Controlling the self-assembly of protein polymers via heterodimer-forming modules

Supramolecular assemblies formed by protein polymers are attractive candidates for future biomaterials. Ideally, one would like to be able to define the nanostructure, in which the protein polymers should self-assemble, and then design protein polymer sequences that assemble exactly into such nanostructures. Despite progress towards ‘programmability’ of protein polymer self-assembly, we do not yet have such control. This holds especially for hierarchical structures such as self-assembled fibril bundles, where one would like to have independent control over the structures at the different length-scales. In this thesis we explore the use of heterodimerization as a strategy to control self-assembly of protein polymers at multiple length-scales. We tested a selected set of heterodimer-forming peptide modules. The heterodimer-forming modules are genetically incorporated at the C-terminus of protein polymers with a previously characterized self-assembly behavior. Several newly constructed protein polymers were biosynthesized in the yeast Pichia pastoris and, for these new protein polymers we investigated whether the inclusion of the heterodimer-forming blocks improved the control over the assembly of nanostructures.   

The incorporation of heterodimer-forming modules into protein polymers is not the only tool that can be used for improving programmability of assembly. In Chapter 2 we present an overview of several tools that can be use, and we highlighted their advantages and disadvantages.     

In Chapter 3 we test de novo designed heterodimerizing coiled coils DA = LEIRAAFLRQRNTALRTEVAELEQEVQRLENEVSQYETRYGPLGGGK and DB = LEIEAAFLERENTALETRVAELRQRVQRLRNRVSQYRTRYGPLGGGK. These peptides were fused to hydrophilic random coil protein polymer (CP4) and homotrimer forming protein polymer (T9-CP4). We present data on the production, characterization and functionality for four new protein polymers: CP4-DA, CP4-DB, T9-CP4-DA and T9-CP4-DB. When the new protein polymers were produced using the fermentation process established previously for other protein polymers such as CP4 (i.e. standard fermentation), we found the new protein polymers to be partly degraded. The use of a protease deficient strain, as well as changes in aeration or pH were found ineffective in preventing degradation, but nearly intact products were obtained from a fermentation in which the induction was done at 20 ˚C and in which the medium was supplemented with casamino acids. With respect to the physical properties of the new protein polymers, size exclusion chromatography (SEC) showed that an equimolar mixture of CP4-DA and CP4-DB contained mostly dimers, whereas unmixed CP4-DA and CP4-DB contained only monomers. However, we also found that CP4-DB forms homooligomers at concentrations ≥100 µM. A mixture of T9-CP4-DA and T9-CP4-DB forms a hydrogel, most probably due to both homotypic and heterotypic DA/DB associations. We conclude that when used at low concentration, this pair of coiled coils seems to be suitable to control self-assembly of protein polymers produced in Pichia Pastoris.   

Next, in Chapter 4 we test another pair of de novo designed coiled coils. These are much shorter and have lower reported values of the association constant as compared to the DA/DB coiled coils. The systems consist of a peptide DE = (EIAALEK)3 and a peptide DK = (KIAALKE)3. The two peptides were C-terminally fused to protein polymers CP4 and T9-CP4. The standard fermentations resulted in intact CP4-DE and T9-CP4-DE, but protein polymers CP4-DK and T9-CP4-DK were found to be partly degraded. The degradation of variants with DK module could not be readily resolved by fermentation at higher pH or using proteases deficient strain. For CP4-DK, ion exchange chromatography showed that about 40% of protein polymer (by mass) was intact. We find that for this pair of coiled-coils, homotypic interactions are so strong that they can drive gel formation in the case of T9-CP4-DE, and a strong increase in viscosity for T9-CP4-DK. Mixtures of the complimentary triblocks also form hydrogels, but it is not yet clear to what extent this is due to homotypic DE/ DE and DK/ DK associations, and to what extent it is due to DE/ DK heterodimer formation.  

A very different type of heterodimer-forming block is the so-called WW domain that is found in many natural proteins, and which forms heterodimers with proline-rich peptides PPxY. In Chapter 5 we test the interaction between a naturally occurring WW domain (DWW) and its proline-rich ligand (DPPxY). Both were C-terminally fused to the hydrophilic random coil protein polymer CP4. The new protein polymers CP4-DWW and CP4-DPPxY were produced intact during standard fermentations, but CP4-DPPxY was shown to be glycosylated. Using genetic engineering, we mutated the CP4-DPPxY protein polymer sequence by the substitution Ser12→Ala. A standard fermentation resulted in an intact and non-glycosylated protein polymer CP4-DPPxY*. Interaction studies (ITC and steady state tryptophan fluorescence quenching), showed that both CP4-DPPxY and CP4-DPPxY* bind to CP4-DWW with an equilibrium dissociation constant on the order of mM.  

Finally, to demonstrate that heterodimer-forming blocks can be used to independently control protein polymer self-assembly at multiple length-scales, we selected the heterodimer-forming modules DA and DB to control the lateral interactions of fibrils self-assembled from the previously designed triblock protein polymer C2-SH48-C2. In Chapter 6 we construct the protein polymers C2-SH48-C2-DA and C2-SH48-C2-DB. The C2-SH48-C2 protein polymers assemble into long and stiff fibrils at neutral pH. The aim of the C-terminal attachment of the DA/DB blocks was to be able to control subsequent physical cross-linking and bundling of the fibrils. Both protein polymers C2-SH48-C2-DA and C2-SH48-C2-DB were produced intact and with high yield during fermentation at optimal conditions as discussed in Chapter 3. Using Atomic Force Microscopy (AFM) we show that at neutral pH, fibrils consisting of 100% C2-SH48-C2-DA or C2-SH48-C2-DB protein polymers bundle up and cross-link via homotypic DA/DA and DB/DB associations. Control over the degree of cross-linking and bundling can be obtained by using mixed fibrils consisting of C2-SH48-C2 with controlled amounts of the newly developed protein polymers C2-SH48-C2-DA and C2-SH48-C2-DB. While the effect of the heterodimers on the structure of the fibril network as judged from AFM is very strong, oscillation rheology shows that the inclusion of the heterodimer forming blocks merely leads to a moderate increase in gel stiffness.       

In order to place the research discussed in this thesis into the broader perspective, in Chapter 7 we provide a General Discussion. We discuss several general strategies that can be used to control protein polymer self-assembly and discuss why and when there is a need for using heterodimer forming blocks. After providing an overview over results obtained in this thesis, we highlight the most urgent questions that need to be answered next. This is followed by a discussion on the benefits that heterodimer-driven self-assembly may bring to possible future applications of protein polymers as biomaterials. We also discuss the possible risks for human health end environment that might arise from the use of protein polymers technology. Finally we present some speculations about the future of the field of self-assembling protein polymers.