**3. Discussion**

Seaweeds are considered a promising alternative ingredient for aquafeeds. In addition to having nutrients with potential quantitative interest, the presence of biologically active compounds like polysaccharides, pigments (chlorophylls and carotenoids), sterols, polyphenols, and vitamins also makes seaweeds a valuable functional ingredient for aquafeeds [18,19]. However, it has been reported that macroalgae contain several anti-nutritional factors as well, such as lectins and protease inhibitors that might interfere with digestive processes [8]. In this regard, the present study explores the presence

of antinutritional factors in *Ulva ohnoi* with a potential inhibitory e ffect on fish digestive proteases and provides an initial overview of their mode of action on those proteolytic enzymes.

Related to *Ulva* ANFs, the results of the present study confirmed the existence of compounds able to reduce the digestive proteolytic activity of di fferent marine fish species. This fact agrees with previous studies pointing to the existence of protease inhibitors in some macroalgae species, such as *Ulva rigida*, *Ulva ohnoi, Gracilaria cornea* and *Sargassum* sp. [7,8,17]. In our study, inhibition plots and zymograms illustrate the response of fish proteases after incubation with crude *Ulva ohnoi*. Seabream, Senegalese sole and seabass digestive proteases showed susceptibility to *Ulva* ANFs, although a high concentration of *Ulva* was needed to cause high inhibition (>50% of protease inhibition). According to the equations obtained from the inhibition assays, the amount of *Ulva* required to reach IC50 would represent a dietary inclusion of approximately 40–53% in feeds, which are unusual levels for feed formulation. From a practical point of view, for 40 g fish consuming feeds containing 15% *Ulva* at an intake rate of 3% of its body weight, a 20% reduction in their digestive protease activity would be expected. In general, fish have physiological mechanisms aimed at compensating the e ffects of dietary antinutrients [20], although the influence of these compounds is species-specific [21] and may depend on di fferent factors like fish physiology, macroalgae species, duration of feeding period with seaweed-supplemented diets, and the dietary inclusion level [8]. In this regard, there are numerous studies reporting the utilization of dietary seaweeds without compromising growth performance [19,22]. However, others described some negative e ffects on fish growth, even using seaweed biomass at low dietary inclusion level [8].

The e ffect of *Ulva* inhibitors on fish digestive proteases were also evidenced in zymograms. Protease inhibition caused by CR-*Ulva* on gilthead seabream enzymes could be classified as "unspecific", owing to the fact that all the protease fractions visualized in the gels were a ffected similarly. However, the inhibitory e ffect seemed to be more "specific" in Senegalese sole and seabass, taking into account that the active fractions with molecular masses below 30 kDa were inhibited even with the lowest amount of crude *Ulva*, whereas heavier proteases were inhibited only using the highest concentration assayed. This reduction in protease activity may negatively a ffect feed intake and nutrient digestibility in fish [23]. In this context, the in vitro digestive simulations also confirmed that CR-*Ulva* hampered the hydrolysis of standard (casein) by digestive proteases of three di fferent species of aquaculture fish. Thus, a clear reduction both in CPD values and the amount of amino acids released was evidenced when the concentration of CR-*Ulva* in the in vitro assay was increased. Considered together, both findings clearly indicate lower protein hydrolysis and also reflect a significant reduction in the hydrolytic action of both the digestive endo- and exo-proteases of fish [24].

The deactivation of ANFs is an important issue in raw materials processing [25]. Basically, ANFs can be divided into two groups: (i) heat-labile ANFs, including protease inhibitors, phytates and lectins; and (ii) heat-stable ANFs, represented by saponins, non-starch polysaccharides and some phenolic compounds [2]. Heat treatment is a simple procedure for inactivating ANFs and improving the nutritional value of raw protein feedstu ffs [26]. The results in our study indicate that thermal treatment is e ffective when it comes to inactivating *Ulva ohnoi* ANFs that a ffect fish digestive proteases. The degree of ANFs inactivation depends on factors like temperature, time, particle size, and moisture conditions [2]. In fact, both time and temperature should be controlled carefully in order to minimize losses of nutritional value of a given feed ingredient (for instance, lower availability of amino acids and vitamins, and reduced protein bioaccessibility) as a result of excessive heat denaturation [15,27]. In this regard, the e ffect of the thermal treatment on the capacity of *Ulva* to inhibit trypsin activity indicates that such inhibitors are susceptible to relatively slight thermal treatment. A thermal treatment of 80 ◦C for 15 min reduced the inhibitory capacity by 50%, and above 75% as prolonged times were applied. In agreement, proteinograms (Figure 6, lanes 6 to 8) also confirmed that *Ulva* protease inhibitors are thermolabile, owing to the lack of detection of the 31 kDa proteinaceous complex following the thermal treatment of *Ulva*. These results sugges<sup>t</sup> that temperatures reached during the standard industrial processing, for instance, in the extrusion of feeds, would be enough for minimizing the inhibitory capacity of *Ulva*. Both the preconditioning of the ingredient mixture and the friction forces of the

extrusion process itself, which squeezed through a cylinder by a specially designed volute [26] can increase the temperature above 100 ◦C. Indeed, many researchers have shown that extrusion is an efficient procedure for decreasing the trypsin inhibitory capacity of pulses like soybean without altering the amino acid composition, transforming soybean into a high-quality product [26].

For a better understanding of the potential mode of action of *Ulva* protease inhibitors, a kinetic study was performed in which commercial trypsin and chymotrypsin were exposed to *Ulva* extracts. Plant protease inhibitors are characterized by either reversible or irreversible mechanisms [28]. In the present study, kinetics studies revealed a potential reversible inhibition. According to the Lineweaver–Burk plots, both trypsin and chymotrypsin inactivation occurred by a mixed-type inhibition. This type of inhibition is characterized by the ability to bind not only to free enzymes but also to with the enzyme-substrate complexes [29]. In kinetics terms, mixed type inhibition causes changes that result in a progressive decrease in Vmax when K m increases occurs [30]. This type of inhibition cannot be reversed by increasing substrate concentration, given that the inhibitor cannot be displaced by the substrate. Therefore, the extent of the inhibition depends on the concentration of the inhibitor. In addition, di fferences in K m indicated that *Ulva* protease inhibitors presented higher affinity for trypsin than for chymotrypsin.

Although plenty of literature regarding the mechanisms of plant protease inhibitors is available, there are no studies assessing the mode of action of seaweed protease inhibitors. Protease inhibitors in soybean and other seeds have been studied extensively. They are grouped into the Bowman–Birk and Kunitz families according to primary structure homology, the position of reactive sites, the number or location of disulfide bonds, and their ability to withstand thermal and acid processing [31]. According to the available literature, Kunitz proteinase inhibitors are usually 18–26 kDa proteins [32]. Overall, they are characterized by several Kunitz domains composed of approximately 60 amino acid residues, stabilized by three conserved disulfide bonds. This family mainly inhibits trypsin and weakly inhibits chymotrypsin [33] and is relatively heat- and acid-sensitive [15]. On the other hand, Bowman–Birk proteinase inhibitors (BBIs) are usually 6–9 kDa proteins with a polypeptide chain bridged by seven conserved disulfide bonds; they have independent sites for trypsin and for chymotrypsin, and they display similar inhibitory capacity for both proteases [31,34]. Disulfide bonds are fundamental for maintaining the structural stability of inhibitors [35]. Unfortunately, this work does not provide information on the structure of purified protease inhibitors of *Ulva*, their molecular weight, or their amino acid profile; however, the ability to inhibit mainly trypsin and also chymotrypsin, as well as the heat lability observed, sugges<sup>t</sup> a certain similarity with the Kunitz type inhibitors family, although further studies are needed to ascertain this hypothesis.

#### **4. Materials and Methods**

#### *4.1. Ulva Biomass*

*Ulva ohnoi* biomass was ground, sieved (<100 μm), and kept at −20 ◦C until use. For inhibitory assays, *Ulva* biomass was separated in two di fferent batches (100 g each), the first batch received no thermal treatment (CR-*Ulva*), and the second was heat-treated at 120 ◦C for 20 min (HT-*Ulva*). Aqueous extracts (0.1 g mL−1) were prepared from CR-*Ulva* and HT-*Ulva*, homogenized in distilled water by shaking for 30 min at room temperature, and then for 24 h at 4 ◦C. The mixture was centrifuged for 20 min at 12,000 g and 4 ◦C. Supernatants were stored at 4 ◦C until used in posterior inhibitory assays.

#### *4.2. Fish Enzyme Extracts*

Juvenile specimens of gilthead seabream (*Sparus aurata*), seabass (*Dicentrarchus labrax*), and Senegalese sole (*Solea senegalensis*) were used as model aquaculture fish species. Nine fish of each species were anesthetized and sacrificed by severing their spine according to the requirements of the Council Directive 2010/63/UE. The abdomen was opened and the whole viscera were obtained. Intestines of each species were pooled (three pools including three intestines each, one per fish species), and manually homogenized in distilled water at 4 ◦C to a final concentration of 0.5 g mL−1. Supernatants were obtained after centrifugation (12,000 rpm for 12 min at 4 ◦C) and stored at −20 ◦C until further use. The total soluble protein in the enzyme extracts was determined using bovine serum albumin as standard [36]. The total alkaline protease activity in the enzyme extracts was measured spectrophotometrically following the procedures described by Alarcón et al. [37], using 5 g L−<sup>1</sup> casein in 50 mM Tris–HCl (pH 9.0) as substrate. One unit of total protease activity of activity (UA) was defined as the amount of enzyme that released 1 μg of tyrosine per min in the reaction mixture, considering an extinction coe fficient for tyrosine of 0.008 μg<sup>−</sup><sup>1</sup> mL−<sup>1</sup> cm<sup>−</sup>1, measured spectrophotometrically at 280 nm.

#### *4.3. Testing the Presence of Protease Inhibitors in Ulva*

The inhibitory e ffects of CR-*Ulva* and HT-*Ulva* on the intestinal proteases of gilthead seabream, Senegalese sole and seabass were determined using a modification of the method described by Alarcón et al. [38]. This method is based on the measurement of the residual proteolytic activity after the preincubation of fish extracts with di fferent volumes of CR-*Ulva* and HT-*Ulva* extracts providing a ratio mg *Ulva* per fish protease activity ranged from 0.0 mg *Ulva* UA−<sup>1</sup> to 1.5 mg *Ulva* UA−1. Enzyme inhibition was expressed as a percentage of protease inhibition after comparing with a control assay carried out without any *Ulva* extract. In addition, the amount of *Ulva* requested for 50% protease inhibition (IC50) was estimated.

In order to visualize the e ffect of *Ulva* on active fish intestinal proteases, substrate-SDS-PAGE electrophoresis gels were performed. Intestinal extracts were preincubated for 60 min with di fferent volumes of CR-*Ulva* or HT-*Ulva* extracts. Then, the samples were mixed (1:1) with SDS sample bu ffer (0.125 M Tris HCl, pH 6.8; 4% ( *w*/*v*) SDS; 20% (*v*/*v*) glycerol; 0.04% ( *w*/*v*) bromophenol blue) and SDS-PAGE was performed according to Laemmli, [39] using 11% polyacrylamide gels (100 V per gel, 45 min at 4 ◦C). Zymograms revealing protease active bands were made according to Alarcón et al. [37]. After electrophoresis, gels were washed with distilled water and incubated in 0.75% ( *w*/*v*) casein solution prepared in 50 mM Tris–HCl bu ffer, pH 9.0, for 30 min at 4 ◦C. The gels were then incubated in the same solution for 90 min at 37 ◦C without agitation. Finally, the gels were washed and fixed in 12% TCA for 10 min to stop the reaction prior to staining with Coomassie Brilliant Blue R-250 in a solution of methanol–acetic acid–water for 12 h. Distaining was done using a methanol–acetic acid–water solution. Clear gel zones revealed the presence of active proteases with caseinolytic activity.

#### *4.4. E*ff*ect of Ulva on Fish Digestive Proteases*

The capacity of *Ulva* to inhibit the hydrolysis of casein by fish intestinal proteases was also assessed using an in vitro assay in the presence of di fferent concentrations of crude (CR) and heat-treated (HT) algae biomass providing 0.5, 1.0 and 1.5 mg *Ulva* UA−1.

The in vitro casein hydrolysis was simulated in 10 mL-jacketed reaction vessels connected to a circulating water bath at 37 ◦C, under continuous agitation by a magnetic stirrer. The temperature was selected in order to increase the activity of the enzymes for reducing the time requested for each analysis [40]. An amount of casein, providing 80 mg of crude protein per vessel, was suspended in 50 mM Tris HCl bu ffer pH 9.0. After 15 min stirring, the hydrolysis was started by the addition of the enzymatic extract providing 200 UA of total alkaline proteolytic activity [24]. The alkaline hydrolysis was maintained for 90 min, and samples of the reaction mixture (0, 15, 30, 60 and 90 min) were withdrawn. The products of the hydrolysis were separated by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE), and total amino acids released were also measured at each sampling time, in order to estimate the sequential degradation of casein [41]. All determinations were performed in triplicate. Blank assays with casein but without *Ulva* biomass were carried out for each fish species.

In order to visualize the casein hydrolysis of SDS-PAGE-separated casein fractions, electrophoresis gels were performed. The procedure was carried out as previously described. The rate of hydrolysis was expressed by a numerical value obtained considering both the percentage of reduction in optical density for each protein after the enzymatic hydrolysis and the relative proportion that such protein represented in the total proteins [38]. The value obtained was called the coe fficient of protein degradation (CPD), and it was estimated using the following mathematical expression:

$$CDP = \sum\_{i=1}^{n} \left[ \frac{OD\_i(t=0) - OD\_i(t=90\text{min})}{OD\_i(t=0)} \ge 100 \right] \ge \frac{OD\_i(t=0)}{\sum\_{i=1}^{n} OD\_i(t=0)}$$

where *i* are the proteins identified, *ODi* is the optical density of the proteins, and *t* is the time of reaction.

In addition, the total released amino acids in each sampling time were also quantified at 340 nm in a spectrophotometer (Shimadzu UV-1800, Shimadzu, Kyoto, Japan), using L-leucine as standard [41]. The results were expressed as accumulated values of amino acid released during the enzymatic hydrolysis (g 100 g protein−1).

#### *4.5. Partial Characterization of Ulva Protease Inhibitors*

#### 4.5.1. E ffect of Temperature on Protease Inhibitors

The e ffect of temperature on *Ulva* protease inhibitors was assessed by heating the aqueous *Ulva* extract (0.1 g mL−1) at di fferent temperatures (25, 40, 60, 80, 90, 100 ◦C) during 60 min and then immediately cooled in a water bath. Samples of *Ulva* from each temperature treatment were withdrawn at 5, 15, 30 and 60 min, and then preincubated with a solution of bovine trypsin (1 μg mL−1. T8003 from Sigma Aldrich, SL. Saint Louis, MO, USA) during 60 min at room temperature at a ratio of 500 μg of *Ulva* per μg trypsin. After that, trypsin activity was assayed according to Erlanger et al. [42] using BAPNA (N α-Benzoyl-DL-arginine 4-nitroanilide hydrochloride) as substrate. Enzyme inhibition was expressed as the percentage of trypsin inhibition after comparing with a control assay carried out without *Ulva*. SBTI was used as positive control of the inhibition assay.

In addition, the formation of proteinaceous enzyme-inhibitor complexes was determined by using substrate-SDS-PAGE electrophoresis gels. Samples were prepared by preincubating crude or heat-treated (100 ◦C, 5 min) *Ulva* extracts with a trypsin solution (1 μg mL−1) at a ratio of 500 μg of *Ulva* per μg trypsin for 0, 30 and 60 min at room temperature. Samples were mixed (1:1) with SDS sample bu ffer (0.125 M Tris HCl, pH 6.8; 4% ( *w*/*v*) SDS; 10% (*v*/*v*) β-mercaptoethanol; 20% (*v*/*v*) glycerol; 0.04% ( *w*/*v*) bromophenol blue and SDS-PAGE was performed according to Laemmli [39] using 12% polyacrylamide gels (100 V per gel, 45 min, 4 ◦C). After electrophoresis, gels were washed with distilled water prior to staining with Coomassie Brilliant Blue R-250 in a methanol-acetic acid solution overnight. Finally, distaining was done with a methanol-acetic acid-water solution. In addition, 5 μL of a wide-range molecular weight marker (S-84445 SigmaMarker ™, St. Louis, MO, USA) were included in each gel. The molecular marker consisted of 12 proteins ranging from 6.5 kDa (aprotinin, bovine lung) to 200 kDa (myosin, porcine heart).

#### 4.5.2. Trypsin and Chymotrypsin Inhibition Kinetics

Inhibition kinetics were conducted according to Bijina et al. [30], with minor modifications, using trypsin and chymotrypsin from bovine pancreas (T8003 and C4129 from Sigma Aldrich, SL) and di fferent concentrations of the synthetic substrate. An aliquot of 10 μL of each protease (1 mg mL−1) was pre-incubated with di fferent concentrations of *Ulva* (from 0 to 500 μg *Ulva* per μg trypsin) for 60 min. Later on, trypsin and chymotrypsin activities of the pre-incubated mixtures were assayed using various concentrations of BAPNA (N α-Benzoyl-DL-arginine 4-nitroanilide hydrochloride) (from 0.10 to 0.75 mM) according to Erlanger et al. [42], or SAPNA (N-succinyl-(Ala)2-Pro-Phe-P-nitroanilide) (from 0.05 to 0.5 mM) according to DelMar et al. [43], respectively, in 50mM Tris-HCl, 10mM CaCl2 bu ffer, pH 8.5.

The activity of the enzymatic reaction (v) based on the rate of change in absorbance (405 nm) of the reaction mixture was determined for each substrate concentration [S] assayed. Lineweaver–Burk curves, 1/v versus 1/[S], were plotted and the Michaelis constant (K*m*) and the maximum rate of reaction (Vmax) were calculated for classifying the pattern of inhibition generated by the *Ulva* extract (competitive, uncompetitive or non-competitive).
