3D Porous Collagen Matrices—A Reservoir for In Vitro Simultaneous Release of Tannic Acid and Chlorhexidine

The treatment of wounds occurring accidentally or as a result of chronic diseases most frequently requires the use of appropriate dressings, mainly to ensure tissue regeneration/healing, at the same time as treating or preventing potential bacterial infections or superinfections. Collagen type I-based scaffolds in tandem with adequate antimicrobials can successfully fulfill these requirements. In this work, starting from the corresponding hydrogels, we prepared a series of freeze-dried atelocollagen type I-based matrices loaded with tannic acid (TA) and chlorhexidine digluconate (CHDG) as active agents with a broad spectrum of antimicrobial activity and also as crosslinkers for the collagen network. The primary aim of this study was to design an original and reliable algorithm to in vitro monitor and kinetically analyze the simultaneous release of TA and CHDG from the porous matrices into an aqueous solution of phosphate-buffered saline (PBS, pH 7.4, 37 °C) containing micellar carriers of a cationic surfactant (hexadecyltrimethylammonium bromide, HTAB) as a release environment that roughly mimics human extracellular fluids in living tissues. Around this central idea, a comprehensive investigation of the lyophilized matrices (morpho-structural characterization through FT-IR spectroscopy, scanning electron microscopy, swelling behavior, resistance against the collagenolytic action of collagenase type I) was carried out. The kinetic treatment of the release data displayed a preponderance of non-Fickian–Case II diffusion behavior, which led to a general anomalous transport mechanism for both TA and CHDG, irrespective of their concentrations. This is equivalent to saying that the release regime is not governed only by the gradient concentration of the releasing components inside and outside the matrix (like in ideal Fickian diffusion), but also, to a large extent, by the relaxation phenomena of the collagen network (determined, in turn, by its crosslinking degree induced by TA and CHDG) and the dynamic capacity of the HTAB micelles to solubilize the two antimicrobials. By controlling the degree of physical crosslinking of collagen with a proper content of TA and CHDG loaded in the matrix, a tunable, sustainable release profile can be obtained.


Introduction
An almost ideal wound dressing must meet some mandatory features: ease of application, bioadhesiveness to the wound surface, absorption of excess exudates, maintenance of and hexadecyltrimethylammonium bromide (HTAB) from Fluka, CHDG (20% aqueous solution, by weight) from Caesar & Loretz GmbH (Hilden, Germany), and ethanol (ACS reagent absolute, ≥99.5%) from EMPARTA were used as received. All the other reagents and chemicals were purchased from Reactivul-Bucharest, Romania and utilized without further processing.

Preparation of Collagen Hydrogels and Matrices
Acidic collagen hydrogels (pH of about 4) with and without TA and CHDG were prepared as described elsewhere [56]. Briefly, a series of nine collagen hydrogels (pH 4) of 1.10% collagen and different contents of TA and CHDG (by taking into account all combinations of 5, 10, 15% TA and 1. 82, 4.55, 9.09% CHDG, the concentrations being considered with respect to the collagen) were prepared by adding proper amounts of aqueous solutions of TA and/or CHDG to a certain quantity of initial collagen hydrogel (2.50% and pH 3.20) under gentle stirring. To maintain the overall pH of the systems at ca. 4 during the collagen-based hydrogels preparation, a solution of 1M NaOH was used. A series of three collagen hydrogels containing TA (5, 10, and 15%) for comparison were also prepared. All the collagen-based hydrogels were equilibrated by keeping them at 4 • C for 24 h. The final porous collagen-based matrices were obtained by freeze-drying collagenbased hydrogels (poured into molds with a flat bottom and open to the air side on top) for 48 h, employing a Delta 2-24 LSC (Martin Christ, Osterode am Harz, Germany) freeze-dryer. A detailed algorithm of the entire lyophilization process was described previously [57].

Matrices Characterization 2.3.1. FT-IR Measurements
Structural characterization of collagen from collagen-based matrices was performed by FT-IR using a JASCO FT-IR 4200 spectrometer equipped with a TGS detector and Spectra Manager software. Data were acquired by ATR technique using a MKII Golden Gate 45 degree diamond crystal with a horizontal ATR plate. All the spectra were corrected for ATR effect and transformed into absorption spectra. Each spectrum is the average of 30 scans, with a resolution of 4 cm −1 .

Morphology Investigations
Morphologies of dry porous collagen-based matrices were examined by scanning electron microscopy. A metal coating applied to a freshly sectioned area of a matrix makes the sample conductive and ready to expose to the electron beam of the microscope gun with no significant damages. The metal deposition (Au-99.99%, normal incidence) step took place by metal vaporization under high vacuum (by resistive heating of a tungsten filament under a high vacuum of 5 × 10 −6 torr) inside an evaporator Jeol-JEE 4C, as described in another paper [58]. The SEM images were acquired by employing a microscope Tescan Lyra 3 XMU FEG at an acceleration voltage of 5 kV.

UV-VIS Absorption Spectroscopy and Steady-State Fluorescence Measurements
In order to quantify the amounts of TA and CHDG released as a function of time, UV-VIS absorption spectra were recorded by using a Jasco-V660 spectrophotometer equipped with a thermostating module.
Steady-state fluorescence of pyrene was considered to be useful in determining the critical micelle concentration (cmc) of HTAB in PBS (sodium phosphate buffer-10 mM, NaCl-0.154 M pH 7.4) at 37 • C. Thus, the emission spectra of pyrene for different HTAB concentrations in the abovementioned conditions were recorded on a Jasco FP-6300 spectrofluorometer also equipped with a thermostating module (operating parameters: band width-5 nm; data pitch-0.5 nm; scanning speed-100 nm/min; spectrum accumulation-3; path length-10 mm using a Quartz SUPRASIL cell). Due to HTAB-based composition of the release environment, all absorption spectra were acquired by using a solution of 3 × 10 −4 M HTAB in sodium phosphate buffer PBS (PBS-HTAB, pH 7.4) at 37 • C as reference.

Buffer Uptake into the Collagen-Based Matrices
Buffer (PBS-HTAB) uptake measurements were performed at 37 • C and the results were expressed as the ratios between the mass of the swollen-in buffer samples and the initial mass of the corresponding freeze-dried matrices (g/g). The weighing of the swollen samples was performed every 5 min within the first 30 min, then every 15 min for 4-5 h. The final weighing was made after 24 h, which practically corresponded to the equilibrium swelling of the samples. Three replicas were used for each of the collagen-based matrices investigated. The data were obtained as mean values ± relative standard deviation (RSD). RSD values were less than 15% for all the experimental data.

Resistance to Collagenase Digestion
The time course of enzymatic degradation of collagen was followed using collagenase type I (from Clostridium histolyticum) in physiological conditions: PBS-HTAB of pH 7.4 at 37 • C. Pieces of 1 × 1 cm porous collagen-based matrices immersed in 1mg/mL collagenase solution were monitored for 9 days. The aspect of solutions and matrix pieces were noticed through visual inspection.

In Vitro Release of TA and CHDG
In vitro release of TA and CHDG contained within the collagen-based matrices was performed by employing USP 2 equipment (paddle apparatus). A cylinder-shaped sample (porous matrix) was immobilized at ca. 3-4 mm above the vessel base. To ensure that the swelling of matrix and releasing of TA/CHDG occurred only by the lateral surface of the sample as a disc-shaped slice (cylindrical shape), both the bottom and upper faces of the matrix were covered by a thin paraffin wax layer (resulting from solvent evaporation of a proper solution of paraffin in petroleum ether). The release medium consisted of 1 L of 3 × 10 −4 M HTAB in PBS (PBS-HTAB, pH 7.4) at 37 • C and a stirring speed of 50 rpm. The fraction of TA/CHDG release was calculated with respect to their amount initially contained within the dry collagen-based matrices. Measurements (performed over a time scale of 11 h) at each time point took place in triplicate and mean and RSD values were calculated. RSD values for all these experimental data were found to be less than 10%.

FT-IR Spectroscopy Analysis of Freeze-Dried Collagen-Based Matrices
The IR absorption bands typically considered to be specific to the peptide functional groups in collagen are amide A (3330-3325 cm −1 ), amide B (3080 cm −1 ), amide I (1650 cm −1 ), and amide II (1550 cm −1 ) [59,60]. Many other published results [61][62][63][64][65][66] take into account the amide III band in characterizing the collagen microstructure. FT-IR spectra discussed below concern the matrices of collagen alone on the one hand, and four series of three collagen-based samples (one series of 5, 10, and 15% TA with no CHDG and three series containing 5, 10, and 15% TA, each of them with the same content of CHDG-1.82, 4.55, and 9.09%) on the other hand (Figures 1 and S1-S3). The assignment data for all the samples investigated are collected in Tables 1 and S1 (in Supplementary Materials).
integrity of collagen triple helix may be empirically confirmed by considering either the difference in wavenumbers of the amide I and amide II bands (νamideI-νamideII) [73], that has to be lower than 100 cm −1 , or by the intensity ratio of the amide III band to that located at 1450 cm −1 (AamideIII/A1450), that has to be close to unity [61,62,[74][75][76]. Accordingly, IR spectra of all collagen-based matrices indicate a nondenatured, native state of collagen: νamideI-νamideII = 85-95 cm −1 and AamideIII/A1450 = 0.9 (see Table 1). Concerning the collagen-TA matrices, due to TA addition, the amide A band shifted to lower wavenumbers, suggesting an enhancement of hydrogen bonding most likely between collagen (NH of peptide groups) and TA (hydroxyl phenolic groups) ( Table 1, Figure S1a). A similar red-shifting tendency is obeyed by the amide II and amide III bands (Table 1, Figure 1), which supports the idea of increasing hydrogen bonding between collagen and TA in these matrices. Moreover, some peculiar features in the IR spectra of collagen have been identified with increasing TA content. Thus, the absorption band centered at ca. 1335 cm −1 , ascribed to in plane OH bending [77] (associated with tyrosine residues), ascends in its relative intensity and broadness as TA concentration increases.  Table 1. Characteristic infrared bands for the collagen-based matrices with TA (5, 10, 15%) and TA (5, 10, 15%)-CHDG (9.09%).

TA-10% CHDG-9.09%
Collagen TA-15% CHDG-9.09% The main IR absorption bands associated with the polypeptidic structure of collagen are clearly revealed in its FT-IR spectrum ( Figure 1 and Figure S1a). Thus, the amide A band is centered at 3301 cm −1 and is considered to be arisen from NH stretching vibrations. This is a valuable indicator of the presence of hydrogen bonds between the collagen chains in the matrix because the free NH vibrations generally take place at 3400-3440 cm −1 and the band position experiences a red shifting when the NH part of a peptide linkage is involved in hydrogen bonding [60][61][62][63][64][67][68][69][70]. Even though the amide I band is sensitive to the secondary structure of a polypeptide [62,63,71,72], much more information can be obtained taking into account the amide I, II, and III bands together. In this respect, preserving the integrity of collagen triple helix may be empirically confirmed by considering either the difference in wavenumbers of the amide I and amide II bands (ν amideI -ν amideII ) [73], that has to be lower than 100 cm −1 , or by the intensity ratio of the amide III band to that located at 1450 cm −1 (A amideIII /A 1450 ), that has to be close to unity [61,62,[74][75][76]. Accordingly, IR spectra of all collagen-based matrices indicate a nondenatured, native state of collagen: ν amideI -ν amideII = 85-95 cm −1 and A amideIII /A 1450 = 0.9 (see Table 1).

Assignment
Concerning the collagen-TA matrices, due to TA addition, the amide A band shifted to lower wavenumbers, suggesting an enhancement of hydrogen bonding most likely between collagen (NH of peptide groups) and TA (hydroxyl phenolic groups) (Table 1, Figure S1a). A similar red-shifting tendency is obeyed by the amide II and amide III bands (Table 1, Figure 1), which supports the idea of increasing hydrogen bonding between collagen and TA in these matrices. Moreover, some peculiar features in the IR spectra of collagen have been identified with increasing TA content. Thus, the absorption band centered at ca. 1335 cm −1 , ascribed to in plane OH bending [77] (associated with tyrosine residues), ascends in its relative intensity and broadness as TA concentration increases. This is reflected by the evolution of the A amideIII /A 1335 ratio from 1.3 to 1.1 when the TA amount varies from 0 to 15% and arises from the corresponding OH phenolic groups (some of them belonging to the pentagalloylglucose core and the others carried by the five galloyl residues within the outer zone of each TA molecule, Figure 2), occurring in the samples. Similarly, the relative intensity of the band located at about 1200 cm −1 assigned to C-O stretching vibrations (C-OH in phenol) [73] follows the same ascendant course as TA concentration increases (A amideIII /A 1200 is 1.48 for collagen and 0.95 for collagen with 15% TA). At the same time, the IR bands centered at 1080 and 1035 cm −1 assigned to C-O stretching vibrations of carbohydrate moieties [64,72,78] enhance in their relative intensities when TA content rises. This is due to the progressively increasing contribution of the ester groups in TA to IR signals at those wavenumbers (C-O-C symmetrical stretching) that overlap the specific bands of carbohydrate moieties bound to collagen.
To assess the simultaneous presence effect of CHDG-TA (at the maximum content of CHDG) on the secondary structure of collagen and the hydrogen bonding in CHDG-TA-collagen systems, a series of three matrices with the 9.09% CHDG and various TA concentrations (5, 10, 15%) were subjected to FT-IR investigations. Surprisingly, neither amide A band, nor amide II and amide III bands indicate supplementary enhancement in hydrogen bonding between collagen molecules themselves and collagen-TA/CHDGs. On the contrary, the blue-shifting in the amide A band for the CHDG-TA-collagen-based matrices with 5 and 10% TA confirm a little diminishment in collagen-collagen hydrogen bonding (Table 1). This does not exclude the possible interactions through hydrogen bonding between collagen and TA/CHDG that may imply types of collagen functional groups other than peptide linkages: −COOH of glutamic and aspartic acid residues, −OH of tyrosine, serine, and threonine. These hydrogen bonds may also be accompanied by hydrophobic interactions mainly established between galloyl residues of TA and the hexamethylene bridge of CHDG on the one hand, and these parts of TA/CDDG molecules and some hydrophobic regions of collagen [18,19]. Therefore, such combined hydrogen bond/hydrophobic interactions could play a significant role in the physical crosslinking of collagen in the matrices. Otherwise, the same picture regarding the changes in IR patterns of collagen-TA matrices (at ca. 1335, 1200, 1080, and 1035 cm −1 ) was also found for the collagen-TA-CHDG samples ( Figure S1). However, due to the specific vibrations of guanidine groups of CHDG [68,79] (C=N symmetrical stretchings at ca. 1630 cm −1 overlapped by the amide I band and C−N stretching at 1235 cm −1 overlapped by the amide III band of collagen), the relative intensities taken as the A amideIII /A 1200 ratio were found to vary from 1.48 for the collagen alone sample to 1.01 for the TA-CHDG-collagen with 15% TA. FT-IR spectra for the rest of the collagen-TA-CHDG matrices exhibit very similar features as those discussed above and can be readily seen in Table S1 and Figures S1-S3.
Pharmaceutics 2023, 15, x FOR PEER REVIEW 8 of bridge of CHDG on the one hand, and these parts of TA/CDDG molecules and some h drophobic regions of collagen [18,19]. Therefore, such combined hydrogen bond/hydr phobic interactions could play a significant role in the physical crosslinking of collagen the matrices. Otherwise, the same picture regarding the changes in IR patterns of coll gen-TA matrices (at ca. 1335, 1200, 1080, and 1035 cm −1 ) was also found for the collagen TA-CHDG samples ( Figure S1). However, due to the specific vibrations of guanidin groups of CHDG [68,79] (C=N symmetrical stretchings at ca. 1630 cm −1 overlapped by th amide I band and C−N stretching at 1235 cm −1 overlapped by the amide III band of coll gen), the relative intensities taken as the AamideIII/A1200 ratio were found to vary from 1.48 f the collagen alone sample to 1.01 for the TA-CHDG-collagen with 15% TA. FT-IR spect for the rest of the collagen-TA-CHDG matrices exhibit very similar features as those di cussed above and can be readily seen in Table S1 and Figures S1-S3.

SEM Investigations of Matrices Morphologies
As pointed out previously [5], the porous structure of collagen and collagen-base matrices is a direct result of a negative replica formation of the ice network (in this ca the dendritic crystals of aqueous acetic acid solution in the frozen state) generated durin the first stage of lyophilization, followed, eventually, by the ice sublimation at a low pre sure and temperature set adequately to the freeze-drying process. In this respect, by co trolling the freezing step of the initial collagen-based solutions/suspensions, a large var ety of pore morphologies could be obtained for the final systems. Thus, it was experime tally established that the lower the temperature of freezing, the smaller the pore size, an

SEM Investigations of Matrices Morphologies
As pointed out previously [5], the porous structure of collagen and collagen-based matrices is a direct result of a negative replica formation of the ice network (in this case the dendritic crystals of aqueous acetic acid solution in the frozen state) generated during the first stage of lyophilization, followed, eventually, by the ice sublimation at a low pressure and temperature set adequately to the freeze-drying process. In this respect, by controlling the freezing step of the initial collagen-based solutions/suspensions, a large variety of pore morphologies could be obtained for the final systems. Thus, it was experimentally established that the lower the temperature of freezing, the smaller the pore size, and, depending on the quality of heat transfer in the course of lyophilization, the pore orientation may be altered [5]. Even though all the matrices were obtained simultaneously during the same freeze-drying procedure, SEM investigations revealed some differences in porous morphology according to the systems' composition. For the sponges consisting of collagen solely, a well-defined lattice-like lamellar morphology was observed in a cross-section taken perpendicular to the air-sided plan (Figure 3a parallel to the air side of the matrix ( Figure S4). Such morphologies are fairly common and have been also reported [5,80,81] for collagen/collagen-based matrices obtained via freezedrying. Instead, the presence of TA at concentrations of 10 and 15% brings some disorder degree both in lamellar and pore orientation (see Figure 3b and Figure S4b) of the collagen-TA sponges. However, this feature of "random" pore orientation is actually an advantage of using this type of matrix for soft tissue (such as skin) regeneration purposes [5]. At the same time, the pore size distribution seems to be far broader in these collagen-TA systems compared to the collagen matrices (except for the collagen-TA (5%) mixtures that look like similar, morphologically speaking, to the collagen ones). By introducing the other low molecular weight compound (CHDG), the morphology of the three-component matrices has progressively changed in the sense that the characteristic lattice-like lamellar structure and pore orientation have been partially lost with an increasing the CHDG percentage. While the systems of collagen-TA (5%)-CHDG (1.82%) resemble the collagen and collagen-TA (5%) matrices to a great extent, all the other three-component freeze-dried mixtures exhibited quite different morphologies. Particularly, the higher the concentrations of TA and CHDG, the larger the disorder in the lamellar structure and pore orientation. Moreover, a high content of TA and CHDG partially induced the pore collapse ( Figure S4c), in addition to a visible disorder in the lattice-like lamellar morphology (Figure 3c) of these three-component sponges. Practically, the most affected matrices from this point of view were the collagen-TA (15%)-CHDG (9.09%) ones. On the other hand, all the matrices investigated revealed a porous aspect ion the surface in contact with the molding support and a relatively smooth air-sided face as a result of pore collapse during freeze-drying (images not shown). These peculiarities were also experimentally demonstrated more than three decades ago [80,82] and, based on that, the reason for why the pan side of a lyophilized collagen matrix should be brought, for example, into direct contact with the skin, particularly when such kind of sponge is intended for wound dressing [80].
tion may be altered [5]. Even though all the matrices were obtained simultaneously during the same freeze-drying procedure, SEM investigations revealed some differences in porous morphology according to the systems' composition. For the sponges consisting of collagen solely, a well-defined lattice-like lamellar morphology was observed in a crosssection taken perpendicular to the air-sided plan (Figure 3a), whereas a network of rather regular open pores (of ca. 200 μm in diameter, on average) was detected when exploring sections parallel to the air side of the matrix ( Figure S4). Such morphologies are fairly common and have been also reported [5,80,81] for collagen/collagen-based matrices obtained via freeze-drying. Instead, the presence of TA at concentrations of 10 and 15% brings some disorder degree both in lamellar and pore orientation (see Figures 3b and S4b) of the collagen-TA sponges. However, this feature of "random" pore orientation is actually an advantage of using this type of matrix for soft tissue (such as skin) regeneration purposes [5]. At the same time, the pore size distribution seems to be far broader in these collagen-TA systems compared to the collagen matrices (except for the collagen-TA (5%) mixtures that look like similar, morphologically speaking, to the collagen ones). By introducing the other low molecular weight compound (CHDG), the morphology of the three-component matrices has progressively changed in the sense that the characteristic lattice-like lamellar structure and pore orientation have been partially lost with an increasing the CHDG percentage. While the systems of collagen-TA (5%)-CHDG (1.82%) resemble the collagen and collagen-TA (5%) matrices to a great extent, all the other three-component freeze-dried mixtures exhibited quite different morphologies. Particularly, the higher the concentrations of TA and CHDG, the larger the disorder in the lamellar structure and pore orientation. Moreover, a high content of TA and CHDG partially induced the pore collapse ( Figure S4c), in addition to a visible disorder in the lattice-like lamellar morphology ( Figure 3c) of these three-component sponges. Practically, the most affected matrices from this point of view were the collagen-TA (15%)-CHDG (9.09%) ones. On the other hand, all the matrices investigated revealed a porous aspect ion the surface in contact with the molding support and a relatively smooth air-sided face as a result of pore collapse during freeze-drying (images not shown). These peculiarities were also experimentally demonstrated more than three decades ago [80,82] and, based on that, the reason for why the pan side of a lyophilized collagen matrix should be brought, for example, into direct contact with the skin, particularly when such kind of sponge is intended for wound dressing [80].

Isotropic Swelling of Collagen-Based Matrices in an Aqueous PBS-HTAB Environment
The remarkable capacity of collagen matrices to retain aqueous medium is due to the numerous hydrophilic moieties of collagen molecules/fibrils, such as amide, amine, carboxyl, and hydroxyl functional groups, as is the case with most polar polymers [83]. Its swelling behavior in aqueous media is very important for the use of collagen matrices as wound dressings, firstly in ensuring the hydration of the tissue surface on which the matrix is applied and the releasing of the drug contained by the matrix placed onto the wound surface. Moreover, swelling with biological fluids is the main compulsory step in the process of matrix degradation.
The swelling measurements were carried out in a PBS-HTAB solution on freeze-dried matrices of collagen solely, collagen with TA (5, 10, 15%), and collagen with TA (5, 10, 15%) and CHDG at its maximum content used (9.09%), obtained from the corresponding acidic hydrogels. The reason why the PBS-HTAB solution was chosen as the swelling solvent is rationally explained later in the section dedicated to TA/CHDG release. The experimental results are graphically shown in Figure 4 and the plotted values were obtained after a correction of the raw data with the amount of PBS-HTAB solution physically retained (by capillarity) into the micrometer-sized pores of the dry matrices. To do that, each porous collagen-based sample intended to be swollen in the PBS-HTAB solution was firstly soaked in anhydrous ethanol (where matrices morphology and volume are practically not affected) in a pycnometer at 20 • C (at which the pycnometer was calibrated) to completely fill the pores. The volume of ethanol retained by the sample (calculated from the measured weight of filling ethanol and its density) was considered to be the same as that of the aqueous solution of PBS-HTAB. This was the first step in assessing the fraction of PBS-HTAB uptake in a matrix corresponding to mechanically filling its pores from the total aqueous solvent retained by the sample during its swelling. In fact, the real swelling data resulted from removing the capillary contribution to reflect the capacity of solvent retaining exhibited by a matrix based only on the hydrophilicity of the sample components and degree of crosslinking. More than that, due to the fact that the swelling measurements were performed at 37 • C by weighing, the evaluation of the capillary contribution to the process and the degree of swelling expressed on the basis of volume fractions (useful in the next section) required the most reliable value of the density of the PBS-HTAB solution at that temperature. Therefore, starting from a well-known change in water density over a given temperature range, and considering the same course of tendency for the temperature dependence of PBS-HTAB density, a density value of 1.093 g/cm 3 at 37 • C was estimated for the buffer solution after an initial experimentally determined value of 1.098 g/cm 3 for the same solution at 20 • C (for specific details, see Figure S5). numerous hydrophilic moieties of collagen molecules/fibrils, such as amide, amine, carboxyl, and hydroxyl functional groups, as is the case with most polar polymers [83]. Its swelling behavior in aqueous media is very important for the use of collagen matrices as wound dressings, firstly in ensuring the hydration of the tissue surface on which the matrix is applied and the releasing of the drug contained by the matrix placed onto the wound surface. Moreover, swelling with biological fluids is the main compulsory step in the process of matrix degradation.
The swelling measurements were carried out in a PBS-HTAB solution on freezedried matrices of collagen solely, collagen with TA (5, 10, 15%), and collagen with TA (5, 10, 15%) and CHDG at its maximum content used (9.09%), obtained from the corresponding acidic hydrogels. The reason why the PBS-HTAB solution was chosen as the swelling solvent is rationally explained later in the section dedicated to TA/CHDG release. The experimental results are graphically shown in Figure 4 and the plotted values were obtained after a correction of the raw data with the amount of PBS-HTAB solution physically retained (by capillarity) into the micrometer-sized pores of the dry matrices. To do that, each porous collagen-based sample intended to be swollen in the PBS-HTAB solution was firstly soaked in anhydrous ethanol (where matrices morphology and volume are practically not affected) in a pycnometer at 20 °C (at which the pycnometer was calibrated) to completely fill the pores. The volume of ethanol retained by the sample (calculated from the measured weight of filling ethanol and its density) was considered to be the same as that of the aqueous solution of PBS-HTAB. This was the first step in assessing the fraction of PBS-HTAB uptake in a matrix corresponding to mechanically filling its pores from the total aqueous solvent retained by the sample during its swelling. In fact, the real swelling data resulted from removing the capillary contribution to reflect the capacity of solvent retaining exhibited by a matrix based only on the hydrophilicity of the sample components and degree of crosslinking. More than that, due to the fact that the swelling measurements were performed at 37 °C by weighing, the evaluation of the capillary contribution to the process and the degree of swelling expressed on the basis of volume fractions (useful in the next section) required the most reliable value of the density of the PBS-HTAB solution at that temperature. Therefore, starting from a well-known change in water density over a given temperature range, and considering the same course of tendency for the temperature dependence of PBS-HTAB density, a density value of 1.093 g/cm 3 at 37 °C was estimated for the buffer solution after an initial experimentally determined value of 1.098 g/cm 3 for the same solution at 20 °C (for specific details, see Figure S5).  Referring back to the data in Figure 4, the collagen matrices displayed the highest solvent uptake (6.80 g/g at 100 min), but they became progressively more gelatinous and, after ca. 120 min, broke apart up into tiny gelatinous pieces dispersed into the aqueous environment. Alternatively, the collagen-based matrices with 5% TA showed a much lower swelling degree. Additionally, the samples retained their physical integrity, becoming gelatinous after a 24 h immersion (buffer uptake at 24 h was 5.66 g/g). This behavior is a direct consequence of collagen crosslinking via TA molecules, especially through hydrogen bonding and even hydrophobic interactions. Increasing TA content to 10% resulted in a significantly lower buffer uptake near to equilibrium (4.30 g/g at 24 h), although the swelling rate was rather similar to that of matrices with 5% TA at the onset of the swelling process (within the first 50 min). Going to 15% TA content, the matrices exhibited almost the same swelling behavior as the samples with 10% TA, with a buffer uptake of 4.11 g/g after 24 h of immersion. The last results support the supposition of a saturated crosslinking of collagen by TA, beyond which the swelling capacity in the PBS-HTAB solution does not change significantly. As partially discussed in the previous section on FT-IR data, TA entities may build up crosslinkages through H-bonding within collagen matrices involving -OH groups linked to the pentagalloyl core and galloyl residues of TA (see Figure 2), and potentially some specific moieties of collagen: side -OH groups of serine and hydroxyproline residues, side -COOH or -COO − groups of aspartic and glutamic acid residues, amine groups of lysine residues, or amide groups of asparagine and glutamine residues [17]. Every single functional group involved may play the role of either hydrogen bond donor or hydrogen bond acceptor. At the same time, the aromatic rings of the galloyl groups of TA structure are considered to be responsible for strong hydrophobic interactions, with nonpolar regions of collagen molecules [18,19] having a direct consequence in an increased hydrothermal stability of TA-treated collagen compared to the TA-free collagen. Thus, computational docking carried out on the TA-collagen-like polypeptide (CLP) system suggested formation of hydrophobic "bonds" between the aromatic rings of TA and glycine, methionine, glutamic acid, serine, arginine, and proline residues on CLP, and a number of the hydrogen bonds between TA and CLP [18].
On the other hand, and quite unexpectedly, all the matrices loaded with TA and with the maximum content of CHDG (9.09%) have shown buffer uptake, with time traces very close to each other irrespective of TA relative amount. At the same time, the values of swelling degrees were found to be 4.16, 4.02, and 4.06 g/g for the TA-CHDG-collagen sponges with TA content of 15, 10, and 5%, respectively, which are practically the same as for that of collagen-based samples loaded with 15% TA (for comparative purposes only, where the capillary contribution dominates, the raw data for the maximum swelling ratios ranged from 111.32 g/g for the collagen matrices, to 92.62 g/g for the most swollen TA-collagen sponges, and to about 68 g/g for the three-component matrices with the maximum content of CHDG). Such behavior proves the existence of the supplementary attractive interactions brought by CHDG over the entire volume of the three-component matrix when the porous collagen network contains lower proportions of TA (5 and 10%). In other words, a saturation degree of crosslinking has been reached for these collagen-based matrices at this content of CHDG (9.09%). In fact, the effect of the chlorhexidine species endowed with so strong a crosslinking ability onto the swelling behavior of the studied porous collagen-based systems could be mainly attributed to two categories of attractive interactions: -Electrostatic interactions between the chlorhexidine dications (CH ++ ) and the collagen entities negatively charged in the swelling solvent (for more details, see the next section), with a substantial enhancement in the general crosslinking of the collagen network, and -So-called Yoshida forces or Yoshida H-bonding [84], as quite strong attractive interactions established between proton-donating groups/molecules and π-electron bases represented by entities with delocalized π-electron clouds, especially aromatics.
Strong sorption of a 4-vinylpyridine-based polymer partially quaternized or of chlorhexidine (CH) on cellulose are two classical examples of how Yoshida forces (between an electron-deficient hydrogen linked to an oxygen atom of an -OH group anchored onto a β-D-glucose residue of cellulose and π electrons belonging to aromatic ring of pyridine/pyridinium [85] and to p-chlorophenyl moieties of CH [86]) act. Adapting this to our collagen-based systems, there are many electron-deficient hydrogens linked to the collagen structure (of peptide bonds, of -OH groups contained in hydroxyproline residues, tyrosine, serine, threonine, of unprotonated primary amine groups of lysine residues) able to interact with π electrons belonging to the aromatic rings of CH via Yoshida H-bonding, leading to increasing of the crosslinking degree of the collagen network. On the other hand, the favorable interactions between CH and TA through the same Yoshida forces, and even electrostatic ones, together with the contribution of the hydrophobic effect caused by the presence of aromatic rings, make this situation more and more complex, which is why it deserves proper attention in further investigations.

Degree of Crosslinking Estimation for Collagen-Based Matrices Revealed by Equilibrium Swelling Measurements in an Aqueous Environment
To estimate the degree of crosslinking (DC, in mol·cm −3 ) associated with the collagenbased matrices, a useful relationship developed by Flory [87] for polyelectrolyte gels at equilibrium swelling could have been applied: where V 1 is the molar volume of the solvent, ∆c = c − c i -the difference in the total concentration of ionic mobile species (in mol·cm −3 ) found inside (c) and outside (c e ) the swollen gel, ν 2eq -the volume fraction of macromolecular component in the gel at equilibrium swelling, and χ 1 -the Flory-Huggins interaction (polymer-solvent) parameter. However, Equation (1) is mainly applicable to those polyelectrolyte gels crosslinked in a bulk state prior to swelling. For a gelation process occurring in solution at a certain volume fraction of a nonionic polymer, ν 20 , Bry and Merrill [88] have derived a corresponding relationship that has then been adapted to a polyelectrolyte network at the same state of equilibrium swelling by Ofner and Bubnis [89]: Given the nature of the solvent (aqueous environment) in Equation (2), the quantity V 1 is the molar volume of water and the dimensionless quantity χ 1 -the Flory-Huggins interaction parameter of the collagen-water pair was considered to be 0.49. This value was chosen in accordance with the results published elsewhere [90] for gelatin and based on the finding that both gelatin and collagen have the same χ 1 value at high relative humidity (>95%) [91] and, most likely, even in aqueous systems.
For all the systems discussed herein, the overall difference in the mobile ion concentration inside and outside the matrix was evaluated by considering the Donnan equilibrium [92] established between the releasing environment (phosphate-buffered saline, PBS) and the polyelectrolyte network at pH 7.40 and 37 • C. Thus, according to the net electrical charge of collagen, the distribution of the mobile ions inside and outside the matrix at equilibrium has to fulfill the condition of equality between the activities of the electrolyte existing in the two zones of the swollen systems. Prior to discussing the consequences of the Donnan effect in these systems at equilibrium swelling, some prerequisites were assumed: Based on the abovementioned conditions, the ionic strength corresponding only to the concentration of 10 mM phosphate buffer in the presence of 154 mM NaCl (considered fully dissociated) was found to be 0.026 (see Supplementary Materials). Thus, the total ionic strength (I) is 0.180, the equivalent concentration of NaCl leading to this I value is 0.180 M, and, going further, half the overall concentration of the mobile ionic species (NaCl was considered fully dissociated), c e , is 0.090 M. Therefore, according to what the Donnan effect mainly means [92] (the fulfillment of the equality condition between the activity of the mobile electrolyte inside and outside the matrix-which is expressed as an equality between the products of the constituent ion activities of the mobile electrolyte-and also of the electroneutrality condition involving all ionic species into the two regions), the following equation will finally result: where c p is the concentration of net fixed electrical charge of collagen in the swollen matrix (ionic strength of 0.180, pH 7.40, and 37 • C). In deriving Equation (3), the mean ionic activity coefficient of the electrolyte (see details in Supplementary Materials) was considered to be the same both inside and outside the matrix. To estimate the c p value, it is necessary to evaluate the number of net fixed electrical charges (N elch ) onto every single collagen molecule (atelocollagen). To do so, based on the amino acid composition, the residues of aspartic acid (Asp), glutamic acid (Glu), histidine (His), arginine (Arg), and lysine (Lys) have been taken under consideration in generating the fixed charges attached to the collagen type I triple helix depending on pH, temperature, and the ionic strength of the swelling environment (see Table S2). For simplicity, the following conjugate acid/base pairs were taken into account: -COOH/-COO − pairs for the residues of Asp and Glu and -NH 3 + /-NH 2 pairs for His, Arg, and Lys residues. Their acidity exponents (pK a ) at 25 • C and with no electrolyte (Table S2) were corrected according to the influences induced by the temperature and ionic strength changes (t = 37 • C, I = 0.180) [4,[93][94][95][96][97]. The corrected pK a values (Table S2, recorded in parentheses) were obtained by applying equation (S13), where the quantity Z is −1 for the -COOH/-COO − pairs and +1 for the -NH 3 + /-NH 2 pairs as a result of a comparison to the equilibrium (S2) (in Supplementary Materials). To account for all the presumable electrical charges per one triple helix of collagen type I as atelocollagen (without terminal telopeptides cleaved enzymatically), the corresponding number of N-terminal and C-terminal amino acid residues (three glycine residues for N-terminus and three proline residues for C-terminus) were also considered (Table S2). This is in accordance with the primary structure of the three polypeptide chains belonging to the triple helical structure of the bovine collagen type I [98]. Starting from the well-known Henderson-Hasselbalch equation applied to all the conjugate acid/base pairs discussed above, the number of net electrical charges carried by a molecule of atelocollagen (N net ) at 37 • C, pH 7.40, and I 0.180 will be given as: where the first term of the right side of the equation denotes the overall number of the negative charges and the second, the total number of the positive charges. The subscripts i refer to each of the identical amino acid residue types carrying carboxylic groups in the protonated form (Asp, Glu, and Pro, respectively) and j, a similar quantity defined for the identical amino acid residues types bearing basic groups in the deprotonated form (His, Arg, Lys, and Gly, respectively). Therefore, N i and N j are the number of identical amino acid residues belonging to each of the residue kinds just mentioned (Asp-90, Glu-150, Pro-3, His-14, Arg-162, Lys-89, and Gly-3, respectively) and (pK a ) i and (pK a ) j are the corresponding acidity exponents (see Table S2). Because the isoelectric region of collagen was found to be 5.0-6.0 (determined from the maximum optical density of 0.03% collagen in 0.180 M NaCl aqueous solution at different pHs, results not shown), a value of 5.5 has been chosen for its isoelectric point (IEP), where N net becomes zero, and, as a consequence, the number of fixed negative charges should be greater than that of the fixed positive charges at pH 7.40. Instead, the calculated IEP, by counting the contributions of the overall fixed charges, was found to be higher than 8.0, a value far greater than that experimentally determined. A possibility of somehow matching the two very different IEP values is to consider a larger number of acid amino acid residues (Asp, Glu) actually carried by the collagen type I molecules used in this study. This is truly feasible if the asparagine (Asn) and glutamine (Gln) residues undergo a deamidation to a certain extent to give additional carboxylic groups. Indeed, by taking into account a degree of deamidation of 24% and keeping the same ratio of Asn/Gln (Table S2), a supplementary number of Asp (11) and Glu (18) could result, and, consequently, an IEP value of 5.5 was calculated as shown in Figure 5 (by using Equation (4)). Such an acidic IEP of collagen could be attained due to the extraction procedures of collagen partially conducted at quite basic pH when some side carboxamide groups of Asn and Gln might be easily hydrolyzed to the corresponding carboxylic groups. Once a very good accordance between the experimental and calculated values of the collagen IEP was established, the next step was to estimate the number of net negative charges carried by a molecule of atelocollagen at pH 7.40, I 0.180, and 37 • C.
As can been seen in Figure 5, this quantity may be easily obtained based on Equation (4), giving a value of 17.6 and, going further (by taking into account the expressions (S17)-(S20) in Supplementary Materials), the c p values (in mol/cm 3 ) will be given by: where Q eq v,coll is the equilibrium volume swelling ratio regarding only the collagen component (with density ρ coll , in g/cm 3 ) of a given collagen-based matrix of density ρ matrix (in g/cm 3 ) and M-the molar mass of collagen (considered 3 × 10 5 g·mol −1 ). The values of the quantities preceding c p , as well as the c p values themselves, and the ∆c values, are listed in Tables 2 and 3 for all the studied matrices. Eventually, with these intermediate steps accomplished, it was possible to calculate the crosslinking degrees of collagen-based matrices (by Equation (2)) supposing that ν 20 is on average 0.20. This value was chosen in accordance with the percentage of moisture content found in the fresh hide/skin of mammals (ca. 70%) and the percentage content of collagen in dry skin (ca. 75%) [98][99][100][101][102][103].
7.40. Instead, the calculated IEP, by counting the contributions of the overall fixed charges, was found to be higher than 8.0, a value far greater than that experimentally determined. A possibility of somehow matching the two very different IEP values is to consider a larger number of acid amino acid residues (Asp, Glu) actually carried by the collagen type I molecules used in this study. This is truly feasible if the asparagine (Asn) and glutamine (Gln) residues undergo a deamidation to a certain extent to give additional carboxylic groups. Indeed, by taking into account a degree of deamidation of 24% and keeping the same ratio of Asn/Gln (Table S2), a supplementary number of Asp (11) and Glu (18) could result, and, consequently, an IEP value of 5.5 was calculated as shown in Figure 5 (by using Equation (4)). Such an acidic IEP of collagen could be attained due to the extraction procedures of collagen partially conducted at quite basic pH when some side carboxamide groups of Asn and Gln might be easily hydrolyzed to the corresponding carboxylic groups. Once a very good accordance between the experimental and calculated values of the collagen IEP was established, the next step was to estimate the number of net negative charges carried by a molecule of atelocollagen at pH 7.40, I 0.180, and 37 °C. As can been seen in Figure 5, this quantity may be easily obtained based on equation (4), giving a value of 17.6 and, going further (by taking into account the expressions (S17)-(S20) in Supplementary Materials), the cp values (in mol/cm 3 ) will be given by:  Tables 2 and 3 for all the studied matrices. Eventually, with these intermediate steps accomplished, it was possible to calculate the crosslinking degrees of collagen-based matrices (by Equation (2)) supposing that ν20 is on average 0.20. This value was chosen in accordance with the percentage of moisture content found in the fresh hide/skin of mammals (ca. 70%) and the percentage content of collagen in dry skin (ca. 75%) [98][99][100][101][102][103].    Due to the fact that the real ν 20 values (the collagen volume fraction in a certain swollen collagen-based matrix at equilibrium) are difficult to find, there are two limiting cases where this problem can be readily solved (by keeping in mind the same matrices densities that were picnometrically determined): the first of them, when neither TA nor CHDG were released at equilibrium swelling (Table 2), and the second, when practically both TA and CHDG were considered almost majority released at equilibrium (Table 3). Accordingly, the degrees of crosslinking (DCs) corresponding to the two limiting cases grew higher from 58 × 10 −5 mol·cm −3 for collagen matrices to (86-103) × 10 −5 mol·cm −3 for collagen-TA (5%) sponges, maintaining this tendency as TA content increased to 15% ((191-306) × 10 −5 mol·cm −3 ). On the other hand, DCs varied from (208-321) × 10 −5 mol·cm −3 to (184-336) × 10 −5 mol·cm −3 and further to (144-297) × 10 −5 mol·cm −3 for the threecomponent porous matrices with the same content of CHDG (9.09%) and TA percentage rising from 5 to 15%, respectively (Tables 2 and 3). The DC values are, most likely, closer to those specific to the swollen collagen-based sponges with the two low molecular weight components (TA and CHDG) almost entirely released. This is because the fractional release after 660 min (denoted as c 11 in Supplementary Materials) ranges from 0.7 to 0.9 for TA and ca. 0.4-0.6 for CHDG depending on the matrix composition, as can be seen later in the Section 3.6 (see also Figure S6). Anyway, irrespective of the limiting cases mentioned above, the ascendant tendency of DC values follows exactly the decrease of Q eq v,coll values for all the matrices investigated (Tables 2 and 3, the 4th and 8th columns). Unlike the sponges made of collagen solely, which are practically non-crosslinked (DC value of 58 × 10 −5 mol·cm −3 corresponding to a swelling time of 100 min is purely informative because these kinds of matrices began to disintegrate into smaller and smaller pieces as time passed, see Figure 5), the collagen-TA matrices behaved completely differently, showing the key role played by TA in making physical crosslinkages between the collagen molecules. Thus, a gradually increasing percentage of TA led to enhancing in DCs of lyophilized collagen-TA sponges, as arisen from the decreasing values of equilibrium of the weight/volume swelling ratio for the same bicomponent systems ( Figure 5 and Tables 2 and 3). On the other hand, the presence of CHDG at its maximum percentage (9.09%) somehow impairs the physical crosslinking of collagen in matrices with TA and CHDG when TA content rises (see the values of Q eq v,coll and DC in Tables 2 and 3). Taking into account the more realistic case of about total release of TA and CHDG from the three-component systems (containing the same proportion of CHDG of 9.09%) at equilibrium swelling (Table 3), both Q eq v,coll and DC values are very close to those associated with TA (15%)-collagen matrices. These data support the idea of reaching the maximum value of the DC of collagen network for all these systems (by the crosslinking contribution brought by TA and CHDG taken together or separately), and, in addition, are consistent with the experimental findings discussed earlier concerning the similarities noticed in swelling behavior. Other specific details on building up the physical crosslinkages in the collagen-based matrices studied are discussed later in the section regarding TA/CHDG releasing.

Collagenase Digestion
The collagen type I (which is the main component in all the samples studied) in its native state resists the attack of most proteolytic enzymes, except for the specific collagenases. These so-called matrix metalloproteinases are able to cleave through each individual triple helix at a point approximately three-quarters of the distance from the N-terminal end along the collagen molecule (usually at a peptide bond between Gly and Leu or Gly and Ile amino acid residues). This pathway of enzymatic digestion was also described for the collagen type II and III [4,[104][105][106][107]. Degradation of collagen fibrils starts from the outside: the enzyme fastens tightly onto molecules from the surface, and the inside ones are accessible by progressive degradation [4,107]. Once the triple helix cleaved, the fragments remain attached to the fibril (in the case of collagen fibrils) by crosslinks occurring in mature collagen (fibrils), and, at body temperature (ca. 37 • C), they start to denature [104]. At the same time, gelatinases or other nonspecific proteinases continue the degradation, digesting the primary fragments into low molecular weight peptides and eventually amino acids [4]. Due to the specific enzymatic action only onto triple helices taken individually, collagenases are frequently used for in vitro collagen digestion.
The collagen-based matrices studied in this paper were subjected to digestion by using collagenase type I (from Clostridium histolyticum) in a PBS-HTAB environment at pH 7.4 and 37 • C.
The matrices consisting of fibrillar collagen solely disappeared within ca. 30 min and the digestion aqueous medium became clear. The integrity and overall shape of the matrices with 5% TA were kept for about 6 h, and then disintegrated into highly swollen fragments, the final aqueous system becoming a little turbid and pale green-yellowish in color.
By doubling the amount of TA, the collagen-TA matrices preserved their own integrity for 24 h, the fragments formed during the subsequent disintegration were more or less swollen, while the turbidity and color of the digestion medium intensified. By increasing the TA amount to 15%, the matrices resisted for 9 days (the maximum period of observation), preserving their integrity and shape, with the same appearance of the digesting medium as that of the previous samples.
The presence of both TA and CHDG into the collagen matrices increased spectacularly their resistance to collagenase: all resisted for 9 days, except those containing 1.82% CHDG and 5% TA, which, on the 9th day, showed a fragmentation into a few relatively bulky fragments. These findings indicate most likely that TA and CHDG, jointly or individually, generate physical crosslinks between collagen molecules throughout the collagen network of the studied matrices limiting the effectiveness of the collagenolytic activity of collagenase, as reported elsewhere [27]. However, beyond these noticed aspects, supplementary investigation should be taken into account to prove whether the presence of TA and/or CHDG inhibits to a greater or lesser extent the collagenase activity, as suggested Velmuragan et al. in an interesting work on TA-collagen type I interaction [18].

TA and CHDG Release in a PBS-HTAB Medium
Generally speaking, drug release from a polymeric matrix, porous or nonporous, involves a number of important aspects that may have a greater or lesser influence [108]: matrix swelling and erosion, drug dissolution, drug-matrix structure interactions, drug distribution and concentration within matrix, matrix shape (cylindrical, spherical, etc.), and matrix polydispersion. Adapting them in terms of the simultaneous release of TA/CHDG from a collagen matrix into a release environment, we took into account four main factors that could affect the releasing process, considering the same matrix shape and operating temperature: (a) TA/CHDG load (concentration) inside the matrix; (b) TA/CHDG-matrix interaction; (c) mutual TA-CHDG interactions that may yield complexes more or less soluble into the release medium; (d) TA/CHDG-release medium interactions. Based on all these features, it becomes clear that TA/CHDG in a dry collagen matrix cannot be released as solid form (micro crystals or other solid form). However, during immersion into the release fluid, the dry matrix may experience a swelling process (accompanied by a relaxation phenomenon of macromolecular network [109][110][111][112][113], followed by a corresponding TA/CHDG dissolution. Anyway, as the swelling process evolves, three moving boundaries can be defined irrespective of the swellable matrix: erosion, diffusion, and swelling front, respectively ( Figure 6)) [108,113].   Figure S7).
At this point, prior to discussing, in detail, TA/CHDG release, it is important to make some considerations on a few structural features of chlorhexidine. First of all, as a biguanide-based compound (i.e., bisbiguanide derivative), chlorhexidine is frequently represented by a misleading structure both in studies of medicinal chemistry and generally in Figure 6. Overall scheme depicting releasing process of TA/CHDG from a collagen-based matrix prepared from an initial hydrogel (pH 4) and a subsequent step of freeze-drying.
However, from a practical standpoint, using PBS (pH 7.4) at 37 • C as the release medium was a difficult task because CHDG could not be dissolved in that medium at all (most probably due to the chlorhexidine phosphate formation involving chlorhexidine dication, CH ++ , and hydrogen phosphate dianion, HPO 4 2− ). To overcome such a drawback, we proposed a phosphate buffer with a carrier able to solubilize both CHDG and TA. Thus, we have chosen a cationic surfactant (to avoid precipitation), hexadecyltrimethylammonium bromide (HTAB), with low critical micelle concentration (cmc) [58,[114][115][116][117][118][119][120][121]. This overall picture seems to be supported by the experimental findings according to which CHDG is soluble in pure water, NaCl solution, HTAB solution, or a mixed PBS-HTAB solution, but very insoluble in PBS solution. Moreover, the presupposition of insoluble CH ++ -HPO 4 2− ionic pairs formation is based on the preponderant proportion of HPO 4  . Accordingly, the release solution consisting of 0.3 mM HTAB in PBS (10 mM sodium phosphate buffer, NaCl-0.154 M, pH-7.4, which imitates human extracellular fluid from the point of view of ionic strength and pH) was employed. Such a medium was in fact a micellar solution of HTAB, with cationic micelles as carriers for TA/CHDG (mimicking somehow the function of the carrier proteins). By using pyrene as the fluorescence probe, the cmc of HTAB was found to be 7.3 × 10 −5 M in PBS (pH 7.4) at 37 • C ( Figure S7).
At this point, prior to discussing, in detail, TA/CHDG release, it is important to make some considerations on a few structural features of chlorhexidine. First of all, as a biguanidebased compound (i.e., bisbiguanide derivative), chlorhexidine is frequently represented by a misleading structure both in studies of medicinal chemistry and generally in chemistry literature. Thus, the "incorrect" structure of biguanide moiety (Table S3, 1st row) proved to be less stable than the "right" tautomer (Table S3, 2nd row) by ca. 10 kcal·mol −1 , which resulted from ab initio calculations [122,123]. This stable structure of biguanide (asymmetric in shape and with intramolecular conjugation between C2-N3 and C4-N5 double bonds, see Table S3, 2nd row) was clearly confirmed more than 40 years ago via single crystal X-ray diffraction [124]. At the same time, the biguanide moiety is protonated in CHDG and, based on the calculated partial atomic charges (from natural population analysis) [123] and the absolute proton affinities [125], the most favored protonation atom to result in the biguanidinium cation is N5 (Table S3, 2nd and 3rd rows). These theoretical calculations performed on biguanide to give the structure after its protonation were also strongly supported by the crystal structure of biguanide hydrochloride revealed through X-ray diffraction [126]. Secondly, in accordance with the structure of the biguanidinium cation postulated both theoretically and experimentally, there are a number of conformers/tautomers of the chlorhexidine dication (CH ++ ) that can adopt more or less extended conformations, as shown in Table S3 (Supplementary Materials). On the other hand, in a PBS environment (pH 7.4), CHDG ought to be most likely in double protonated form because of the pK a value evaluated to be of 9.55 for the chlorhexidine dication [86]. As already mentioned, the poorly soluble (actually, quite insoluble) complexes of CHDG in the releasing solution (complexes with strong ion pairing between CH ++ and HPO 4 2− , similar to those established between guanidinium cations and sulfate dianions [127]) will be embedded preferentially into the HTAB micelles. The way the chlorhexidine is accommodated into the HTAB micelles depends to the utmost extent on how well the chlorhexidine units and surfactant micelles fit together, considering their geometrical size and how favorable the mutual interactions between the components of the mixed micelles can be. Taking into account the first aspect, the length of the extended CH ++ conformation (considering the shape of the linear 1,6-hexamethylene linker), expressed as the Cl-Cl distance (ca. 2.4-2.9 nm, see Table S3, structures (1)-(3)), and the overall diameter of spherical HTAB micelles [128][129][130] (from ca. 5 to 6.6 nm within a temperature range of 25-40 • C) do not match at all. However, from this structural standpoint, CH ++ may adopt an overall V-shaped conformation (folded state, see Table S3, conformations (5)-(10)) able to be inserted into the HTAB micelles corona, as schemed simplistically in Figure 2. This possible mechanism was inspired by the interesting data published elsewhere [131,132]. Thus, these studies, based on neutron scattering measurements and molecular dynamics (carried out on the mixed systems of CHDG-phosphatidyl choline (PC) bilayers, with CHDG deuterium-labelled or unlabeled at the hexamethylene moiety), evinced convincingly both the place where CHDG entities are favorably located (near the headgroups region of the lipid bilayers) and the shape of CHDG (bent in half in a wedge-like fashion; Table S3, folded conformations).
Moreover, the molecular dynamics simulations performed on CH ++ in water led to a broad distribution of Cl-Cl (belonging to the two chlorophenyl moieties of CH ++ ) distance, while the same procedure performed in the PC bilayer exhibited a sharp distribution, with the most likely conformation having a Cl-Cl distance of ca. 1.5 nm [131]. These results support our assumption regarding the V-shaped conformations of CH ++ inserted into HTAB micelles. In this respect, the folded conformations (5)-(8) (Table S3) seem to best match the requirement of 1.5 nm for the Cl-Cl distance.
The next step was to properly adjust a UV-VIS method for quantifying TA/CHDG released at a desired time with a satisfactory accuracy. In this way, analyzing UV-VIS absorption spectra of numerous TA/CHDG mixtures (with known compositions, see some in Figure S8), and based on the individual UV-VIS spectra of the releasing components in HTAB-sodium phosphate buffer (Figure 7a), it was found that the most reliable approach to estimate TA/CHDG amount released into the release environment is monitoring the values of absorbances at ca. 325 nm for TA and 260 nm for CHDG, respectively. Then, a subsequent experimental algorithm was obeyed to obtain TA/HTAB concentrations in release medium with good enough reliability and accuracy (for more details, see the text on page 13 with Figure S9 (Table S2). Every single site of the bidimensional matrix represents a specific, discrete composition expressed as the concentration ratio of TA/CHDG in the releasing solution. Any composition obtained so, taken individually, should follow the Beer-Lambert law by diluting it correspondingly using the same releasing solvent ( Figure S9); -From the UV-VIS spectrum of a certain composition (extracted from the release medium at a desired time), the ratio of absorbances belonging to the two components (A 325 nm /A 260 nm ), corrected by the corresponding extinction coefficients (found to be constant for a broad variety of TA-CHDG compositions), was compared to the values from the data matrix (Table S4). The closest value found in the table defines the relative composition most likely to be in the real system (extracted from the release medium); -A series of 5-6 solutions of known concentrations (keeping the same ratio of TA/CHDG) were used to generate a calibration curve (a straight line, according to the Beer-Lambert law) on which the specific values of A 325nm and A 260nm allowed for finding the TA and CHDG concentrations in the real sample. With the aid of this set of experimental rules, the time dependences of simultaneous TA and CHDG release have been obtained.
At the same time, the analysis of the overall kinetics of TA/CHDG release was based on some preliminary considerations allowing for the two particular cases of diffusional transport involved in the course of the releasing process. Therefore, the Fick's second law of three-dimensional diffusion considering a constant diffusion coefficient (D) can be written as [133]: where c is the concentration of a diffusing substance. In cylindrical coordinates, Equation (6) can be rewritten in accordance with the corresponding Laplacian operator form: where, generally, r∈[0,∞), θ∈[0,2π), and z∈(-∞,∞). Supposing a diffusion phenomenon occurring within a cylindrical volume (through the lateral area of the cylinder), it is useful to take into account the simplest case of radial diffusion within a plane located at a constant z coordinate and irrespective of the angular coordinate (θ). At given constant values for the quantities z and θ, Equation (7) defines so-called one-dimensional radial diffusion into a cylinder [133,134]: For this kind of penetrant diffusion in a cylinder of radius a, the initial and boundary conditions, respectively, will be: where c1 is the constant diffusant concentration initially existing within the cylindrical volume (generally c1 = 0) and c0 is a constant concentration of diffusant located outside the cylinder. Allowing for water (or solvent) uptake into a cylindrical sample, the concentrations defined above are referred to as the water/solvent (diffusant) concentration. A useful solution of Equation (8) under conditions (9) and (10) at "small times" has been demonstrated to be [134,135]: With the aid of this set of experimental rules, the time dependences of simultaneous TA and CHDG release have been obtained. At the same time, the analysis of the overall kinetics of TA/CHDG release was based on some preliminary considerations allowing for the two particular cases of diffusional transport involved in the course of the releasing process. Therefore, the Fick's second law of three-dimensional diffusion considering a constant diffusion coefficient (D) can be written as [133]: where c is the concentration of a diffusing substance. In cylindrical coordinates, Equation (6) can be rewritten in accordance with the corresponding Laplacian operator form: where, generally, r∈[0,∞), θ∈[0,2π), and z∈(−∞,∞). Supposing a diffusion phenomenon occurring within a cylindrical volume (through the lateral area of the cylinder), it is useful to take into account the simplest case of radial diffusion within a plane located at a constant z coordinate and irrespective of the angular coordinate (θ). At given constant values for the quantities z and θ, Equation (7) defines so-called one-dimensional radial diffusion into a cylinder [133,134]: For this kind of penetrant diffusion in a cylinder of radius a, the initial and boundary conditions, respectively, will be: where c 1 is the constant diffusant concentration initially existing within the cylindrical volume (generally c 1 = 0) and c 0 is a constant concentration of diffusant located outside the cylinder. Allowing for water (or solvent) uptake into a cylindrical sample, the concentrations defined above are referred to as the water/solvent (diffusant) concentration. A useful solution of Equation (8) under conditions (9) and (10) at "small times" has been demonstrated to be [134,135]: where M t denotes the amount of water/solvent that has been retained by the cylinder in time, t, and M ∞ is the corresponding quantity at equilibrium (after infinite time). The term "small times" defines an approximation that is valid for the first 60% of water/solvent uptake (M t /M ∞ ≤ 0.6). Apart from the Fickian diffusion (Case I) described by Equation (11) for "small times", Enscore et al. proposed a generalized model of the Case II transport in a polymeric sample based on its shape [136]: where k 0 is the Case II relaxation constant of polymer. The quantity b is the half-thickness for planar samples or the radius of the cylinder/sphere (a) for cylindrical/spherical samples, respectively. Keeping the same order of sample geometry, the numerical values of N are 1 (for plane), 2 (for cylinder), or 3 (for sphere). Thus, the Case II water/solvent absorption by a cylindrical sample will be given by adapting Equation (12) accordingly:  (14) where D ap is an apparent constant diffusion coefficient for the non-Fickian diffusion (Case II). The reverse phenomenon of the release of a solute contained into a cylindrical sample may be treated analogously, in accordance with Equations (11) and (14) by suitably redefining the significance of quantities M t and M ∞ : M t -the amount of solute released from the sample up to time, t, and M ∞ -the amount of solute desorbed into the releasing environment after infinite time. Depending on the type of solute loaded into the cylindrical sample (in our case, TA and CHDG, respectively), the times at which M t reaches its equilibrium values (when fractional release is considered to be practically equal to unity) differ significantly. Thus, a time period of 24 h (1440 min) to attain the equilibrium release for TA was experimentally observed, and 72 h (4320 min) for CHDG. As a result, the two types of solute release described by Equations (11) and (14) require two distinct values for the quantity D ap /a 2 : 1/1440 min −1 for TA and 1/4320 min −1 for CHDG release, respectively ( Figure 8). The real release behavior of a solute contained by a real polymer-based matrix does strictly obey neither Fickian nor non-Fickian diffusion (Case II) but follows an intermediate mechanism. Thus, Peppas et al. [111,134,137,138] proposed an empirical equation, socalled the power law model, as under: where k is a kind of kinetic constant related to both the structural and geometrical features of the matrix and n-diffusional exponent that is an indicator of the transport/diffusion mechanism of the solute/drug through the swollen matrix. When n = 0.5, the release of solute is considered Fickian, as demonstrated for diffusion from a plane matrix, based on Fick's laws [111,134]. Moreover, by accounting the same system of drug/solute release, the Case II transport may be carried out when n = 1. For the real matrices, the release of solute/drug is a result of the coupling of diffusion and relaxation phenomena, both of them leading to a non-Fickian (anomalous) intermediate behavior. While the specific values of n for a plane, very thin matrix are 0.5 and 1.0 for Fickian and non-Fickian (Case II) diffusion behavior, respectively, by changing the shape of the delivery system into a cylindrical one, these values become 0.45 and 0.89, respectively [134,139]. Any other values of n within the 0.45-0.89 range define the so-called anomalous diffusion. In our specific case, we performed the release experiments upon a cylinder-shaped matrix, with the upper and The real release behavior of a solute contained by a real polymer-based matrix does strictly obey neither Fickian nor non-Fickian diffusion (Case II) but follows an intermediate mechanism. Thus, Peppas et al. [111,134,137,138] proposed an empirical equation, so-called the power law model, as under: where k is a kind of kinetic constant related to both the structural and geometrical features of the matrix and n-diffusional exponent that is an indicator of the transport/diffusion mechanism of the solute/drug through the swollen matrix. When n = 0.5, the release of solute is considered Fickian, as demonstrated for diffusion from a plane matrix, based on Fick's laws [111,134]. Moreover, by accounting the same system of drug/solute release, the Case II transport may be carried out when n = 1. For the real matrices, the release of solute/drug is a result of the coupling of diffusion and relaxation phenomena, both of them leading to a non-Fickian (anomalous) intermediate behavior. While the specific values of n for a plane, very thin matrix are 0.5 and 1.0 for Fickian and non-Fickian (Case II) diffusion behavior, respectively, by changing the shape of the delivery system into a cylindrical one, these values become 0.45 and 0.89, respectively [134,139]. Any other values of n within the 0.45-0.89 range define the so-called anomalous diffusion. In our specific case, we performed the release experiments upon a cylinder-shaped matrix, with the upper and bottom face of the cylinder covered by a thin wax layer as a water-repellent barrier. Indeed, by fitting the first 60% of solute release shown in Figure 8 (curves 1 and 2) with Equation (15) (red curves 1 and 2 ), the abovementioned values of n were fully confirmed (both n and k values are collected in Table 4). The release data for all the collagen-based matrices prepared and entirely fitted with only the Peppas-type Equation (15) are summarized in Tables S5-S7. However, for a preliminary assessment of TA/CHDG release, it was useful to establish the contribution of each type of diffusion mechanism to the overall solute release. Accordingly, any fractional release-time dependence (up to 0.6 fractional release) may be written as a weighted sum of the two types of diffusion mechanisms as follows: for TA release and M t M ∞ = k·t n = a·0.04009·t 0.4497 + b·0.00088·t 0.88703 (17) for CHDG release. Before discussing the release behavior from the collagen-based matrices, it is worth noting for the purpose of the present work that all the values of diffusional exponents fall within 0.45-0.89, which demonstrates an anomalous transport mechanism for both TA and CHDG during the releasing process.
As for TA release from the TA-collagen matrices, all four (a)-(d) factors abovementioned seem to have influenced the process. Referring back to the preparation of porous TA-collagen matrices from corresponding acidic hydrogels presented elsewhere [140] (see Figure 6), it became clear enough that the crosslinker role of TA was achieved mainly by hydrogen bonding between the polyphenolic -OH groups of TA and -COOH, -OH, and peptide groups belonging to collagen, on one hand, and by Yoshida-type H-bonding previously discussed. Consequently, at a low TA concentration (5%), the majority of TA may be bound onto collagen fibrils/molecules. As a result, a low release rate may occur (Figure 7b). Increasing the TA concentration to 10%, the amount of "nonbound" TA increased, becoming more available to be solubilized by the HTAB micelles of the release solution. That is why the relative amount of TA released also increased and the time dependence release for 10% TA is placed above that for 5% TA. The amount of TA released per unit area per unit time continued to intensify until reaching its maximum value imposed by the TA solubilization rate into HTAB micelles. Beyond this point, as in the case of the collagen matrix with 15% TA, the process practically became rate-controlled. Thus, the relative amount of TA released per unit area per unit time (relative flux) diminished, even though the release flux (not relative) remained constant. It is also worth mentioning that the non-Fickian mechanism of TA release is by far prevalent, with relative weights (based on the value of dimensionless coefficients a and b in Equation (16)) easily exceeding 90%, as exhibited in Table S8 (Supplementary Materials). This behavior is consistent with an anomalous Fickian transport much closer to the Case II mechanism as mirrored by the values of general diffusion exponent (n) ranging between 0.792 and 0.846 (Table S8). Instead, by fitting the entire experimental dataset of swelling (not only those corresponding to the first 60% of TA release as above) with the Peppas-type Equation (15), the values of diffusional exponent descended significantly (0.686-0.735, Table S5), which might suggest a lower contribution of relaxation phenomena of the collagen-based network to the releasing process.
The presence of CHDG drastically changed TA release from the TA-CHDG-collagen matrices. Thus, for the matrices with the minimum CHDG content (1.82%, see Figure S10a), the most significant factor seems to be (b) CHDG interaction with the collagenous substrate under release conditions. Indeed, collagen molecules of the swollen matrix (pH 7.4) became negatively charged. This is because the extraction technology of collagen extraction performed by our supplier involved an intermediate stage of sodium hydroxide treatment. As a result, some asparagine and glutamine residues of collagen might have undergone deamidations, leading to the corresponding aspartate and glutamate residues. These newlyformed amino acid residues could have been responsible for lowering the isoelectric point of collagen into an acidic domain (pH 4.6-6.0), as reported elsewhere [57]. Thus, electrostatically induced interactions between chlorhexidine dication and negatively charged collagen (at pH 7.4) significantly altered the collagenous substrate with respect to TA. The direct consequence was a low TA binding (and low degree of TA binding) onto collagen. In this way, based on the experimental data ( Figure S10a), TA release did not depend to a large extent on its loading concentration inside the matrix, but mainly on the solubilization rate in HTAB micelles. Such a rather rate-controlled process is equivalent to an almost constant release flux of TA (irrespective of its concentration) and decrease in its relative release as TA content ascends from 5 to 15%. This behavior is also mirrored by the majority proportion of Case II diffusion (with a percentage of non-Fickian Case II behavior between ca. 63 and 76% as shown in Table S9). However, a relative decreasing contribution of the collagen network relaxation to the transport mechanism is sustained by the diffusion exponent values (0.595, 0.628, 0.673) closer to 0.45 than to 0.89 when it comes to a fractional TA release up to 60% (Table S7). This course is supported by the values of n (0.562, 0.628, 0.630) being a little bit smaller when 100% of releasing data were considered.
Similarly, the crosslinking effect induced by CH onto the collagen network as a result of attractive electrostatic interactions CH ++ -negatively charged collagen led to the same relatively low degree of TA binding on collagen for TA-CHDG(4.55%)-collagen matrices, as just described for TA-CHDG(1.82%)-collagen sponges. At the same time, the rate-controlled process of TA release imposed by the same rate of TA solubilization by HTAB micelles gave a roughly similar descendant order of time dependent TA releases when TA load increased from 5 to 15% ( Figure S10b). The kinetics data obeyed the tendency noticed in the case of previous TA-CHDG-collagen sponges loaded with 1.82% TA, which practically reveals the prevailing proportion of the non-Fickian Case II diffusion (Table S9) corresponding to a general anomalous transport mechanism (Table S6) for TA release in the aqueous release medium.
The possibility of mixed HTAB-based micelles to enhance continuously their ability of TA solubilization with increasing CH content, coupled with the formation of a TA-CH complex, could be a realistic one when inspecting TA release from porous TA-CHDG (9.09%)-collagen matrices. Actually, the structural evolution of the mixed TA-CH-HTAB micelles should be strongly related to the amount of CH embedded, which, in turn, may generate thermodynamic instability of micellar assemblage due to the repulsive interactions between the cationic heads of the surfactant and CH ++ dications. Accordingly, to improve thermodynamic stability, such micellar structures require insertion of additional TA entities as relatively non-charged or partially charged with negative charges spacers (see the scheme of process in Figure 2). Indeed, a general value of protonated constant (acidity exponent, pK a ) estimated by Ghigo et al. [141] for TA is of 7.5, which is equivalent to say that about half the TA species are statistically in protonated form (non-charged) and the rest of them are in the deprotonated state (negatively charged), favoring mixed micelles formation. At the same time, the similarity in ionic size of Br − (0.196 nm in radius) and hexadecyltrimethylammonium cation (HTA + , considered to possess a radius close to that of tetramethylammonium cation, ca. 0.2 nm) [142], as parts of every single HTAB species, partially lead to development of undissociated inner sphere ion pairs HTA + /Br − in an aqueous solution according to the Collins' theory of matching water affinity with respect to a given cation-anion pair [143,144]. Such a behavior is partly supported by a value of 0.42 conductometrically determined in pure water for the degree of counterions (bromides) binding in HTAB micelles [145]. This phenomenon determines not only an increase in micelle size, but also an enhanced capacity of TA solubilization by the newly mixed micelles when CH content becomes greater.
Secondly and simultaneously, a possible nonstoichiometric TA-CHDG complex formation (due to reversible electrostatic interactions between CH ++ and negatively charged TA at pH 7.4, Yoshida H-bonding, and π-π stacking between TA and CHDG) might have played a significant role in altering TA release from three-component matrices with a maximum content of CHDG (9.09%): where n and m are unknown compositional indices (probably m ≥ n). Thus, we hypothesized that at a large CHDG content, a certain amount of TA could have been strongly "immobilized" in the complex TA-CHDG formed. Typically, at low level of TA, most of the "immobilized" TA was more difficult to solubilize in the mixed HTAB-based micelles, which could explain the lowest fractional release ( Figure S10c) over the entire time period of data acquisition (660 min). On the other hand, the increasing capacity of the mixed HTAB micelles (able to incorporate much more CH) could have led to an enhanced solubilization of TA when its percentage was 10 or 15% in a way somehow proportional to the available (less "immobilized") TA content. Such a behavior could explain the high similarity of TA release traces when its concentration increased and also the greater fractional release when compared to the system TA(5%)-CHDG(9.09%)-collagen. From a kinetic point of view, the prevailing Case II diffusion (involving the relaxation contribution of the collagen network) over the Fickian transport mechanism is in line with the similar peculiarity noticed in the other cases of TA release from TA-CHDG-collagen matrices considered up to 60% of fractional release (Table S9). This aspect is asserted by the corresponding anomalous mechanism of TA diffusion outside the matrix into the releasing environment, when all the dataset of TA release fitted with the Peppas-like Equation (15) led to a diffusional index of about 0.6 ( Table S6).
The most important features concerning CHDG release from TA/CHDG-collagen matrices are reflected in (a) maintaining the release dependences very close to each other (practically superimposed) at a certain CHDG concentration and different TA contents (see Figure S10a-c) and (b) a well-defined order of CHDG release dependences from the matrices where TA concentration was fixed and CHDG amount was changed. The first aspect confirms the predominance of chlorhexidine dication-collagen interaction over the TA-collagen one under releasing conditions (pH 7.4, 37 • C). Therefore, the rate of relative CHDG release is mainly determined by the strength of CHDG binding (electrostatic in nature) onto the collagenous swollen matrix. On the other hand, the solubilization capacity of the mixed micelles seems not to influence CHDG release. This is supported by the decreasing tendency in CHDG release when the CHDG concentration varies from 1.82% CHDG to 9.09% at the same TA content (Figure S6a-c). The behavior experimentally observed is indicative of the existence of a rate-controlled transport process, irrespective of CHDG content. Consequently, the absolute release flux of CHDG into the release medium at a certain moment (taken the same for all the systems) is practically the same for all TA-CHDG-collagen matrices, which means that the relative amount of CHDG released at that time decreases as CHDG concentration rises. In all these cases of chlorhexidine release (simultaneously with tannic acid), the corresponding three-component collagen-based matrices have shown an important contribution derived from the relaxation of the collagen network during the anomalous diffusion, as can be seen in Table S10 (contribution of the Case II diffusion was found to be of ca. 64-78%, with a value of diffusional index of about 0.6 by considering the first 60% of CHDG release) and Table S7 (where the general quantity n, by taking into account the entire CHDG releasing process, has a similar value of ca. 0.6).

Conclusions
The in vitro simultaneous release of TA and CHDG from three-component TA-CHDGatelocollagen type I porous matrices into a solution of PBS containing cationic micellar carriers of HTAB (pH 7.4, 37 • C): -Was monitored spectrophotometrically (UV-VIS) according to an original algorithm designed and successfully applied; -Was kinetically studied and, to the best of our knowledge, an adapted kinetic approach developed in this study led, for the first time, to general results of anomalous diffusion able to be rationally split into the two contribution parts related to the pure Fickian and non-Fickian Case II regime of diffusion/release, respectively; -Was strongly influenced, and eventually able to be controlled, by both the density of non-covalent crosslinkages of the collagen network produced by TA and CHDG and the dynamic capacity of HTAB micelles contained in the release environment to solubilize TA and CHDG.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pharmaceutics15010076/s1. 10 tables ((1) Table S1: Characteristic infrared bands for some collagen-based matrices; (2) Table S2: The amino acid residues and their features mainly involved in generating the fixed electrical charges borne by a single triple helix of collagen type I extracted from calf skin; (3) Table S3: Planar and spatial structures of some extended and folded conformers/tautomers of chlorhexidine dication, highlighting the distances between chlorine atoms; (4) Table S4: The values matrix for the TA/CHDG compositions used to guide the experimental measurements. The values inside the nonbolded cells are the ratios of TA to CHDG concentrations; (5) Table S5: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and fractional releases after 11 h (c 11 ) for the entirely (up to 100%) determined TA release from the TA-collagen matrices; (6) Table S6: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and fractional releases after 11 h (c 11 ) for the entirely (up to 100%) determined TA release from the TA-CHDG-collagen matrices; (7) Table S7: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and fractional releases after 11 h (c 11 ) for the entirely (up to 100%) determined CHDG release from the TA-CHDG-collagen matrices; (8) Table S8: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and weights of (a) Fickian and (b) non-Fickian-Case II diffusion for the TA release (up to 0.6) from the TA-collagen matrices; (9) Table S9: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and weights of (a) Fickian and (b) non-Fickian-Case II diffusion for the TA release (up to 0.6) from the TA-CHDG-collagen matrices; (10) Table S10: The values of the kinetic constants (k), diffusional exponents (n), coefficients of determination (R 2 ) and weights of (a) Fickian and (b) non-Fickian-Case II diffusion for the CHDG release (up to 0.6) from the TA-CHDG-collagen matrices), 10 figures ((11) Figure S1: FT-IR spectra (extended and fingerprint region) for the collagen-based matrices with the specified compositions (CHDG-free and 9.09% CHDG); (12) Figure S2: FT-IR spectra (extended and fingerprint region) for the collagen-based matrices with the specified compositions (4.55% CHDG); (13) Figure S3: FT-IR spectra (extended and fingerprint region) for the collagen-based matrices with the specified compositions (1.82% CHDG); (14) Figure S4: SEM micrographs (taken from sections parallel to air-sided face) for (a) collagen matrix and collagen-based matrices with (b) TA(10%) and with (c) TA(10%)-CHDG(9.09%); (15) Figure S5: Estimated temperature evolution of buffer density (upper graph) obeying the same tendency as for water density (bottom plot) (buffer density of 1.098 g/cm 3 at 20 • C was pycnometrically determined) (see www.vip-ltd.co.uk for water density values; accessed on November 19, 2022); (16) Figure S6: CHDG release from the collagen-based matrices with (a) 5% TA, (b) 10% TA and (c) 15% TA into PBS-HTAB solution (pH 7.4) at 37 • C; (17) Figure S7: Concentration dependence of I 1 /I 3 of Py in HTAB solution at 37 • C (5 × 10 −7 M Py in 10 mM buffer phosphate); (18) Figure S8: Changes in UV-VIS spectra of TA in different aqueous environments (a-d), the corresponding spectra of TA and CHDG (taken separately) in PBS and HTAB (e) and the UV-VIS spectra for three TA-CHDG mixtures in the same release solution (PBS+HTAB) at the specified compositions (f); (19) Figure S9: