Fluorescent Anisotropy Evaluation of Bicelle Formation Employing Carboxyl BODIPY and Pyrromethene

Bicelles are extensively used as the parent assemblies of functional membrane materials. This study characterizes membrane fluidity in fatty acid/detergent bicelles containing carboxyl boron-dipyrromethene (BODIPY C12) and pyrromethene as fluorescent probe molecules. The anisotropy value of BODIPY C12 and pyrromethene in the phospholipid vesicles depended on the phase state of the vesicles. The anisotropy of the fluorescent probe molecules in bicelles of oleic acid/3-[(3-cholamidopropyl) dimethylammonio]-2-hydroxypropane sulfonate (OA/CHAPSO) was then evaluated. The OA/CHAPSO bicelles were prepared by mixing CHAPSO detergent solution with OA vesicles at different molar ratios, X OA (= [OA]/([OA]+[CHAPSO])). The anisotropies of the probes in the OA/CHAPSO bicelles increased with decreasing X OA. BODIPY C12 in the range 0.30 ≤ X OA ≤ 0.70 exhibited a distinctly larger anisotropy than pyrromethene. This result agreed with the increase in packing density associated with the adsorption of CHAPSO molecules on the OA bilayer membrane in the OA/CHAPSO bicelle, revealing that the anisotropy of BODIPY C12 molecule enables membrane-fluidity evaluation in OA/CHAPSO bicelles.

the DMPC/DHPC bicelle is dispersed in DHPC solution through interactions between the DMPC and DHPC molecules 4 , but after dilution, the hydrated DHPC molecules leak from the DMPC/DHPC bicelles into the DHPC solution 4 . Finally, a phospholipid vesicle is formed by fusion of the DMPC/DHPC bicelles 4 . The DMPC bilayer membrane in the DMPC/DHPC bicelles maintains its initial membrane fluidity after dilution 4 . The result showed that the bicelles initial membrane properties, such as their membrane fluidity and morphology, are important factors in the design strategy of functional membrane materials with bicelle parents. Bicelle morphology is usually monitored by transmission electron microscopy TEM or cryo-TEM 6, 7 . However, morphology must be analyzed from different characterization techniques to avoid microscope image artifacts. The authors of combined static light scattering and dynamic light scattering DLS to evaluate the persistence of warm micelles in aqueous solution 8 . This technique detects the scattering vectors of various concentrations of micelles, but is not practical for morphological analysis because the morphology of bicelles depends on their con-centration.
Fluorescent probe analysis, which is independent of lipid concentration, can reproducibly characterize the membrane properties such as membrane fluidity 4,9,10 . 1,6-Diphenyl-1,3,5-hexatriene DPH is a fluorescent probe of polarization that detects local mobilities in a biomembrane or protein 11 . When excited under polarized light, a DPH molecule in a biomembrane develops a dipole moment and the magnitude of its emission light versus rotation profile depends on the viscosity in biomembranes 12 . Fluorescent polarization using DPH has estimated the membrane fluidities of vesicles, bicelles, and micelles 10 . The membrane fluidity is calculated as the inverse polarization 1/P, which determines whether the membrane phase is ordered or disordered. 1/P 6 is the threshold value between the disordered and ordered phases where 1/P 6 suggests an ordered phase 10 . We previously confirmed the bilayer membrane structure of DMPC/DHPC bicelles using the probes DPH and 6-dodecanoyl-N,N-dimethyl-2-naphthylamine Laurdan 10 . Laurdan characterizes the membranous packing density in DMPC/DHPC bicelles based on the solvation of DMPC/DHPC bicelles 10 . At a mixed molar ratio X DMPC DMPC / DMPC DHPC of 0.33-0.67, the low membrane fluidity and high packing density of DMPC/ DHPC bicelles measured by DPH and Laurdan, respectively confirmed the presence of an ordered DMPC bilayer membrane in the bicelles 10 . However, these analyses are inapplicable to more fluid bilayer structures than DMPC e.g., fatty acid molecular assemblies of oleic acids OAs with unsaturated chains because fluid detergents are not easily distinguished from OA assemblies. Fatty acid molecules decomposed from food coexisting with bile salts in the human digestive process 13 form an oblate bicelle which is a potentially suitable phospholipid parent. In an improved Laurdan analysis, bicelles of OA mixed with 3-3-cholamidopropyl dimethylammonio -2-hydroxypropane sulfonate CHAPSO were found to possess an OA heterogeneous bilayer membrane with an ordered phase and a disordered phase 14 . The size distributions of OA/ CHAPSO bicelles with broad peaks showed multi-dispersed bicelles and those sizes were reduced upon the addition of CHAPSO solutions with decreasing a mixed molar ratio X OA OA / OA CHAPSO of 0.70-0.30 14 . The results indicate the interaction between the CHAPSO molecule and the OA heterogeneous bilayer membrane. Thus, the improved Laurdan analysis can detect the packing density of a membrane. Here, the few OA molecules beneath the electrostatically adsorbed CHAPSO molecule are linked to a single CHAPSO molecule and the OA molecules beneath the CHAPSO molecule are difficult to fluidize 11 . As a result, adsorption of CHAPSO molecules to the carboxy group of OA molecules slightly improves the packing density of the OA membrane surface in an aqueous solution. However, there aren t many cases where the DPH-based fluorescent polarization method is applied to the OA/ CHAPSO bicelles. Although it is expected that the OA molecule forms a disordered membrane, there is not enough discussion about the difference in the membrane fluidities.
In the present work, we improve the fluorescent polarization method for evaluating the formation of bicelles containing fatty acid molecular assemblies. The polarization is reported to change as the anisotropy of the molecular assembly changes from spherical to rod-shaped 15 . In the reference 15 , the molecular assembly was a reversed-phase micelle of didodecyldimethylammonium bromide DDAB with 8-hydroxypyrene-1,3,6-trisulfonate HPTS as a probe. With increasing anisotropy, the authors reported a change in the overall rotation of the molecular assembly and the local rotation of HPTS 15 . It is due to the interaction between the trisulfonate group of HPTS and the dimethylammonium group of DDAB. In other words, by fixing a probe molecule to a hydrocarbon chain and suppressing its rotation, it is possible to evaluate slight changes in the membrane fluidity of fluidic molecular assemblies such as micelles. Therefore, it is necessary to employ a probe molecule having a fluorescent site at the end of the hydrophobic chain in order to evaluate the membrane fluidity in the center of an OA bilayer membrane at different X OA values. We employ the fluorescent fatty acid 4,4-difluoro-5-2thienyl -4-bora-3a,4a-diaza-s-indacene-3-dodecanoic acid BODIPY C12 , which has a carboxyl group Fig. 1 . BODIPY dyes have been used as fluorophores in the imaging and tracking of natural cells 16 . BODIPY cores with various substituents that provide the desired physical properties have also been developed 17 . For example, hydrocarbon chains have been substituted on the two pyrrole rings of the BODIPY core and the BODIPY core has been substituted at the end of the hydrocarbon chain of a phospholipid, forming molecules for characterizing the properties of lipid membranes 18,19 . The BODIPY cores of the first and second of the above molecules are positioned at the hydrophilic surface and the hydrophobic interior of the membrane, respectively. Both molecules provide information on the phase state of the membrane because the solvent polarity affects the emission intensity of BODIPY cores 18,19 . An excited BODIPY dye is also known to possess a transition dipole moment and polarization characteristics like the DPH molecule.

Phospholipid vesicle and bicelle preparation
Phospholipid vesicles and bicelles were prepared as references of the phase states and morphologies Fig. 2 10, 14 . A chloroform solution of phospholipids was dried in a round-bottomed flask by rotary evaporation under vacuum conditions. The obtained lipid film was re-dissolved in chloroform and the solvent was evaporated. The obtained lipid thin film was hydrated with phosphate buffer 50 mM, pH 7.0 . The obtained vesicle solution was frozen at 213 K and then thawed at 337 K. This freeze-thaw cycle was performed five times. LiposoFast extruder Avestin Inc., Ottawa, Canada was used to prepare the vesicles sizes to 100 nm or 50 nm. The vesicles were characterized at 298 K, where the bilayer membrane is ordered in the DPPC vesicle but disordered in the DOPC vesicle 14 . Near or above the phase-transition temperature of DMPC vesicles 295 K , the morphology of DMPC/DHPC bicelles may not be oblate 6, 10 , and the DMPC membranes of the DMPC vesicles and DMPC/DHPC bicelles are ordered at 289 K. At X DMPC 0.33, 0.60, and 0.67, DMPC/DHPC bicelles were reported to be oblate 10 .

OA/CHAPSO bicelle preparation
First, OA vesicles pH 8.51, NaCl 50 mM were prepared as the source of the OA bilayers. The OA vesicles were then mixed with BODIPY C12 or pyrromethene molecules at 1.0 mol . OA/CHAPSO bicelles with different X OA values 0.3, 0.5, and 0.7 were prepared by mixing 20 mM CHAPSO stock solution with 20 mM OA vesicle solution Fig. 2 . The total concentration 20 mM exceeded the critical micelle concentration of CHAPSO CMC CHAPSO 8 mM . The TEM images HF-2000, Hitachi Ltd., Osaka, Japan were obtained after staining the bicelles with uranium acetate.

Anisotropy evaluation of the OA/CHAPSO bicelles
Anisotropy of the OA/CHAPO bicelles was evaluated by the fluorescent probes BODIPY C12, pyrromethene, and DPH. The fluorescent intensity was measured using a fluorescence spectrophotometer F-2500; Hitachi, Osaka, Japan . The BODIPY C12 molecule in the bicelles was excited at 558 nm and its emission intensities were recorded at 576 nm. The emission intensity I of the polarized exciting probe was measured, and the anisotropy r and G factor were respectively determined as follows Fig. 2  In these expressions, the subscripts V and H denote the orientations of the polarizer in the vertical and horizontal directions, respectively. For example, I VH is the intensity with the excitation and emission polarizers oriented vertically and horizontally, respectively. The G factor suppresses the polarization bias introduced by the fluorescence spectrophotometer 12 . In addition, the P value is defined as: The r value is related with the P value according the following equation 12 .
The threshold value of 1/P 6, which corresponds to the boundary between the disordered and ordered phases using the probe DPH, is approximately converted to r 0.12 10 . The anisotropy characteristics of pyrromethene excitation wavelength Ex. 490 nm; emission wavelength Em. 534 nm and DPH Ex. 360 nm, Em. 430 nm were evaluated by the above calculation and the concentration of the probe molecules employed for the measurements was fixed at 2 μM. A high r value implies that the membrane fluidity of the molecular assembly is low 10, 15 .

Size evaluation of the OA/CHAPSO bicelles
To examine the size of OA/CHAPSO bicelles at 298 K, we attempted DLS employing a He-Ne laser with a wavelength of 632 nm 21 . The intensity of the scattered laser was transiently measured using a charge-coupled device detector to obtain its auto-correlation functions, G τ . The size distribution of the OA/CHAPSO bicelles was calculated by an approximate solution of the G τ based on the CONTIN algorithm 22 . The calculation was performed in MATLAB R2019b, The MathWorks Inc., Natick MA, USA .

Results and Discussion
3.1 Effect of probes on anisotropy of phospholipid vesicles We first evaluated the anisotropy of BODIPY C12 in phospholipid vesicles, for which the packing densities of the bilayer membranes are known. The anisotropy value of BODIPY C12 was compared with those of pyrromethene and DPH. Figure 3 displays the spectra of these probes in a DPPC vesicle at different directions of light propagation. Changing the direction of the polarizing plate changed the intensities of the fluorescence but not the overall spectral shapes. The r value was calculated from the peak intensity in each spectrum. Figure 4 compares the r values of the probes in the DOPC and DPPC vesicles. The different anisotropies of DPH r DPH in the DOPC and DPPC vesicles can be explained by the different phase states of the bilayer membranes 10 . Whereas a DOPC molecule has an unsaturated bond in its hydrophobic chain and a disordered phase in the bilayer membrane 14 , DPPC is a saturated phospholipid that builds an ordered bilayer membrane 14 . Consequently, the rotation of the DPH molecule was more suppressed in the DPPC vesicle than in the DOPC vesicle. Similar results were obtained for the anisotropy of pyrromethene r Pyrromethene , indicating that a pyrromethene molecule is also suitable for membrane-fluidity evaluation. However, the threshold value may be newly set for the type of probe molecule. On the other hand, the rotation of BODIPY C12 was thought to be more suppressed than those of pyrromethene and DPH, and an ordered phase as in the DPPC vesicle was assumed in the r BODIPY calculation. However, the results suggested a more fluidic membrane in the DPPC vesicle. The difference between the anisotropies of BODIPY C12 r BODIPY in the DOPC and DPPC vesicles was also smaller than that of the other probes. As the BODIPY C12 molecule has a fluorescent site and a hydrophobic chain terminated with a carboxyl group, it was probably inserted into the vesicles like phospholipids Fig. 2 . In general, the center of a bilayer membrane is disordered and mobile 23 . It was suggested that the BODIPY core located at the most hydrophobic center of the bilayer membranes of the DPPC and DOPC vesicles; thus, the difference between the r BODIPY values of DOPC and DPPC represented a slight difference in fluidity near the center of the membrane. Figure 5 shows the r BODIPY values in a DMPC vesicle, in DMPC/DHPC bicelles at X DMPC 0.67, 0.60, and 0.33. The DMPC vesicles of 100 nm and 50 nm showed similar r BODIPY values; it was considered that r BODIPY doesn t depend on the vesicle size. DMPC-rich bicelles of X DMPC 0.67 and 0.60 also showed similar r BODIPY values to the vesicles. Below the phase-transition temperature, the DMPC bilayer membranes in both the DMPC vesicle and DMPC/DHPC bicelle at X DMPC 0.67 and 0.60 are in the ordered state 10 . Therefore, these r BODIPY values were similar to that of the DPPC vesicle Fig. 4 . Then, the r BODIPY was smaller in the DMPC/DHPC bicelle at X DMPC 0.33 than in the DMPC vesicles and the other DMPC/DHPC bicelles at X DMPC 0.67 and 0.60. A DHPC molecule has much shorter chains than a DMPC molecule Fig. 1 , and acts as a weak detergent against the bilayer membrane of long-chain phospholipids, such as DMPC molecules 10 . Our previous report has shown that the DMPC/DHPC bicelles are disc-shaped at X DMPC 0.60, suggesting that DMPC molecules and DHPC molecules are separated in a DMPC/ DHPC bicelle 10 . Increasing the proportion of DHPC molecules is expected to eliminate the separation, resulting in smaller, disordered bicelle like a mixed micelle 10 . It was presumed that the r BODIPY value at X DMPC 0.33 decreased because DHPC molecules constructed a bilayer membrane together with DMPC molecules.
The characteristics of the parameters of the molecular assembly evaluated by each probe are shown in Fig. 6. The phase state of a phospholipid molecular assembly has been reproducibly obtained by conventional membrane fluidity using DPH molecules 10 . Recall that the threshold of the disordered state is r DPH 0.12 10 . The heterogeneous phase of the OA bilayer membrane in an OA vesicle and an OA/ CHAPSO bicelle has been observed in the improved packing-density evaluation using a Laurdan molecule 14,24,25 . The fluorescent spectrum of Laurdan in molecular assemblies has three moments in the range of 400-600 nm 14 ; those moments show the hydrophobic moment i , hydrophilic molecular assembly moment ii , and more hydrophilic molecular assembly like a micellar assembly moment iii , respectively. The packing density of molecular assemblies was calculated based on the area ratio of the deconvoluted moment i to ii, A i /A ii 14 . The threshold of packing density was defined by comparing the previously revealed phase states of DOPC and DPPC vesicles 14 . In contrast, al- though the exact anisotropy threshold of r BODIPY is currently unclear, the anisotropy differences detected by the BODIPY C12 molecule were attributable to differences in membrane fluidity of the center of the bilayer membrane.

Morphologies of OA/CHAPSO bicelles prepared at different X OA values
The morphologies of the OA/CHAPSO bicelles were observed in TEM images Fig. 7 . The initial OA vesicles were destroyed by mixing with CHAPSO solution. The OA/ CHAPSO bicelles formed various shapes for the ratio of OA to CHAPSO. The assemblies seen in X OA 0.70-0.30 were bicelles made by the interaction of the initial OA vesicles and the CHAPSO molecules. A CHAPSO molecule has a hydrophobic steroid skeleton with three hydroxy groups on its surface and a hydrophilic tail Fig. 1 . The hydrophobic surface of the steroid skeleton adsorbs to the hydrophilic surface of an OA vesicle; thereby, the CHAPSO molecule acts as a spacer that destroys the OA vesicle 14, 26 28 . The edge of the disrupted OA vesicle is exposed to aqueous solution and becomes covered by other CHAPSO molecules 11, 27,29 . As X OA decreased, the size of the OA/ CHAPSO bicelles also decreased Fig. 7 , indicating that the CHAPSO molecules further destroyed the bilayer membrane derived from the OA vesicle 14 . The micelles were dominant at X OA 0.10, and the presence of OA/ CHAPSO bicelle was clearly unconfirmed in the TEM image Fig. 7 . As determined in DLS measurements Fig. 8 , the mode sizes of the OA/CHAPSO bicelles ranged from 135 nm at X OA 0.70 to 66.8 nm at X OA 0.30; meanwhile, the particle size distributions at X OA 0.70 and 0.30 obtained from the TEM images correlated with their DLS results 24,25 . After analyzing the packing densities of Laurdan molecules in bicelles, we previously reported that the bilayer membrane structures in OA/CHAPSO bicelles 0.30 ≤ X OA ≤ 0.70 are derived from the OA vesicles 14,24 . The packing densities were slightly improved in the OA/CHAPSO bicelles obtained after mixing with CHAPSO molecules 14,24 , suggesting that the OA bilayer membrane derived from the OA vesicle included a heterogeneous coexistence of  ordered and disordered phases. Here, the packing densities of the bicelles formed at 0.30 ≤ X OA ≤ 0.70 were not significantly different, suggesting that bilayer membrane structures existed even in the smaller bicelles 14,24 . This result is probably explained by adsorption of CHAPSO molecules on the OA bilayer membranes 28 . A single CHAPSO molecule adsorbed on the hydrophilic surface of an OA vesicle is linked to several OA molecules 11 . Because the OA molecules beneath the CHAPSO molecule are difficult to fluidize 11, 28 , the OA bilayer membrane is thought to persist even after mixing the CHAPSO solution. From the previous report and the present TEM images, it was inferred that the anisotropic OA/CHAPSO bicelles formed by interactions between the OA bilayer membranes and the CHAPSO molecules.

Anisotropy of OA vesicle and OA/CHAPSO bicelles
The r values in the OA vesicles are shown in Fig. 9a. Each r DPH value indicated a disordered phase in the OA vesicles. Similar results were obtained for the values of r Pyrromethene and r BODIPY C12 . As an OA molecule has an unsaturated chain Fig. 1 , an OA vesicle tends to form a disordered bilayer structure similarly to DOPC vesicles . The r values remained similar after diluting the OA vesicles Fig.  9a , indicating that the r value of each probe was independent of OA concentration. After mixing with CHAPSO solution at X OA 0.70 and 0.30, the final OA concentrations were 6.0 and 14 mM, respectively. Figure 9b compares the r values after mixing the CHAPSO solutions with OA vesicles at different X OA . In all cases, the r value increased with decreasing X OA . Here, in the OA vesicle and OA/CHAPSO bicelle, the r DPH values indicated only the disordered phase. The r Pyrromethene tended to increase with decreasing X OA as well as the r DPH . However, as X OA decreased from 0.70 to 0.30, the change in r BODIPY was much more obvious than the changes in r Pyrromethene and r DPH . This result correlated with our previous finding that when CHAPSO molecules are mixed with an OA bilayer membrane derived from an OA vesicle, the packing density of the OA bilayer membrane increases in X OA 0.70-0.30 14 ; consequently, the decrease in membrane fluidity and the increase in packing density correlated. DHPC molecules are inserted into a bilayer membrane to fluidize the bilayer membrane, whereas CHAPSO molecules are expected to adsorb on the hydrophilic groups of a bilayer membrane and act as spacers to fluidize the bilayer membrane Fig. 2 2, 20 . The adsorption of CHAPSO molecules on the hydrophilic groups of the OA bilayer membrane likely caused an in-  crease of the packing density around their hydrophilic groups and a decrease in membrane fluidity. BODIPY C12 can be employed for representing changes in membrane fluidity within the OA bilayer membrane of OA/CHAPSO bicelle. The higher r values of all probes in the CHAPSO micelle than in the OA vesicle can be explained by the internal structure of the CHAPSO micelle. The molecular-assembling morphology of CHAPSO molecules is thought to depend on the CHAPSO concentration in aqueous media 30 , and above CMC CHAPSO , a two-layer spherical structure is expected 31 . In the hydrophobic core of a CHAPSO micelle, the aliphatic groups of the inner layer interact with the steroid groups of the outer layer 31,32 . This interaction within the core is expected to be stronger than the hydrophobic interaction between the unsaturated chains within the OA vesicle, and possibly suppresses rotation of the pyrromethene and DPH molecules inserted in the core. The prolate-ellipsoid shapes of CHAPSO micelles are also  thought to increase the r BODIPY value 32 . Moreover, all r values of the OA/CHAPSO bicelles remained almost unchanged for at least two weeks Fig. 9b , suggesting that the OA/CHAPSO bicelles had dispersed in the aqueous solution without aggregation or precipitation.

Conclusion
The membrane fluidity of OA/CHAPSO bicelles was inferred from the anisotropy values of two fluorescent probe molecules, BODIPY C12 and pyrromethene. As the OA/ CHAPSO bicelle has a heterogeneous bilayer membrane structure, its anisotropy could not be discussed using the conventional spectroscopic membrane-property evaluation method with DPH or Laurdan probes. The anisotropy value of pyrromethene, which has no carboxyl group, did depend on the membranes phase state, similarly to that of DPH. However, the core of BODIPY C12, which has a single chain with a terminal carboxyl group, was inserted in the center of the phospholipid bilayer membranes; consequently, the anisotropy of BODIPY C12 showed a smaller difference in the phase states of the phospholipid bilayer membranes than the other probes. Then, the anisotropy values of the probes in the OA vesicle were very similar, but were increased by mixing OA with CHAPSO solution. It suggested a decrease in membrane fluidity in more fluid bilayer structures such as OA/CHAPSO bicelles. Especially, the anisotropy of BODIPY C12 increased more significantly from X OA 0.70 to 0.30 than those of pyrromethene and DPH. The increased anisotropy values correlated well with the increase in membrane packing density. By combining the anisotropies of BODIPY C12 and pyrromethene, we could evaluate a membrane fluidity of the OA/CHAPSO bicelles. This technique is applicable to bicelles composed of more fluid bilayer structures and is useful for designing membrane materials from bicelle parents.