Superheated droplets for protein thermal stability analyses of GFP, BSA and Taq-polymerase

Abstract

Here we describe a novel method for the study of protein thermal stability using superheated aqueous samples within virtual reaction chambers. Virtual reaction chambers consist of an aqueous sample droplet encapsulated by an oil droplet on a hydrophobic surface. Such samples can be superheated due to the lack of nucleation sites. The thermal denaturation of proteins is induced through the application of a temperature gradient using a bespoke silicon heating chip. The unfolding of proteins is followed through the addition of a hydrophobic dye that attaches to protein hydrophobic domains that become exposed during denaturation. Using this method, we investigated the thermal stability of green fluorescence protein and Taq-polymerase. A possible screening application of the method was demonstrated by evaluating the effect of ionic concentration on the thermal stability of bovine serum albumin.

---

Introduction

During the early part of the 20th century, it was established that proteins lose their function when heated, although the cause for this was as yet unknown. Around 1930, several groups began predicting the secondary and tertiary structures of proteins, among them Dorothy Wrinch¹ and Linus Pauling.² With these models in place, the first experimental validations were attempted by denaturating proteins through the addition of reagents that selectively break bonds or weaken intra-molecular attractions.³ Today protein structures are mainly determined using X-ray crystallography,⁴ cryo electron-microscopy⁵ or NMR.⁶

While methods such as NMR, X-ray crystallography and cryo electron-microscopy are good for determining the structure of a protein, they provide only limited information on protein interactions with other molecules. Furthermore, the thermodynamics of these interactions are often important when screening for potential drug candidates.⁷ To access this thermodynamic information, a variety of thermodynamic methods (e.g., isothermal titration calorimetry [ITC] or differential scanning calorimetry [DSC]) have been developed.⁸

Because of the highly sensitive nature of fluorescence spectroscopy, fluorescent dye-based approaches have been developed to research proteins. These dyes typically interact non-covalently with the protein or protein degradation products through hydrophobic or electrostatic interactions. Thus, protein aggregation, fibrillation, chemical degradation and conformational changes can be detected optically.⁹ 1-Anilinonaphtalene-8-sulfonate [ANS], 4,4′-dianilino-1,1′-binaphthyl-5,5′-disulfonic acid [bis-ANS]¹⁰ and Nile red¹¹ are examples of dyes used for this purpose. Due to their ability to control temperature while measuring fluorescence, real-time polymerase chain reaction (RT-PCR) thermocyclers, which are commonly available in research laboratories, are often used for these experiments.

While many proteins will denaturate at temperatures covered by these commercial thermocyclers thermostable proteins are of particular interest. As enzyme catalysed reactions are highly selective and thermostable proteins are often resistant to harsh conditions these are of interest for many industrial applications like in the hydrolysis of starch or cellulose for example.¹² Many of the industrially used enzymes are even stable and active at temperatures higher than the optimum growth temperature of the organism they are extracted from. To extend the range of available proteins the mechanisms increasing thermostability have to be understood. Many have been suggested, like improved electrostatic interactions, increased occurrence of hydrophobic residues with branched side chains or higher portions of charged residues.¹³

Recently, our group has introduced a micro-machined, silicon chip for heating PCR samples, mountable to a fluorescence microscope.¹⁴ The sample is placed as a droplet with a volume of 100 nL on a hydrophobic surface, thus forming a virtual reaction chamber (VRC)¹⁵ (Fig. 1). Superheating within VRCs has already been applied to peptide analysis.¹⁶ Here, we demonstrated how this system can be used to investigate the thermal stability of proteins. The use of a micro-fabricated heating chip and VRCs has the advantage of having significantly lower sample consumption (100–300 nL) compared to real time PCR cyclers (5–20 μL) or calorimeters (0.1–1 mL). Furthermore, aqueous samples in VRCs can be superheated to temperatures of up to 200 °C.¹⁷ This is due to the sample being completely encapsulated through the oil phase, preventing the formation of the nucleation sites necessary for boiling.¹⁸ Through this feature it is possible to study proteins stable at 100 °C or more, a temperature range not accessible by real-time PCR cyclers due to sample boiling. Thus, the combination of miniaturization with VRCs helps to reduce sample volumes and allows samples to be heated to temperatures in excess of the boiling temperature of water, without the need of pressurization.

Fig. 1 Image of the heating chip and two VRCs. The chip was designed for the parallel heating of up to four samples in parallel. For experiments only one heating position was used at a time. For visualisation purposes the sample has been replaced with fluorescein solution in this image.

To demonstrate the usefulness of VRCs, initial experiments were made using green fluorescent protein (GFP). Since GFP has an intrinsic fluorescence dependent on its folded state, its denaturation can be easily observed, without the requirement of adding a dye. For comparison, these experiments were carried out on a chip and in a commercial RT-PCR cycler. Second, a screening application was demonstrated by testing the influence of the ionic strength of the buffer on the stability of bovine serum albumin. Lastly, experiments were carried out measuring the thermal denaturation of Taq polymerase. As Taq polymerase is stable at 95 °C, super-heating of the solution is necessary for observing the unfolding of this protein.

Experimental section

Setup

The virtual reaction chambers were prepared by pipetting 300 nL of sample onto a microscope cover slip coated with (1H,1H,2H,2H-perfluorooctyl)trichlorosilane (Sigma-Aldrich).¹⁷ To prevent evaporation samples were covered with 2 μL of a mixture consisting out of 5 g of FC-40 (Sigma-Aldrich) and 5 g FC-70 (Sigma-Aldrich). The viscosity of the oil mixture was increased through the dissolution of 3 wt% Teflon-beads (Sigma-Aldrich), preventing the spreading of the oil phase on the glass slide. Before use the fluorinated oils were filtered using a syringe filter (VWR). Prepared VRCs were placed on a custom made, micro-machined silicon heater developed for a portable PCR device.¹⁴ A temperature ramp of 0.5 °C s⁻¹ starting from 25 °C was applied while fluorescence was continuously monitored. The temperature was measured with a resistive sensor next to the heater on the bespoke silicon chip. The temperature correlation between the measured temperature and sample temperature was determined earlier.¹⁷

Fluorescence amplitude was captured using a fluorescence microscope (Zeiss) equipped with a blue LED (ThorLabs) and FAM-Filterset (Chroma) for experiments with GFP or an amber LED (ThorLabs) with corresponding filters (Chroma) for all other experiments. Detection was done using a photomultiplier tube (Hamamatsu) and oscilloscope (Tektronix). The LEDs for excitation were modulated and a LockIn amplifier (AmTek) was used for detection to increase the signal to noise ratio.

Samples consisted of 5 μL of Protein Thermal Shift buffer with a pH of 6.8 (ThermoFischer) and 5 μL of 4× Protein Thermal Shift dye (ThemoFischer). Either 2 μL of bovine serum albumin solution (200 mg mL⁻¹) (Sigma-Aldrich), 1 μL of native polymerase (5 U μL⁻¹) (ThermoFischer) or 4 μL of extracted GFP were added. To the BSA samples, aqueous sodium chloride solution (Sigma-Aldrich) was added to achieve final concentrations of 0, 0.15, 0.3, 0.5 and 1 M. Finally, volumes were adjusted to 20 μL using Milli-Q water (ProgradT3 column, Millipore). For samples containing GFP the Protein Thermal Shift dye was replaced with water.

When applicable, samples were also measured using a Roche LightCycler (Roche Molecular Diagnostics) with a temperature ramp of 0.5 °C s⁻¹ ranging from 37 to 95 °C with continuous fluorescence measurement.

Isolation of GFP

Green fluorescent protein was harvested from Escherichia coli transfected with GFP by first washing the culture with water, followed by centrifugation at 3250g for 10 min. The resulting cell pellet was resuspended in 1 mL of Laemmli lysis buffer and sonicated on ice 6 times for 30 s. The cell debris were removed from the solution through centrifugation (10 min, 12 100g at 4 °C). The supernatant was transferred into another microcentrifuge tube and diluted with a mixture of acetone (Sigma-Aldrich) and methanol (Sigma-Aldrich) (8 : 1) cooled to −20 °C. After 2 h of incubation at −80 °C, proteins were pelleted by centrifugation (10 min, 1000g at 4 °C). Pellets were washed with acetone (Sigma-Aldrich) at −20 °C three times before resuspension in difference gel electrophoresis [DIGE] label buffer.

Results and discussion

For comparison, experiments were first carried out on both our custom-made chip and a LightCycler. When GFP was heated, protein unfolding results in a loss of its fluorescence (Fig. 2). On both systems, the loss of fluorescence occurs in a single step, as recognizable from the single peak in the derived signals. This confirms reports of a single domain unfolding in the denaturation of GFP.¹⁹ The peak maximums from the light cycler was at 87.0 °C while the one measured by the chip was at 88.8 °C, indicating the comparability of results from the two systems. The graphs also show how a broader temperature range can be accessed by the chip approach than by the LightCycler. While the minimum temperature achievable by the LightCyler is 37 °C, the lower limit of the chip is ambient temperature. Other commercial thermal cyclers, based on Peltier elements, offer the option to cool the sample, decreasing the lower limit of the accessible temperature range. To prevent boiling in the capillaries, the LightCyclers maximum temperature is limited to 95 °C. Whereas, through the use of VRCs, the chip aqueous samples can be superheated to temperatures in excess of 100 °C (up to 125 °C in this study).

Fig. 2 Fluorescence signal as function of temperature as recorded by the LightCycler (blue curve) and out chip (red) (left graph). First derivative of fluorescence signals with respect to temperature for the LightCycler (blue) and our chip calculated from the smoothed signal (red) (right graph). The single step in which GFP denatures can be recognized from the single peak in the derived fluorescence signal. Due to the shorter optical path length of the VRC the signal to noise ratio is worse compared to the LightCycler, as apparent by the artefacts to the left of the peak and above 100 °C.

To demonstrate the usefulness of the chip approach for screening applications, we tested the influence of ionic strength on the thermal stability of bovine serum albumin. In these experiments, a dye was necessary to follow the unfolding process. Fig. 3 shows how the fluorescence increases at different buffer concentrations of sodium chloride, as well as the first derivative with respect to temperature calculated from the smoothed signal. As the concentration of sodium chloride increases, the ionic strength of the buffer also increases, thus delaying the rise in fluorescence to a higher temperature. Hence, the peak maximums of the derived fluorescence increases from 72 °C with no sodium chloride to 87.5 °C at sodium chloride concentrations of 1 M. The strong stabilizing effect of sodium chloride was previously reported by DSC²⁰ and could be confirmed through control experiments on the LightCycler (S2†). Additionally experiments with only the dye and buffer were carried out on the LightCycler to test for background noise originating from the dye (S3†).

Fig. 3 Normalized fluorescence signals as function of temperature for experiments with BSA protein and different concentrations of sodium chloride (left). First derivative of fluorescence with respect to temperature for experiments with BSA and various concentrations of sodium chloride (right). The graphs show how a higher ionic strength of the solution leads to a thermal stabilization of the protein.

The ability of VCRs to superheat aqueous samples was used for the analysis of Taq polymerase. As it is used for polymerase chain reaction, and thus repeatedly heated to 95 °C, the stability of this protein is well-known. Therefore, super-heating is necessary to thermally denature the protein. Through calorimetric measurements the denaturation of Taq polymerase in two distinctive steps was reported with the first domain unfolding at 88.9 °C and the second at 99.1 °C.²¹ As the first domain denatures at temperatures exceeded during polymerase chain reaction, this unfolding process is thought to be reversible.²¹ In the experiments conducted with our chip, an unfolding process in two steps was observed (Fig. 4). The first derivative of the fluorescence signal with respect to temperature shows two peaks, one at 81 °C and the second one at 104 °C. The presence of the two peaks confirms the observations made by differential scanning calorimetry of a two-step denaturation process.²¹ The denaturation temperatures found by differential scanning calorimetry could not be reproduced. However, the reference measurements were conducted at a pH of 9.5 while our measurements were conducted at a pH of 6.8, furthermore there might be a difference in buffer salt concentration. This could account for the observed difference, especially as stabilization effects of up to 12 °C have been reported by Arakawa et al.²² Furthermore, it is possible that the proteins were extracted from different strains of Thermus aquaticus. However, we confirmed that the second, irreversible, denaturation step occurs at a temperature higher than the boiling temperature of water and thus is not reached during PCR.

Fig. 4 Diagram showing fluorescence amplitude as a function of temperature for experiments conducted with native Taq polymerase (left). First derivative of the fluorescence signal with respect to temperature for experiments with Taq polymerase calculated from the smoothed signal (right). In the derivative two peaks can be seen corresponding to the denaturation of two separate domains of Taq polymerase.

Interestingly the fluorescence amplitude is decreasing with increasing temperature for experiments with Taq polymerase (a decrease in fluorescence could also be observed on control experiments on the Light Cycler (ESI S3†)). We suspect that the origin for this effect is linked to the thermal stability of the protein. Thermostability is a product of various factors such as shorter alpha-helixes, a more densely packed hydrophobic core and a higher amount of charged residuals compared to non-thermostable proteins.^23,24 Furthermore thermostable proteins form networks of salt bridges stabilizing the protein at high temperatures.²⁵ These factors could lead to the hydrophobic dye binding having more binding sites on the folded protein compared to the unfolded state, leading to a decrease in fluorescence during unfolding.

Conclusion

Here we have shown how VRCs to analyse the thermal stability of proteins. It was demonstrated that thermostable proteins can be easily analysed using the ability of VRCs to superheat water to temperatures >100 °C. While other methods (e.g., scanning calorimetry) require extensive pressurized set-ups to measure above 100 °C, the experimental set-up used here is relatively simple and measurements could be performed in as little as 3 to 5 min, depending on the temperature range investigated. Furthermore, sample consumption is much lower using VRCs with sample volumes (in the range of 100 to 300 nL) than for other fluorescence methods performed using commercial machines.

Through the use of the hydrophobic dye the extraction of thermodynamic data is hardly possible. Furthermore, fluorescence measurements are more prone to noise due to temperature effects on fluorescence yield and unspecific binding of the dye, especially at low temperatures.

However, this novel chip system is suitable for screening tasks, for example when searching for buffer conditions stabilising or destabilising for a protein or when screening for the effects of mutations on stability. Because of the low sample consumption, fast analysis and simple set-up, potential candidates can be screened first using the chip method, with promising candidates being subsequently analysed by DSC.

Acknowledgements

The authors would like to thank Ahyeon Gyeon and Mathias Altmeyer for help with bacteria culture and protein extraction. This work was financed by Korea Institute of Science and Technology – Europe basic research program.

References

1. D. M. Wrinch, Nature, 1936, 137, 411–412.
2. L. Pauling, R. B. Corey and H. R. Branson, Proc. Natl. Acad. Sci. U. S. A., 1951, 37, 205–211.
3. W. Kauzmann, in Advances in Protein Chemistry, ed. M. L. Anson, K. B. C. B. Anfinsen and T. E. John, Academic Press, 1959, vol. 14, pp. 1–63.
4. A. T. Gres, K. A. Kirby, V. N. KewalRamani, J. J. Tanner, O. Pornillos and S. G. Sarafianos, Science, 2015, 349, 99–103.
5. R. Y.-R. Wang, M. Kudryashev, X. Li, E. H. Egelman, M. Basler, Y. Cheng, D. Baker and F. DiMaio, Nat. Methods, 2015, 12, 335–338.
6. L. Brown, C. Cox, J. Baptiste, H. Summers, R. Button, K. Bahlow, V. Spurrier, J. Kyser, B. Luttge, L. Kuo, E. Freed and M. Summers, Viruses, 2015, 7, 2210.
7. N. C. Garbett and J. B. Chaires, Expert Opin. Drug Discovery, 2012, 7, 299–314.
8. N. Garbett, in Metallointercalators, ed. J. Aldrich-Wright, Springer, Vienna, 2011, ch. 12, pp. 299–324.
9. A. Hawe, M. Sutter and W. Jiskoot, Pharm. Res., 2008, 25, 1487–1499.
10. A. Bothra, A. Bhattacharyya, C. Mukhopadhyay, K. Bhattacharyya and S. Roy, J. Biomol. Struct. Dyn., 1998, 15, 959–966.
11. D. L. Sackett and J. Wolff, Anal. Biochem., 1987, 167, 228–234.
12. G. D. Haki and S. K. Rakshit, Bioresour. Technol., 2003, 89, 17–34.
13. S. Kumar and R. Nussinov, Cell. Mol. Life Sci., 2001, 58, 1216–1233.
14. C. D. Ahrberg, A. Manz and P. Neuzil, Sci. Rep., 2015, 5, 11479.
15. P. Neuzil, J. Pipper and T. M. Hsieh, Mol. BioSyst., 2006, 2, 292–298.
16. M. O. Altmeyer, A. Manz and P. Neužil, Anal. Chem., 2015, 87, 5997–6003.
17. T. K. Pavel Neuzil, W. Sun and A. Manz, Appl. Phys. Lett., 2015, 106, 024104.
18. D. Zahn, Phys. Rev. Lett., 2004, 93, 227801.
19. D. M. Sagar, S. Aoudjane, M. Gaudet, G. Aeppli and P. A. Dalby, Sci. Rep., 2013, 3, 2130.
20. C. Giancola, C. De Sena, D. Fessas, G. Graziano and G. Barone, Int. J. Biol. Macromol., 1997, 20, 193–204.
21. I. Karantzeni, C. Ruiz, C.-C. Liu and V. J. Licata, Biochem. J., 2003, 374, 785–792.
22. T. Arakawa and S. N. Timasheff, Biophys. J., 1985, 47, 411–414.
23. A. Goldman, Structure, 1995, 3, 1277–1279.
24. S. Korolev, M. Nayal, W. M. Barnes, E. Di Cera and G. Waksman, Proc. Natl. Acad. Sci. U. S. A., 1995, 92, 9264–9268.
25. S. Kumar, B. Ma, C.-J. Tsai and R. Nussinov, Proteins: Struct., Funct., Bioinf., 2000, 38, 368–383.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra07079e

‡ Current address: Department of Mechanical Engineering, Sogang University, Seoul 121-742, Korea.

---