The lumpfish used in this study were sampled from salmon net cages in March 2020 from Karanes (Karlsøy, Troms, and Finnmark county, Norway) as described in Ageeva et al. (2021). Whole lumpfish and lumpfish skin was cut into 2 × 2 cm pieces, packed in vacuum bags (Scancell, Kuppenheim, Baden-Wurttemberg, Germany), and frozen at −40°C until further processing.

The proteases used for enzymatic hydrolysis were Corolase 8,000 (batch number R205464ST, 100000 BPU/g, AB Enzymes, Darmstadt, Germany), Corolase 7,089 (batch number F181360ST, 840 Uhb/g, AB Enzymes, Darmstadt, Germany), Food Pro PNL (batch number 4863924615, 1,600 U/g, DANISCO, Copenhagen, Denmark), Flavourzyme (batch number HPN00549, 1000 LAPU/g, Novozymes, Bagsværd, Denmark) and Endocut 01 L (batch number 1043, 180 NU/g, Tailorzyme, Herlev, Denmark). All the enzymes used comply with the recommended purity specifications for food-grade enzymes issued by the joint FAO/WHO Expert Committee on Food Additives and the Food Chemicals Codex.

Peptide standards for molecular weight (MW) distribution were purchased from Sigma-Aldrich: Carbonic Anhydrase from bovine erythrocytes (C7025, 29 kDa), Lysozyme from chicken egg white (L7651, 14.3 kDa), Cytochrome c from the bovine heart (C2037, 12.3 kDa), Aprotinin from bovine lung (A1153, 6.51 kDa), Insulin Chain B oxidized from bovine pancreas (I6383, 3.5 kDa), Renin Substrate Tetradecapeptide porcine (R8129, 1.76 kDa), Angiotensin II human (A9525, 1.05 kDa), Bradykinin Fragment 1–7 (B1651, 0.757 kDa), [D-Ala²]-Leucine enkephalin (E5008, 0.569 kDa), Val-Tyr-Val (V8376, 0.379 kDa), and L-Tryptophane (T0254, 0.204 kDa). All cell lines and microbes used for bioactivity screening were purchased at LGC standards: HepG2 (ATCC HB-8065™), MRC5 (ATCC CCL-171™), Enterococcus faecalis (ATCC 29212), Escherichia coli (ATCC 25922), Pseudomonas aeruginosa (ATCC 27853), Staphylococcus aureus (ATCC 25923), Streptococcus agalactiae (ATCC 12386), Candida albicans (ATCC 90028) and Staphylococcus epidermidis (ATCC 35984).

All chemicals used in this study have been purchased from Sigma Aldrich and VWR and solutions for electrophoresis through Thermo Scientific.

The proximate composition of the lumpfish used in this study has been published previously (Ageeva et al., 2021). The results showed that the lumpfish had high moisture levels (91.5% w/w) and low levels of protein (5.7%), fat (1.3%), and ash (1.6%). For comparison, a previous study by Dave et al. (2019) and (Lin et al. (2020) observed an almost identical proximate composition with 91.08% moisture, 5.02% protein, 1.59% fat, and 1.72% ash. The total amino acid content of whole lumpfish used in this study was 46.67 g/100 g (analyzed on dry weight), and it contained all essential amino acids (Supplementary Table 3).

Lumpfish skin and whole lumpfish were used as the starting point for gelatin extraction. The gelatin samples contained significant components in the form of α-chains and β-chains, as shown in Figure 4. The molecular weights of the α1- and α2-chains were larger than 100 kDa. The β-chain had a molecular weight > 200 kDa. This result suggests that the extraction conditions did not cause any degradation to the α-chains and β-chains. The mass yield of LSG and WLG were 1.61 and 0.63%, respectively. Vate et al. (2023) have previously shown a mass yield of 1.24% of native collagen isolated from lumpfish (gutted with head). The total amino acid content of LSG and WLG indicated higher protein content in the LSG than in the WLG sample (Table 1). LSG also has a higher amount of glycine (Gly), Hyp and proline (Pro) compared to WLG, which is expected since connective tissue (like skin) contains a more significant amount of collagen compared to filets (Liaset and Espe, 2008). Based on the amino acid composition of lumpfish biomass, the recovery of Hyp was 0.17 and 1.71% in extracted WLG and LSG, respectively. Based on the low mass yield and low recovery, it is believed that the process used for gelatin extraction is not satisfactory from an economic point of view. The continued work focused on complete protein extraction from whole lumpfish, using different pretreatments (both chemical and mechanical) and enzymatic hydrolysis.

The hydrolysis trials produced two fractions: fish protein hydrolysate and sediment. There were no visible lipid layers or emulsions in the hydrolysates, indicating that the amount of lipids not removed during the filtration is incorporated either in the fish protein hydrolysates or in the sediment (not analyzed).

Using the different commercial enzymes at a concentration of 0.5%, the mass yield ranged between 5.35 and 6.73% (Table 2). Reducing enzyme concentration led to a correlating decrease in mass yield. In a commercial setting, a reduction in yield and product loss must be evaluated against the costs saved on the reduced use of enzymes.

The lyophilized fish protein hydrolysates had between 4.9–7.4% moisture and 72–81% Kjeldahl protein (Table 3). There were no significant differences in Kjeldahl protein content when comparing the proteases used at 0.5% (w/w). Surprisingly, a reduction in enzyme concentration from 0.5 to 0.1% gave higher Kjeldahl protein for all enzymes tested (FP, C8, and C7). Decreasing the enzyme concentration to 0.05% showed a reduction in Kjeldahl protein content compared to 0.1% enzyme. However, the 10-fold decrease in the enzyme concentration did not significantly impact the Kjeldahl protein measurements. The sum of total amino acids is the most accurate estimate for protein content in a product (Mæhre et al., 2018), and the levels of total amino acids (Table 3) were lower than Kjeldahl protein in the samples. This is expected when using the protein factor 6.25 in Kjeldahl analysis, which was previously shown to be inaccurate for fish fractions. It has been found that using a protein factor of 5.6 is more accurate (Aspevik et al., 2021). When factor 5.6 is applied to the current results (data not shown), the levels are more similar than the sum of total amino acids.

However, the actual total amino acid content is expected to be higher than displayed in Table 3 since some amino acids, such as asparagine and glutamine, are hydrolyzed to their aspartic and glutamic acid forms, respectively. In addition, other amino acids cannot be determined 100% correctly using this method. For example, tryptophan is completely degraded during the reaction, while sulfur-containing amino acids (e.g., cysteine, methionine) cannot be determined with certainty due to partial degradation of the amino acids. Furthermore, amino acids such as tyrosine, serine, and threonine may have lower recovery due to the nature of acid hydrolysis. The hydrolysates produced in this study contained all the essential amino acids. The choice of enzyme did not notably influence the amino acid contents in the different hydrolysates.

SEC is a method to elucidate the relative size of compounds in a complex sample, e.g., a proteinaceous sample after hydrolysis. The analysis can be used to find the hydrolysis process with the most desired peptide size distribution. Most bioactive peptides have molecular mass in the range of 400–2,000 Da (Zaky et al., 2022), and intestinal digestibility is affected by the size of the peptides (Korhonen and Pihlanto, 2006). In this study, SEC was used to evaluate the effect of enzyme choice and concentration on the peptide size distribution of the respective hydrolysates. All hydrolysates produced at different enzyme concentrations (without pretreatment) were analyzed, and the total average MW was calculated (Table 4). At 0.5% enzyme concentration, the average MW of the peptides in the hydrolysates ranged from 961 Da (FP) to 2,763 Da (FL). At 0.1% enzyme concentration, the same trend can be seen. However, at 0.05% enzyme concentration, C8 seems more efficient in peptide degradation than FP, with a total average MW of 2,318 and 2,670, respectively. The effect of enzyme concentration can be measured as the change in total average MW for peptides produced by the same enzyme. A reduction in total average MW can be observed for all tested enzymes with increased enzyme concentration. This can be expected as higher concentrations of enzymes typically increase protein hydrolysis. The same can be observed in the relative size distribution of peptides in each hydrolysate, shown in Table 4, where the peptides have been divided into categories based on size. Enzyme kinetics can be affected by concentrations (Juárez-Enríquez et al., 2022), possibly explaining this phenomenon.

Samples treated chemically and using PEF pretreatment on non-chemically pretreated samples (Tables 5, 6) resulted in a higher overall protein content with relatively higher amounts of the amino acids Gly, Pro, and Hyp, which are characteristic of collagen (Vate et al., 2023). This indicates that these hydrolysates contain higher amounts of collagen peptides. The mass yield, however, is lower in these samples compared to non-chemically pretreated samples, which is expected as the chemicals used to remove compounds such as non-collagen proteins and fat. During PEF pretreatment, the water changes continuously, and the mass yield results indicate that some compounds are removed during this treatment. A test sample from the PEF-processing drip water showed high amounts of protein on a dry weight basis. Drip loss during preprocessing steps can give lower yields but might result in a product with higher collagen purity.

The samples that underwent chemical pretreatment have lower ash values compared to non-treated samples (Table 6). This implies that the chemical pretreatment removes some non-protein compounds, such as mineral residues. The biomass that was not chemically pretreated was not washed with fresh water before processing. Therefore, the remaining minerals from seawater could also be a factor in increasing the ash values.

The five fish protein hydrolysates produced with 0.5% (w/w) enzyme (FP, C7, C8, E1, and FL) were evaluated for bioactivity in a selection of assays (Figure 5). The two human cell lines, HepG2 and MRC-5, were screened mainly to exclude any potential toxic activities of the hydrolysates and to investigate any potential growth-promoting effects. Compared to the HepG2 cells, an increase in cell survival (up to 134%) was observed, indicating growth-promoting effects on the cells. The assays were performed with one biological replicate and three technical replicates, so further studies must be conducted for confirmational data. The same was observed in the non-malignant cell line MRC-5 (up to 171% cell survival). The results strongly indicate that the hydrolysates do not exert any toxic effects on the cells. On the contrary, they are boosting cell growth and survival. Further studies investigating potential growth-promoting effects should be considered for an extended panel of relevant cell lines, including primary muscle cells. An increase in cell proliferation has been observed after treatment with protein hydrolysates in several previous studies against a variety of cell lines (Yang et al., 2019; Chotphruethipong et al., 2021; Jung et al., 2022). The hydrolysates were assayed for antioxidative activities in a FRAP assay, and no activity was observed (Supplementary Table 4).

The hydrolysates produced with 0.5% (w/w) enzyme (FP, C7, C8, E1, and FL) were screened against five bacterial strains, three Gram-positive and two Gram-negative, at 100, 50, and 25 μg/mL in a minimal inhibitory concentration assay. None of the hydrolysates gave any considerable growth inhibition against any tested bacteria (Supplementary Table 5). No growth inhibition was observed against Candida albicans at any assayed concentrations (Supplementary Table 5). In the assay for biofilm inhibition, some effects were observed for the hydrolysates produced using the Corolase enzymes (C7 and C8) (Figure 6). At the highest concentration (100 μg/mL), the C7 hydrolysate resulted in a biofilm formation of 43%. No growth reduction was observed for the bacterium, indicating that the activity was specific for biofilm formation, not targeting bacterial growth.