Giant Polymersome Protocells Dock with Virus Particle Mimics via Multivalent Glycan-Lectin Interactions

Despite the low complexity of their components, several simple physical systems, including microspheres, coacervate droplets and phospholipid membrane structures (liposomes), have been suggested as protocell models. These, however, lack key cellular characteristics, such as the ability to replicate or to dock with extracellular species. Here, we report a simple method for the de novo creation of synthetic cell mimics in the form of giant polymeric vesicles (polymersomes), which are capable of behavior approaching that of living cells. These polymersomes form by self-assembly, under electroformation conditions, of amphiphilic, glycosylated block copolymers in aqueous solution. The glycosylated exterior of the resulting polymeric giant unilamellar vesicles (GUVs) allows their selective interaction with carbohydrate-binding receptor-functionalized particles, in a manner reminiscent of the cell-surface docking of virus particles. We believe that this is the first example of a simple protocell model displaying cell-like behavior through a native receptor-ligand interaction.

have been developed. Marguet et al. 15 combined both concepts of compartmentalization and a gel cavity in vesicles to achieve a more structurally advanced cell model.
The second rational step towards cell biomimicry is to introduce some "living" functional aspects (such as metabolism, replication or adaptability) to the existing cellular structural models. One such aspect is cellular internalization, in which cells take up a variety of external species including macromolecules, nanoparticles (e.g. viruses) and bacteria. Internalization occurs by various mechanisms, including endocytosis, the key stage in which is the docking of an external species to the cell membrane, followed by an invagination of the fluid bilayer and complete wrapping of the species in question and ultimately its transportation to the intracellular milieu encapsulated within a vesicle 16,17 . A sub-set of different endocytosis mechanisms is initiated by specific ligand-receptor interactions 18 . These receptor-mediated endocytosis (RME) processes are used by the cell to internalize a variety of nutrients, hormones, growth factors and other macromolecules, and are exploited by viruses as a means to gain entry into the cell 19 .
Carbohydrates are commonly encountered ligands for cell surface receptor proteins (lectins) and, indeed, many biological processes in mammalian cells, such as initiation of the inflammatory cascade, virus docking, fertilization and cancer cell metastasis, are mediated by carbohydrate-lectin interactions 20,21 . In many cases, carbohydrate-lectin binding leads to RME and internalization of the sugar-bearing cargo. Sugar-lectin binding typically displays high specificity despite the fact that interactions between individual sugars and lectins are unusually weak (K a ca. 10 3 M −1 ) 22 . This high specificity occurs through the 'cluster glycoside' effect, whereby many copies of the same sugar are presented to the lectin, leading to much higher K a values (10 9 -10 12 M −1 ) 23 . Consequently, multivalent glycosylated macromolecules, such as dendrimers (glycodendrimers) and linear polymers (glycopolymers), bearing many copies of the same sugar 24 , have been demonstrated to give binding to lectins that is massively enhanced compared to the individual sugar 23,25 .
At present, no structural cell mimics that can interact specifically with extracellular species in solution via receptor-ligand binding have been reported. Successful internalization of nanoparticles into liposomes 26 and polymersomes 27 has been shown recently as an attempt to mimic the phagocytosis process of living cells. However, in both cases, an external stimulus, such as a large concentration gradient 27 or an optical trap 26 , was required to induce the uptake process. Here, we present the spontaneous and selective interaction between stable and robust cell-sized polymersomes, which have sugar moieties presented on their surface, and lectin-functionalized particles. The polymersomes are formed by self-assembly of amphiphilic glycopolymers, which were prepared using the RAFT 28 polymerization technique.

Results and Discussion
We first utilized RAFT to polymerize an activated ester monomer, pentafluorophenyl acrylate (PFPA), followed by chain extension with n-butyl acrylate (n-BA) to produce a reactive block copolymer precursor for subsequent modification with amine-functionalized sugars (Fig. 1A). PFPA was first polymerized using benzyl 2-hydroxyethyl carbonotrithioate (BHECTT) as a chain transfer agent (CTA) ( Table S1). The P(PFPA) as mac-roRAFT agents were used to polymerize n-BA to produce block copolymers with different compositions. After purification by reprecipitation, the block copolymers were analyzed by SEC which showed a monomodal distribution with dispersities of ca. 1.2 (Table S2). Prior to coupling with aminoethyl glucoside, the CTA end group was removed by treatment with AIBN. Under optimized experimental conditions, high yields with total consumption of pentafluorophenyl ester as revealed by 19 F-NMR spectroscopy, were achieved. Further evidence of successful attachment of the sugar moieties was provided by FTIR spectroscopy (see SI). Giant vesicles were prepared by self-assembly of the amphiphilic p(Nβ GluEAM-b-BA) glycopolymers using the electro-formation method (Fig. 1B), which has been shown to be efficient for producing giant unilamellar vesicles (GUVs) in high yields with narrow size distribution and few defect structures 29,30 . An AC field was applied across a conducting substrate onto which the glycopolymer was coated, causing vesicles to bud off from the surface. Application of optimized electro-formation conditions on one of the synthesized glycopolymers, namely p(Nβ GluEAM 5 -b-BA 50 ), led to the formation of stable glycosylated GUVs (glyco-GUVs) with high yields (77 ± 8 vesicles per square mm) and average diameter of 20.0 ± 2.0 μ m (Fig. 1C,D).
In order to utilize these glyco-GUVs as cell mimics, we needed to understand their response to changeable environmental conditions and permeability to various substances. We found that the glyco-GUVs responded to changing osmotic pressure; hypertonic conditions trigger shrinking of the vesicles while hypotonic conditions induce swelling. The glyco-GUVs are approximately 2.5 times more susceptible to negative osmotic pressure than positive. The average vesicle diameter decreases linearly by 19.7 ± 2.0% with an increase of negative osmotic pressure to − 24.4 atm; however an increase in negative osmotic pressure beyond this value does not induce further changes in the average diameter of vesicles. Vesicles are able to withstand a negative osmotic shock higher than − 24.4 atm and adapt to the altered osmolality; however, upon applying an osmotic shock lower than − 24.4 atm the majority of the glyco-GUV population collapses and the remainder adjusts their average diameter to reduce the osmotic gradient.
Before employing these glyco-GUVs in interaction studies with receptor (lectin) -functionalized particles, it was necessary to demonstrate the availability of the pendent glucose moieties present on the vesicles' surface for lectin binding. A turbidity assay was performed whereby 240 μ l of a GUV solution was added to 600 μ l of a Concanavalin A (Con A) solution in HEPES buffer (2 mg/mL). A steady increase in A 450nm was observed over 60 minutes caused by increasing sample turbidity ( Figure S9). This is caused by agglomeration of glyco-GUVs, which present a multivalent display of glucose units to Con A which is itself multivalent (a tetramer at pH = 7.4).
Con A-functionalized polystyrene (PS) beads were prepared as model extracellular receptor functionalized species to study their binding interactions with our glyco-GUVs ( Fig. 2A,B). Commercially available carboxylate-modified PS beads were conjugated with Con A using carbodiimide coupling chemistry. Con A has a strong affinity for glucose-containing glyco-conjugates 31 . In order to probe the specificity of interactions between Scientific RepoRts | 6:32414 | DOI: 10.1038/srep32414 Con A-functionalized PS beads and glycopolymers, we conducted a microscopic assay whereby we added an aqueous solution of glucose-or fucose-containing multivalent glycopolymers to a suspension of Con A-functionalized PS beads in HEPES buffer (fucose has no binding affinity for Con A). On addition of the glucosidic polymer, the lectin-functionalized PS beads were seen to agglomerate very rapidly; conversely, on addition of the fucosidic polymer, no change in the agglomerated status of the beads was apparent (Fig. 2C-F). This agglomeration is due to specific binding interactions between the glucoside and subsequently potential crosslinking. The experiment was repeated using the carboxylate-modified PS beads, whereupon no agglomeration occurred, confirming that binding is caused specifically by carbohydrate-lectin interactions (Fig. 2C-F).
We next studied the interaction between our glyco-GUVs and Con A-functionalized PS beads as model extracellular objects. Confocal microscopy was used to visualize the interactions. In order to eliminate any potential errors and misinterpretations of data produced by non -lectin mediated interactions, two types of control experiments were performed: glyco-GUVs incubated with unfunctionalized PS beads (the original carboxylate-modified PS beads); and glyco-GUVs incubated with RCA 120 -functionalized PS beads (RCA 120 has no affinity to β -linked glucose moieties). All experiments were replicated in triplicate with an incubation time of 18 h, to allow significant numbers of interactions between beads and GUVs to occur. Upon overnight incubation of the glyco-GUVs with the unfunctionalized PS beads, very few examples of a bead next to a GUV were observed; however, the majority of the beads were distributed randomly and remained at the bottom of the visualization chamber. The percentage of interaction between the glyco-GUVs and the unfunctionalized beads, defined as the percentage of glyco-GUVs with an adjacent bead, did not exceed 6.5% in each of the observed samples. Similarly, upon overnight incubation of the glyco-GUVs with the RCA 120 -functionalized PS beads, a small number of interactions between the two species were observed; however the majority of RCA 120 -functionalized PS beads were dispersed randomly in the sample. The percentage of interaction between the glyco-GUVs and the RCA 120 -functionalized PS beads varied from 6 to 9%, which is slightly higher than that determined for the unfunctionalized PS beads.
Following these control experiments, we incubated our glyco-GUVs with the Con A-functionalized PS beads under the same conditions used for the control experiments. We observed in this case many examples whereby a bead appeared to attach to the surface of a glyco-GUV. Repeat experiments (n = 4) gave consistent results. Based on the collected data, the average percent of interaction between the glyco-GUVs and the Con A -functionalized PS beads was determined to be 42.0 ± 7.8% which is approximately five times higher than those with the RCA 120 -functionalized PS (8.2 ± 1.4%) and eight times higher than those with the unfunctionalized PS beads (4.9 ± 1.0%) (Fig. 3A). The strength and stability of the ligand -receptor interactions was assessed by recording the behavior of the species over a period of time. Figure 3B-D shows a glyco-GUV that is attached to a group of beads via a single bead -GUV connection. We presume that bead aggregation is caused by some free glycosylated polymer chains or nanostructures (eg micelles) that are too small to be observed by confocal microscopy. Time-lapse images show that the beads and GUVs move in concert, demonstrating that the strength and stability of the sugar-lectin binding interaction is sufficient to withstand translation from Brownian motion. Furthermore, the precise location of beads relative to GUVs was investigated by microscopy. Successive confocal microscopy images at different focal planes (Z-stack images) indicated that beads located adjacent to GUVs were indeed interacting strongly with the vesicle membrane ( Fig. 3E-J). As the focal plane is lowered from roughly mid-way through the large GUV in the centre of the image (Fig. 3E), the bead appears (Fig. 3F) then increases in intensity (Fig. 3G), indicating that the bead is located next to the lower half of the GUV. Also seen in these images is a smaller GUV interacting with a bead (Fig. 3F,G -lower right, arrow). Evidence of a bead becoming embedded in a GUV membrane is presented in Fig. 3H-J (in the video in the SI, the GUV attempts to engulf the bead). At the lowest focal plane, it appears that the bead is to some extent buried in the GUV membrane (Fig. 3J). It should be noted that GUV aggregation induced by lectin-coated beads is unlikely due to the restricted motion of the GUVs in the confocal visualisation chamber.
There are four possible locations of beads relative to GUVs (Fig. 4). GUVs have an internal aqueous pool consisting of a sucrose solution which causes them to sink to the bottom of the viewing chamber and so the GUVs rest on a substrate. We expect that confocal microscopy would easily reveal when beads are well-separated from GUVs (Fig. 4A). Beads internalized by GUVs (Fig. 4B) would be revealed by confocal microscopy at a focal plane mid-way through the GUV. An image in which the bead is clearly within the GUV membrane would be expected if internalization occurred. There is no clear evidence for such internalization in Fig. 3. A bead may be located adjacent to the GUV membrane whilst also resting on the substrate (Fig. 4C). We suspect that this is the situation described by Fig. 3E-G, where the fluorescence intensity of the bead is greatest at the lowest focal plane. The final possible orientation is when the bead is embedded in the GUV membrane, but not necessarily resting on the visualization chamber surface (Fig. 4D). Evidence for this relative orientation is provided in Fig. 3H-J. In particular, on lowering the focal plane it appears that the bead is interacting strongly with the GUV (Fig. 3J) and may indeed be buried in the GUV membrane.
In summary, we show that the outer membrane of giant polymersome protocells formed from glucose-bearing amphiphilic block copolymers are able to bind to microparticles that are decorated with the glucose-specific lectin Concanavalin A. Binding only occurs when both glucose and Con A are present on the surface of the polymersomes and microparticles, respectively. This behaviour mimics the binding of virus particles (e.g. influenza) to the surface of mammalian cells, which leads to viral particle entry and infection. This study, which we believe is the first to demonstrate receptor-mediated particle binding to giant polymersome protocells, may provide important insights for future research on protocells and minimal cell systems.