The Use of Defatted Tenebrio molitor Larvae Meal as a Main Protein Source Is Supported in European Sea Bass (Dicentrarchus labrax) by Data on Growth Performance, Lipid Metabolism, and Flesh Quality

Objective This study aims to determine the maximal inclusion level of defatted (d-) Tenebrio molitor larvae meal (TM) able to replace dietary fishmeal (FM) without compromising growth performance, general metabolism, and flesh quality traits in European sea bass, and to evaluate the major underlying physiological mechanisms. Materials and Methods Fish (55 ± 2 g) were fed with diets containing increasing levels of dTM: 0, 40, 80 and 100% (CTRL, TM40, TM80, and TM100, respectively) to replace FM. After 10 weeks of feeding, the growth performance, nutrient and energy balance, intestinal integrity, plasma metabolites and the expression of genes related to growth and nutrient metabolism, in liver and muscle were determined. The fatty acids (FA) profile, textural properties and color were also evaluated in muscle. Results Protein and lipids digestibility remained unaltered up to 80% dTM inclusion. Growth performance parameters were similar among dietary treatments. The dTM inclusion increased the hepatosomatic index in fish fed TM100. Muscle eicosapentaenoic acid, docosahexaenoic acid and n-3 long-chain polyunsaturated FA levels were maintained up to 80% dTM inclusion, but total cholesterol and non-esterified FA increased with dietary dTM inclusion. In liver, the expression of elongation of very long-chain FA protein 6 (elovl6) and FA desaturase 2 (fads2) did not change in fish fed TM40 and TM80, but elovl6 decreased whilst fads2 increased in fish fed TM100 when compared to those fed CTRL. The expression of cholesterol 7 alpha-monooxygenase (cyp7a1) decreased with dietary dTM inclusion. In muscle, the expression of myoblast determination protein-2 (myod2) decreased in fish fed TM80 and TM100. Conclusion It is feasible to substitute dietary FM by dTM up to 80% in European sea bass without detrimental effects on nutrient digestibility, growth performance and associated genetic pathways, whilst assuring fillet nutritional value for human consumption.


INTRODUCTION
The use of insect protein as sustainable alternative to animal and plant protein sources has been encouraged for direct human consumption and incorporation into animal feeds (Nogales-Mérida et al., 2018;Salter, 2019). Insects have several advantages when compared with the conventional protein sources, as they grow fast and reproduce easily, have low feed conversion ratio (FCR) and small need of arable land and water (Gasco et al., 2020). Additionally, insects are rich protein sources, with a wellbalanced essential amino acid (EAA) profile; being also rich sources of fat and some vitamins and minerals (Koutsos et al., 2019). Insects can also be valuable sources of healthy compounds, such as chitin and antioxidant and antimicrobial peptides. Recent studies showed that insect-based diets modulate fish microbiota and improve the immune system, which may reduce the use of antibiotics in aquaculture (Bruni et al., 2018: Henry et al., 2018Antonopoulou et al., 2019;Rimoldi et al., 2019;Stenberg et al., 2019;Li et al., 2021). In 2017, European Union (EU) authorized the use of insect meal (IM) from seven species, including yellow mealworm (Tenebrio molitor), in aquafeeds (European Commission, 2017). Basto et al. (2020) have recently demonstrated that defatted (d-) T. molitor larvae meal (TM) is a highly digestible protein source able to meet European sea bass (Dicentrarchus labrax) EAA requirements (Basto et al., 2020). The potential of TM to partially replace fishmeal (FM) in aquafeeds has been previously assessed in various marine fish species, such as European sea bass or sparids (Gasco et al., 2016;Piccolo et al., 2017;Henry et al., 2018;Antonopoulou et al., 2019;Iaconisi et al., 2019;Ido et al., 2019), whilst total replacement was only evaluated in red sea bream (Pargus major) (Ido et al., 2019). Results among authors are controversial, and the maximal replacement level of FM by TM are extremely variable and never exceeding 71% in diets for European sea bass (Gasco et al., 2016).
The main nutritional limitation of including insects in aquafeeds is their low content of long-chain polyunsaturated fatty acids (LC-PUFA), such as the eicosapentaenoic (EPA; 20:5n−3) and docosahexaenoic (DHA; 22:6n−3) acids. Fish muscle is the main dietary source of these two n-3 LC-PUFA for humans, which are associated with beneficial health effects (Lands, 2014). Thus, IM may compromise not only flesh quality, but also growth performance when included at high levels in diets for marine fish species. Fish oil (FO) and FM are the main dietary sources of n-3 LC-PUFA, such as EPA and DHA. Thus, high FM replacement levels, or the concomitant replacement of FM and FO, have to be carefully addressed to not compromise the recommended dietary levels of n-3 LC-PUFA for marine fish species (approximately 0.7% on dry matter (DM) basis) (Skalli and Robin, 2004). In blackspot sea bream (Pagellus bogaraveo), full-fat TM did not affect fillet EPA or DHA content, when included at 21% of feed, but decreased the EPA content when included at 40% . This suggest that dietary IM may induce changes in lipid metabolism, depending on its dietary inclusion level. Besides, IM also has a limited content of phosphorus (P), which may affect not only lipid metabolism, but also fish growth performance (Sugiura et al., 2004;Prabhu et al., 2014).
The present study aimed to explore the impact of partial and total replacement of FM by dTM in a comprehensive approach focusing not only on growth performance and nutrient utilization, but also on the underlying mechanisms involved in nutrient metabolism, namely lipid metabolism, in European sea bass. This is one of the most important fish species in Mediterranean aquaculture, where production surpassed 194,000 tons in 2018 (Eurostat, 2020). Muscle quality traits and fillet nutritional value for human consumption were also evaluated.

Diets
Extruded practical diets were formulated and produced by SPAROS Lda. (Portugal) to meet all the known nutritional requirements of European sea bass (NRC, 2011). All diets contained FO (12-13% inclusion level) as the main dietary lipid source. In the control diet (CTRL), FM was added at 45% and was progressively replaced in experimental diets (40%, TM40; 80%, TM80; 100%, TM100) by dTM from Entomo Farm (France). The inclusion level of dTM increased up to 60% in TM100, and was due to the concomitant replacement of FM and plant proteins, resulting in increased protein and lipid content ( Table 1). All experimental diets were properly supplemented with DLmethionine and mono-calcium phosphate. For determination of the apparent digestibility coefficients (ADC), 1% chromium oxide (Cr 2 O 3 ) was added, as an inert marker, to each experimental diet. Details of dietary fatty acid (FA) and amino acid (AA) profile are shown in Table 2

Digestibility Trial
Fish (58 ± 1 g) from the initial stock were randomly distributed in 8 tanks of 50 L (18 fish per tank) equipped with feces sedimentation columns (Cho and Slinger, 1979). Fish were fed the experimental diets (with 1% chromium oxide (Cr 2 O 3 ) until apparent satiation, once a day, during 5 days for adaptation to the diets, before the feces collection began. Approximately 30 min after feeding, every tank was carefully cleaned to assure that no remains of uneaten feed were left in the bottom of the tank or in the sedimentation column. Feces were collected from the sedimentation column every morning, before feeding, and then centrifuged at 3000 g, to eliminate water excess, and kept at −20 • C until chemical analysis. Daily collection of the feces was performed for 10 consecutive days. In order to test each diet in quadruplicate and since the RAS was only constituted by 12 tanks equipped with feces sedimentation columns, this procedure was repeated over two consecutive rounds. In each round, each diet was tested in duplicate. In the second round, diets were allocated to different tanks and fed to new groups of 18 fish per tank from the same initial stock and with the same size range (58 ± 1 g) used in the first round.

Growth Trial
Twelve homogeneous groups of 25 fish (55 ± 2 g) from the initial stock were randomly distributed in 160 L tanks. Each diet was distributed to triplicate groups of fish, by automatic feeders until visual satiation, three times daily, for 10 weeks. The amount of feed supplied to each tank was adjusted daily, and when some uneaten feed remained in the bottom of the tank, the total amount of feed distributed each day was reduced by 5%, until no feed losses were recorded. When no feed losses were observed, the amount of feed was maintained for 2 days, and then augmented by 5%.

Fish Sampling
After overnight fasting, ten fish from the initial stock, and five fish from each tank at the end of growth trial were sampled, sacrificed by anesthetic overdose (2-phenoxyethanol, 500 µL/L), and kept at −20 • C until whole-body composition analysis. Also, at the end of the growth trial, five additional fish per tank were slightly anesthetized with 2-Phenoxyethanol (200 µL/L) for blood and tissue sampling. Blood was collected from the caudal vein with heparinized syringes, centrifuged at 10,000 g for 5 min at 4 • C, and the collected plasma was stored at −80 • C until metabolite analysis. Prior to sampling the remaining tissues, fish were sacrificed by a sharp blow on the head. Liver and white skeletal muscle (dorsal right side) portions (100 mg) were then collected, frozen in liquid nitrogen and stored at −80 • C until RNA extraction. Sections of anterior intestine after the pyloric caeca (5 mm thick) were excised, washed, fixed in phosphatebuffered formalin 4% (pH 7) for 24 h, and posteriorly preserved in ethanol 70% until histological analysis. A cross-sectional slab with skin (5 mm thick) was taken just before the dorsal fin, and photographed (with a scale reference) for determination of the cross-sectional area (CSA). A portion from the left part of the fillet (1 × 1 cm) was frozen in isopentane, cooled by dry ice, and stored at −80 • C for later histological analysis. Two other muscle samples, from the left part of the fillet, were taken, frozen in liquid nitrogen and stored at −80 • C for DM, total lipid content and FA profile analyses. Additionally, the right dorsal fillet was collected with skin for instrumental color and texture analysis.

Proximate Composition Analysis
Freeze-dried fish, diets and feces were ground and homogenized prior to proximate composition analysis. Ash, DM, crude protein (N × 6.25), lipids, P and energy were determined according to AOAC methods (AOAC, 2006) as described by Basto et al. (2020). The Cr 2 O 3 content in diets and feces was determined according to Bolin et al. (1952).

Total Lipids and Fatty Acids Analyses
Muscle total lipids were extracted and quantified gravimetrically according to Folch et al. (1957), using dichloromethane instead of chloroform. The muscle FA methyl esters (FAME) were obtained through transesterification of lipid extracts, whilst in diets FAME were obtained by direct transesterification. The identification and quantification of FAME was performed as described by Campos et al. (2017), using the tricosaenoic acid (23:0) as internal standard.

Histological Analysis
Samples of anterior intestine were embedded in paraffin and processed according to standard histological procedures.
Intestinal cross-sections (3 µm) were stained with specific Alcian Blue/PAS staining (pH 2.5) and examined under a light microscope (Olympus BX51, GmbH, Germany) coupled with a camera (Olympus DP50, GmbH, Germany). An imaging software (Olympus cellSens Dimension Desktop) was used to determine the following parameters: intestine CSA (mm 2 ); width of muscularis externa (µm), outer longitudinal and inner circular layers (OLL and ICL, respectively); submucosa and lamina propria width (µm); villus length and width (µm); number of neutral and acid goblet cells (NGC and AGC, respectively) per villus. The OLL, ICL, and submucosa width were measured in eight points of each transverse section analyzed, and the mean was considered. The lamina propria width (measured in the middle of the villus), and villus length (measured from the top to the bottom following their natural curves) and width (measured at the base) were measured in the eight highest villus. Goblet cells were counted on these selected highest villi. All measurements are exemplified in Figure 1.
Frozen cross-sections of muscle (12 µm) were stained with hematoxylin-eosin and examined using the previously mentioned imaging software. The following parameters were calculated: muscle CSA (mm 2 ); total number of fibers per CSA; fiber number per unit of area (density; n • /mm 2 ); and frequency of fibers classes according to their diameter (≤20 µm and >140 µm) as reported in Lopes et al. (2017).

Gene Expression
Total RNA from liver and skeletal muscle was extracted using the MagMAX-96 total RNA isolation kit (Life Technologies, United States) after tissue homogenization in TRI reagent following manufacturers' instructions. The RNA quantity and purity were determined by Nanodrop (Thermo Fisher Scientific, United States) with absorbance ratios (A260/A280) of 1.9-2.1. Reverse transcription (RT) of 500 ng of total RNA was performed with random decamers using the High-Capacity cDNA Archive Kit (Applied Biosystems, United States) following manufacturer's instructions. The RT reactions were incubated for 10 min at 25 • C and 2 h at 37 • C. Negative control reactions were run FIGURE 2 | Daily nitrogen (A), phosphorous (B), and lipids (C) balance of European sea bass fed experimental diets. Bars represent means ± SD; n = 3 (pools of 5 fish/replicate). Labeled means without a common superscript letter differ significantly, p < 0.05. Values are means ± SEMs; growth performance: n = 75 (25 fish/replicate); somatic indexes and whole-body composition: n = 15 (5 fish/replicate). Labeled means without a common superscript letter differ significantly, p < 0.05. DGI, daily growth index; FCR, feed conversion ratio; HSI, hepatosomatic index; K, condition factor; PER, protein efficiency ratio; VFI, voluntary feed intake; VSI, viscerosomatic index; WW, wet weight.  Table 2). Two customized PCR-array layouts were designed for the simultaneous gene expression profiling of 49 genes covering a number of markers of growth-hormone/insulin-growth-factors (GH/IGF) system (10), lipid metabolism (12), energy metabolism (11), muscle cell proliferation and differentiation (12), and protein turnover (4) ( Table 3). The program used for qPCR reactions included an initial denaturation step at 95 • C for 3 min, followed by 40 cycles of denaturation for 15 s at 95 • C and annealing/extension for 60 s at 60 • C. All the pipetting operations were made by means of an EpMotion 5070 Liquid Handling Robot (Eppendorf, Germany) to improve data reproducibility. The efficiency of qPCRs was checked, and the specificity of reactions was verified by analysis of melting and linearity of serial dilutions of RT reactions. Fluorescence data acquired during the extension phase were normalized by the delta-delta CT method (Livak and Schmittgen, 2001), using β-actin as housekeeping gene due to its stability among different experimental conditions (average CT among experimental groups varied less than 0.2).

Calculations
The ADC of the experimental diets were calculated according to Maynard et al. (1979)

Statistical Analysis
Data were tested for normality and homogeneity of variances by Kolmogorov-Smirnov and Levene's tests, respectively, and log-transformed whenever required before submitted to a oneway ANOVA using IBM SPSS Statistics 26.0 (IBM corporation, United States). When one-way ANOVA showed significance (p < 0.05), individual means were compared using Student-Newman-Keuls post-test. Fold-changes of genes were submitted to a student's t-test with significance set at p < 0.05. A bivariate spearmen's rank correlation coefficient (rs) test was applied to all variables. Significant correlations were considered at the bilateral level of 0.05.

Digestibility and Nutrient Balance
Overall, all diets were well digested (DM ADC > 76%; protein ADC ≥ 89%; lipids ADC ≥ 97%; energy ADC > 86%; P ADC > 62%), but differences on nutrient digestibility were found among groups in the digestibility trial (Table 4). Fish fed TM100 had the lowest protein apparent digestibility, and the highest lipid and P digestibility. As a result, N intake and metabolic N losses decreased with total FM replacement, and the opposite pattern was found for N retention efficiency. Fecal N losses increased both in fish fed TM100 and in those fed TM80. Digestible P intake decreased with FM replacement, being significantly lower in fish fed TM80 and TM100 compared to fish fed CTRL. Fecal P losses decreased significantly with the concomitant increase of dietary dTM. Total P losses of fish fed dTM decreased significantly compared to fish fed CTRL, regardless of the dietary dTM level. Conversely, in comparison to fish fed CTRL and other experimental groups, both lipid and energy gain and retention efficiency were consistently higher in fish fed TM100, whilst lower branchial and urinary E losses, total E losses and total heat production were observed. Fecal E losses increased significantly in fish fed TM40 and TM80. The net balance (total gain, fecal and metabolic losses) of N, P, and L is summarized in Figure 2.

Growth Performance and Blood Biochemistry
As shown in Table 5, fish almost tripled their initial weight, and no mortality was registered during the entire growth trial. All groups grew at the same rate and a significant decrease of FCR from 1.2 in fish fed CTRL to 1.0 in those fed TM100 was observed. Final K remained similar among dietary treatments. The wholebody lipids and energy content increased in fish fed TM100 with a concomitant decrease of moisture. This group also showed a statistically significant increase of HSI that was positively correlated with whole-body lipid content (0.59; p = 0.04) and lipid gain (0.59; p = 0.04). The VSI also displayed a tendency to increase in fish fed TM100 fish, being this biometric index negatively correlated with metabolic (−0.64; p = 0.03) and total lipid (−0.64; p = 0.03) losses (Supplementary Table 3). Regarding blood analysis, plasma levels of total cholesterol and NEFA increased with the replacement of FM by dTM, regardless of the inclusion level. Plasma triglycerides were only significantly affected by TM100, resulting in the highest values (Figure 3). Cholesterol and/or triglycerides showed positive correlations with HSI and VSI (Supplementary Table 3). Concerning the fish-in:fish-out ratio (Fi:Fo), the calculated values were reduced by 20% in fish fed TM100 in comparison to those fed CTRL or the two other experimental groups (Figure 4).

Muscle Total Lipids and Fatty Acids Analyses
Despite the increase of lipids in TM100 diet, the muscle total lipids content was similar among groups ( Table 6). The muscle saturated (SFA) and monounsaturated FA (MUFA) profile reflected the dietary FA profile ( Table 2). Up to 80% of FM replacement by dTM, the muscle PUFA profile also reflected the dietary FA profile. On the other hand, and despite the highest percentage of PUFA in TM100 diet, muscle PUFA levels in fish fed TM100 were comparable to those of all other treatments.
FIGURE 4 | Fi:Fo ratio. Bars represent means ± SD; n = 3 (pools of 5 fish/replicate). Labeled means without a common superscript letter differ significantly, p < 0.05. Fi:Fo was calculated according to Tacon and Metian (2008), assuming yield from forage fish of 24% for FM and 4.8% for FO.
The relative percentage of muscle EPA, DHA, and n-3 LC-PUFA of fish fed TM40 and TM80 was similar to those fed CTRL, despite their dietary percentage being decreased with the dTM inclusion. Although fish fed TM100 had the lowest relative percentage of EPA and DHA, when expressed in wet weight (g/100 g WW), both muscle EPA and EPA + DHA final contents were similar among all fish. As observed in the experimental diets, the percentage of muscle n-6 LC-PUFA increased with dTM inclusion and consequently the n-3/n-6 ratio decreased.

Intestine and Muscle Histological Analysis
The morphology of anterior intestine was well-preserved in all fish, with no signs of villus fusion or enterocyte vacuolization. Intestine CSA varied between 11 and 15 mm 2 . The OLL and ICL varied between 32 and 41 µm, and 58 and 69 µm, respectively. The lamina propria width varied between 27 and 30 µm. The villus length and width ranged from 1749 to 2001 µm and 161 to 180 µm, respectively. The number of AGC and NGC per villus ranged from 147 to 192, and 70 to 113, respectively. The morphometric parameters analyzed in the anterior intestine did not vary significantly among dietary treatments, with the exception of submucosa thickness that increased concomitantly with dietary dTM inclusion level: the submucosa thickness increased from 27 µm in fish fed CTRL to 29, 30, and 33 µm in those fed TM40, TM80 and TM100, respectively (Supplementary Table 4). White muscle CSA varied between 459 and 496 mm 2 , total number of fibers ranged from 104 to 108 thousand, and fiber density from 217 to 228 (Supplementary Table 5). Fiber diameter varied between 65 and 67 µm, with 5-6% of the fibers having below 20 µm and 2-3% above 140 µm. None of the morphometric parameters analyzed in muscle had significant differences among dietary treatments.

Muscle Instrumental Color and Texture Analysis
Texture and color parameters were similar among dietary treatments (Supplementary Table 6). Hardness ranged from 0.9 to 1 N, adhesiveness from −0.01 to −0.004 J, springiness from 1.2 to 1.6, cohesiveness from 0.4 to 0.5, chewiness from 0.5 to 0.8 J, and resilience was 0.4 irrespectively of the dietary treatment. In the muscle, L * varied from 41 to 42, a * from −0.5 to −0.2, b * from −0.7 to 0.1, H • from 183 to 250 and C • from 1 to 1.2.

Gene Expression
The relative gene expression profile of key selected markers in liver and muscle is shown in Supplementary Tables 7, 8, with arbitrarily assigned values of 1 for hormone sensitive lipase (hsl) in liver of fish fed CTRL, and calpain 2 (capn2) in muscle of fish fed CTRL. Among all the analyzed genes, up to three markers of hepatic lipid metabolism and four markers of muscle growth were differentially expressed with the FM replacement, and the corresponding log 2 fold-changes (dTM/CTRL) were graphically represented. In liver, the expression of elongation of very long-chain FA protein 6 (elovl6) was significantly downregulated in fish fed TM100, but not in those fed TM40 and TM80 ( Figure 5A). Conversely, FA desaturase 2 (fads2) was markedly up-regulated in fish fed TM100, remaining almost unaltered in those fed TM40 and TM80 ( Figure 5B). A strong nutritional regulation was also found for cholesterol 7-alphamonooxygenase (cyp7a1) with a significant down-regulation in all dTM groups ( Figure 5C). A similar trend was observed for the expression of stearoyl-CoA desaturase 1b (scd1b), although this down-regulation was not statistically significant ( Figure 5D). The expression of lipolytic peroxisome proliferator-activated receptor α (pparα) also evidenced a non-significant downregulation trend for fish fed TM100 ( Figure 5E). The expression of elovl6, cyp7a1 and pparα was negatively correlated with plasma cholesterol levels (−0.64 (p < 0.01), −0.46 (p < 0.01), and −0.37 (p = 0.02), respectively) (Supplementary Table 9). Myoblast determination protein 2 (myod2) was down-regulated in TM80 and TM100 (Figure 6A), whereas a significant down-regulation of muscle atrophy F-box (mafbx/atrogin-1) was only found in fish fed TM100 ( Figure 6B). Contrarily, the expression of myostatin (mstn) ( Figure 6C) and myoblast fusion factor (mymk) (Figure 6D) tended to increase with FM replacement level, but this increase was only statistically significant in TM100.
substitution of FM by dTM can be achieved in European sea bass without impairing growth performance or condition factor and even improving FCR after 10 weeks of feeding. To the best of our knowledge, the maximum replacement of FM by fullfat IM, successfully tested in European sea bass, varied between 36% and 50%, in trials of similar duration (Gasco et al., 2016;Abdel-Tawwab et al., 2020). Skalli and Robin (2004) pointed out that growth impairment at higher FM replacement levels is primarily due to nutrient deficiencies in n-3 LC-PUFA, which are fixed for marine fish close to 0.7% on DM basis (Skalli and Robin, 2004). In fact, a previous study in sea bass with 71% FM replacement by TM did not fulfil this requirement, impairing growth (Gasco et al., 2016), whilst all experimental diets tested in the present study contained 3.8-4.3% DM n-3 LC-PUFA, probably due to the high quality of FO, and did not affect growth performance. In blackspot sea bream and gilthead sea bream (Sparus aurata), it was possible to replace FM by full-fat TM up to 49% and 74%, respectively, without affecting growth performance or FCR, after 18-23 weeks of feeding Piccolo et al., 2017). In red sea bream, the total replacement of FM by dTM was successfully achieved without affecting FCR and even increasing growth after 4 weeks of feeding (Ido et al., 2019). Contrarily, the use of defatted black soldier fly (Hermetia illucens) to replace FM at levels above 18% in turbot (Psetta maxima) and 35% in meagre (Argyrosomus regius), for 7-9 weeks, did not affect FCR, but impaired growth performance (Kroeckel et al., 2012;Guerreiro et al., 2020). Thus, optimal FM substitution levels could vary among fish, insect species, rearing conditions, processing methods of IM, and duration of the experimental trials making comparisons difficult (Osimani et al., 2016;Iaconisi et al., 2019). Therefore, standardization of insect protein production is needed to assure the correct formulation of diets not only for fish, but also for other animal species.
this trade-off was apparently altered in fish fed TM100; these fish conserved a higher N retention compared to control by reducing metabolic N losses. Digestibility disturbances with the use of IM are attributed to the chitin, which has been negatively correlated with protein digestibility in in vitro digestibility studies (Marono et al., 2015). Although chitinolytic activity was reported in some fish species (Gutowska et al., 2004;Krogdahl et al., 2005;Kawashima et al., 2016), there are evidences that high FIGURE 6 | Fold-changes (dTM/CTRL) of differentially expressed genes in muscle tissue: myod2 (A), mstn (B), mafbx/atrogin 1 (C), and mymk (D). The asterisks indicate statistically significant differences (*p < 0.05; t-student) between European sea bass fed dTM and CTRL diets. Values >1 indicate up-regulated genes in dTM fish; values <1 indicate down-regulated genes in dTM fish. n = 9 (3 fish/replicate).
dietary chitin levels have negative impacts on fish growth and nutrient utilization (Alegbeleye et al., 2012;Kroeckel et al., 2012). In the present study, this negative impact seems to have apparently been mitigated by the low chitin content (4.8% DM) in dTM. Moreover, the integrity of anterior intestine was well preserved in all fish, without major morphological changes, with the exception of submucosa thickening in fish fed dTM. Li et al. (2020) have observed a higher degree of submucosa cellularity in the anterior intestine of seawater-phase Atlantic salmon (Salmo salar) fed a H. illucens based diet, but the submucosa of mid and posterior intestine remained unaltered after 8 weeks of feeding such diets to freshwaterphase fish (Li et al., 2019). According to Sitjà-Bobadilla et al. (2005), the increase in thickness of submucosa is usually due to infiltrations of granular eosinophil cells, but this could not be confirmed in the present study. In contrast to the present results, in rainbow trout (Oncorhynchus mykiss), the replacement of FM by 25-50% full-fat or 100% defatted H. illucens reduced villus length in the anterior intestine after 12-14 weeks of feeding (Dumas et al., 2018;Cardinaletti et al., 2019). These controversial results may be associated to different species, feeds and processing methods used for IM production, leading to different nutritional profiles.
It was also observed that P ADC increased from 62% in fish fed CTRL to 72-80% in those fed the dTM diets. This higher P digestibility resulted in a pronounced reduction of total P losses, which suggests that the use of dTM as a FM replacer in European sea bass could help reducing P output into the environment. Likewise, Wang et al. (2017) demonstrated in Nile tilapia that FM replacement by increasing levels of M. domestica from 18% to 100% resulted in lower concentrations of total phosphate in the water. This metabolic feature might be part of the adaptive response of fish facing a reduced dietary P supply. Indeed, P deficiencies are a common nutritional disturbance of FM replacement by alternative proteins in aquafeeds (Sugiura et al., 2004;Prabhu et al., 2014). In the case of IM, this is a relevant issue because the mineral content of this feedstuff is markedly lower than FM or other conventionally FM substitutes, such as plant proteins. Indeed, ash content was reduced herein from 11.7% in CTRL diet to 6.5% in TM100 diet.
With total FM replacement, the increase of dietary lipids and energy content was concomitant with an improved digestibility of lipids that would favor increased whole-body lipid content by the end of the growth trial. This trend was not found in previous European sea bass studies where FM was replaced by IM from 25% to 71% (Gasco et al., 2016;Magalhães et al., 2017;Abdel-Tawwab et al., 2020). Likewise, in rainbow trout, Rema et al. (2019) did not observe any impact of dTM on lipid digestibility and retention. Regardless of this, our data of tissue body fat depots suggest an enhanced flux of lipids toward liver and secondly mesenteric fat when FM is totally replaced by dTM. This assumption is supported by data on muscle lipid content, and circulating lipid levels that were positively correlated with HSI and VSI. This is not surprising since European sea bass is a fish with low to moderate lipid deposition rates in the fillet . Although lipid content of liver has not been assessed, the strong and positive correlation between circulating triglycerides and HSI, and the moderate and positive correlation between final whole-body lipids content, lipid gain and HSI may indicate liver steatosis which is a sign of deficiencies in minerals and n-3 LC-PUFA (Ballester-Lozano et al., 2015). In contrast, enhanced deposition rate in mesenteric fat depots is highly informative of P deficiencies (Sugiura et al., 2004). In any case, both mineral and lipid metabolism seem to be highly sensitive to FM replacement by IM, which might be exacerbated at long-term through the production cycle, especially in the case of lipid-enriched diets (see below).
Liver plays a major role in fish lipid metabolism, and its gene expression profile is greatly influenced by dietary composition and feeding levels (Monroig et al., 2018). In the present study, all diets were formulated with 12-13% of FO and the n-3 LC-PUFA levels (3.8-4.3% DM) were above recommended level for marine fish species (approximately 0.7% DM) (Skalli and Robin, 2004). When revisiting the gene expression profile of gilthead sea bream fed isolipidic diets with a maximal replacement of FO by vegetable oils or semisynthetic diets formulated to be deficient in n-3 LC-PUFA, the most responsive hepatic enzymes of lipid metabolism are scd1 and elovl6, and secondly fads2 . This will contribute to mitigate the signs of deficiencies in n-3 LC-PUFA, though it is very important to limit hepatic lipogenesis to avoid the lipotoxic effects of excessive fat accumulation (Perera et al., 2019). Indeed, scd1 is the rate limiting enzyme in the synthesis of MUFA, especially oleic acid (OA, 18:1n-9) and palmitoleic acid (16:1n-7) from stearoyl-CoA and palmitoyl-CoA, respectively. Likewise, elovl6 is responsible for the elongation of SFA and MUFA of 12, 14 and 16 carbons to form 18-carbon FA (Weiss-Hersh et al., 2020). Therefore, the up-regulation of scd1b and elovl6 enhances the biosynthesis of MUFA, which in turn increases the unsaturation index of FA membrane phospholipids. This is reinforced by the recent observation of Perera et al. (2019) that scd1a is epigenetically regulated in gilthead sea bream by broodstock nutrition with alpha-linolenic acid (ALA, 18:3n-3) enriched diets.
In the present study, no changes in ALA were found among diets, whilst both OA and linoleic acid (LA, 18:2n-6) were 3-7 times higher in TM100 than in the CTRL diet. The observed down-regulation of elovl6 in fish fed TM100, accompanied by a global decreasing trend of scd1b expression with FM replacement, support a reduced de novo lipogenic activity. In this scenario, the simultaneous up-regulated expression of fads2 in fish fed TM100 will serve to further desaturate PUFA, but the capacity of de novo synthesis of n-3 LC-PUFA is limited in marine fish and European sea bass in particular, due to the lack of elovl2 and reduced delta 5 desaturase activity (Monroig et al., 2018). Thus, although all dietary treatments provided European sea bass recommended n-3 LC-PUFA levels, TM100 diet may compromise the muscle n-3 LC-PUFA content at long term.
Cholesterol metabolism is also affected by FM replacement, and the hypocholesterolemic effect of plant proteins as FM replacers has been observed in a wide range of fish species, including gilthead sea bream (Gómez-Requeni et al., 2004;Benedito-Palos et al., 2016), European sea bass (Messina et al., 2013), rainbow trout (Romarheim et al., 2008) and salmon (Hartviksen et al., 2014). Likewise, feeding trials conducted with semisynthetic diets highlighted that FO replacement with vegetal oils acts lowering triglycerides and cholesterol in gilthead sea bream (Ballester-Lozano et al., 2015). In the same study, diets formulated for deficiencies in phospholipids and vitamins also showed a hypocholesterolemic effect, whereas P deficiency acted as a hypercholesterolemic factor. Since this same trend was found herein with FM replacement by dTM, the availability of bioactive P becomes again a limiting factor for the successful replacement of FM, though signs of growth and health impairment were mostly masked at short-term. The precise mechanism remains unknown, but it would be favored by the down-regulated expression of hepatic cyp7a1, the rate limiting enzyme of the bile acid synthesis pathway (Jelinek et al., 1990). This assumption was supported by the negative correlation between circulating cholesterol and the expression level of hepatic cyp7a1, as cyp7a1 deficiency has been reported to lead to hypercholesterolemia (Pullinger et al., 2002;Erickson et al., 2003;Qayyum et al., 2018). The ultimate physiological consequences of these findings remain elusive, but intriguingly the suppression of the hepatic cyp7a1 expression in mice fed cholesterolemic diets mainly involves the activation of inflammatory cytokines (Henkel et al., 2011).
Growth of fish is intrinsically related to protein deposition in muscle, which is regulated by systemic and local signaling pathways involving the GH/IGF system (Mommsen, 2001). In this study, no differences in the expression of gh receptors or igfs transcripts were observed among diets, which is in line with growth performance results. Nonetheless, myod2 plays a pivotal role in myoblast proliferation and differentiation (Valente et al., 2013), and its down-regulated expression in fish fed TM80 and TM100 should be indicative of some muscle growth derangement. Moreover, mstn is a negative regulator of muscle growth through inhibition of proliferation and differentiation of myogenic progenitor cells (Bonnieu et al., 2007), and its up-regulation coupled to the down-regulated expression of myod2 highly supports muscle growth inhibition at the transcriptional level. The extent to which this negative feature is counter-regulated by compensatory growth mechanisms remains uncertain, but the expression of key transcripts of the ubiquitin proteasome pathway, with a main role in the protein turnover of skeletal muscle (Foletta et al., 2011), remained unaltered (muscle RING-finger protein-1; murf1) or were even down-regulated (mafbx/atrogin-1) with the total FM replacement. Furthermore, a recent study in zebrafish demonstrated that mymk plays a pivotal role in myoblast fusion and consequently in muscle growth (Shi et al., 2018), and interestingly the expression of this growthpromoting factor was consistently up-regulated with the total FM replacement. The net result at short-term would be, thereby, the preservation of muscle growth with the use of IM as a main protein source in European sea bass. This was supported by the lack of changes in skeletal muscle cellularity and instrumental texture properties. To our best knowledge, the muscle cellularity of fish fed IM diets was never assessed before, but previous studies in different farmed fish species have evidenced the preservation of textural properties with high levels of FM replacement by IM (Lock et al., 2016;Borgogno et al., 2017;Iaconisi et al., 2017;Wang et al., 2017;Secci et al., 2019;Bruni et al., 2020).
In conclusion, the results of the present study demonstrate that FM replacement by dTM is largely feasible in European sea bass without detrimental effects on growth performance, nutrient utilization, intestinal integrity, and flesh nutritional and textural quality. Indeed, FM replacement by dTM resulted in fair levels of EPA and DHA in muscle of European sea bass (0.38-0.46 g/100 g WW), which are above those recommended by EFSA (2010) for human consumption to decrease the risk of cardiovascular diseases (0.25 g of EPA plus DHA per 100 g of fish). Also, most of the possible negative effects on P and lipid metabolism are largely mitigated at this high replacement level. Nonetheless, further research is needed to fully validate such nutritional approach at farm level throughout the production cycle, giving special attention to the total energy content of the diet since lipid-energized diets can exacerbate the disturbance of lipid metabolism. It is also important to mention that the reduction of P emissions to the environment, coupled with the reduction in the Fi:Fo ratio, prove that dTM is an environmentally sustainable alternative to FM.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by IATS-CSIC and CSIC Review Boards, European (2010/63/EU) Animal Directives, and Spanish Laws (Royal Decree RD53/2013) on the handling of experimental animals.

AUTHOR CONTRIBUTIONS
LV, JP-S, JC-G, and EM conceived and designed the study. AB, BO, LP, TS, MM, and SF conducted research, analyzed data, and performed statistical analysis. AB, JP-S, and LV wrote the article and had primary responsibility for final content. All authors read and approved the final manuscript.

FUNDING
This work was funded by the EU H2020 Research Innovation Program under the TNA Program (project AE090027) at IATS-CSIC Research Infrastructure within AQUAEXCEL 2020 Project (652831). Additional financial support was received by the structured program of R&D&I ATLANTIDA -Platform for the monitoring of the North Atlantic Ocean and tools for the sustainable exploitation of the marine resources (reference NORTE-01-0145-FEDER-000040), supported by the North Portugal Regional Operational Programme (NORTE2020), through the European Regional Development Fund (ERDF) and by ANIMAL4AQUA Project, funded by Portugal 2020, financed by ERFD through COMPETE -POCI-01-0247-FEDER -017610.