Photodynamic Efficiency of Xanthene Dyes and Their Phototoxicity against a Carcinoma Cell Line : A Computational and Experimental Study

The aim of this study is to assess the insights of molecular properties of the xanthene dyes [fluorescein (FL), Rose Bengal (RB), erythrosin B (EB), and eosin Y (EY)] to correlate systematically their photodynamic efficiency as well as their phototoxicity against a carcinoma cell line.The phototoxicity was evaluated by comparing the values of themedium inhibitory concentration (IC50) upon HEp-2 cells with the xanthene corresponding photodynamic activity using the uric acid as a chemical dosimeter and their octanolwater partition coefficient (logP). RB was the more cytotoxic dye against HEp-2 cell line and the most efficient photosensitizer in causing photoxidation of uric acid; nevertheless it was the only one characterized as being hydrophobic among the xanthenes studied here. On the other hand, it was observed that the halogen substituents increased the hydrophilicity and photodynamic activity, consistent with the cytotoxic experiments. Furthermore, the reactivity index parameters, electric dipolemoment,molecular volume, and the frontier orbitals were also obtained by the Density Functional Theory (DFT). The lowest dipole moment and highest molecular volume of RB corroborate with its highest hydrophobicity due to heavy atom substituents like halogens, while the halogen substituents did not affect expressively the electronic features at all.


Introduction
The photodynamic therapy (PDT) is a promising cancer treatment that involves the systemic or topic administration of a photosensitizer followed by a period of time in which the PS accumulates preferentially in the tumors cells followed by irradiation with visible light of compatible wavelength and absorption of the photosensitizer.Each factor is harmless by itself but when combined with molecular oxygen leads to the generation of lethal cytotoxic species and consequently to cell death and tissue destruction [1][2][3][4].The photodynamic therapy is an interesting treatment for cancer due to dual selectivity produced by both preferential uptake of the photosensitizer by the diseased tissue and the ability to confine activation of the photosensitizer to this diseased tissue by restricting the illumination to that specific region.Therefore, PDT allows for the selective destruction of tumors while leaving normal tissue intact [1][2][3][4][5].
Photofrin5, a complex mixture of porphyrin monomers, dimmers, and oligomers, was the first compound approved by the Food and Drug Administration (FDA) to be used in PDT for the treatment of some types of cancer.Although Photofrin and other porphyrin-related sensitizers show a weak absorbance in the red region of the spectrum (≥600 nm), where penetration of light in tissue is optimal, these sensitizers induce long-lasting skin photosensitivity (4-6 weeks) through retention in cutaneous tissue [6][7][8].Hence, the search for pure photosensitizers with light absorption in the range of 600 nm or higher, together with the absence of side effects exhibited by Photofrin, has led to the search and development of series of compounds, including chlorins and phtalocyanines with adequate absorption coefficients, that is, above 650 nm [6][7][8].
In order to understand the mechanisms involved in PDT, so that more effective photosensitizers can be developed, some synthetic xanthene dyes have been considered.That is the case of cyclic anionic xanthene dyes with three aromatic rings in linear arrangement and an atom of oxygen in the central ring as, for example, Rose Bengal (RB), erythrosin B (EB), eosin Y (EY), and fluorescein (FL) (Figure 1).These dyes possess interesting properties for PDT such as considerable quantum yield of singlet oxygen, intense light absorption in the visible region of the spectrum (500-570 nm), and low costs.Moreover, since these dyes do not absorb light in the wavelength of maximum tissue penetration (600-900 nm), they have been indicated for the treatment of superficial injuries [9][10][11].
Rose Bengal is known to produce singlet oxygen with a quantum yield of nearly 100% under illumination [12][13][14].It is also reported that the Rose Bengal induces damage in rabbit corneal (in vitro) [13] and in retinal pigment epithelial cells [15].Notwithstanding, RB shows a high potential to be used as photosensitizer and its clinical application has been limited since RB is immediately excreted right after being introduced into the living body and accumulated in the liver [12,13,15].In addition, RB has been used topically in ophthalmology for diagnoses and in photodynamic inactivation of different microorganisms [14,[16][17][18].On the other hand, erythrosin B and eosin Y possess photodynamic activity on bacteria and yeast [10,16,17,19].In the case of fluorescein, it shows high fluorescence making it an ideal visualizing agent for optical diseases leading as corneal trauma detection [20]; however it is not therapeutically used due to its low toxicity.
The interest in these dyes continues to be new as their derivatives are still nowadays being prepared and intensely studied.For example, Rose Bengal acetate [21] improves diagnostics and therapeutic effects due to the addition of a quencher group [21][22][23][24][25][26][27][28][29][30][31][32].Other examples are the investigation of fluorescein [33], erythrosin [34], and eosin derivatives [35] that have favorable characteristics for use in photodynamic therapy.Therefore, the photodynamic efficiency of four xanthene dyes (Rose Bengal, erythrosin B, eosin Y, and fluorescein) was determined in the present study by comparing their medium inhibitory concentration (IC 50 ) values in HEp-2 cells, their photodynamic activity using uric acid as a chemical dosimeter, and their octanol-buffer partition coefficient (log ).These results were then related to molecular and electronic properties of these dyes obtained with Density Functional Theory (DFT) aiming to provide insights for a rational design of new and better compounds to be used as photosensitizers in PDT [36][37][38][39][40][41].

Photosensitizers and Light Source. The xanthene dyes
Rose Bengal, erythrosin B, eosin Y, and fluorescein, the reagents 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and 1-octanol, were purchased from Sigma-Aldrich and were used without further purification.The dyes stock solutions were made by weighting and diluting (until the desired concentrations) using phosphate buffered saline (PBS) and storing at 4 ∘ C in the dark.The green light source was a BIOTABLE containing an array of diode emitting light (LED) at 532 nm providing a fluence rate of 19 mW cm −2 .

Phototoxicity and Dark Toxicity.
In order to study the intrinsic cellular toxicity of xanthene dyes, the cells were harvested using trypsin and incubated overnight for complete cell adhesion in 96-well plates at a concentration of 1 × 10 5 cells mL −1 (2 mL well −1 ).A range of concentrations of the dyes were added and incubated for 6 h.The medium with the dyes was then removed and fresh medium was added.For phototoxicity measurements, plates were illuminated with LED at 532 nm for 15 min, giving a light dose of 18 J cm −2 .Following this treatment, the cells were maintained at 37 ∘ C, 5% CO 2 , and 95% air for additional 48 h.To evaluate cell viability and thus calculate the toxicity, 50 L of MTT was added to each well and this was incubated for 3 h [42].The medium and MTT were aspirated; then 50 L of ethanol and 150 L of a solution containing PBS and 2-propanol were added to each well in order to solubilize the crystals.The absorbance was read on a plate reader (Benchmark-BIO-RAD) at 570 nm.

Photodynamic Activity (PA)
. Uric acid (UA) is a known singlet oxygen scavenger used as a chemical dosimeter to determine the photodynamic activity of photosensitizers, since this parameter is related to quantum yields for singlet oxygen production.When a solution of uric acid and photosensitizer is irradiated, the uric acid absorption band at 292 nm decreases as an evaluation of relative photodynamic activity of the photosensitizer [43][44][45].The samples containing 10 g mL −1 of uric acid and 5 g mL −1 of the photosensitizer were irradiated with green LED during 360 s.Before and after the irradiation, the absorbance spectra were registered and photodynamic activity could be determined by Fischer's expression [44]: where PA is the photodynamic activity (m 2 J −1 ), Δ UA is the absorbance decrease at 292 nm in a mixture of uric acid and photosensitizer after irradiation,  0 is the fluence rate (W m −2 ),  is the irradiation time (s), and  PS.irr refers to the absorbance of the photosensitizer in uric acid and photosensitizer solution at irradiation wavelength.

Partition Coefficients. The partition coefficients (log 𝑃)
were taken as a measure of the hydrophobic character of the dyes.They were calculated using the method of Pooler and Valenzeno [46].Phosphate buffered saline and 1-octanol were used as immiscible solvents initially saturated to each other.The dyes were dissolved in PBS (phosphate buffered saline) to give a final concentration of 3 g mL −1 ; then 10 mL of each of these solutions was individually placed in a flask with equal volume of 1-octanol.After 1 h of agitation, the mixture was left to rest overnight and then centrifuged.The dye concentration in PBS before and after being mixed with octanol was measured spectrophotometrically in a Hitachi U-2800.The logarithm of the partition coefficient, log , was calculated according to where   and   are the measured absorbance of the dye solution in PBS before and after partitioning, respectively. PBS and  OCT are the volumes of the PBS and octanol phases, respectively.The partition coefficient was determined five times for each dye.
2.6.Computational Methodology.The chemical structure of the four xanthene dyes Rose Bengal, erythrosin B, eosin Y, and fluorescein studied in this work (Figure 1) was also investigated with Density Functional Theory (DFT) as implemented in Gaussian 03 package [47].In the first step, a fully unconstrained geometry optimization was performed on the molecular structure of all xanthene dyes using B3LYP [48,49] (Becke's three-parameter hybrid exchange functional and Lee, Yang, and Parr's correlation functional) and 6-31+G(d) basis sets [50,51].It was followed by a vibrational frequency analysis and the absence of imaginary frequencies was used as a criterion to ensure that each optimized structure corresponds to a true minimum on their respective potential energy surface.The solvent effects were treated with the Integral Equation Formulation of the Polarizable Continuum Model (IEF-PCM) method [52][53][54] simulating the water environment, which has been successfully applied to study other tricyclic molecules [55,56].
Once the optimized structures were properly obtained, the energies of frontier molecular orbitals HOMO (highest occupied molecular orbital) and LUMO (lowest unoccupied molecular orbital), as well as electric dipole moment, molecular electrostatic potential, area, and volume of the four xanthene dyes, were then calculated.

Results and Discussion
Table 1 shows the phototoxicity and dark toxicity of the xanthenes in HEp-2 after 6 h incubation with the dye.Under dark conditions, the tested dyes showed Dark-IC 50 values indicating some level of inherent toxicity to this cell line.RB presented the highest levels of such toxicity with IC 50 of 58 g mL −1 , while FL showed the lowest intrinsic toxicity observed with IC 50 of 630 g mL −1 .It is important to comment that the incubation time affected the xanthene dyes cytotoxicity, where the decrease of the incubation time leads to an increase of the IC 50 value, suggesting that more dyes accumulate in the cell.
Another inspection on Table 1 demonstrates that the light greatly improved the cytotoxicity of the dyes in the cell line tested.In this aspect, RB was found to be the most phototoxic dye with IC 50 of 11 g mL −1 , which was the lowest IC 50 under illumination, followed by EB, EY, and FL, as shown by Masamitsu et al. [57].In contrast, FL reaches 40% of its potentialization by light, whereas RB shows a large decreasing in IC 50 under illumination (≈80%), which is compared with the other photosensitizers that reach their maximum potentialization in the used light dose.
Figure 2 shows the variation in absorbance bands assigned to uric acid (292 nm) and the minor changes at the absorption bands of the photosensitizers, probably due to some photodecomposition of the dyes.This method is based on the suppression of singlet oxygen by uric acid ( = 3.6 × 10 8 mol L −1 ), and it is important to mention that this reaction leads to formation of triuret, sodium oxalate, allantoxaidin, and CO 2 , according to Fischer et al. [44].In addition, Table 2 presents the obtained values for photodynamic activities.In the case of RB, this dye demonstrated the highest photodynamic activity result (6.0 m 2 /J), while FL showed the lowest one (0.4 m 2 /J).These results agree with the This relation is possible of verifying, once the singlet oxygen is considered the main responsible agent to cause tumor cells death and irradiated photosensitizers that produces higher amounts of singlet oxygen and consequently lead to high cytotoxic effects.Thus, the results of the photooxidation of uric acid and determination of the IC 50 experiments converge to the following order of photodynamic efficiency for the dyes studied here: Rose Bengal (RB) > erythrosin B (EB) > eosin Y (EY) > fluorescein (FL).Table 3 shows the logarithm of the partition coefficients of the photosensitizers.Here it is important to remember that the hydrophilic character of a compound is associated with the poor distribution in octanol/water (then log  < 0), while lipophilic substances are characterized by log  > 1.5.On the other hand, species that present log  in the intermediate range are considered amphiphilic [58].Therefore, the log  results indicate RB as being amphiphilic (0 < log  < 1.5), while the other dyes demonstrate a hydrophilic nature (log  <0).These results are in agreement with the photodynamic activity and IC 50 values, once RB is the most hydrophobic dye and demonstrated the highest photodynamic activity value being the most cytotoxic.EB and EY have high values of photodynamic activity (5.6 e 4.2 m 2 J −1 ) but they are not the most cytotoxic, being this result in agreement with log  values for these dyes, suggesting that they are hydrophilic and have the tendency to locate in plasma membrane.Nevertheless, a different picture is seen for FL which shows a hydrophilic character and possess the lowest value of photodynamic activity (0.4 m 2 J), which may explain the fact of being the least efficient photosensitizer among the dyes studied here.This hydrophilic/lipophilic balance of the photosensitizer is very important in determining its cell localization and therefore its site of action [59][60][61], where a hydrophobic photosensitizer can diffuse across the plasma membrane and then to relocate to other intracellular membranes.In contrast, those ones that are less hydrophobic tend to be too polar to diffuse across the plasma membrane, having its action restricted to this site, or are taken up by endocytosis [61].Moreover, lipophilic photosensitizer tends to be localized in the mitochondria [62,63] and the activation of a photosensitizer located in the mitochondria leads to apoptosis [59,64].
From another glance at Figure 1, it can be seen that the 3-oxo-xanthen-9-yl system is planar in all cases and the angle between this plane and the phenyl ring is ca.60 ∘ for EB, EY, and FL while RB shows a right angle possibly due to hindrance effect of the ortho-chlorine atoms bonded to the phenyl ring.In addition, plots of the frontier molecular orbitals are presented in Figure 3.The HOMO plots are very similar for all xanthene dyes studied, except for the different contributions of carboxylic acid group bonded to the phenyl ring.The atoms localized at the extremity of 3-oxo-xanthen-9-yl ring system contribute to the HOMO of all systems, while the HOMO of the RB and EB also shows contribution from the carboxylic acid group.The LUMO shows contributions from the atoms of the center of the 3-oxo-xanthen-9-yl ring system, while the oxygen atom from the carboxylic group of the EB, EY, and FL, positioned next to the 3-oxo-xanthen-9yl ring system, also contributes to the LUMO of these systems.For the RB and EB, all iodine atoms participate in HOMO but not in LUMO.The same occurs to the bromine atoms of EY.Hence, the main transition (HOMO-LUMO) for the dyes studied in this work corresponds to a charge transfer from the extremity, including the substituent atoms, to the center of the ring system.Meanwhile, the LUMO plots depend on the compound.For example, only the LUMO of the FL shows contribution from the atoms of the substituents at three phenyl rings, while the 3-oxo-xanthen-9-yl ring system atoms are more important for the LUMO of the EB than for the LUMO of the EY.
The calculation of the reactivity index parameters was also performed based on the methodology of Pearson [65] and Parr et al. [66].Table 4 presents the chemical hardness (), electronic chemical potential (), electrophilicity (), and the Kohn-Sham eigenvalue of HOMO and LUMO.An interesting aspect in RB and EB is their small HOMO-LUMO gap energy, which may be caused from the stabilization induced by the iodine substituents over the tricyclic moiety that induces the electron-acceptor effect in these molecules.An interesting aspect is that the chloride atoms of the RB participate neither in the HOMO nor in the LUMO of this system indicating that the phenyl ring substituents do not contribute to increase of its HOMO-LUMO energy gap.The  result is an important indication of the resistance of  compounds to change its electronic configuration, while  is a parameter to predict the escaping tendency of electrons in the specie, which is also associated with the electronic charge rearrangement associated with any chemical process.Among the four structures, the lowest  value is estimated for RB, while the lowest  result is seen for EB.In addition, it is noted that EB shows a better electrophilic character than the other dyes examining the electrophilicity index, while FL and EY showed a large nucleophilic character.In general aspects, severe changes are not observed when we compare FL/EY or even EB/RB.Moreover, this large nucleophilic nature from FL and EY may also explain the fact that these two dyes present the highest IC 50 results.
Since the area and the molecular volume are directly related to the experimental data (i.e., RB > EB > EY ≫ FL), the photodynamic activity decreases from RB to FL due to the presence of heavy atoms.It has been often demonstrated that the replacement of an atom in a structure by other with higher atomic number leads to an increase of the quantum yield.The hydrophilic/lipophilic character of the molecules can be evaluated by means of the quantum chemical calculations from the following properties: dipole moments, atomic charges, area, and molecular volumes.The dipole moment results show that RB has the lowest dipole moment obtained, what could be assigned to the presence of chloride atoms, which present negative charges.As the 3-oxoxanthen-9-yl ring system essentially does not contribute to the RB dipole moment, this is the most hydrophobic molecule explaining its highest log  value.Contrarily, the results presented in this work indicate that the presence of heavy atoms does not change the electronic properties of the dyes studied but increases their hydrophobicity as demonstrated by the decrease in the electric dipole moments (FL > EY > EB > RB), whereas in addition an increase in the molecular volume is estimated (RB > EB > EY ≫ FL).
Another important issue, which needs to be addressed here, is the molecular electrostatic potential as presented in Figure 4. Examining RB, the nucleophilic and the electrophilic regions are localized in the phenyl group and in the peripheral areas, respectively.In RB, a different picture is noted when we look at the central region, where the neutral area is localized with a small point presenting a nucleophilic character.In contrast, EB shows a similar picture with RB; however EB demonstrates a decrease in the nucleophilic nature of the phenyl group, whereas in the central area a large positive region is observed.Nevertheless, FL and EY present a neutral region around each molecule with a large negative area in the carboxyl group.Thus, the neutral/nucleophilic character at the phenyl group may explain the interaction with uric acid and the photocytotoxicity assay.On the other hand, it does not mean, however, that RB and EB should show similar behavior as photosensitizers, since the presence of the chloride atoms is the only difference in the composition of these compounds.As the presence of the chloride atoms is the only difference in the composition of RB and EB, it does not mean that these compounds should present similar behavior as photosensitizers Figure 4: Molecular electrostatic potential of each xanthene dye (isovalue of 0.0004 a.u.).The right-side view shows the front of the solid surface, and the left-side view shows the back of the surface and displays its molecular disposition.The regions in red refer to areas with an electrophilic character, while the regions in blue are areas with a nucleophilic nature.A different case is the area in green which is related to neutral regions.
since the intrinsic characteristics of these chloride atoms can significantly change other properties of RB such as aggregation and volume, hence altering their efficiency in photodynamic activity.Moreover, as RB shows higher photodynamic efficiency than EB, the presence of iodine atoms could be even better substituent instead of chloride.

Conclusions
The purpose of this investigation was to perform an assessment of two methods in order to characterize the photodynamic efficiency of four dyes with potential use in photodynamic therapy.The medium inhibitory concentration (IC 50 ) in tumoral cell line HEp-2 was determined aiming to evaluate the dyes cytotoxicity and phototoxicity.The photoxidation of uric acid was used to estimate the production of singlet oxygen of the dyes, while the determination of the octanol-buffer partition coefficient (log ) allowed measuring the lipophilic character of the compounds considered.RB showed the higher phototoxicity among the xanthene dyes studied in this work; however this dye has the highest toxicity without illumination against Hep-2.Although the other dyes are less effective in killing cells under illumination, they have much lower intrinsic dark cytotoxicity.Moreover, a quantum chemical investigation was also carried out to provide valuable insights for the design of new and better compounds to be employed as photosensitizers in photodynamic therapy.

Figure 2 :
Figure2: Absorption spectra of solutions containing uric acid (10 g mL −1 ) and dye (5 g mL −1 ) dissolved in PBS before (black) and after (red) the irradiation with LED.The absorption spectra show the decrease of the band in 292 nm attributed to the acid uric photoxidation.

Figure 3 :
Figure 3: Plots of frontier orbitals (HOMO and LUMO) for the four xanthene dyes studied.

Table 1 :
IC 50 (g mL −1 ) values under illumination (light dose of 18 J cm −2 ) and in the dark for the xanthene dyes incubated in HEp-2 cells for 6 h * .
*The fluorescence rate of the used LED was 19 mW cm −2 .The mean values and the standard deviations (m ± sd) for  = 7.

Table 2 :
Photodynamic activity (m 2 J −1 ) of the photosensitizers [the mean values and the standard deviations (m ± sd) for  = 4].

Table 3 :
The partition coefficient values for the xanthene dyes studied [the mean values and the standard deviations (m ± sd) for  = 5].
50values are detected by RB and FL, respectively.