Enhanced protein encapsulation in polymer vesicles with hydrophobic membrane-anchor peptides

Document Type
Doctoral Thesis
Issue Date
Issue Year
Mertz, Michael

Over the last two decades, hollow spherical particles made from single or double layer(s) of block-copolymers, called polymersomes, have been used in various concepts from nano-medicine over bio-catalysis to synthetic biology. So far, most studies have addressed the proofs of concepts, laying out a great number of promising designs. However, further work is needed to bring polymersomes to a wider application. One critical aspect is the production of polymersomes in a scalable manner. Recent work on this topic resulted in the establishment and thorough characterization of a polymersome formation process in stirred tank reactors (STR). The defined environment of the reactor allowed scaling polymersome production from the milliliter- to the liter scale (Poschenrieder et al. 2017b). The second critical issue is the functionalization of polymersomes with proteins. Proteins can be located either outside the polymersomes (outer compartment), within the polymer membrane, or in the vesicle lumen (inner compartment). The drawback of polymersome production in STRs is the low encapsulated volume that forms the inner compartment. For the triblock-copolymer poly(2-methoxazoline)15-b-poly(dimethylsiloxane)68-b-poly(2-methoxazoline)15 (PMOXA15-PDMS68-PMOXA15) the encapsulated volume is only 0.53 %, resulting in a low upper limit for statistical protein encapsulation (of 0.53 %). Since hydrophobic proteins or proteins with hydrophobic membrane anchors can exert attraction to the hydrophobic phase of the polymer membrane, the combination of the enhanced green fluorescent protein (eGFP) with C-terminal membrane anchoring peptides from rabbit liver cytochrome b5 (Cytb5’), the SM2 phage lysis protein L (L’), the syntaxin Vam3p (Vam3p’) and ubiquitin-conjugating enzyme 6 (UBC6’) from yeast were used to test the capability of the peptide anchoring peptides to increase protein encapsulation in PMOXA-PDMS-PMOXA polymersomes. Polymersomes were formed in the presence of all four, membrane anchoring eGFP variants, and analysis of the samples by size-exclusion chromatography and fluorescence measurement demonstrated the increased attraction of eGFP with membrane anchors to the polymersomes in contrast to the protein without membrane anchors. Quantification of the proteins that were immobilized per polymersome revealed that the polymersomes were functionalized with up to 60.46 % of the applied protein (functionalization efficiency, FE%). All four membrane anchors showed good overall FE% of the applied protein, with UBC6’ presenting itself as the most efficient (FE% between 39.70 to 60.46 %). Figure 8.1 gives an overview of the functionalization (Figure 8.1 A and B) and encapsulation (Figure 8.1 C and D) of polymersomes with soluble proteins and proteins with hydrophobic membrane anchors. Regarding the encapsulation of proteins in the lumen of polymersomes, the moderate hydrophobic peptides Cytb5’ and L’ enabled encapsulation of around 40 eGFP per polymersome (43.6 and 39.4 respectively) at 0.78 and 0.75 g L-1 protein. The shorter and highly hydrophobic peptides Vam3p’ and UBC6’ encapsulated even higher numbers of 78.3 and 89.9 eGFP per polymersome at 0.40 g L-1 protein. Factoring in the respective protein concentrations applied, this increased protein encapsulation by 299.3 fold compared to the statistical encapsulation of eGFP.

Figure 8.1: Adapted from (Mertz et al. 2021). A: schematic representation of polymersome functionalization. Proteins with hydrophobic membrane anchors (left) or soluble proteins (right) were mixed with the polymer solution during polymersome formation. Proteins with hydrophobic membrane anchors were integrated into the polymer membrane, and soluble proteins were encapsulated in the polymersome lumen. Not integrated proteins were then removed by size-exclusion chromatography (SEC). B: functionalization of polymersomes with eGFP or eGFP with membrane anchors at a protein concentration of 0.5 g L-1 (except for UBC6’ where 0.4 g L 1 was used). C: Schematic representation of the analysis of encapsulated proteins in polymersomes with membrane anchoring eGFP. After functionalization of the polymersomes during vesicle formation, the proteins on the outer surface of the polymersomes were digested by proteinase K. Polymersomes were then separated from the digested proteins by SEC. D: number of proteins per polymersome when encapsulating soluble eGFP and eGFP with hydrophobic membrane anchors in polymersomes. The numbers given in D represent the highest results for the individual protein species achieved in the experiments by using the following protein concentrations: L’ 0.75 g L 1, Cytb5’ 0.78 g L-1, Vam3p’ 0.30 g L-1, UBC6’ 0.40 g L-1, and eGFP 4.30 g L-1.

The formation of polymersomes in the presence of membrane anchoring eGFP resulted, however, in polymersomes that present proteins on both the inner- and outer surface. For a specific removal of the protein domains from the outer polymer surface, an intein domain, as well as a TEV protease recognition site, were introduced into the membrane anchoring proteins. In principle, the intein domain can give the additional benefit to release protein domains immobilized to the inner surface from the polymer membrane since small uncharged molecules like the thiol dithiothreitol (DTT) could diffuse across the polymer membrane. The fusion protein of eGFP, the intein domain from Mycobacterium xenopi gyrase A and the membrane anchor Cytb5’, eGFP-Int-Cytb5’, could be spliced with 50 mM DTT, and splicing was not majorly impacted by the temperature. Because the amino acid before the intein (-1) can significantly affect the splicing efficiency, it was optimized by saturation mutagenesis. For eGFP-Int-Cytb5’ asparagine showed the best results, showing low unspecific splicing in the absence of DTT and the highest percentage of spliced protein when DTT was added. The protein could be spliced in solution as well as when it was immobilized to the surface of polymersomes. However, the immobilization resulted in slower splicing. Whereas only around 10 % of unspliced protein remained after 4 h when eGFP-Int-Cytb5’ was in solution, the immobilized protein was reduced over 48 h only to 50.4 %. The passive diffusion of DTT across the polymersome membrane was tested to see if splicing within polymersomes could be possible. This is particularly of interest when no protein channels are integrated into the polymer membrane. The results of the experiments indicated that DTT diffused rapidly over the polymersome membrane within the time frame of a size exclusion chromatography run (5 to 7 minutes). It is, therefore, likely that the PMOXA-PDMS-PMOXA polymer membrane does not present a high diffusion barrier for DTT. A downside of the integration of the intein domain into the protein presented itself within the protein encapsulation in polymersomes. It was drastically reduced compared to the protein without an intein domain (eGFP-Cytb5’). The reduction was about 3.5 times lower, managing only 4.6 encapsulated proteins per polymersome with 0.45 g L-1 eGFP-Int-Cytb5’. Inclusion of a TEV protease recognition sequence between the eGFP and Cytb5’ membrane anchor, flanked by flexible (GSSSS; GS) or rigid (EAAAK; EA) linkers, created a number of fusion proteins that could be specifically cleaved via proteolysis with TEV protease. When proteins with the GS and EA linkers and linker lengths of one to three repeats were immobilized to polymersomes, the removal of the eGFP domain with TEV protease was demonstrated. Immobilized proteins with the flexible linker were cleaved more efficiently than the ones with the rigid linker. The protein with the shortest flexible linker, one repeat of GS (eGFP-GS1/TEV-Cytb5’), was removed almost completely from the polymersome surface (98.9 %). This construct was additionally tested for protein encapsulation, where it performed comparably to the eGFP-Cytb5’ construct it was derived from. With eGFP-GS1/TEV-Cytb5’, up to 35.1 +/- 1.7 proteins per polymersome were encapsulated at 0.60 g L-1 as opposed to 28.4 +/- 9.4 eGFP-Cytb5’ per polymersome at 0.58 g L-1.

After exploring protein encapsulation and the release of protein domains from the polymersome surface with fusion proteins of eGFP, the domain was exchanged for the enzyme N-acetylglucosamine 2-epimerase (AGE) K160I. The activity of the fusion enzyme with the intein and Cytb5’ membrane anchor (AGE(K160I)-Int-Cytb5’) was reduced by 56.8 % compared to AGE K160I. When splicing was induced by DTT, separation of the enzyme was observed on SDS-PAGE, but the activity did not increase, as would be expected from the release of the AGE K160I domain. The reduced activity did not show as a result from the addition of three amino acids (GSN), which remained at the C-terminus after splicing but were not present in AGE K160I. To see if protein folding was impaired by expression in the fusion enzyme, circular dichroism (CD) spectroscopy was used to measure AGE K160I and the AGE K160I domain after splicing from the full fusion enzyme. The CD measurements did not show a distinctly different spectrum compared to the one of soluble AGE K160I, indicating that overall, the AGE domain in the AGE-Int-Cytb5’ fusion protein was at least partially folded. However, minor deviations in protein folding could have been present, which might not have been detectable by this method. The fusion enzymes of AGE K160I with TEV protease cleavable GS and EA linker and the Cytb5’ membrane anchor had reduced enzymatic activity compared to soluble AGE K160I. The rigid EA linker retained more activity than the flexible GS linkers. The best construct was AGE(K160I)-EA3/TEV-Cytb5’ (three repeats of the EAAAK linker sequence), which had 83.2 % of AGE K160I’s activity. Effects like the insertion of the Cytb5’ membrane anchor into the polymer membrane showed only minor changes in activity with increases between 2.5 and 27.4 %. Proteolysis of the membrane anchor, on the other hand, had varying effects on protein activity from decreasing activity (enzymes with EA2 and GS2 linker), little change (EA3 linker), and activity increases (EA1, GS1, and GS3 linkers).

In summary, it was demonstrated that the combination of proteins with C-terminal hydrophobic membrane anchoring peptides could increase protein encapsulation in PMOXA-PDMS-PMOXA polymersomes. Protein constructs, which can selectively release the N-terminal domain of the immobilized protein from the polymer surface, were created. Lastly, the exchange of the N-terminal protein domain (eGFP) for a catalytic active enzyme domain (AGE K160I) was demonstrated. However, the fusion protein of AGE K160I and the Cytb5’ membrane anchor could not be expressed and purified in sufficient amounts to perform a series of encapsulation experiments in 12 mL STRs. Encapsulation of AGE-(K160I)-EA3-Cytb5’ at a protein concentration of 0.35 g L-1 was tested, but this only resulted in polymersomes with around one-third of the catalytic activity compared to polymersomes encapsulating soluble AGE K160I at 1.2 g L-1 protein concentration. Although no sufficient amount of the fusion enzymes could be produced to test encapsulation at the appropriate protein concentrations to create polymersomes with higher catalytic activity compared to the encapsulation of soluble proteins, the principle was demonstrated.

Faculties & Collections
Zugehörige ORCIDs