We tested four artificial organelles in E. coli, with the goal of creating rubber compartmentalisation. Key findings include: successful protein expression for all strategies; effective encapsulation visualization, particularly in VLP and ASIP designs; strong performance of ASIP in hydrophobic compartment formation and rubber production; and advantages of fusion protein strategies (ASIP and LLPS) over dual-protein approaches (VLP and PHA). UV-vis spectroscopy confirmed higher rubber yield in ASIP compared to other designs. These outcomes provide a foundation for developing more efficient, sustainable rubber production methods in both prokaryotic and eukaryotic systems.
Click to view details of each section.
Within H Brasiliensis, the rubber transferase enzyme is part of spherical organelles called the rubber particle. When expressed heterologously, the transferase is thought to integrate into the cell membrane in the absence of this organelle 1 2. The enzyme's association to the membrane decreases its activity and reduces the length of the polymer, which worsens its material properties.
Inspired by rubber particles of H Brasiliensis, we explored whether engineering compartments to isolate the enzyme from the cytoplasm could prevent its association to membranes and enhance its activity.
We hypothesized that four different strategies from the literature, which are liquid-liquid phase separation (LPS) 3, virus-like particles (VLP) 4 5, PHA-inspired granules (PHA) 6, and amphipathic spider-silk inspired proteins (ASIP) 7 could compartmentalise rubber transferase within E. coli and improve its activity. In other words, we investigated strategies for assembling artificial organelles that produce natural rubber.
The ASIP strategy, where micelles from through the self-assembly of a fusion protein composed of spider silk and rubber transferase domains, showed the most promising results with in encapsulating natural rubber and enhancing the activity of the transferase. Through computational modelling, we provide explanations for why ASIP was more robust and performant than other encapsulation methods.
Figure 1: We investigated four different encapsulation strategies. HRT2trunc was modified for each strategy’s plasmids; namely HRT2-PP for LLPS, HRT2-SP for VLP, PhaC-HRT2 for PHA and HRT2-ASIP for ASIP. The compartments that emerge from the self-assembly of these proteins are illustrated on the right-hand side. HRT2-PP from the LLPS plasmid was used for liquid-liquid phase separation, HRT2-SP from VLP for virus-like particles, PhaC-HRT2 from PHA for PHA-inspired granularization, and HRT2-ASIP from ASIP for spider-silk inspired micelle formation.
Figure 2: Designs for the 6 major plasmids used in our experiments. Refer to Table 1 for an explanation of their function.
To test our four strategies, we designed 21 different plasmids that we tested out in E. coli , which yielded varying levels success. A key of the prominent plasmids that will be mentioned in this page can be found here:
Table 1: Key of code, short names, and functions for the major plasmids relevant to our results page.
| Plasmid code | Plasmid short name | Function |
|---|---|---|
| IC2404_pQE60-DXS-IspH | IC04 | Overexpression of IPP -- uncontrollable |
| IC240401_pQE60-DXS-IspH | IPP | Overexpression of IPP – Arabinose induced |
| IC2405_pET28a_HRT2trunc | HRT2T (Hevea brasilensis rubber transferase truncated) | Expression of the truncated rubber transferase enzyme HRT2trunc |
| IC2403_pET28a_PhaP_PhaC-HRT2trunc | PHA | Expression of Pha-HRT2 composite protein |
| IC2406_P22_HRT2-SP | VLP (virus-like protein) | Expression of P22 HRT2-SP |
| ICASIP_HRT2-NT-ASIP | ASIP (amphipathic spider inspired protein) | Expression of HRT2-ASIP |
| pET28a | pET | Produce the backbone pET28a for plasmid assembly |
| ICPP_MCP-HRT2trunc | LLPS (liquid-liquid phase separation) | Produce the LLPS system in E. coli |
While some of our designs were directly obtained from the manufacturer, some of the composite designs required specific sequences from other plasmids for assembly to get a larger, more complex plasmid. For instance, the HRT2trunc plasmid, which is used to express our truncated design of the rubber transferase required sections from two plasmids, PHA and pET, to be concatenated (Fig. 3). We targeted the necessary sequences via specifically designed primers and amplified these sections using PCR. Before the Gibson Assembly protocol was carried out, we analysed the sequences that underwent PCR using gel electrophoresis.
Figure 3: Diagram of the formation of HRT2trunc. It involves the amplification of the HRT2trunc CDS from the fusion protein used in the PHA design and insert it into the linearized PET28a backbone.
Figure 4: Gel image of PCR products from pET and PHA. From left to right: Lane 1: Generuler 1kb plus ladder marker (Thermo Fisher, UK); lane 2: linearized pET vector (5.1kbp); Lane 3: HRT2trunc fragment from PHA plasmid (~0.5kbp).
We identified the appropriate bands at 0.5kbp and 5 kbp locations from PHA and pET respectively (Fig. 4). Then, we cut out the bands from the gel and extracted the DNA inside these sections using a Gel Extraction Kit (Qiagen, UK). The cleaned-up DNA was then used for Gibson assembly where the assembly product was used to transform DH5a cells. Colony PCR was then performed with the 8 transformants picked from the agar plate. These PCR products were also analysed using gel electrophoresis (Fig. 5).
Figure 5: Gel image of samples 1 – 8 of assembled HRT2trunc plasmids. Lane L=1kb ladder, Lanes 1 – 8=HRT2trunc samples from transformed colonies. Blue line indicates the desired size for the insert (500 bps).
This gel demonstrated that a 0.5kbp insert from the colony template in lanes 4, 6 and 8 were successful amplified (Fig. 5). The plasmids from these three colonies were later sequenced through the full-plasmid sequencing service (Full Circle Labs, London) and successful insertions of the HRT2trunc sequence into the pET backbone were confirmed for all. While the IPP design was also assembled and verified using these steps, other designs–such as IC04, VLP and ASIP–were ordered and synthesized by a DNA manufacturer (GenScript).
Although it was a simpler plasmid in terms of its sequence, the IC04 plasmid posed a significant difficulty when we attempted to transform it intoE. coli . After several attempts where no growth was observed, we identified that the backbone had a leaky Tac-Promoter (pTac) which was overexpressing IC04 producing enzymes in the cells to a lethal extent even when no inducers were added. To restore the inducibility and tunability of the expression these enzymes, the IC04 plasmid was modified into the IPP plasmid by changing the pTac with an arabinose-inducible promoter, pBAD (Fig. 6). The edit was made via Gibson Assembly and verified through gel electrophoresis followed by sequencing. Thus, we prevented the uncontrolled expression of IPP producing enzymes in our new plasmid.
Figure 6: We edited IC04 into the IPP plasmid through Gibson Assembly.
Figure 7: Gel image of preparing fragments for assembling IPP. Left: linearization of IC04 and discarding the old pTac promoter. Lane L: Generuler 1kb plus ladder, lane 1: IC04 linearized (~5.2kbp). Right: amplifying pBAD promoter sequence. From left to right: Lane L=1kb ladder. Lanes 1 and 2= pBAD amplified from pS361-mcherry plasmid (provided kindly by our advisor Kathakali)
From our plasmid designs, an encapsulation plasmid needed to be inserted along with the IPP over-expressing plasmid for the E. coli to produce long-chain rubber. This co-transformation process was repeated for all four strategies.
Co-transformants were grown with and without the presence of inducers (IPTG and arabinose) over 15 hours. We examined the metabolic burden imposed by each of our plasmids on the cells to assess the differences in expression efficiency and biological compatibility across the designs. To achieve this, we regularly measured the OD600 (optical density at 600 nm) values of each co-transformed colony over this growth period. These measurements provide information on the growth of E. coli transformed with our different designs.
Figure 8: Growth curves for control (pET) colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG
Figure 8: Growth curves for control (pET) colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG
Figure 9: Growth curves for HRT2trunc colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG.
All growth curves for colonies expressing HRT2trunc resemble the trend seen in negative control curves. Both the non-induced and IPTG induced curves in Fig. 9 have a similar trend, and plateau at the same OD600 value. This suggests that the HRT2trunc plasmid does not cause a significant amount of stress to the cells. The addition of arabinose appears to boost growth activity, as both graphs B and C reach optical density of 1.2 after induction over the full growth period.
The rest of the growth curves obtained are used for discussing the effect of our encapsulation techniques on E. coli.
Figure 10: Growth curves for PHA colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG.
All curves for the PHA design colonies are similar to those of pET and HRT2trunc, with a similar lag phase and plateau. This suggests that like the HRT2trunc, the PHA plasmid has negligible burden on E. coli cells.
Figure 11: Growth curves for VLP colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG.
The growth curve for IPTG induced colonies in figure 11 appears to have a lower OD600 value than the rest. Although the non-induced curve has an endpoint similar to graph D, graph A trends upwards after induction and decreases slightly towards the end of the growth period. The larger error bars for the data points which have curved downwards, compared with those of graph D, suggest that VLP has exerted observable stress onto E. coli cells that have stunted their growth. While graphs B and C in figure 11 show a similar end point of 1.2 optical density, unlike the B and C graphs of figure 8 and figure 10 the colonies do not have as sharp of an increase after induction. This suggests that although the addition of Arabinose has boosted the growth, the VLP plasmid’s burden still stunted the growth of the colonies.
Figure 12: Growth curves for LLPS colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG.
The growth curves of LLPS colonies demonstrate that the plasmid has clear burden on cells, as none of the colonies reached optical density value of 1.2 even with the addition of arabinose.
Figure 13: Growth curves for ASIP colonies. A=no inducers, B=Arabinose, C=Arabinose and IPTG, D=IPTG.
As for the ASIP colonies, they have shown the most amount of fluctuation between repeats, as seen in error bars throughout growth period after induction, or the 5-hour mark in the case of graph A. This suggests that the ASIP plasmid has the highest variation in the amount of burden it poses onto cells. Although graph D suggests that it exerts an observable amount of metabolic burden, curve C resembles that of control. From this, we can deduce that although ASIP stunts the growth of cells when induced only by IPTG, this effect is compensated when they are induced by IPTG and arabinose simultaneously.
Overall, after studying the entire dataset and comparing the growth curves of our designs with those of pET, we concluded that although they may cause slight variations in cell growth, our designs do not pose a consequential burden on E. coli.
Alongside our co-transformation growth curves, we also transformed individual plasmids independently into E. coli cells and carried out an SDS-PAGE to verify whether our desired proteins were successfully being expressed.
Figure 14: SDS-PAGE profile of all BL21 transformants with different plasmids, induced overnight with 0.5mM IPTG. From left to right, lane L: Precision Plus Protein™ Ladder marker (Bio-rad, UK); Lane 1 and 2: Lysate supernatant and pellet from PHA transformant, PhaC-HRT2trunc is identified with a bands around 45kDa; Lane 3 and 4: lysis supernatant and pellet from HRT2trunc construct, the truncated rubber transferase enzyme expressed with a band around 20kDa; Lane 5 and 6: supernatant and pellet of VLP expressing both P22 capsid and HRT2-SP, bands identified at 46 and 26kDa; Lane 7 and 8: supernatant and pellet from ASIP with bands identified at 45kDa for HRT2-NT-asip; Lane 9 – 12: lysis supernatant and pellet from LLPS and ICPP-NH. Where the fusion protein of HRT2trunc with the MCP protein in their constructs were identified as dimers and trimers, potentially indicating the formation of disulphide bonds. All desired bands are indicated with an arrow. S means lysate supernatant, and P signifies lysate pellet. Samples were sonicated during lysis unless noted otherwise.
The results from this SDS-PAGE showed successful expression of all proteins for each of our strategies. Specifically, the band pointed with a blue arrow (43kDa) on lane 1 points to the PhaCRT2 protein that is used for the PHA encapsulation strategy, whereas the band pointed by the green arrow on lane 4 shows HRT2trunc (22kDa) expression. For our VLP approach, the P22 capsid protein (46kDa) is located with the upper band (shown with purple arrow) on lane 6, and the HRT2-SP (26kDa) protein is observed on the same lane at the bottom band pointed the second purple arrow. This revealed the expression of proteins by the recombinant VLP plasmid was successful. Our HRT2-NTasip (45kDa) approach also demonstrated a positive outcome, based on the band on lane 8 pointed with the yellow arrow. Finally, the HRT-PP protein was able to be located on lane 10 with two pink arrows at 100kDa and 150kDa respectively. This indicated the presence of disulphide bonds which force the formation of dimer and trimers of HRT2trunc-MCP fusion protein.
Figure 15: SDSPAGE analysis result of IPTG-induced VLP overnight culture. From left to right: lane L: Transgen 10-180kDa protein marker; lane 1: cell pellet after lysed with BugBuster with 0.2mg/ml lysozyme for 30min at room temperature; lane 2: lysate supernatant centrifuged at 3k rpm for 10min; lane 3: lysate supernatant centrifuged at 12k rpm for 10min; lane 4: cells sampled directly from overnight culture; lane 5: 0.45um PDMV-filtered lysate (after centrifuged at 3k rpm); lane 6: first wash (with PBS) after protein loaded to Ni-NTA columns; lane 7: second wash (with Rinsing buffer) of the Ni-NTA column; lane 8: 5ul of 20x concentrated elute; lane 9: 80x concentrated lysate supernatant 5ul. All lanes except lane 8 and 9 were loaded with 10ul of samples. The P22 capsid protein around 46kDa and HRT2trunc protein around 26kDa were identified with red arrows in the gel.
After obtaining the SDS-PAGE results on figure 14, we wondered if it was possible for the P22 protein to encapsulate the HRT2-SP protein. We also wanted to test whether it would be possible to purify this construct - the HRT2-SP protein inside its P22 shell. For this, we conducted a more detailed SDS-PAGE analysis to trace the presence of our virus-like-particle through an entire lysis process. In Fig. 15, we identified our VLP and HRT2-SP proteins to be present both inside the E. coli and in the lysate supernatant after the cell is disrupted (lanes 1 and 2). After spinning at both 3k and 12k rpm for 10 minutes (lanes 2 and 3), the VLP and HRT2-SP remained in the liquid phase, which was the same case in lane 5, after passing the 0.45µm filter. This suggests that the VLP is highly separated from the cell debris in the lysate. We hence applied ion-exchange chromatography using columns of Ni-NTA resin to selectively purify the His-tagged P22 capsid protein from the filtered lysate. The presence of P22 was successfully detected after eluting with 200mM imidazole, yet it was untraceable in the flow-through of washing buffers (PBS with 1% glycerol). After spinning and concentrating the eluted VLP through a 30kDa filter-spin column, we were able to identify a meaningful amount of HRT2-SP protein with the VLP (lane 8). Since HRT2-SP does not carry a His-tag and is smaller than the 30kDa pores on the filter, it is highly evident that the HRT2-SP was present inside to VLP to throughout the lysis and purification procedures.
After obtaining the SDS-PAGE results on figure 14, we wondered if it was possible for the P22 protein to encapsulate the HRT2-SP protein. We also wanted to test whether it would be possible to purify this construct - the HRT2-SP protein inside its P22 shell. For this, we conducted a more detailed SDS-PAGE analysis to trace the presence of our virus-like-particle through an entire lysis process. In Fig. 15, we identified our VLP and HRT2-SP proteins to be present both inside the E. coli and in the lysate supernatant after the cell is disrupted (lanes 1 and 2). After spinning at both 3k and 12k rpm for 10 minutes (lanes 2 and 3), the VLP and HRT2-SP remained in the liquid phase, which was the same case in lane 5, after passing the 0.45µm filter. This suggests that the VLP is highly separated from the cell debris in the lysate. We hence applied ion-exchange chromatography using columns of Ni-NTA resin to selectively purify the His-tagged P22 capsid protein from the filtered lysate. The presence of P22 was successfully detected after eluting with 200mM imidazole, yet it was untraceable in the flow-through of washing buffers (PBS with 1% glycerol). After spinning and concentrating the eluted VLP through a 30kDa filter-spin column, we were able to identify a meaningful amount of HRT2-SP protein with the VLP (lane 8). Since HRT2-SP does not carry a His-tag and is smaller than the 30kDa pores on the filter, it is highly evident that the HRT2-SP was present inside to VLP to throughout the lysis and purification procedures.
We wanted to visualise our encapsulation techniques within the cells to confirm that they were forming successfully. One method that made this possible was using transmission electron microscopy (TEM). Once our SDS-PAGE from Fig. 15 confirmed that our VLP construct was successfully encapsulating the HRT2-SP protein, we sent a colony transformed with VLP to be imaged. TEM allowed us to visualise the intricate constructs within these cells with great detail thanks to its powerful magnification.
Figure 16: Negatively stained transmission electron microscopy (TEM) imaged with 80kV at 15k magnification for the VLP concentrated in PBS. Spherical VLPs have been identified in the sample (with red arrows) of approximately 50nm in diameter. Regions are zoomed out on the left side of the image where the bar corresponds to 50nm.
The concentrated VLP after purification was sent to our external contractor (Service Bio, China), where the TEM was then pictured. The figure 16, shows several spots of spherical objects approximately 50nm in diameter exhibiting the characteristics to be our VLP 8 . This proved our encapsulation strategy is functioning.
Our hypothesis stated that our artificial organelles could harbour a hydrophobic pocket inside them. Theoretically, these hydrophobic compartments would allow for polymerization of rubber chains inside the host organism. Once we visually confirmed that one of our designs (VLP) had successfully accomplished encapsulation, we aimed to test for the presence of hydrophobic compartments in our designs.
We chose to investigate this with a lipophilic staining technique using BODIPY 493/503, due to its ability to stain hydrophobic compartments, such as lipid droplets, at intracellular scales 9.
To compare the number of hydrophobic compartments formed within cells using different artificial organelles, the fluorescence of different strains after stained with BODIPY was measured in 96-well plates using a plate reader.
Figure 17: Fluorescence assay with excitation at 485nm and emission at 515nm (gain = 1000) for transformants with pET (control), PHA, HRT2trunc, VLP, ASIP and LLPS plasmids. Positive cells were induced with IPTG while negative ones were not. p<0.05; p<0.005. The medium of each group is represented with the cross in the plot.
The emission values of all non-induced colonies were similar, around 15000. Both the induced and non-induced samples of our control colonies transformed with pET produced similar fluorescence readings, which can be accepted as the autofluorescence of E. coli cells. The PHA design has not increased in fluorescence after induction, which indicates that it was not able to form non-polar vesicles within the cells. Similarly, both HRT2trunc samples have emission values with no statistically significant difference (p > 0.05). However, this result for the HRT2trunc design was expected as it expresses only the truncated rubber transferase enzyme without any encapsulation design. This signifies that the presence of unencapsulated rubber transferase within the cytoplasm affects E. coli fluorescence by a negligible amount.
The fluorescent emission of IPTG induced VLP, ASIP and LLPS colonies were higher than their non-induced counterparts. From this, we can infer that the cells transformed with these strains were able to form non-polar capsules with varying degrees of success. While the success of the VLP design is statistically lower due to the large error bars of the positive VLP colonies (shown darker purple in figure 17), the ASIP and LLPS strategies have demonstrated significantly higher fluorescence when induced. Specifically, ASIP strategy showed the highest amount of emission compared to our negative control, as its t-test results demonstrated a statistically significant increase from the negative group with p<0.005. Thus, our findings supported ASIP as the most successful design out of our four strategies in expressing proteins and forming non-polar compartment withinE. coli .
Following our results from the fluorescence emission assay, we decided to image our stained colonies under a confocal microscope to see whether the capsules can be identified within the cells.
Figure 18: Confocal images of stained ASIP colonies under 63x magnification. Left=ASIP negative (no induction), Right=ASIP positive (IPTG inducted). The red and orange boxes in both images signify the zoomed-in image for a single cell on both samples.
Our confocal images demonstrated the formation of lipid bodies in our ASIP and HRT2trunc co-transformed colonies. As seen in figure 18 the IPTG induced colonies showed greater local fluorescence than the non-induced negative controls. Aside from the two bright points in the negative image (left), which are caused by the auto-fluorescence of the cluster of cells overlapping, the overall image has a low average brightness. The image on the right showed individual cells with bright points within the cells, which is further verified by the high overall brightness value in the bulk of cells found at the top half of the image. Additionally, the z-stack images from the positive sample verified that the bright points were located inside of the cells, whereas the negative controls maintained their low brightness throughout.
After confirming the extent of hydrophobic encapsulation achieved by our designs, we proceeded to investigate the amount of rubber they could produce. We achieved this by measuring the absorbance of our most promising fusion protein (ASIP) and dual-protein (VLP) designs over the UV-vis spectra. We chose to focus our investigation between the 200-250nm spectrum, as the UV-vis absorbance of pure polyisoprene (rubber) has a peak around the 210nm wavelength 10 11. By comparing this spectrum with the absorbance of our ASIP and VLP colonies, co-transformed with IPP plasmid, after the rubber extraction process, we can assess how much rubber the colonies were able to synthesize. To eliminate any influence from cell debris or rubber produced outside the encapsulation, we compared these data to colonies of pET and IPP co-transformed cells that underwent the same rubber purification process.
Figure 19: The UV-vis absorbance assay carried out with purified rubber from 100ml culture of BL21 strains. All colonies were co-transformed with IPP IC04 plasmid. pET is used as a negative control that is transformed with the pET28a empty vector. The VLP and ASIP are transformed and induced with 0.5mM IPTG and cultured at 25oC for 48hrs. Purified rubber is dissolved in cyclohexane. The rubber is measured at absorbance at 210nm (indicated with the dashed-line) and blanked by subtracting the baseline absorbance of cyclohexane.
Figure 20: Table showing the comparison of rubber extracted from different <i>E. coli </i> cultures with known concentrations of polyisoprene. All samples are resuspended in 3mL of pure cyclohexane.
All transformants were cultured in 100ml quantities, followed by the rubber extraction method we adapted using procedures from existing literature and team SDU-Denmark 2013 (check our contribution page to see how it was refined) 1012. We were able to quantify the rubber extracted from E. coli through the UV-vis absorbance values at about 210nm 10 11. Figure 19 clearly depicts an increase of absorption for both the VLP and ASIP designs compared to the negative control, which suggests that our designs are producing rubber. The absorbance values at the 210nm point from this graph were collected and compared with the 210nm absorbance values of known concentrations of polyisoprene added to pET samples. Based on these comparisons, we were able to approximate the amount of produced from our ASIP and VLP designs to be 1500ug and 30ug respectively.
To further confirm our results from the UV-vis spectroscopy, we sent purified rubber extracted from our sample colonies to be analysed by Nuclear Magnetic Resonance (NMR). Each sample was extracted from colonies co-transformed with one of the four encapsulation strategies along with HRT2trunc. However, our results from NMR were inconclusive. One possible reason is that the functional groups in polyisoprene that we can detect through NMR, such as CH3 or CH2, are abundant within the cell, introducing considerable noise to our NMR data.
Our project's long-term goal is to develop a reliable rubber synthesis method using moss cells, aiming for sustainable implementation of our technology. We've already taken significant steps towards this goal by conducting preliminary work on moss culturing and transformation protocols alongside our E. coli experiments. This parallel approach allows us to build a foundation for translating our findings from bacteria to a more complex and environmentally sustainable plant-based system. As we progress, we plan to further refine and expand our moss-based research, paving the way for a green alternative in rubber production.
In collaboration with George Greiff at the University of Bristol, we acquired expertise in moss cultivation and genetic modification. This partnership enabled us to implement a rigorous weekly moss maintenance regimen and master an intricate transformation protocol. The precision demanded by these procedures underscored the complexity of working with moss as a chassis for synthetic biology applications.
To advance our moss transformation efforts, we collaborated with Team SZ-SHD, leveraging their expertise in tracked-transformation, translation, and trans-regulation (TTTT) in plants 13. Their advisor, Mr. Yingjie Lei, provided us with a GFP-expressing plasmid and a Nucleator plasmid, crucial tools for our investigations.
Our primary objective was to transform moss with the GFP plasmid and assess protein expression using fluorescence microscopy. This approach would allow us to quantify protein expression efficiency, regardless of the specific gene inserted.
We prepared the plasmids through molecular cloning in E. coli, followed by extraction and concentration optimisation. The transformation protocol involved creating four GFP-only samples and two samples co-transformed with both GFP and Nucleator plasmids.
Despite our careful approach, we encountered challenges in the transformation process. After a week of cultivation, we observed no visible cell growth or fluorescence. This outcome, while not producing quantifiable data, provided valuable insights into the complexities of moss transformation and allowed us to streamline the procedure based on the challenges we faced.
This experience has been crucial for us to identify key obstacles and potential improvements in our moss transformation procedure and our moss-based chassis approach, setting the stage for greater success in the future.
We aimed to create an encapsulation technique that could facilitate the effective production of rubber by the Hevea rubber transferase (HRT2) expressed heterologously. Therefore, the data we gathered from assessing our strategies after BODIPY staining were essential to our investigation. The fluorescence assay data demonstrated the varying levels of success of encapsulation for our four strategies. The ASIP and LLPS strategies had the most statistically significant results, whereas the VLP strategy had high variability and PHA strategy showed negligible difference from the control. We can use these differences in performance to understand the strengths and weaknesses of each method by dissecting their molecular interactions.
On a molecular level, the key difference between these strategies is the number of proteins being used for the encapsulation process. As also shown in the diagram in figure 2, the PHA and VLP plasmids use two separate proteins for their encapsulation and rubber synthesis strategies, while ASIP and LLPS both use an individual fusion protein. Due to this difference, the PHA and VLP strategies introduce two additional degrees of freedom in the system: the timing between capsule and enzyme expression, and the relative levels of protein expression.
For instance, premature expression of VLP capsid proteins (P22) can form many envelopes without enzymes, but delayed expression causes majority of rubber transferase (HRT2-SP) to integrate into the cell membrane. While the first scenario imposes metabolic burden to the cell, the latter hinders enzyme activity due to non-specific binding. The PHA plasmid also functions in a similar mechanism where the integration of HRT2trunc to the membrane could highly disrupt formation of the granule. The importance of these parameters to these 2-protein mechanisms can be explored using computational models. Such an investigation can potentially led to finding an optimum time separation where enzyme activity is enhanced without incurring high metabolic cost. However, this optimum needs to be experimentally investigated, which renders the methods involving fusion proteins more efficient.
The ASIP and LLPS strategies, while using single fusion proteins, function in different ways. ASIP is a variant of micelle formation. There is a competition between the energy gained from the aggregation of hydrophobic groups and the energy lost in forming an ordered micelle. 14 The free energy of forming a micelle depends on the length of the hydrophobic domain. The ASIP strategy relies on production of the aliphatic polymer to form a hydrophobic core, however it also introduces a large hydrophilic domain which can interfere with the enzyme. Based on this, we can speculate that this strategy introduces a risk of protein misfolding. However, it still holds a significant advantage over the dual-protein strategies since it doesn’t pose any additional variables to the mechanism. This makes the ASIP strategy a more robust method for encapsulation and rubber synthesis.
LLPS on the other hand, utilises a non-membrane bound organelle strategy. The liquid-liquid phase separation occurs when molecules undergo separation in the cellular environment, shifting from a homogeneous solution to two distinct phases: a dense phase (the droplet) and a dilute phase (the surrounding cytoplasm). 15 The LLPS plasmid utilises the MCP-HRT2trunc protein, which has specific structural domains that can self-associate, resulting in the formation of a droplet-like organelle. However, unlike the ASIP method, the LLPS strategy disrupts nuclear condensates due to its CG repeats and therefore is not viable in eukaryotic cells such as moss. This renders the strategy only viable to be used in prokaryotic cells.
These discussions are further supported by our findings. The VLP plasmid showed high levels of variance as indicated by the large error bars in figure 17. This can be a firm sign caused by the variances in the dual protein strategy. Meanwhile, the PHA strategy showed low variance. The ASIP fluorescence assay demonstrated significant fluorescence difference compared to the control sample. Furthermore, when visualised under a confocal microscope the ASIP sample appeared to have increased brightness compared to the control, with distinctive bright points within IPTG induced cells. This strongly suggests that it was the most robust design out of our four strategies, that was able to successfully create hydrophobic encapsulation suitable for rubber production. While LLPS plasmid also showed significant results, the p-value was greater than that of ASIP, meeting only a lower standard of proof. Additionally, the induced LLPS samples did not demonstrate a clear increase in brightness when visualised using confocal microscopy.
The efficiency of fusion-protein (ASIP) strategies over dual-protein encapsulation (VLP) is also supported by our UV-vis spectroscopy findings. The absorbance of from figure 19 demonstrated a significant increase in absorption at 210nm for ASIP compared to VLP. This suggests that the fusion protein ASIP strategy is much more efficient in producing rubber than VLP’s dual protein design.
Based on all the above, we conclude that fusion protein strategies are much more efficient in creating encapsulations. Our fusion protein strategies - LLPS and ASIP designs - have performed in accordance with these discussions. While both can be viable for prokaryotic cells, the ASIP strategy is the most promising if the technique is to be used for eukaryotic cells, such as moss!
Yamashita S, Yamaguchi H, Waki T, Aoki Y, Mizuno M, Yanbe F, et al. Identification and Reconstitution of the Rubber Biosynthetic Machinery on Rubber Particles from Hevea Brasiliensis. Okada K, editor. eLife. 2016 Oct 28;5:e19022.
Yamashita S, Takahashi S. Molecular Mechanisms of Natural Rubber Biosynthesis. Annual Review of Biochemistry. 2020 Jun 20;89(Volume 89, 2020):821–51.
Guo H, Ryan JC, Song X, Mallet A, Zhang M, Pabst V, et al. Spatial engineering of <i>E. coli </i> with addressable phase-separated RNAs. Cell. 2022 Sep 29;185(20):3823-3837.e23.
Patterson DP, Schwarz B, El-Boubbou K, Oost J van der, Prevelige PE, Douglas T. Virus-like particle nanoreactors: programmed encapsulation of the thermostable CelB glycosidase inside the P22 capsid. Soft Matter. 2012 Sep 19;8(39):10158–66.
Patterson DP, Prevelige PE, Douglas T. Nanoreactors by Programmed Enzyme Encapsulation Inside the Capsid of the Bacteriophage P22. ACS Nano. 2012 Jun 26;6(6):5000–9.
Bresan S, Sznajder A, Hauf W, Forchhammer K, Pfeiffer D, Jendrossek D. Polyhydroxyalkanoate (PHA) Granules Have no Phospholipids. Sci Rep. 2016 May 25;6(1):26612.
Kronqvist N, Sarr M, Lindqvist A, Nordling K, Otikovs M, Venturi L, et al. Efficient protein production inspired by how spiders make silk. Nat Commun. 2017 May 23;8(1):15504.
McCoy K, Selivanovitch E, Luque D, Lee B, Edwards E, Castón JR, Douglas T. Cargo retention inside P22 virus-like particles. Biomacromolecules. 2018 Aug 9;19(9):3738-46.
Chen J, Liu W, Fang X, Qiao Q, Xu Z. BODIPY 493 acts as a bright buffering fluorogenic probe for super-resolution imaging of lipid droplet dynamics. 2022 Dec 1;33(12):5042–6.
Asawatreratanakul K, Zhang YW, Wititsuwannakul D, Wititsuwannakul R, Takahashi S, Rattanapittayaporn A, Koyama T. Molecular cloning, expression and characterization of cDNA encoding cis‐prenyltransferases from Hevea brasiliensis: a key factor participating in natural rubber biosynthesis. European Journal of Biochemistry. 2003 Dec;270(23):4671-80.
Abu Hassan A. Experimental and computational study of latex clearing protein LcpK30 for rubber degradation (Doctoral dissertation, University of Nottingham).
Team:SDU-Denmark - 2013.igem.org [Internet]. Igem.org. 2024 [cited 2024 Oct 2]. Available from: https://2013.igem.org/Team:SDU-Denmark/
Plant | SZ-SHD - iGEM 2024 [Internet]. Igem.wiki. 2024 [cited 2024 Oct 2]. Available from: https://2024.igem.wiki/sz-shd/plant#Accelerated%20plant%20synbio%20research
Maibaum, L., Dinner, A. R. & Chandler, D. Micelle Formation and the Hydrophobic Effect. The Journal of Physical Chemistry B 108, 6778–6781 (2004).
Alberti S, Gladfelter A, Mittag T. Considerations and Challenges in Studying Liquid-Liquid Phase Separation and Biomolecular Condensates. Cell. 2019 Jan;176(3):419–34.
