*2.1. Alkaline Pre-Treatment of Skin*

The average (±standard deviation (SD)) chemical composition of non-treated skin from the small spotted catshark expressed as dry weight is shown in Table 1.

**Table 1.** Approximate composition (media ± standard deviation (SD)) expressed as percentage of dry weight of non-treated skin from the small-spotted catshark.


Hydroxyproline (HPro) content was used as an estimation of initial collagen content in the nontreated skins, considering that the ratio of HPro in collagen is 12.5 g of HPro/100 g of collagen [20]. Thus, the determined collagen content was 34.22% (g collagen/100 g dried skin). Collagen recovered (g collagen/100 g of collagen in non-treated skins) was estimated in the solid skin residues and in the filtrated liquid for the 20 experiments carried out during the experimental design, from the Kjeldahl determined nitrogen using a factor of 5.4 [21] (Table S1, Supplementary Material).

Experimental data from Table S1 were modelled using second-order equations (Table 2). These polynomial models describe the correlation between variables and the corresponding response followed the general form defined by Equation (1).

**Table 2.** Second-order equations describing the effect of temperature (T), time (t) and concentration of NaOH (M) on the efficiency of collagen recovery (%) from the skin of the small-spotted catshark. The coefficient of adjusted determination (R2 adj) is also shown. Optimum (opt) values of each independent variable to obtain maximum responses are also shown.


The R2 adj values revealed good agreement among experimental and predicted data described by the second-order equations proposed (a high proportion of variability, more than 81% for both solid residue skins and filtrated liquid, was achieved). The consistency of the polynomial equations was validated since the F1 and F2 ratios from F-Fisher test were satisfied in all cases (data not shown). The results of the multivariate analysis showed significant quadratic terms for temperature, NaOH concentration and time (Student's *t*-test, *p* < 0.05) in the estimated collagen present in both fractions. In the solid fraction, this outcome is graphically translated as a concave surface where the collagen recovery increases with lower temperature, lower concentration of NaOH and low reaction times (Figure 1). The inverse response obtained for temperature, NaOH concentration and time in the filtrated liquid (convex surface) is in agreement with the fact that collagen recovered in the solid fraction is not present in the filtrated liquid fraction. Among the three independent variables, NaOH seems to have a slightly higher effect on collagen recovery in both fractions.

The variables maximizing the recovery of collagen in the solid fraction were 4 ◦C, 2 h and 0.1 M NaOH. However due to industrial constraints, mainly due to the high cost of low temperature processes, the temperature of 8.3 ◦C was selected for the next optimization step. Thus, the consensus values for the subsequent acid-soluble collagen extraction step were a temperature of 8.3 ◦C, a treatment time of 2 h and a NaOH concentration of 0.1 M.

Although previous studies have also shown that a low impact NaOH pre-treatment has a positive effect on the collagen yield, this is the first time that an optimization study has been carried out regarding the skin NaOH pre-treatment. Our results show that as little as two hours of treatment is enough to condition the skin, making it suitable for the posterior acid treatment. Thus, Woo et al. [12] have found that treatment times between 12 h and 36 h and NaOH concentration values between 0.5–1.3 M positively affects the achievement of maximum values of collagen content extracted from yellowfin tuna skin. Zhou and Regenstein [22] found that significant amounts of collagens are lost when pre-treatment conditions include concentration values higher than 0.5 M NaOH, reaction time of 4 days and temperature of 4 ◦C. Liu et al. [15] have also studied the effect of different alkaline pre-treatment conditions on the acid-soluble collagen obtained from grass carp, concluding that temperature ranges of 4–20 ◦C for pre-treatment conditions and NaOH concentration between 0.05 and 0.1 M were adequate. Wang et al. [13] also employed 0.1 M NaOH to remove non-collagenous proteins from the skin of grass carp with low temperature (4 ◦C) but higher reaction time (6 h). These results suggest that the efficiency of alkaline pre-treatment may vary between fish species and also between temperature, time and NaOH concentration conditions, highlighting the importance of specific two-step optimization studies for different species including these three variables.

**Figure 1.** Combined effect of alkali concentration (M), time (t) and temperature (T) on the removal of collagen from the skin of the small-spotted catshark. Collagen recovered in the solid fraction (**a**–**c**). Collagen recovered in the liquid fraction (**d**–**f**).

#### *2.2. Acid-Soluble Collagen (ASC) Extraction Stage*

The next experiment was designed for the optimization of collagen extraction in acidic media using NaOH pre-treated skin. In this case, the combined effect of acetic acid concentration, temperature and time of processing on collagen production was studied. The average (±SD) chemical composition of NaOH treated skins (under the optimal consensus values obtained in the first experimental optimization stage, expressed as percentage of dry weight) used for this second experimental design was 76.55 ± 1.22% of moisture content; 56.71 ± 0.61% of protein content; 0.59 ± 0.10% of lipid content and 46.49 ± 0.38% of ash content. Compared to the approximate composition of non-treated skins, the protein and lipid content decreased significantly (Kruskal–Wallis test for protein: chi-square = 5.398, d.f. = 1, *p* = 0.020; ANOVA for lipid: F1,4 = 299.483, *p* < 0.01), confirming the removal of unwanted materials [23]. The significantly higher ash content observed in NaOH treated skin (ANOVA for ash: F1,4 = 1758.801, *p* < 0.01 is due to the NaOH added. A representation of the lyophilized collagen obtained in some of the 20 experiments is shown in Figure 2. The corresponding amino acid composition from all collagens is summarized in Table 3. In addition, the yields of lyophilized collagen recovered varied between 18.33% and 49.65% and are defined in Table S2.

**Figure 2.** Dialyzed (**a**) and lyophilized (**b**) collagens obtained in each of the 20 experiments developed for the acid-soluble collagen extraction stage experimental design.

**Table 3.** Hydroxyproline (HPro), Proline (Pro) and Glycine (Gly) content in lyophilized extracted collagen obtained in each of the 20 experiments developed for the acid-soluble collagen extraction stage of the experimental design. Real values of independent variables are indicated, as well as the codified values (in brackets).


The dependent variables (responses) evaluated were HPro, Gly, Pro and the sum of Pro + HPro (imino acids) as well as the yield of collagen recovered. Table 4 summarizes the equations obtained from the mathematical modelling and multivariable statistical analysis of the experimental responses mentioned. The accuracy between experimental and theoretical data were remarkable with values of R<sup>2</sup> adj > 0.85. The robutness of the different response selected and the reproducibility of collagen production was confirmed by the fact that equations and theoretical three-dimensional (3D) surfaces were similar in all cases studied (Figure 3). As in the previous factorial design, the consistency of the equations was also found: all ratios F1–F4 were validated (data not shown). Finally, the values of the independent variables which maximize the recovery of collagen were a temperature of 25 ◦C, a time of 34 h and a concentration of 1 M acetic acid. Using these optimal extraction conditions, the yield of collagen obtained was 61.24% (g of collagen/100 g of initial collagen in skin), which is higher than that obtained previously [6].

**Table 4.** Second-order equations describing the effect of temperature (T), time (t) and concentration of AcOH (M) on the collagen recovery by means of HPro, Gly, Pro, HPro + Pro and yield determination, from the skin of the small-spotted catshark. The coefficient of adjusted determination (R2 adj) is also shown. Optimum values of each independent variable to obtain maximum responses are also shown.


a b c

**Figure 3.** Combined effect of acetic acid (AcOH), time (t) and temperature (◦C) on HPro released (**a**–**c**) and collagen yield (**d**–**f**) produced from *S. canicula* skins.

Previously published results on acid-soluble collagen extraction from the skin of the small-spotted catshark [6] showed lower yield values as the extraction conditions were different than the optimum values presented here. Several studies have also focused on the extraction and characterization of acid-soluble collagen from different marine fish species, traditionally using 0.5 M acetic acid at around 4 ◦C [22] without a previous optimization study [24,25]. In recent decades, several manuscripts have addressed the study of the optimization conditions of collagen extraction from different sources; however, not many of those optimize the complete extraction procedure, including both the alkaline and the acidic stages. Thus, Wang et al. [13] found higher yields of acid-soluble collagen from the skin of grass carp with increased acetic acid concentration (up to 0.5 M) and increased reaction times (up to 32 h), while the optimum temperature differs with different levels of acetic acid or reaction time. The collagen yield reported by these authors was lower than the one obtained in this study. As in the previous alkaline pre-treatment optimization stage, the efficiency of the acidic extraction stage varies between fish species and also with temperature, time and concentration of acetic acid, suggesting the great importance of specific optimization studies for different species including the three variables involved in the process.

As shown in Figure 4, ASC extracted under the different experimental conditions used in this work resulted in similar electrophoretic patterns, which consisted of the typical heterotrimer collagen structure containing two identical α<sup>1</sup> chains (approximately 120 kDa) and one α<sup>2</sup> chain (approximately 110 kDa) in the molecular form of [α1(I)]2 α2(I)], and one β dimer of about 200 kDa [6]. The band intensity of the α<sup>1</sup> chain was not two-fold higher than that of α<sup>2</sup> chain; in fact, the α<sup>2</sup> chain is hardly visible. This fact, together with the high intensity of the β dimer, might suggest the existence of higher crosslink degree between α<sup>2</sup> chain in elasmobranchs This has also been found for other elasmobranchs where the α<sup>2</sup> chains are scarcely visible while the β dimer bands are stronger than in other teleosts [17,26,27]. A γ-component can be also seen in all ASC obtained, similarly to previously reported results by Sotelo et al. [6]. The collagen obtained in some of the experimental conditions (corresponding to Experiments 6, 8, 10 and 12) present a few bands below 100 kDa. These low molecular weight components might be the result of the particular extraction conditions on which those collagens were obtained: the highest temperatures, times and AcOH concentrations or the combination of them (further research on the characterization of those components using ionic exchange chromatography might be interesting, however it exceeds the objectives of this study).

**Figure 4.** SDS-PAGE (7%) showing acid-soluble collagen (ASC) obtained in each of the 20 experiments developed for the acid-soluble collagen extraction experimental design. MWM: molecular weight marker.

#### **3. Material and Methods**

#### *3.1. Biological Samples and Compositional Analysis*

Small-spotted catshark (*Scyliorhinus canicula*) individuals obtained approximately 12 h after capture from a local market in Vigo (Northwestern Spain) were manually skinned (ES360570202001 by Galicia Government). Skins were stretched and aligned on top of each other (Figure 5a) in order to select only the central part of the skins (Figure 5b) with the aim of obtaining a homogeneous material. The selected central parts were mechanically cut into small pieces (5 × 5 cm2) (Figure 5c) and then each of those pieces were manually cut into smaller pieces (0.5 × 0.5 cm2) (Figure 5d), mixed thoroughly, separated in sealed plastic bags each containing 5 g of skin and stored at −20 ◦C until used for the experimental designs.

**Figure 5.** Small spotted catshark skin sampling: Skins stretched and aligned on top of each other (**a**); selected central portions of skins (**b**); homogeneized cutsobtained from the selected central parts of skins using scissors and divided in three for sampling purposes (**c**); small pieces finally obtained using a scalper (**d**).

The chemical composition of the skin was evaluated in triplicate by analyzing crude protein, ash, moisture and fat content. Total nitrogen was determined with the Kjeldahl method [28] in a DigiPREP HT digestor (SCP Science, Baie-d'Urfe, QC, Canada), DigiPREP 500 fully automatic steam distillation (SCP Science, Baie-d'Urfe, QC, Canada) and a TitroLine easy titration unit (Schoot, Mainz, Germany), and crude protein content was calculated as total nitrogen multiplied by 6.25. Fat content was determined by the method of Bligh and Dyer [29]. Moisture was determined after heating the sample at 105 ◦C for 24 h, and ash content was determined after heating the sample for 24 h at 550 ◦C [28].

The hydroxyproline content in the skin was determined according to the procedure described in Blanco et al. [30] and used for the estimation of initial collagen content in the untreated skin, considering that the ratio of HPro in collagen is 12.5 g of HPro/100 g of collagen [20].

#### *3.2. Experimental Design and Statistical Analysis*

In this work, two experimental designs were performed to analyze the influence of chemical treatment concentration, temperature and time on the extractability of collagen from the skin of the small-spotted catshark. First, the effect of temperature (T), concentration of NaOH (M) and time (t) on the efficiency of removing non-collagenous proteins was studied (alkaline pre-treatment). Then, the effect of temperature (T), concentration of acetic acid (M) and time (t) on the efficiency of extracting acid-soluble collagen (ASC) was optimized (acid-soluble collagen extraction stage). In both cases, the factorial experiments were rotatable second-order designs with six replicates in the center of the experimental domains [31].

#### 3.2.1. Alkaline Pre-Treatment Experimental Design

The conditions of the independent variables studied in the pre-treatment experimental design were: temperature in the range of 4–25 ◦C, concentration of NaOH in the range of 0.1–2 M and intervals of time between 2–48 h (Table 5). The values of independent variables were selected from previously reported studies to cover a wide range of conditions in order to obtain the values that maximize the isolation of collagen and to reduce the times and concentrations needed for the bioproduction of *S. canicula* collagen. The most common times and NaOH concentration values used in the literature

ranged from 1–36 h and 0.05–1.3 M NaOH. The temperature preferentially chosen in the literature for removing other non-collagen proteins ranged from 4–20 ◦C [11–13].


**Table 5.** Experimental domain and codification of independent variables in the second-order rotatable designs developed for collagen extraction from *S. canicula* skin.

The conditions which were maintained as constants were the solid (skin):alkaline solution ratio of 1:10 and high agitation (200 rpm). The reactions were developed in a stirred and thermostated reactor (100 mL). After each of the 20 alkali treatments, the solutions were filtered using a 35 μm membrane. The filtrate was measured, centrifuged and the supernatant collected to be analyzed in terms of collagen content which was determined by means of total nitrogen content according to the Kjeldahl method [28]. The skin residue of the filter was weighed and also analyzed in terms of total nitrogen content. The dependent variable studied was collagen/initial collagen in skin rate, for both the collagen recovered in the skin residues and the collagen measured in the filtered solution.

#### 3.2.2. Acid-Soluble Collagen Extraction Stage Experimental Design

Skins (500 g) obtained as explained in Section 2.1 were introduced in a stirred and thermostated 5 L reactor connected to a pH electrode and a temperature probe (Afora S.A., Barcelona, Spain). Based on the consensus values obtained in the alkaline pre-treatment experimental design, skins were treated with NaOH and then filtered using a 200 μm membrane. The liquid was removed, and the skins were washed with distilled water until neutral pH was achieved (Figure 6), weighed and divided into 5 g sealed plastic bags which were frozen at −20 ◦C until used for the experimental design. NaOH pre-treated skins (10 g) were used for approximate compositional analysis. Differences in approximate composition between non-NaOH treated skins and NaOH treated skins were statistically analyzed. Prior to analysis, data were checked for normality and homoscedasticity using the Kolmogorov–Smirnov and Levene tests, respectively. The Kolmogorov–Smirnov test showed that protein data and their transformations did not fit with the assumptions of normality. As a consequence, non-parametric statistics were used for these data. Differences in lipids and ashes were compared by one-way ANOVA, with NaOH treated or non-treated skins as the between-subject effect. The Kruskal–Wallis test, the non-parametric equivalent of a one-way ANOVA, was used to examine variations between treated and non-treated skins in protein content. Significance levels were set at *p* < 0.05. Statistical tests were performed with IBM SPSS Statistics 23. 0 (IBM Corp., Armonk, NY, USA).

The conditions of the independent variables studied in the collagen extraction stage of the experimental design were: temperature in the range of 4–25 ◦C, concentration of acetic acid (AcOH) in the range of 0.2–1 M and intervals of time between 2–48 h (Table 4). The conditions that were maintained as constants were solid (skin residue): alkaline solution ratio of 1:10 and high agitation (200 rpm). The values of independent variables were again selected from previously reported studies to cover a wide range of conditions in order to obtain the values that maximize the isolation of collagen and to reduce the time and concentration needed for the bioproduction of *S. canicula* collagen. The most common times and acetic acid concentration values used in the literature ranged from 1–16 h and

0.1–1 M acetic acid. The temperature preferentially chosen in the literature for the acid extraction of collagen ranged from 4–30 ◦C [12,13,24].

**Figure 6.** Filtered and washed NaOH treated skins used for the acid-soluble collagen extraction stage of the experimental design.

The reactions were developed in the same reactor as the pre-treatment experimental design. After each one of the 20 acid experiments, the solutions were filtered using a 200 μm membrane. The filtrated solution was collected, measured, dialyzed and freeze-dried. The freeze-dried collagen was weighed and characterized in terms of collagen content (determined by means of hydroxyproline (HPro), proline (Pro) and glycine (Gly) content), yield of collagen recovered and SDS-PAGE characterization (the content of other amino acid was also analyzed). The dependent variables studied were yield of collagen (as a percentage of collagen recovered/initial collagen in skin) and contents of the collagen characteristic amino acids: HPro, Pro and Gly.
