Article pubs.acs.org/Biomac
Predicting the Loading of Virus-Like Particles with Fluorescent Proteins W. Frederik Rurup,† Fabian Verbij,† Melissa S. T. Koay,† Christian Blum,‡ Vinod Subramaniam,‡ and Jeroen J. L. M. Cornelissen*,† †
Laboratory for Biomolecular Nanotechnology and ‡Nanobiophysics (NBP), MESA+Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands S Supporting Information *
ABSTRACT: The virus-like particle (VLP) of the Cowpea Chlorotic Mottle Virus (CCMV) has often been used to encapsulate foreign cargo. Here we show two different rational design approaches, covalent and noncovalent, for loading teal fluorescent proteins (TFP) into the VLP. The covalent loading approach allows us to gain control over capsid loading on a molecular level. The achieved loading control is used to accurately predict the loading of cargo into CCMV VLP. The effects of molecular confinement were compared for the differently loaded VLPs created with the covalent method. We see that the loading of more than 10 fluorescent proteins in the 18 nm internal cavity of the CCMV capsid gives rise to a maximum efficiency of homo-FRET between the loaded proteins, as measured by fluorescence anisotropy. This shows that already at low levels of VLP loading molecular crowding starts to play a role.
■
INTRODUCTION
ability to gain control over capsid loading on a molecular level, which is essential for all applications in nanotechnology. In the past, we have shown that the heterodimerization of leucine zipper-like “E-coil” and “K-coil” coiled-coil peptides effectively guide fluorescent proteins inside the CCMV capsid.6 Despite the benefits of the E−K coiled-coil design, it is also known that the coiled coils disassemble and form E-coil homotrimers during capsid assembly (at pH 5).16 This formation of homotrimers significantly reduces the efficiency of loading at higher protein concentrations. To overcome this problem, we sought to optimize the design for the rational loading of CCMV by creating more stable “cargo−capsid” complex variants. We investigated two different approaches: (1) increasing the intrinsic ratio between the cargo and CCMV capsid protein (i.e., its valency) and (2) designing a flexible linker whereby the cargo is fused to the CCMV capsid protein. In our model systems, we use Teal Fluorescent Protein (mTFP)17 as our cargo of choice. We restored the natural dimerization of the monomeric TFP to form dTFP with E-coil (dTE). In the first approach, dTE should form a more stable complex with the capsid protein dimer with K-coil (CK) (dTECK) compared to the monomeric form, therefore, directing a larger number of fluorescent proteins into the CCMV capsids. Alternatively, genetically engineering a TFPlinker-CCMV fusion protein (hereby referred to as HTC) abolishes the influence of dissociating E−K coiled-coils and should therefore further increase the loading efficiency in
In nature, the packaging of molecules into discrete compartments is an essential feature for maintaining cellular efficiency. Organelles are a classical example, where simple building blocks are used to assemble complex, well-defined nanostructures. The structure, function, and efficiency of organelles in compartmentalization and catalysis serve as a constant source of inspiration. However, there is still a limited understanding of how molecules behave in such molecularly confined spaces. Although various approaches to mimic nanoconfinement have been demonstrated,1−7 there are few examples of nanocontainers that are as well-defined and monodisperse as virus-like particles (VLPs). Here, we use the virus capsid of the Cowpea Chlorotic Mottle Virus (CCMV) to study the effects of molecular confinement. CCMV has a well-known assembly/ disassembly pathway that can be controlled by pH and ionic strength. This allows us to gain molecular control over loading of cargo into the CCMV.8 Although the assembly of native CCMV relies on electrostatic interactions between the positively charged N-terminus of CCMV and the negatively charged cargo (e.g., viral RNA), CCMV has been used to encapsulate various non-native materials in vitro, such as anionic polymers and nanoparticles.9−11 For the encapsulation of non-negative materials, such as neutral or positively charged proteins, noncovalent affinity tags have been previously reported and used to direct loading inside CCMV.6 Other VLPs, such as Lumazine synthase from Aquifex aeolicus, Qβ, and the capsid of bacteriophage P22, have also been successfully used as nanocontainers for the encapsulation of proteins.12−15 However, one of the major challenges is the © 2013 American Chemical Society
Received: October 25, 2013 Revised: December 20, 2013 Published: December 23, 2013 558
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
Design and Molecular Cloning of mTFP−CCMV Capsid Fusion Protein. The XhoI site (underlined) was introduced into the pET15b CCMV K-coil2 (CK) vector, between the K-coil and the capsid protein, by site-directed mutagenesis using the following primer: 5′-CGC CGC CCT GAA GGA GCT CGA GAT GTC TAC AGT CGG AAC AGG G-3′. The sequence for His6-tagged TFP was amplified by PCR from the pET15b TFP-E-coil vector using the following forward primer: 5′-GGA ATT GTG AGC GGA TAA-3′ and reverse primer: 5′-GCG GCT CGA GGC CAC CAG AGC CGC TCC CGC GCG GCA CCA GAC CCT GCC ACC CGT GTA CAG CTC GTC CAT GCC GTC-3′ (XhoI is underlined). These primers also introduce the sequence for flexible linker between TFP and the capsid protein and a unique XhoI site in the PCR product. Double digestion by XbaI and XhoI of both the PCR product and the mutated pET15b CCMV K-coil vector and subsequent ligation resulted in the pET15b His-TFP-Capsid (HTC) vector. The amino acids sequence of the product is shown in the Supporting Information, Figure S1. Recombinant Protein Expression and Purification. The pET15b plasmids containing CK, TE, dTE, and HTC were transformed into Escherichia coli BL21(DE3)pLysS (Novagen) for protein expression. Starting cultures were grown overnight at 37 °C from glycerol stock cells in 7 mL of LB medium (Sigma) containing 100 μg/mL ampicillin and 34 μg/mL chloramphenicol (Sigma). The overnight cultures were used to inoculate 0.5 L of LB medium containing ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL) and grown to an optical density of OD600 = 0.6−0.8. Protein expression was induced following addition of isopropyl β-Dthiogalactoside (IPTG) to a final concentration of 0.1 mM at 30 °C for 4−5 h. The cells were harvested by centrifuging (10.000 g, 15 min) and the cells were lysed using BugBuster according to the manufacturer’s protocol (Novagen). The TE, dTE, and HTC proteins (all bearing a His-tag) were purified using nickel-affinity column chromatography with a modified version of the suppliers protocol (Novagen). TE, dTE, and HTC were bound and washed with 0.1 M phosphate buffer, 0.3 M NaCl, 12.5 mM Imidazole, pH 8.0. For the noncovalent variants, the nickel immobilized TE and dTE were then mixed for 1 h with the CK to form the TECK or dTECK complexes, respectively. TECK, dTECK, and HTC were then washed with 10 column volumes of 0.1 M phosphate buffer, 1.5 M NaCl, 12.5 mM imidazole, pH 8.0, and eluted with 0.1 M phosphate buffer, 1.5 M NaCl, 0.25 M imidazole, pH 8.0. TECK, dTECK, and HTC were stored in 50 mM Tris, 0.5 M NaCl, 10 mM MgCl2, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 7.5. All formed complexes, namely, TECK, dTECK, and HTC, were further purified using size exclusion chromatography (SEC) using a fast protein liquid chromatography (FPLC) Ä KTA purifier. The complexes were purified using a Superdex 75 column (GE healthcare), 1 mL injection volume, and the same buffer as they were stored in to remove noncomplexed fluorescent protein, and their purity was confirmed by SDS and native-PAGE analysis (see Supporting Information, Figure S2). Quantification of Encapsulation Efficiency. Quantification of fluorescent proteins per capsid was determined based on absorbance spectroscopy measurements and calculations based on the Beer− Lambert law (see Supporting Information, Figure S3 and equations). Standard calibration curves were performed to establish the extinction coefficient of the fluorescent protein at both pH 7.5 (during the assembly/mixing step) and pH 5.0 (after encapsulation; see Supporting Information, Figure S4). Based on the unique absorbance peak at 462 nm (corresponding to the fluorophore of TFP), the concentration of TFP could be determined. For all monomeric, dimeric, and fusion TFP variants, it was assumed that each TFP was bound to a single coat protein of CCMV. The encapsulation efficiency and quantification of TFP per capsid were determined after further purification by size-exclusion chromatography (at pH 5.0). Capsid Assembly. The growth and purification of the native wildtype CCMV virus (native CCMV coat protein) from Vigna unguiculata leaves was performed as reported previously.11 To reassemble the CCMV capsid, the isolated native CCMV coat proteins were mixed
CCMV (Scheme 1). In this work, we performed systematic studies to compare the loading capacity and efficiency into Scheme 1. Schematic Representation of the Different Design Principles Described in This Work for the Rational Loading of CCMVa
a The first approach utilizes leucine zipper like E-coils tethered to (a) monomeric TFP or (b) dimeric TFP, which form electrostatic interactions at pH 7.5 with the complementary K-coil tethered to CCMV. The second approach utilizes a genetically engineered TFPpeptide linker-CCMV fusion protein. In all cases, lowering the pH to 5.0 promotes the encapsulation of TFP cargo inside CCMV.
CCMV capsids and investigate the effects of molecular confinement. With the new design we believe that it is possible to improve molecular control over quantitative capsid loading and accurately predict the average loading of cargo into CCMV.
■
EXPERIMENTAL SECTION
Site-Directed Mutagenesis and Molecular Cloning of pET15b-mTFP. Using site-directed mutagenesis (QuikChange mutagenesis, Stratagene), a unique restriction site for BamHI was introduced into the pET15b EGFP-E-coil expression vector using the following primers: forward 5′-GGT GCC GCG CGG GAT CCA TAT GCT CGA GAA AAG AG-3′ and its reverse complement (BamHI restriction site underlined). The DNA fragment encoding mTFP was excised from the pNCSmTFP DNA vector (Allele BioTech, U.S.A.) by double digestion with the BamHI and BsrGI restriction enzymes. pET15b EGFP-E-coil was double-digested using the same enzymes, hence, removing EGFP for subsequent replacement by TFP. The frameshift caused by the cloning was corrected by site-directed mutagenesis with primer: 5′-GGT GCC GCG CGG GAG CCA TAT GGT GAG CAA GGG CG-3′ and its reverse complement. This resulted in a pET15b TFP-E-coil (TE) vector. All sequences were confirmed by DNA sequencing (Eurofins MWG Operon, Germany). The native dimerization of TFP was restored by reintroducing the mutations into the A-C dimerization site (D144E, A145P, R149I, K162S, K164S) 11 by site-directed mutagenesis, using the primers: 5′CCA CCG GCT GGG AGC CCT CCA CCG AGA TCA TGT ACG TGC GCG-3′ (D144E, A145P, R149I, mutations are underlined), 5′GCT GAA GGG CGA CGT CAG CCA CAG CCT GCT GCT GGA GGG CG-3′ (K162S, K164S, mutations are underlined), and their reverse complementary strands. The dimeric TFP-E-coil (dTE) variant was confirmed by DNA sequencing (Eurofins MWG Operon, Germany). 559
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
with the recombinant protein complexes (TECK, dTECK, HTC) in various ratios in 50 mM Tris-Cl, 500 mM NaCl, 1 mM EDTA (pH 7.5). For all capsid reassembly mixtures, a final concentration of 75 μM CCMV protein was maintained (combined native CCMV coat protein and recombinant CK proteins). The mixtures were dialyzed overnight to 50 mM sodium acetate, 1 M NaCl, 1 mM NaN3 (pH 5.0) to induce capsid formation. The assembled capsids were purified and analyzed by FPLC using a Superose 6 column (see Supporting Information, Figure S5). PAGE Analysis. Polyacrylamide gel electrophoresis (PAGE) was performed both in the presence of sodium dodecylsulfate (SDS) and under native conditions. For SDS-PAGE, samples were heated to 99 °C for 5 min in the presence of 2-mercaptoethanol and 1% SDS. UV−vis Analysis. All UV−visible measurements were performed on a Perkin-Elmer Lambda 850 Spectrometer. Standard quartz cuvettes were used. Fluorescence Anisotropy. A Varian Cary Eclipse spectrophotometer was used to measure fluorescence anisotropy of all samples. Samples were measured in quartz cuvettes with 450 nm wavelength light excitation. The anisotropy was averaged over the range of 480 to 520 nm. Transmission Electron Microscopy (TEM). TEM micrographs were recorded on an analytical FEG-TEM (Phillips CM 30) operated at 300 kV acceleration voltages. Samples were prepared by placing a 5 μL drop of the samples on Formvar carbon-coated copper grids (Electron Microscopy Sciences). The sample drop was left on the grid for 5 min, after which time the excess buffer was blotted away with filter paper. Samples were negatively stained by applying 5 μL of stain (1% w/v uranyl acetate in Milli-Q water) onto the grid and removing the excess stain away after 1 min with filter paper. The samples were dried overnight before imaging.
maximum upon increasing mixing ratios, which also suggests an optimal loading. When the amount of TFP in the assembly mix is higher than 7.5 μM, the total number of TECK and dTECK that are effectively encapsulated approaches a maximum, suggesting an apparent saturation point exists (Figure 1, red and green curves, respectively). Overall, this trend shows that the loading efficiency of dTECK is significantly improved compared to that of the monomeric TECK construct, but still not exceeds ∼6 guest proteins per capsid. These observations also suggest that the E-coil homotrimerization is more pronounced in the monomeric TECK, which subsequently decreases the overall loading efficiency, in line with previous results with the monomeric system.16 By comparison, the HTC fusion variant shows a systematic increase in loading number and efficiency, with increasing concentration ratio. This further confirms that the saturation points observed for both TECK and dTECK are likely due to the homotrimerization of the coiled-coils at higher concentrations of guest material. However, when the concentration range for HTC loading is extended over 40 mol % HTC and 60 mol % native CCMV coat protein, the loading efficiency decreases dramatically and a large amount of erroneous capsids are formed (see Supporting Information, Figure S5). This is attributed to internal overcrowding, which does not allow more cargo in the VLP interior, and consequently, no additional capsid proteins needed to complete the VLP shell can associate with the assembly. Improving the Directed Loading. Figure 1 shows the linear correlation between the mixing ratios (0−40 mol %) and the number of HTC encapsulated. This calibration curve was subsequently used to precalculate and predict a desired number of TFPs to be encapsulated. Indeed, we consistently obtain excellent agreement between the predicted and the obtained number of TFPs encapsulated, confirming that HTC loading is not only reliable but also highly controllable (Figure 2). Interestingly, upon increasing the loading of TFPs inside CCMV, the yield of CCMV VLPs formed decreases dramatically. Based on the relative concentration ratios between HTC and CCMV, in the presence of 10 mol % HTC, 18 TFP should be loaded (10% of 180 CCMV monomers). However, we observe approximately 25% efficiency in effective loading
■
RESULTS AND DISCUSSION Determining the Loading Efficiency. Figure 1 shows the assembly behavior and efficiency of the different complexes,
Figure 1. Typical capsid TFP loading as a function of percentage of complex (TECK, dTECK, or HTC) in assembly.
whereby the amount of complex was varied from 0 to 40 mol % in combination with 100 to 60 mol % native CCMV coat protein. Larger amounts of complex were attempted, however, a significant decrease in assembly efficiency was observed (see Supporting Information, Figure S6). Since dTECK forms a homodimer between two TFP molecules as well as the heterodimer between each TFP and CCMV, a more stable complex is expected. Indeed, dTECK appears to be more efficient compared to the monomeric TECK, leading to a higher number of fluorescent proteins being encapsulated (Figure 1). Interestingly, the loading ratio approaches a
Figure 2. Plot showing (left axis) the average vs predicted loading of TFPs (HTC) per CCMV capsid (blue curve) and (right axis) the normalized number of capsids formed calculated based on the amount of capsids obtained relative to the total amount of protein in each sample (gray curve). 560
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
(∼5 TFP; Figure 1). Since the 25% effective loading is consistent within the 10−40 mol % range, this may suggest that cargo loading may be a (equilibrium) process in which three additional native CCMV coat proteins (CP) per TFP are needed in order to include the latter. At higher TFP/CP ratios, insufficient CP is available in line with the lower yield and incomplete capsid formation (vide infra). Transmission electron microscopy (TEM) of CCMV loaded with 2 TFPs show highly monodisperse particles with an average diameter of ∼27 nm, which is consistent with native T = 3 particles (Figure 3A). The inclusion of the uranyl acetate
Table 1. Fluorescent Anisotropy Values for the Different CCMV Loading Components sample free FPs
Complexed FPs
encapsulated FPs (1−3 FPs per capsid)
a
TFP TFP−E-coil dTFP−E-coil TECK dTECK HTC TECK encaps. dTECK encaps. HTC encaps.
anisotropya (490−520 nm) 0.31 0.32 0.33 0.33 0.33 0.33 0.36 0.34 0.35
The standard deviation in the data ≤0.01.
did not show significant changes compared to the free TFP and dTFP forms (with E-coil). We suspect that the increase in anisotropy typically attributed to increased restricted rotational freedom is offset in this case by the decrease in anisotropy observed due to homo-FRET. Upon encapsulation of low amounts of fluorescent proteins (1−3 TFPs) into virus capsids, we observe a distinct increase in anisotropy values for all complexes. This increase is attributed to the successful incorporation of FPs into the capsid, which gives rise to a decrease in rotational mobility. Under these conditions, the bulk of the assembled capsid (3.7 MDa, 28 nm) restricts the movement of the fluorescent proteins (26.9 kDa, 3 nm) and increases the anisotropy, suggesting that the fluorescent proteins are still tethered to the capsid wall. Homo-FRET could decrease the anisotropy if multiple fluorescent proteins are in close proximity. However, at these low loading levels, homo-FRET is expected to play a minor role since fluorescent proteins encapsulated within one capsid are likely to be further apart than the effective distance for homoFRET. We only expect a difference from this trend for dTECK, which has the tendency to dimerize. This is also consistent with the anisotropy data, in which dTECK displays the lowest anisotropy of the encapsulated complexes although the measured differences are small. Additional fluorescence anisotropy measurements were performed using the HTC variant, since this showed the most efficient, predictable and controllable loading. Upon increasing the amount of encapsulated HTC, a decrease in the anisotropy value is observed (Figure 4). Increasing the number of encapsulated HTC per capsid results in a decrease of average distance between HTCs (Figure 4). This molecular crowding of the fluorophores gives rise to an increase in homo-FRET between the fluorophores within the capsid interior, as was also seen by Mullaney et al. and O’Neil et al.18,19 The anisotropy approaches a steady-state for samples with an average loading of 10 encapsulated fluorescent proteins. This indicates that for 10 or more encapsulated TFPs, the effect on the anisotropy has already reached its maximum. Nonetheless, the efficiency of homo-FRET is strongly dependent on the interfluorophore distance and the fluorophore specific Förster radius (R0).20 TFP has no reported Förster radius for homoFRET, but based on interactions of TFP with other fluorescent proteins,20 an R0 ∼ 5.5 nm is estimated (references within report R0 = 4.6−6.2 nm). When the interfluorophore distance is less than 1.5 R0, homo-FRET is known to be much more
Figure 3. Transmission electron microscopy (TEM) images of CCMV loaded with (A) 2 TFPs per capsid and (B) 20 TFPs per capsid. The samples were negatively stained with 1% uranyl acetate.
staining suggests a significant void volume inside these VLPs. The particles filled with 20 TFPs also show mostly intact T = 3 assemblies, however, some irregularity in size and shape are also observed (particles of 32−38 nm). This may suggest that some exhibit a partially swollen morphology and that the capsids are filled, which is also supported by the relative intensity of uranyl staining. Studying Molecular Crowding by Fluorescence Anisotropy. To obtain a better understanding of the molecular crowding effects and the rotational freedom of the TFP localized inside the CCMV capsids, we performed steadystate fluorescence anisotropy measurements. Fluorescence anisotropy (r) describes the extent of depolarization of the fluorescence upon excitation with polarized. A fluorophore with rotational correlation time shorter than its fluorescence lifetime undergoes multiple rotations within the lifetime of the molecule, consequently emitting fluorescence with a polarization orientation that is not aligned with that of the excitation light. This emission depolarization translates into a lower steady-state fluorescent anisotropy value (r). Depolarization of emission can also result from energy transfer processes between identical fluorophores in nanometer proximity, a process called homo-Förster resonance energy transfer (homo-FRET). In case that the two fluorophores are not aligned parallel, this energy transfer will result in a depolarization of the emission and thus a decrease of anisotropy when the light is emitted from the fluorophore that the energy was transferred to. Here, we employed fluorescence anisotropy to monitor changes in TFP rotational mobility during encapsulation as well as the effects of molecular crowding on TFP resulting in homo-FRET inside the capsids. Before complexation to native CCMV coat protein, we observe a higher anisotropy for the dimeric TFP than the monomeric TFP, suggesting that the restricted rotational freedom due to dimerization of the TFPs has a more profound effect than the homo-FRET (Table 1). Surprisingly, the fluorescence anisotropy for the TECK and dTECK complexes 561
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
Figure 4. Relationship between the fluorescence anisotropy vs number of fluorescent proteins encapsulated showing the decrease in anisotropy with increasing number of TFP molecules. The solid gray line serves as a guide to the eye.
efficient.20 For TFP, this threshold distance is calculated to be 8.2 nm or less. CCMV has an average internal diameter of 18 nm. Assuming a perfect sphere, the maximum distance in between the fluorophores can be approximated by a ‘Tammes problem’.21 This mathematical model calculates the maximal possible distance between a number of points on a sphere. If we take into account the size of the TFP molecule (3 nm in diameter, based on the 2HQK PDB entry) with the fluorophores located in the middle of the protein molecule, the fluorophores can be distributed over a sphere with a 15 nm diameter. Using the solution from the Tammes problem, we find that a loading of 10 TFPs is sufficient to ensure that the fluorophores are at distances below 8.2 nm, which will allow for efficient homoFRET. It should be noted that this is based on the assumption that all TFPs are equally distributed and oriented against the inner capsid wall. However, the TFPs are more likely to be randomly oriented (rather than being docked along the capsid shell wall) and due to the flexible linker, a distribution diameter of 10−12 nm is more likely to be the case (Figure 5, top). In this case, according to the Tammes curves, 4−6 fluorescent proteins are already sufficient for possible homo-FRET to occur. Finally, the distances derived from solving the Tammes problem are the distances for an equal distribution of the TFPs over the inner surface. A deviation from the equal distribution, which we expect in our case, will bring some of the TFPs closer together and will result in increased homo-FRET efficiency.
Figure 5. (top) Schematic representation of TFP location inside the 18 nm diameter CCMV internal cavity; (bottom) the solutions of the “Tammes” problem for different sphere diameters.
work demonstrated the encapsulation of up to 100 supercharged GFPs or GFP-mCherry fusion pairs within the capsid of lumazine synthase or P22, respectively, our strategy provides a unique means to control the cargo loading number in a highly predictive manner. Our approach offers two distinct advantages: (1) it allows the native cargo to be encapsulated and does not require the introduction of surface charges to promote cargo loading and (2) this strategy can be readily extended to other cargo proteins by simply replacing the TFP cargo for alternative cargo of choice. From our fluorescence anisotropy studies, it is apparent that loading of 10 or more TFPs per capsid induces molecular crowding effects and that the packing is sufficiently close that maximal homo-FRET occurs. Figure 4 shows that the fluorescence anisotropy approaches a minimum, which is attributed to homo-FRET effects and provides an effective means to measure the loading capacity of guest materials in virus-like assemblies. Similar molecular crowding effects have been reported by O’Neil et al., whereby covalently tethered FRET pairs (GFP and mCherry) were encapsulated in the bacteriophage procapsid P22; however, in this example, the studies were not quantitative since contributions from both intra- and intermolecular FRET were observed. Interestingly, the values obtained in this work correlate to 19% volume occupancy, which is close to the 20−30% volume occupancy predicted for macromolecules in cells and organelles22 and could suggest that CCMV serves as an excellent model system for understanding molecular crowding in cells and organelles.
■
CONCLUSIONS Based on our present understanding of the CCMV VLP, we evaluated in detail the loading of this VLP with FPs following two strategies. Using one approach based on the leucine zipperlike “E−K” coiled-coil peptides, we observe a distinct nonlinear behavior for the encapsulation of monomeric and dimeric TFP variants. While we observe a 3-fold improvement in the loading efficiency for the dimeric TFP leading to 6 TFPs being encapsulated, a clear saturation point is observed. Based on the theoretical volume of CCMV (18 nm diameter sphere), we calculate that 6 TFPs only occupy 5% of the interior volume (assuming the volume of a cylinder). The second strategy, in which TFP was covalently bound to the monomeric capsid protein (HTC), allows for the predictive and more controlled loading of fluorescent proteins inside CCMV. The loading efficiency of TFP was significantly improved through this method, resulting in up to 20 TFPs loaded inside CCMV while maintaining a reasonable yield of assembly. While previous 562
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
■
Article
(17) Ai, H. W.; Henderson, J. N.; Remington, S. J.; Campbell, R. E. Biochem. J. 2006, 400, 531−540. (18) Mullaney, J. M.; Thompson, R. B.; Gryczynski, Z.; Black, L. W. J. Virol. Methods 2000, 88, 35−40. (19) O’Neil, A.; Prevelige, P. E.; Basu, G.; Douglas, T. Biomacromolecules 2012, 13, 3902−3907. (20) Sun, Y.; Wallrabe, H.; Seo, S. A.; Periasamy, A. ChemPhysChem 2011, 12, 462−474. (21) Erber, T.; Hockney, G. M. J. Phys. A: Math. Gen. 1991, 24, L1369−L1377. (22) Ellis, R. J. Curr. Opin. Struct. Biol. 2001, 11, 114−119.
ASSOCIATED CONTENT
S Supporting Information *
The complete amino acid sequence of the fusion construct HTC is provided in Figure S1. Additional native gel electrophoresis and SDS-PAGE analyses are presented in Figure S2. UV−visible data and calibration curves are shown in Figures S3 and S4, respectively. Size-exclusion chromatograms for the different assembly mixing ratios for each construct are shown in Figure S5. An alternative representation of Figure 2 showing the full range of assembly mixing ratios from 0−100 mol % HTC is provided in Figure S6. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Fax: 0031 53489 4645. Tel.: 0031 53489 4380. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge financial support by the Chemical Council of The Netherlands Organization for Scientific Research (NWO−CW), the European Science Foundation (ESF), and The Netherlands Royal Academy for Arts and Sciences (KNAW).
■
REFERENCES
(1) Vriezema, D. M.; Comellas Aragonès, M.; Elemans, J. A. A. W.; Cornelissen, J. J. L. M.; Rowan, A. E.; Nolte, R. J. M. Chem. Rev. 2005, 105, 1445−1490. (2) Kang, S.; Uchida, M.; O’Neil, A.; Li, R.; Prevelige, P. E.; Douglas, T. Biomacromolecules 2010, 11, 2804−2809. (3) Baumler, H.; Georgieva, R. Biomacromolecules 2010, 11, 1480− 1487. (4) Rhee, J.-K.; Hovlid, M.; Fiedler, J. D.; Brown, S. D.; Manzenrieder, F.; Kitagishi, H.; Nycholat, C.; Paulson, J. C.; Finn, M. G. Biomacromolecules 2011, 12, 3977−3981. (5) van Dongen, S. F.; Nallani, M.; Cornelissen, J. J. L. M.; Nolte, R. J. M.; van Hest, J. C. Chemistry 2009, 15, 1107. (6) Minten, I. J.; Hendriks, L. J. A.; Nolte, R. J. M.; Cornelissen, J. J. L. M. J. Am. Chem. Soc. 2009, 131, 17771−17773. (7) Bode, S. A.; Minten, I. J.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Nanoscale 2011, 3, 2376−2389. (8) Lavelle, L.; Gingery, M.; Phillips, M.; Gelbart, W. M.; Knobler, C. M.; Cadena-Nava, R. D.; Vega-Acosta, J. R.; Pinedo-Torres, L. A.; Ruiz-Garcia, J. J. Phys. Chem. B 2009, 112, 3813−3819. (9) Douglas, T.; Young, M. Nature 1998, 393, 152−155. (10) de la Escosura, A.; Verwegen, M.; Sikkema, F. D.; ComellasAragones, M.; Kirilyuk, A.; Rasing, T.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Chem. Commun. (Cambridge, U. K.) 2008, 1542. (11) Sikkema, F. D.; Comellas-Aragones, M.; Fokkink, R. G.; Verduin, B. J.; Cornelissen, J. J. L. M.; Nolte, R. J. M. Org. Biomol. Chem. 2007, 5, 54. (12) Seebeck, F. P.; Woycechowsky, K. J.; Zhuang, W.; Rabe, J. P.; Hilvert, D. J. Am. Chem. Soc. 2006, 128, 4516−4517. (13) Worsdorfer, B.; Woycechowsky, K. J.; Hilvert, D. Science 2011, 331, 589−592. (14) O’Neil, A.; Reichhardt, C.; Johnson, B.; Prevelige, P. E.; Douglas, T. Angew. Chem., Int. Ed. 2011, 50, 7425−7428. (15) Fiedler, J. D.; Brown, S. D.; Lau, J. L.; Finn, M. G. Angew. Chem., Int. Ed. 2010, 49, 9648−9651. (16) Minten, I. J.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Macromol. Biosci. 2010, 10, 539. 563
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Predicting the Loading of Virus-Like Particles with Fluorescent Proteins W. Frederik Rurup,† Fabian Verbij,† Melissa S. T. Koay,† Christian Blum,‡ Vinod Subramaniam,‡ and Jeroen J. L. M. Cornelissen*,† †
Laboratory for Biomolecular Nanotechnology and ‡Nanobiophysics (NBP), MESA+Institute for Nanotechnology, University of Twente, P.O. Box 217, 7500 AE Enschede, The Netherlands S Supporting Information *
ABSTRACT: The virus-like particle (VLP) of the Cowpea Chlorotic Mottle Virus (CCMV) has often been used to encapsulate foreign cargo. Here we show two different rational design approaches, covalent and noncovalent, for loading teal fluorescent proteins (TFP) into the VLP. The covalent loading approach allows us to gain control over capsid loading on a molecular level. The achieved loading control is used to accurately predict the loading of cargo into CCMV VLP. The effects of molecular confinement were compared for the differently loaded VLPs created with the covalent method. We see that the loading of more than 10 fluorescent proteins in the 18 nm internal cavity of the CCMV capsid gives rise to a maximum efficiency of homo-FRET between the loaded proteins, as measured by fluorescence anisotropy. This shows that already at low levels of VLP loading molecular crowding starts to play a role.
■
INTRODUCTION
ability to gain control over capsid loading on a molecular level, which is essential for all applications in nanotechnology. In the past, we have shown that the heterodimerization of leucine zipper-like “E-coil” and “K-coil” coiled-coil peptides effectively guide fluorescent proteins inside the CCMV capsid.6 Despite the benefits of the E−K coiled-coil design, it is also known that the coiled coils disassemble and form E-coil homotrimers during capsid assembly (at pH 5).16 This formation of homotrimers significantly reduces the efficiency of loading at higher protein concentrations. To overcome this problem, we sought to optimize the design for the rational loading of CCMV by creating more stable “cargo−capsid” complex variants. We investigated two different approaches: (1) increasing the intrinsic ratio between the cargo and CCMV capsid protein (i.e., its valency) and (2) designing a flexible linker whereby the cargo is fused to the CCMV capsid protein. In our model systems, we use Teal Fluorescent Protein (mTFP)17 as our cargo of choice. We restored the natural dimerization of the monomeric TFP to form dTFP with E-coil (dTE). In the first approach, dTE should form a more stable complex with the capsid protein dimer with K-coil (CK) (dTECK) compared to the monomeric form, therefore, directing a larger number of fluorescent proteins into the CCMV capsids. Alternatively, genetically engineering a TFPlinker-CCMV fusion protein (hereby referred to as HTC) abolishes the influence of dissociating E−K coiled-coils and should therefore further increase the loading efficiency in
In nature, the packaging of molecules into discrete compartments is an essential feature for maintaining cellular efficiency. Organelles are a classical example, where simple building blocks are used to assemble complex, well-defined nanostructures. The structure, function, and efficiency of organelles in compartmentalization and catalysis serve as a constant source of inspiration. However, there is still a limited understanding of how molecules behave in such molecularly confined spaces. Although various approaches to mimic nanoconfinement have been demonstrated,1−7 there are few examples of nanocontainers that are as well-defined and monodisperse as virus-like particles (VLPs). Here, we use the virus capsid of the Cowpea Chlorotic Mottle Virus (CCMV) to study the effects of molecular confinement. CCMV has a well-known assembly/ disassembly pathway that can be controlled by pH and ionic strength. This allows us to gain molecular control over loading of cargo into the CCMV.8 Although the assembly of native CCMV relies on electrostatic interactions between the positively charged N-terminus of CCMV and the negatively charged cargo (e.g., viral RNA), CCMV has been used to encapsulate various non-native materials in vitro, such as anionic polymers and nanoparticles.9−11 For the encapsulation of non-negative materials, such as neutral or positively charged proteins, noncovalent affinity tags have been previously reported and used to direct loading inside CCMV.6 Other VLPs, such as Lumazine synthase from Aquifex aeolicus, Qβ, and the capsid of bacteriophage P22, have also been successfully used as nanocontainers for the encapsulation of proteins.12−15 However, one of the major challenges is the © 2013 American Chemical Society
Received: October 25, 2013 Revised: December 20, 2013 Published: December 23, 2013 558
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
Design and Molecular Cloning of mTFP−CCMV Capsid Fusion Protein. The XhoI site (underlined) was introduced into the pET15b CCMV K-coil2 (CK) vector, between the K-coil and the capsid protein, by site-directed mutagenesis using the following primer: 5′-CGC CGC CCT GAA GGA GCT CGA GAT GTC TAC AGT CGG AAC AGG G-3′. The sequence for His6-tagged TFP was amplified by PCR from the pET15b TFP-E-coil vector using the following forward primer: 5′-GGA ATT GTG AGC GGA TAA-3′ and reverse primer: 5′-GCG GCT CGA GGC CAC CAG AGC CGC TCC CGC GCG GCA CCA GAC CCT GCC ACC CGT GTA CAG CTC GTC CAT GCC GTC-3′ (XhoI is underlined). These primers also introduce the sequence for flexible linker between TFP and the capsid protein and a unique XhoI site in the PCR product. Double digestion by XbaI and XhoI of both the PCR product and the mutated pET15b CCMV K-coil vector and subsequent ligation resulted in the pET15b His-TFP-Capsid (HTC) vector. The amino acids sequence of the product is shown in the Supporting Information, Figure S1. Recombinant Protein Expression and Purification. The pET15b plasmids containing CK, TE, dTE, and HTC were transformed into Escherichia coli BL21(DE3)pLysS (Novagen) for protein expression. Starting cultures were grown overnight at 37 °C from glycerol stock cells in 7 mL of LB medium (Sigma) containing 100 μg/mL ampicillin and 34 μg/mL chloramphenicol (Sigma). The overnight cultures were used to inoculate 0.5 L of LB medium containing ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL) and grown to an optical density of OD600 = 0.6−0.8. Protein expression was induced following addition of isopropyl β-Dthiogalactoside (IPTG) to a final concentration of 0.1 mM at 30 °C for 4−5 h. The cells were harvested by centrifuging (10.000 g, 15 min) and the cells were lysed using BugBuster according to the manufacturer’s protocol (Novagen). The TE, dTE, and HTC proteins (all bearing a His-tag) were purified using nickel-affinity column chromatography with a modified version of the suppliers protocol (Novagen). TE, dTE, and HTC were bound and washed with 0.1 M phosphate buffer, 0.3 M NaCl, 12.5 mM Imidazole, pH 8.0. For the noncovalent variants, the nickel immobilized TE and dTE were then mixed for 1 h with the CK to form the TECK or dTECK complexes, respectively. TECK, dTECK, and HTC were then washed with 10 column volumes of 0.1 M phosphate buffer, 1.5 M NaCl, 12.5 mM imidazole, pH 8.0, and eluted with 0.1 M phosphate buffer, 1.5 M NaCl, 0.25 M imidazole, pH 8.0. TECK, dTECK, and HTC were stored in 50 mM Tris, 0.5 M NaCl, 10 mM MgCl2, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 7.5. All formed complexes, namely, TECK, dTECK, and HTC, were further purified using size exclusion chromatography (SEC) using a fast protein liquid chromatography (FPLC) Ä KTA purifier. The complexes were purified using a Superdex 75 column (GE healthcare), 1 mL injection volume, and the same buffer as they were stored in to remove noncomplexed fluorescent protein, and their purity was confirmed by SDS and native-PAGE analysis (see Supporting Information, Figure S2). Quantification of Encapsulation Efficiency. Quantification of fluorescent proteins per capsid was determined based on absorbance spectroscopy measurements and calculations based on the Beer− Lambert law (see Supporting Information, Figure S3 and equations). Standard calibration curves were performed to establish the extinction coefficient of the fluorescent protein at both pH 7.5 (during the assembly/mixing step) and pH 5.0 (after encapsulation; see Supporting Information, Figure S4). Based on the unique absorbance peak at 462 nm (corresponding to the fluorophore of TFP), the concentration of TFP could be determined. For all monomeric, dimeric, and fusion TFP variants, it was assumed that each TFP was bound to a single coat protein of CCMV. The encapsulation efficiency and quantification of TFP per capsid were determined after further purification by size-exclusion chromatography (at pH 5.0). Capsid Assembly. The growth and purification of the native wildtype CCMV virus (native CCMV coat protein) from Vigna unguiculata leaves was performed as reported previously.11 To reassemble the CCMV capsid, the isolated native CCMV coat proteins were mixed
CCMV (Scheme 1). In this work, we performed systematic studies to compare the loading capacity and efficiency into Scheme 1. Schematic Representation of the Different Design Principles Described in This Work for the Rational Loading of CCMVa
a The first approach utilizes leucine zipper like E-coils tethered to (a) monomeric TFP or (b) dimeric TFP, which form electrostatic interactions at pH 7.5 with the complementary K-coil tethered to CCMV. The second approach utilizes a genetically engineered TFPpeptide linker-CCMV fusion protein. In all cases, lowering the pH to 5.0 promotes the encapsulation of TFP cargo inside CCMV.
CCMV capsids and investigate the effects of molecular confinement. With the new design we believe that it is possible to improve molecular control over quantitative capsid loading and accurately predict the average loading of cargo into CCMV.
■
EXPERIMENTAL SECTION
Site-Directed Mutagenesis and Molecular Cloning of pET15b-mTFP. Using site-directed mutagenesis (QuikChange mutagenesis, Stratagene), a unique restriction site for BamHI was introduced into the pET15b EGFP-E-coil expression vector using the following primers: forward 5′-GGT GCC GCG CGG GAT CCA TAT GCT CGA GAA AAG AG-3′ and its reverse complement (BamHI restriction site underlined). The DNA fragment encoding mTFP was excised from the pNCSmTFP DNA vector (Allele BioTech, U.S.A.) by double digestion with the BamHI and BsrGI restriction enzymes. pET15b EGFP-E-coil was double-digested using the same enzymes, hence, removing EGFP for subsequent replacement by TFP. The frameshift caused by the cloning was corrected by site-directed mutagenesis with primer: 5′-GGT GCC GCG CGG GAG CCA TAT GGT GAG CAA GGG CG-3′ and its reverse complement. This resulted in a pET15b TFP-E-coil (TE) vector. All sequences were confirmed by DNA sequencing (Eurofins MWG Operon, Germany). The native dimerization of TFP was restored by reintroducing the mutations into the A-C dimerization site (D144E, A145P, R149I, K162S, K164S) 11 by site-directed mutagenesis, using the primers: 5′CCA CCG GCT GGG AGC CCT CCA CCG AGA TCA TGT ACG TGC GCG-3′ (D144E, A145P, R149I, mutations are underlined), 5′GCT GAA GGG CGA CGT CAG CCA CAG CCT GCT GCT GGA GGG CG-3′ (K162S, K164S, mutations are underlined), and their reverse complementary strands. The dimeric TFP-E-coil (dTE) variant was confirmed by DNA sequencing (Eurofins MWG Operon, Germany). 559
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
with the recombinant protein complexes (TECK, dTECK, HTC) in various ratios in 50 mM Tris-Cl, 500 mM NaCl, 1 mM EDTA (pH 7.5). For all capsid reassembly mixtures, a final concentration of 75 μM CCMV protein was maintained (combined native CCMV coat protein and recombinant CK proteins). The mixtures were dialyzed overnight to 50 mM sodium acetate, 1 M NaCl, 1 mM NaN3 (pH 5.0) to induce capsid formation. The assembled capsids were purified and analyzed by FPLC using a Superose 6 column (see Supporting Information, Figure S5). PAGE Analysis. Polyacrylamide gel electrophoresis (PAGE) was performed both in the presence of sodium dodecylsulfate (SDS) and under native conditions. For SDS-PAGE, samples were heated to 99 °C for 5 min in the presence of 2-mercaptoethanol and 1% SDS. UV−vis Analysis. All UV−visible measurements were performed on a Perkin-Elmer Lambda 850 Spectrometer. Standard quartz cuvettes were used. Fluorescence Anisotropy. A Varian Cary Eclipse spectrophotometer was used to measure fluorescence anisotropy of all samples. Samples were measured in quartz cuvettes with 450 nm wavelength light excitation. The anisotropy was averaged over the range of 480 to 520 nm. Transmission Electron Microscopy (TEM). TEM micrographs were recorded on an analytical FEG-TEM (Phillips CM 30) operated at 300 kV acceleration voltages. Samples were prepared by placing a 5 μL drop of the samples on Formvar carbon-coated copper grids (Electron Microscopy Sciences). The sample drop was left on the grid for 5 min, after which time the excess buffer was blotted away with filter paper. Samples were negatively stained by applying 5 μL of stain (1% w/v uranyl acetate in Milli-Q water) onto the grid and removing the excess stain away after 1 min with filter paper. The samples were dried overnight before imaging.
maximum upon increasing mixing ratios, which also suggests an optimal loading. When the amount of TFP in the assembly mix is higher than 7.5 μM, the total number of TECK and dTECK that are effectively encapsulated approaches a maximum, suggesting an apparent saturation point exists (Figure 1, red and green curves, respectively). Overall, this trend shows that the loading efficiency of dTECK is significantly improved compared to that of the monomeric TECK construct, but still not exceeds ∼6 guest proteins per capsid. These observations also suggest that the E-coil homotrimerization is more pronounced in the monomeric TECK, which subsequently decreases the overall loading efficiency, in line with previous results with the monomeric system.16 By comparison, the HTC fusion variant shows a systematic increase in loading number and efficiency, with increasing concentration ratio. This further confirms that the saturation points observed for both TECK and dTECK are likely due to the homotrimerization of the coiled-coils at higher concentrations of guest material. However, when the concentration range for HTC loading is extended over 40 mol % HTC and 60 mol % native CCMV coat protein, the loading efficiency decreases dramatically and a large amount of erroneous capsids are formed (see Supporting Information, Figure S5). This is attributed to internal overcrowding, which does not allow more cargo in the VLP interior, and consequently, no additional capsid proteins needed to complete the VLP shell can associate with the assembly. Improving the Directed Loading. Figure 1 shows the linear correlation between the mixing ratios (0−40 mol %) and the number of HTC encapsulated. This calibration curve was subsequently used to precalculate and predict a desired number of TFPs to be encapsulated. Indeed, we consistently obtain excellent agreement between the predicted and the obtained number of TFPs encapsulated, confirming that HTC loading is not only reliable but also highly controllable (Figure 2). Interestingly, upon increasing the loading of TFPs inside CCMV, the yield of CCMV VLPs formed decreases dramatically. Based on the relative concentration ratios between HTC and CCMV, in the presence of 10 mol % HTC, 18 TFP should be loaded (10% of 180 CCMV monomers). However, we observe approximately 25% efficiency in effective loading
■
RESULTS AND DISCUSSION Determining the Loading Efficiency. Figure 1 shows the assembly behavior and efficiency of the different complexes,
Figure 1. Typical capsid TFP loading as a function of percentage of complex (TECK, dTECK, or HTC) in assembly.
whereby the amount of complex was varied from 0 to 40 mol % in combination with 100 to 60 mol % native CCMV coat protein. Larger amounts of complex were attempted, however, a significant decrease in assembly efficiency was observed (see Supporting Information, Figure S6). Since dTECK forms a homodimer between two TFP molecules as well as the heterodimer between each TFP and CCMV, a more stable complex is expected. Indeed, dTECK appears to be more efficient compared to the monomeric TECK, leading to a higher number of fluorescent proteins being encapsulated (Figure 1). Interestingly, the loading ratio approaches a
Figure 2. Plot showing (left axis) the average vs predicted loading of TFPs (HTC) per CCMV capsid (blue curve) and (right axis) the normalized number of capsids formed calculated based on the amount of capsids obtained relative to the total amount of protein in each sample (gray curve). 560
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
(∼5 TFP; Figure 1). Since the 25% effective loading is consistent within the 10−40 mol % range, this may suggest that cargo loading may be a (equilibrium) process in which three additional native CCMV coat proteins (CP) per TFP are needed in order to include the latter. At higher TFP/CP ratios, insufficient CP is available in line with the lower yield and incomplete capsid formation (vide infra). Transmission electron microscopy (TEM) of CCMV loaded with 2 TFPs show highly monodisperse particles with an average diameter of ∼27 nm, which is consistent with native T = 3 particles (Figure 3A). The inclusion of the uranyl acetate
Table 1. Fluorescent Anisotropy Values for the Different CCMV Loading Components sample free FPs
Complexed FPs
encapsulated FPs (1−3 FPs per capsid)
a
TFP TFP−E-coil dTFP−E-coil TECK dTECK HTC TECK encaps. dTECK encaps. HTC encaps.
anisotropya (490−520 nm) 0.31 0.32 0.33 0.33 0.33 0.33 0.36 0.34 0.35
The standard deviation in the data ≤0.01.
did not show significant changes compared to the free TFP and dTFP forms (with E-coil). We suspect that the increase in anisotropy typically attributed to increased restricted rotational freedom is offset in this case by the decrease in anisotropy observed due to homo-FRET. Upon encapsulation of low amounts of fluorescent proteins (1−3 TFPs) into virus capsids, we observe a distinct increase in anisotropy values for all complexes. This increase is attributed to the successful incorporation of FPs into the capsid, which gives rise to a decrease in rotational mobility. Under these conditions, the bulk of the assembled capsid (3.7 MDa, 28 nm) restricts the movement of the fluorescent proteins (26.9 kDa, 3 nm) and increases the anisotropy, suggesting that the fluorescent proteins are still tethered to the capsid wall. Homo-FRET could decrease the anisotropy if multiple fluorescent proteins are in close proximity. However, at these low loading levels, homo-FRET is expected to play a minor role since fluorescent proteins encapsulated within one capsid are likely to be further apart than the effective distance for homoFRET. We only expect a difference from this trend for dTECK, which has the tendency to dimerize. This is also consistent with the anisotropy data, in which dTECK displays the lowest anisotropy of the encapsulated complexes although the measured differences are small. Additional fluorescence anisotropy measurements were performed using the HTC variant, since this showed the most efficient, predictable and controllable loading. Upon increasing the amount of encapsulated HTC, a decrease in the anisotropy value is observed (Figure 4). Increasing the number of encapsulated HTC per capsid results in a decrease of average distance between HTCs (Figure 4). This molecular crowding of the fluorophores gives rise to an increase in homo-FRET between the fluorophores within the capsid interior, as was also seen by Mullaney et al. and O’Neil et al.18,19 The anisotropy approaches a steady-state for samples with an average loading of 10 encapsulated fluorescent proteins. This indicates that for 10 or more encapsulated TFPs, the effect on the anisotropy has already reached its maximum. Nonetheless, the efficiency of homo-FRET is strongly dependent on the interfluorophore distance and the fluorophore specific Förster radius (R0).20 TFP has no reported Förster radius for homoFRET, but based on interactions of TFP with other fluorescent proteins,20 an R0 ∼ 5.5 nm is estimated (references within report R0 = 4.6−6.2 nm). When the interfluorophore distance is less than 1.5 R0, homo-FRET is known to be much more
Figure 3. Transmission electron microscopy (TEM) images of CCMV loaded with (A) 2 TFPs per capsid and (B) 20 TFPs per capsid. The samples were negatively stained with 1% uranyl acetate.
staining suggests a significant void volume inside these VLPs. The particles filled with 20 TFPs also show mostly intact T = 3 assemblies, however, some irregularity in size and shape are also observed (particles of 32−38 nm). This may suggest that some exhibit a partially swollen morphology and that the capsids are filled, which is also supported by the relative intensity of uranyl staining. Studying Molecular Crowding by Fluorescence Anisotropy. To obtain a better understanding of the molecular crowding effects and the rotational freedom of the TFP localized inside the CCMV capsids, we performed steadystate fluorescence anisotropy measurements. Fluorescence anisotropy (r) describes the extent of depolarization of the fluorescence upon excitation with polarized. A fluorophore with rotational correlation time shorter than its fluorescence lifetime undergoes multiple rotations within the lifetime of the molecule, consequently emitting fluorescence with a polarization orientation that is not aligned with that of the excitation light. This emission depolarization translates into a lower steady-state fluorescent anisotropy value (r). Depolarization of emission can also result from energy transfer processes between identical fluorophores in nanometer proximity, a process called homo-Förster resonance energy transfer (homo-FRET). In case that the two fluorophores are not aligned parallel, this energy transfer will result in a depolarization of the emission and thus a decrease of anisotropy when the light is emitted from the fluorophore that the energy was transferred to. Here, we employed fluorescence anisotropy to monitor changes in TFP rotational mobility during encapsulation as well as the effects of molecular crowding on TFP resulting in homo-FRET inside the capsids. Before complexation to native CCMV coat protein, we observe a higher anisotropy for the dimeric TFP than the monomeric TFP, suggesting that the restricted rotational freedom due to dimerization of the TFPs has a more profound effect than the homo-FRET (Table 1). Surprisingly, the fluorescence anisotropy for the TECK and dTECK complexes 561
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
Article
Figure 4. Relationship between the fluorescence anisotropy vs number of fluorescent proteins encapsulated showing the decrease in anisotropy with increasing number of TFP molecules. The solid gray line serves as a guide to the eye.
efficient.20 For TFP, this threshold distance is calculated to be 8.2 nm or less. CCMV has an average internal diameter of 18 nm. Assuming a perfect sphere, the maximum distance in between the fluorophores can be approximated by a ‘Tammes problem’.21 This mathematical model calculates the maximal possible distance between a number of points on a sphere. If we take into account the size of the TFP molecule (3 nm in diameter, based on the 2HQK PDB entry) with the fluorophores located in the middle of the protein molecule, the fluorophores can be distributed over a sphere with a 15 nm diameter. Using the solution from the Tammes problem, we find that a loading of 10 TFPs is sufficient to ensure that the fluorophores are at distances below 8.2 nm, which will allow for efficient homoFRET. It should be noted that this is based on the assumption that all TFPs are equally distributed and oriented against the inner capsid wall. However, the TFPs are more likely to be randomly oriented (rather than being docked along the capsid shell wall) and due to the flexible linker, a distribution diameter of 10−12 nm is more likely to be the case (Figure 5, top). In this case, according to the Tammes curves, 4−6 fluorescent proteins are already sufficient for possible homo-FRET to occur. Finally, the distances derived from solving the Tammes problem are the distances for an equal distribution of the TFPs over the inner surface. A deviation from the equal distribution, which we expect in our case, will bring some of the TFPs closer together and will result in increased homo-FRET efficiency.
Figure 5. (top) Schematic representation of TFP location inside the 18 nm diameter CCMV internal cavity; (bottom) the solutions of the “Tammes” problem for different sphere diameters.
work demonstrated the encapsulation of up to 100 supercharged GFPs or GFP-mCherry fusion pairs within the capsid of lumazine synthase or P22, respectively, our strategy provides a unique means to control the cargo loading number in a highly predictive manner. Our approach offers two distinct advantages: (1) it allows the native cargo to be encapsulated and does not require the introduction of surface charges to promote cargo loading and (2) this strategy can be readily extended to other cargo proteins by simply replacing the TFP cargo for alternative cargo of choice. From our fluorescence anisotropy studies, it is apparent that loading of 10 or more TFPs per capsid induces molecular crowding effects and that the packing is sufficiently close that maximal homo-FRET occurs. Figure 4 shows that the fluorescence anisotropy approaches a minimum, which is attributed to homo-FRET effects and provides an effective means to measure the loading capacity of guest materials in virus-like assemblies. Similar molecular crowding effects have been reported by O’Neil et al., whereby covalently tethered FRET pairs (GFP and mCherry) were encapsulated in the bacteriophage procapsid P22; however, in this example, the studies were not quantitative since contributions from both intra- and intermolecular FRET were observed. Interestingly, the values obtained in this work correlate to 19% volume occupancy, which is close to the 20−30% volume occupancy predicted for macromolecules in cells and organelles22 and could suggest that CCMV serves as an excellent model system for understanding molecular crowding in cells and organelles.
■
CONCLUSIONS Based on our present understanding of the CCMV VLP, we evaluated in detail the loading of this VLP with FPs following two strategies. Using one approach based on the leucine zipperlike “E−K” coiled-coil peptides, we observe a distinct nonlinear behavior for the encapsulation of monomeric and dimeric TFP variants. While we observe a 3-fold improvement in the loading efficiency for the dimeric TFP leading to 6 TFPs being encapsulated, a clear saturation point is observed. Based on the theoretical volume of CCMV (18 nm diameter sphere), we calculate that 6 TFPs only occupy 5% of the interior volume (assuming the volume of a cylinder). The second strategy, in which TFP was covalently bound to the monomeric capsid protein (HTC), allows for the predictive and more controlled loading of fluorescent proteins inside CCMV. The loading efficiency of TFP was significantly improved through this method, resulting in up to 20 TFPs loaded inside CCMV while maintaining a reasonable yield of assembly. While previous 562
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
Biomacromolecules
■
Article
(17) Ai, H. W.; Henderson, J. N.; Remington, S. J.; Campbell, R. E. Biochem. J. 2006, 400, 531−540. (18) Mullaney, J. M.; Thompson, R. B.; Gryczynski, Z.; Black, L. W. J. Virol. Methods 2000, 88, 35−40. (19) O’Neil, A.; Prevelige, P. E.; Basu, G.; Douglas, T. Biomacromolecules 2012, 13, 3902−3907. (20) Sun, Y.; Wallrabe, H.; Seo, S. A.; Periasamy, A. ChemPhysChem 2011, 12, 462−474. (21) Erber, T.; Hockney, G. M. J. Phys. A: Math. Gen. 1991, 24, L1369−L1377. (22) Ellis, R. J. Curr. Opin. Struct. Biol. 2001, 11, 114−119.
ASSOCIATED CONTENT
S Supporting Information *
The complete amino acid sequence of the fusion construct HTC is provided in Figure S1. Additional native gel electrophoresis and SDS-PAGE analyses are presented in Figure S2. UV−visible data and calibration curves are shown in Figures S3 and S4, respectively. Size-exclusion chromatograms for the different assembly mixing ratios for each construct are shown in Figure S5. An alternative representation of Figure 2 showing the full range of assembly mixing ratios from 0−100 mol % HTC is provided in Figure S6. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail: [email protected]. Fax: 0031 53489 4645. Tel.: 0031 53489 4380. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge financial support by the Chemical Council of The Netherlands Organization for Scientific Research (NWO−CW), the European Science Foundation (ESF), and The Netherlands Royal Academy for Arts and Sciences (KNAW).
■
REFERENCES
(1) Vriezema, D. M.; Comellas Aragonès, M.; Elemans, J. A. A. W.; Cornelissen, J. J. L. M.; Rowan, A. E.; Nolte, R. J. M. Chem. Rev. 2005, 105, 1445−1490. (2) Kang, S.; Uchida, M.; O’Neil, A.; Li, R.; Prevelige, P. E.; Douglas, T. Biomacromolecules 2010, 11, 2804−2809. (3) Baumler, H.; Georgieva, R. Biomacromolecules 2010, 11, 1480− 1487. (4) Rhee, J.-K.; Hovlid, M.; Fiedler, J. D.; Brown, S. D.; Manzenrieder, F.; Kitagishi, H.; Nycholat, C.; Paulson, J. C.; Finn, M. G. Biomacromolecules 2011, 12, 3977−3981. (5) van Dongen, S. F.; Nallani, M.; Cornelissen, J. J. L. M.; Nolte, R. J. M.; van Hest, J. C. Chemistry 2009, 15, 1107. (6) Minten, I. J.; Hendriks, L. J. A.; Nolte, R. J. M.; Cornelissen, J. J. L. M. J. Am. Chem. Soc. 2009, 131, 17771−17773. (7) Bode, S. A.; Minten, I. J.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Nanoscale 2011, 3, 2376−2389. (8) Lavelle, L.; Gingery, M.; Phillips, M.; Gelbart, W. M.; Knobler, C. M.; Cadena-Nava, R. D.; Vega-Acosta, J. R.; Pinedo-Torres, L. A.; Ruiz-Garcia, J. J. Phys. Chem. B 2009, 112, 3813−3819. (9) Douglas, T.; Young, M. Nature 1998, 393, 152−155. (10) de la Escosura, A.; Verwegen, M.; Sikkema, F. D.; ComellasAragones, M.; Kirilyuk, A.; Rasing, T.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Chem. Commun. (Cambridge, U. K.) 2008, 1542. (11) Sikkema, F. D.; Comellas-Aragones, M.; Fokkink, R. G.; Verduin, B. J.; Cornelissen, J. J. L. M.; Nolte, R. J. M. Org. Biomol. Chem. 2007, 5, 54. (12) Seebeck, F. P.; Woycechowsky, K. J.; Zhuang, W.; Rabe, J. P.; Hilvert, D. J. Am. Chem. Soc. 2006, 128, 4516−4517. (13) Worsdorfer, B.; Woycechowsky, K. J.; Hilvert, D. Science 2011, 331, 589−592. (14) O’Neil, A.; Reichhardt, C.; Johnson, B.; Prevelige, P. E.; Douglas, T. Angew. Chem., Int. Ed. 2011, 50, 7425−7428. (15) Fiedler, J. D.; Brown, S. D.; Lau, J. L.; Finn, M. G. Angew. Chem., Int. Ed. 2010, 49, 9648−9651. (16) Minten, I. J.; Nolte, R. J. M.; Cornelissen, J. J. L. M. Macromol. Biosci. 2010, 10, 539. 563
dx.doi.org/10.1021/bm4015792 | Biomacromolecules 2014, 15, 558−563
