Fluorescence anisotropy (FA) of anionic dyes bound to ionic and zwitterionic micelles

Abstract

Anionic fluorescein and 8-hydroxy-1,3,6-pyrenetrisulfonate (POH), bind to cationic and zwitterionic micelles, experience hindered rotation and exhibit fluorescence anisotropy (FA). Fluorescein emits three lines from its dianion, carboxylate, and phenolate forms. POH emissions are from excited POH* and its deprotonated form, PO⁻. Fluorescence was excited by vertically polarized (V) light. Spectra recorded with vertical (IVV) and horizontal (IVH) polarizers in the emitted beam were corrected for instrument response and polarization bias. Corrected line shapes were fit to Gaussians availing the computationally derived second harmonic for better fit precision. FA of the individual forms of the same dye was calculated from the IVV and IVH intensities of each component line. For fluorescein, FA phenolate > carboxylate > dianion and FA PO⁻> POH*. Micelle-bound dye conformations, consistent with this order, are presented. Distinguishing between FA of different forms is novel and significant to elucidation of dye-host interactions.

1. Introduction

The phenomenon of Fluorescence Polarization (FP) has been known to exist since the early 19th century. Polarization of fluorescence from a dye molecule excited with polarized light, informs on properties of the molecule itself and interactions with its environment. Properties that cause fluorescence polarization include molecular size, binding to other large molecules like proteins, and DNA, folded states of proteins, and viscosity of the medium. This capability has led to the development of FP as a diagnostic tool in biology, biochemistry and clinical science [1]. A review by Jameson and Ross traces the origins of applications of FP to early twentieth century [2].

A fluorescent dye molecule or fluorophore, when excited by linearly polarized light, will emit fluorescence polarized in the same direction as the incident excitation if it does not reorient by tumbling or rotation during the lifetime of the excited state. However, if the rotational correlation time, τ_c, is shorter than the excited state lifetime, τ_f, then the molecule tumbles several times while emitting, resulting in depolarization of the emission [3]. Rotational correlation time increases with medium viscosity or local dye neighborhood viscosity, molecular size, fluorophore binding to a large molecule such as a protein or molecular aggregates such as micelles and vesicles [4]. Restriction to dye motion can simply be a size effect, viscosity of medium, or specific interactions that prevent rapid isotropic motion [1,5,6]. Existence of restricted dye motion in structured interfacial regions and in the neighborhoods of large molecules like proteins and DNA is well known. High FA of some probes in water speak to the formation of probe-solvent cage like structures that prevent free rotation [6]. Presence of FP therefore indicates slow rotational motion of the molecule on time scales longer than the fluorescence lifetime. Any medium and probe property and interaction that increases τ_c will increase FP.

Fluorescence anisotropy (FA), denoted by r and defined by [1,7]

$$r=\frac{IVV-G.IVH}{IVV+2G.IVH}$$ (1)

is a metric for FP. In a typical FA measurement, the dye molecule is excited with light polarized vertically (V). IVV and IVH are the intensities of emission with polarization parallel and perpendicular to the excitation polarization respectively. The quantity G is the geometric factor of the emission monochromator and detection system [4,7,8]. It accounts for the polarization bias of the emission monochromator grating and the detection system.

For vertically polarized excitation, IVV is greater than IVH when molecular motion is restricted to a rate that is lower than the fluorescence decay rate, resulting in non-zero values for r. FA experiments are usually carried out with a spectrometer equipped with polarizers in the excitation and emission beams. Emission intensities, IVV and IVH, are measured at a selected wavelength or a series of wavelengths and FA calculated according to eq.1 [9]. Measurement at any one wavelength or a band of wavelengths includes intensities from more than one elementary emission because of overlapping emissions from different coexisting emitting forms. FA determined in such a manner does not refer to any single emitting chemical form. Fluorescence spectra of organic dyes are rarely a single line. Spectra comprise overlapping lines from more than one excited state and excitation of more than one ground state form. Intensity at any wavelength is a composite of emitted intensities from different forms. This work introduces a novel methodology that recognizes the fact that dye molecules emit from more than one excited state, can exist in more than one ground state and that, each of these forms can exhibit different values of τ_c due to differing interactions with large molecular entities in solution. FA determined from these emitted intensities at some wavelength is not representative of any one form. While a bulk property like viscosity may affect motion of the different forms in a like manner, interactions of the various forms with large molecules can differ specifically and nonspecifically from each other. Rotational motion of the different forms can therefore be different. FA is thus form dependent. The present approach resolves the composite spectra of IVV and IVH into its components and calculates the FA of each individual line. This provides better insight into the presence of specific and non-specific interactions between each of the different molecular forms of the dye and the large molecules.

Interactions of anionic fluorescein (disodium salt) and anionic 8-hydroxy-1,3,6-pyrenetrisulfonate (POH) (trisodium salt) with cationic cetyltrimethylammonium chloride (CTAC), anionic sodium dodecyl sulfate (SDS), and the two zwitterionic 3-(N-Hexadecyl-N,N-dimethylammonium)-propanesulfonate (HPS), and lysophosphatidylcholine (LPC) micelles, were investigated by measurements of FA exhibited by fluorescein and POH in these micellar solutions. Anionic fluorescein exists in three forms: dianion, monoanionic carboxylate, and phenolate and emits three lines. POH emits four lines. The two shorter wavelength emissions are from the POH excited state (POH*) to POH ground state transition. The two longer wavelength lines are from PO⁻ formed by a proton transfer from the excited state, POH*, to water resulting in PO⁻ and hydronium ion [10–12]. IVV and IVH spectra of the dye-micelle solution were each resolved into their components using fitting methods that involves the computationally derived second harmonic (SH) of the measured zeroth order spectra in determination of the best fit [13,14]. The advantages of SH are in: (i) identifying the number of peaks present in a composite emission, so that the measured spectra (referred to as zeroth order spectra) can be fit with that number of peaks. The number of peaks is not often obvious in the zeroth order. (ii) gauging with confidence the precision of the fit from the residuals of the computationally derived SH of the measured data and SH of the fit. The FA of the individual emission was determined using a modified but equivalent form of eq. 1. The G-factor was determined experimentally.

Clear differences observed between the FA values of the various dye forms and the methodology itself to determine FA of the individual forms are new contributions from this work. The phenolate form of fluorescein has higher anisotropy than the dianion and carboxylate. PO⁻ fluorescence exhibits greater anisotropy than POH*. FA values also depend on the type of micelle headgroup.

2. Materials and methods.

2.1. Materials

The water-soluble dyes anionic fluorescein and 8-hydroxy-1,3,6-pyrenetrisulfonate (POH) and the surfactants: Cetyltrimethylammoniumchloride (CTAC); 3-(N,N-Dimethylpalmitylammonio)propanesulfonate (HPS); sodiumdodecylsulfate (SDS) were from Sigma. The lipid, 1-pal-mitoyl-2-hydroxy-sn-glycero-3-phosphocholine (LPC) was from Avanti Lipids. Fig. 1 show the chemical structures of the dyes. The two ring systems in fluorescein are closer to orthogonal than planar ((http://www.chemspider.com/Chemical-Structure.10163.html) and the rings in POH are planar (http://www.chemspider.com/Chemical-Structure.55317.html)

Fig. 1.

Chemical structures of the dyes: (a) three forms of anionic fluorescein showing the two orthogonal ring systems. (b) 8-Hydroxypyrene-1,3,6-trisulfonic acid trisodium salt (POH).

Fig. 2 presents the lipid and non-lipid surfactant structures. The headgroups are shown in the orientation they adopt in the interface when in micellar form. These materials differ in their headgroup charge distributions and orientation of the headgroup. The CTA⁺ headgroup in CTAC renders a positively charged CTAC micelle surface; SDS is negatively charged due to the sulfate headgroup; HPS and LPC are zwitterionic. The negatively charged sulfonate of HPS headgroup is the outer group and the positive moiety −(CH₂)₃N⁺(CH₃)₂ is the inner group. Computer simulations show that the positive choline −CH₂N⁺(CH₃)₃ group in the zwitterionic LPC headgroup bends inward toward the inner surface of the micellar interface exposing the negatively charged phosphate $\left({\text{PO}}_4^{-}\right)$ to the surface [11,15]. Molecular dynamics simulations show this to be the configuration in phosphatidylcholine bilayer vesicles as well [16,17].

Fig. 2.

Chemical structures of surfactants. The structures are per the conformation adopted in the micelle.

2.2. Sample preparation

Samples were prepared by dissolving the surfactants in fresh distilled water to yield a required concentration. The dye was added to the solution to a concentration of 5 μM. Another solution of the dye in water at the same concentration was prepared. The two solutions were mixed to yield a dilution series of different surfactant concentrations while keeping the same dye concentration. Samples were also made with the same dye to surfactant ratio. The solution pH of the samples was about 6.3. Three ground state anionic forms: dianion, monoanions carboxylate and phenolate, with measurable fluorescence emission, exist at this pH and at pH between about 5.5 and 8 [18]. Experimental conditions were designed to observe fluorescence, with good signal to noise, from the three forms. A pH between 6 and 7 was found to be ideal for the present goal

2.3. Fluorescence experiments and corrections

Fluorescein was excited at 460 nm and POH at 400 nm. A Fluouromax-4 Spectrofluorometer (Horiba Scientific), equipped with polarizers for anisotropy measurements, recorded fluorescein emission spectra from 465 to 700 nm and POH spectra from 406 to 700 nm, at a step size of 1 nm and a bandpass slit width of 1 nm.

Measurements were conducted at an ambient temperature of 23 °C with vertically polarized excitation and with vertical (for IVV) and horizontal (for IVH) polarizers in the emission beam and with no polarizers. The instrument has the automated capability of swiveling between the parallel and perpendicular settings of the emission beam polarizer, reading the emission intensity at each setting, and calculating and outputting FA values at any selected wavelength. This value of FA however, as mentioned in the introduction, is not exclusive to any one single form of the dye because of contributions to the emission intensity from all forms. Herein lies the advantage of steady state spectra, where the emission can be resolved into individual components and emission intensities of the individual forms computed.

The polarizers were aligned using a dilute solution of glycogen in the sample compartment according to the instructions in the instrument manual. Instrument response correction was applied through the built-in software. Blank subtraction was performed with background spectra measured on solutions without the dyes for every polarization configuration and concentration. The blank subtractions provided a flat base line that aided better fitting.

Instrument polarization bias corrections were performed using fluorescein and POH emission spectra in water for obtaining the line shape and intensity correction parameter with the rationale that these probes in water exhibit low to no FA because the rotational correlation time in water is quite shorter than the fluorescence lifetime [8,11]}. Fluorescence lifetimes of fluorescein in water is about 4 ns and the rotational correlation time is about 110 ps [8,10,19]. Measured FA of fluorescein in water is 0.022 [2]. The no-polarizer, IVV, and IVH spectra in water should exhibit the same line shape if there were no polarization bias in the instrument. If a difference in line shape is present, then it indicates existence of wavelength dependent polarization bias. It is well-known that gratings do exhibit polarization bias and it is important to correct the spectra when conducting fluorescence polarization experiments [4,7,8]. The water spectra, with no polarizers in the excitation and emission beams, corrected for instrument response are taken as the true or control spectra. However, the IVV and IVH spectra as measured do not include the right instrument response correction because polarization bias is not accounted for. Relevant measurements and analyses in the methodology to obtain the corrections are illustrated in Fig. 3. The first step is to examine the normalized no-polarizer (NP), IVV, and IVH spectra. Normalized spectra for fluorescein are shown in Fig. 3a. The line shapes lack complete overlap. The next step is to determine the line shape correction factor for IVV, and IVH spectra by requiring their line shapes be the same as that of the no-polarizer control spectra. A tacit assumption in this condition is that the anisotropy in a homogeneous solvent like water does not depend on the form of fluorescein and is independent of wavelength. So, all the line shape difference is from wavelength dependent part of instrument polarization bias. The thus determined wavelength dependent line shape correction factor is used to correct the observed IVV and IVH line shapes. Fig. 3b presents the line shape corrected IVV and IVH. The correction at this point is only for line shape and not for the differences in sensitivity to the polarization state. The IVV and IVH intensity values in Fig. 3b are clearly different. Existing difference in IVV, and IVH after correction for line shape is due to the difference in signal strength response of grating and detector toward vertical and horizontal polarizations. Apparently IVH is greater than IVV. This indicates that the horizontal polarization is detected more efficiently than the vertical. IVH needs to be reduced by a factor, referred to as G-factor, to bring the line shape corrected IVV and IVH in water into quantitative agreement. This could simply be the line shape corrected intensity ratio IVV / IVH if the FA in water is taken as zero. For technical correctness the FA value of 0.022 in water was used in eq. 1 with line shape corrected values of IVV and IVH to determine the wavelength independent part of the polarization bias [2]. This is referred to as G in this work. The G-factor determined was 0.8.

Fig. 3.

Spectra of 8 μM fluorescein in water: (a) as measured no-polarizer (NP), IVV and IVH spectra, normalized to peak intensity = 1 before correction and (b) line shape corrected IVV and IVH.

The line shape correction factor obtained with fluorescein in water is for the fluorescein emission spectral region of 465 to 700 nm and does not include the shorter wavelength region of the POH fluorescence spectra that extends from 406 to 700 nm. Correction factors were also obtained for this wavelength region with POH in 18 % by volume of ethanol in water. The ethanol–water mixture was chosen because POH* emission is very weak in water but strong enough in ethanol–water to work with for a good signal to noise ratio. Fig. 4 a and b of normalized NP, IVV, and IVH spectra serve to illustrate the difference. Data of Fig. 4b were used to determine the wavelength dependent line shape correction factor that would bring IVV and IVH spectra into quantitative agreement with the no-polarizer spectra; just as was done for fluorescein in water.

Fig. 4.

Spectra of 5 μM POH as measured no-polarizer (NP), IVV and IVH spectra, normalized to peak intensity = 1 (a) in water where PO⁻ emission is quite stronger than that from POH*. (b) in 18 % by volume of ethanol in water with strong PO⁻ emission and reasonable POH* emission.

The line shape correction factor is an instrument property irrespective of probe-solvent system. In the wavelength region of 480 to 580 nm common to fluorescein and POH where there is sufficient emission signal the values of the correction factor determined with POH in 18 % ethanol–water, POH in water, and fluorescein in water differ by about 3 %.

The steps in the protocol for line shape correction and determination of FA of the dyes in surfactant solutions are:

1. Correct the as measured IVV and IVH spectra of the dyes in surfactant solutions using the wavelength dependent line shape correction obtained from spectra in water for fluorescein in micelles and POH in ethanol–water for POH in micelles.

2. Calculate IVVcorr and IVHcorr from IVV measured and IVH measured using

   $$\text{IVVcorr}=\text{line}\text{shape}\text{correction}\text{for}\text{IVV}\text{X}\text{IVV}\text{measured}$$ (2)

   $$\text{IVHcorr}=\text{G}\text{X}\text{line}\text{shape}\text{correction}\text{for}\text{IVH}\text{X}\text{IVH}\text{measured}$$ (3)

3. Fit the IVVcorr and IVHcorr composite spectra to Gaussians according to the procedure described below in Section 2.4 and obtain the IiVV and IiVH (i = 1,2,3 for fluorescein and i = 1,2,3,4 for POH) spectra of the resolved lines.

4. use the product of the fit values of peak intensity and line width to represent the total intensity in each of the resolved lines. Calculate the FA of each line using,

   $$r_i=\frac{I_{iVV}-I_{iVH}}{I{i}_{VV}+2I{i}_{VH}}$$ (4)

IiVV and IiVH are the total intensities of emission from each of the different forms of the dye. Equation 4, where G is subsumed in the corrected IVHcorr (eq. 3) and thereby in IiVH, is equivalent to eq. 1

2.4. Fitting

Wavelength (λ) spectra were converted to wavenumber (k = 1/λ) spectra by the Jacobian transformation, I(k) = λ² I(λ) [20,21]. SH of spectra were computed first as described in previous work [13,14,22]. SH is the computationally derived second harmonic by convolution of the measured spectra with the second harmonic response function. For any variation y(x) of a dependent signal variable y with an independent variable x, the first harmonic, Y₁, of y(x) may be experimentally obtained by modulating the x variable and measuring the change Δy in y. The second harmonic (SH), Y₂, is then given by the change in Δy. Higher harmonics are useful in discerning underlying subtle variations in y(x) that are not visible in the original y(x). It is not always possible to modulate the x variable experimentally. Computational methods may then be employed to determine Y₂ from y(x). It is given by a convolution of a response function R and y(x)

$$Y_2(x)=R^{*}y(x),$$ (5)

where * denotes convolution. The functional form of R derived theoretically and verified experimentally for transitions between spin states observed in electron spin resonance experiments is given by [23],

$$R=-\sum _{n=-\infty }^{\infty }\gamma J_n(\beta )(J_{n-2}(\beta )+J_{n+2}(\beta ))$$ (6)

where J_n(β) is the Bessel function of order n, β = γB_m/ω_m is the unitless index of modulation, and the gyromagnetic ratio, γ, is 1.76 * 10⁷ (sG)⁻¹ [24,25].

The form, eq.6, was found to work for any signal line shape y(x). This was tested by applying eq. 5 and 6 to a simulated signal y(x) that was a sum of two closely spaced Gaussians. The number of terms in the series in R was 50. The quantity γ is a constant and β = γ x_m/ω_m, where x_m is the modulation amplitude of x and ω_m is the modulation frequency. γ and β are adjustable parameters. Too high a modulation amplitude can broaden Y₂(k) and smear the overlapped peaks. The best parameters are obtained by trial and error. The SH derived computationally and analytically were found to match closely. The methodology was applied to determine SH of the experimental emission signal y(x) ≡ I(k) computationally because the variable k cannot be modulated experimentally. Eq. 5 was employed with R as in eq. 6 to obtain the SH, Y₂(k). The required convolution was performed in MATLAB using the function CONVN.

The SH indicated three peaks for fluorescein and four for POH. Therefore, fluorescein spectra were fit to three and POH spectra to four Gaussians as a function of the wavenumber k, each in the form,

$$I(k)=I_m\text{exp}\{-0.693{(\frac{k-k_m}{\mathit{\text{HWHM}}})}^{2}6$$ (5)

where I_m, k_m and HWHM are the peak height, peak position, and the half width at half maximum, respectively. After fitting, the SH of the measured zeroth order spectra, SH of the fit and their residual were examined to ascertain the quality of the fit. The very nature of the greater sensitivity of SH to noise makes it more stringent than the zeroth order spectra in confirming the quality of fit.

3. Results

Graphs of spectra are presented first, followed by numerical values of FA in Table 1. Fluorescence intensities were linear with dye concentrations up to 8 μM. Fig. 5a and 6a present corrected IVV and IVH spectra, denoted by IVVcorr and IVHcorr, their fits and residuals, and their resolved lines for 5 μM fluorescein respectively in 25 mM CTAC and in 25 mM SDS. Fig. 5b and 6b are the computationally derived SH spectra of IVHcorr and IVHfit. The SH confirms the presence of three peaks, indicated by 1,2,3 in Fig. 5b. The measured zeroth order spectra in Fig. 5a or 6a suggest two peaks but do not immediately reveal an additional peak. Fluorescein emission spectra were fit to three lines. The component lines, given by fitting, are also shown in Fig. 5a and 6a. The strongest peak from fluorescein at ≈19147 cm⁻¹ (515 nm) is from the dianion, the middle peak at ≈18587 cm⁻¹ (538 nm) from the carboxylate, and the phenolate emission at ≈17391 cm⁻¹ (575 nm) is the weakest. The overlap of the SH of measured and corrected spectra and fit and the SH residual provide assurance of the quality of fit.

Table 1.

Sample and Concentration Averages and Standard Deviations of FA. rPOH* and rPO⁻ denote the FA of POH* and PO⁻ respectively.

Fluorescein in	r1 Dianion	r2 Carboxylate	r3 Phenolate
CTAC	0.12 ± 0.006	0.11 ± 0.003	0.15 ± 0.01
HPS	0.16 ± 0.009	0.18 ± 0.013	0.20 ± 0.032
LPC	0.105 ± 0.026	0.114 ± 0.022	0.15 ± 0.019
SDS	0.048 ± 0.006	0.054 ± 0.007	0.052 ± 0.016
POH in	rPOH*	rPO −
CTAC	0.027 ± 0.0024	0.094 ± 0.003
HPS	0.063 ± 0.004	0.145 ± 0.009

Fig. 5.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM fluorescein emission in 25 mM CTAC. (b) Computationally derived second harmonics of corrected spectra (SH IVH) and fit (SH IVH fit) with numbers indicating regions of the three lines, and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

Fig. 6.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM fluorescein emission in 25 mM SDS. (b) Computationally derived second harmonics (SH) of corrected spectra (SH IVH) and fit (SH IVH fit), and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

The total IVH and the individual component line intensities, denoted by I1VH, I2VH, and I3VH are less than the total IVV and the corresponding I1VV, I2VV and I3VV. Fig. 5c of only the resolved lines provides better clarity in the relative intensities of the VV and VH lines and presence of FA. Each of the Ii (i = 1, 2, 3)VV is normalized to peak height = 1 and the Ii (i = 1, 2, 3)VH are each normalized by the same factor as their respective Ii (i = 1, 2, 3) VV. The Ii (i = 1, 2,3)VH are less than their respective Ii (i = 1, 2, 3)VV, signaling presence of FA. The purpose of the figure is to illustrate the difference in the amount of decrease of each of the Ii(i = 1, 2, 3)VH from the respective Ii(i = 1, 2, 3) VV. I3VH is less than I3VV by a greater fraction than the fractional decrease of I1VH from I1VV and I2VH from I2VV. This means that the phenolate form responsible for I3 has a higher FA than the other two forms.

Fig. 5b illustrates the value of SH in fitting. The residual quantifies the agreement between the SH of data and SH of fit. Notably in Fig. 5c, I3VH is reduced from I3VV by a greater factor than I2VH from I2VV or I1VH from I1VV. The display in Fig. 5c thus exemplifies the main result of this work, namely that FA differs from form to form.

Comparison of Fig. 5 with Fig. 6 shows the FA in SDS is weaker than in CTAC, but still observable. IVH intensities in SDS are less reduced from IVV than in CTAC, signaling a smaller FA than in CTAC.

Figs. 7 and 8 present the results for fluorescein in zwitterionic HPS and LPC micelles respectively. FA in the zwitterionic micelles appear to be greater than in the ionic micelles, from a visual comparison of Figs. 5c and 6c to Figs. 8c and 9c.

Fig. 7.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM fluorescein emission in 25 mM HPS. (b) Computationally derived second harmonics (SH) of corrected spectra (SH IVH) and fit (SH IVH fit) and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

Fig. 8.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM fluorescein emission in 6.25 mM LPC. (b) Computationally derived second harmonics (SH) of corrected spectra (SH IVH) and fit (SH IVH fit) and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

Fig. 9.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM POH emission in 25 mM HPS. (b) Computationally derived second harmonics (SH) of corrected spectra (SH IVH) and fit (SH IVH fit) with numbers indicating regions of the four lines, and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

POH spectra in HPS and CTAC, shown in Fig. 9a and 11a, were best fit to four lines. The SH indicated four peaks (Fig. 9b), not obvious in the zeroth order spectra. PO⁻ as well as POH* emissions are present in micelles. The two peaks at about k = 18832 cm⁻¹ (λ ≈ 531 nm) are from the PO⁻ to POH transition and the two at about k = 23100 cm⁻¹ (λ = 433 nm) are due to POH* deexcitation to POH [10,11]. The presence of the higher k peak (weak to absent in water) is evidence that POH is sequestered with the micelle. Dissociation of POH* in the water-poor micelle interface is not as efficient as in water, allowing for the presence of both PO⁻ as well as POH*. PO⁻ emission is dominant in water, because of the near complete deprotonation of the excited POH*. The resolved lines of IVVcorr and IVHcorr are included in Figs. 9a and 10a.

Fig. 11.

FA of fluorescein vs surfactant concentration. Lines are concentration and sample averages.

Fig. 10.

(a) The corrected spectra, fits, residuals and resolved lines of 5 μM POH emission in 25 mM CTAC. (b) Computationally derived second harmonics (SH) of corrected spectra (SH IVH) and fit (SH IVH fit) and residual. The residuals in a and b are shifted to below the spectra for clarity. (c) The resolved lines. The IiVV are each normalized to a peak height of unity and the corresponding IiVH are normalized by the same factor.

The observed reductions of IiVH from IiVV in Fig. 9c and 10c illustrate the effect of the finite FA of POH. PO⁻ exhibits greater FA than POH* as evidenced by the greater decrease in the areas of I3(4)VH from I3(4)VV than I1(2)VH from I1(2)VV.

Surfactant concentration dependences of FA are presented in Figs. 11 and 12. FA was calculated according to eq. 4 with the intensities calculated as areas (peak X line width) of each of the lines resolved from the corrected spectra. In the case of POH, the emission areas of peaks 1 and 2 of POH* and peaks 3 and 4 of PO⁻ were combined to calculate their respective FA. FA values are averages of measurements on three or more samples of each concentration.

Fig. 12.

FA of POH* and PO⁻ vs surfactant concentration. Lines are concentration and sample averages. PO⁻ exhibits a higher FA than POH*. There is no remarkable concentration dependence.

The quantities r1, r2, r3 in Fig. 11 denote the FA of dianion, carboxylate, and phenolate respectively. The FA, r3, of the phenolate form of fluorescein is higher than r1 and r2.

Table 1 presents the numerical values of FA averaged over repeated measurements on different samples and concentrations.

4. Discussion

FA values greater than that in water are indicative of dye association with micelles. Local viscosity in the dye neighborhood, binding of dye to large molecules or molecular aggregates, or binding that confers a preferred orientation are some reasons for restricted dye mobility that result in increased rotational correlation time and consequently FA. The main contributor to decrease of FA of dyes in micelles, from the theoretical limit of 0.4, is hindered rotation [26–28]. Micelle rotation and lateral diffusion of the probe in the interface are slow processes that also contribute to FA decay, but at a slower rate than probe wobbling or rotation [26–29]. Therefore, slow probe rotation is mainly responsible for the observed value of FA [26–28]. Fluorescein and POH, although water soluble, appear to behave as hydrophobic ions in the presence of micelles preferring to interact through their hydrophobic rings with the less polar interface and non-polar hydrocarbons of micelles. This interaction restricts rotation leading to fluorescence polarization. A higher FA in HPS, for either probe, than in ionic and zwitterionic micelles indicates an environment that is less water-like.

Significance of FA variation with the chemical form of the same dye is its potential to reveal the nature of specific interactions and the conformations of the reactants that could cause the variation. Electrostatic binding of probe to surfactant headgroup alone does not appear to be the cause of restricted probe motion, although it might play a major role in the present case of anioic fluorescein, as discussed later. FA of a cationic probe in micelles was found to be in the order: FA in non-ionic Triton > anionic SDS > cationic dodecytrimethylammoinum bromide (DTAB) [28]. A reasonable argument can be made based on electrostatics for FA of the cationic probe in anionic SDS > cationic DTAB; but not for FA in non-ionic Triton > anionic SDS. This and the present results of the order of FA of fluorescein: FA monoionic phenolate > FA monoionic carboxylate > FA dianion argue for role of other interactions in addition to electrostatics.

FA in SDS was not expected because the anionic dyes supposedly do not interact with anionic SDS. However, the values measured, albeit low, are greater than that in water and speak to the presence of interaction. Some interaction of fluorescein with SDS has been reported [30]. This is possibly a weak interaction due to the hydrophobic part of the dyes seeking the low water region close to the micelle interface. Absorbance and fluorescence were found to increase with [SDS]. A 1:1 SDS-fluorescein complex formation, driven mainly by a hydrophobic interaction, was proposed and the equilibrium constant for the complexation was determined. In AOT reverse micelles presence of anionic probes near the anionic interface within the water pool was noted [31]. For all the other surfactants: CTAC, HPS, and LPC, electrostatic binding between the anionic dyes and the cationic CTAC interface or the −(CH₂)₃N⁺(CH₃)₂ group of HPS or the choline −CH₂N⁺(CH₃)₃ group of LPC, the highly oriented structure of water in the interfacial region and the greater local microviscosity restrict dye mobility and the resulting FA is higher than in water. Location of the probes in the interface is a reasonable expectation based on the higher value of observed FA in the experiments and as proposed in several works on micelles and reverse micelles [11,32]. The zwitterionic headgroup in a single LPC molecule has the positively charged choline group on the outside. Computer simulations of LPC micelles and phosphatidylcholine (PC) membranes show however that the headgroup is not along the micelle or vesicle radial direction but bends inward toward the micelle core making an angle that is greater than 90° to the surface normal [11,15]. The (N⁺ (CH₃)₃) group bends toward the inner surface of the micelle or vesicle interface and the negative ${\text{PO}}_4^{-}$ group is exposed to the surface (see Fig. 2) [11,15]. Surface potential and charge of PC membranes were measured to be negative supporting such a conformation [33]. In HPS micelles, the −(CH₂)₃N⁺(CH₃)₂ to ${\text{SO}}_3^{-}$ bond is along the radial direction [11]. FA of the anionic dyes in this work confirm their access to the positively charged moieties of the headgroups in LPC and HPS within the interface and localized on the inner surface as modeled in earlier work [11]. A model for the incorporation of the three anionic forms of fluorescein in LPC and HPS micelles is proposed in Fig. 13.

Fig. 13.

(a) Distribution of the negative charges in each form of fluorescein along with a geometric model placed to the right of each form’s structure. (b) Scheme for anionic fluorescein in LPC, HPS, and CTAC. The numbers 1, 2, 3 refer to dianion, carboxylate, and phenolate respectively.

The negative charged single carboxylate ring of the carboxylate (form 2) and dianion (form 1) and the distributed charge in the three-ring system in phenolate (form 3) are each positioned near the positive charged nitrogen. The two systems (three planar rings and the phenylcarboxylate ring) of the fluorescein are nearly orthogonal to each other. Their orientation in the micelle, along with the charge distribution should impact on the conformation and location of the dye, tightness of binding and freedom of rotation in the micelle.

The three-ring anionic site of the dianion should prefer to be further away from the negative oxygen of the phosphate or sulfate. This would put this anionic site of the dianion in the encumbered neighborhood of the hydrocarbons. Charged species in non-polar regions might be expected to experience more wobbling than in polar locations. Dianion fluorescence (form 1) is thus subject to greater depolarization than that from the other two forms. Anionic parts of monoionic phenolate (form 3) and carboxylate (form 2) are near the positive nitrogen. The neutral rings of carboxylate and phenolate should bind to the hydrophobic sites of LPC and HPS. The charge distribution in phenolate is spread wider across the three rings than in carboxylate and this could cause phenolate to bind to more than one monomer in the micelle. Phenolate rotation, because of such two-site binding, would be hindered to a greater extent than the single site bound carboxylate. In CTAC, the dianion binds to the positive nitrogen through its negative charged single carboxylate just as in LPC and HPS. There is no negative charge in the CTAC micelle monomer, unlike in LPC and HPS, to repel the three-ring anionic site of the dianion, which then protrudes toward the micelle surface. In such a configuration, the dianion will experience less hindrance to rotation and thereby exhibit less FA than the other two forms. This model is consistent with the observation FA phenolate > FA carboxylate > FA dianion.

The positive nitrogen is on the inner surface of the interface in HPS. Preference of the anionic parts of fluorescein for proximity to nitrogen pushes the fluorescein deeper into the tighter core region which could be the reason for a longer rotational correlation time and higher FA in HPS over that in the other micelles.

Electrostatic and hydrophobic interactions are responsible for association of the dye with micelle. Once bound to the micelle, the dye experiences a lack of freedom in rotation or a wobbling motion. The nature and strength of binding and conformation depend on the dye chemical form, which this work shows are different for the different forms of the same dye. Ionic attraction between the N⁺ of the micelle monomers and the anionic charge sites of the dye appear to be mainly responsible for the conformation of the dye forms. The conformations are further facilitated by the hydrophobic attraction of the dye rings for the hydrophobic tails of the monomers in the cases of carboxylate and phenolate. Stronger the attraction, more hindered or less wobbly is the rotation and greater is the FA. The partial charge distribution in phenolate facilitates ionic interaction with more than one neighboring micelle monomer and thus phenolate’s ionic binding is stronger than that of carboxylate binding. The hydrophobic attraction of carboxylate and phenolate toward the micelle’s hydrophobic tails is similar for the two forms. FA of phenolate is higher than FA of carboxylate because of the stronger ionic binding of phenolate. The anionic rings of the dianion are in the micelle non-polar region because of ionic repulsion from the phosphate in LPC or sulfate in HPS. The charged nature of the rings in the dianion increases the wobble of this form because of the unfavorable non-polar neighborhood in the micelle, which then decreases the rotational correlation time leading to low FA. There is no negative charge in the CTAC headgroup to repel the dianion rings and this portion of the molecule prefers the interfacial water region where there may be greater rotational freedom than within the micelle, but less than in bulk water because of higher microviscosity and structured nature of interfacial water. The dianion in CTAC may then be expected to exhibit a lower FA than phenolate. It appears therefore that electrostatics plays a greater role in dye conformations and the resulting FA.

A stronger binding and less wobble for PO⁻ than for POH* is suggested by the greater FA of PO⁻. The four negative charges on PO⁻ localized in the four corners of the planar ring system in PO⁻ enables a stronger binding than the triple negative charge in POH*.

5. Summary and conclusions

Fluorescent dye molecules exist in more than one chemical form emitting multiple lines that overlap. This work introduces a new methodology to extract the fluorescence anisotropy of each of the forms. The different forms of the dyes embed in the interface region in different conformations because of which they exhibit differing fluorescence anisotropies. Existing methodologies are not able to discriminate between FA of the different forms of the same dye.

The present methodology involves (i) Determination of the wavelength dependent polarization bias of the instrument and spectral correction of the measured spectra for this bias; (ii) recruiting the computationally derived second harmonic of the corrected line shape to determine the number of lines as suggested by the number of peaks in the SH; (iii) fitting the corrected IVV and IVH spectra; (iv) examining fit of the SH of fitted line to the computationally derived SH of the data to better determine the best fit parameters; (v) obtaining FA of the dye form responsible for the individual emission line from their IVV and IVH intensities. This methodology was applied to fluorescein and POH emission in micelles of CTAC, LPC, HPS, and SDS.

HPS recorded the highest FA and values in SDS were low. FA of the phenolate form of fluorescein was greater than that of carboxylate and dianion. PO⁻ exhibited a higher FA than POH*. Finite values indicate the presence of strong interactions between the dye and the micelle-water headgroup interface and non-polar parts of the micelle that restrict dye mobility and increase rotational correlation times. A model consistent with the observed order of the FA values is proposed. The model in Fig. 13 predicts the type of binding and location of the dye forms that could result in hindered rotation or wobbling motion and finite FA that are such that FA phenolate > FA carboxylate > FA dianion. Presence of FA in HPS and LPC show that the dyes have access to the positive charge in the inner part of the zwitterionic interface. Conformation adopted by the dye form is dictated by electrostatic coupling as well as hydrophobic interactions of the non-polar rings with the non-polar hydrocarbons of the micelle. These interactions together account for the degree of wobbling of the dye in the micelle and order of rotational correlation times of the forms: τ_c phenolate > τ_c carboxylate > τ_c dianion and τ_c PO⁻ > τ_c POH.

Data availability

Data will be made available on request.