Results

Summary

On this page you will find the results of the experiments we have conducted this year and our interpretation of the data. This page will overlap with the Engineering Success page, the latter presenting a refined summary of our design-build-test-learn cycles and key positive results. The Engineering Success page provides the (iterative) rationale behind our experimental design. Meanwhile this Results page provides additional information about several preliminary optimisation experiments, further detail on purification and cloning, and other miscellaneous data.

---

Expression & Purification

Standard expressions

Our team was provided stocks of DH5α containing Reb1 and Reb206 plasmids created by Polka et al. (2016). We amplified the plasmids, transformed into BL21 E. coli and used mNeonGreen fluorescence per cell as a proxy measure of R body expression (Figure 1), with Reb1 serving as a control. It should be noted however that without data on relative rates of fusion versus wildtype RebB monomer expression, rates of monomer incorporation into assembled R bodies, or rates of successful R body assembly given monomer expression, drawing correlation between monomer expression (fluorescence data) and R body yield is only approximate.

We observed that codon optimisation did not improve R body yield although it appeared to accelerate the rate of bacterial growth. Enriched media more than doubled R body yield. Expression in TPM and codon-optimisation of the 8 composite parts created in this iGEM project is therefore an important consideration for future entrepreneurship work, particularly with respect to scaling-up operations.

---

R body purification

Our next set of experiments aimed to replicate the R body purification protocol described in Polka et al. (2016) for a fluorescent R body construct where the RebB monomer is fused to mNeonGreen, named Reb206. The protocol is unconventional as it aims to solubilise everything but the R bodies, the inverse of typical FPLC procedure. It is a harsh protocol that exploits the insolubility and density of R bodies to separate them from cellular components by lysis, enzymatic cleanup, centrifugation and repeated washing with detergent. We could not successfully replicate said protocol for Reb206. We found that all reported protocols use variations of the same key steps, and therefore began optimising these steps until sufficient purity was achieved (Lalucat, 1988; Schrallhammer et al., 2012). We experimented for six months with combinations of longer incubation lengths, multiple enzyme incubations, different wash detergents and concentrations, centrifugation speeds and other parameters. Although we ultimately achieved high purity, we could not observe Reb206 extension in acid. We therefore gained access to a more advanced microscope and repeated purification on wildtype Reb1, which was imaged with DIC and phase contrast microscopy. This finally yielded positive results.

These results primarily demonstrate the effect of an iterative experimental process, where successful strategies were preserved and unsuccessful ones discarded. The impact of these strategies were synergistic: for instance, we believe Triton-X100 solubilised more contaminants than SDS and therefore a second enzymatic incubation step proved doubly effective when the optimal detergent was used. It might be that Triton-X100, as a non-ionic detergent with greater micelle diameter, interacts favourably with amphipathic or hydrophobic contaminants than anionic SDS (Chae et al., 2013). Our team is also surprised we were unable to replicate the Polka et al. (2016) protocol. One contributing factor may be the different cell lines used for protein expression: Polka et al. used C43 whereas our team used BL21 cells, which might have altered the identity or quantity of contaminants and necessitated harsher purification.

---

Purified R body behaviour

Apart from the expected pH-dependent contraction-extension behaviour (see Parts or Figure 2F and 2G), we also observed the time-dependent clumping of R bodies in purified samples. This finding reinforces that vortexing samples may permit smoother sample transfer, and reinforces good microscopy technique. Otherwise we present it as a characteristic behaviour of pure R body samples that we have not yet seen described in the literature.

---

Positive controls for amber codon suppression

Later in the project, our team expressed and purified most new Reb constructs (Cys-N RebB, RebA-LPETG, etc.) in an identical manner to Reb1 and Reb206 (see above). The TAG-RebA and TAG-RebB constructs for amber codon suppression were the only exceptions. Given our limited experience with amber codon suppression we conducted a positive control using E. coli cotransformed with pEVOL_AzF and pCDF RFP amber plasmid. The former encodes the p-azido-L-phenylalanylyl-tRNA synthetase and tRNACUA for the amber codon. The latter encodes a modified RFP where the first amino acid is p-azido-L-phenylalanine (AzF). Inspired by Ma et al. (2014) we qualitatively tested whether RFP expression was enhanced by the addition of organic solvents, which helps overcome resistance to AzF uptake through the cell membrane (Figure 4).

We observed little difference with or without the presence of organic solvent. This may be due to the qualitative nature of our analysis and due to the small culture volumes (2 mL), as Ma et al. (2014) report significant effects on protein yield. Optimising amber codon suppression using organic solvents could be of interest in scaling-up operations, but presently we were hesitant to introduce an untested compound into R body expression medium. We therefore chose to use AzF alone.
The very purification of sizeable pellets from TAG-RebB and TAG-RebA expression cultures suggests that amber codon suppression was achieved, given that the amber codon was positioned close to the N-terminus of RebA or RebB. This alone cannot definitely claim success amino acid misincorporation may produce false positives. However the success of CuAAC conjugation implies that AzF had been incorporated into RebA or RebB monomers. Amber codon suppression can be a challenging technique to implement. Although powerful and easy to execute, its implementation in the lab is commonly hampered by unpredictable variations in incorporation efficiencies at different amber codon positions. Flanking sequences, cell lines, ribosomal fidelity and even permeability of the non-canonical amino acid (ncAA) all influence rates of misincorporation or translation termination. Bartoschek et al. (2021) have developed a predictive tool to identify permissive amber codon sites, which we could use in the future to identify which contextual factors enabled our success with ncAA incorporation.

---

Genetic Assembly

Overlap extension PCR

As mentioned in our Engineering Success section, overlap extension PCR was used to create fragments for Gibson assembly. As shown in Figure 4, an insertion sequence can be encoded at any site using two pairs of primers, in each of which there is a primer with an overhang. Standard Gibson assembly can then ligate the short and long fragments (sometimes known as insert and backbone fragments). The small size of the Reb plasmids (~8000 base pairs) was crucial for this bipartite cloning strategy.

All our primer designs are tabulated in the Protocols page and all successfully produced PCR products, which were subsequently Gibson assembled. The Gibson products were transformed, amplified and then sequenced to confirm correct assembly (example in Figure 5). All our Gibson sequencing data is publicly available at this link.

In the following drop-down menu we provide details of our entire PCR process for all Cys constructs (Cys-N RebA, Cys-C RebB, etc.), all LPETG constructs (RebA-LPETG, RebB-LPETG) and GGG-mNeonGreen. The PCR gels for TAG-RebA and TAG-RebB are not shown.

PCR Gel Results

We found that the shorter fragments had amplified poorly in the old thermocycler, and the longer fragments, amplified in a newer thermocycler, had amplified sufficiently well. The fragments all appeared to be close to the expected size by looking at the 1kb ladder, indicating that at least our fragments had been produced successfully. Suspecting that the PCR conditions for the short fragments was less-than-ideal, we tested a range of T_m temperatures and DMSO conditions in the next PCR round, as well as amplifying and producing the remaining cysteine-ready longer reb constructs.

We switched thermocyclers, and found that the short fragments mostly amplified well in the newer thermocycler, with most conditions being quite viable. The long samples, amplified in the older thermocycler but in the same conditions as previous (T_m=58°C, no DMSO) came out poorly — thus, the thermocycler was most likely the culprit in the faulty PCR. We then amplified the remaining long fragments in the new thermocycler in the same conditions.

Viewing the amplification results above, both thermocyclers worked similarly well, with all the samples appearing quite clearly on the gel. Thus, it remains unclear what caused the previous amplification failures in the previous gel runs — one possibility is that the older thermocycler was more complicated to configure, and it was then incorrectly programmed for amplification the first two times. In any case, at this point, we had amplified samples of all the long and short fragments for both Sortase A conjugation and cysteine-maleimide conjugation. We pooled the amplified replicates of each sample and ran them on a gel for extraction.

All the samples ran well on the gel. We cut the bands out of the gel, and used a gel extraction kit to purify the DNA for downstream purposes.

We also modified mNeonGreen in a pCDFDuet-1_His_mNeon_TmTP backbone. To make mNeonGreen compatible with sortase A-catalysed conjugation, an N-terminal triglycine motif was added, preceded by a TEV protease recognition sequence to cleave off the N-terminal methionine residue.

---

Assembly challenges

Below we also report the difficulty of cloning and transforming certain constructs:

1. RebA-LPETG, RebB-LPETG and GGG-mNeonGreen for sortase A bioconjugation were assembled and transformed into DH5α and BL21 on the first attempt.



2. Cys-C RebB and Cys-N RebB required troubleshooting with different short fragment to long fragment ratios and preparing fresh Gibson mastermix for assembly to work. This helped us obtain Cys-C RebB and Cys-N RebB transformants.



3. TAG-RebA and TAG-RebB constructs assembled correctly, which we confirmed by running products on a gel, but obstinately refused to transform into DH5α. Troubleshooting by varying the amount of DNA delivered to cells was not fruitful. Transformation was only successful after commercial One Shot™ TOP10 Chemically Competent E. coli from ThermoFisher were used.



4. Cys-N RebA and Cys-C RebA could not transform into DH5α. We were limited in the supply of commercial competent cells, and given our success with Cys-N RebB and Cys-C RebB chose to save this genetic construct for future efforts.
5. Cotransformation of pEVOL_AzF with any other plasmid (pCDF RFP amber, TAG-RebA or TAG-RebB) into BL21 proved impossible, regardless of variations in quantities and ratios of DNA. Commercial BL21(DE3) Competent Cells from ThermoFisher however were able to compensate for the inefficiency of the transformation process.

We therefore observe that some of our DNA constructs exhibit poor transformation efficiency or poor assembly, likely for reasons peculiar to each construct.

---

Extension testing

A key question our team needed to answer is whether our N-terminus or C-terminus modifications influence the extension ability of R bodies. At present, we have confirmed that RebB-LPETG, RebA-LPETG and Cys-C RebB extend readily at pH 5 (Figure 8). Cys-N RebB has not been observed to extend, although an unusual pellet appearance (not shown) and low concentration of R bodies (Figure 8B) leads us to believe a purification error has occurred. We would replicate the protocol to be certain. Cai (2023) proposes a physical model whereby a coil-to-helix transition of the C-termini of R body monomers induces a tension vector change in the lattice structure and unrolls it. We believe our success with preserving extension ability in our C-terminus may therefore lie in the use of linkers that are averse to forming secondary structures (oligoglycine tract in GGGGSC for Cys-C RebB, or Pro in LPETG). This should be an important consideration for further expansion of the R body conjugation toolbox.

---

Bioconjugation

Thiol-maleimide conjugation

Thiol-maleimide conjugation was performed on Cys-C RebB and Cys-N RebB. We observed that Cys-C RebB repeatedly did not conjugate with sulfo-Cy5 maleimide, whereas Cys-N RebB did so readily. This result cannot be a product of chemical accessibility since C-terminus conjugation to mNeonGreen was successful, nor is disulfide bridge formation a likely culprit given that Cys-N RebB can be conjugated. Further investigation is necessary to understand the cause underlying failed conjugation. Other than sulfo-Cy5 maleimide, we successfully conjugated aldoxorubicin to Cys-N RebB for tests in cell culture.

Our team is not aware of any biochemical characterisation technique that can quantify R body concentration and therefore conjugation efficiency. This limits our conclusions to a qualitative answer; whether conjugation is possible or not. For practical applications, conjugation efficiency will need to be defined.

---

Copper-catalysed azide-alkyne cycloaddition (CuAAC) conjugation

Our first CuAAC reaction onto TAG-RebA and TAG-RebB was unsuccessful. As we were quite confident in ncAA incorporation, we troubleshooted the CuAAC reaction by adding a water-soluble ligand, THPTA, which coordinates Cu(I) and accelerates the reaction by preventing Cu(I) oxidation into Cu(II). We tried this approach as the original CuAAC paper (Rostovtsev et al. 2002) did not utilise a Cu-coordinating agent and reported good yields, and we believed that degassing solutions would sufficiently prevent Cu(I) oxidation. The addition of THPTA however drastically improved the efficiency of the reaction and enabled conjugation.

---

2-pyridinecarboxyaldehyde (2-PCA) conjugation

Our first conjugation attempt was unsuccessful. After reviewing optimisations reported by Bridge et al. (2021) we concluded that the reaction was temperature sensitive and retried at higher temperatures (whereas the first attempt was conducted at 4°C). This time both our negative control and experimental pellets were blue (Figure 12A). We hypothesized that the sulfo-Cy5 azide dye might associate strongly and non-specifically with R bodies (e.g. due to π-π stacking) and washed thoroughly with 20% ethanol to better solubilise the dye. We discarded very blue supernatants and obtained the expected blue experimental and white control pellets. Given that we discovered, incidentally, that R bodies are stable in 100% ethanol we would recommend a 20% ethanol wash for all conjugation protocols and will undertake this in future cycles.

---

Sortase A conjugation

Our modified GGG-mNeonGreen was expressed in BL21 cells. Refer to Protocols for GGG-mNeonGreen expression and GGG-mNeonGreen purification. Both were highly successful (Figure 20) as evidenced by the fluorescent yellow pellets and cell lysates.

Several variants of the sortase enzyme are commercially available. We selected a sortase pentamutant (Addgene #75144), which was engineered via directed evolution to have 140-fold enhancements in ligation activity (Chen et al., 2011). This sortase mutant is translationally fused to an N-terminal polyhistidine tag. This tag was used to perform Ni-NTA affinity chromatography on lysates of cells overexpressing the enzyme. Lysates were incubated with the resin for 3 hours in a binding buffer (50mM Tris pH 7.5, 300mM NaCl, 10mM imidazole). The unbound lysate was collected as a flow-through sample. The resin was washed 3 times with a washing buffer (50mM Tris pH 7.5, 300mM NaCl, 20mM imidazole). Finally, bound proteins were eluted multiple times using an elution buffer (50mM Tris pH 7.5, 300mM NaCl, 500mM imidazole). Aliquots were spared at each stage and run on an SDS-PAGE gel (Figure 8A). The eluted samples were combined and dialysed overnight with the buffer for size exclusion chromatography (25 mM tris pH 7.5, 150mM NaCl). Before size exclusion chromatography (SEC), the sample was concentrated to a volume less than 5mL, suitable for input into the SEC column. The absorbance at 280 nm for each 1mL SEC fraction was measured. Fractions with peaks at the expected size of sortase were collected and run in an SDS-PAGE gel to confirm the presence of sortase A (Figure 8B). SEC visibly removed contaminants enriched during the Ni-NTA chromatography. The pure sortase A samples were then combined and concentrated, forming our stock sortase solution for subsequent conjugation reactions.

Coincidentally, as we were conducting our conjugation experiments, we discovered that a member of the Structural Biology Group (the group that supplied one of the lab spaces that we used to conduct our experiments) had also been working on protein conjugation using sortase A. As such, we were granted the opportunity to learn some more specific details about sortase A conjugation. In particular, we discovered that sortase A conjugation is a highly reversible reaction and that the most efficient method of pushing the equilibrium to completion is to incorporate an additional GHHHHHH motif following the LPETG sequence (Reed et al., 2020). This would result in cleavage of a 6xHis tag which could then be sequestered by the addition of free Ni2+ ions in solution. Removing one of the products in this manner would greatly inhibit the reverse sortase reaction, resulting in significantly improved conjugation efficiency.

Unfortunately, at this point, we had already designed and produced the LPETG-modified Reb1 constructs and did not have enough time to re-make our plasmids. Nevertheless, Sebastian provided us with a positive control substrate (a truncation of the protein CHD4 - spanning residues 1380-1810 - which possessed a C-terminal LPETG motif) for us to test our sortase enzyme sample with.

We performed a conjugation test using our purified eSrtA sample, GGG-mNeonGreen and Sebastian’s positive control substrate. The reaction was carried out with 10 µM sortase enzyme, 100 µM GGG-mNeonGreen and 100 µM LPETG-substrate, and allowed to incubate at room temperature for 1 hour before being moved to the cold room (4°C) overnight to prevent substrate degradation. The products were analysed by SDS-PAGE (Figure 23), which revealed successful conjugation - evidenced by the appearance of a higher molecular weight product at ~66 kDa, corresponding to the sum of the weights of mNeonGreen (~27 kDa) and the LPETG-substrate (38 kDa) - albeit with very low efficiency (likely < 5%). Despite the low efficiency, this still confirmed that both our sortase and mNeonGreen samples appeared functional for conjugation, giving us confidence to proceed with R body conjugation trials - although with the knowledge that we would have to add a large excess of substrate in an attempt to push the equilibrium forward.

The method for sortase A ligation onto LPETG-modified R bodies can be found on the Protocols page. In short, we calculated the number of LPETG binding sites using an approximate R body pellet mass and the molar mass of a monomer, and added more than 10-fold molar excess of GGG-mNeonGreen to drive the reaction to completion. This immediately yielded bright yellow pellets and fluorescent R bodies clearly distinguishable under a microscope (Figure 23).

Our success with sortase A conjugation is a novel result. It demonstrates successful aqueous-phase bioconjugation to a resuspended insoluble protein, which is typically hindered by limited solvent access to reactive sites and aggregation. R bodies are also one of the largest (if not the largest) natural protein assemblies on which sortase conjugation has been attempted to the extent of our knowledge. For example, Kim et al. (2015) describe “large” 40 kDa protein conjugation, whereas for comparison Cai’s (2023) physical model describes over 800,000 10 kDa monomers being packed into a single R body ribbon. We believe that demonstrating efficient conjugation at this scale demonstrates the feasibility of bioconjugation in a wider range of contexts than previously assumed.

---

In Vitro Application

In order to assess endocytic efficiency we collated a large number of images that demonstrate RebA-LPETGGG-mNeonGreen endocytosis into EXPI293 at different dilutions and incubation times. We pelleted 2 mL of resuspended R bodies (see protocol) and diluted it 1:8 and 1:16 then incubated for 24h or 48h in an EXPI293 cell suspension. The cells were visualised under a Zeiss Axio Imager Z1 in trypan blue, which quenches extracellular fluorescence (Figure 24).

The mean number of R bodies per cell scaled with dilution factor averaged 4 R bodies per cell (when a 2 mL pellet was diluted 1:8 times). However, the mean number of R bodies yields a misleading impression: as seen in Figure 14 the number of R bodies per cell follows a bimodal distribution. At high concentration (1:8 dilution), a majority of cells internalised 5-10 R bodies and the remaining proportion internalised 0-1 R bodies (see Figure 15D). The cause and therapeutic implications of this heterogeneity merit further investigation. Figure 24A demonstrates the intrinsically non-cytotoxic nature of our platform, matching reports by Pond et al. (1989). We conclude that the cytotoxic activity of R bodies should depend on what cargo it carries, mimicking their role as natural toxin delivery systems in Paramecia spp.. Further, the average intracellular accumulation is ostensibly dose-dependent, which is highly promising for therapeutic application. We also believe the potential of R bodies to facilitate internalisation of membrane-impermeant molecules is highly compatible with the development of large macromolecules such as cyclic peptides into therapeutic options.

We also tested an aldoxorubicin-Cys-N RebB conjugate in vitro. Our original intention was to use a membrane-impermeant cargo molecule attached to R bodies by an acid-labile or protease-cleavable linker. This would allow us to assess whether R bodies enable endosomal escape in vitro (as the endosome acidifies or matures into an endolysosome the cargo should be released after which the R body ruptures the endosomal membrane). However, as deadlines approach there were no candidate molecules that could ship to Australia in time. Aldoxorubicin was available, had a maleimide group and was fluorescent. Effectively, the aldoxorubicin experiment was a repetition of the RebA-LPETGGG-mNeonGreen experiment (Figure 26), the key difference being the presence of an acid-labile linker which would release aldoxorubicin from R bodies in the endosome.

We wanted to know whether the diffuse intracellular fluorescence observed in EXPI293 cells could be attributed to cleavage of aldoxorubicin at its acid-labile hydrazone linker. Therefore we imaged a resuspended sample of aldoxorubicin-Cys-N RebB R bodies that had been stored at room temperature for 24 h in MQW. We observed diffuse fluorescence throughout the solvent which suggested that premature cleavage was releasing aldoxorubicin into solution. This instability is a notorious feature of hydrazone groups (Yousefpour et al., 2019). Aldoxorubicin is highly membrane permeant. Therefore we could not conclude that endosomal release (and subsequent escape) of aldoxorubicin was responsible for the intracellular concentrations observed in EXPI293 cells. Ultimately, we could only conclude that R bodies could deliver aldoxorubicin into cell vesicles.

---

References

Bartoschek, M. D., Ugur, E., Nguyen, T. A., Rodschinka, G., Wierer, M., Lang, K., & Bultmann, S. 2021. Identification of permissive amber suppression sites for efficient non-canonical amino acid incorporation in mammalian cells. Nucleic Acids Research. 49(11): e62. https://doi.org/10.1093/nar/gkab132.

Cai, G. (2023). An Interdisciplinary Investigation of Conformational Changes in Quasi-Crystalline Protein Array R-Bodies in Response to pH [Ph.D., University of Washington]. In ProQuest Dissertations and Theses (2837944634). ProQuest One Academic. https://www.proquest.com/dissertations-theses/interdisciplinary-investigation-conformational/docview/2837944634/se-2?accountid=14757

Chae, P. S., Wander, M. J., Cho, K. H., Laible, P. D., & Gellman, S. H. 2013. Carbohydrate-containing Triton X-100 analogues for membrane protein solubilization and stabilization. Molecular BioSystems. 9(4): 626-629. https://doi.org/10.1039/c3mb25584k.

Chen, I., Dorr, B. M., & Liu, D. R. 2011. A general strategy for the evolution of bond-forming enzymes using yeast display. Proceedings of the National Academy of Sciences of the United States of America. 108(28): 11399-11404. https://doi.org/10.1073/pnas.1101046108.

Lalucat, J. 1988. 4 Analysis of refractile (R) bodies. Methods in Microbiology. 20: 79-90. https://doi.org/10.1016/S0580-9517(08)70048-5.

Ma, X., Wei, B., & Wang, E. 2022. Efficient incorporation of p-azido-l-phenylalanine into the protein using organic solvents. Protein Expression and Purification. 200: 106158. https://doi.org/10.1016/j.pep.2022.106158.

Polka, J. K., & Silver, P. A. 2016. A tunable protein piston that breaks membranes to release encapsulated cargo. ACS Synthetic Biology. 5(4): 303–311. https://doi.org/10.1021/acssynbio.5b00237.

Reed, S. A., Brzovic, D. A., Takasaki, S. S., Boyko, K. V., & Antos, J. M. 2020. Efficient sortase-mediated ligation using a common C-terminal fusion tag. Bioconjugate Chemistry. 31(5). https://doi.org/10.1021/acs.bioconjchem.0c00156.

Rostovtsev, V. V., Green, L. G., Fokin, V. V., & Sharpless, K. B. 2002. A stepwise Huisgen cycloaddition process: Copper(I)-catalyzed regioselective “ligation” of azides and terminal alkynes. Angewandte Chemie International Edition. 41(14): 2596-2599. DOI: https://doi.org10.1002/1521-3773(20020715)41:14<2596::AID-ANIE2596>3.0.CO;2-4

Schrallhammer, M., Galati, S., Altenbuchner, J., Schweikert, M., Görtz, H. D., & Petroni, G. 2012. Tracing the role of R-bodies in the killer trait: Absence of toxicity of R-body producing recombinant E. coli on paramecia. European Journal of Protistology. 48(4): 290-296. https://doi.org/10.1016/j.ejop.2012.01.008.