A natural deep eutectic solvent (NADES) as potential excipient in collagen-based products

Article history: Received 17 December 2019 Received in revised form 26 March 2020 Accepted 5 April 2020 Available online 11 April 2020 Natural deep eutectic solvents (NADES) have previously shown antibacterial properties alone or in combination with photosensitizers and light. In this study, we investigated the behavior of the structural protein collagen in a NADES solution. A combination of collagen and NADES adds the unique wound healing properties of collagen to the potential antibacterial effect of the NADES. The behavior of collagen in a NADES composed of citric acid and xylitol and aqueous dilutions thereofwas assessed by spectroscopic, calorimetric and viscositymethods. Collagen exhibited variable unfolding properties dependent on the type of material (teloor atelocollagen) and degree of aqueous dilution of the NADES. The results indicated that both collagen types were susceptible to unfolding in undiluted NADES. Collagen dissolved in highly diluted NADES showed similar results to collagen dissolved in acetic acid (i.e., NADES network possiblymaintained). Based on the ability to dissolve collagenwhilemaintaining its structural properties, NADES is regarded as a potential excipient in collagen-based products. This is the first study describing the solubility and structural changes of an extracellular matrix protein in NADES. © 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/).


Introduction
We hereby propose a new sustainable concept for potential use in antimicrobial products, e.g., for treatment of infected wounds. This concept combines collagen from rest raw material and a natural deep eutectic solvent (NADES; used as both singular and plural in the following). This combination benefits the unique wound healing properties of collagen with the potential antibacterial effect of NADES [1,2]. The concept focuses on reducing environmental pollution and using sustainable resources, like "green", natural solvents and rest raw material. A successful combination depends on the molecular stability of collagen in the solvent while keeping the eutectic properties of NADES intact. The following paper addresses the structural and thermal properties of collagen in a selected NADES and maintenance of eutectic properties. Both issues are of great importance in the preformulation of a potential therapeutic product based on collagen and NADES.
Collagen is the major fibrillar component and protein in both human and animal connective tissue. Collagen has a triple helix structure formed by three α-chains held together by hydrogen bonds. The chains consist of repeating triplets of the amino acid glycine, followed by often proline and hydroxyproline. Collagen has nonhelical telopeptides attached to the ends of the molecule in its post-translational form [3,4].
The telopeptides can be cleaved off by pepsin digestion to produce atelocollagen. The amino acid tyrosine is located at the telopeptides, and cleavage of these will result in a low tyrosine content [4,5]. While telocollagen can produce immunogenicity, atelocollagen is considered biocompatible and well tolerated by the human body. Telocollagen is soluble in weak acids, while atelocollagen is soluble in both pepsin and in weak acids [6]. Both collagen and collagen peptides have demonstrated excellent wound healing properties by the attraction of cells involved in the rebuilding of the extracellular matrix and skin [2,7].
NADES are regarded as a third class of liquids in organisms, different from water and lipids, which is present in all living cells. NADES were first described by Choi et al. in 2011 [8]. They solely consist of natural compounds, i.e., primary metabolites (e.g., organic acids, amino acids, sugars, polyols, and tertiary amines). It is postulated that NADES have a central role in plants' ability to survive extreme conditions, such as cold and drought [8]. Apart from solubilizing plant metabolites, they have been reported to solubilize both small molecules such as itraconazole, curcumin and porphyrins, and proteins, such as gluten and laccases [9][10][11][12][13][14][15][16]. NADES are considered as "green" solvents compared to conventional organic solvents. NADES have shown antimicrobial effect in the absence and presence of light. The antimicrobial effect is present in pure NADES, in aqueous dilutions of NADES up to 1:200, or in combination with photosensitizers and light under the production of toxic reactive oxygen species. This was first reported by Wikene et al. in 2017 [1]. NADES have also shown antioxidative properties [17].
A polyol compound is an important component of some NADES. Polyols have been reported to stabilize the triple helix of collagen to various extents, depending on the number of carbon atoms present in the polyol molecule. The stabilizing effect is suggested to be achieved through the binding of the polyol to the surface of the collagen molecule followed by the formation of additional hydrogen bonds [18,19]. Although this should indicate a stabilization of collagen in NADES, the properties of polyols when part of a eutectic mixture could be different from a polyol solution. In NADES, the components are tightly bound in a network of hydrogen bonds, which can affect the way the solutes reacts with the surrounding media.
NADES have been proposed as potential excipients in pharmaceutical preparations and drug delivery systems, particularly because of their solubilizing properties, varying viscosity and antibacterial properties [1,16]. Collagen has good biocompatibility, and both collagen and collagen peptides have as mentioned above, beneficial properties for wound healing. The combination of NADES and collagen has the potential to be included in different types of topical formulations, e.g., spray formulation, personalized products from 3D printing, and wound dressings. Collagen to be used in wound products should retain the chemotactic properties important for wound healing, either as a triple helix or in fragmented form as collagen peptides [2,7]. For other purposes like 3D printing, the collagen should be able to be crosslinked, either physically by pH and temperature, or by a crosslinker. This requires an intact triple helical structure [20].
The aim of the present study was to investigate the physicochemical properties of collagen in a selected NADES and aqueous dilutions thereof to identify potential combinations suitable in pharmaceutical preparations. Both pepsin soluble collagen (atelocollagen) and acid soluble collagen (telocollagen) were studied. The selected NADES contained an organic acid (citric acid) and a polyol (xylitol). This NADES has shown antibacterial effect combined with unique solvent properties and is, therefore, a candidate excipient in antimicrobial products [1]. The unfolding, thermal properties, fragmentation and viscosity of telo-and atelocollagen in NADES were assessed. Freeze-dried collagen sheets with NADES were prepared as a potential wound dressing. The structure and mechanical properties of the sheets were evaluated. Our data reveals the potential of the selected NADES and aqueous dilutions thereof as excipients in collagen-based products. This is to our knowledge the first study describing solubility and behavior of an extracellular matrix protein in NADES.

Materials and methods
All experiments were performed at 25°C unless other stated. The data were presented as mean ± highest deviation from three independent experiments.

Materials
Pepsin soluble collagen (atelocollagen) isolated from industrially produced turkey (Meleagris gallopavo) rest raw materials was prepared as described in Grønlien et al. [21]. Acid soluble collagen (telocollagen) from calf skin (Sigma-Aldrich, C9791) and other reagents were of analytical grade and were purchased from either Sigma-Aldrich Chemical Company (St. Louis, MO, USA) or Merck KGaA (Darmstadt, Germany).

Preparation of the natural deep eutectic solvent (NADES)
The selected NADES was prepared by a solvent evaporation method, according to Wikene et al. [12]. The two components of the NADES were dissolved in warm Milli-Q water (~50°C) and evaporated at 45°C for 20 min with a rotary evaporator. The liquid obtained was transferred to polypropylene tubes with a tight cap. Water content was determined by Karl Fischer titration (C20 Coulometric KF Titrator, Mettler Toledo Inc., Schwerzenbach, Switzerland). The NADES prepared contained citric acid/xylitol (molar ratio 1:1) (abbreviated CX). The NADES was used in the undiluted form or after dilution in Milli-Q water 1:1, 1:10, 1:50, 1:100 and 1:200. The pH was measured using a pH 526 MultiCal® pH meter (WTW GmbH, Weilheim, Germany).

Intrinsic viscosity and estimated average molecular weight of collagen
Collagen isolated from turkey tendon was dissolved in 0.5 M acetic acid to a concentration of 0.5 mg/ml. The solution was diluted to concentrations between 0.1 and 0.5 mg/ml. Acetic acid (0.5 M) was used as negative control. The viscosity of the solutions was determined with an Anton Paar Rheometer (Anton Paar Physica MCR301 Rheometer, Germany). Samples of 10 ml were measured with a double gap concentric cylinder (DG 26.7) at 25°C with a shear rate of 10 s −1 . The intrinsic viscosity of collagen was calculated from the dynamic viscosity [22]. The average molecular weight was then estimated from the intrinsic viscosity according to the Kuhn-Mark-Houwink-Sakurada equation (Eq. (1)): where η is the intrinsic viscosity, M is the average molecular weight, K and α are values specific for the polymer or protein (1.86 × 10 −19 and 1.8 for collagen, respectively) [23].

Preparation of samples for spectroscopy
Collagen isolated from turkey filet tendon or collagen from calf skin was dissolved at 1.5 mg/ml in NADES, aqueous dilutions of NADES or 0.02 M acetic acid. The samples were gently stirred overnight at room temperature (IKA® RO 15, IKA Werke, GmbH, Staufen, Germany, 400 rpm) protected from light and filtered (5 μm Versapor® Membrane, Pall Corporation, MI, USA) prior to the spectroscopic measurements.

UV-Vis spectrophotometry
Absorption spectra were recorded between 190 and 700 nm on a Shimadzu UV-2101 PC (Kyoto, Japan) UV-Vis scanning spectrophotometer using a quartz cuvette with a 1 cm cell path.

Fluorescence spectroscopy
Fluorescence measurements were performed on a Photon Technology International modular fluorescence system (London, Ontario, Canada), Model 101 monochromator with f/4 0.2-m Czerny-Turner configuration. The instrument was equipped with a red-sensitive photomultiplier. The excitation source was a 75 W xenon lamp. The emission and excitation spectra were automatically corrected for both the lamp spectral radiance and the detector quantum efficiency by means of the acquisition software (FeliX32, PTI). The excitation and emission monochromator bandpasses were set at 2 or 5 nm for recording of emission spectra and fluorescence quenching measurements respectively, and at 10 nm for anisotropy measurements. The excitation wavelength was 270 nm or 295 nm (anisotropy measurements of turkey collagen). Correction for the difference in absorbance between samples at the excitation wavelength was performed when relevant. The measurements were performed in quartz cuvettes with 1 × 1 cm cell path or in micro cuvettes at 25 ± 0.1°C (n = 3). The anisotropy measurements were performed by the L-format (single channel) method.

Viscosity measurements
Viscosity measurements were performed on a Brookfield DV2T viscometer (Brookfield Engineering Laboratories, Inc., Middleboro, MA, USA) with spindles CPA-40Z (low viscosity samples b 10 mPa s; accuracy: ±0.1 mPa s; sample volume: 1.5 ml) and CPA-52Z (high viscosity samples ≥ 10 mPa·s; accuracy: ±3.1 mPa·s; sample volume: 0.5 ml). The temperature was kept constant at 25 ± 0.1°C during the viscosity measurements (Grant LTD6G water bath, Grant Instruments, Cambridge, Ltd., Royston, UK). A single point viscosity measurement was performed with the end condition parameter fixed at 2 min and speed 30 rpm for all samples.

Differential scanning calorimetry
Differential scanning calorimetry (DSC) experiments were performed using a Nano DSC 602000 differential scanning calorimeter (TA Instruments, Lindon, UT, USA) with a capillary cell volume of 0.300 ml. DSC thermograms were recorded for 1.5 mg/ml collagen solutions in selected media at a constant heating rate of 2°C/min and 3 atm in the temperature range of 20-60°C. The respective NADES in their current dilutions were used as baseline references. The transition temperature of collagen was defined at the maximum of the transition peak after baseline subtraction and processing.
The results were processed by application of the NanoAnalyze software (TA Instruments). The partial specific heat capacity of collagen was determined from the estimated average molecular weight and a partial specific volume of 0.700 ml/g [26]. The results were fitted to a two-state trimer-to-monomer model. The deviation between the recorded data and the modulated data was ≤3%.

Preparation of collagen-NADES sheets
Collagen-NADES sheets were prepared by a freeze-drying method. Collagen (both qualities) was solubilized overnight in NADES CX diluted 1:1, 1:10, 1:50, 1:100 and 1:200 with Milli-Q water to a final concentration of 3.0 mg/ml. The solutions were transferred to a 6-well multiplate (Corning Life Sciences, Tewksbury, MA, USA) (2 ml) and freeze-dried in order to form the sheets. The solutions were frozen at −80°C for 1 h prior to freeze-drying at 0.0019 mbar for 20 h, including 1 h final drying at 0.0010 mbar (Alpha 2-4 LD Plus Freeze Dryer, Martin Christ, Osterode am Harz, Germany). Collagen dissolved in undiluted NADES is not applicable, as undiluted NADES will neither freeze nor freeze-dry under the actual conditions.

Fourier transform infrared spectroscopy (FT-IR) of collagen-NADES sheets
FT-IR spectra of undiluted NADES CX and the freeze-dried collagen-NADES sheets were acquired using a Nicolet™ iS™ 5 FTIR Spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with an iD5 diamond ATR Accessory. For each sample, 16 scans from 4000 cm −1 to 550 cm −1 were collected in single beam mode with a spectral resolution of 4 cm −1 . The smoothed spectra are presented.

Mechanical force-displacement studies of collagen-NADES sheets
Mechanical properties (force-displacement) of the collagen-NADES sheets were studied using a TA-XT2i Texture Analyser (Stable Micro Systems, Haslemere, UK) in compression mode. The sheets were exposed to constant pressure at 0.5 mm/s with a 2 mm probe. Forcedisplacement curves of the collagen-NADES sheets were compared to collagen sheets without NADES (n = 3-6). The force was defined as the maximum force measured and displacement was defined as the displacement at the maximum force.

Intrinsic viscosity and estimated average molecular weight of collagen
The intrinsic viscosity of collagen isolated from turkey tendon was found to be 1721 cm 3 /g. The average molecular weight was estimated to be 445 kDa. Commercially available collagen from calf skin has no reported molecular weight, but the specified production method indicated an isolation without pepsin, producing acid soluble (telo) collagen. In the current experiments, the average molecular weight was set to 300 kDa.

SDS-PAGE
The SDS-PAGE pattern of collagen (both qualities) in acetic acid indicated the presence of α-chains from both collagen type I and III [25,27,28]. For collagen dissolved in aqueous dilutions of NADES, the pattern showed the most extensive fragmentation of the molecule in dilution 1:10 in case of calf skin collagen and dilutions ≤1:50 in case of turkey collagen. The pattern for dilutions N1:50 appeared similar to collagen dissolved in acetic acid (Fig. 1).

Absorption spectra
Collagen (both qualities) in acetic acid displayed an absorption maximum around 220 nm and a broad shoulder in the range 250-290 nm with a diffuse maximum around 275 nm (Table 1). Dissolution in undiluted NADES resulted in a broad peak with low absorbance and a maximum at 265 nm and shoulders at 258, 268, 275, and 281 nm (calf) or 281 nm with shoulders at 269, 265, and 259 nm (turkey). Dissolution in NADES diluted 1:1 induced a blue shift of the main peak of 32 nm and 16 nm for turkey and calf skin collagen, respectively. The resulting absorption maximum was similar in both samples (249 nm). The absorption maximum was further blue-shifted upon further dilution (Fig. 2). The absorption maximum in the individual NADES dilutions was similar for both collagen qualities. A hyperchromic effect occurred at dilutions ≥1:50 in the case of calf collagen and ≥1:1 in case of collagen from turkey. A shoulder at approximately 280 nm remained virtually constant in all the diluted samples (Fig. 2). There was a linear relationship (r 2 = 0.998, turkey collagen; r 2 = 0.975, calf skin collagen) between the absorption maximum (Table 1) of collagen and the pH of the NADES solution (Table 2) in the dilution range 1:1-1:100.

Fluorescence spectra
An excitation wavelength (270 nm) corresponding to the tyrosine absorption was selected. The emission maximum was 297-303 nm in all the samples containing collagen from calf skin. The emission maximum in samples made from turkey collagen showed a bathochromic shift and was detected between 319 nm and 339 nm with a shoulder around 298 nm (Table 1, Fig. 3). The fluorescence excitation spectrum at emission 340 nm was recorded in acetic acid and in NADES diluted    (Table 1).

Steady state fluorescence anisotropy
The fluorescence anisotropy was independent of the solvent in samples containing turkey collagen but did increase upon dilution of the NADES in calf skin samples. The latter showed a linear increase as a function of NADES dilution (r 2 = 0.992; except 1:50); i.e., by a decrease in sample viscosity (Table 1).

Differential scanning calorimetry
The thermal transitions of collagen from turkey tendon and calf skin were dependent on the dissolution medium. Collagen exhibited both a minor (T s ) and a major thermal transition (T m ) in acetic acid and aqueous dilutions of NADES N 1:50. In NADES diluted 1:10 in the case of calf skin collagen, only T m was present (Table 1). No signals were detected in undiluted and 1:1 dilution of NADES. The denaturation temperature showed a linear increase as a function of increased pH between dilution 1:10 and 1:100 (r 2 ≥ 0.999, turkey: T s and T m ; calf: T m ).

Characterization of collagen-NADES sheets
Freeze-drying of collagen dissolved in aqueous dilutions of NADES CX between 1:50 and 1:200 formed sheet-like sponges with different plasticity. Dilution 1:10 resulted in a sticky fibrous layer, which was difficult to handle. Dilution 1:1 resulted in a transparent product (i.e., no sheet formation).
Force-displacement studies demonstrated a plasticizing effect by aqueous dilutions of NADES on collagen sheets. However, the mechanical strength of the sheets was also dependent on the collagen source. Only sheets made from turkey tendon collagen could be handled, whereas sheets made from calf skin collagen were sticking to the container and appeared too fragile to be tested. The results are summarized in Table 3 and representative figures are shown in Supplementary Material, Fig. S1a-d.

Discussion
Type I collagen has a low content of aromatic amino acids, which are represented by tyrosine and phenylalanine. The tyrosine residues are located at the non-helical telopeptides [30,31]. Collagen from calf skin contains telopeptides, while turkey collagen has been treated with pepsin to remove these peptides. It has, however, been demonstrated that telopeptides are not completely excised by pepsin treatment [30]. This can explain why tyrosine was detected in both types of collagen. An absorption peak at~276 nm and a shoulder at~282 nm would reflect the presence of tyrosine residues while peaks at 268, 265, and 258 nm indicate the presence of phenylalanine [30]. All these peaks were identified in collagen samples dissolved in undiluted NADES. The absorption peak of phenylalanine (i.e., 265 nm) emerged as the highest intensity peak in the calf collagen spectrum in undiluted NADES, while the tyrosine absorption (i.e., 281 nm) was dominating in samples of turkey collagen under similar conditions. The hypsochromic effect showed a linear   dependency of the solution pH which increased from 1.0 to 2.4 upon NADES dilution. The pK a value of tyrosine and phenylalanine is approximately 2.2 and 2.0, respectively. A hyperchromic effect in turkey collagen occurred at dilutions ≥1:50 which corresponded to a pH similar to the tyrosine pK a value. A similar hyperchromic effect in calf collagen was observed at lower dilutions, i.e., at a pH closer to the pK a value of phenylalanine. It has previously been reported that hypsochromic and hyperchromic shifts in collagen samples could indicate an uncoiling of the triple helix, i.e., a denaturation of the protein [31,32]. However, since the components in NADES and the solutes are tightly bound together in a network of hydrogen bonds, the hyperchromic effect observed upon dilution could also indicate a weakening in the hydrogen bonding between collagen and the NADES network, leading to collagen stabilization as observed in acetic acid. This is consistent with a bathochromic shift in fluorescence in the case of turkey collagen (i.e., tyrosine/tyrosinate fluorescence). Tyrosine side chains are often distributed from the interior to the surface of a protein, while phenylalanine residues are highly apolar and thereby buried in the interior of the protein, making them less sensitive to changes in the solvent (e.g., polarity, H-bonding) [33]. Collagen belongs to the group of tryptophan-free proteins. Excitation at 270 nm should, therefore, induce tyrosine fluorescence (maximum emission near 305 nm) without any interference from tryptophan (maximum emission near 350 nm). However, unusual tyrosine emission at longer wavelengths has previously been reported in proteins [34]. This can be observed as a weak shoulder on the tyrosine band in the range 330-350 nm and has been ascribed to tyrosinate emission [35]. In the present study, the tyrosinate emission was intense and emerged as the main peak in turkey collagen samples while absent in calf skin collagen. Ground state tyrosine has a pK a value of 10.3 and tyrosinate fluorescence is, therefore, most easily observed at high pH. The pK a value does, however, decrease to about 4 in the excited state. Tyrosinate emission can for that reason occur at neutral pH from the singlet excited tyrosine [36]. This is facilitated by the presence of proton acceptors like acetate in the solvent but is also dependent on to which extent the tyrosine is exposed to the aqueous phase. Tyrosinate emission can also be achieved by excited-state proton transfer to adjacent acceptor groups within the protein, e.g., carboxylate, histidyl, and lysyl residues or the side chains of aspartate and glutamate residues [35,37]. The excitation spectrum for the tyrosinate emission is expected to be independent of pH below the pK a of ground-state tyrosine if the emission arises from singlet excited tyrosine. This was the case in the present study when the pH varied from 1.0 to 3.2. The intensity of the tyrosinate emission (corrected for variation in absorbance) showed a linear decrease as a function of decreased viscosity in the diluted NADES samples. The fact that tyrosinate emission occurred only in turkey collagen (including samples in 0.02 M acetic acid) might indicate that tyrosine was more exposed to the aqueous phase than in calf skin collagen, which is in agreement with the UV absorption measurements (see above) and/or the presence of nucleophilic group(s) in the tyrosyl microenvironment in turkey collagen. Further, it has previously been reported that tyrosine residues in calf skin collagen mainly are located in hydrophobic regions of the protein that are poorly accessible to the surrounding solvent, which supported the above results [38]. The intensity of the tyrosine emission from calf skin collagen was quite independent of NADES concentration (i.e., viscosity and pH). This emphasized that the rate of deactivation of the tyrosine excited state in calf skin collagen was fast relative to the rate of deprotonation (i.e., formation of tyrosinate).
The anisotropy in dilute non-viscous solutions is primarily determined by the rotational motion of the fluorophore. In the case of proteins, these motions will rely on e.g., the size, shape, and extent of aggregation of the molecule. The samples in the present study varied from low to medium viscosity dependent on the extent of NADES dilution, which complicated the interpretation of the results. Further, all the samples were slightly turbid in spite of filtration prior to the spectroscopic measurements. The observed anisotropy can be expected to decrease linearly with an increase in turbidity. The contribution of scattering to depolarization can be evaluated by changing into a cuvette with smaller dimensions. Both conventional cuvettes (1 × 1 cm) and micro cuvettes (2 × 2 mm) were applied in the present work giving similar results (data not shown). It is therefore unlikely that depolarization due to scattering made a major contribution to the observed values.
The steady-state fluorescence anisotropy of tyrosinate emission in turkey collagen remained constant independent of NADES dilution; i.e., independent of the supramolecular structure, pH, and viscosity of the solvent. Further, the anisotropy was virtually independent of the excitation wavelength, which indicated that the angle between the absorption and emission dipoles did not change upon changes in the macro environment. Fluorophores bound to proteins can be independent of the overall rotational diffusion by displaying fast segmental  motions [39]. The result is that the anisotropy becomes rather insensitive to macroscopic viscosity [34]. This could explain the observations on turkey collagen and emphasized the assumption that the tyrosine residues were not deeply embedded in this protein. On the other hand, the fluorescence anisotropy from tyrosine emission in calf skin collagen was dependent on the NADES dilution and was apparently an inverse function of the viscosity. It is evident from the calorimetric measurements that the conformation of calf skin collagen changed upon dilution of the NADES. This had apparently a palpable effect on the emission properties of the tyrosine fluorophores in calf skin collagen. Collagen normally exhibits two thermal transitions when heated. The minor transition has previously been ascribed to collagen fibril depolymerization, unfolding and melting of small parts of the triple helix structure or been related to the thermally labile regions with hydroxyproline deficient sequences, while the major transition has been ascribed to denaturation of the triple helix to form monocoils [40,41]. The thermograms demonstrated a lack of protein unfolding prior to denaturation in undiluted NADES and samples diluted 1:1 and 1:10 in case of calf skin collagen; i.e., only the denaturation temperature could be detected. This indicated either that collagen is thermostable in concentrated NADES, or that unfolding of the triple helix had occurred at room temperature in these solvents. The latter hypothesis was supported by SDS-PAGE, where collagen in NADES diluted 1:10 showed higher fragmentation than dilutions ≥1:50. It is reasonable to infer that dilution of NADES will cause conformational changes in the protein due to changes in the supramolecular network formed in deep eutectic solvents combined with the change in pH. The triple helix, therefore, seemed to be maintained at dilutions ≥1:10 illustrated by the occurrence of two peaks in the thermograms that are typical for unfolding and denaturation, respectively. Further, this is consistent with the increase in fluorescence anisotropy observed by dilution of the NADES, best illustrated by calf skin collagen. In very dilute NADES samples (i.e., 1:200); the anisotropy was virtually similar to samples in 0.02 M acetic acid representing a stable collagen structure. This is as well consistent with the hypothesis that the hyperchromic effect observed upon dilution of the NADES was caused by a weakening of the eutectic network, rather than denaturation of collagen. The presence of xylitol in the eutectic mixture had apparently no stabilizing effect on the collagen triple helix, which was an effect hypothesized from studies where collagen exhibited increased thermostability in the presence of polyols [18,19].
The denaturation temperature showed a linear increase as a function of increased pH between dilution 1:10 and 1:100. This was valid for both the minor and major transition of turkey tendon collagen, but only for the major transition of calf skin collagen. Collagen has shown low protein solubility, water binding capacity and apparent viscosity at pH ≤ 1, caused by a partial denaturation [42]. This can explain why the collagen lacked an unfolding transition prior to denaturation in the thermograms and the fragmentation observed by the SDS.
The correlations demonstrated by the overall results were valid up to a NADES dilution between 1:100 and 1:200 with water, e.g., the hyperchromic effect of collagen absorbance at the absorption maximum observed upon dilution of NADES was present up to a 1:200 dilution. Further, at this dilution, the anisotropy became virtually similar to collagen dissolved in 0.02 M acetic acid. Together, these results indicated that the eutectic network was preserved upon dilution to this ratio. This is in accordance with previous results obtained in our lab where the eutectic network appeared to be preserved upon dilution ≤1:200 [1]. However, a report in the literature claims that the network was absent in dilutions ≥1:1 [43]. The NADES network structure is probably dependent on a series of factors, e.g., ionic strength, pH and chemical properties of the solute.
The results on collagen-NADES sheets were consistent with the spectroscopic and thermal observations on collagen dissolved in NADES solutions, indicating that the structural properties of collagen was maintained when the NADES was diluted ≥1:10. This can be illustrated by e.g., comparing the FT-IR spectra of the freeze-dried sheets with the undiluted NADES. The main peaks of collagen (amide A, B, I, II and II) were clearly present in collagen-NADES sheets prepared with NADES diluted 1:50-1:200, whereas the sheets prepared with NADES diluted 1:1 and 1:10 showed weak or no amide I or II signals. Conformational changes in the secondary structure of the collagen molecule can be followed by the changes in the amide profile in the FT-IR spectrum. The profile of amide I is associated to the α-helix conformation, the βsheet conformation and the β-turn. An apparent loss in these structures is indicated by the lower intensity of this peak, which was observed for collagen-NADES sheets prepared with NADES diluted 1:1 and 1:10. A shift of the C_O stretching signal of the citric acid component of the NADES (i.e., from 1692 to approximately 1714 cm −1 ) can be a measure of hydrogen bonding between the reagents [44]. Such a shift was observed in the plain NADES (undiluted) and in all the NADES samples containing collagen. The water will evaporate upon freeze-drying of the collagen sheets and leave concentrated NADES within the protein structure. This was confirmed by the FT-IR results obtained for collagen-NADES sheets when compared to the undiluted NADES. The presence of concentrated NADES within the sheets will maintain the solubilizing properties and potential antibacterial properties of this solvent.
Mechanical force-displacement studies of the formed sheets clearly showed a plasticizing effect by the NADES and aqueous dilutions thereof compared to a sponge of collagen without NADES. The plasticizing effect can be attributed to the high concentration of NADES, the presence of polyols, a crosslinking effect of the NADES components or a combination of all. Some of the force-displacement results expressed quite large standard deviations. This can be caused by a variation in the organization of the collagen fibers formed during freeze-drying or how the NADES is organized within the collagen structure. Highly concentrated NADES may dissolve the collagen and result in varying mechanical properties. The mechanical properties of collagen sheets prepared with highly diluted NADES seem to approach the properties of collagen sheets prepared in acetic acid, although some of the plasticizing effects by the NADES are maintained. Further, the NADES may work as a plasticizer as demonstrated by the force-displacement results discussed above, to form bioplastics, offering an alternative to toxic or nonbiodegradable plasticizers [45]. Citric acid has previously been investigated as a potential natural crosslinker for collagen, increasing the mechanical properties of collagen sheets. Andonegi et al. [46] investigated how compressed collagen sheets were influenced by different concentrations of citric acid. Addition of low amounts of citric acid did not change the structure, but higher contents changed the structural order of the collagen sheets [46]. In a NADES based on citric acid, the acid is maintained within a hydrogen-bonded network, resulting in supersaturated solutions. The high concentration of citric acid may induce a change in the collagen structure. Xylitol has previously been tested for a plasticizing effect on squid protein films, where it showed promising properties, although the films became brittle over time [47]. It is however, likely that the properties of xylitol or other components can be modified when they are part of a NADES network. A choline chloride and glycerol based NADES has been evaluated for a potential plasticizing effect on pectin films and showed promising properties as excipients in bio-based plastics formulations [48]. The addition of plasticizers to a collagen sheet can be used to control the mechanical properties of the constructs. It appeared that a 1:100 dilution of the NADES was the optimal concentration in this study, maintaining both the collagen structure and a plasticizing effect. A collagen-NADES sheet formulation has to be optimized further to find the optimal NADES composition and concentration.
Based on the above results and previous knowledge about the antibacterial properties of NADES and the unique wound healing properties of collagen, a combination of NADES CX and collagen represents a promising formulation strategy for topical preparations. Inclusion of a photosensitizer could allow for additional application in antimicrobial photodynamic therapy (aPDT). In cases where there is a need to preserve the collagen triple helical structure, an aqueous dilution of the NADES between 1:10 and 1:200 can be selected. Although collagen seems fragmented in undiluted NADES, it is still possible that the chemotactic properties are preserved. The application of undiluted or slightly diluted NADES in collagen preparations should therefore not be ruled out.

Conclusions
Collagen exhibited variable unfolding properties based on the type and method of isolation and the degree of aqueous dilution of the added NADES. The results indicated that both telo-and atelocollagen were more susceptible to molecular changes when dissolved in undiluted NADES than in acetic acid. However, a formulation based on undiluted NADES and collagen can benefit from both chemotactic properties by the attraction of cells involved in wound healing by collagen peptides and the unique antibacterial properties of the NADES. The results obtained in highly diluted samples approached the data obtained in acetic acid. Further, the results indicated that the intramolecular NADES network was maintained up to a dilution ≤1:200, which preserves the unique effects of NADES. The eutectic solvents have previously shown antibacterial properties in aqueous dilutions up to 1:200 [1]. Further, an increased mechanical strength of freeze-dried collagen-NADES sheets and a plasticizing effect of NADES at low concentration, were demonstrated in the present work. The combination of diluted NADES and collagen seemed suitable for further development into a topical preparation.

Declaration of competing interest
The authors declare that they have no conflict of interest.