Stabilizing viral nano-reactors for nerve-agent degradation

Abstract

Virus like particles, and other naturally occurring protein containers, have emerged as excellent building blocks for nanomaterials design and synthesis. Here we exploit a directed assembly and encapsulation approach to sequester multiple copies of a phosphotriesterase (PTE) enzyme within the capsid of bacteriophage P22. Phosphotriesterase, from Brevundimonas diminuta, is an intriguing enzyme as it is highly active against a wide range of harmful insecticides and nerve agents such as Soman and Sarin. However, difficulty in expressing large quantities of the active recombinant enzyme has limited efforts to scale-up its use. Additionally, as a mesophilic enzyme its low heat tolerance and susceptibility to proteolysis makes it a less than ideal candidate as a practical bioremediation tool. Through encapsulation of the PTE within the P22 capsid, we demonstrate a greatly enhanced thermal tolerance of the enzyme, maintaining 50% of its activity to 60 °C. Additionally, the P22 capsid confers protection to the enzyme from proteases, as well as stabilizing the enzyme against desiccation. Thus, our engineered P22 encapsulation system greatly enhances the stability of the mesophilic phosphotriesterase and results in a robust and active nanoparticle reactor.

---

Introduction

Viruses have evolved robust shells that function to transport and protect their nucleic acid cargo from degradation. Viral capsids devoid of their nucleic acid cargo, and other nanoscale containers, have been exploited for their protective properties and precise spatially controlled assembly, for nanomaterials design and synthesis.^1–3 These assembled protein-based capsids have been utilized to entrap a wide range of metals, polymers, and proteins with beneficial effect on both the cargo and the resultant composite material.^4–8 Through encapsulation of cargo molecules within virus-like particles (VLPs), the well-defined volumes of these architectures can be exploited for cargo occupation, restricted and defined particle growth, and unique interfacial chemistry utilizing the high internal surface area of the capsid. The cargo, once sequestered inside the capsid, is forced into proximity with its neighbours in this crowded, confined space. This confinement and crowding may alter the cargo protein properties when compared to the unencapsulated cargo.^5,9,10

Upon encapsulation of enzymes in protein containers we, and others, have observed varied effects on the enzyme kinetics. In some cases, there is little to no effect,^5,11 in others the overall rate seemed to be enhanced upon encapsulation,^9,12 while in other examples the rate (k_cat) and the K_M were decreased.⁵ The kinetic trends and outcomes of high density encapsulation is not yet predictable for this promising new class of biomaterials and at present appear to be enzyme specific.

Viral capsids are very robust and can be utilized to protect non-native cargo, analogous to their natural nucleic acid protection, in a biomimetic approach to developing new active nanomaterials. Many virus like particles (VLPs) assemble into precise architectures, can be stable for long periods of time, and some can withstand dramatic pH changes.¹³ In the case of the bacteriophage P22 capsid, the VLP is soluble to high concentrations (30 mg mL⁻¹), highly resistant to proteases, maintains the capsid assembly across a wide pH (pH 4–10) and temperature (−80 to 85 °C) range, can be lyophilized and rehydrated, and can withstand high concentrations of organics all without disrupting the cage like architecture of the VLP capsid. By exploiting this capsid as a nanocontainer for directed encapsulation of an enzymatic cargo, we hoped to bestow these stabilizing qualities on the cargo enzyme and use this advantageously towards the engineering of novel active nanocatalyst materials.

We have investigated the encapsulation of the highly active phosphotriesterase (PTE) from Brevundimonas diminuta, which has been shown to break down organophosphates, including chemical warfare agents and commercial insecticides.¹⁴ The homodimeric, mesophilic enzyme has been shown to contain a bi-nuclear zinc active site, which can be replaced with cobalt to generate an even more active enzyme.^15,16 The high catalytic activity, and toxic substrate remediation, makes PTE an intriguing candidate for large-scale bioremediation. However, large-scale isolation of the enzyme for practical applications has been hindered due to the difficulty in scale-up expression of active enzyme.¹⁷ When over-expressed in E. coli, the recombinant PTE was found largely in an aggregated inactive form and active enzyme made up only ∼0.1% of the total cell mass.¹⁷ Additionally, as a mesophilic enzyme, PTE has low heat tolerance (loses activity above 45 °C) when free in solution or immobilized on a resin.¹⁸ Therefore, as promising a bioremediation tool as the PTE may be, the poor expression yields and low heat tolerance makes it less applicable for ultimate industrial use. These properties make PTE a prime candidate for protein expression rescue and thermal stability enhancement via incorporation into our synthetic bio-materials approach. Through the encapsulation of PTE within the P22 capsid, we expand the usefulness of this enzyme by increasing the active protein yield in E. coli and by greatly increasing the thermal tolerance of PTE. Our engineered system confers additional advantages by protecting the mesophilic enzymes from proteolytic degradation and lyophilization.

Materials and methods

P22 phosphotriesterase fusion assembler plasmid design

The background plasmid for the design was the pET-Duet vector (Novagen). To create the fusion, first the gene that encodes amino acids 1–141 of P22 scaffold protein were amplified using standard PCR such that a 5′ EcoRI (Fwd primer: aag̲a̲a̲t̲t̲c̲aagggggtgaaggaagtaatgaagatc) and a 3′ SacI (Rvs primer: aag̲a̲g̲c̲t̲c̲cttgttcttctcaggcagcagaga) recognition sequences were added. The digested product was inserted into MCS1 of the pET-Duet vector between the EcoRI and SacI sites. Similarly, the gene for phosphotriesterase (PTE) was amplified off a vector graciously given from Prof. Frank Raushel (Texas A and M University) adding a 5′ BamHI and a 3′ EcoRI site. This gene was also cloned into MCS1 between the BamHI and EcoRI sites such that an N-terminal 6xHis tag was added to the PTE gene. Lastly, the gene for the P22 coat protein was amplified to add a 5′ NdeI site and a 3′ XhoI site and ligated into MCS2 of the pET-Duet vector.

The ligated plasmid was transformed into XL-2Blue ultra competent cells (Agilent Technologies, Santa Clara, CA) and colonies were screened via single pass sequencing for the incorporation of the three genes. The confirmed sequence fusion plasmids were subsequently transformed into BL21 E. coli (EMD Chemicals) for protein expression.

Protein purification

Transformed BL21 (DE3) E. coli were grown to an OD_{600 nm} = 0.6 at 30 °C. The culture was then induced with IPTG (0.1 mmol L⁻¹) and CoCl₂ (1 mmol L⁻¹). The culture was allowed to grow an additional 18 hours at 30 °C with vigorous shaking. The bacteria were separated from the media by centrifugation at 4500 rpm for 10 minutes. The cell pellets were resuspended in HEPES (50 mmol L⁻¹) buffer pH 8.5 and frozen at −20 °C overnight. The next day, the cell pellets were thawed and incubated with DNAse (20 mg mL⁻¹), RNAse (30 mg mL⁻¹), and lysozyme (15 mg mL⁻¹) (Sigma-Aldrich) for 30 minutes at room temperature. Cells were lysed further by sonication (40% duty cycle, 5 minutes three times with a 2 minute cool down between) on ice. The cell debris was removed via centrifugation at 14 500 rpm for 25 minutes.

For the P22 encapsulated samples, clarified lysate (25 mL) was layered on top of a 35% (w/v) sucrose cushion (5 mL) and centrifuged at 48 000 rpm for 50 minutes in an ultra centrifuge (Sorvall WX Ultra). The resulting virus pellet was resuspended in 50 mmol L⁻¹ HEPES buffer pH 8.5 and spun at 17 000g for 20 minutes to remove any particulates and lipid. The virus was then dialyzed against the aforementioned buffer overnight.

The unencapsulated free phosphotriesterase fusion was purified out of the ultra centrifugation supernatant produced in the step above. The clarified supernatant was loaded onto a HisTrap HP Column (Amersham) using an AKTA FPLC (GE Healthcare). Samples were washed with imidazole (20 mmol L⁻¹) in phosphate (20 mmol L⁻¹, 500 mmol L⁻¹ NaCl, pH 8). Protein was eluted with imidazole (250 mmol L⁻¹) and dialyzed against HEPES buffer (50 mmol L⁻¹, pH 8.5) extensively.

Size exclusion chromatography

Protein samples were additionally purified via SEC. Encapsulated samples were centrifuged at 17 000 rpm for 10 minutes and the supernatant was loaded on to a preparative S-500 Sephadex (GE Healthcare) 80 mL column using an AKTA Pharmacia FPLC. The flow rate was 1 mL per minute using HEPES buffer (50 mmol L⁻¹, pH 8.5) and the total elution volume was 150 mL. Fractions were taken from the later half of the 60 mL elution peak.

Size exclusion chromatography – multi angle light scattering

P22 capsid samples were separated by HPLC (Agilent 1200) size exclusion chromatography (WTC-0200S column, Wyatt Technologies) at a flow rate of 0.7 mL min⁻¹. The mobile phase was phosphate buffer (50 mmol L⁻¹, 100 mmol L⁻¹ NaCl, 200 ppm sodium azide, pH 7.2). Samples (25 μL) were injected on the column and run for 30 minutes and elution was monitored using UV-vis detector (Agilent), a refractive index detector (Wyatt), light scattering, and quasi-elastic light scattering was detected with a Dawn 8 (Wyatt). Average M_w, R_g, and R_H were determined from a fit of the data using a Zimm plot in the Astra software from Wyatt.

Enzymatic rate determination

To determine the K_M and V_max of the encapsulated PTE, the sample was diluted to a total protein concentration of 0.198 mg mL⁻¹ and subjected to 11 concentrations of the substrate paraoxon from 6.7 μmol–1.2 mmol L⁻¹. Acetonitrile was kept at a constant 8% of the total reaction mixture and HEPES (50 mmol L⁻¹, pH 8.5) was used as the buffer. The absorbance of the paraoxon degradation product was monitored at 405 nm over five minutes in triplicate for each substrate concentration. The change in Abs 405 nm, was plotted against time and the slope of the linear portion of the graph was determined to be the initial rate of each sample. The triplicate-averaged rates were then plotted against the substrate concentrations and fit to a Michaelis–Menten kinetics model. From this fit, V_max (5.66 × 10⁻⁶) and K_M (136 μmol) were determined. Using the method previously described (ref. 1), the concentration of the PTE enzyme only in the encapsulated sample was calculated to be 5.3 × 10⁻⁷ M. By dividing the V_max by the encapsulated enzyme concentration, the k_cat was determined to be 10.67 M s⁻¹.

Enzyme sensitivity assays

For the heat sensitivity assay, encapsulated and free PTE were heated to 45 °C, 50 °C, 55 °C, or 60 °C for twenty minutes and then allowed to cool to room temperature before assayed. For the proteolysis assay, encapsulated and free PTE were treated with excess trypsin (10 units trypsin per mg of protein) for either 1 or 18 hours at room temperature, in triplicate, before being assayed against an untreated sample. Additionally, encapsulated and free PTE were quickly frozen by immersion in liquid nitrogen and then subjected to lyophilization to complete dryness. The samples were then rehydrated in HEPES (50 mmol L⁻¹, pH 8.5) to the starting volume and assayed for activity. All treated and untreated samples, in triplicate, were assayed for activity with paraoxon (300 μmol L⁻¹) by measuring the change in A405 nm over five minutes. Rates were calculated as described above and compared to the untreated sample average to report a percent activity.

Results and discussion

P22-PTE characterization

The bacteriophage P22 requires the presence of a scaffolding protein (SP) and coat protein (CP) for heterologous assembly of T = 7 capsids. As we have previously reported, genetic fusions of a cargo gene to a truncated SP allow the directed encapsulation of a gene product cargo, in high copy number, within the P22 capsid. The gene for phosphotriesterase (PTE) from Brevundimonas diminuta was cloned upstream of a truncated form of the P22 scaffold protein (SP) as an N-terminal fusion (PTE-SP). The PTE-SP fusion was able to template the assembly of P22 capsids in vivo, when co-expressed with CP, resulting in the encapsulation of the PTE-SP inside the assembled P22 (Fig. 1a). When imaged by transmission electron microscopy, the P22 capsids have an interesting, slightly irregular, morphology (Fig. 1b, see Fig. S1† for more examples). In our experience with the P22 encapsulation system, non-perfect capsids are obtained when the scaffold-cargo fusion expression is low relative to CP expression. With the PTE-SP fusion, E. coli growth conditions had to be optimized to obtain consistent, single shelled, capsids, consistent with poor expression of the PTE enzyme itself as seen by others,¹⁷ and relatively poor expression of the PTE-SP fusion compared to other examples using this approach.^5,10,19–21 Even when growth conditions are optimized, we observed lower than expected cargo encapsulation numbers (∼40 compared to ∼300 for a similarly sized fusion, EGFP-SP141¹⁹) and more mis-formed capsids by size-exclusion chromatography (Fig. S2†).

Fig. 1 Through the P22 encapsulation system, phosphotriesterase (PTE) is specifically sequestered on the interior of the P22 capsid and maintains its catalytic activity. (a) Cartoon representation illustrating the plasmid design on the left and the spontaneously assembled product on the right. (b) Negatively stained TEM image of the P22-PTE capsids. Bar represents 100 nm. (c) As the P22-PTE cleaves the paraoxon substrate, a p-nitrophenol group is released which can be monitored as an increase in the absorbance at 405 nm.

To further understand the role that the scaffold protein expression plays in the capsid assembly, a construct expressing only the P22 coat protein was made. Identical expression conditions were used and the resulting “coat protein only” materials were subjected to size-exclusion chromatography to assess the relative amount of properly assembled capsids. The SEC profile revealed a significant peak that eluted early and corresponds to a population that is mis-assembled into a wide range of irregular structures. A slightly later eluting peak corresponds to well-formed (T = 7) P22 capsid particles. This profile is similar to what was observed with the P22-PTE construct grown under non-optimized conditions. In both cases the ratio of the early (mis-assembled/aggregate) peak to the later eluting (properly assembled) peak, was much higher than what is seen in constructs where the scaffold fusions are well expressed and present in high copy number during the assembly process (Fig. S3†). With the poorly expressing (or no) scaffold protein constructs, the mis-assembled and assembled peaks are not perfectly resolved by SEC, indicating some heterogeneity in the material. When examined by TEM, the images of the poorly expressing (or no scaffold protein) constructs revealed many multiple shelled and irregular capsids (Fig. S4†). These results indicate that the scaffold protein fusion expression level is central to the proper templating of our P22 nanomaterial assembly.

However, despite these challenges, after the optimization of growth conditions and size-exclusion purification, the P22 encapsulation system still produces homogeneous material with good yields (∼150 mg of purified material per liter of E. coli grown or ∼5 g cell paste). Analysis of the purified P22-PTE material by HPLC-SEC coupled to quasi-elastic and multi-angle light scattering (MALS) detectors, revealed that the purified capsids were homogeneous in size with a radius of hydration (R_H) of 30.57 ± 0.12 nm and a radius of gyration (R_g) of 27.87 ± 0.06 nm. The ratio of R_g/R_H for empty P22 is 0.95, which is characteristic of an empty sphere, where as with P22-PTE this ratio is 0.91 suggesting packaging of the cargo enzyme. From MALS, the average total molecular weight for the P22-PTE capsids was 21.8 ± 0.56 MDa as compared to 19.65 MDa for empty P22, indicating packaging of 40 ± 10 copies of the PTE-SP fusions per capsid (see ESI† for calculation).

The enzymatic activity of the encapsulated PTE was monitored at room temperature across 11 substrate concentrations ranging from 6.7 μmol L⁻¹ to 1.2 mmol L⁻¹. Paraoxon was used as the substrate because its enzymatic degradation product p-nitrophenol gives a characteristic absorbance at 405 nm. For each substrate concentration, the rate of paraoxon hydrolysis was determined based on the change in absorbance at 405 nm over 5 minutes (Fig. 1c). Fitting this data to a Michaelis–Menten model gave a K_M of 136 ± 15 μmol L⁻¹ and a k_cat of 10.7 ± 0.5 s⁻¹ (Fig. 2). The reported values of K_M and k_cat, for the unencapsulated cobalt substituted enzyme are 100 μmol and 4870 ± 90 s⁻¹ respectively.¹⁴ Encapsulation of enzymes in protein containers has been observed to produce varied effects on the kinetics of the enzyme. In the case of P22-PTE, the enzymes K_M is slightly increased and the turnover (k_cat) severely decreased. This observation may indicate that not all of the encapsulated enzymes are actually functional or are distorted such that they do not function as well as the PTE-fusion free in solution. However, like classical enzymology, characterizing the individual number of active proteins within a purified preparation is difficult. Considering the P22-PTE as a multi-component nanoreactor, with 40 active enzymes per capsid, the per capsid turnover rate is 400 s⁻¹, still an excellent active nanoparticle catalyst.

Fig. 2 P22-PTE kinetics were measured using paraoxon as a substrate between 6.7 μM–1.2 mM. The plotted rates were fit to a Michaelis–Menten model (R² = 0.99) to determine a V_max of 0.096 ± 0.003 and a K_M of 136 ± 15 μM.

P22-PTE enzyme stability enhancement

The native and recombinant PTE free in solution is temperature sensitive and is reported to lose all activity after heating to 60 °C. However, the temperature sensitivity of the PTE enzyme is greatly improved upon encapsulation within the P22. P22-PTE and unencapsulated PTE were heated to a range of temperatures (up to 60 °C) for 20 minutes, cooled to room temperature, and then assayed for activity. The free PTE lost significant activity upon heating, maintaining only 45% activity when heated to 45 °C, while the P22-PTE maintained more than 90% of its original activity (Fig. 3). After heating to 55 °C, roughly 15% of the activity of the free PTE survived while the encapsulated PTE retained greater than 70% of activity. After heating to 60 °C, the free enzyme had lost almost all activity while the P22-PTE was still more than 40% active. Through sequestration of the enzyme inside the P22 capsid, PTE exhibits a much higher thermal tolerance. This is a promising enhancement of the properties for future materials use.

Fig. 3 Encapsulation of phosphotriesterase within the P22 capsid greatly decreases its thermal sensitivity. Samples were heated at the indicated temperatures for 20 minutes, cooled to room temperature and assayed for activity. While the unencapsulated enzyme quickly loses activity with heat, the encapsulated enzyme maintains >70% activity to 55 °C. At 60 °C, the unencapsulated enzyme has lost almost all activity while the encapsulated maintains ∼45% activity.

The P22 capsid confers an additional level of protection to the encapsulated PTE in that it shields the enzyme from access to proteolytic digestion. P22-PTE and the unencapsulated PTE were treated with an excess of the protease trypsin for either one hour or over-night at room temperature and then assayed for activity. After one hour trypsin treatment, the unencapsulated PTE lost 67% of its activity while the activity of the encapsulated PTE remained unchanged (Fig. 4). Overnight (18 hours) trypsin treatment reduced the unencapsulated PTE activity to 19% of the untreated while the P22 encapsulated PTE maintained greater than 90% of its original activity (Fig. 4). This enhancement of the enzyme stability makes the P22-PTE a significantly better candidate for remediation applications. If PTE were used to remediate insecticide-contaminated wastewater, the enzyme would likely encounter many proteases from an environmental sample. By protecting the enzyme inside the P22 capsid, which is porous enough for small substrate molecules, but restrictive of larger proteases, PTE is better suited for practical applications.

Fig. 4 Encapsulation of the phosphotriesterase (PTE) within the P22 capsid confers heightened resistance to proteolytic degradation and loss of activity after rehydration. Treatment with trypsin for either one hour or overnight greatly decreased the activity of the unencapsulated PTE while the encapsulated PTE maintained activity with either length of treatment. Upon lyophilization and subsequent rehydration, the unencapsulated enzyme lost all activity while the encapsulated enzyme retained 91% of its activity.

Additionally, when the P22-PTE and the free enzyme were lyophilized and then rehydrated, the encapsulated enzyme retained 91% activity while the free enzyme lost all activity (Fig. 4). For long-term storage and use, it may be advantageous to be able to freeze dry active P22-PTE material. Through encapsulation within the P22 capsid, this storage method is now compatible with the PTE enzyme. It is also noteworthy that during the several months of this study, the P22-PTE stored in solution at 4 °C did not lose any measurable activity. However the free-PTE was unstable in solution and slowly precipitated under the same conditions and exhibited inconsistent enzyme activity. The encapsulation of PTE in the P22 capsid thus confers many desirable materials properties to the enzyme making it a more advantageous candidate for incorporation into larger scale materials synthesis.

Conclusions

By genetically fusing the PTE to the truncated SP of P22, and co-expressing this with coat protein, the system is programmed for the templated assembly and encapsulation of the PTE within the assembled P22 capsid. Thus, neither the scaffold protein directed assembly of P22 capsids, nor the PTE enzymatic activity are disrupted. The capsid acts as a protective shell for the PTE and upon encapsulation, the PTE exhibits greatly enhanced thermal stability, is resistant to proteolytic degradation, and can withstand the process of lyophilization and rehydration with little loss of activity. This approach takes advantage of the natural catalytic activity of the PTE enzyme against harmful nerve agents and insecticides and the stabilizing effects of the P22 encapsulation in the construction and design of a promising new material. The enhancement of the stability of the PTE enzyme now makes it a more promising material for applications as a remediation catalyst.

The authors would like to thank Professor Frank M. Raushel (Texas A and M University) for supplying the phosphotriesterase gene. Ms O'Neil held an AAUW American Fellowship while conducting this research. This research was supported by a grant from the National Science Foundation (BMAT-1104876).

Notes and references

1. K. Renggli, P. Baumann, K. Langowska, O. Onaca, N. Bruns and W. Meier, Adv. Funct. Mater., 2011, 21, 1241–1259.
2. A. Scheffel, N. Poulsen, S. Shian and N. Kröger, Proc. Natl. Acad. Sci. U. S. A., 2011, 108, 3175–3180.
3. Y. Jeong, B. Duncan, M.-H. Park, C. Kim and V. M. Rotello, Chem. Commun., 2011, 47, 12077–12079.
4. J. Lucon, S. Qazi, M. Uchida, G. J. Bedwell, B. LaFrance, P. E. Prevelige and T. Douglas, Nat. Chem., 2012, 4, 781–788.
5. D. P. Patterson, P. E. Prevelige and T. Douglas, ACS Nano, 2012, 6, 5000–5009.
6. B. Worsdorfer, Z. Pianowski and D. Hilvert, J. Am. Chem. Soc., 2011, 134, 909–911.
7. M. Kwak, I. J. Minten, A. Diana-Milena, A. J. Musser, M. Brasch, R. J. M. Nolte, K. Müllen, J. J. L. M. Cornelissen and A. Herrmann, J. Am. Chem. Soc., 2010, 132, 7834–7835.
8. J. D. Fiedler, S. D. Brown, J. L. Lau and M. G. Finn, Angew. Chem., Int. Ed., 2010, 49, 9648–9651.
9. I. J. Minten, V. I. Claessen, K. Blank, A. E. Rowan, R. J. M. Nolte and J. J. L. M. Cornelissen, Chem. Sci., 2011, 2, 358–362.
10. A. O'Neil, P. E. Prevelige, G. Basu and T. Douglas, Biomacromolecules, 2012, 13, 3902–3907.
11. J. E. Glasgow, S. L. Capehart, M. B. Francis and D. Tullman-Ercek, ACS Nano, 2012, 6, 8658–8664.
12. N. Bruns and J. C. Tiller, Nano Lett., 2004, 5, 45–48.
13. S. G. Morillo and M. d. C. S. T. Timenetsky, Rev. Assoc. Med. Bras., 2011, 57, 462–467.
14. D. P. Dumas, S. R. Caldwell, J. R. Wild and F. M. Raushel, J. Biol. Chem., 1989, 264, 19659–19665.
15. D. P. Dumas, H. D. Durst, W. G. Landis, F. M. Raushel and J. R. Wild, Arch. Biochem. Biophys., 1990, 277, 155–159.
16. E. Ghanem and F. M. Raushel, Toxicol. Appl. Pharmacol., 2005, 207, 459–470.
17. C. Roodveldt and D. S. Tawfik, Protein Eng., Des. Sel., 2005, 18, 51–58.
18. S. R. Caldwell and F. M. Raushel, Biotechnol. Bioeng., 1991, 37, 103–109.
19. A. O'Neil, C. Reichhardt, B. Johnson, P. E. Prevelige and T. Douglas, Angew. Chem., Int. Ed., 2011, 50, 7425–7428.
20. D. P. Patterson, B. Schwarz, K. El-Boubbou, J. van der Oost, P. E. Prevelige and T. Douglas, Soft Matter, 2012, 8, 10158–10166.
21. C. Reichhardt, M. Uchida, A. O'Neil, R. Li, P. E. Prevelige and T. Douglas, Chem. Commun., 2011, 47, 6326–6328.

---

Footnote

† Electronic supplementary information (ESI) available: Calculations, SEC chromatograms, and TEM micrographs. See DOI: 10.1039/c3bm60063g

---