analyze Mx and Qt tyrosinase-encapsulin fusion constructs, as well as lysates from HEK cells co-transfected with Mx pro-encapsulins, the rapamycin-inducible MESA receptor, and their corresponding tyrosinase cargo. Following Native PAGE separation, we will apply the reagents of the tyrosinase assay directly onto the gel. After an incubation period, we plan to image the gel to evaluate both nanocage assembly and enzymatic activity in these nanocages. Successful assembly will be indicated by distinct fluorescent bands in the expected region of the gel. If the tyrosinases enclosed within the nanocages are active, we expect to see these same bands in a brown color when imaged under normal light, confirming melanin synthesis within the assembled nanocages.‘

Tyrosinases

Design Of Experiment

One of the core elements of our project are encapsulated tyrosinases, which produce the pigment melanin. These enzymes convert L-tyrosine to melanin in a copper-dependent manner. To find the optimal conditions for efficient melanin production, we transfected HEK293T cells in a 96-well plate with a plasmid encoding a Bm Tyrosinase, coupled with a destabilization domain to regulate protein levels in the cytoplasm, a cargo signal for encapsulation into Mx encapsulin nanocages, and the sequence for the corresponding Mx encapsulin to allow co-assembly inside the cells.

We used a Design of Experiment (DOE) approach to systematically test different combinations of L-tyrosine and copper concentrations. To make sure our results are reliable, we applied a central composite design, where we repeated each condition twice and included five average values as reference points - so called central points.

Our tested L-tyrosine concentrations ranged from 0 mM to 10 mM, and copper concentrations from 0 µM to 65 µM.
We evaluated melanin production by measuring absorbance at 900 nm and normalized the measured values by substracting the absorbance of wells containing transfected cells without supplemented copper or L-tyrosine.
Since both factors, copper and L-tyrosine, showed a two-way interaction and a second order relaionship, their combined effect on absorbance, and therefore melanin synthesis, can be described using the following model:
$y = 0.03 + 0.01 \cdot [\text{Copper}] + 0.15 \cdot [\text{L-Tyrosine}] - 0.00 \cdot [\text{Copper}]^2 - 0.01 \cdot [\text{L-Tyrosine}]^2 - 0.00 \cdot [\text{Copper}] \cdot [\text{L-Tyrosine}]$

To visualize melanin synthesis under varying copper and L-tyrosine concentrations, we generated a 3D plot shown in Fig. 3.1.

Fig. 3.1 / DOE with titration assay. Effect of different L-tyrosine and copper concentrations on melanin synthesis.

This graph shows that at low L-tyrosine concentrations, increasing copper levels lead to rising in absorbance, indicating higher activity of the tyrosinases and therefore higher melanin synthesis. Similarly, rising L-trosinase level at a low copper concentration also lead to rising absorbance, indicating higher activity of the tyrosinases and therefore again higher melanin synthesis. This observation changes when looking at higher copper concentrations paired with higher L-tyrosine concentrations. In this case, absorbance decreases again after achieving a maximum. This phenomena could be explained by copper and L-tyrosine being toxic for cells if supplemented in a high concnetration, leading to cells dying and consequentely less melanin synthesis. Furthermore, it is important to note that with this experimental setup, we could not confrim the tyrosinase being successfully encapsulated. Therefore, it is possible that higher melanin production occured in the cytosol, leading to cell death as well. Lastly, even with the tyrosinases being efficiently loaded into the nanocages, there are still some present in the cytosol. Increasing L-tyrosine and copper concentration may enhance their activity to a level where the cells can’t tolerate the cytosolic melanin production.

Cell-free approach

To evaluate which of our designed tyrosinases leads to efficient melanin sysnthesis, we used a cell-free TXTL-based expression system for rapid prototyping. Our pipeline consisted of cloning the constructs via Gibson Assembly, amplifying them via RCA, and expressing them with TXTL before performing BCA and a tyrosinase activity assay. To confirm that our constructs correctly assmebled, we linearized them and verifiedied the presence of the insert via agarose-gel electrophoresis (as shown in Fig. 3.2) before expressing them cell-free.

Our verification gel showed bands at the expected length for the wild-type BmTyrs (lane 1), as well as the BmTyrs carrying targeted mutations (lanes 2—4), the circular permuted BmTyr (lane 5), and the three different split Tyrs (lane 6—11). In contrast to that, even though our HcTyr (lane 12) showed a band at the expected height, the corresponding HcTyr with deleted LID (lane 13) showed no signal at all, indicating failed assembly or RCA amplification. The wildtype LsTyr (lane 14) did show a faint band at the expected height, but also a much stronger band at a smaller length, indicating wrong assembly. The SkMelC1 tyrosinase (lane 15) showed a band at the expected height, as well as VsTyr with a LID domain (lane 16). As seen for HcTyr, the respective VsTyr with deleted LID (lane 17) showed no fluorescent signal at all, indicating again wrong assembly or failed amplification. The tyrosinase SkMelC2 (lane 19) shows a band at the expected height, as well as the tyrosinase variants of SavMelC1 (lane 18 and 20). In contrast to that, only one of the SavMelC2 tyrosinase variants showed the expected signal (lane 22), while the other one showed none (lane 21), likely due to wrong assembly or insufficient amplification. Lastly, we analyzed repeats of the Gibson assembly for the failed HcTyr and VsTyr deletion variants (lane 23 and 24). Both showed faint bands at the expected heigth, together with additional bands in the same intensity, hinting to unspecific assembly. Overall, we achieved the desired bands for all tyrosinases except for SavMelC2 (I42Y).

Bicinchoninic acid assay (BCA) for protein quantification

To quantify the total protein yield after cell-free TXTL expression, we performed a BCA. Our goal was assessing expression efficiency together with enabling downstream rapid prototyping of our tyrosinase constructs. We generated a standard curve using bovine serum albumin (BSA) standards. As a negative control, we performed a TXTL reaction without adding DNA template to account for background protein synthesis inherent to the TXTL system. We normalized the measured absorbance values against blank samples containing only BCA assay buffer. All our BCA measurements were done in duplicates. The results of the BCA are portrayed in Fig. 3.3.

The concentrations of the standards are plotted against their respective absorbance, generating the standard curve, yielding an R² value of 0.99, indicating great linearity and statistical reliability of the calibration. The absorbance values of the tyrosinase TXTL reactions were plotted on this curve to determine their protein concentrations.
Unfortunately, our tyrosinase samples exhibited very low absorbance values, corresponding to concentrations below those of the negative control. This suggests that either tyrosinase expression was inefficient or that the expressed protein concentrations were below the BCA detection limit. Under successful expression conditions, the tyrosinase samples would be expected to yield higher total protein levels compared to the template-free control, due to the additional production of the target enzyme.

Western Blot

After the BCA assay with the cell-free TXTL reactions resulted in no detectable protein, we decided to perform a Western Blot using the wild-type BmTyr as a representative construct to assess whether any protein was expressed. For that, we purified BmTyr via a strep column and analyzed the lysate, flowthrough, wash fraction, and both elution fractions by Western Blot. To validate the detection system, we included an anti-strep GFP control. The membrane after Ponceau S staining, as well as the Western Blot with chemiluminescent imaging, are shown in Fig 3.4.

Unfortunately, the Western Blot confirmed our assumption that the cell-free expression of our tyrosinases was unsuccessful. While we were still able to see unspecific protein bands in the lysate and flowthrough with the Ponceau S staining, the membrane was empty apart from one distinct band for our reference. Additionally, chemiluminescent imaging also showed no detectable signal apart from the anti-strep GFP control, indicating that the antibody with the conjugated horseradish peroxidase did not bind any target protein, most likely because almost no strep-tagged tyrosinase was produced during TXTL expression.

Expression in E. coli

After the attempt to prototype tyrosinases cell-free using E.coli extract failed due to insufficient protein expression, we switched towards expressing them in E.coli BL21 (DE3) from a pET21b vector. We purified the enzymes using Strep-Tactin Spin columns, determined protein concentration by measuring absorbance at 280 nm using nanodrop, tested purity of samples on SDS-PAGE and run colorimetric tyrosinase assay.

The SDS-PAGE results were partially inconclusive. Due to time constraints we decided to test all of our samples in a tyrosinase assay to qualitatively determine whether the enzymes were functional. Since the samples were not pure, the measured protein concentration did not correspond solely to tyrosinase. Absorbance values were measured at 490 nm. The blank was subtracted from each well, and the resulting values were normalized to the total protein content to account for differences in absorbance due to varying protein concentrations. After 60 minutes, the measurements did not show any changes in absorbance. We therefore added 0.1 μM Cu²⁺, an essential cofactor for tyrosinases, and extended the incubation time by 12 h hours (Kipouros et al., 2022). Following this adjustment, differences became visible, as shown in the results below.

Fig. 3.5 Verification of tyrosinases expression levels in E.coli BL21 (DE3). The gel was stained in Coomassie Blue.

Split - Tyrosinases

Unfortunately, no full length BmTyr sample yielded detectable amounts of melanin. In light of the absent bands in lanes 2-4 on the SDS-PAGE (Fig. 3.5), this outcome was anticipated. The molecular weight of the split tyrosinases (lanes 5-10) was found to be too low for reliable verification on the the 12% gel. Nonetheless, the tyrosinase assay was performed on each N- and C-terminal domain in isolation, and the resulting absorbances were compared with those of a sample comprising both N- and C-terminal fragments. With modest optimism, an increase in the degree of absorption was observed (Fig. 3.6), which could not be fully explained by the estimated increase in protein content in the combined sample. However, in view of the evident constraints in evaluating sample concentration and purity, these findings are unmistakably preliminiary, necessesitating further investigation.

Fig. 3.6 Tyrosinase activity of split BmTyr: (A) NBmTyr(85)(BBa_2502R9NP) and CBmTyr (85) (BBa_25LKZLQD, (B) NBmTyr(157)(BBa_25O84IGR) and CBmTyr (157) (BBa_25JLEAQE). Absorbance at 490 nm indicates enzymatic activity.

Lid - Tyrosinases

Our setup for both lid tyrosinases (HcTyr and VsTyr) differed slightly. We expressed both tyrosinases with and without the respective lid domain. A comparison between the different constructs was impeded by the signigicant differences in yield and purity observed between the constructs. Instead, aliquots of the same samples (*HcTyr and VsTyr with the corresponding lid domain) where examined, with and without the addition of 10 U of TEV protease. Despite promising identification on the SDS page (lanes 12 and 13, 21 and 22), no detectable amount of melanin was synthesized by any of the VsTyr samples, irrespective of TEV addition. In contrast, HcTyr in conjunction with TEV protease resulted in the most significant increase of absorption, which was perceptible to the unaided eye (Fig. 3.7). While still requiring further research, these results suggest a potential working mechanism for TEV protease-mediated activation of tyrosinases.

Fig. 3.7 Tyrosinase activity of HcTyr1 (BBa_25OLOGCU) with and without TEV cleavage. Absorbance at 490 nm indicates enzymatic activity. The ”+TEV” sample contains the lid-cleavable tyrosinase treated with TEV protease, while the other sample was not treated.

Caddie - Tyrosinases

It was not possible to discern LsTyr (lane 14), SavMelC2 (16 and 18) and SkMelC2 (lane 20) on the SDS-PAGE. The previously described problem of low molecular weight was encountered again for SavMelC1 (lane 15 and 17) and SkMelC1 (lane 19). Hence, we were not able to properly determine the success of the purification. Unsurprinsingly, since caddie proteins alone are not sufficient to synthesize melanin, the respective assays did not show any absorbance.

Future tyrosinase experiments

As indicated by the SDS-PAGE, significant challanges were encountered during the purification process with the Strep-Tactin XT 4Flow Spin Column Kit. In subsequent steps, these difficulties may be addressed by optimizing the purification protocol, which may be achieved thorugh the testing of different incubation times, the variation of the amount of washing steps, and the adjustment of the elution volume of each fraction. Furthermore, it is evident that the SDS-PAGE method was not optimal for the detection of all tested proteins. In subsequent iterations, an effort will be made to modify the PAGE conditions by decreasing the voltage and transitioning to higher concentrated gels (15%—20%), with the objective of enhancing the resolution, particularly for low molecular weight proteins, such as the caddies and split fragments. Finally, the tyrosinase assay itself requires optimization. Due to the limited availability of purified protein, it was not feasible to perform tripliates of all measurements. Furthermore, the quantity of protein could not be normalized in this iteration of the assay. In consideration of the supplementary bands discernible on the SDS-PAGE, it is evident that the enzyme levels of some samples were below the requisite threshold, consequently resulting in protracted reaction times, we observed. Nonetheless, we observed slight increases in the degree of absorption for the mixtures that contained both N- and C-terminal fragments of split-Tyrosinases as well as a relatively high increase in absorbance for one of the TEV-activated activated lid-tyrosinases.

References