Laurdan and Di-4-ANEPPDHQ probe different properties of the membrane

Laurdan and di-4-ANEPPDHQ were obtained from ThermoFisher. RBL-2H3 cells were cultured in MEM supplied with 10% FCS and 1% L-glutamine (all supplied from Sigma-Aldrich). Lipids were bought from Avanti Polar Lipids.

GPMVs were prepared as previously described [27]. Briefly, cells were incubated with 25 mM paraformaldehyde and 10 mM DTT (Sigma) at 37 °C for 1 h. They were collected and incubated with 250 nM laurdan or di-4-ANEPPDHQ for 30 min at room temperature and then transferred to BSA-coated 8-well glass bottom Ibidi chambers.

Laurdan or di-4-ANEPPDHQ labelled GPMVs were imaged using the spectral imaging mode of a Zeiss 780 microscope equipped with a 40×, 1.2NA objective as reported previously in detail [28].

Spectra of laurdan and di-4-ANEPPDHQ in different solvents were measured in a quartz 96-well plate (Hellma). 385 nm and 488 nm were used for excitation of laurdan and di-4-ANEPPDHQ, respectively. Emission was collected between 400 and 700 nm for laurdan and 500–700 nm for di-4-ANEPPDHQ.

Chloroform solutions of POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine) and cholesterol were combined in the appropriate amounts. The organic solvents were then evaporated under a stream of nitrogen. For thorough removal of the solvent, the lipid films were left under vacuum overnight. HEPES buffer (10 mM HEPES, 150 mM NaCl, pH 7.0, 0.2 mM EDTA) was then added to the dried lipid film (lipid concentration of 1 mM), which was left to hydrate for 30 min. The resulting suspension was vortexed for at least 4 min and then extruded through polycarbonate membranes with a nominal pore diameter of 100 nm (Avestin, Ottawa, Canada). Laurdan (methanol solution) was added to chloroform/lipids solution before the evaporation under nitrogen. Di-4-ANEPPDHQ (aqueous solution) was added at the end of the hydration stage. The molar ratio of fluorescent probes to lipids was 1:100. For measurements, the vesicle suspensions were diluted to an overall lipid concentration of 0.5 mM.

The temperature in the cuvette holders was maintained using a water-circulating bath. Steady-state excitation and emission spectra were acquired using a Fluorolog-3 spectrofluorometer (model FL3-11; Jobin Yvon Inc., Edison, NJ) equipped with a xenon arc lamp. The steady-state spectra were recorded in steps of 1 nm (bandwidths of 1.2 nm were chosen for both the excitation and emission monochromators) in triplicate and averaged. Fluorescence decays were recorded on a 5000 U single-photon counting setup using a cooled Hamamatsu R3809U-50 microchannel plate photomultiplier (IBH, Glasgow, UK) and either a NanoLED 11 laser diode (375 nm peak wavelength, 1 MHz repetition rate) or a PicoQuant pulsed diode laser (470 nm peak wavelength, 2.5 MHz maximum repetition rate), for measurements with laurdan and di-4-ANEPPDHQ respectively. A 399 nm, or 499 nm, cut-off filter was used to eliminate scattered light, for measurements with laurdan and di-4-ANEPPDHQ respectively. The signal was kept below 1% of the repetition rate of the light source. Data were collected until the peak value reached 5000 counts. The full width at half maximum (FWHM) of the instrument response function was 78 ps and 84 ps, for measurements with laurdan and di-4-ANEPPDHQ respectively.

Fluorescence emission decays were recorded at a series of wavelengths spanning the steady-state emission spectrum ((400–550 nm) in steps of 10 nm for Laurdan and (540–700) nm in steps of 14 nm for di-4-ANEPPDHQ). The fluorescence decays were fitted to a multi-exponential function via the deconvolution method using the IBH DAS6 software. Three exponential components were necessary to obtain a satisfactory fit of the data. The purpose of the fit is to deconvolve the instrumental response from the data, and should not be over-parameterized. The fitted decays together with the steady-state emission spectrum were used for the reconstruction of time-resolved emission spectra (TRES) by a spectral reconstruction method [29]. The reconstruction routine was implemented in Matlab. Measurements were done at 23 °C (room temperature) and 37 °C (physiological temperature).

The generalized polarization spectra of laurdan (GP_laurdan) was calculated according to [26]: $\text{GP}=\frac{{{I}_{440}}-{{I}_{490}}}{{{I}_{440}}+{{I}_{490}}}$, where I₄₄₀ and I₄₉₀ represent the fluorescence intensity emitted at 440 nm and 490 nm, respectively. The generalized polarization spectra of di-4-ANEPPDHQ (GP_di-4) was calculated according to refs [24, 30]: $\text{GP}=\frac{{{I}_{560}}-{{I}_{650}}}{{{I}_{560}}+{{I}_{650}}},~$ where I₅₆₀ and I₆₅₀ represent the fluorescence intensity emitted at 560 nm and 650 nm, respectively.

We first questioned how well both probes empirically distinguish differences in membrane order in phase-separated model membranes, specifically in cell-derived giant plasma membrane vesicles (GPMVs). We performed spectral imaging where fluorescence images of the equatorial plane of the vesicles were recorded in  ≈10 nm wide wavelength intervals, i.e. a fluorescence emission spectrum was acquired for each image pixel with 10 nm spectral resolution (figure 1(a)) and separate spectra in the ordered and disordered phases could be obtained (figures 1(b) and (c)). Laurdan displayed a higher intensity in the liquid ordered phases at all wavelengths, while that of di-4-ANEPPDHQ was stronger in the liquid disordered phases (figure 1(a)). Both dyes exhibited a significant spectral shift in fluorescence emission between the ordered and disordered environment (figures 1(b) and (c)), from which the generalized polarization index (GP_laurdan and GP_di-4) was calculated (figure 1(d)). The final GP map (i.e. the GP values determined for each image pixel) of the vesicles showed an efficient distinction between ordered and disordered phases for both probes. We conclude that both probes are empirically capable of distinguishing ordered versus disordered membrane environments in cellular model membranes.

Zoom In Zoom Out Reset image size

In order to investigate the pure influence of polarity on the emission characteristics of the dyes, we characterized their fluorescence emission spectra in solvents with varying polarity. Unfortunately, there is no universal polarity index for solvents. As an example, we selected three solvents with significantly different polarity, namely ethanol, DMSO and chloroform. These solvents have different polarity values according to different polarity indices (see table 1), namely the dielectric constant and the Dimroth and Reichard polarity scales. Chloroform is considered the most non-polar solvent in both scales. However, the relative position of DMSO and ethanol depends on which index is considered. DMSO is more polar than ethanol in terms of dielectric constant, while it is less polar than ethanol in the Dimroth and Reichard polarity index (table 1). The reason for this difference lies in the definition of these polarity scales. The Dimroth and Reichard polarity index is based on the measure of the ionizing power of a solvent; those with ability to form hydrogen bonds have higher solvation capacity, i.e. stabilizing interactions between the solute and the solvent. On the other hand, the dielectric constant is a relative measure of the chemical polarity of a solvent, or its polarizability. Therefore, the polarity scaling according to the dielectric constant does not consider the possibility of hydrogen bonding between solvent and solute.

The spectra of laurdan and di-4-ANEPPDHQ in the three different solvents exhibit different solvatochromic behaviour. While the spectrum of Laurdan was red-shifted with the emission maxima λ in the order λ_Ethanol > λ_DMSO > λ_Chloroform, the spectra of di-4-ANEPPDHQ followed the order λ_DMSO > λ_Eth > λ_Chloroform (figure 2). This shows that the photophysical behaviour of the two probes is not identical. The characteristics of laurdan follow what is expected for a dye obeying the Dimroth and Reichardt index, while that of di-4-ANEPPDHQ is in line with the polarity scale based on the dielectric constants. These different behaviours reflect the ability of laurdan to form hydrogen bonds with the solvent, which does not seem to occur with di-4-ANEPPDHQ. Thus, the solvatochromic behaviour of laurdan follows the scale that accounts for the hydrogen bonding effect, and the solvatochromic behaviour of di-4-ANEPPDHQ correlates simply with the dielectric constants of the solvents. This differential behaviour of Laurdan and di-4-ANEPPDHQ points out that these probes may react differently on different membrane environments.

Zoom In Zoom Out Reset image size

Another well-defined system, large unilamellar vesicles (LUVs), was used to investigate the dependence of the fluorescence emission spectra of laurdan and di-4-ANEPPDHQ dyes on membrane lipid packing. For this, we acquired fluorescence emission spectra of these dyes in LUVs composed of POPC or POPC  +  10% cholesterol (POPC/Chol) at 23 °C and 37 °C. From these spectra we determined dye-dependent GP values; GP_laurdan and GP_di-4 (table 2). As a reminder, the more negative the GP values the more disordered or less packed the bilayer (see experimental procedures). The increase in membrane fluidity caused by the increase in temperature is clearly measurable with laurdan. GP_laurdan value is a very sensitive fluidity indicator in both POPC and POPC/Chol bilayers. In contrast, the GP_di-4 values of di-4-ANEPPDHQ hardly decrease with temperature in both POPC and POPC/Chol LUVs. Consequently, GP_di-4 seems to be a less sensitive indicator of changes in fluidity or membrane order. For example, GP_laurdan values in POPC/Chol bilayers change from 0.026 at 23 °C to  −0.254 at 37 °C, i.e. ΔGP_laurdan  =  0.280, while the GP_di-4 values vary by only ΔGP_di-4  =  0.045. This is a surprising observation, since the lipid packing of the POPC/Chol bilayer is known to be very different between 23 °C and 37 °C [35].

The observations at different temperatures clearly indicate that the steady-state emission spectrum of di-4-ANEPPDHQ does not report efficiently on the bilayer's fluidity or lipid packing. However, GP_di-4 is very sensitive to cholesterol, increasing by ΔGP_di-4  =  0.187 (at 23 °C) and 0.175 (at 37 °C) between POPC and POPC/Chol bilayers, even a larger increase than for laurdan (ΔGP_laurdan  =  0.151 (at 23 °C) and 0.106 (at 37 °C)). The sensitivity to cholesterol but insensitivity to temperature suggests that the change in GP_di-4 is caused by a specific effect of cholesterol [34], rather than a consequence of the known cholesterol-induced increase in lipid packing [36–40].

In order to investigate in detail the dipolar relaxation properties of di-4-ANEPPDHQ and laurdan, we performed time-dependent fluorescence shift (TDFS) measurements of the dyes in the same LUV systems. It is well known that values of GP_laurdan are affected by both the lipid packing and the hydration level of the membrane bilayer. Nonetheless, TDFS measurements have shown that GP_laurdan reflects predominantly the mobility of the hydrated sn  −  1 carbonyls (i.e. lipid order) and not the extent of hydration of a lipid bilayer in the liquid crystalline phase [31]. Yet, such correlation has not been investigated for di-4-ANEPPDHQ.

TDFS experiments are based on the ultrafast change in the dipole moment of a fluorophore upon electronic excitation, to which its solvation shell must respond. This dipolar relaxation causes a time(t)-dependent shift of the peak maximum ν(t) of the emission spectrum, i.e. TDFS probes the time-resolved emission spectra (TRES). The analysis of ν(t) uniquely reveals independent information on the polarity (hydration) and on the molecular mobility in the immediate environment of the dye. The total amount of fluorescence shift Δν is proportional to polarity [29], and is calculated as:

$$\begin{eqnarray} &&\Delta \nu ={{\nu}_{(0)}}-{{\nu}_{\left(\infty \right)}}\end{eqnarray} \tag{ 1 }$$

where ν₍₀₎ is the position of TRES maximum at t  =  0 estimated using the method of Fee and Maroncelli [41] and ν_(∞) is the position of the TRES at the fully relaxed state. The TDFS kinetics depend on the dynamics [42] of the polar moieties in the vicinity of the probe and can be expressed as the integrated relaxation time

$$\begin{eqnarray}&&{{\tau}_{\text{r}}}={\int}_{0}^{\infty}\frac{{{\nu}_{(t)}}-{{\nu}_{\left(\infty \right)}}}{ \Delta \nu}\text{d}t.\end{eqnarray} \tag{ 2 }$$

In the particular case of di-4-ANEPPDHQ, we were unable to estimate ν₍₀₎ since this dye is insoluble in non-polar solvents. Therefore, it is not possible to calculate the total amount of fluorescence shift, Δν, or the integrated relaxation time, τ_r. The latter problem can be circumvented by inspecting the spectral half widths (full width at half-maximum, FWHM) of the reconstructed TRES. The evolution of the FWHM values with time provides useful information on the observed dipolar relaxation phenomenon. It has been shown that the spectral half width, i.e. the FWHM value passes through a maximum during the solvation process [43–45], which is in line with a non-uniform distribution of solvent response times [45, 46]. The solvent shells of individual fluorophores distributed in a spatially heterogeneous system respond to changes in the local electric fields at different rates. This creates a transient distribution of phases of the individual dipolar relaxation phenomena. The heterogeneity increases significantly during the relaxation process and then decreases once the fluorophores reach the equilibrated excited state. The FWHM of the TRES gives, therefore, a measure of the heterogeneity of the microenvironment of the dye and shows if the entire dipolar relaxation phenomenon was captured within the time-frame of the measurement [29, 43, 47]. Moreover, the time at which the FWHM time-evolution reaches its maximum is a good estimate of the average time taken to complete the dipolar relaxation process and therefore can be used in place of the integrated relaxation time, τ_r.

The values of the fluorescence shift Δν did not significantly shift for laurdan in the POPC and POPC/Chol LUVs at both temperatures, 23 and 37 °C (figure 3(a), table 3). This implies that the polarity probed by laurdan is identical for the two different lipid bilayers, at both temperatures. However, the relative temporal kinetics are significantly different, e.g. revealed by the temporal development of the maximum or the spectral half width FWHM of the TRES (figure 3(b) and table 3). Table 3 lists the time at which the values of the FWHM of the TRES of laurdan reach their maximum. These values correlate well with the values of GP_laurdan (compare table 2). It becomes obvious that: (a) the average time for the completion of the relaxation process is shorter at the higher (37 °C) temperatures (compared to 23 °C), i.e. there is a higher mobility of the polar moieties in the vicinity of the probe and the lipid packing is decreased at higher temperature; (b) addition of cholesterol slows down the process of dipolar relaxation, i.e. there is an increased rigidity in the vicinity of the probe and higher packing of lipids in the cholesterol containing bilayers. These data highlight that, while TDFS measurements are more informative than GP values, the values of GP_laurdan are still a very good indicator of the order of a lipid bilayer in the liquid crystalline phase.

Zoom In Zoom Out Reset image size

Figure 4 shows the time dependence of the TRES maximum for di-4-ANEPPDHQ in the POPC and POPC/Chol bilayers at both 23 °C and 37 °C. Due to reasons outlined before, it is not possible to estimate the Δν for this dye. One remedy to this limitation may be to correlate the differences in the energy of the equilibrated excited state, i.e. the asymptote ν_(∞) of the TRES (figure 4) to the Δν, which would for example indicate that di-4-ANEPPDHQ shows significant differences in hydration with temperature (table 4). However, the energy ν₍₀₎ of the Frank-Condon state is not the same in all cases. In fact, constant values of ν₍₀₎ in all cases would contradict the results on hydration of laurdan, as revealed by the TDFS experiments. Moreover, from both the solvatochromic behaviour of laurdan and di-4-ANEPPDHQ (table 1) and from the inability of di-4-ANEPPDHQ to form hydrogen bonds one expects that di-4-ANEPPDHQ has a lower sensitivity to changes in polarity compared to laurdan. Please note that the presence of cholesterol shifts the TRES of di-4-ANEPPDHQ to higher energies, which is consistent with previously reported steady-state fluorescence data of di-4-ANEPPDHQ [34].

Zoom In Zoom Out Reset image size

The time-evolution of the FWHM values of the TRES for di-4-ANEPPDHQ exhibits complex photophysical characteristics of the dye at all conditions (figure 5). Instead of the usual single maximum as observed for laurdan, multiple maxima are detected, which suggests the existence of several underlying processes; (1) a very fast process occurs at times  <0.1 ns which is hardly resolved by our instrument due to missing temporal resolution. Still, our data indicates no significant difference in the fast dynamics between the different LUV samples. Such a fast kinetic is likely to be due to an intramolecular process rather than an effect of the environment of the dye. (2) The presence of two additional maxima at longer times suggests an additional process due to the dipolar relaxation and/or different locations of the dye within the lipid bilayer.

Zoom In Zoom Out Reset image size

The maxima at the longest time are most sensitive to temperature, shifting for example for POPC/Chol LUVs from 4.08  ±  0.05 ns at 23 °C to 2.00  ±  0.05 ns at 37 °C (table 4). In fact, the shift with temperature is stronger for POPC/Chol compared to POPC LUVs, most probably since the lipid order changes more dramatically for POPC/Chol. In contrast, the peak at around 1 ns hardly changes with temperature in POPC/Chol and POPC LUVs. These observations suggest that the process that causes the slowest modulation of the FHWM (at around 5 ns), likely stems from the dipolar relaxation of the dye. Interestingly, while temperature significantly affects the kinetics of this process, it hardly affects the values of GP_di-4 (compare table 2). This is contradictory, since changes in the relaxation kinetics do reflect changes in membrane lipid order for laurdan; yet we can conclude that the GP_di-4 parameter is not an accurate enough indicator of the order of a lipid bilayer in the liquid crystalline phase.

The addition of cholesterol to the lipid bilayers causes an unexpected effect in the kinetics of the relaxation process of di-4-ANEPPDHQ. Instead of slowing down the dynamics of the observed processes (due to the known increase in lipid packing with cholesterol addition), we observed faster relaxation dynamics. The fastest process at  <0.1 ns seems unaffected (at least as far as TRES can resolve, figure 5). However, the other two maxima are both shifted towards shorter times (figure 5 and table 4). The faster kinetics may be explained by a different positioning of di-4-ANEPPDHQ in the presence of cholesterol in the bilayer. If, for instance, the dye is pushed slightly out of the membrane upon addition of cholesterol, this would increase the mobility of its immediate environment [48]. Yet, the faster relaxation kinetics (i.e. obviously decreased lipid packing) of di-4-ANEPPDHQ upon cholesterol addition is in contrast with the increase in GP_di-4 values. Most probably, the observed increase in GP_di-4 values is related to a specific influence of cholesterol on the energy levels of the Frank–Condon (ν₍₀₎) and fully relaxed (ν_(∞)) states of di-4-ANEPPDHQ.

It is worth noting that di-4-ANEPPDHQ belongs to a family of dyes with known electrochromic characteristics [49, 50]. These electrochromic characteristics introduce an enhanced sensitivity to membrane potential, i.e. a change in the voltage or electric field across the membrane will cause a spectral shift in the fluorescence emission, which results from a direct interaction between the electric field and the dipole moments of the dye's ground and excited states. This electrochromic shift is present in both the absorption and emission spectra [51]. Cholesterol is known to either directly or indirectly (via its induced change in lipid packing) increase the internal electrical dipole potential of a lipid membrane bilayer by changing the density of dipoles and the dielectric constant of the membrane interface region [52–54].