Frauke
Schibilla
a,
Linda
Stegemann
b,
Cristian A.
Strassert
b,
Fabio
Rizzo
*c and
Bart Jan
Ravoo
*a
aOrganic Chemistry Institute and CeNTech, Westfälische Wilhelms-Universität Münster, Corrensstr. 40, D-48149 Münster, Germany. E-mail: b.j.ravoo@uni-muenster.de
bPhysical Institute and CeNTech, Westfälische Wilhelms-Universität Münster, Heisenbergstr. 11, D-48149 Münster, Germany
cIstituto di Scienze e Tecnologie Molecolari (ISTM), CNR, Via Golgi 19, I-20133 Milano, Italy. E-mail: fabio.rizzo@istm.cnr.it
First published on 4th January 2016
Fluorescent β-cyclodextrin vesicles (β-CDV) that display host cavities available for host–guest interactions at the vesicle surface were prepared by incorporation of the hydrophobic spirobifluorene-based dye 1 into the membrane of unilamellar vesicles. Fluorescence quenching of dye 1 was observed in the presence of different quenchers. Methyl viologen 2 does not quench dye 1 because it does not bind to β-CDV. 4-Nitrophenol 3 and 4-nitrophenol covalently connected to adamantane 4 quench the fluorescence of dye 1 in neutral solution, but by different mechanisms according to lifetime measurements. The quenching efficiency of 3 is pH dependent due to the presence of the phenolate form. Competition experiments with excess host and guest showed that 3 is likely to diffuse in and out of the membrane, while 4 forms an inclusion complex with β-CDV leading to close contact and efficient quenching. Our findings confirm that this dynamic supramolecular system is a versatile model to investigate quenching and recognition processes in bilayer membranes.
Almost all biological processes depend on molecular recognition. Vesicles are widely used as models for biological membranes.8,9 Vesicles can encapsulate water-soluble molecules in the aqueous interior and incorporate hydrophobic molecules into the bilayer membrane. Synthetic amphiphiles used to form vesicles can additionally include a binding motif which selectively binds complementary molecules. Vesicles are attractive systems to control photophysical processes such as quenching.10,11 Photochemical reactivity is often enhanced by binding the quencher and the fluorophore to vesicles compared to homogeneous solutions.12 For example, electron transfer from hydrophobic molecules incorporated into the membrane of liposomes to molecules bound electrostatically at the surface of the vesicle was investigated.13,14 Other reports describe the preparation of an artificial photosynthetic reaction center incorporated into the bilayer of a liposome15 and the improvement of the efficiency of electron transfer by using charged photosensitizers adsorbed on charged vesicles.16 Also luminescent chemosensing vesicles have been described.17–19
Among the amphiphilic compounds used to form vesicles, modified cyclodextrins (CDs) appear very interesting as stimuli-responsive nanomaterials due to their host–guest properties.20 In fact, CDs are able to act as a host for hydrophobic molecules in aqueous solution and for example are used as molecular encapsulators in pharmaceutical applications.21–23 In addition, amphiphilic CDs form self-assembled nanostructures.24 Vesicles consisting of β-CD amphiphiles (β-cyclodextrin vesicles, β-CDV) can encapsulate molecules in their aqueous interior, incorporate hydrophobic molecules into the membrane and form selective inclusion complexes at the surface of the vesicle.25 For example, the incorporation of hydrophobic, magnetic nanoparticles into the membrane of β-CDV and directed assembly by magnetic forces and photoresponsive hydrophobic guest molecules have been reported.26 Studies on the modification of host–guest interactions at β-CDV in the presence of competitive host or guest are also reported in the literature.27,28 Förster resonance energy transfer (FRET) between a hydrophobic dye incorporated into the membrane of β-CDV acting as a donor and an acceptor fluorophore bearing adamantane, a well-known guest for β-CD, was used to investigate the dynamic multivalent molecular recognition of CDV.29 Thus, β-CDV appears as a suitable system to control photophysical processes by the combination of hydrophobic co-partitioning and host–guest interactions. Recently, there have been a number of reports on the combination of chromophores with β-CDV or CD nanoparticles. The integration of photosensitizers in β-CDV or CD nanoparticles may potentially find application in nanomedicine as a system for photodynamic therapy.21,30–32 However, systematic photophysical studies of these structures are rare.33 In this paper we use the β-CDV as a model membrane to investigate the quenching of the fluorescence of an incorporated dye with guest compounds interacting in a different way with β-CDV. In particular, we describe the incorporation of spirobifluorene chromophore 134 (2,7-bis-(4-(N,N-diphenylamino)phen-1-yl)-9,9-spirobifluorene) into the membrane of β-CDV and the investigation of fluorescence quenching by three different quenchers depicted in Chart 1.
Methyl viologen 2 (commonly called paraquat) is widely used as a herbicide and it is highly toxic to humans and animals. People working with this herbicide are more likely to develop Parkinson's disease.35,36 Also 4-nitrophenol 3 is toxic, hazardous for the environment and suspected to be carcinogenic. It is used as an important intermediate in industrial synthesis for explosives, pesticides and drugs and is eventually released to the environment as waste.37 The detection of paraquat and 4-nitrophenol and their derivatives is discussed in the recent literature.38–41 To increase the affinity of 3 to CD, we synthesized 4-(2-((adamant-1-yl)oxy)ethoxy)nitrobenzene 4 in which 3 is covalently connected to adamantane. The fluorescence quenching of 1 was investigated using fluorescence spectroscopy and lifetime measurements.
Unilamellar β-CDV with incorporated dye 1 were prepared by sonication and extrusion in phosphate buffer at pH 7.2. The average diameter of the vesicles was around 120 nm according to dynamic light scattering (DLS) measurements. The concentration of heptakis[6-deoxy-6-dodecylthio-2-oligo(ethyleneoxide)]-β-cyclodextrin (5) was 100 μM, while the average amount of incorporated dye 1 was around 6 μM calculated by absorption measurements (the dye's absorbance lies approximately between 0.40 and 0.45). It is important to note that this calculation is strongly affected by scattered light due to the presence of vesicles. Fig. 1 shows the absorption and emission spectra of 1 incorporated into β-CDV. The absorption spectrum shows the π–π* band with a maximum at 375 nm. The emission band has a peak at 420 nm for an excitation wavelength (λex) of 375 nm, slightly blue shifted compared to the fluorescence in solution.34
Fig. 1 Absorption (dashed line) and normalized fluorescence intensity (solid line) spectra (λexc = 375 nm) of 1 (6 μM) incorporated into β-CDV in phosphate buffer (pH 7.2). |
To study the distribution of the dye in the membrane microscopically, giant unilamellar vesicles (GUVs) with incorporated 1 were prepared by electroformation. Microscopy images of these GUVs (Fig. 2) indicate spherical and elongated vesicles with slightly irregular surfaces. Similar GUVs were previously described by our group for pure β-CDV.43 GUVs show fluorescence exclusively in the region of the membrane and thus these observations confirm that dye 1 is confined to the membrane.
Fig. 2 Microscopy images of GUVs with incorporated 1: left: light microscopy with UV-irradiation. Right: fluorescence microscopy. |
Although the dye and the cyclodextrin are in close proximity, we can reasonably exclude the formation of inclusion complexes. In fact, the rigid spirobifluorene derivative 1 appears too large to bind the cavity of β-CD and the peripheral phenyl rings show low binding constant values.44,45 Thus all cavities along the vesicle can be considered available for host–guest interactions. The vesicles are stable for more than one day and their stability is not affected by the addition of quenchers 2, 3 or 4 up to 100 μM.
Compounds 2–4 all quench the fluorescence of 1 in organic solutions, but each has a different affinity for β-CD. Dicationic quencher 2 is highly hydrophilic and thus is not able to form host–guest complexes with β-CD. Quencher 3 is soluble in water, but shows moderate affinity towards β-CD (Ka ≈ 102–103 M−1),46 while the adamantane derivative 4 strongly binds β-CD47 (Ka ≈ 104 M−1) and is not soluble in water. On the basis of these data, we assumed a dynamic partitioning of nitrophenol 3 between the solution and β-CDV and a preferential localization of 4 at the surface of β-CDV close to the dye. To investigate the fluorescence quenching, emission spectra of 1 incorporated into β-CDV were measured in the presence of 2–4 at different concentrations (see ESI Fig. S1 and S2†). The concentration dependent quenching of 1 by 2–4 shown in Fig. 3 is reported as the ratio F/F0 at the emission maximum (420 nm), where F and F0 are the emission intensities of 1 incorporated into β-CDV in the presence and absence of quenchers, respectively. This ratio allows a direct comparison of the quenching efficiency among different quenchers. The fluorescence data F and F0 were corrected using the formula reported in the ESI,† which takes into account the co-absorption of 1 and the quencher at the excitation wavelength (absorption at λexc = 375 nm) and the inner filter effect at the emission wavelength (absorption at λem = 420 nm) (see ESI Fig. S3†).1 We emphasize the importance of the correction factor, since in fact the uncorrected data would suggest that 3 and 4 have the same quenching efficiency (see ESI Fig. S4†). In the text, we refer to the corrected values of F and F0.
As depicted in Fig. 3, the presence of 2 does not influence the fluorescence in vesicle solution (black squares). As expected, the dicationic and hydrophilic methyl viologen remains in solution resulting in a large distance between 1 and 2, while close contact is necessary for quenching. In contrast, 3 (red circle) and 4 (blue triangle) quench the fluorescence of the dye. Despite their similar structure, 4 results to be more efficient than 3.
To investigate the quenching process in more detail the Stern–Volmer plot was used. The classical Stern–Volmer equation can be written as F0/F = 1 + KSV [Q], where F0 and F is the intensity of the fluorophore in the absence and presence of the quencher, KSV is the Stern–Volmer constant and [Q] is the concentration of the quencher. The equation shows linear behavior and describes dynamic (τ0/τ = F0/F) or static (τ0/τ = 1) quenching (τ0 and τ are the lifetime of the dye in the absence and in the presence of quencher Q, respectively). In some cases dynamic and static quenching appear simultaneously resulting in a positive deviation from linearity. Therefore, the adjusted Stern–Volmer plot, F0/F = (1 + KD [Q]) (1 + Ks [Q]), can be divided into a linear dynamic (KD) and linear static (KS) part, whereby the dynamic part (1 + KD [Q]) is equal to τ0/τ.1
To analyze the Stern–Volmer plot, the fluorescence lifetime of 1 incorporated into β-CDV was measured as a function of the concentration of 2, 3 or 4 (Fig. 4). The average lifetime (τ0) of dye 1 incorporated into β-CDV in aqueous solution is 1.0 ns, which is slightly lower than the value reported for 1 in organic solvent (1.3 ns).34 Upon increasing the concentration of 2 in solution, the lifetime remains almost constant. The addition of 3 leads to a lower decrease in lifetime (τ (100 μM 3) = 0.86 ns) compared to the addition of 4 (τ (100 μM 4) = 0.45 ns) (Fig. 4). The different behaviours of 3 and 4 indicate the operation of dissimilar quenching mechanisms. The Stern–Volmer plots for quenching of 1 incorporated into β-CDV with 3 and 4 are shown in Fig. 5. As depicted in Fig. 5, the slope of F0/F in the presence of 3 or 4 shows a positive deviation from linearity, which refers to a complex quenching process compared to the classic behaviour.
In order to interpret these data, the dynamic and static contributions of the quenching have to be evaluated separately. The dynamic part of the quenching can be determined by lifetime measurements and showed linear behaviour, as indicated in the following equation:
The dynamic Stern–Volmer constant KD was calculated to be 1.75 × 103 M−1 for 3 and 1.2 × 104 M−1 for 4. In both cases the contribution of dynamic quenching to the decrease of luminescence is high. The second quenching contribution ((F0/F)/(τ0/τ)) could be assigned to apparent static quenching, but shows an exponential growth instead of linear behaviour (blue triangle in Fig. 5). The upward curvature cannot be attributed to classic static quenching with the formation of a ground-state non-fluorescent complex, but to the existence of a sphere of effective quenching, as predicted by the Perrin's model.48 In fact, the fluorophore is strictly confined in the membrane with low diffusion freedom. As a consequence, the quencher can act only when it is located near the dye at the moment of excitation. Because of the high local concentration in the vesicle, the positive deviation from linearity in the Stern–Volmer plot with exponential behaviour can be associated with a simultaneous dynamic and static quenching in which the sphere of quenching has to be considered. The equation of the sphere of quenching is F0/F = (1 + KD[Q]) × eVNA[Q] where V is the volume of the sphere and NA the Avogadro's number.1 By plotting vs. [Q] it is possible to calculate the radius of the quenching sphere. We obtained a value for R of 15 Å and 19 Å for 3 and 4 respectively, which are in agreement with the thickness of the membrane (42 Å).25
Additionally, the quantum yield (QY) was measured to calculate the rates of radiative and nonradiative decay (kr and knr).
The QY of the spiro-compound incorporated into β-CDV without the quencher is 45% and thus lower than for 1 in DMF (87%).34 The QY decreases with increasing concentration of 3 or 4, whereby the absorbance of quenchers 3 and 4 overlaps the absorbance of 1 and thus the measured quantum yield had to be corrected (see ESI Fig. S3†).
In the absence of the quencher, the rate of radiative decay of 1 incorporated into β-CDV in buffer solution (4.5 × 108 s−1) is smaller than in DMF (6.8 × 108 s−1), while the rate of nonradiative decay is much higher (β-CDV solution: 5.5 × 108 s−1, DMF: 1.0 × 108 s−1).34 Although the formation of excimers can be excluded, the incorporation into the membrane influences the emission of 1. However, because knr is strictly connected to the quenching process, we focused our attention on its variations by increasing the concentration of quenchers in buffer solution (Table 1). By adding different amounts of quencher 3 or 4, an increase in the rate of nonradiative decay was measured. The increase is much higher for 4 than for 3, which strengthens the interpretation of different quenching mechanisms involved. The high increase of knr for 4 can be explained by taking into account the high affinity of 4 to the cavity of β-CDV.
μM | k nr(3) [108 s−1] | k nr(4) [108 s−1] |
---|---|---|
0 | 5.5 | |
50 | 9.6 | 14.6 |
100 | 10.8 | 21.8 |
Due to the host–guest interaction between the cyclodextrin cavity and adamantane, almost all quencher molecules are located in the proximity of the membrane, which leads to a high local concentration of 4 close to the dye. We can then conclude that knr data are in agreement with the sphere of quenching interpretation.
In the case of 3, the distribution of the quencher in the proximity of 1 appears very different for 4. The 4-nitrophenol has low affinity for the hydrophobic cavity of β-CD (Ka ≈ 102–103 M−1).46 However, it might also be able to diffuse through the hydrophobic membrane of the vesicles due to its aromatic and apolar structure. Thus, 3 can come in close contact to 1 and quench its fluorescence. On the other hand, 3 is soluble in water, which probably leads to a partitioning of 3 between the vesicles and aqueous solution.
To interpret the quenching behaviour of 3, also the pH has to be considered. In phosphate buffer at pH = 7.2, 4-nitrophenol (pKa = 7.15)49 exists in its protonated (phenol) and deprotonated (phenolate) form, while more than 99.9% of 3 is protonated in acetate buffer at pH = 4.5 and deprotonated in carbonate buffer at pH = 10.3, respectively. To investigate the effect of pH, β-CDV with incorporated dye 1 were additionally prepared in acetate buffer and carbonate buffer. Absorption and emission spectra were recorded at different concentrations of 3. In Fig. 6 the normalized fluorescence intensity of 1 in β-CDV is reported at various concentrations of 4-nitrophenol at different pH values. The general trend seems to indicate that the quenching is more efficient if 3 is deprotonated (phenolate) than if it is protonated (phenol), even though the association constant of 4-nitrophenol and 4-nitrophenolate with β-CD is approximately the same.47 This effect can be explained by considering the absorption spectra of 4-nitrophenol as a function of pH (see ESI Fig. S5†).
The absorption maximum of 3 is at 310 nm in acid solution (phenol) and at 405 nm in basic solution (phenolate), while the emission peak of 1 is at 420 nm (Fig. 7). Thus, the variation of pH results in a more extended overlap between the emission of 1 and the absorption of 3 and in an increasing of the inner filter effect. Although the data were corrected by a factor which takes into account the co-absorption and the reabsorption of the emitted light, at pH 7.2 and 10.3 the high absorption band of 3 at higher concentrations can give an overestimation of the quenching. In fact, as reported in the literature48,50 the correction factors regarding the inner filter effect are affected by uncertainties difficult to estimate.
Fig. 7 Absorption of 3 (100 μM) added to 1 (6 μM) incorporated into β-CDV at different pH values and normalized fluorescence intensity (λexc = 375 nm) of the system 1/β-CDV without quenchers. |
However, an alternative explanation in agreement with the recorded data can be also given. The charged phenolate can interact with 1 leading to the formation of a non-emitting ground-state complex (static quenching). This hypothesis is justified by examining the Stern–Volmer plot at different pH values (see ESI Fig. S6†). F0/F appears linear at pH 4.5 suggesting a pure dynamic quenching contribution, while at pH 7.2 and 10.3 the plot leads to an exponential fit. The increased upwards curvature can be associated with an additional static quenching contribution described by the above-mentioned sphere of quenching model.
Finally, the quenching effect of 3 and 4 was also analyzed by the addition of a competitive, nonquenching host (β-CD) and guest (1-adamantane methanol) to obtain more information about host–guest interactions involved in the supramolecular system. In the case of quencher 3, the addition of an excess of β-CD (up to 8 mM) or 1-adamantane methanol (50 μM) to 1 in β-CDV quenched by 3 (50 μM) does not influence the fluorescence intensity. This strengthens the interpretation that 3 is not bound to the CD, but diffuses through the hydrophobic membrane. Thus, the quenching action exerted by 3 can be definitely attributed to its partitioning and diffusion between the aqueous phase and the membrane and not by host–guest interaction.
A completely different behaviour was observed in the case of quencher 4, as depicted in Fig. 8. Upon addition of an excess of β-CD (up to 8.3 mM) to 1 in β-CDV quenched by 4 (50 μM), the fluorescence intensity increased, even though the initial intensity of 1 in the absence of the quencher cannot be fully restored. (A higher concentration of β-CD cannot be achieved due to its limited solubility.) Thus, we can reasonably suppose that β-CD in solution binds competitively to 4 decreasing the amount of the quencher in β-CDV. The lower concentration of 4 in proximity to 1 results in less efficient quenching.
In contrast, the addition of a competitive guest (3.5 mM of 4-diamantane carboxylic acid, 10 mM of 1-adamantane carboxylic acid or 50 μM of 1-adamantane methanol) after quenching with 4 (50 μM) induces a further decrease of the fluorescence intensity. This effect could be explained with a partial displacement of 4 from the β-CD cavities by the presence of the competitive guest. Due to its hydrophobic character, the nitrophenolate-adamantane derivative diffuses into the membrane leading to a higher effective concentration of the quencher in close proximity to the dye. This strengthens the interpretation that the effective quenching of 4 is due to host–guest interactions.
For the preparation of giant unilamellar vesicles (GUVs) 0.50 mM 5 and 0.05 mM 1 were dissolved in chloroform. Two electrode slices coated with ITO were covered with a thin layer of the solution (22.5 μL). The solvent was evaporated in a vacuum oven at 50 °C. After adding buffer solution (1 mM HEPES, 300 mM sucrose, pH 7.2), an alternating electric field was applied (1 V, 10 Hz) at 50 °C for 1 h. The solution was spread on a microscope slide coated with bovine serum albumin to reduce surface adsorption and investigated by light and fluorescence microscopy.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5pp00226e |
This journal is © The Royal Society of Chemistry and Owner Societies 2016 |