Enzyme Stabilization by Virus-Like Particles

Abstract

The properties of enzymes packaged within the coat protein shell of virus-like particles (VLPs) were studied to provide a comprehensive assessment of such factors. Such entrainment did not seem to perturb enzyme function, but it did significantly enhance enzyme stability against several denaturing stimuli such as heat, organic solvents, and chaotropic agents. This improvement in performance was found to be general and independent of the number of independent subunits required and of the number of catalytically active enzymes packaged. Packaged enzymes were found by measurements of intrinsic tryptophan fluorescence to retain some of their native folded structure even longer than their catalytic activity, suggesting that protein folding is a significant component of the observed catalytic benefits. While we are unable to distinguish between kinetic and thermodynamic effects − including inhibition of enzyme unfolding, acceleration of refolding, and biasing of folding equilibria − VLP packaging appears to represent a useful general strategy for the stabilization of enzymes that operate on diffusible substrates and products.

Graphical Abstract

Given their ubiquity in therapeutics, catalysis, and a host of other applications, the production and stabilization of proteins has been extensively studied.^1–6 Since proteins are large molecules that require proper folding to achieve their function, many of the manufacturing or operational conditions used for nonbiological molecules or materials, such as elevated temperatures, nonaqueous solvents, proteolytic reagents, high salt concentrations, and extreme pH, often compromise the desired structures and properties of proteins.⁷ The most common strategies for their stabilization have been immobilization, chemical modification, and the engineering of their amino acid sequences or the media in which they are used.^8–14 Among these methods, immobilization has proven to be the most widely employed for enzymes in chemical synthesis.^15,16

A variety of supporting scaffolds can be used, ranging from macroscopic surfaces to particulate nanostructures, and enzyme immobilization is usually accomplished by one of three methods: adsorption, covalent cross-linking, and entrapment/encapsulation.^17–19 The first can be compromised by enzyme leaching and surface-enhanced denaturation, and the second often results in decreased enzyme activity. Encapsulation offers a potential solution to both problems, and many different containers have been used, including liposomes,^20–23 mesoporous silicates,^24–26 metal−organic frameworks,²⁷ and polymer matrices.^28–30 However, the enzyme stabilizing effects of entrainment are less certain.

Because protein stability is usually the most serious issue, it is no accident that the enhancement of tertiary and quaternary structure stability toward heat or (less commonly) chaotropic agents is among the first and most important uses of directed evolution in protein engineering.^31–33 In general, stabilization can be achieved by enhancing the overall stability of the desired structure, slowing the rate of its unfolding, or a combination of these thermodynamic and kinetic factors.^34–36 Several years ago we described a technique to produce enzymes in E. coli and simultaneously encapsulate them inside highly stable virus-like particles (VLPs) by RNA-mediated noncovalent association.^37,38 The recombinant VLPs are derived from the Leviviridae Qβ or PP7, each comprised of 180 identical copies of a small coat protein.^39–41 These VLPs are stable to extremes of temperature and pH and can act as a protective shell for encapsidated enzymes against denaturation or precipitation upon long-term storage, proteolysis, and hydrophobic adsorption.^37,38 When entrained inside a VLP, the method for the isolation and purification of every enzyme is the same convenient operation as for the VLP alone, making the system highly modular for the production of different catalytic nanoparticles as long as the substrates and products of the enzyme are small enough to diffuse through the capsid shell.³⁸

We describe here an exploration of the stability of Qβ-encapsidated enzymes toward a greater range of denaturing stimuli than before, including heat, organic solvents, and chaotropic agents. We have found significant enhancements in enzyme stability including the ability to perform reactions at greatly accelerated rates because of greater enzyme tolerance toward elevated temperatures. We also provide spectroscopic data that suggest that loss of enzymatic activity need not be due to the complete unfolding of the encapsidated catalyst. Enhancements in performance were found for all three enzymes tested, and for different average numbers of enzymes per capsid, indicating that the beneficial effects of VLP packaging may be general.

MATERIALS AND METHODS

Cloning, Production, and Purification of Enzyme-Packaged VLPs.

We have previously described the simultaneous production and trapping of proteins inside VLP capsids by virtue of noncovalent associations.³⁷ In this method, expressed mRNA acts as a bridge by associating with the interior surface of the capsid protein (using the canonical “packaging hairpin” sequence) and a positively charged Rev peptide added to the enzyme N-terminus. While some packaging typically occurs without the adapter RNA, the packaging efficiency is usually enhanced with its use. We employed bicistronic plasmids coding for both Qβ coat protein (CP) and Rev peptide-tagged enzymes in the pCDF-1b parent vector as previously described (Figure 1).³⁸ The RNA hairpin loop sequence was appended immediately after the enzyme coding sequence.

Figure 1.

Schematic representation of the method used to package enzymes inside Qβ VLPs. Compatible T7 expression vectors drove simultaneous expression of the capsid protein, Rev-tagged cargo enzyme, and bifunctional mRNA.

The pJF32 plasmid coding for His₆-CD was used as a starting vector for constructing other free 6His-tagged enzymes. All sequences were verified by sequencing before expression. E. coli BL21 (DE3) (Biogen) cells harboring the appropriate plasmids were grown in either super optimal broth (SOB, Amresco) supplemented with 20 mM magnesium sulfate and 50 μg/mL streptomycin. Starter cultures were grown overnight at 37 °C and used to inoculate larger expression cultures. Expression was induced with 1 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) when OD₆₀₀ reached approximately 1.0, and the induced culture was kept at room temperature for overnight expression. Cells were harvested by centrifugation in a JA-10 rotor at 6000 rpm (10 620g), and the pellets were either processed immediately or stored at −80 °C. The cell lysate was prepared by resuspending the cell pellet with 100 mL of 100 mM phosphate buffer (pH 7.0) and sonicating at 30 W for 10 min with 5 s bursts and 5 s intervals. Cell debris was pelleted in a JA-17 rotor at 14 000 rpm (32 776g), and 0.265 g/mL ammonium sulfate was added to the supernatant to precipitate the VLPs. The crude VLP pellet from precipitation was resuspended in 3 mL of 100 mM phosphate buffer (pH 7.0). Organic extraction with 1:1 n-butanol/chloroform was performed to remove lipids and other cellular debris from VLPs. The aqueous layer containing VLPs were further purified by sucrose density ultracentrifugation (10−40% w/v). Particles were pelleted out by ultracentrifugation in a 70Ti rotor (Beckman) at 68 000 rpm (489 600g) for 2 h.

Characterization of Enzyme-Packaged VLPs.

The purity of assembled VLPs was assessed by isocratic size exclusion chromatography with a Superose 6 column on a fast protein liquid chromatography (FPLC) instrument. Non-aggregated Qβ particles were eluted at about 12 mL after the void volume-associated peaks. Dynamic light scattering (DLS) was used to measure the hydrodynamic radius, and a Bioanalyzer 2100 Protein 80 microfluidics chip was used to analyze the average number of enzymes packaged inside the particles. The latter value was determined by normalizing the integration area of coat protein and cargo protein peaks to their respective molecular weights to estimate the molar ratio of coat protein to cargo protein, assuming that equal molar amounts of the two proteins give the same intensity in the staining and detection in the microfluidic Bioanalyzer instrument. A factor of 180 was used to obtain the final calculation of the number of cargo proteins loaded per VLP, as each VLP is composed of 180 coat protein copies. The overall protein concentration was determined with Coomassie Plus Protein Reagent (Pierce) according to the manufacturer’s instructions.

Free Enzyme Production and Purification.

Free (unpackaged) enzymes were expressed as His₆ C-terminal fusions as above. The cleared E. coli cell lysate (from a 250 mL culture) was passed through a cobalt-NTA Talon resin column (0.5 mL bed volume), and the column was washed and eluted (by gravity) according to the manufacturer’s instructions. The fractions containing free enzyme (checked by protein 280 nm absorption) were pooled and dialyzed against three changes of 1 L of 100 mM potassium phosphate buffer (pH 7.0) and concentrated with an Amicon Ultra centrifugal filtration unit (10 kDa MWCO, Millipore). Purity was assayed by chip-based electrophoresis as described above.

Enzyme Activity Assays.

All experiments were performed in triplicate with independently purified particles. All runs with encapsidated enzymes were performed in parallel with purified free enzyme for comparison under the same reaction conditions. Yeast cytosine deaminase (CD) activity was measured by monitoring the conversion of 5-fluorocytosine (5-FC) to 5-fluorouracil (5-FU) using the UV absorbance at 234 nm using a Flex-Station 3 Flash plate reader.^42,43 The molar absorptivities (mM⁻¹ cm⁻¹) at 234 nm of 5-FC and 5-FU were determined by a standard curve to be 6.62 and 2.35, respectively. For determinations of kinetic parameters, 40 μL of a 2× enzyme solution of His₆CD or Qβ@CD was added to 40 μL of 0−2.0 mM substrate in 100 mM potassium phosphate buffer (pH 7.0) and read immediately. A Michaelis−Menten nonlinear fit was used to obtain K_M and k_cat values.

Peptidase E (PepE) activity was analyzed with the fluorescent substrate aspartate-4-amino-7-methyl-coumarin (Asp-AMC, purchased from Bachem). The increase in fluorescence was monitored by a Thermo Varioskan Flash plate reader (excitation 352 nm, emission 438 nm, 5 nm slit, 100 ms read time). For determinations of kinetic parameters, 50 μL of the 2× enzyme solution of His₆PepE or Qβ@PepE was added to 50 μL of 0−1.4 mM substrate in potassium phosphate buffer (pH 7.0) and read immediately.

Purine nucleoside phosphorylase (PNP) activity was assayed by measuring the absorbance at 360 nm of the liberated product (2-amino-6-mercapto-7-methylpurine) from 2-amino-6-mercapto-7-methylpurine riboside (MESG) in the Flex-Station 3 flash plate reader. The kinetic values of His₆PNP or Qβ@PNP were determined by a range of substrate concentrations (0−0.6 mM) in 50 mM Tris-HCl (pH 7.4) with 1 mM potassium phosphate. The activity was initiated by mixing 40 μL of a 2× enzyme solution into 40 μL of substrate solution with the required amount of inorganic phosphate. The molar absorptivity of the product at 360 nm was found to be 1.93 mM⁻¹ cm⁻¹ and was used to convert absorbance to concentration. The initial rates were plotted against the substrate concentration, and a Michaelis−Menten nonlinear fit was used to obtain K_M and k_cat values.

Thermostability, Temperature-Dependent Activity, and Thermodynamic Analysis.

To determine the half-life of the His₆-tagged or Qβ-packaged enzymes at various temperatures, several aliquots of 2× enzyme solutions from a single stock were incubated at the desired temperature. At each time interval, one of the solutions was placed in a room-temperature water bath for 15 min. Then, 40 μL of the resulting enzyme solution was added to 40 μL of the substrate in varying concentrations (0−0.7 mM). Activity measurements were plotted vs time, and a first-order exponential decay nonlinear fit was used to obtain half-life values.

The influence of increasing temperature on activity was performed with an Evolution 200 UV−vis spectrophotometer equipped with a Peltier stage for temperature control. In a typical experiment, 995 μL of 0−0.35 mM substrate in PBS buffer was allowed to equilibrate at each temperature for 5 min, followed by the addition of 5 μL of 200× His₆-tagged or Qβ-packaged enzyme solution (final enzyme concentration 50 nM) with readings starting immediately. The temperature-dependent activity of PepE was determined with a similar technique using a fluorescence plate reader equipped with temperature control. For these experiments, 50 μL of 0−1.4 mM substrate solution in PBS buffer was placed into the well plate and incubated for 5 min at the desired temperature in the plate reader. After this incubation, 50 μL of a 2× enzyme solution (preincubated at the same temperature as the substrates) was added, and readings were commenced immediately.

The natural log of the observed catalytic turnover (k_cat) for each averaged set of experimental data was plotted against the reciprocal of the absolute temperature (T). To calculate the activation energy (E_a), the data were then fit to the Arrhenius eq (eq 1) and Arrhenius plot (ln (k_cat) vs 1/T) using the linear fitting function in Origin software. Activation enthalpy (ΔH^‡) and activation entropy (ΔS^‡) were calculated from the Eyring eq (eq 2) and Eyring plot (ln (k_cat/T) vs 1/T),

$$\text{ln}(k_{\text{cat}})=-\frac{E_a}{R}\left(\frac{1}{T}\right)+\text{ln}A$$ (1)

$$\text{ln}(k_{\text{cat}}/T)=-\frac{\Delta H^{‡}}{R}\left(\frac{1}{T}\right)+\text{ln}\left(\frac{\kappa k_B}{\hslash }\right)\times \frac{\Delta s^{‡}}{R}$$ (2)

where k_cat = catalytic turnover; E_a = activation energy; R = gas constant; A = frequency factor; T = absolute temp; κ = transmission coefficient; k_B = Boltzmann constant; ℏ = Planck’s constant.

Stability of Enzymes in Organic Solvents and Chaotropic Agents.

The stabilities of free and packaged enzymes against organic solvents and chaotropic agents were assayed by incubating aliquots of the enzyme with these additives for different periods of time, followed by kinetic measurements of enzyme activity. Generally, samples were prepared by mixing a buffered solution of 100 nM enzyme (in 100 mM phosphate buffer, pH 7.4, or 50 mM Tris-HCl, pH 7.4, as required for the specific enzymatic reaction) with the candidate organic solvent or guanidine hydrochloride (GuHCl) as the chaotropic agent at the desired concentration. The activity measurement was performed by the same methods as above for each enzyme without purification from the additives. All measurements were referenced to the t = 0 sample, which reports enzyme activity in the presence of the freshly added additive with minimal incubation time.

Monitoring Enzyme Unfolding by Intrinsic Fluorescence.

Steady-state fluorescence measurements were performed on a Horiba Fluorolog 3–21 instrument. Fluorescence was measured at 280 nm excitation wavelength for tryptophan⁴⁴ with a slit width of 5 nm for both excitation and emission. The emission spectra were recorded from 290 to 450 nm. The spectra were recorded three times for each sample, and the data were presented as the mean of such measurements on three independent samples. Samples were prepared by mixing a buffered solution (100 mM potassium phosphate, pH 7.4) of 100 nM enzyme with organic solvents or stock solution of GuHCl at the desired concentration and for the desired time, followed by direct fluorescence measurement in the presence of the additive. For free enzymes, the background spectrum was taken of the buffered solution without an enzyme; for packaged enzymes, the background spectrum was taken of the Qβ VLP lacking any packaged protein, but at the same particle concentration as the enzyme-containing sample.

RESULTS AND DISCUSSIONS

Catalytic Activities of Free and Qβ-Packaged Enzymes.

We expressed three different enzymes encapsidated separately in Qβ particles, each differing in the oligomerization state of the catalytically active form: PepE (28 kDa, a functional monomer), CD (21 kDa, a functional dimer),⁴⁵ and PNP (26 kDa, a functional hexamer),^46,47 as previously described (Figures S1 and S2).³⁸ The enzyme-packaged VLPs are designated as Qβ@(enzyme)_n, averaging n = 12−18 enzyme monomers per VLP. The details of the kinetic analyses are provided in Supporting Information; all of the free and packaged enzymes gave good fits to the single-site Michaelis−Menten equation, suggesting that entrainment of the enzymes did not produce catalysts of different activity within each particle (Figure S3).

Table 1 summarizes the results, showing that the kinetic parameters for free and packaged enzymes were quite similar, consistent with our previous report.²⁹ Thus, while K_M increased for PepE upon encapsidation, but not for CD or PNP, and while k_cat decreased for PepE and CD, but not for PNP, the differences were not very large. Values of catalytic efficiency (k_cat/K_M) of free vs packaged enzymes were within a factor of 3 for all cases. Furthermore, kinetic analysis of the inhibition of free and packaged CD by 2-hydroxypyrimidine, a well-known competitive ligand for the enzyme active site,⁴⁸ gave the same values of K_i for the two forms of the enzyme: 13.5 ± 1.1 μM against free CD and 14.1 ± 1.2 μM against packaged CD (Supporting Information, Figure S4). Taken together, these observations show that the environment of the VLP interior and the requirement for substrate and product to traverse the capsid shell do not substantially affect the ability of these enzymes to form active noncovalent complexes and to perform their molecular transformations. It is worth noting that analysis of one-year-old samples of all of these Qβ@Enz_n particles (stored at 4 °C) showed no change in the number of packaged enzymes (Figure S5), confirming our expectation that enzymes are not able to leak through the capsid, unlike other systems in which enzymes are trapped inside ceramic or polymer nanoparticles.^49,50

Table 1.

Kinetic Parameters of Free and Packaged Enzymes Reported on a Per-Enzyme-Monomer Basis

entry	enzyme	oligomeric state	average packaging number (monomer per VLP)	KM (μM)	kcat (sec−1)	kcat/KM (sec−1mM−1)
1	His6PepE	1		201 ± 14	4.1 ± 0.2	19.9
2	Qβ@PepE12	1	12	284 ± 12	1.8 ± 0.1	6.3
3	His6CD	2		115 ± 8	8.1 ± 0.5	70.4
4	Qβ@CD18	2	18	118 ± 7	4.4 ± 0.3	37.3
5	His6PNP	6		42 ± 3	4.5 ± 0.2	107.1
6	Qβ@PNP16	6	16	37 ± 5	5.4 ± 0.3	145.9

Note that we earlier reported³⁸ far lower values of activity and half-life of packaged PNP: k_cat = 0.85 s⁻¹ and half-life = 12 h, compared to 5.4 s⁻¹ and 500 h. We believe the prior report to be in error, as the measurements were performed under less robust conditions (smaller reaction volumes and no waiting period between heating and activity measurement). The values reported here were reproduced with completely independent preparations of catalytic particles.

Thermostability, Temperature-Dependent Activity, and Thermodynamic Analysis.

With the notable exception of enzymes from thermophilic organisms and those evolved in the laboratory to tolerate high temperatures, enzymatic catalysis is not often done at temperatures significantly greater than 37 °C. For most enzymes, high temperatures induce catalytic inactivation by unfolding of the native conformation of the protein. Consequently, irreversible thermal inactivation is a common parameter for assessing the kinetic stability of enzymes. We characterized the thermostability of free and packaged enzymes in terms of the half-life of kinetic activity (assessed at room temperature, relative to that of the freshly prepared enzyme) after incubation at different temperatures for different lengths of time. These decay processes were all found to be apparent first-order phenomena (Figure 2a,b), and packaging conferred considerable stability. For example, His₆CD activity decayed with a half-life of approximately 1 h at 35 °C, compared to 120 h for Qβ@CD₁₈.⁵¹ At 40 and 45 °C, the packaged enzyme retained approximately 100-fold greater thermal stability than the free CD (half-lives of 49 and 23 h, respectively, vs 30 and 15 min). The packaged CD also showed significant stability toward incubation at 50 and 55 °C (half-lives of 4.8 h and 50 min, respectively), whereas His₆CD was rapidly inactivated under these conditions. These results compare favorably to the thermal stabilization of CD by PEGylation, protein attachment, or immobilization on calcium alginate, which have been reported to enhance stability by factors from 2- to 8-fold,^52,53 and exceed the performance of a mutated sequence computationally designed for thermal stability.^54,55 PepE and PNP were also found to benefit greatly from Qβ encapsidation in terms of thermal stability, showing enhancements of at least a factor of 50 at temperatures up to 50 °C (Figure 2d,e), again consistent with our prior report^37,38 but validated here over a larger temperature range. The packaging inside this type of VLP may, therefore, be a general way to transform thermolabile proteins into more thermostable formulations without changing the enzyme structure.

Figure 2.

Effect of temperature on packaged and free enzymes. (a) Half-lives of free (red) and packaged (green) CD upon incubation at the indicated temperatures. (b) Representative first-order exponential decay of the activity of free and packaged CD after incubation at 40 °C. Kinetic measurements for panels a and b were performed at 25 °C and normalized to the activities of freshly prepared free and packaged enzymes at that temperature. (c) Observed values of k_cat at different temperatures; see Supporting Information for numerical values. (d) Half-lives of free (red) and packaged (green) PepE upon incubation at the indicated temperatures. (e) Half-lives of free (red) and packaged (green) PNP upon incubation at the indicated temperatures. All reported values are averages of three independent measurements ± the standard deviation.

Furthermore, the packaged CD was shown to sustain very high rates of catalysis at elevated temperatures, as shown in Figure 1c. Thus, the saturated rate constants of His₆CD and Qβ@CD₁₈ were essentially the same up to 40 °C, but at higher temperatures, the free enzyme lost activity, whereas the packaged enzyme continued to gain speed, achieving a maximum rate 14 times greater than its own rate at room temperature, and 7 times greater than the rate of a free CD at room temperature (Table S1). In addition to its greater thermal stability, the packaged enzyme may benefit from the easier substrate and product diffusion through the capsid shell at higher temperatures.

The temperature dependence of enzymatic turnover (k_cat) of free and packaged CD gave fairly good fits to linearized Arrhenius and Eyring treatments as shown in Figure 3; Table 2 summarizes the kinetic constants extracted from these plots, along with the same data for PepE (Figure S6). Consistent with the observation of similar Michaelis−Menten parameters, the observed activation energies, pre-exponential factors, and thermochemical parameters (energies and enthalpies of activation) derived here were similar between free and packaged enzymes. This has been generally true for other cases of enzyme immobilization on a variety of materials; for example, activation energies within ±1.2 kcal/mol of those exhibited by the free enzymes.^51,56,57 Therefore, by every criterion tested, packaging within Qβ particles does not seem to perturb enzyme function, with the notable and very useful exception of stabilizing them against thermal deactivation.

Figure 3.

Arrhenius (a) and Eyring (b) plots for free and packaged CD.

Table 2.

Kinetic Parameters of Enzymes

enzyme	Ea(kcal/mol)	A (s−1)	ΔG‡ at 25 °C(kcal/mol)	ΔH‡ (kcal/mol)	ΔS‡(cal/mol·K)
His6CD	13.6 ± 0.l	10.9 × 1010	16.0 ± 0.2	13.2 ± 0.1	−9.7 ± 0.2
Qβ@CD18	13.5 ± 0.l	5.9 × 1010	16.2 ± 0.2	13.3 ± 0.2	−9.9 ± 0.2
His6PepE	11.0 ± 0.2	8.7 × 108	16.8 ± 0.1	10.9 ± 0.1	−19.8 ± 0.l
Qβ@PepE12	10.6 ± 0.l	2.1 × 108	16.0 ± 0.2	8.4 ± 0.1	−25.6 ± 0.3

Stability of Enzymes toward Organic Solvents.

Hydration shells represent an integral part of the protein structure and are essential for protein function.⁵⁸ Consequently, displacement of bound water by organic solvent may result in denaturation or a dramatic change of the protein structure.² Because organic solvents can be helpful to chemical synthesis operations, however, improving enzyme stability in organic solvents has been an industrially relevant target for some time.^3,6

Ethanol is both a protein denaturant and a common solvent for many organic reactions. We found that, in the presence of 1.5 M (8.7% v/v) ethanol in potassium phosphate buffer (0.1 M, pH 7.0) at 25 °C, free His₆CD rapidly lost its activity with a half-life of 1.8 h, becoming completely inactive after 4 h (Figure 4a). In contrast, the packaged enzyme maintained consistent performance (k_cat and K_M) for more than 8 days in this solvent mixture (Figure S7). Higher concentrations of EtOH were found to be more damaging to the enzyme, but packaged CD (in the form of Qβ@CD₁₈) still showed substantial activity in mixtures containing up to 4 M ethanol. For example, in 3 M EtOH (17.5% v/v), the half-life of packaged CD was about 30 h, whereas free CD was completely inactivated within minutes. Thus, entrainment in the VLP provides CD with significant protection from ethanol-induced denaturation, compared to the free enzyme.

Figure 4.

Enzyme activity in the presence of ethanol. (a) Relative activities of free and packaged CD in buffer with 1.5 M EtOH at 25 °C. (b) Relative activities of packaged CD in the presence of increasing concentrations of EtOH. (c) Relative activities of free and packaged PepE in buffer with 4 M EtOH at 25 °C. (d) Relative activities of packaged PepE in the presence of increasing concentrations of EtOH. (e) Relative activities of free and packaged PNP in buffer with 4 M EtOH at 25 °C. Activities of 1.0 denote the same rate as the freshly prepared enzyme in buffer with no organic cosolvent. Measurements at 0 h were obtained immediately after mixing with EtOH. All reported values are averages of three independent measurements ± the standard deviation.

The same kind of behavior was observed for Qβ@PepE₁₂, which lost only a few percent of its activity in 4 M ethanol (23.4% v/v) over 8 days, maintaining the same K_M and k_cat values throughout. In contrast, the free PepE enzyme lost more than 85% of its activity within that time period (Figure 4c,d). Similar results were observed for packaged and free PNP (Figure 4e). This implies that the stabilization effect of Qβ packaging is quite versatile.

tert-Butanol, along with relatively hydrophobic water-miscible solvents such as 1,4-dioxane and tetrahydrofuran, is a strong denaturant.⁵⁹ It proved to be more damaging to enzyme function in our hands as well, instantly eliminating the activity of free CD at 1.0 M (9.4% v/v) and allowing for only 30 min of activity at 0.5 M (data not shown). Again, VLP packaging conferred protection: Qβ@CD₁₈ exhibited a room-temperature half-life of more than 120 and 16 h in the presence of 0.5 and 1.0 M t-BuOH, respectively (Figure 5a). Although reported to stabilize proteins at low to moderate concentrations,^60,61 2,2,2-trifluoroethanol (TFE) was also found to quickly deactivate free CD at either 1.0 M (7.2% v/v) or 0.5 M (3.6% v/v). Again, the packaged enzyme was quite resistant to denaturation at these TFE concentrations (Figure 5b). The same comparative trend was observed for PepE, although this enzyme is overall more resistant to denaturation (Figure 5c,d; Figure S18).

Figure 5.

Enzyme activity in the presence of t-BuOH and TFE. (a) Relative activities of packaged CD in the presence of different amounts of t-BuOH at 25 °C. (b) Relative activities of packaged CD in the presence of different amounts of TFE at 25 °C. (c) Relative activities of free and packaged PepE in buffer with 2.5 M t-BuOH at 25 °C. (d) Relative activities of free and packaged PepE in buffer with 1.5 M TFE at 25 °C. Activities of 1.0 denote the same rate as the freshly prepared enzyme in the absence of an organic cosolvent. Measurements at 0 h were obtained immediately after mixing with t-BuOH or TFE. All reported values are averages of three independent measurements ± the standard deviation.

Measurements of the effects of dimethyl sulfoxide (DMSO), another well-known protein denaturant, on CD were partially frustrated by overlapping absorbance of the solvent and the substrate 5-FC, which gave unacceptably high background signals at DMSO concentrations greater than 0.5 M. We were able to observe that 0.5 M DMSO (3.5% v/v) inactivates free CD within 7 h but has little effect on packaged CD over 8 days (Figure 6a). PepE and PNP responded somewhat curiously to the presence of DMSO, as shown in Figure 6. In most cases, a relatively rapid drop in activity was observed, resulting in a loss of approximately 40% activity in the first 2 h for both free and packaged enzymes. Thereafter, the two forms of both PepE and PNP diverged, with the free enzymes continuing to lose activity (albeit at a slower rate) and the packaged enzymes being much more resistant. As with the other organic cosolvents, K_M values for the packaged enzymes were not significantly different after 8 days in the presence of DMSO than the enzymes in a pure buffer.

Figure 6.

Relative activities of free and packaged enzymes in the presence of the indicated concentrations of DMSO. (a) Cytosine deaminase, (b) peptidase E, (d) purine nucleoside phosphorylase, (c, e) packaged PepE and PNP, respectively, in different DMSO concentrations. Activities of 1.0 denote the same rate as the freshly prepared enzyme in buffer with no organic cosolvent. Measurements at 0 h were obtained immediately after mixing with DMSO. All reported values are averages of three independent measurements ± the standard deviation.

Stability of Enzymes toward a Chaotropic Agent.

A common way to assess the conformational stability of a protein is the measurement of denaturation or catalytic activity in the presence of a chaotropic agent such as GuHCl.^62–64 This species induces protein unfolding by the disruption of hydrogen bonding of structured water molecules, thereby weakening hydrophobic interactions.⁶⁵ We found VLP packaging to stabilize enzymes against this form of attack as well. Thus, in the presence of 75 mM GuHCl, His₆CD lost all catalytic activity within 5 h, whereas Qβ@CD₁₈ retained 90% of its activity after 8 days (Figure 7a). Higher concentrations of GuHCl could not be conveniently tested with CD because of competing UV−vis absorbance by the additive, compromising the kinetic assay. Packaged PepE retained substantial activity in 1.5 M GuHCl for 1−2 days and for 2−4 h in 2.0 M GuHCl. Similarly, packaged PNP exhibited significant activity for up to 1 day in the presence of 1.2 M GuHCl (Figure 7a,c).

Figure 7.

Relative activities of free and packaged enzymes in the presence of the indicated concentrations of chaotropic agent GuHCl. (a) Cytosine deaminase, (b) peptidase E, (d) purine nucleoside phosphorylase, (c, e) packaged PepE and PNP, respectively, in the presence of GuHCl at different concentrations. Activities of 1.0 denote the same rate as the freshly prepared enzyme in the absence of GuHCl. Measurements at 0 h were obtained immediately after mixing with GuHCl. All reported values are averages of three independent measurements ± the standard deviation.

Monitoring Enzyme Unfolding by Intrinsic Fluorescence.

We were able to gain some insight into the folding state of packaged enzymes using the intrinsic fluorescence properties of tryptophan (Trp).^66–69 Since Trp residues in the hydrophobic interiors of proteins lose the fluorescence quantum yield when exposed to solvent, we expected decreases in fluorescence emission intensity at 335 nm (excitation at 280 nm) to reflect partial or complete denaturation of the encapsidated protein. Fortunately, the Qβ capsid protein lacks Trp residues and so does not interfere in this measurement; PNP also has no Trp residues and so was not studied in this manner (Figure S8). For CD and PepE, the fluorescence intensities of 100 nM solutions of the packaged enzyme were very similar to those of the corresponding free enzymes at the same concentration (Figure S8).

Heat-induced enzyme unfolding of cytosine deaminase was followed at 40 °C. The intrinsic fluorescence of free His₆CD enzyme decayed rapidly to a minimum constant value upon extended incubation, suggesting substantial unfolding of the protein (Figure S9a). Independent measurements of protein concentration confirmed that the denatured enzyme did not precipitate during this process (Figure S10). The loss of catalytic activity of the free enzyme correlated quite well with this relative loss in Trp fluorescence emission intensity (Figure 8a). In contrast, packaged CD lost approximately 70% of its catalytic activity at 40 °C within 120 h, but only about 15% of its fluorescence emission intensity over that time (Figure 8b) or longer (Figure S10). A similar result was also observed at 75 °C: both free and packaged CD were quickly inactivated (Figure S11), but the packaged enzyme retained much more of its initial fluorescence intensity even after prolonged incubation. The presence of a high concentration of RNA had little effect on the fluorescence of both folded and denatured unpackaged CD (Figure S12), although it is impossible to replicate the RNA-rich environment of the particle interior in bulk solution. Interestingly, both free and packaged PepE showed significantly slower loss of fluorescence intensity than catalytic activity upon incubation at elevated temperatures, although this disconnect was more pronounced for the packaged enzyme (Figure 8c,d).

Figure 8.

Monitoring enzyme unfolding by intrinsicfluorescence (ex. 280 nm, em. 335 nm). Relative activity and decrease in fluorescence intensity for (a) free or (b) packaged CD at 40 °C and (c) free or (d) packaged PepE at 50 °C.

Enzyme unfolding in the presence of organic solvents (t-BuOH, TFE, DMSO) and GuHCl was also studied by monitoring Trp fluorescence for a subset of samples and conditions (Figures S14−16). To highlight the effect of the additive, values of fluorescence intensity were normalized to those obtained for the same amount of enzyme incubated in the absence of the additive at respective time points. For packaged enzymes, fluorescence intensity, just like enzymatic activity, was highly preserved over extended periods of time under these challenges. The data for unpackaged (free) enzymes were more variable, but in general, overall protein folding (Trp fluorescence) and catalytic activity roughly correlated with each other.

It is therefore clear that encapsulation in the VLP inhibits enzyme unfolding in response to multiple different challenges. The above results also suggest that, for CD and PepE, it may be possible to lose the structural integrity required for proper catalytic function, presumably at the active site,⁷⁰ without full unfolding of the enzyme. Such a multistage process of enzyme denaturation is schematically represented in Figure 9. Alternatively, it may also be the case that Trp residues on fully denatured protein inside the VLP are able to interact with the capsid or packaged protein in such a way as to remain sequestered from the solvent and thereby retain similar fluorescence properties as the properly folded structure.

Figure 9.

Proposed staged loss of enzyme structural and functional integrity, starting with (a) active site destabilization and followed by (b, c, d) unfolding of the rest of the structure. VLP encapsidation apparently slows down both, the latter to a greater extent.

Insight into the Stabilization of Enzymes in Qβ Capsids.

To investigate the nature of VLP-encapsidated enzyme stabilization, we found the activity and stability profiles toward heat and ethanol cosolvent of free cytosine deaminase to be unchanged in the presence of empty Qβ particles (Qβ-wt, Figure 9a,b), ruling out a nonspecific buffering effect of the capsid. To test potential steric crowding effects,⁷¹ inactivation assays were performed with VLPs containing averages of 5 vs 18 copies of cytosine deaminase per particle. These particles displayed very similar UV-absorbance spectra (A₂₆₀/A₂₈₀ ratios) and thus were found to package similar amounts of RNA. No differences in catalytic performance vs time or additive concentration were observed for thermal or organic cosolvent challenge (Figure 10c−e), eliminating the possibility that enzyme concentration within the particle is a significant factor, at least within the range tested. We are left, then, with an inability to assign the general stabilization exhibited by enzymes packaged inside the Qβ to either kinetic or thermodynamic factors.

Figure 10.

Additional stability tests of free and packaged CD. (a) Stability (half-life of catalytic activity) of His₆CD in the absence or presence of wild-type Qβ particles at increasing temperatures. (b) Similar measurements as panel a for incubation in the presence of 1.5 M EtOH. (c) Thermal stability comparison of VLPs with different average numbers of packaged CD. (d) Stability profiles of Qβ@CD₁₈ and Qβ@CD₅ in the presence of EtOH. (e) Stability profiles of Qβ@CD₁₈ and Qβ@CD₅ in the presence of tBuOH and TFE.

Prior studies and computation contain precedents for both phenomena, but thermodynamic factors are more often discussed. Macromolecular crowders like polyethylene glycol (PEG), dextran, protein, and DNA are believed to restrict the conformationally extended states of protein due to volume exclusion,^72,73 leading to entropy-driven stabilization of the folded state.^74–76 Entrapment in tight spaces may also simply bias the available thermodynamic population toward the more compact folded structure.⁷¹ On the other hand, intermolecular interactions between the protein and surrounding species, including surfaces or trapped water molecules, can be either stabilizing or destabilizing and thus may work in opposition to the excluded volume effect.^77,78 Thus, while the RNA concentration inside the Qβ capsid is quite high, perhaps enough to provide an excluded volume environment, it is likely not to be homogeneously distributed⁷⁹ and thus is of uncertain relevance. Likewise, we cannot affirm or rule out the possibility of kinetic stabilization of packaged enzymes by the formation of surface interactions with the capsid protein or packaged RNA. Note also that packaged enzyme stabilization may simply be a byproduct of the fact that the packaged enzymes are prevented by their entrapment from irreversibly aggregating and precipitating after they unfold, and so have an opportunity to refold.

CONCLUSIONS

Among the many strategies commonly used to enhance enzyme stability and activity in aqueous−organic media,³ including protein engineering,⁸⁰ sampling from organic-tolerant microorganisms,^81,82 chemical modification of enzymes,^83,84 and physical immobilization,^15,16 VLP encapsidation during coexpression is process-friendly and conceptually modular, in that it can be applied to many enzymes without altering their amino acid sequences. This work extends previous observations to a larger range of challenges, showing general stabilization against a variety of denaturation stimuli. We believe it is worthwhile to further explore this technique for any enzyme that operates on molecules able to traverse the protective shell. Higher turnover rates are thereby made accessible at elevated temperatures, and slower denaturation may also offer advantages in future studies of the enzyme structure.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.biochem.0c00435.

Particle characterization data, enzyme kinetics and inhibition data, evaluation of protein stability upon extended storage, UV−visible absorption, and intrinsic fluorescence emission data to monitor protein folding stability (PDF)