Embedding a membrane protein into an enveloped artificial viral replica

Natural enveloped viruses, in which nucleocapsids are covered with lipid bilayers, contain membrane proteins on the outer surface that are involved in diverse functions, such as adhesion and infection of host cells. Previously, we constructed an enveloped artificial viral capsid through the complexation of cationic lipid bilayers onto an anionic artificial viral capsid self-assembled from β-annulus peptides. Here we demonstrate the embedding of the membrane protein Connexin-43 (Cx43), on the enveloped artificial viral capsid using a cell-free expression system. The expression of Cx43 in the presence of the enveloped artificial viral capsid was confirmed by western blot analysis. The embedding of Cx43 on the envelope was evaluated by detection via the anti-Cx43 antibody, using fluorescence correlation spectroscopy (FCS) and transmission electron microscopy (TEM). Interestingly, many spherical structures connected to each other were observed in TEM images of the Cx43-embedded enveloped viral replica. In addition, it was shown that fluorescent dyes could be selectively transported from Cx43-embedded enveloped viral replicas into Cx43-expressing HepG2 cells. This study provides a proof of concept for the creation of multimolecular crowding complexes, that is, an enveloped artificial viral replica embedded with membrane proteins.


Introduction
Enveloped viruses such as the influenza virus, human immunodeficiency virus, and coronavirus are nano-sized multimolecular crowding complexes consisting of nucleocapsids covered by a lipid bilayer. [1][2][3] Membrane proteins embedded in the outer surface of enveloped viruses are involved in diverse functions, including adhesion and infection of host cells. For example, spike proteins embedded in the envelope of the coronavirus play an important role during host cell infection. 4,5 The influenza virus has two different membrane proteins, haemagglutinin and neuraminidase, on its envelope, which are involved in infection and budding, respectively. 6,7 Over the past two decades, natural self-assembling protein nanocapsules, such as viral capsids, 8-16 lumazine synthase, 17 ferritin, 18 carboxysome, 19 clathrin, 20 encapsulin, 21,22 and their variants, have emerged as attractive organic materials of discrete size, unique morphology, and constant assembly number.
Protein nanocapsules have been exploited as nanocarriers for drug delivery systems (DDSs), nanotemplates, and nanoreactors. Recently, artificial protein nanocapsules mimicking natural viral capsids have been actively developed by self-assembly of rationally designed proteins. [23][24][25][26][27][28] For example, Baker et al. designed de novo artificial protein subunits with various symmetries and succeeded in constructing protein nanocapsules that are stable against denaturation reagents and heat. 27,28 The de novo-designed protein nanocapsules were utilised as a scaffold for vaccine candidates. The display of the receptor binding domain of the spike protein from SARS-CoV-2 and haemagglutinin from the influenza virus on the surface of artificial protein nanocapsules activated antibody production by the immune system, and induced a strong neutralising antibody response. [29][30][31][32] However, direct expression of membrane protein motifs on protein scaffolds may significantly affect the structure and properties of the nanocapsules and membrane proteins. If nano-architecture similar to natural enveloped viruses embedded with various functional membrane proteins can be readily constructed, it may be possible to overcome the previous challenges. To date, there have been no examples of construction of artificial enveloped viral capsids embedded with membrane proteins, which remains a challenging task.
On the other hand, the progressive development of nanoarchitectures, self-assembled from rationally designed peptides, have enabled the construction of viral capsid-like nanocapsules consisting of peptides. [33][34][35][36][37][38][39][40] We previously found that a 24-residue b-annulus peptide (INHVGGTGGAIMAP-VAVTRQLVGS), which is involved in the formation of the dodecahedral inner skeleton of the tomato bushy stunt virus, self-assembles spontaneously into a hollow artificial viral capsid with a size range of 30-50 nm in water. 36,37,41 The artificial viral capsid can encapsulate various guest molecules inside and can be modified with functional molecules on the external surface. 36,[41][42][43][44][45] We recently succeeded in constructing an enveloped artificial viral capsid by complexing an anionic artificial viral capsid self-assembled from the b-annulus-EE peptide (INHVGGTGGAIMAPVAVTRQLVGSEE) with 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP)/1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) mixed lipids through electrostatic interaction. 46 This enveloped virus-mimicking complex has a relatively uniform particle size and forms a stable structure, as the critical aggregation concentration is 100 times lower than that of the capsid alone, suggesting that it is a candidate for DDS carriers and artificial models of enveloped viruses.
Membrane protein-embedded liposomes (proteoliposomes) can be constructed by expressing membrane proteins using a cell-free protein expression system (PURE system) and simultaneously embedding them into liposomes. [47][48][49][50][51] The PURE system synthesises the target proteins by mixing purified initiation factors, elongation factors, ribosome regeneration factors, necessary enzyme proteins, amino acids, NTPs, and tRNAs involved in the translation reaction in E. coli. The advantages of this method are that its composition can be freely modified and there are almost no proteins irrelevant to protein synthesis because it does not contain cell extracts, but only purified factors. Akiyoshi et al. demonstrated that Connexin-43 (Cx43), a membrane protein involved in the transport of substances between cells, can be embedded in liposomes using the PURE system. 50 The fluorescent dye, calcein, encapsulated in the Cx43embedded proteoliposomes may be transported into the cells through the gap junctions between Cx43 molecules. 50 The question of whether functional membrane proteins can be embedded on the enveloped artificial capsid using the PURE system is unexplored. In this study, we constructed an enveloped artificial viral capsid embedded with the functional membrane protein, Cx43, using the PURE system (referred to as an enveloped artificial viral replica). The function of the embedded Cx43 on the enveloped capsid was evaluated by western blot analysis, fluorescence correlation spectroscopy (FCS), and transmission electron microscopy (TEM). In addition, the transport of fluorescent small dyes from the Cx43-embedded enveloped viral replica into Cx43expressing HepG2 cells was evaluated by confocal laser scanning microscopy (CLSM). The results provide a new proof of concept for creating multimolecular crowding complexes, which are enveloped artificial viral replica embedded with membrane proteins.

Results and discussion
Expression of Cx43 in the presence of an enveloped artificial viral capsid To construct the enveloped artificial viral capsid containing Cx43, we employed a 26-residue b-annulus-EE peptide Fig. S1, ESI †) as the scaffold, which possesses two anionic Glu residues in the C-terminal region directed towards the outer surface of the capsid. 46 The artificial viral capsid possessing an anionic surface was complexed with a mixture of a cationic lipid, DOTAP, and a zwitterionic lipid, DOPC, using a hydration method to construct an enveloped artificial viral capsid with a size of 84 AE 23 nm. 46 Then, cell-free protein expression of the plasmid encoding Cx43 (pURE-Cx43) was conducted using the PURE system in the presence of the enveloped artificial viral capsid (Fig. 1A). The TEM image of the post-expression solution ([pURE-Cx43] = 17.3 nM), showed spherical structures of 50-100 nm (Fig. 1B), and the size distribution obtained from TEM images and dynamic light scattering (DLS) were 103 AE 46 nm and 58 AE 11 nm, respectively ( Fig. S2 and S3A, ESI †). The difference in size distribution obtained from TEM and DLS might be caused by the difference in the observation conditions, that is, the particle size observed by TEM in the dry condition is likely to be large. On the other hand, at higher plasmid concentrations (34.6, 69.1 nM), aggregates of spherical assemblies of approximately 100 nm were primarily observed ( Fig. 1B and Fig. S4, ESI †). The size distribution of the Cx43-expressed solution ([pURE-Cx43] = 34.6 nM) obtained from DLS was 26 AE 4 nm and 571 AE 122 nm (Fig. S3B, ESI †). The small particle sizes might be derived from the PURE system reagents, and the large ones from the aggregates of the spherical assemblies. Cx43 is a four-fold transmembrane protein that is involved in the transport of substances between cells. It forms a hexamer (connexon) on the cell membrane, which in turn forms a gap junction with the connexon of other cells. 52,53 The aggregation reaction may be caused by the accumulation of increased amounts of Cx43 on the envelope and by the formation of excessive gap junction structures among the spherical structures.
Western blot analysis using the anti-Cx43 primary antibody and horseradish peroxidase (HRP)-conjugated secondary antibody showed expression of Cx43 (43 kDa) in the presence of the enveloped artificial viral capsid ( Fig. 2A). The expression of Cx43 in the presence or absence of enveloped capsids was quantified by band density. In the presence of the enveloped capsid consisting of DOTAP/DOPC = 0.2/10, the expression was similar to that of the control (i.e. in the absence of the enveloped capsid). However, as the ratio of DOTAP increased, the expression level decreased. These results indicate that the optimal ratio of DOTAP/DOPC for the expression of Cx43 is 0.2/10, and that higher ratios of cationic lipids inhibit expression. It is likely that translation was inhibited by the adsorption of ribosomes or pURE-Cx43 on the excess cationic surface of the enveloped capsid. In addition, the expression level of Cx43 on the enveloped capsid (DOTAP/DOPC = 1/10) increased with increasing plasmid concentration (Fig. 2B). The expression of Cx43 in the presence of b-annulus-EE peptide or DOTAP/DOPC liposome was also observed using western blot analysis; however, no decrease in the expression level of Cx43, caused by the increase of DOTAP, was observed (Fig. S5, ESI †). Probably, the difference in the expression of Cx43 on the enveloped capsid and liposome might be caused by differences in membrane curvature and fluidity.

FCS analyses of enveloped artificial viral replica embedded with Cx43
To confirm whether the expressed Cx43 is embedded on the surface of the enveloped artificial viral capsid,   and Table S1, ESI †). In addition, FCS analyses of the mixture of antibody and Cx43 expressed in the presence of the b-annulus-EE peptide alone or DOTAP/DOPC liposome did not show any components with an apparent hydrodynamic diameter of 100 nm ( Fig. S6 and Table S1, ESI †). Therefore, the enveloped artificial viral capsid may be a suitable platform for embedding membrane proteins.
As the concentration of the Alexa Fluor 488-labelled anti-Cx43 antibody increased, the ratio of the antibody component binding to free Cx43 ( Fig. 4A and Fig. S7, Table S2, ESI †) or Cx43embedded viral replica ( Fig. 4B and Fig. S8, Table S3, ESI †) increased. The binding curves were analysed by the least-squares   That is, the antibody can bind to the C-terminal of Cx43 exposed outside of the envelope, but not to the C-terminal shielded inside.
TEM observation of the binding of the antibody to the enveloped artificial viral replica embedded with a membrane protein To further confirm the presence of Cx43 on the enveloped capsid, we directly observed TEM images of the Cx43embedded viral replica with the anti-Cx43 primary antibody and gold nanoparticle-labelled secondary antibody (Fig. 5). TEM images of the complex of the anti-Cx43 primary antibody and gold nanoparticle-labelled secondary antibody as a control showed dot-like structures of approximately 5 nm (Fig. 5A). The enveloped artificial viral capsid without Cx43 did not show the presence of gold nanoparticles on the surface of the spherical structures, even with the mixing of the gold nanoparticlelabelled secondary antibody (Fig. 5B). In contrast, TEM images of the Cx43-embedded viral replica (expressed from 17.3 nM pURE-Cx43) showed many dot-like structures on the surface of the spherical structures (Fig. 5C). This indicates that Cx43 is embedded on the enveloped artificial viral replica. TEM images of Cx43-expressing liposomes consisting of DOTAP/DOPC at a 1/10 ratio showed the presence of a few dot-like structures per liposome (Fig. 5D). Next, we increased the concentration of the plasmid used for expressing Cx43 ([pURE-Cx43] = 34.6, 69.1 nM) and observed TEM images using the same method. Interestingly, spherical structures connected to each other were abundant (Fig. 6). These images strongly support the formation of gap junction structures among Cx43-embedded viral replicas. In addition, as the plasmid concentration increased, the density of the gold nanoparticle-labelled secondary antibody bound to the Cx43-embedded viral replica surface appeared to increase, suggesting the increase in the amount of Cx43 embedded on the envelope (Fig. S9, ESI †). Compared to  the TEM images without the antibody (Fig. 1B and Fig. S4, ESI †), the aggregation of spherical structures was minimally observed. These results suggest that the antibodies bound to Cx43 on the envelope inhibit the formation of excessive gap junctions between Cx43-embedded viral replica. The average distance between the gap junctions of the spherical structures was estimated to be approximately 20 nm, based on the TEM images. This suggests that the enveloped artificial viral replicas are tightly adherent to one another (Fig. S10, ESI †). The particle size distribution from the TEM images of the Cx43embedded viral replica at each plasmid concentration was 140 AE 76 nm at 17.3 nM, 143 AE 38 nm at 34.6 nM, and 165 AE 25 nm at 69.1 nM (Fig. S11-13, ESI †). These particle sizes were larger than those estimated from the FCS (Table 1), and those observed in the TEM image without antibody (Fig. 1B). The increase in particle size may be caused by the binding of primary and secondary antibodies to the Cx43-embedded viral replica.

Transport of 5-TMR from the Cx43-embedded viral replica to HepG2 cells
Evaluation of the functionality of the Cx43-embedded viral replica is important to show that a functional Cx43 is properly embedded on the envelope. Therefore, we performed the transport experiments of fluorescent small dyes into Cx43expressing cells through the formation of gap junctions between Cx43-embedded viral replicas and Cx43-expressing cells. A 5-carboxytetramethylrhodamine (5-TMR)-encapsulated artificial viral capsid was prepared by adding a solution of 5-TMR in DMSO to b-annulus-EE powder, then dissolving it in 10 mM Tris-HCl buffer (pH 7.0). The capsid was then complexed with DOTAP/DOPC to construct a 5-TMR-encapsulated enveloped capsid. The particle size of the 5-TMR-encapsulated capsid obtained from DLS was 30 AE 12 nm (Fig. S14A, ESI †), whereas that of the 5-TMR-encapsulated enveloped capsid was 47 AE 15 nm (Fig. S14B, ESI †), indicating that particle size was increased by complexation. After removing free 5-TMR by dialysis, the concentration of 5-TMR encapsulated in enveloped capsid was determined by UV/vis spectroscopy. Cx43 was embedded into the 5-TMR-encapsulated enveloped viral replica by PURE system using 17.3 nM pURE-Cx43. We used HepG2 cells (human hepatoma cell line) that are known to express Cx43. 57 The expression of Cx43 was confirmed by binding of the Alexa Fluor 488-labelled anti-Cx43 antibody to the cell surface in the CLSM image (Fig. 7A). After addition of the 5-TMRencapsulated Cx43-embedded viral replica or non-embedded viral replica to HepG2 cells and further incubation for 1 h, the transport activity of 5-TMR was evaluated by CLSM. As a result, 5-TMR-derived fluorescence was observed inside the cells by adding the Cx43-embedded viral replica, while the fluorescence was minimally observed by adding the enveloped capsid without Cx43 (Fig. 7B). The fluorescence intensity distribution of the Cx43-embedded viral replica showed a significant difference compared with the enveloped capsid without Cx43 (Fig. 7C, S15, and S16, P o 0.001). These results indicate that 5-TMR could be transported from the Cx43-embedded viral replica into Cx43-expressing cells through the gap junctions.

Conclusions
We successfully constructed a membrane protein Cx43embedded viral replica using a cell-free protein expression system (PURE system). Western blot analysis revealed that Cx43 was well expressed in the presence of the enveloped artificial viral capsid. FCS analyses confirmed the presence of a component of the anti-Cx43 antibody bound to the Cx43 embedded on the envelope surface. TEM images obtained using the anti-Cx43 antibody with a gold nanoparticlelabelled secondary antibody strongly support the existence of many Cx43 molecules embedded on the envelope. Interestingly, as the plasmid concentration increased, gap junction structures between Cx43 were observed by TEM. In addition, we succeeded in selectively transporting fluorescent dyes from the Cx43embedded viral replica into Cx43-expressing cells, probably through gap junctions. These results demonstrated that functional membrane proteins were embedded on the lipid bilayer surface of artificial enveloped capsids. In the future, drug transport using Cx43-embedded viral replicas into Cx43-expressing cells will be the next challenge. Our strategy provides a new concept for constructing multimolecular crowding complexes equipped with cellular proteins.

General
Reversed-phase HPLC was performed at ambient temperature using a Shimadzu LC-6AD liquid chromatography system equipped with a UV/vis detector (220 nm, Shimadzu SPD-10AVvp) and an Inertsil WP300 C18 column (250 Â 4.6 mm and 250 Â 20 mm, GL Science). MALDI-TOF mass spectra were obtained using an Autoflex-T2 instrument (Bruker Daltonics) in linear/positive mode with a-cyano-4-hydroxy cinnamic acid (a-CHCA) as a matrix. Ultrapure water with high resistivity (418 MO cm) was purified using a Millipore Purification System (Milli-Q water), and was used as a solvent for the peptides. Reagents were obtained from a commercial source and used without further purification.

DLS
The DLS of the Cx43-embedded viral replica (expressed from 17.3 nM pURE-Cx43) in 10 mM Tris-HCl buffer (pH 7.0) was measured at 25 1C using Zetasizer Nano ZS (MALVERN) with an incident He-Ne laser (633 nm) and ZEN2112-Low volume glass cuvette cell. During measurements, count rates (sample scattering intensities) were also recorded. Correlation times of the scattered light intensities G(t) were measured several times and their means were calculated for the diffusion coefficient. Hydrodynamic diameters of scattering particles were calculated using the Stokes-Einstein equation.

TEM
A solution (42 nM, 5 mL) of the enveloped viral replica embedded with Cx43 in 10 mM Tris-HCl buffer (pH 7.0) and the mouse IgG anti-Cx43 monoclonal antibody (BD Transduction) in the same buffer were mixed and incubated overnight at 25 1C.
An aqueous glycerol solution of gold nanoparticle-labelled goat anti-mouse IgG secondary antibody (Sigma-Aldrich, 14.8 nM, 10 mL) was then added to the mixture and was incubated for 3 h at 25 1C. Aliquots (5 mL) of the sample aqueous solutions were applied to the hydrophilised carbon-coated Cu-grids (C-SMART Hydrophilic TEM grids, ALLANCE Biosystems) for 1 min, and were then removed. Subsequently, the TEM grids were immersed in the staining solution (5 mL): 25% EM stainer (Nisshin EM Co., Ltd) for 10 min, and were then removed. After the sample-loaded carbon-coated grids were dried in vacuo, they were observed using TEM (JEOL JEM 1400 Plus) at an accelerating voltage of 80 kV.

Paper
RSC Chemical Biology 0.1% dilution (v/v) with 10% Blocking One in TBS-T] for the anti-Cx43 monoclonal antibody. After being washed with TBS-T four times, blots were developed using a chemiluminescence method (ECL; GE Healthcare). The band images were obtained with a LAS-4000EPUVmini (Fujifilm Co., Ltd) with an exposure time of 30 min.

FCS
The Alexa Fluor 488-labelled anti-Cx43 antibody (Bioss Inc.) was added to the aqueous solution of the enveloped viral replica embedded with Cx43 (the final concentration of the antibody was 10 nM), and was then incubated at 25 1C for 1 h. FCS analyses of the solutions were conducted on an FCS compact BL/GR (Hamamatsu Photonics Co., Ltd) in a microwell slide (25 mL) using a 473-nm laser (75 mW) at 25 1C. Each measurement was recorded for 150 s (period of a single measurement: 5 s; looptime: 30 times). The diffusion time (t) and ratio (R) of the fast (t 1 , R 1 ) and slow (t 2 , R 2 ) components were obtained by curvefitting the auto-correlation function G(t) obtained from FCS measurements based on (eqn (1)) 60 where t is the diffusion time of the Alexa Fluor 488-labelled anti-Cx43 antibody in the detection area, N is the average number of Alexa Fluor 488-labelled anti-Cx43 antibodies in the detection area, k is the structural parameter, and F is triplet component ratio. The hydrodynamic radiuses (r 1 , r 2 ) of the Alexa Fluor 488labelled anti-Cx43 antibody were calculated using eqn (2) and (3). The hydrodynamic diameters (d 1 , d 2 ) were calculated by doubling r 1 and r 2 : where o is the radius of the detection area, T is the absolute temperature, k B is the Boltzmann constant, Z is the viscosity of the solvent, and D is the diffusion coefficient of the Alexa Fluor 488labelled antibody. The value of o was evaluated through reference measurement using Alexa 488 (D = 414 mm 2 s À1 ). 61 The diffusion time of Alexa 488 was 0.0356 ms.
To calculate the dissociation constant K d for the complex of Cx43 and Alexa Fluor 488-labelled anti-Cx43 antibody, the following equation (Langmuir's equation) was used.
where [A] is the antibody concentration, R is the ratio of antibody bound to Cx43 at the antibody concentration, and R max is the maximum ratio of antibody bound to Cx43. Transport of 5-TMR from the Cx43-embedded viral replica to HepG2 cells

Encapsulation of 5-TMR into the
HepG2 cells (RIKEN BioResource Research Center, Japan) were cultured in Dulbecco's modified Eagle's medium (DMEM). The medium contained 10% fetal bovine serum (FBS, v/v), 100 mg mL À1 streptomycin, 100 units per mL penicillin, 1 mM sodium pyruvate, and 1% MEM nonessential amino acids (v/v, Sigma M7145). Cells were maintained at 37 1C in a 5% CO 2 incubator, and a subculture was performed every 3-4 days. HepG2 cells were seeded onto a single-well glass bottom dish at 2.0 Â 10 4 cells per well in a final volume of 100 mL, and were incubated for 24 h at 37 1C, 5% CO 2 . Solutions of the Alexa Fluor 488-labelled anti-Cx43 antibody (5 mM, 60 mL) in DMEM (+10% FBS), the Cx43-embedded viral replica, or enveloped viral capsid without Cx43 (50 mL) were added to the cells and these were incubated for 1 h, at 37 1C, 5% CO 2 . After removal of the solution, 10 mg mL À1 Hoechst 33342 (80 mL) was added to the cells and another incubation phase was carried out for 10 min at 37 1C, 5% CO 2 . After being washed with PBS, the medium was added to the cells and CLSM was performed. Alexa Fluor 488 was excited at 499 nm and was observed through a 520 nm emission band-pass filter (green). 5-TMR was excited with 553 nm and was observed through a 577 nm emission band-pass filter (magenta). Hoechst 33342 was excited with 352 nm and was observed through a 455 nm emission band-pass filter (cyan). The fluorescence intensity of 5-TMR transported into the HepG2 cells was measured from the fluorescence images by subtracting the background intensity using Image J software (N = 60).