- Email: [email protected]
galactose-binding protein
Biosensors and Bioelectronics 24 (2009) 3229–3234
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Fluorescence lifetime spectroscopy and imaging of nano-engineered glucose sensor microcapsules based on glucose/galactose-binding protein Tania Saxl a,∗ , Faaizah Khan a , Daniel R. Matthews b , Zheng-Liang Zhi a , Olaf Rolinski c , Simon Ameer-Beg b , John Pickup a a
Metabolic Unit, King’s College London School of Medicine, Guy’s Hospital, London SE1 9RT, UK Randall Division of Cell and Molecular Biophysics, Richard Dimbleby Department of Cancer Research, King’s College London School of Medicine, Guy’s Hospital, London SE1 1UL, UK c Dept of Physics, University of Strathclyde, Glasgow G4 0NG, UK b
a r t i c l e
i n f o
Article history: Received 8 January 2009 Received in revised form 26 March 2009 Accepted 6 April 2009 Available online 14 April 2009 Keywords: Fluorescence Lifetime FLIM Glucose sensor Diabetes Microcapsule Nanolayers Badan
a b s t r a c t We aimed to develop microsensors for eventual glucose monitoring in diabetes, based on fluorescence lifetime changes in glucose/galactose-binding protein (GBP) labelled with the environmentally sensitive fluorophore dye, badan. A mutant of GBP was labelled with badan near the binding site, the protein adsorbed to microparticles of CaCO3 as templates and encapsulated in alternating nano-layers of polyl-lysine and heparin. We used fluorescence lifetime imaging (FLIM) with two-photon excitation and time-correlated single-photon counting to visualize the lifetime changes in the capsules. Addition of glucose increased the mean lifetime of GBP-badan by a maximum of ∼2 ns. Analysis of fluorescence decay curves was consistent with two GBP states, a short-lifetime component (∼0.8 ns), likely representing the open form of the protein with no bound glucose, and a long-lifetime component (∼3.1 ns) representing the closed form with bound glucose and where the lobes of GBP have closed round the dye creating a more hydrophobic environment. FLIM demonstrated that increasing glucose increased the fractional proportion of the long-lifetime component. We conclude that fluorescence lifetime-based glucose sensing using GBP encapsulated with nano-engineered layer-by-layer films is a glucose monitoring technology suitable for development in diabetes management. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Blood glucose monitoring is an essential component of modern diabetes care, particularly in type 1 diabetes and insulin-requiring type 2 diabetes. It allows detection of hyper- and hypoglycaemia, and adjustment of insulin, diet and exercise to achieve and maintain near-normal blood glucose concentrations, thereby slowing or preventing the development of serious microvascular complications (Bergenstal and Gavin, 2005; Pickup et al., 2005b). The ideal glucose monitoring technology for diabetes must measure glucose continuously whilst also being minimally invasive, or non-invasive. Present commercially available glucose sensors for continuous monitoring in vivo are needle-type probes implanted in the subcutaneous tissue and are based on either amperometric enzyme electrodes (Heller and Feldman, 2008; Mastrototaro, 2000; Wilson and Gifford, 2005) or microdialysis (Maran et al., 2002). They can be used only in the short term (without re-implantation), and suffer from unpredictable drift and less than optimal accuracy,
∗ Corresponding author. Tel.: +44 207 188 1909; fax: +44 207 188 0146. E-mail address: [email protected] (T. Saxl). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.04.003
thereby requiring frequent re-calibration. A promising alternative to electrochemistry that is being researched by several groups is fluorescence-based glucose sensing; fluorescence has the advantage of being highly sensitive and free from interference from electro-active components in the tissues (Pickup et al., 2005a). In previous work, we described a glucose-sensing strategy based on engineered bacterial glucose/galactose-binding protein (GBP) labelled with the environmentally sensitive fluorophore, badan (6-bromo-acetyl-2-dimethylaminonaphthalene) at a cysteine mutation at position 152 near the binding site (Khan et al., 2008). Badan displays an increased fluorescence intensity in a hydrophobic environment, such as in the interior of a protein. GBP has been used for glucose sensing in work by several others (e.g. Ballerstadt and Schultz, 2000; Deuschle et al., 2006; Fehr et al., 2003; Ge et al., 2004; Marvin and Hellinga, 1998; Salins et al., 2001; Scognamiglio et al., 2004; Tolosa and Rao, 2006) but not using badan as the reporter of glucose binding. The single polypeptide chain of GBP folds into two domains, connected by a hinge; GBP specifically binds glucose and galactose, and binding of glucose at a site between the domains induces a significant conformational change as the domains close around the ligand (Borrock et al., 2007; Dwyer and Hellinga, 2004; Marvin and Hellinga, 1998). The decrease in
3230
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
the polarity in the micro-environment of the badan as the lobes close around the badan at the glucose binding site produces a maximal 300% increase in fluorescence intensity on addition of glucose (Khan et al., 2008). This compares with a less than 20% change in fluorescence intensity in the more researched fluorescence energy transfer (FRET)-based systems where GBP is labelled with donor and acceptor fluorophores at different sites in the protein molecule (Khan et al., 2008; Ye and Schultz, 2003). Time-resolved fluorescence measurements are known to be relatively independent of light scattering, fluorophore concentration and photobleaching and have therefore been used as an alternative to intensity measurements in molecular glucose sensing (Lakowicz, 2006). In this paper, we report studies on the fluorescence of GBPbadan which will underpin the later use of this protein as a glucose monitoring device in diabetes. Firstly, we studied the fluorescence lifetime of badan-labelled GBP and its variation with glucose. We also explore two further contributions to fluorescence-based glucose sensing in the in vivo environment. Firstly, we test whether GBP-badan maintains its function when encapsulated in nanoengineered microvesicles, using our recently reported technique for layer-by-layer electrostatic deposition of alternately charged polypeptide onto a protein-loaded calcium carbonate template (Zhi and Haynie, 2006a,b). Such self-assembled, layer-by-layer glucose sensors have several potentially useful properties for in vivo application, including tunable permeability to the measured analyte (glucose) and the possibility of enhanced biocompatibility. They may thus be optimised for impregnation in the dermis as a ‘smart tattoo’ for minimally invasive glucose sensing (Pickup et al., 2008). Secondly, we make use of a custom-built fluorescence lifetime imaging (FLIM) system with two-photon excitation to visualise lifetime changes in the GBP-badan glucose sensor particles; in this technique, differences in lifetime are used to generate contrast in the image (Ameer-Beg et al., 2005), particularly suitable for determination of localised fluorescence lifetime changes in one or several microcapsules. 2. Material and methods 2.1. Materials Most chemicals used for engineering the GBP mutant were of molecular biology grade and purchased from Sigma–Aldrich (St. Louis, USA). The pTZ18U-mglB vector containing the GBP gene was a kind gift from Dr. S. D’Auria. The plasmid pET303/CT-His vector was purchased from Invitrogen (Paisley, UK). E. coli DH5␣ cells were used as host cells for plasmid proliferation. LB media supplemented by antibiotics (50 g/ml of kanamycin or 100 g/ml of ampicillin) were employed to grow cells. E. coli BL21(DE3) was from BD Biosciences (Franklin Lakes, NJ, USA). All restriction enzymes were purchased from New England Biolabs (Hitchen, UK). Quick-change site-directed mutagenesis kit was purchased from Stratagene (La Jolla, CA, USA). The Rapid DNA ligation kit and long-template polymerase chain reaction (PCR) enzyme (Expand) were from Roche Applied Science (Basel, Switzerland). The kit used for plasmid extraction and Ni-NTA agarose was from Qiagen (West Sussex, UK) and the kit used to purify PCR products or restriction reactions was from Qbiogene (Morgan Irvine, CA, USA). The fluorescent probe badan was from Invitrogen and 5 m-diameter spherical CaCO3 particles were from PlasmaChem GmbH (Germany). 2.2. Construction of expression vector pET303-GBP and purification of H152C mutant of GBP Details of the methodology are given in Khan et al. (2008). In brief, the GBP gene (mglB) was isolated from the plasmid pTZ18U-
mglB by PCR and ligated into pET303/CT-His vector using a Rapid DNA ligation kit to form pET303-GBP. For the H152C mutant, pET303-GBP was used as a template. Site-directed mutagenesis was performed using the Quick-change mutagenesis kit. DNA sequencing data verified the presence of the desired point mutation. A single colony of E. coli BL21(DE3) transformed with the pET303-GBP plasmid was inoculated in LB media containing 100 g/ml of ampicillin and grown at 37 ◦ C. Expression of the protein was induced by adding isopropyl-2-d-thiogalactopyranoside to a final concentration of 1 mM. Bacterial cells were lysed and the cell extract was clarified by centrifugation. Affinity chromatography was performed in a glass column packed with 5 ml Ni-NTA agarose. The protein was eluted with buffer containing imidazole. The purity of GBP was determined by SDS-PAGE using 10% acrylamide gels that were viewed by Coomassie Blue staining. 2.3. Fluorophore labelling To label GBP with badan, 50 M protein was dissolved in 5 mM Tris(2-carboxyethyl)phospine in phosphate-buffered saline (PBS) pH 7.4, and then a 10-fold excess of dye (500 M) was added and the mixture incubated overnight at room temperature, after which it was purified on a Sepadex G-25 gel filtration column. 2.4. Nano-engineered microcapsules containing GBP-badan The encapsulation method of Zhi and Haynie (Zhi and Haynie, 2006a,b) was modified as follows. All buffers were 50% (v/v) 10 mM Tris (pH 7.0)/polyethylene glycol 300, unless otherwise stated. 50 l of 20 M GBP-badan in 10 mM PBS pH 7.4 was mixed with 5 mg of 5 m-diameter spherical CaCO3 particles. CaCO3 acts as a template to adsorb the protein. The mixture was left on a roller at 4 ◦ C for 30 min to allow the GBP-badan to adsorb onto the template. Particles were then centrifuged at 1000 × g for 2 min and washed with buffer. 50 l of 1 mg/ml poly-l-lysine was added and particles were left to stand for 5–10 min on ice. The particles were then centrifuged and washed in 100 l buffer. This process was repeated with the GBP-badan solution and followed by a further layer of poly-l-lysine. The final protein incorporated into the particle was quantified by absorption spectroscopy at 40% of the original 20 M GBP-badan solution. Six further bilayers of 50 l of 1 mg/ml heparin in buffer and poly-l-lysine were added sequentially as above. The resulting particles were centrifuged and re-suspended in 50 l PBS. 2.5. Steady-state fluorescence measurements Steady-state fluorescence measurements were recorded on a PerkinElmer LS50B fluorimeter (PerkinElmer Instruments, Beaconsfield, UK). The excitation and emission wavelength of badan was at 400 and 550 nm respectively. All data were obtained at room temperature using quartz cuvettes with sample volume of 100 L. The labelled protein was incubated with increasing amounts of dglucose for 15–20 min at room temperature before fluorescence was recorded. 2.6. Fluorescence lifetime measurement and imaging Excitation for the solution measurements was provided by a femtosecond titanium-sapphire laser (Coherent, Santa Clara, CA, USA), operating at 800 nm (120 fs) and up-converted by second harmonic generation to 400 nm and directed to a cuvette holder. Fluorescence emission from the cuvette was collected using a lens, bandpass filtered (550 ± 10 nm) and detected using a fast photomultiplier tube (PMT) (PMH-100, Becker and Hickl GmbH, Berlin, Germany) and time-correlated single-photon counting (TCSPC) electronics (SPC830, Becker and Hickl GmbH, Berlin, Germany).
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
3231
Fluorescence lifetime imaging (FLIM) was performed using a custom-built multi-photon system constructed around an upright 90i fluorescence microscope (Nikon, Tokyo, Japan) and similar to that described elsewhere (Parsons et al., 2005; Peter et al., 2005). In brief, excitation was provided by a titanium-sapphire laser (Coherent, Santa Clara, CA, USA), operating at 800 nm. Time-resolved detection was provided by a non-descanned detection channel with a fast PMT (PMH-100, Becker and Hickl GmbH, Berlin, Germany) and a TCSPC personal computer plug-in board (SPC830, Becker and Hickl GmbH, Berlin, Germany). The instrument response was measured from the hyper-Rayleigh scattering of highly attenuated excitation in a suspension of 20 nm colloidal gold (Sigma–Aldrich Company Ltd., G-1652) and used in the analysis and fitting of data by iterative re-convolution. Data were collected at 550 ± 10 nm. The laser power was adjusted to give average photon counting rates of the order 104 –105 photons s−1 (0.0001–0.001 photons per excitation event) below the maximum counting rate afforded by the TCSPC electronics to avoid pulse pile-up. All measurements were made in 10 mM PBS pH 7.4. Cuvette measurements of GBP-badan in glucose were performed by sequentially adding aliquots of 1 mM glucose solution to 100 l of 12 M GBPbadan. The solution was mixed and left to equilibrate for 10 min between each measurement. For multi-photon FLIM, glucose was added sequentially as in the cuvette experiments to 100 l capsules in buffer (made using 10 l of capsule stock). For FLIM, 0.5 l of capsule suspension was placed into a cavity on a microscope slide and the cavity sealed with a cover glass. 2.7. Data analysis Analysis of the fluorescence transients was performed using the TRI2 analysis package (courtesy of Dr. Paul Barber, Gray Cancer Institute of Radiation Oncology and Biology, Oxford University, Oxford, UK). The analysis can be performed on a pixel-by-pixel, image-by-image basis or by considering the whole data set globally (Barber et al., 2005, 2009; Verveer et al., 2000). Global analysis assumes one or more lifetimes across all the images, whilst enabling the pre-exponential factors to be determined for each decay. The fractional contribution of the lifetime was calculated from the preexponential factor, see Eqs. (1) and (2). f (t) = Z + Fi =
˛i exp
−t i
˛
i
˛t
(1) (2)
where ˛i are the pre-exponential factors, Fi the fractional contribution of lifetime i to the transient and Z is a constant corresponding to the background noise. FLIM images were generated from the fractional contribution, F, for the long-lifetime state at each pixel. 3. Results and discussion 3.1. Fluorescence intensity and lifetime of free badan in solvents of different polarity We used the environmentally (polarity) sensitive fluorophore badan for detecting glucose-induced changes in the conformation of GBP. The fluorescence emission maximum for free badan was 500 nm in the dipolar, aprotic solvent dimethyl formamide (DMF), shifting to 550 nm in 2% DMF, 98% water (Fig. 1a), but with a marked reduction in fluorescence intensity in the aqueous environment. Fluorescence lifetime decays for badan in DMF and water were best fitted by a bi-exponential model (Fig. 1b), confirmed by the residuals for mono-exponential and bi-exponential fits (Fig. 1c).
Fig. 1. (a) Fluorescence emission spectra for free badan in 100% DMF (solid line) and 98% water, 2% DMF (dashed line), plotted with measured intensity and normalised intensity. (b) Lifetime fluorescence decays for badan in 100% DMF (at 500 ± 10 nm) and 98% water, 2% DMF. The bi-exponential fit to each decay is shown with corresponding values for the two lifetimes of 2.1 and 0.5 ns for 100% DMF, and 0.8 and 0.2 ns for 98% aqueous solution. The instrument response function is shown for comparison with the fluorescence decay. (c) Residuals for bi- and mono-exponential models.
Lifetimes generated were estimated as 2.4 and 1.4 ns in DMF, and 0.8 ns and <0.2 ns in water, indicating the presence of two emissive states of the free dye in these solvents, the polarity of the solvent affecting the fluorescence intensity. The two lifetime components of badan in water and DMF, is in agreement with the dual fluorescence described recently (Koehorst et al., 2008). Exposed to polar protic water, badan has reduced fluorescent lifetimes and intensity compared to the environment of dipolar aprotic DMF, indicating that a change to a more hydophobic environment would be expected to increase the lifetime, e.g. as the badan label near the binding site of GBP is enclosed by the polypeptide lobes as glucose binds to the protein (Borrock et al., 2007). We have previously found that there is no corresponding shift in emission wavelength when free GBP-badan is compared to GBP-badan in the presence of glucose, both emission maxima being at 550 nm (Khan et al., 2008). 3.2. Glucose-dependent fluorescence changes of GBP-badan On binding to GBP, the photophysical properties of the sensor dye were modified due to close proximity of the protein but as Fig. 2a illustrates, the environmental sensing property was retained.
3232
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
Table 1 Fluorescence lifetimes () and corresponding fractional contributions (F) for GBP-badan, obtained by fitting data using mono-, bi- and tri-exponential models. Best fits are highlighted in bold. [Glucose] (M)
1 (ns)
Fits to individual fluorescence decays 0 0.9 1.3 1.5 310
2.9 3.4 3.2
(F1)
2 (ns)
(F2)
– (0.41) (0.25)
0.5 0.7
(0.59) (0.57)
– (0.83) (0.75)
1.7 2.9
(0.17) (0.21)
Global fit to fluorescence decays from nine glucose concentrations (0 and 310 M shown) 0 2.5 – 310 2.5 – 0 3.1 (0.10) 0.8 (0.90) 310 3.1 (0.95) 0.8 (0.05) 0 3.2 (0.03) 1.0 (0.58) 310 3.2 (0.71) 1.0 (0.23)
The fluorescence lifetime decay of GBP-badan in PBS (zero glucose), and in a saturating glucose concentration of 310 M are shown for comparison. Both individual analysis of each decay and global analysis of nine decays from nine increasing glucose concentrations were performed. Fitting the fluorescence decay at zero glucose with a bi-exponential model produced decay constants of 1.3 and 0.5 ns, with a mean lifetime of 0.8 ns. The two states were similar to those found in water and consistent with an open conformation of GBP
3 (ns)
0.1
1.2
0.3 0.3
(F3)
2
(0.17)
9.4 1.4 1.2
mono bi tri
(0.04)
1.1 1.0 1.1
mono bi tri
6.9
mono
1.1
bi
4.2
tri
(0.39) (0.06)
with badan exposed to the polar environment. With a saturating glucose concentration, where a larger proportion of the closed form of GBP-badan is expected, a mono-exponential fit produced a lifetime of 2.9 ns. This corresponds to an increase in mean lifetime of 2.1 ns with the addition of saturating glucose. Short-lifetime states are generated in the presence of hydrogen bonding, as would be found in water (Józefowicz et al., 2005). On addition of glucose, the increase in fluorescence lifetime that we observed can be explained by considering the interior of the protein to be a hydrophobic environment, the lobes of GBP closing around the glucose, effectively shielding the fluorophore from the surrounding medium. The apparent change from two short-lifetime states at zero glucose, to an increasing fraction of a single long-lifetime state as glucose is added, is possibly due to badan being bound in different orientations (rotomers) at zero glucose and restriction in the geometry occurring as the conformation of the protein changes, excluding one emissive sates and favouring another. A similar trend was seen in by Gilardi et al. in maltose binding protein labelled with nitrobenzoxadiazole (Gilardi et al., 1997). On global analysis of the data set obtained from nine glucose concentrations, we obtained lifetimes for a two-component model of 3.1 and 0.8 ns, the relative fraction of each varying with glucose concentration. The two short-lifetime states, seen in the absence of glucose, are approximated to one short-lifetime state (0.8 ns), the contribution of which decreases with addition of glucose. The two lifetimes obtained by global analysis are in agreement with those recovered from the individual fits, as is the increase in mean lifetime of approximately 2 ns on addition of glucose (summarised in Table 1). The short-lifetime state of 0.8 ns is consistent with an open conformation of GBP-badan and the long-lived, 3.1 ns state as the closed conformation of GBP-badan, generated by binding of glucose. Fig. 2b shows the fractional contribution of the two global lifetimes as the glucose concentration was increased: the proportion of the open form decreased and the proportion of the closed form increased. 3.3. Glucose-dependent changes of GBP-badan encapsulated in microcapsules
Fig. 2. (a) Cuvette-based measurements of fluorescence emission for GBP-badan in PBS and 310 M glucose (saturated glucose). The mono-exponential fit to the saturating glucose and bi-exponential fit to the zero glucose decays are indicated, with corresponding lifetime values of 2.9, 1.3 and 0.5 ns respectively. (b) Global analysis of the data from 9 glucose concentrations, including those shown in a. The fractional contribution of lifetime states 3.1 and 0.8 ns is plotted against glucose concentration. The two lifetimes were determined by globally fitting a bi-exponential model.
GBP-badan was adsorbed onto spherical calcium carbonate particles, 5 M in diameter, and then encapsulated in six bi-layers of alternating poly-l-lysine and heparin (layer-by layer technique) stabilised with PEG. Fig. 3 shows images of the spherical capsules obtained by two-photon FLIM, taken for 0, 10, 25, 50 and 100 M glucose. Fig. 3a shows intensity profiles for each glucose concentration. Global lifetimes were determined as 2.3 and 0.4 ns by fitting
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
3233
Fig. 3. Fluorescence lifetime microscopy images generated from the time-correlated fluorescence data at varying glucose concentrations and their analysis: (a) diagram illustrating a micro-capsule (not to scale); (b) fluorescence intensity images of the GBP-badan microcapsules; (c) the fractional contribution (F) of the long-lifetime state (2.3 ns), generated from global analysis and displayed as a change in colour of the image from blue (low fraction of closed [glucose-bound] form) to red (high fraction of closed [glucose-bound] form). (d) Pixel histograms for each image. The maxima have been normalised to display the histograms on the same scale. (e) Graph showing the mean value of F against glucose concentration obtained from the images in (a). Pixels were binned for each capsule and the error is the standard deviation between capsules in each image.
the entire set of images to a two-component model, as in the free protein case. In the set of FLIM images, the increasing fractional contribution of the long-lifetime state, corresponding to the increasing proportion of the bound GBP-badan glucose, is displayed as a colour change from blue to red on addition of glucose (Fig. 3b). An increase
in the long-lifetime state on addition of glucose can be seen in the pixel histograms (Fig. 3c). Fig. 3d shows the fractional contribution of the long-lifetime 2.3 ns state (GBP bound to glucose) for each glucose concentration, obtained from global analysis of the images. Lifetime data were summed over all pixels for each image.
3234
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
Layer-by-layer techniques which involve encapsulation and particle covering (Donath et al., 1998) have been reviewed recently (Ariga et al., 2007) and the present studies where the sensor is encapsulated, extends our previous work, in which glucose oxidase was encapsulated in nano-engineered polypeptide layers (Zhi and Haynie, 2006a,b). Trau and Renneberg have also reported layer-bylayer encapsulation of a sensing protein (glucose oxidase), though using micro-crystals of the enzyme rather than adsorption to a template of CaCO3 , alternating layers of polyelectrolytes rather than polypeptide and amperometric detection technology rather than fluorescence (Trau and Renneberg, 2003). Also, Chinnayelka and McShane have described FRET-based glucose sensors using concanavalin A (Chinnayelka and McShane, 2004) or apo-glucose oxidase (Chinnayelka and McShane, 2005) as the glucose receptor and layer-by-layer encapsulation in polyelectrolytes. Microcapsules similar to ones used in our study may be suitable for a ‘smart tattoo’ type sensor, where particles are impregnated in the dermis or subcutaneous tissue and glucose measured from outside the body (Pickup et al., 2008). Our work has limitations. The operating range is relatively restricted and an order of magnitude below the typical pathophysiological range of blood glucose concentrations in diabetes (<30 mM), even when the GBP-badan was encapsulated. Several approaches to extending the range of sensing into the mM range exist (which will be needed for clinical use) and will be pursued in future work. For example, a recently reported Phe16Ala mutation increased the Kd of GBP to 3.9 mM (Sakaguchi-Mikami et al., 2008). In this study, we used FLIM to visualize and quantify microcapsule fluorescence. FLIM, especially in combination with multi-photon excitation, is a powerful biological imaging tool with emerging applications (Festy et al., 2007). A special advantage is that it allows spacially resolved analyte sensing, i.e. highly localised measurements within fabricated structures (such as the glucose microsensors here) and within cells and tissues. This is likely to prove valuable in diabetes research for mapping glucose and other analyte changes in various tissue compartments, in interstitial fluid and in different cells and cell organelles. 4. Conclusion In this paper, we demonstrate for the first time proof-of-concept for glucose sensing based on fluorescence lifetime changes of GBP labelled with the environmentally sensitive dye badan. We also show that GBP-badan can be encapsulated in microcapsules constructed by a layer-by-layer technology, and that the glucose sensing capacity is preserved in the capsules. As glucose concentration was increased, we observed an increasing mean lifetime of GBP-badan, due to binding of glucose. This offers a robust methodology for measurement of glucose concentration using a GBP-badan sensor. In future research it will be necessary to construct sensors based on this approach which are suitable for testing vivo and to check requirements such as freedom from interference and quenching and long-term stability. Acknowledgements JCP is grateful to EPSRC (EP/D062861/1), the Diabetes Foundation and the Wellcome Trust for generous grant support.
SAB and DRM gratefully acknowledge the support of the EPSRC (EP/C546105/1), Cancer Research UK (C1519/A6906), the Royal Society and Dimbleby Cancer Care. OR thanks the EPSRC for grant support (EP/D062861/1). Many thanks to Dr. P. Barber (Oxford University) for providing the TRI2 FLIM imaging software for global image analysis, Dr. S. D’Auria for providing the vector contining the GBP gene and Prof. DJS Birch (Strathclyde University) for valuable discussion and assistance. References Ameer-Beg, S.M., Peter, M., Keppler, M.D., Prag, S., Barber, P.R., Ng, T., Vojnovic, B., 2005. Proc. SPIE 5700, 152–161. Ariga, K., Hill, J.P., Ji, Q., 2007. Phys. Chem. Chem. Phys. 9, 2319–2340. Ballerstadt, R., Schultz, J.S., 2000. Meth. Biotechnol. 7, 89–98. Barber, P.R., Ameer-Beg, S.M., Gilbey, J.D., Edens, R.J., Ezike, I., Vojnovic, B., 2005. Proc. SPIE 5700, 171–181. Barber, P.R., Ameer-Beg, S.M., Gilbey, J., Carlin, L.M., Keppler, M., Ng, T.C., Vojnovic, B., 2009. Multiphoton time-domain FLIM: practical application to protein-protein interactions using global analysis. J. R. Soc. Interface 6, S93–S105. Bergenstal, R.M., Gavin, J.R., 2005. Am. J. Med. 118 (Suppl. 9A), 1S–6S. Borrock, M.J., Kiessling, L.L., Forest, K.T., 2007. Prot. Sci. 16, 1032–1041. Chinnayelka, S., McShane, M.J., 2004. Biomacromolecules 5, 1657–1661. Chinnayelka, S., McShane, M.J., 2005. Anal. Chem. 77, 5501–5511. Deuschle, K., Chaudhuri, B., Okumoto, S., Lager, I., Lalonde, S., Frommer, W.B., 2006. Plant Cell 18, 2314–2325. Donath, E., Sukhorukov, G.B., Caruso, F., Davis, S.A., Möhwald, H., 1998. Angew. Chem. Int. Ed. Engl. 37, 2201–2205. Dwyer, M.A., Hellinga, H.W., 2004. Curr. Opin. Struct. Biol. 14, 495–504. Festy, F., Ameer-Beg, S.M., Ng, T., Suhling, K., 2007. Mol. Biosyst. 3, 381–391. Ge, X., Tolosa, L., Rao, G., 2004. Anal. Chem. 76, 1403–1410. Gilardi, G., Mei, G., Rosato, N., Agro, A.F., Cass, A.E., 1997. Protein Eng. 10, 479–486. Fehr, M., Lalonde, S., Lager, I., Wolff, M.W., Frommer, W.B., 2003. J. Biol. Chem. 278, 19127–19133. Heller, A., Feldman, B., 2008. Chem. Rev. 108, 2482–2505. Józefowicz, M., Kozyra, K.A., Heldt, J.R., Heldt, J., 2005. Chem. Phys. 320, 45–53. Khan, F., Gnudi, L., Pickup, J.C., 2008. Biochem. Biophys. Res. Commun. 365, 102–106. Koehorst, R.B., Spruijt, R.B., Hemminga, M.A., 2008. Biophys. J. 94, 3945–3955. Lakowicz, J.R., 2006. Principles of Fluorescence Spectroscopy, 3rd edn. SpringerVerlag, New York, p. 954. Maran, A., Crepaldi, C., Tiengo, A., Grassi, G., Vitali, E., Pagano, G., Bistoni, S., Calabrese, G., Santeusanio, F., Leonetti, F., Ribaudo, M., Di Mario, U., Annuzzi, G., Genovese, S., Riccardi, G., Previti, M., Cucinotta, D., Giorgino, F., Bellomo, A., Giorgino, R., Poscia, A., Varalli, M., 2002. Diabetes Care 25, 347–352. Marvin, J.S., Hellinga, H.W., 1998. J. Am. Chem. Soc. 120, 7–11. Mastrototaro, J.J., 2000. Diabetes Technol. Ther. (Suppl. 1), S13–S18. Parsons, M., Monypenny, J., Ameer-Beg, S.M., Millard, T.H., Machesky, L.M., Peter, M., Keppler, M.D., Schiavo, G., Watson, R., Chernoff, J., Zicha, D., Vojnovic, B., Ng, T., 2005. Mol. Cell. Biol. 25, 1680–1695. Peter, M., Ameer-Beg, S.M., Hughes, M.K., Keppler, M.D., Prag, S., Marsh, M., Vojnovic, B., Ng, T., 2005. Biophys. J. 88, 1224–1237. Pickup, J.C., Hussain, F., Evans, N.D., Rolinski, O.J., Birch, D.J., 2005a. Biosens. Bioelectron. 20, 2555–2565. Pickup, J.C., Hussain, F., Evans, N.D., Sachedina, N., 2005b. Biosens. Bioelectron. 20, 1897–1902. Pickup, J.C., Zhi, Z.L., Khan, F., Saxl, T., Birch, D.J., 2008. Diabetes Metab. Res. Rev. 24, 604–610. Sakaguchi-Mikami, A., Taneoka, A., Yamoto, R., Ferri, S., Sode, K., 2008. Biotechnol. Lett. 30, 1453–1460. Scognamiglio, V., Staiano, M., Rossi, M., D’Aurio, S., 2004. J. Fluoresc. 14, 491–498. Salins, L.L., Ware, R.A., Ensor, C.M., Daunert, S., 2001. Anal. Biochem. 294, 19–26. Tolosa, L., Rao, G., 2006. In: Geddes, C.D., Lakowicz, J.R. (Eds.), Top. Fluoresc. Spectrosc. 11, 323–332. Trau, D., Renneberg, R., 2003. Biosens. Bioelectron. 18, 1491–1499. Verveer, P.J., Squire, A., Bastiaens, P.I., 2000. Biophys. J. 78, 2127–2137. Wilson, G.S., Gifford, R., 2005. Biosens. Bioelectron. 20, 2388–2403. Ye, K., Schultz, J.S., 2003. Anal. Chem. 75, 3451–3459. Zhi, Z.L., Haynie, D.T., 2006a. Chem. Commun. (Camb.), 147–149. Zhi, Z.L., Haynie, D.T., 2006b. Artif. Cells Blood Substit. Immobil. Biotechnol. 34, 189–203.
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Fluorescence lifetime spectroscopy and imaging of nano-engineered glucose sensor microcapsules based on glucose/galactose-binding protein Tania Saxl a,∗ , Faaizah Khan a , Daniel R. Matthews b , Zheng-Liang Zhi a , Olaf Rolinski c , Simon Ameer-Beg b , John Pickup a a
Metabolic Unit, King’s College London School of Medicine, Guy’s Hospital, London SE1 9RT, UK Randall Division of Cell and Molecular Biophysics, Richard Dimbleby Department of Cancer Research, King’s College London School of Medicine, Guy’s Hospital, London SE1 1UL, UK c Dept of Physics, University of Strathclyde, Glasgow G4 0NG, UK b
a r t i c l e
i n f o
Article history: Received 8 January 2009 Received in revised form 26 March 2009 Accepted 6 April 2009 Available online 14 April 2009 Keywords: Fluorescence Lifetime FLIM Glucose sensor Diabetes Microcapsule Nanolayers Badan
a b s t r a c t We aimed to develop microsensors for eventual glucose monitoring in diabetes, based on fluorescence lifetime changes in glucose/galactose-binding protein (GBP) labelled with the environmentally sensitive fluorophore dye, badan. A mutant of GBP was labelled with badan near the binding site, the protein adsorbed to microparticles of CaCO3 as templates and encapsulated in alternating nano-layers of polyl-lysine and heparin. We used fluorescence lifetime imaging (FLIM) with two-photon excitation and time-correlated single-photon counting to visualize the lifetime changes in the capsules. Addition of glucose increased the mean lifetime of GBP-badan by a maximum of ∼2 ns. Analysis of fluorescence decay curves was consistent with two GBP states, a short-lifetime component (∼0.8 ns), likely representing the open form of the protein with no bound glucose, and a long-lifetime component (∼3.1 ns) representing the closed form with bound glucose and where the lobes of GBP have closed round the dye creating a more hydrophobic environment. FLIM demonstrated that increasing glucose increased the fractional proportion of the long-lifetime component. We conclude that fluorescence lifetime-based glucose sensing using GBP encapsulated with nano-engineered layer-by-layer films is a glucose monitoring technology suitable for development in diabetes management. © 2009 Elsevier B.V. All rights reserved.
1. Introduction Blood glucose monitoring is an essential component of modern diabetes care, particularly in type 1 diabetes and insulin-requiring type 2 diabetes. It allows detection of hyper- and hypoglycaemia, and adjustment of insulin, diet and exercise to achieve and maintain near-normal blood glucose concentrations, thereby slowing or preventing the development of serious microvascular complications (Bergenstal and Gavin, 2005; Pickup et al., 2005b). The ideal glucose monitoring technology for diabetes must measure glucose continuously whilst also being minimally invasive, or non-invasive. Present commercially available glucose sensors for continuous monitoring in vivo are needle-type probes implanted in the subcutaneous tissue and are based on either amperometric enzyme electrodes (Heller and Feldman, 2008; Mastrototaro, 2000; Wilson and Gifford, 2005) or microdialysis (Maran et al., 2002). They can be used only in the short term (without re-implantation), and suffer from unpredictable drift and less than optimal accuracy,
∗ Corresponding author. Tel.: +44 207 188 1909; fax: +44 207 188 0146. E-mail address: [email protected] (T. Saxl). 0956-5663/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.bios.2009.04.003
thereby requiring frequent re-calibration. A promising alternative to electrochemistry that is being researched by several groups is fluorescence-based glucose sensing; fluorescence has the advantage of being highly sensitive and free from interference from electro-active components in the tissues (Pickup et al., 2005a). In previous work, we described a glucose-sensing strategy based on engineered bacterial glucose/galactose-binding protein (GBP) labelled with the environmentally sensitive fluorophore, badan (6-bromo-acetyl-2-dimethylaminonaphthalene) at a cysteine mutation at position 152 near the binding site (Khan et al., 2008). Badan displays an increased fluorescence intensity in a hydrophobic environment, such as in the interior of a protein. GBP has been used for glucose sensing in work by several others (e.g. Ballerstadt and Schultz, 2000; Deuschle et al., 2006; Fehr et al., 2003; Ge et al., 2004; Marvin and Hellinga, 1998; Salins et al., 2001; Scognamiglio et al., 2004; Tolosa and Rao, 2006) but not using badan as the reporter of glucose binding. The single polypeptide chain of GBP folds into two domains, connected by a hinge; GBP specifically binds glucose and galactose, and binding of glucose at a site between the domains induces a significant conformational change as the domains close around the ligand (Borrock et al., 2007; Dwyer and Hellinga, 2004; Marvin and Hellinga, 1998). The decrease in
3230
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
the polarity in the micro-environment of the badan as the lobes close around the badan at the glucose binding site produces a maximal 300% increase in fluorescence intensity on addition of glucose (Khan et al., 2008). This compares with a less than 20% change in fluorescence intensity in the more researched fluorescence energy transfer (FRET)-based systems where GBP is labelled with donor and acceptor fluorophores at different sites in the protein molecule (Khan et al., 2008; Ye and Schultz, 2003). Time-resolved fluorescence measurements are known to be relatively independent of light scattering, fluorophore concentration and photobleaching and have therefore been used as an alternative to intensity measurements in molecular glucose sensing (Lakowicz, 2006). In this paper, we report studies on the fluorescence of GBPbadan which will underpin the later use of this protein as a glucose monitoring device in diabetes. Firstly, we studied the fluorescence lifetime of badan-labelled GBP and its variation with glucose. We also explore two further contributions to fluorescence-based glucose sensing in the in vivo environment. Firstly, we test whether GBP-badan maintains its function when encapsulated in nanoengineered microvesicles, using our recently reported technique for layer-by-layer electrostatic deposition of alternately charged polypeptide onto a protein-loaded calcium carbonate template (Zhi and Haynie, 2006a,b). Such self-assembled, layer-by-layer glucose sensors have several potentially useful properties for in vivo application, including tunable permeability to the measured analyte (glucose) and the possibility of enhanced biocompatibility. They may thus be optimised for impregnation in the dermis as a ‘smart tattoo’ for minimally invasive glucose sensing (Pickup et al., 2008). Secondly, we make use of a custom-built fluorescence lifetime imaging (FLIM) system with two-photon excitation to visualise lifetime changes in the GBP-badan glucose sensor particles; in this technique, differences in lifetime are used to generate contrast in the image (Ameer-Beg et al., 2005), particularly suitable for determination of localised fluorescence lifetime changes in one or several microcapsules. 2. Material and methods 2.1. Materials Most chemicals used for engineering the GBP mutant were of molecular biology grade and purchased from Sigma–Aldrich (St. Louis, USA). The pTZ18U-mglB vector containing the GBP gene was a kind gift from Dr. S. D’Auria. The plasmid pET303/CT-His vector was purchased from Invitrogen (Paisley, UK). E. coli DH5␣ cells were used as host cells for plasmid proliferation. LB media supplemented by antibiotics (50 g/ml of kanamycin or 100 g/ml of ampicillin) were employed to grow cells. E. coli BL21(DE3) was from BD Biosciences (Franklin Lakes, NJ, USA). All restriction enzymes were purchased from New England Biolabs (Hitchen, UK). Quick-change site-directed mutagenesis kit was purchased from Stratagene (La Jolla, CA, USA). The Rapid DNA ligation kit and long-template polymerase chain reaction (PCR) enzyme (Expand) were from Roche Applied Science (Basel, Switzerland). The kit used for plasmid extraction and Ni-NTA agarose was from Qiagen (West Sussex, UK) and the kit used to purify PCR products or restriction reactions was from Qbiogene (Morgan Irvine, CA, USA). The fluorescent probe badan was from Invitrogen and 5 m-diameter spherical CaCO3 particles were from PlasmaChem GmbH (Germany). 2.2. Construction of expression vector pET303-GBP and purification of H152C mutant of GBP Details of the methodology are given in Khan et al. (2008). In brief, the GBP gene (mglB) was isolated from the plasmid pTZ18U-
mglB by PCR and ligated into pET303/CT-His vector using a Rapid DNA ligation kit to form pET303-GBP. For the H152C mutant, pET303-GBP was used as a template. Site-directed mutagenesis was performed using the Quick-change mutagenesis kit. DNA sequencing data verified the presence of the desired point mutation. A single colony of E. coli BL21(DE3) transformed with the pET303-GBP plasmid was inoculated in LB media containing 100 g/ml of ampicillin and grown at 37 ◦ C. Expression of the protein was induced by adding isopropyl-2-d-thiogalactopyranoside to a final concentration of 1 mM. Bacterial cells were lysed and the cell extract was clarified by centrifugation. Affinity chromatography was performed in a glass column packed with 5 ml Ni-NTA agarose. The protein was eluted with buffer containing imidazole. The purity of GBP was determined by SDS-PAGE using 10% acrylamide gels that were viewed by Coomassie Blue staining. 2.3. Fluorophore labelling To label GBP with badan, 50 M protein was dissolved in 5 mM Tris(2-carboxyethyl)phospine in phosphate-buffered saline (PBS) pH 7.4, and then a 10-fold excess of dye (500 M) was added and the mixture incubated overnight at room temperature, after which it was purified on a Sepadex G-25 gel filtration column. 2.4. Nano-engineered microcapsules containing GBP-badan The encapsulation method of Zhi and Haynie (Zhi and Haynie, 2006a,b) was modified as follows. All buffers were 50% (v/v) 10 mM Tris (pH 7.0)/polyethylene glycol 300, unless otherwise stated. 50 l of 20 M GBP-badan in 10 mM PBS pH 7.4 was mixed with 5 mg of 5 m-diameter spherical CaCO3 particles. CaCO3 acts as a template to adsorb the protein. The mixture was left on a roller at 4 ◦ C for 30 min to allow the GBP-badan to adsorb onto the template. Particles were then centrifuged at 1000 × g for 2 min and washed with buffer. 50 l of 1 mg/ml poly-l-lysine was added and particles were left to stand for 5–10 min on ice. The particles were then centrifuged and washed in 100 l buffer. This process was repeated with the GBP-badan solution and followed by a further layer of poly-l-lysine. The final protein incorporated into the particle was quantified by absorption spectroscopy at 40% of the original 20 M GBP-badan solution. Six further bilayers of 50 l of 1 mg/ml heparin in buffer and poly-l-lysine were added sequentially as above. The resulting particles were centrifuged and re-suspended in 50 l PBS. 2.5. Steady-state fluorescence measurements Steady-state fluorescence measurements were recorded on a PerkinElmer LS50B fluorimeter (PerkinElmer Instruments, Beaconsfield, UK). The excitation and emission wavelength of badan was at 400 and 550 nm respectively. All data were obtained at room temperature using quartz cuvettes with sample volume of 100 L. The labelled protein was incubated with increasing amounts of dglucose for 15–20 min at room temperature before fluorescence was recorded. 2.6. Fluorescence lifetime measurement and imaging Excitation for the solution measurements was provided by a femtosecond titanium-sapphire laser (Coherent, Santa Clara, CA, USA), operating at 800 nm (120 fs) and up-converted by second harmonic generation to 400 nm and directed to a cuvette holder. Fluorescence emission from the cuvette was collected using a lens, bandpass filtered (550 ± 10 nm) and detected using a fast photomultiplier tube (PMT) (PMH-100, Becker and Hickl GmbH, Berlin, Germany) and time-correlated single-photon counting (TCSPC) electronics (SPC830, Becker and Hickl GmbH, Berlin, Germany).
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
3231
Fluorescence lifetime imaging (FLIM) was performed using a custom-built multi-photon system constructed around an upright 90i fluorescence microscope (Nikon, Tokyo, Japan) and similar to that described elsewhere (Parsons et al., 2005; Peter et al., 2005). In brief, excitation was provided by a titanium-sapphire laser (Coherent, Santa Clara, CA, USA), operating at 800 nm. Time-resolved detection was provided by a non-descanned detection channel with a fast PMT (PMH-100, Becker and Hickl GmbH, Berlin, Germany) and a TCSPC personal computer plug-in board (SPC830, Becker and Hickl GmbH, Berlin, Germany). The instrument response was measured from the hyper-Rayleigh scattering of highly attenuated excitation in a suspension of 20 nm colloidal gold (Sigma–Aldrich Company Ltd., G-1652) and used in the analysis and fitting of data by iterative re-convolution. Data were collected at 550 ± 10 nm. The laser power was adjusted to give average photon counting rates of the order 104 –105 photons s−1 (0.0001–0.001 photons per excitation event) below the maximum counting rate afforded by the TCSPC electronics to avoid pulse pile-up. All measurements were made in 10 mM PBS pH 7.4. Cuvette measurements of GBP-badan in glucose were performed by sequentially adding aliquots of 1 mM glucose solution to 100 l of 12 M GBPbadan. The solution was mixed and left to equilibrate for 10 min between each measurement. For multi-photon FLIM, glucose was added sequentially as in the cuvette experiments to 100 l capsules in buffer (made using 10 l of capsule stock). For FLIM, 0.5 l of capsule suspension was placed into a cavity on a microscope slide and the cavity sealed with a cover glass. 2.7. Data analysis Analysis of the fluorescence transients was performed using the TRI2 analysis package (courtesy of Dr. Paul Barber, Gray Cancer Institute of Radiation Oncology and Biology, Oxford University, Oxford, UK). The analysis can be performed on a pixel-by-pixel, image-by-image basis or by considering the whole data set globally (Barber et al., 2005, 2009; Verveer et al., 2000). Global analysis assumes one or more lifetimes across all the images, whilst enabling the pre-exponential factors to be determined for each decay. The fractional contribution of the lifetime was calculated from the preexponential factor, see Eqs. (1) and (2). f (t) = Z + Fi =
˛i exp
−t i
˛
i
˛t
(1) (2)
where ˛i are the pre-exponential factors, Fi the fractional contribution of lifetime i to the transient and Z is a constant corresponding to the background noise. FLIM images were generated from the fractional contribution, F, for the long-lifetime state at each pixel. 3. Results and discussion 3.1. Fluorescence intensity and lifetime of free badan in solvents of different polarity We used the environmentally (polarity) sensitive fluorophore badan for detecting glucose-induced changes in the conformation of GBP. The fluorescence emission maximum for free badan was 500 nm in the dipolar, aprotic solvent dimethyl formamide (DMF), shifting to 550 nm in 2% DMF, 98% water (Fig. 1a), but with a marked reduction in fluorescence intensity in the aqueous environment. Fluorescence lifetime decays for badan in DMF and water were best fitted by a bi-exponential model (Fig. 1b), confirmed by the residuals for mono-exponential and bi-exponential fits (Fig. 1c).
Fig. 1. (a) Fluorescence emission spectra for free badan in 100% DMF (solid line) and 98% water, 2% DMF (dashed line), plotted with measured intensity and normalised intensity. (b) Lifetime fluorescence decays for badan in 100% DMF (at 500 ± 10 nm) and 98% water, 2% DMF. The bi-exponential fit to each decay is shown with corresponding values for the two lifetimes of 2.1 and 0.5 ns for 100% DMF, and 0.8 and 0.2 ns for 98% aqueous solution. The instrument response function is shown for comparison with the fluorescence decay. (c) Residuals for bi- and mono-exponential models.
Lifetimes generated were estimated as 2.4 and 1.4 ns in DMF, and 0.8 ns and <0.2 ns in water, indicating the presence of two emissive states of the free dye in these solvents, the polarity of the solvent affecting the fluorescence intensity. The two lifetime components of badan in water and DMF, is in agreement with the dual fluorescence described recently (Koehorst et al., 2008). Exposed to polar protic water, badan has reduced fluorescent lifetimes and intensity compared to the environment of dipolar aprotic DMF, indicating that a change to a more hydophobic environment would be expected to increase the lifetime, e.g. as the badan label near the binding site of GBP is enclosed by the polypeptide lobes as glucose binds to the protein (Borrock et al., 2007). We have previously found that there is no corresponding shift in emission wavelength when free GBP-badan is compared to GBP-badan in the presence of glucose, both emission maxima being at 550 nm (Khan et al., 2008). 3.2. Glucose-dependent fluorescence changes of GBP-badan On binding to GBP, the photophysical properties of the sensor dye were modified due to close proximity of the protein but as Fig. 2a illustrates, the environmental sensing property was retained.
3232
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
Table 1 Fluorescence lifetimes () and corresponding fractional contributions (F) for GBP-badan, obtained by fitting data using mono-, bi- and tri-exponential models. Best fits are highlighted in bold. [Glucose] (M)
1 (ns)
Fits to individual fluorescence decays 0 0.9 1.3 1.5 310
2.9 3.4 3.2
(F1)
2 (ns)
(F2)
– (0.41) (0.25)
0.5 0.7
(0.59) (0.57)
– (0.83) (0.75)
1.7 2.9
(0.17) (0.21)
Global fit to fluorescence decays from nine glucose concentrations (0 and 310 M shown) 0 2.5 – 310 2.5 – 0 3.1 (0.10) 0.8 (0.90) 310 3.1 (0.95) 0.8 (0.05) 0 3.2 (0.03) 1.0 (0.58) 310 3.2 (0.71) 1.0 (0.23)
The fluorescence lifetime decay of GBP-badan in PBS (zero glucose), and in a saturating glucose concentration of 310 M are shown for comparison. Both individual analysis of each decay and global analysis of nine decays from nine increasing glucose concentrations were performed. Fitting the fluorescence decay at zero glucose with a bi-exponential model produced decay constants of 1.3 and 0.5 ns, with a mean lifetime of 0.8 ns. The two states were similar to those found in water and consistent with an open conformation of GBP
3 (ns)
0.1
1.2
0.3 0.3
(F3)
2
(0.17)
9.4 1.4 1.2
mono bi tri
(0.04)
1.1 1.0 1.1
mono bi tri
6.9
mono
1.1
bi
4.2
tri
(0.39) (0.06)
with badan exposed to the polar environment. With a saturating glucose concentration, where a larger proportion of the closed form of GBP-badan is expected, a mono-exponential fit produced a lifetime of 2.9 ns. This corresponds to an increase in mean lifetime of 2.1 ns with the addition of saturating glucose. Short-lifetime states are generated in the presence of hydrogen bonding, as would be found in water (Józefowicz et al., 2005). On addition of glucose, the increase in fluorescence lifetime that we observed can be explained by considering the interior of the protein to be a hydrophobic environment, the lobes of GBP closing around the glucose, effectively shielding the fluorophore from the surrounding medium. The apparent change from two short-lifetime states at zero glucose, to an increasing fraction of a single long-lifetime state as glucose is added, is possibly due to badan being bound in different orientations (rotomers) at zero glucose and restriction in the geometry occurring as the conformation of the protein changes, excluding one emissive sates and favouring another. A similar trend was seen in by Gilardi et al. in maltose binding protein labelled with nitrobenzoxadiazole (Gilardi et al., 1997). On global analysis of the data set obtained from nine glucose concentrations, we obtained lifetimes for a two-component model of 3.1 and 0.8 ns, the relative fraction of each varying with glucose concentration. The two short-lifetime states, seen in the absence of glucose, are approximated to one short-lifetime state (0.8 ns), the contribution of which decreases with addition of glucose. The two lifetimes obtained by global analysis are in agreement with those recovered from the individual fits, as is the increase in mean lifetime of approximately 2 ns on addition of glucose (summarised in Table 1). The short-lifetime state of 0.8 ns is consistent with an open conformation of GBP-badan and the long-lived, 3.1 ns state as the closed conformation of GBP-badan, generated by binding of glucose. Fig. 2b shows the fractional contribution of the two global lifetimes as the glucose concentration was increased: the proportion of the open form decreased and the proportion of the closed form increased. 3.3. Glucose-dependent changes of GBP-badan encapsulated in microcapsules
Fig. 2. (a) Cuvette-based measurements of fluorescence emission for GBP-badan in PBS and 310 M glucose (saturated glucose). The mono-exponential fit to the saturating glucose and bi-exponential fit to the zero glucose decays are indicated, with corresponding lifetime values of 2.9, 1.3 and 0.5 ns respectively. (b) Global analysis of the data from 9 glucose concentrations, including those shown in a. The fractional contribution of lifetime states 3.1 and 0.8 ns is plotted against glucose concentration. The two lifetimes were determined by globally fitting a bi-exponential model.
GBP-badan was adsorbed onto spherical calcium carbonate particles, 5 M in diameter, and then encapsulated in six bi-layers of alternating poly-l-lysine and heparin (layer-by layer technique) stabilised with PEG. Fig. 3 shows images of the spherical capsules obtained by two-photon FLIM, taken for 0, 10, 25, 50 and 100 M glucose. Fig. 3a shows intensity profiles for each glucose concentration. Global lifetimes were determined as 2.3 and 0.4 ns by fitting
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
3233
Fig. 3. Fluorescence lifetime microscopy images generated from the time-correlated fluorescence data at varying glucose concentrations and their analysis: (a) diagram illustrating a micro-capsule (not to scale); (b) fluorescence intensity images of the GBP-badan microcapsules; (c) the fractional contribution (F) of the long-lifetime state (2.3 ns), generated from global analysis and displayed as a change in colour of the image from blue (low fraction of closed [glucose-bound] form) to red (high fraction of closed [glucose-bound] form). (d) Pixel histograms for each image. The maxima have been normalised to display the histograms on the same scale. (e) Graph showing the mean value of F against glucose concentration obtained from the images in (a). Pixels were binned for each capsule and the error is the standard deviation between capsules in each image.
the entire set of images to a two-component model, as in the free protein case. In the set of FLIM images, the increasing fractional contribution of the long-lifetime state, corresponding to the increasing proportion of the bound GBP-badan glucose, is displayed as a colour change from blue to red on addition of glucose (Fig. 3b). An increase
in the long-lifetime state on addition of glucose can be seen in the pixel histograms (Fig. 3c). Fig. 3d shows the fractional contribution of the long-lifetime 2.3 ns state (GBP bound to glucose) for each glucose concentration, obtained from global analysis of the images. Lifetime data were summed over all pixels for each image.
3234
T. Saxl et al. / Biosensors and Bioelectronics 24 (2009) 3229–3234
Layer-by-layer techniques which involve encapsulation and particle covering (Donath et al., 1998) have been reviewed recently (Ariga et al., 2007) and the present studies where the sensor is encapsulated, extends our previous work, in which glucose oxidase was encapsulated in nano-engineered polypeptide layers (Zhi and Haynie, 2006a,b). Trau and Renneberg have also reported layer-bylayer encapsulation of a sensing protein (glucose oxidase), though using micro-crystals of the enzyme rather than adsorption to a template of CaCO3 , alternating layers of polyelectrolytes rather than polypeptide and amperometric detection technology rather than fluorescence (Trau and Renneberg, 2003). Also, Chinnayelka and McShane have described FRET-based glucose sensors using concanavalin A (Chinnayelka and McShane, 2004) or apo-glucose oxidase (Chinnayelka and McShane, 2005) as the glucose receptor and layer-by-layer encapsulation in polyelectrolytes. Microcapsules similar to ones used in our study may be suitable for a ‘smart tattoo’ type sensor, where particles are impregnated in the dermis or subcutaneous tissue and glucose measured from outside the body (Pickup et al., 2008). Our work has limitations. The operating range is relatively restricted and an order of magnitude below the typical pathophysiological range of blood glucose concentrations in diabetes (<30 mM), even when the GBP-badan was encapsulated. Several approaches to extending the range of sensing into the mM range exist (which will be needed for clinical use) and will be pursued in future work. For example, a recently reported Phe16Ala mutation increased the Kd of GBP to 3.9 mM (Sakaguchi-Mikami et al., 2008). In this study, we used FLIM to visualize and quantify microcapsule fluorescence. FLIM, especially in combination with multi-photon excitation, is a powerful biological imaging tool with emerging applications (Festy et al., 2007). A special advantage is that it allows spacially resolved analyte sensing, i.e. highly localised measurements within fabricated structures (such as the glucose microsensors here) and within cells and tissues. This is likely to prove valuable in diabetes research for mapping glucose and other analyte changes in various tissue compartments, in interstitial fluid and in different cells and cell organelles. 4. Conclusion In this paper, we demonstrate for the first time proof-of-concept for glucose sensing based on fluorescence lifetime changes of GBP labelled with the environmentally sensitive dye badan. We also show that GBP-badan can be encapsulated in microcapsules constructed by a layer-by-layer technology, and that the glucose sensing capacity is preserved in the capsules. As glucose concentration was increased, we observed an increasing mean lifetime of GBP-badan, due to binding of glucose. This offers a robust methodology for measurement of glucose concentration using a GBP-badan sensor. In future research it will be necessary to construct sensors based on this approach which are suitable for testing vivo and to check requirements such as freedom from interference and quenching and long-term stability. Acknowledgements JCP is grateful to EPSRC (EP/D062861/1), the Diabetes Foundation and the Wellcome Trust for generous grant support.
SAB and DRM gratefully acknowledge the support of the EPSRC (EP/C546105/1), Cancer Research UK (C1519/A6906), the Royal Society and Dimbleby Cancer Care. OR thanks the EPSRC for grant support (EP/D062861/1). Many thanks to Dr. P. Barber (Oxford University) for providing the TRI2 FLIM imaging software for global image analysis, Dr. S. D’Auria for providing the vector contining the GBP gene and Prof. DJS Birch (Strathclyde University) for valuable discussion and assistance. References Ameer-Beg, S.M., Peter, M., Keppler, M.D., Prag, S., Barber, P.R., Ng, T., Vojnovic, B., 2005. Proc. SPIE 5700, 152–161. Ariga, K., Hill, J.P., Ji, Q., 2007. Phys. Chem. Chem. Phys. 9, 2319–2340. Ballerstadt, R., Schultz, J.S., 2000. Meth. Biotechnol. 7, 89–98. Barber, P.R., Ameer-Beg, S.M., Gilbey, J.D., Edens, R.J., Ezike, I., Vojnovic, B., 2005. Proc. SPIE 5700, 171–181. Barber, P.R., Ameer-Beg, S.M., Gilbey, J., Carlin, L.M., Keppler, M., Ng, T.C., Vojnovic, B., 2009. Multiphoton time-domain FLIM: practical application to protein-protein interactions using global analysis. J. R. Soc. Interface 6, S93–S105. Bergenstal, R.M., Gavin, J.R., 2005. Am. J. Med. 118 (Suppl. 9A), 1S–6S. Borrock, M.J., Kiessling, L.L., Forest, K.T., 2007. Prot. Sci. 16, 1032–1041. Chinnayelka, S., McShane, M.J., 2004. Biomacromolecules 5, 1657–1661. Chinnayelka, S., McShane, M.J., 2005. Anal. Chem. 77, 5501–5511. Deuschle, K., Chaudhuri, B., Okumoto, S., Lager, I., Lalonde, S., Frommer, W.B., 2006. Plant Cell 18, 2314–2325. Donath, E., Sukhorukov, G.B., Caruso, F., Davis, S.A., Möhwald, H., 1998. Angew. Chem. Int. Ed. Engl. 37, 2201–2205. Dwyer, M.A., Hellinga, H.W., 2004. Curr. Opin. Struct. Biol. 14, 495–504. Festy, F., Ameer-Beg, S.M., Ng, T., Suhling, K., 2007. Mol. Biosyst. 3, 381–391. Ge, X., Tolosa, L., Rao, G., 2004. Anal. Chem. 76, 1403–1410. Gilardi, G., Mei, G., Rosato, N., Agro, A.F., Cass, A.E., 1997. Protein Eng. 10, 479–486. Fehr, M., Lalonde, S., Lager, I., Wolff, M.W., Frommer, W.B., 2003. J. Biol. Chem. 278, 19127–19133. Heller, A., Feldman, B., 2008. Chem. Rev. 108, 2482–2505. Józefowicz, M., Kozyra, K.A., Heldt, J.R., Heldt, J., 2005. Chem. Phys. 320, 45–53. Khan, F., Gnudi, L., Pickup, J.C., 2008. Biochem. Biophys. Res. Commun. 365, 102–106. Koehorst, R.B., Spruijt, R.B., Hemminga, M.A., 2008. Biophys. J. 94, 3945–3955. Lakowicz, J.R., 2006. Principles of Fluorescence Spectroscopy, 3rd edn. SpringerVerlag, New York, p. 954. Maran, A., Crepaldi, C., Tiengo, A., Grassi, G., Vitali, E., Pagano, G., Bistoni, S., Calabrese, G., Santeusanio, F., Leonetti, F., Ribaudo, M., Di Mario, U., Annuzzi, G., Genovese, S., Riccardi, G., Previti, M., Cucinotta, D., Giorgino, F., Bellomo, A., Giorgino, R., Poscia, A., Varalli, M., 2002. Diabetes Care 25, 347–352. Marvin, J.S., Hellinga, H.W., 1998. J. Am. Chem. Soc. 120, 7–11. Mastrototaro, J.J., 2000. Diabetes Technol. Ther. (Suppl. 1), S13–S18. Parsons, M., Monypenny, J., Ameer-Beg, S.M., Millard, T.H., Machesky, L.M., Peter, M., Keppler, M.D., Schiavo, G., Watson, R., Chernoff, J., Zicha, D., Vojnovic, B., Ng, T., 2005. Mol. Cell. Biol. 25, 1680–1695. Peter, M., Ameer-Beg, S.M., Hughes, M.K., Keppler, M.D., Prag, S., Marsh, M., Vojnovic, B., Ng, T., 2005. Biophys. J. 88, 1224–1237. Pickup, J.C., Hussain, F., Evans, N.D., Rolinski, O.J., Birch, D.J., 2005a. Biosens. Bioelectron. 20, 2555–2565. Pickup, J.C., Hussain, F., Evans, N.D., Sachedina, N., 2005b. Biosens. Bioelectron. 20, 1897–1902. Pickup, J.C., Zhi, Z.L., Khan, F., Saxl, T., Birch, D.J., 2008. Diabetes Metab. Res. Rev. 24, 604–610. Sakaguchi-Mikami, A., Taneoka, A., Yamoto, R., Ferri, S., Sode, K., 2008. Biotechnol. Lett. 30, 1453–1460. Scognamiglio, V., Staiano, M., Rossi, M., D’Aurio, S., 2004. J. Fluoresc. 14, 491–498. Salins, L.L., Ware, R.A., Ensor, C.M., Daunert, S., 2001. Anal. Biochem. 294, 19–26. Tolosa, L., Rao, G., 2006. In: Geddes, C.D., Lakowicz, J.R. (Eds.), Top. Fluoresc. Spectrosc. 11, 323–332. Trau, D., Renneberg, R., 2003. Biosens. Bioelectron. 18, 1491–1499. Verveer, P.J., Squire, A., Bastiaens, P.I., 2000. Biophys. J. 78, 2127–2137. Wilson, G.S., Gifford, R., 2005. Biosens. Bioelectron. 20, 2388–2403. Ye, K., Schultz, J.S., 2003. Anal. Chem. 75, 3451–3459. Zhi, Z.L., Haynie, D.T., 2006a. Chem. Commun. (Camb.), 147–149. Zhi, Z.L., Haynie, D.T., 2006b. Artif. Cells Blood Substit. Immobil. Biotechnol. 34, 189–203.