New Materials Based on Cationic Porphyrins Conjugated to Chitosan or Titanium Dioxide: Synthesis, Characterization and Antimicrobial Efficacy

The post-functionalization of 5,10,15-tris(1-methylpyridinium-4-yl)-20-(pentafluorophenyl)porphyrin tri-iodide, known as a highly efficient photosensitizer (PS) for antimicrobial photodynamic therapy (aPDT), in the presence of 3- or 4-mercaptobenzoic acid, afforded two new tricationic porphyrins with adequate carboxylic pending groups to be immobilized on chitosan or titanium oxide. The structural characterization of the newly obtained materials confirmed the success of the porphyrin immobilization on the solid supports. The photophysical properties and the antimicrobial photodynamic efficacy of the non-immobilized porphyrins and of the new conjugates were evaluated. The results showed that the position of the carboxyl group in the mercapto units or the absence of these substituents in the porphyrin core could modulate the action of the photosensitizer towards the bioluminescent Gram-negative Escherichia coli bacterium. The antimicrobial activity was also influenced by the interaction between the photosensitizer and the type of support (chitosan or titanium dioxide). The new cationic porphyrins and some of the materials were shown to be very stable in PBS and effective in the photoinactivation of E. coli bacterium. The physicochemical properties of TiO2 allowed the interaction of the PS with its surface, increasing the absorption profile of TiO2, which enables the use of visible light, inactivating the bacteria more efficiently than the corresponding PS immobilized on chitosan.


Introduction
Nowadays, antimicrobial photodynamic therapy (aPDT) is considered to be a promising approach to inactivate a broad spectrum of pathogens, namely, microorganism strains with high resistance to In recent years, our studies have focused on the design of porphyrin derivatives with adequate structural features to be used as PS in aPDT, namely of Gram-negative and Gram-positive bacteria [7,14,20,22,28,[41][42][43][44][45]. In this context, the development of photoactive functional materials, as well as their reusability, represents an important challenge for a sustainable development of aPDT protocols [22].
During our studies concerning aPDT, we found that the tricationic porphyrin 5,10,15-tris(1-methylpyridinium-4-yl)-20-(pentafluorophenyl)porphyrin tri-iodide (P2 in Scheme 1) bearing a pentafluorophenyl group is a highly efficient PS due to its ability to photoinactivate a wide range of microorganisms under a lower irradiation protocol than the one needed for other PS [46][47][48][49][50][51]. In an attempt to develop new materials based on that porphyrinic core and solid supports with antimicrobial capabilities such as chitosan and titanium dioxide, we decided to decorate the meso-pentafluorophenyl group of P2 with other substituents containing carboxylic groups which will improve the interaction of the macrocycle with chitosan and titanium dioxide. It is expected that the synergistic combination of the P2 derivatives with the antimicrobial capability of the materials could lead to an effective inactivation of both the Gram-positive and the Gram-negative bacteria upon irradiation with visible light.

Synthesis of Porphyrin Derivatives
The general synthetic procedure used to obtain the new cationic porphyrin derivatives P3 and P4 bearing carboxyl groups is provided in Scheme 1. Both porphyrins were obtained using the tricationic porphyrin P2. The neutral precursor P1 was prepared by reacting pyrrole with the adequate stoichiometric proportion of pentafluorobenzaldehyde and pyridine-4-carbaldehyde in a mixture of acetic acid/nitrobenzene according to a literature procedure [46]. The quaternization of the pyridyl groups was carried out with methyl iodide and the required cationic porphyrin P2 [5,10,15tris(1-methylpyridinium-4-yl)-20-(pentafluorophenyl)porphyrin tri-iodide] was quantitatively obtained [46]. The following aromatic nucleophilic substitution took place with 3-or 4mercaptobenzoic acid in the presence of pyridine and DMF at room temperature. The required In the present study, the further functionalization of the efficient photosensitizer P2 with carboxylic groups was achieved by involving nucleophilic aromatic substitution of the p-fluorine atom of the C 6 F 5 unit with 3-or 4-mercaptobenzoic acid (Scheme 1); it was considered that the position of the carboxylic group could influence the photoinactivation efficiency of the PS and/or of the PS hybrid materials. The new porphyrins obtained, P3 and P4, both bearing carboxylic groups, and also (for comparison) the template P2 were subsequently immobilized on chitosan films (CF) or on TiO 2 giving rise to materials Px-CF or Px-TiO 2 , respectively (where x stands for the porphyrin derivative number). The efficacy of the photodynamic action of P2, P3 and P4 and of the new materials was assessed in the presence of recombinant bioluminescent E. coli. This Gram-negative bacterium is an excellent model to monitor the effectiveness of a photoinactivation process, since its light output is a sensitive sign of its metabolic activity [16,19,20,[52][53][54][55]. Additionally, E. coli is one of the most common Gram-negative pathogens in humans, causing several serious illnesses, and it is known to develop multidrug resistance [38,56].

Synthesis of Porphyrin Derivatives
The general synthetic procedure used to obtain the new cationic porphyrin derivatives P3 and P4 bearing carboxyl groups is provided in Scheme 1. Both porphyrins were obtained using the tricationic porphyrin P2. The neutral precursor P1 was prepared by reacting pyrrole with the adequate stoichiometric proportion of pentafluorobenzaldehyde and pyridine-4-carbaldehyde in a mixture of acetic acid/nitrobenzene according to a literature procedure [46]. The quaternization of the pyridyl groups was carried out with methyl iodide and the required cationic porphyrin P2 [5,10,15-tris(1-methylpyridinium-4-yl)-20-(pentafluorophenyl)porphyrin tri-iodide] was quantitatively obtained [46]. The following aromatic nucleophilic substitution took place with 3-or 4-mercaptobenzoic acid in the presence of pyridine and DMF at room temperature. The required porphyrins P3 and P4 have been obtained with 90% and 93% yield, respectively [28,57,58]. The structures of the obtained products were confirmed by spectroscopic techniques (Figures S1-S15). The analytical data obtained for the intermediates is in accordance with the data previously published [59].
The 1 H NMR spectra of compounds P3 and P4 showed the expected signals assigned to their structures. The insertion of 3-mercaptobenzoic acid residue rendered a non-symmetric pending group giving rise to three signals depicted at δ 7.69, δ 8.05-7.97 and δ 8.23 ppm, assigned to the resonances of H5, H4 and H6, and H2 of the 3-mercaptobenzoic unit. On the other hand, the insertion of the 4-mercaptobenzoic group gave rise to two doublets depicted at δ 8.03 and δ 7.78 ppm. All these assignments were confirmed through correlations found in 2D NMR spectra (COSY). The 19 F NMR spectra of both derivatives also confirmed the success of the nucleophilic substitution by the disappearance of the resonance signal of the p-fluorine nucleus.
The absorption, emission and excitation spectra of P3 and P4 were recorded in DMF solutions (1 µM) at 298 K and were compared with those from the precursors P1 and P2 (see Table 1 and Figures S13 and S14, Supporting Information). The absorption spectra of P3 and P4 show the typical profile expected for free base meso-tetraarylporphyrins with the Soret band at 420 nm and four weak Q bands in the 500-650 nm range. The phyllo type Q-bands (IV > II > III > I) evidence a significant reduction in the intensity of bands III and I, which is a typical feature of tetraarylporphyrins containing electronegative substituents at the ortho positions of the phenyl rings [60].
The steady-state fluorescence emission spectra of all compounds, after their excitation at ca. 420 nm, show essentially two bands: one centered at ca. 648 nm and the other at 711 nm (Table 1 and Figure S16). These emission bands can be assigned to Q x (0-0) and Q x (0-1) transitions, typical of the free-base porphyrins with a D 2h symmetry due to a nearly unchanged vibronic state upon excitation [61]. The perfect match between the absorption and the excitation spectra observed for all the compounds rules out the presence of any emissive impurities (Figures S14 and S17, Supporting Information).
The Stokes shift for P3 and P4 suggest a change in the electronic nature of the excited state compared to the ground state [62]. The fluorescence yields (Φ F ) were determined using Equation (1), and 5,10,15,20-tetraphenylporphyrin (H 2 TPP) as reference (0.11 in DMF) [63,64]. In this equation, Φ F is the emission quantum yield of the sample, Φ st the emission quantum yield of the standard, A st represents the absorbance of the standard and A s the absorbance of the sample at the excitation wavelength. Furthermore, S st and S s represent the integrated emission band areas of the standard H 2 TPP and of the sample, respectively. The results summarized in Table 1 show that P1 and P3 presented a slightly higher fluorescence quantum yield than P2 and P4.

Immobilization of Porphyrins on Chitosan
The immobilization of the cationic porphyrins P2, P3 and P4 on chitosan was performed by adding the adequate amount of each porphyrin dissolved in THF to a solution of chitosan in an aqueous acetic acid solution (1% v/v), according to a procedure previously reported by our group [28]. The amount of porphyrin immobilized was calculated indirectly by UV-Vis, considering the amount of unreacted porphyrin in the supernatants resulting from the washing process. The chitosan-porphyrin films, designated as P2-CF, P3-CF and P4-CF, were prepared by dissolving the porphyrins and chitosan in a 1% acetic acid solution followed by casting in a ventilated oven at 30 • C. The percentage of each porphyrin immobilized on the chitosan support is summarized in Table 2. In Table 2 it is possible to observe that P3 and P4 lead to higher immobilization percentages than P2, most certainly due to the strong interactions (electrostatic and/or hydrogen bond) between the porphyrinic carboxylic unit and the chitosan cationic ammonium groups. On the other hand, although P2 interacts with chitosan in a lesser extent, this interaction seems also to be facilitated by hydrogen bonding involving the fluorine atoms and the chitosan OH and NH 2 groups; additionally, one cannot exclude the interactions between the π-conjugated electron cloud of the macrocycle rings and the hydroxyl groups of chitosan [65].
All the chitosan-porphyrin films were characterized using a wide array of solid-state spectroscopic techniques such as UV-Vis and fluorescence, ATR-FTIR and powder X-ray diffraction (PXRD).
The presence of the porphyrin on the chitosan films P2-CF, P3-CF and P4-CF was promptly confirmed through UV-Vis and fluorescence spectra in the solid state. All the absorption spectra ( Figure 1) displayed the typical Soret band accompanied by four Q-bands. All these bands are red-shifted and broader when compared with the corresponding absorption bands in solution (Table 1) [66]. A similar situation was observed with the UV-Vis of P2, P3 and P4 in the solid state ( Figure S15, Supporting Information). The Soret bands suffered a red-shift of ca. 10 nm and in the Q-band region, depending on the porphyrin, red-shifts of ca. 5-16 nm were also detected. These shifts are attributed to small alterations of porphyrin molecular environment due to its interaction with chitosan [28,67].
( Figure 1) displayed the typical Soret band accompanied by four Q-bands. All these bands are redshifted and broader when compared with the corresponding absorption bands in solution (Table 1) [66]. A similar situation was observed with the UV-Vis of P2, P3 and P4 in the solid state ( Figure S15, Supporting Information). The Soret bands suffered a red-shift of ca. 10 nm and in the Q-band region, depending on the porphyrin, red-shifts of ca. 5-16 nm were also detected. These shifts are attributed to small alterations of porphyrin molecular environment due to its interaction with chitosan [28,67].  (Table 1) a red-shift of ca. 9-15 nm was observed in the first emission band. The low emission observed in the solid state for P2-P4 can probably be attributed to their tendency to form aggregates, leading to fluorescence quenching. Figure 3 shows the visual appearance of chitosan and of the P2-CF, P3-CF and P4-CF materials under ultraviolet light. The fluorescence intensities have not a linear dependence with the concentration of the porphyrin present in chitosan (P3 and P4 lead to higher immobilization percentages than P2, Table 2) but are dependent on the substituents present on the porphyrin core. This behavior is probably due to self-absorption and self-quenching of the PS supported in the films as described previously by Tian and co-workers [61].  (Table 1) a red-shift of ca. 9-15 nm was observed in the first emission band. The low emission observed in the solid state for P2-P4 can probably be attributed to their tendency to form aggregates, leading to fluorescence quenching. Figure 3 shows the visual appearance of chitosan and of the P2-CF, P3-CF and P4-CF materials under ultraviolet light. The fluorescence intensities have not a linear dependence with the concentration of the porphyrin present in chitosan (P3 and P4 lead to higher immobilization percentages than P2, Table 2) but are dependent on the substituents present on the porphyrin core. This behavior is probably due to self-absorption and self-quenching of the PS supported in the films as described previously by Tian and co-workers [61].    The ATR-FTIR spectra of the Px-CF conjugates displayed the characteristic bands of chitosan ( Figure S18, Supporting Information). The band at 3450 cm −1 is attributed to overlapped OH and N-H vibrations, while the bands at 2916 cm −1 , 2867 cm −1 are associated with C-H and -C=O of the amide group CONHR vibrations of chitosan.
The bands at 1589, 1412 and 1332 cm −1 are associated with the bending vibrations of the N-H, C-H, O-H bonds; the vibration at 1152 cm −1 is due to ν as (C=O) oxygen bridges and the band at 893 cm −1 to ω(C-H) from the polysaccharide's structure. Most of the characteristic peaks of the porphyrin core are significantly reduced, when compared with chitosan, due to the low porphyrin immobilization on CF films ( Table 2).
The PXRD of P3-CF and P4-CF showed some differences in the crystallinity when compared with CF. The peak observed at 42.9 • for CF due to the crystalline phase of chitosan was not observed in P3-CF and P4-CF, indicating that the porphyrin immobilization led to a loss of crystallinity [68]. More specifically, it is possible to note for P3-CF a new set of peaks at 8.5, 11.8 and 15.2 • , attributed to the chitosan chain rearrangements during the porphyrin immobilization process. Most of the peaks in the PXRD of these materials are characteristic peaks of chitosan while the presence of peaks related to the supported porphyrins are not visible [28,69]. Interestingly, the PXRD of P2-CF shows a predominant amorphous phase of chitosan ( Figure S20, Supporting Information).

Immobilization of Porphyrins on Titanium Dioxide
The immobilization of the cationic porphyrins P2, P3 and P4 on TiO 2 was performed by adding an adequate amount of each porphyrin dissolved in methanol to solid TiO 2 (see details in the experimental section).
The amount of each porphyrin immobilized on TiO 2 was also calculated by UV-Vis, considering the amount of porphyrin in the supernatants resulting from the washing process. The immobilization percentages of each porphyrin on the TiO 2 support are summarized in Table 3. It is possible to note that the presence of the carboxylic groups in the porphyrin core is a fundamental requisite for an efficient electrostatic interaction with the TiO 2 solid. Porphyrin P2, without the acidic groups, was completely removed after the washing process, while P3 and P4 showed higher than 90% immobilizations. The UV-Vis spectra of the solids P3-TiO 2 and P4-TiO 2 show the typical porphyrin absorption bands confirming the success of P3 and P4 immobilization on TiO 2 ( Figure 4). The wavelengths of the Soret band at 420 nm and the Q-bands at 512, 547, 588, 649 nm are similar to those of the non-immobilized porphyrins in solution. However, when the spectra are compared with the absorption spectra of the porphyrins in the solid state, a blue shift can be observed ( Figure S15, Supporting Information) [70,71]. This is probably due to interactions between the porphyrin plane and the support in an attempt of the macrocycle to acquire a more planar conformation in order to maximize the electrostatic interaction [72].
the Soret band at 420 nm and the Q-bands at 512, 547, 588, 649 nm are similar to those of the nonimmobilized porphyrins in solution. However, when the spectra are compared with the absorption spectra of the porphyrins in the solid state, a blue shift can be observed ( Figure S15, Supporting Information) [70,71]. This is probably due to interactions between the porphyrin plane and the support in an attempt of the macrocycle to acquire a more planar conformation in order to maximize the electrostatic interaction [72].   Figure 6 shows the corresponding emission spectra in the solid state. Contrarily to the low emission observed with P3 and P4 in the solid state, both hybrids showed a strong fluorescence. Hybrid P3-TiO2 shows two emission bands at 658 and 715 nm, with the first one more pronounced than the second one ( Figure  6e), while P4-TiO2 exhibits the same profile but with the two bands at 667 and 717 nm (Figure 6d). The fluorescence excitation spectra (λem at 715 or 717 nm) show the typical features of porphyrins. These emission bands are slightly shifted when compared with those obtained when the same porphyrins were immobilized on chitosan. This behavior can be attributed to the distinct immobilization modes on the support and agrees with the distortions in the structure of the porphyrins indicated by the UV-Vis results. García-Sánchez and co-workers [73] observed a shift in the emission spectra for porphyrins immobilized on silica. The authors attributed this behavior to the different polarity existing inside the silica pores and the interactions with the surface groups.    The ATR-FTIR of P3-TiO 2 and P4-TiO 2 exhibit the typical vibration bands of titanium dioxide in the region of 3400 cm −1 (ν OH), at 1630 cm −1 attributed to the bending vibration of the hydroxyl group on the surface of TiO 2 and at 481 cm −1 corresponding to the vibration of the Ti-O-Ti bonds [74]. In these solids, some characteristic bands due to the presence of the porphyrins were also observed at approximately 1593 and 1408 cm −1 arising from the symmetric and asymmetric vibrations of the C=O group and at 1472 cm −1 due to C=C vibrations ( Figure S19, Supporting Information).
P3-TiO 2 and P4-TiO 2 hybrids displayed a PXRD pattern similar to those obtained from the pure support ( Figure S21, Supporting Information). The two intense peaks at 25.3 • and 27.4 • (2θ) attributed to the anatase (101) and rutile (110) phases of TiO 2 , can be clearly observed in the hybrids, showing that the porphyrins do not interfere with the TiO 2 crystallinity structure.

Singlet Oxygen Generation and Photostability
Knowing that the ability of a porphyrin or a material to act as a photosensitizer is related with its ability to generate ROS, namely 1 O 2 , and also with its stability, these essential features were also considered in the characterization of the new compounds and materials.
The efficacy of compounds P3 and P4 to generate singlet oxygen was verified using 1,3-diphenylisobenzofuran (DPiBF) as a scavenger of singlet oxygen ( 1 O 2 ), and further compared with the ability of the precursors P1 and P2 (Figure 7). The degradation of DPiBF was monitored by the decrease of its band at 415 nm, after its irradiation in the presence of each porphyrin and atmospheric oxygen. The reaction between 1 O 2 and DPiBF in a 4+2 process affords 1,2-dibenzoylbenzene giving an indirect evidence of 1 O 2 generation.  The DPiBF decomposition provided by the materials PS-CF and PS-TiO2 was also investigated and the results are summarized in Figure 8. The results show that P3-TiO2 is a more efficient 1 O2 producer than P4-TiO2. This result is in line with the better performance of P3 to generate 1 O2 when compared with P4. Considering the chitosan materials, the higher DPiBF decay was caused by P2-CF and none of the other materials (P3-CF and P4-CF) were able to generate 1 O2 under the tested conditions.
Interestingly, P2-CF showed almost the same 1 O2 production as P2 in solution. This behavior might be related with an aggregation attenuation or an increased photostability of the porphyrin after its immobilization on the chitosan support.
The photostability studies were performed with P2-P4 in PBS, monitoring the decrease in absorbance of the corresponding Soret bands after different times of irradiation with a LED array at an irradiance of 10 mW·cm⁻ 2 . P4 suffered a decay of 64%, followed by P2 (41%) and P3 (22%) during The results summarized in Figure 7 show that all the porphyrins are good generators of 1 O 2 , but the best efficacy was found for the new porphyrins P3 and P4. These two new porphyrins can generate 1 O 2 more efficiently than H 2 TPP, also considered an excellent 1 O 2 generator [75]. The best 1 O 2 generator was P3, followed by P4 and H 2 TPP, and finally by P1 and P2, both with a similar efficacy. These results suggest that the presence of the acid group and its position are important features for the efficiency in 1 O 2 generation and the lower efficacy of P1 and P2 is probably associated with aggregation phenomena. The DPiBF decomposition provided by the materials PS-CF and PS-TiO 2 was also investigated and the results are summarized in Figure 8. The results show that P3-TiO 2 is a more efficient 1 O 2 producer than P4-TiO 2 . This result is in line with the better performance of P3 to generate 1 O 2 when compared with P4. Considering the chitosan materials, the higher DPiBF decay was caused by P2-CF and none of the other materials (P3-CF and P4-CF) were able to generate 1 O 2 under the tested conditions.

Photoinactivation Studies
It is commonly accepted that the success of photoinactivation closely depends on the structure of the photosensitizing agent [14]. As previously stated, the charge of the PS is extremely important for the inactivation of the microbial agents and, generally, positively charged PS are more efficient for Gram-negative strains, even if they are at low concentrations when compared to neutral or negatively charged PS [52]. Studies with cationic PS demonstrate that these derivatives are more likely to inactivate both Gram-positive and Gram-negative bacteria without the help of membrane disrupting agents [47,76].
The biological assays started by evaluating the photodynamic action of the new cationic porphyrins P3 and P4 at the concentration of 5.0 µ M, under white light towards the recombinant bioluminescent E. coli ( Figure 9). As mentioned above, this Gram-negative bacterium is an excellent bacterial model to monitor the effectiveness of a photoinactivation process, since its light output is a highly sensitive reporter of its metabolic activity. These set of experiments were not performed in the presence of the non-immobilized P2 since its efficacy has already been studied by some of us [46,47]. Interestingly, P2-CF showed almost the same 1 O 2 production as P2 in solution. This behavior might be related with an aggregation attenuation or an increased photostability of the porphyrin after its immobilization on the chitosan support.
The photostability studies were performed with P2-P4 in PBS, monitoring the decrease in absorbance of the corresponding Soret bands after different times of irradiation with a LED array at an irradiance of 10 mW·cm −2 . P4 suffered a decay of 64%, followed by P2 (41%) and P3 (22%) during the irradiation period ( Figure S22). Based on the observed degradation profiles, the studies were repeated in the absence of light, to verify if the decrease was not associated with an aggregation phenomenon. In fact, P4 showed a decrease of 41%, P2 (28%) and P3 (22%) in dark conditions, showing that somewhere up to a 23% decrease, the decrease is related to the photodecomposition of the macrocycle. In sum, P2 and P4 are more susceptible to photodegradation upon illumination, while P3 was quite photostable over the irradiation period investigated (30 min).

Photoinactivation Studies
It is commonly accepted that the success of photoinactivation closely depends on the structure of the photosensitizing agent [14]. As previously stated, the charge of the PS is extremely important for the inactivation of the microbial agents and, generally, positively charged PS are more efficient for Gram-negative strains, even if they are at low concentrations when compared to neutral or negatively charged PS [52]. Studies with cationic PS demonstrate that these derivatives are more likely to inactivate both Gram-positive and Gram-negative bacteria without the help of membrane disrupting agents [47,76].
The biological assays started by evaluating the photodynamic action of the new cationic porphyrins P3 and P4 at the concentration of 5.0 µM, under white light towards the recombinant bioluminescent E. coli ( Figure 9). As mentioned above, this Gram-negative bacterium is an excellent bacterial model to monitor the effectiveness of a photoinactivation process, since its light output is a highly sensitive reporter of its metabolic activity. These set of experiments were not performed in the presence of the non-immobilized P2 since its efficacy has already been studied by some of us [46,47]. porphyrins P3 and P4 at the concentration of 5.0 µ M, under white light towards the recombinant bioluminescent E. coli ( Figure 9). As mentioned above, this Gram-negative bacterium is an excellent bacterial model to monitor the effectiveness of a photoinactivation process, since its light output is a highly sensitive reporter of its metabolic activity. These set of experiments were not performed in the presence of the non-immobilized P2 since its efficacy has already been studied by some of us [46,47].  The results summarized in Figure 9 show that the bioluminescence of E. coli was not affected by light in the absence of the PS (light control) neither by the direct effect of any of the tested PS in the absence of light (dark control). These results indicate that any variation in the bioluminescence was due to the photodynamic effect of the PS. Photosensitizers P3 and P4 showed significant differences in terms of inactivation efficiency (ANOVA-p < 0.05) (Figure 9). P3 at a concentration of 5.0 µM was more effective than P4, causing a reduction of approximately 3.6 log in the bioluminescence, after only 30 min of irradiation. At this concentration, P3 inactivated completely E. coli (decrease of the bioluminescent signal below the quantification limit, 2 log) after 120 min of irradiation (with a total light dose of 21.6 J·cm −2 ) whilst P4 caused an inactivation of 3 log after 120 min of irradiation. Compound P3 can be considered a good PS against E. coli, since it showed a similar efficiency as P2 towards E. coli at the same concentration (5.0 µM) and under the same light irradiance [47]. The difference observed for each PS can be justified by a balance between the efficiency to produce 1 O 2 (P2~P3 > P4) and their affinity to the bacteria which is more dependent on the PS structure.
Using different approaches (e.g., scanning electron microscopy (SEM), transmission electron microscopy (TEM), spectrophotometry, 1 D electrophoresis, atomic force microscopy, proteomic, lipidomic, infrared spectroscopy) to evaluate the ultrastructural, morphological and functional changes in bacterial cells, at initial stages and during the course of the PDI process, it seems that the most affected cellular constituents in E.coli are external targets, namely proteins and lipids, which can affect normal cell membrane functionality or lead to significant cell membrane disintegration [77][78][79][80].
The results obtained for the photoinactivation efficacy of the new conjugates towards the bioluminescent E. coli were evaluated and are compiled in Figure 10 (Px-CF) and Figure 11 (Px-TiO 2 ). It is known that chitosan can act as an antimicrobial agent, although Gram-negative bacteria are less susceptible to its effects [81]. Figure 10 shows for P2-CF and after 90 min of irradiation, a noticeable decrease of the bioluminescent signal to approximately 4.1 log. However, no significant differences were detected in the bioluminescent signal when the bacteria were exposed to P3 and P4 supported on chitosan (decrease of 0.63 and 0.31 log, respectively).
These results are most certainly related to the inability of these porphyrins to produce oxygen singlet, after being immobilized in chitosan. The porphyrins photodynamic performance in solution follows the order P2~P3 > P4; however, when supported on chitosan films, only P2-CF, with the capacity to generate oxygen singlet, has a significant effect on the bacterial cells. Studies show that chitosan might have either bactericidal or bacteriostatic properties, depending on the type of bacterium, which is related to the thickness of the cell wall. Recent data shows that chitosan tends to act as bacteriostatic rather than bactericidal, which can also explain the visible changes in the dark controls [82]. affect normal cell membrane functionality or lead to significant cell membrane disintegration [77][78][79][80].
The results obtained for the photoinactivation efficacy of the new conjugates towards the bioluminescent E. coli were evaluated and are compiled in Figures 10 (Px-CF) and 11 (Px-TiO2). It is known that chitosan can act as an antimicrobial agent, although Gram-negative bacteria are less susceptible to its effects [81].  Figure 10 shows for P2-CF and after 90 min of irradiation, a noticeable decrease of the bioluminescent signal to approximately 4.1 log. However, no significant differences were detected in the bioluminescent signal when the bacteria were exposed to P3 and P4 supported on chitosan (decrease of 0.63 and 0.31 log, respectively).
These results are most certainly related to the inability of these porphyrins to produce oxygen singlet, after being immobilized in chitosan. The porphyrins photodynamic performance in solution follows the order P2 ~ P3 > P4; however, when supported on chitosan films, only P2-CF, with the capacity to generate oxygen singlet, has a significant effect on the bacterial cells. Studies show that chitosan might have either bactericidal or bacteriostatic properties, depending on the type of bacterium, which is related to the thickness of the cell wall. Recent data shows that chitosan tends to

Materials and Methods
All analytical grade chemicals used in this study were purchased from Sigma-Aldrich or Merck. Chitosan was purified following the procedure described in the literature [28]. H2TPP was synthesized as described in the literature [85]. TiO2 was Evonik-Degussa P25. 1 H, 19 F and 13 C solution NMR spectra were recorded on a Bruker Avance 300 (300.13 and 282.38 MHz, respectively), 500 (125.76 MHz for 13 C) spectrometer. DMSO-d6 was used as solvent and tetramethylsilane (TMS) as the internal reference; the chemical shifts are expressed in ppm and the coupling constants (J) in Hertz (Hz). Unequivocal 1 H assignments were made using 2D COSY ( 1 H/ 1 H). Electronic spectra (UV-Vis) were obtained on a Shimadzu UV-2501PC spectrophotometer, in the 350-800 nm range. Spectra of the solid samples were recorded on a Jasco V-560 spectrophotometer with reflectance accessory (JASCO ISV-469) in absorbance mode in the 250-850 range. Attenuated Total Reflectance Transmission Fourier Transform Infrared (ATR-FTIR) spectra were registered on a FTIR Bruker Tensor 27 spectrophotometer with ATR accessory ATR (ATR Golden Gate Diamond, Specac) in the 350-4000 cm⁻ 1 range; the spectra were collected with a resolution of 4 cm −1 and accumulation of 256 scans. For the X-ray diffraction (PXRD) measurements, self-oriented solids were placed on neutral glass sample holders. The measurements were performed in the reflection mode using a Panalytical Empyrean diffractometer (Cu Kα1,2 X-radiation, 1 =1.540598 Å and 2 =1.544426 Å ), equipped with an X'Celerator detector and a flat-plate sample holder in a Bragg-Brentano parafocusing optics configuration (40 kV, 50 mA). Intensity data were collected by the step counting method (step 0.02) in continuous mode. The fluorescence emission spectra were recorded in DMF in 1 × 1 cm quartz optical cells under normal atmospheric conditions on a computer-controlled Horiba Jobin Yvon FluoroMax-3 spectrofluorimeter. The widths of both excitation and emission slits were As shown in Figure 10, neither P3 nor P4 immobilized on chitosan are good photosensitizers. P3 and P4, when combined with chitosan, show almost no effect on the cells. Durantini and co-workers showed that chitosan was toxic to E. coli by itself at high concentrations (0.75 or 1.00 mg/mL) [81]. The concentration used in this work was 0.34 mg/mL and, as expected, no cytotoxic effect was observed for E. coli in the dark. Similar results were observed for P2-CF at a concentration of 3.0 µM after a total light dose of 10.8 J·cm −2 . The porphyrins 5,10,15,20-tetrakis(4-aminophenyl)porphyrin (p-TAPP) and TMPyP incorporated into chitosan showed a considerable bactericidal effect under irradiation at 590 nm (2 log reduction after 120 min for p-TAPP and 4 log reduction after 140 min for TMPyP) [40]. The best results were obtained when P2 was incorporated into chitosan. The similarities seen in the UV-Vis spectra before and after the PDI biological assays show that PS-CF materials were photostable ( Figure S23).
The positive effect of chitosan was also indicated when Luksiene and co-workers studied the photodynamic inactivation of the Gram (−) food pathogen Salmonella enterica using the chlorophyllin-chitosan hybrid as PS [30]. The authors verified that the inactivation caused by chlorophyllin (1.8 log, at 15 µM, light dose 38 J·cm −2 ) and chitosan (ca 1.8 log) separately gave rise in the presence of photoactivated Chl-CHS complex to a remarkable and swift decrease of Salmonella enterica viability (7 log) [30]. Considering that the inactivation of E. coli was achieved at lower PS concentration (3.0 µM vs 15 µM) and light dose (16.2 vs 38 J·cm −2 ), we can conclude that the new hybrid P2-CF is also a very promising material. No cytotoxic effect observed for chitosan alone can be justified to the low concentrations used since high concentrations (0.75 or 1.00 mg/mL) are required for E. coli inactivation, as showed by Durantini [81].
As stated before, TiO 2 can be used as PS for the inactivation of some pathogens under ultraviolet light irradiation. The electronic properties of TiO 2 can be easily modulated by linking a PS to its surface and the immobilization of the PS induces an extension of the absorption profile of TiO 2 enabling the use of the visible light [83]. Macyk and co-workers showed that a TiO 2 suspension did not result in any significant E. coli photoinactivation under visible light [84]. When P3 and P4 are supported on TiO 2 , it is noticeable that P3 and P4 at a concentration of 10 µM cause a reduction of approximately 2 log in the E. Coli bioluminescence after 180 min (32.4 J/cm 2 ) ( Figure 11). P3-TiO 2 and P4-TiO 2 have a higher effect on the bacterial cells compared to P3-CF and P4-CF. Compared to PS-TiO 2 , the 1 O 2 production from these PS-CF was much less efficient and could be disregarded.
TiO 2 -graphene functionalized with 5,10,15,20-tetrakis(4-carboxyphenyl)porphyrin showed a decrease of 64% in the concentration of viable cells of E. coli after 440 min under light irradiation (450 W xenon lamp) [27]. In this study, both porphyrins P3 and P4 supported on TiO 2 showed a decrease of 99% in the concentration of viable cells of E. coli after 180 min.
Conversely, higher amounts of PS-TiO 2 did not increase the photodynamic effect. In fact, the increase of material as a powder in these tests can obstruct light from reaching the PS, hampering the photosensitization process.

Materials and Methods
All analytical grade chemicals used in this study were purchased from Sigma-Aldrich or Merck. Chitosan was purified following the procedure described in the literature [28]. H 2 TPP was synthesized as described in the literature [85]. TiO 2 was Evonik-Degussa P25. 1 H, 19 F and 13 C solution NMR spectra were recorded on a Bruker Avance 300 (300. 13  spectrofluorimeter. The widths of both excitation and emission slits were set at 2.0 nm. The fluorescence quantum yield (ϕ F ) of porphyrins was calculated in DMF by comparison of the area below the corrected emission spectra using H 2 TPP as standard (λ exc at 420 nm, ϕ F = 0.11 in DMF) [64]. Fluorescence spectra of the solid samples were recorded using a fiber optic system connected to a Horiba-Jobin Yvon FluoroMax-3 spectrofluorimeter for appropriately λ (nm) of the solid compounds.

Synthesis of Porphyrin Derivatives
Porphyrin P1 was obtained by the reaction of the appropriate aldehydes in acetic acid and nitrobenzene, according to a literature procedure (see Scheme 1) [59]. The purity was confirmed by TLC and by 1 H and 19 F NMR spectroscopy.
Porphyrin P2. To a solution of P1 (0.106 mmol, 0.075 g) in N,N-dimethylformamide (DMF; 1 mL), an excess of methyl iodide (0.2 mL) was added in a sealed tube. The mixture was stirred overnight at 40 C. After cooling, diethyl ether was added to precipitate the porphyrin. The solid was isolated by filtration and washed with diethyl ether. The desired porphyrin was crystallized from CH 2 Cl 2 :CH 3 OH/hexane to yield P2. After that, the solid was dried under vacuum and then characterized. Porphyrin P4. 4-Mercaptobenzoic acid (0.03 mmol, 0.004 g) was dissolved in DMF containing 0.5 mL of pyridine. The reaction mixture was kept under stirring at ambient temperature for 30 min. Afterwards, porphyrin P2 (0.03 mmol, 0.031 g) was added, and the reaction mixture was stirred for 1 h at ambient temperature. Then, diethyl ether was added to precipitate the porphyrin, the solid was filtered and washed with diethyl ether. The solid was dissolved in methanol and, after concentration, it was re-precipitated by the addition of acetone.

Immobilization of Porphyrins on Chitosan
The porphyrin-chitosan (Px-chitosan where x = 2, 3 or 4) solids was prepared in two steps: (i) 250 mg of chitosan were dissolved in 7.0 mL of 1% (v:v) aqueous acetic acid. Afterwards, porphyrins P2, P3 or P4 (5.0 mg) were added and the mixtures were stirred for 24 h at 30 • C. The solvent was evaporated under reduced pressure and the resulting solids were filtered and exhaustively washed with water and DMF. The solutions resulting from the washing process were analyzed by UV-Vis in order to quantify the Px that leaked from the chitosan and the loading of the porphyrin in each solid was calculated. The solids were dried in the oven at 55 • C for 72 h.
(ii) Each sample obtained (about 100 mg) was dissolved in 10 mL of 1% (v:v) aqueous acetic acid. The suspension was maintained at 40 • C under stirring until complete solubilization of the solid (3 h). The solution was poured into molds (5.0 × 5.0 cm) and allowed to dry at room temperature for 72 h. The films were denominated as P2-CF, P3-CF and P4-CF and characterized by ATR-FTIR, UV-Vis, PXRD and fluorescence of solid samples.

Immobilization of Porphyrins on Titanium Dioxide
To a solution of P2, P3, or P4 in methanol (0.789 µM, 10 mL) 0.5 g of TiO 2 were added. The resulting suspension was stirred at ambient temperature for 6 h. The solids were isolated by centrifugation, washed with methanol and dried in the oven at 55 • C for 24 h. UV-Vis analysis of the solutions resulting from the washing process was performed and the loading of the porphyrin in each solid was calculated. The solids were named as P3-TiO 2 and P4-TiO 2 and characterized by the techniques described above. P2 was completely removed after the washing process of P2-TiO 2 , so this hybrid was not obtained.

Singlet Oxygen Generation and Photostability
In a 1 cm x 1 cm quartz cell, and to a solution of DPiBF at 50 µM, 0.5 µM of each PS in DMF were added and irradiated with a LED array at an irradiance of 10 mW cm −2 . The LED array is composed of a matrix of 5 × 5 LEDs making a total of 25 light sources with an emission peak centered at 640 nm and a bandwidth at half maximum of ± 20 nm. Irradiation was conducted at ambient temperature under stirring. The decreasing of DPiBF was monitored by UV-Vis at 415 nm at intervals of 1 min during 15 min. The absorption decay is proportional to the production of 1 O 2 . For the Px-CF and Px-TiO 2 materials, a similar protocol was adopted, but the PS concentration used was 2.0 µM and the monitorization by UV-Vis was performed after centrifugation of the suspension irradiated and only the times 0, 5, 10 and 15 min were registered. Controls were also included. The experiments were performed in duplicate. H 2 TPP, an efficient singlet oxygen generator, was included as a reference.
The photostability studies of the porphyrin derivatives were performed by irradiation of a solution of each porphyrin in DMF (5.0 µM) with a LED array. The solutions were stirred and kept at ambient temperature. The stability of each porphyrin derivative was verified by UV-Vis at regular intervals up to 60 min.

Bacterial Strain and Culture Growth Conditions
The efficiency of the porphyrins in solution and after being incorporated on the materials to photoinactivate a bioluminescent E. coli was evaluated using a reliable and quick method that allows the monitorization of the activity of this bacterial strain in real time. Bioluminescent E. coli was transformed as described in a previous work [46] and stored at −80 • C in 10% glycerol. For the transformation, the plasmids pHK724 and pHK555 were inserted into competent cells of E. coli Top 10 (Invitrogen, USA), resulting in a bioluminescent strain. These plasmids contain the luxoperon from the bioluminescent marine bacterium Vibrio fischeri, required to produce light without the addition of exogenous substrates [46]. Before each assay, an aliquot of bioluminescent bacteria stored at −80 • C in 10% glycerol was aseptically plated on tryptic soy agar (TSA, Merck) supplemented with 50 mg·mL −1 of ampicillin (Amp) and with 34 mg·mL −1 of chloramphenicol (Cm), and stored at 4 • C. Before each assay, one isolated colony was aseptically transferred to 10 mL of tryptic soy broth (TSB, Merck) medium previously supplemented with Amp and Cm and was grown overnight at 25 • C under stirring (120 rpm). Afterwards, an aliquot was transferred into 10 mL TSB under the same growth conditions to reach stationary growth phase. An optical density at 600 nm (OD 600 ) of 1.6 ± 0.1 corresponded to ≈10 8 colony forming units (CFU) mL −1 [46].

Irradiation Conditions for Photosensitization
All samples were exposed, in parallel, to white light (PAR radiation, 13 OSRAM 21 lamps of 18 W each, 380 700 nm) with a fluence rate of 3.0 mW cm −2 (measured with a light meter LI-COR model LI-250, Li-Cor Inc., Lincoln, NE, USA), at ambient temperature of less than 25 • C under 100 rpm mechanical stirring.

Photosensitization Procedure
Bacterial cultures mentioned above were grown overnight and then tenfold diluted in phosphate buffered saline (PBS), at pH 7.4, so that~10 7 relative light units (RLUs) were reached. Five milliliters of the bacterial suspension were evenly distributed in glass beakers (previously sterilized) and the appropriate volume of the PS was added in order to attain final concentrations of 5.0 µM. For the immobilized PS, the corresponding mass of PS-CF or PS-TiO 2 was added in order to achieve the final concentration of 3.0 or 10 µM, respectively. For the non-immobilized PS, the suspensions were incubated for 15 min in the dark to allow the binding of the PS to the bacterial cells, and subsequently exposed to irradiation conditions, during a maximum period of 120 min. For the materials prepared, the dark incubation time was 60 min and after which they were exposed to the same irradiation conditions, during a maximum period of 90 min and 180 min for Px-CF and Px-TiO 2 , respectively.
Dark and light controls were included in the experiments. In the light control, the bacterial suspensions without PS were exposed to the same irradiation protocol as the test suspensions. In the dark control, the PS was added and protected from light with aluminum foil. At the initial time, and after established times of irradiation, the bioluminescence was measured in the luminometer (GloMax ® 20/20 Luminometer, Promega, Madinson, WI, USA). The percentage of decay of the bioluminescence is proportional to the concentration of viable cells (CFU). Three independent experiments with three replicates were performed and the results were averaged.

Statistical Analysis
Statistical analysis was performed in GraphPadPrism 7.0. The significance of the PDI effect of each PS and of the irradiation time on bacterial cell viability was assessed by an unvaried analysis of variance (ANOVA) model with the Bonferroni post hoc test. Normal distributions were assessed by the Bartlett's test and homogeneity of variances was assessed by the Brown-Forsythe test. A value of p < 0.05 was considered significant.

Conclusions
We successfully accomplished the chemical modification of the 5,10,15-tris(1-methylpyridinium-4yl)-20-(pentafluorophenyl)porphyrin. Moreover, two cationic porphyrins were obtained as a result of the substitution of the p-fluorine atoms on the meso-phenyl with thio-carboxylate acids. These cationic porphyrins were successfully immobilized on two different supports: chitosan and titanium oxide. The efficiency of the aPDT process against E. coli seems dependent on the structure of the porphyrin and on its ability to produce singlet oxygen. The results showed that the choice of the support and of the adequate porphyrin can be modulated to improve the efficiency of aPDT. The electronic properties of TiO 2 allow linking the PS to its surface, increasing the absorption profile of TiO 2 , which enabled the use of the visible light, inactivating bacteria more efficiently than the PS-CF.