Physiological responses of Siberian sturgeon (Acipenser baerii) juveniles fed on full-fat insect-based diet in an aquaponic system

Over the last years, the potential use of Black Soldier Fly meal (BSF) as a new and sustainable aquafeed ingredient has been largely explored in several fish species. However, only fragmentary information is available about the use of BSF meal-based diets in sturgeon nutrition. In consideration of a circular economy concept and a more sustainable aquaculture development, the present research represents the first comprehensive multidisciplinary study on the physiological effects of a BSF diet during sturgeon culture in an aquaponic system. Siberian sturgeon (Acipenser baerii) juveniles were fed over a 60-days feeding trial on a control diet (Hi0) and a diet containing 50% of full-fat BSF meal respect to fish meal (Hi50). Physiological responses of fish were investigated using several analytical approaches, such as gas chromatography-mass spectrometry, histology, Fourier Transformed Infrared Spectroscopy (FTIR), microbiome sequencing and Real-time PCR. While aquaponic systems performed optimally during the trial, Hi50 group fish showed lower diet acceptance that resulted in growth and survival reduction, a decrease in hepatic lipids and glycogen content (FTIR), a higher hepatic hsp70.1 gene expression and a worsening in gut histological morphometric parameters. The low feed acceptance showed by Hi50 group sturgeon highlighted the necessity to improve the palatability of BSF-based diet designed for sturgeon culture.

to animal ones such as poultry by-products 7 . Nowadays, with the goal of a further reduction of aquaculture's environmental footprint, insect species like the Black Soldier Fly (Hermetia illucens; BSF) represent very promising candidates as FM alternatives 13 . The great interest in the BSF meal as aquafeed ingredient is due to their eco-friendly rearing in terms of land use, water consumption, CO 2 emissions and on high feed conversion efficiency (BSF larvae are able to grow on low value organic by-products converting them into valuable biomass) 14,15 . Furthermore, insects like Diptera and Coleoptera are part of the natural diet of Siberian sturgeon 16 .
It is well known that these insects possess bioactive compounds like chitin, which at certain concentrations are able to boost the fish immune system and promote gut microbiota diversification 17,18 . From the nutritional point of view, BSF larvae have a suitable protein content and the amino acid composition is similar to that of FM 19 . However, BSF meal fatty acid profile has also some disadvantages, such as a high content of saturated fatty acids (SFA) and an extremely low content in polyunsaturated (PUFA) ones 20 . PUFA are particularly important for fish since deficiencies in these compounds may cause a general deterioration of fish health, poor growth, low feed efficiency and often high mortality [21][22][23] . Previous studies demonstrated that a proper PUFA dietary content is essential to sustain both larval and adult Siberian sturgeon growth and welfare 24,25 ; these compounds play a pivotal role in sturgeon's fillet and caviar quality 26 . Some recent studies tested different defatted BSF meal inclusion levels in aquafeed formulation for several fish species, but results on fish physiological responses are still controversial [27][28][29][30] . This topic, however, has scarcely been investigated in sturgeon aquaculture and most of the results are limited to zootechnical analyses [31][32][33] . Nowadays, several laboratory approaches (histology, molecular biology, gas chromatography and infra-red spectroscopy) are available to evaluate fish welfare and quality and represent valid tools to assess the inclusion of new ingredients, like insect meal, in aquafeed production 23,[34][35][36] . In addition, the use of full-fat BSF meal is preferable to the highly defatted in order to reduce manufacturing costs 37,38 . This aspect has been recently addressed by Truzzi et al. 39 . These authors developed an enrichment procedure to increase insects' PUFA content that allowed to include up to 50% of BSF prepupae meal compared to FM in zebrafish diet without impairing fish growth and welfare 38 . Because of this positive result, this same enriched full-fat BSF dietary inclusion percentage was chosen for the present study, expecting to obtain more promising results respect to Caimi et al. 32 that evidenced a significant reduction of feed consumption and growth performance in Siberian sturgeon juveniles fed on a diet in which FM was 50% replaced with highly defatted BSF larvae meal.
In the present study, Siberian sturgeon juveniles were fed over a 60-days feeding trial on a control diet (based on FM and FO; Hi0) and a diet containing 50% of enriched BSF meal (according to Truzzi et al.) 39 respect to FM (Hi50). Results obtained on zootechnical performances, fillet fatty acid composition, liver and gut integrity, expression of genes involved in fish growth, stress and immune response and gut microbiome represent the first multidisciplinary investigation on the physiological effects of BSF-based diets in sturgeon juveniles. Furthermore, this is the first feeding trial using insect-based diets performed in an aquaponics system. This green technology combines aquaculture (production of fish) with horticulture (vegetables production) saving energy, water and nutrients 40 , representing an important step for the development of a sustainable aquaculture in a future zerowaste generation 41,42 .
No significant differences were detected between the two experimental groups. Ammonia values were lower than 0.05 mg/L for both Hi0 and Hi50 at each sampling time.
Histology. Histological analyses at t 0 were performed in order to evaluate liver and small intestine histological integrity at the beginning of the experiment. Sturgeons exhibited a homogeneous hepatic parenchyma with hepatocytes characterized by a moderate degree of intra-cytoplasmic lipid deposition ( Supplementary  Fig. S1a,b). The percentage of fat fraction (PFF) in the liver parenchyma did not show significant differences between Hi0 (49.2 ± 7.6%) and Hi50 (51.5 ± 9.4%) groups at t 0 . Histology of the small intestine ( Supplementary  Fig. S1c,d) evidenced a regular morphology of mucosal folds, with finger-shaped folds formed by a mono-stratified epithelial layer of enterocytes intercalated with goblet cells, followed by a thin submucosal layer surrounded by the outer muscular layer.
Representative histological images of small intestine (SI), pyloric caecum (PC) and spiral valve (SV) are shown in Fig. 4. In the morphometric analysis of these gut tracts (Fig. 4m), no significant differences were detected between the groups at t 0 , while mucosal folds atrophy, with a significant reduction of folds length, was observed at t 1 in Hi50 SI (Fig. 5c,d; p < 0.0001), PC (Fig. 4g,h, p < 0.0001) and SV (Fig. 4k,l, p < 0.05) compared to the Hi0 group (Fig. 4a,b,e,f,i,j for SI, PC and SV, respectively). In addition, a significant (p < 0.0001) reduction of supranuclear vacuoles in SI and PC and a significant (p < 0.01) reduction in the relative abundance of goblet cells in SI and SV were observed in Hi50 compared to Hi0. FTIR analysis. The spectral analysis of liver samples collected from Siberian sturgeons at t 1 showed differences in the biochemical composition between Hi0 and Hi50. The IR maps (Fig. 5a), as well as the statistical analysis of specific band area ratios (Fig. 5b), showed a significant decrease of total lipids (LIP maps; LIP/TBM, p < 0.001), fatty acids (FA maps; FA/TBM, p < 0.05) and glycogen (GLY maps; GLY/TBM, p < 0.01) in Hi50 compared to Hi0. An increase in unsaturated lipids (CH maps; CH/TBM, p < 0.05) and proteins (PRT maps; PRT/ TBM, p < 0.01) in Hi50 was also observed.
For the small intestine samples, both the IR maps (LIP, PRT and CARBO maps; Fig. 5c) and the statistical analysis of specific band area ratios (LIP/TBM, PRT/TBM and CARBO/TBM; Fig. 5d) did not show significant modifications between Hi0 and Hi50.  (Fig. 6) of the biological replicates at both sampling times were found to be very similar. The taxonomic analysis showed the dominance (> 58%) of Mycoplasma in all samples analysed, followed by Clostridium, with relative abundances comprised between 22.64% (Hi50, t 1 ) and 28.27% (Hi0, t 1 ). Aeromonadacean bacteria were found in gut samples from both Hi0 and Hi50 at t 0 (about 6%), and exclusively in Hi50 sampled at t 1 with the relative abundance of 2.08%. Bacteria of the genus Deefgea were present in both experimental groups exclusively at t 0 with relative abundance of about 2%. Additional bacteria were detected sporadically in some samples, with a relative abundance < 1%. Lactobacillus, Paracoccus, Propionibacterium and Streptococcus were identified solely at t 1 in both experimental groups, while Listeria was found only in Hi50.

Real-time PCR results.
Real-time PCR analyses were performed on liver samples in order to test the expression of genes involved in fish growth (igf1) and stress response (hsp70.1). Gene expression of tnfa was investigated in intestine samples. As shown in Fig. 7, the expression of the genes analysed did not show significant differences between the experimental groups at t 0 . At t 1 , results evidenced a significant (p < 0.01) downregulation for igf1 (Fig. 7a) and a significant (p > 0.01) upregulation for hsp70.1 (Fig. 7b) in Hi50 compared to Hi0. For tnfa (Fig. 7c), no significant differences in gene expression were detected between the experimental groups.

Discussion
The inclusion of BSF meal in aquafeed, as well as the physiological responses of fish, have been recently investigated in several important commercial species like Atlantic salmon (Salmo salar), rainbow trout (Oncorhinchus mykiss) and European seabass (Dicentrarchus labrax) 28,30,43 .
Information available in this field for the Siberian sturgeon is still fragmentary, in fact completely lacking if related to aquaponic systems. Previous studies, based on a limited number of laboratory approaches, highlighted that a defatted BSF prepupae meal dietary inclusion level higher than 25% impaired fish growth and welfare [31][32][33] . Based on these results and previous studies which demonstrated the possibility to use higher full-fat dietary BSF meal inclusion levels 23,38 after an enrichment procedure of the insect biomass 39 , the present study aimed to: (i) test, for the first time, a 50% enriched full-fat BSF prepupae meal inclusion level in a practical diet for juvenile The results showed that the inclusion of a 50% enriched full-fat BSF prepupae meal negatively affects fish growth and survival. These conclusions are supported by the expression of the growth markers analysed and agree with previous studies reporting that inclusion of BSF meal levels higher than 40% often impairs fish growth and welfare 32,36 . A possible explanation of these differences between the two experimental groups can be related to the FA composition of the diets. Şener et al. 44 reported high EPA and DHA levels in Russian sturgeon fed on diets rich in linoleic acid. In agreement with this study, our FA analyses showed the ability of sturgeon to convert linoleic acid (18:2n6) and α-linolenic acid (18:3n3) to EPA and DHA by desaturation and elongation enzymatic pathways 45,46 . Since biochemical conversions require expenditure of energy by the fish, they can explain the observed growth delay in Hi50 compared to Hi0 47 .
It should be pointed out, however, that a lower diet acceptance was observed in the Hi50 group compared to Hi0 and the laboratory analysis performed suggest that the fish entered a fasting condition. Fasting is normally characterized by a growth reduction due mainly to a decrease in IGFs production by hepatocytes 48 . In turn, these changes induce hepatic lipolysis to provide peripheral tissue with free FA as energy source 48,49 . A similar scenario was observed in the present study: Hi50 growth reduction was coupled with a lower igf1 gene expression  were analysed in liver samples; (c) tnfa was analysed in intestine samples. Fish fed diets including 0 and 50 of BSF meal (Hi0 and Hi50, respectively). Significant differences between Hi0 and Hi50, compared within the same sampling time, are indicated as follows: ns non-significant; *p < 0.05; **p < 0.01; ***p < 0.001 and ****p < 0.0001. Values are shown as mean ± SD (n = 5). www.nature.com/scientificreports/ and a scarce lipid accumulation in the hepatic parenchyma. These results are also supported by PFF calculation and FTIR analysis which showed an overall decrease in both total lipids (LIP/TBM) and fatty acids (FA/TBM) in the Hi50 liver samples compared to Hi0. Furthermore, a severe reduction of hepatic glycogen, which represents the first energy reserve mobilized to face food restrictions 50,51 , was detected by FTIR in Hi50 compared to Hi0. Furné et al. 52 demonstrated that the Adriatic sturgeon (Acipenser naccarii) responded to fasting with a precocious mobilization of hepatic glycogen and a high hepatic lipid-degradation capacity. Accordingly, our results represent a strong evidence that fish entered a fasting period. In addition, since malnutrition or fasting are nowadays considered stressors 22,53,54 , this stressful situation is fully supported by the higher hepatic hsp70.1 gene expression detected in Hi50 compared to Hi0. Histological analyses of intestinal tracts are also useful to provide evidence of fasting status in fish. Fasting is usually associated to a reduction in mucosal fold number and height, reduction in supra-nuclear lipid droplets and reduction in goblet cells numbers 55,56 . Accordingly, our Hi50 group showed atrophy of mucosal folds and a dramatic decrease of enterocyte vacuolization and goblet cells number compared to Hi0. In a study performed by Caimi et al. 33 on Siberian sturgeon over a 118-days period, lower (37.5%) levels of defatted BSF meal dietary inclusion did not show these negative effects on spiral valve and liver histology. On the other hand, in a 60-days feeding trial, a 15% dietary inclusion of full-fat BSF prepupae caused a thinning of the mucosa and a parallel thickening of the muscular layer in the proximal intestine while not affecting villus height 31 . Overall, our analyses of the intestine histology and tnfa gene expression did not reveal any sign of inflammation in either of the experimental groups. Similar results were reported in other studies 30,36,57 that evidenced a positive role of BSF meal dietary inclusion on fish gut welfare. BSF meal contains lauric acid and chitin, that possess anti-inflammatory and immune-boosting properties 58 . BSF meal is also known to rise biodiversity in the fish microbiome, necessary to improve fish health, metabolism, nutrition and immunity 17,59 . Our Hi50 diet contained higher percentages of lauric acid and chitin and supported a higher number of bacterial groups (at genus or family level) in the fish gut compared to Hi0, possibly explaining the absence of intestinal inflammatory events. The dominance of Mycoplasma, followed by Clostridium, was observed in all samples analysed, regardless of the diet. An increased relative abundance of Mycoplasma was detected by Rimoldi et al. 59 in the autochthonous gut microbiota of rainbow trout fed with BSF-based diets. This author attributed the beneficial action on host health to the production of antibacterial compounds, such as lactic and acetic acid.
In conclusion, the present study demonstrated the feasibility of aquaponic systems for sturgeon culture and for testing new aquafeed ingredients like insect meal. However, the general low feed acceptance showed by the Hi50 group fish compared to Hi0 indicates the need of further studies aimed at improving the palatability of BSF-based diets intended for sturgeon culture.

Methods
All methods were carried out in accordance with relevant guidelines and regulations. Insects rearing and fish diet production. For details on insects rearing and fish diet production, see Supplementary Information section. Fish, aquaponic system and experimental design. The 60-days feeding trial was conducted at the aquaponics facility "Cooperativa Agricola Tanto Sole" (Treia, Macerata, Italy). Juvenile Siberian sturgeons, purchased from Azienda Agricola Pisani Dossi s.s., Cisliano, MI, Italy), were acclimated for 1 week in a single 500 L tank equipped with mechanical, biological and UV filtration (Panaque, Viterbo, Italy). Tank temperature was 18 ± 0.5 °C; ammonia (NH 3 ) and nitrite (NO 2 − ) were < 0.05 mg/L and nitrate (NO 3 − ) 10 mg/L, according to sturgeon rearing requirements 60 . At the end of the acclimation period, fish were randomly allocated into six Media Based Aquaponic Systems (80 specimens per tank). Each aquaponic system consisted of a 1.56 m 2 hydroponic unit for plants cultivation and a 600 L fish tank, for a total volume of 720 L of water.

Ethics.
Fish unit. The six systems were maintained at constant temperature (18.0 ± 0.5 °C) by chillers TK500 (Teco, Ravenna, Italy). Evaporated water was replaced on request and the systems were subjected to a natural photoperiod (11L/13D). Water samples were collected weekly in order to test ammonia (NH 3  Hydroponic unit. Each hydroponic unit was filled with expanded clay with biological and mechanical filtration function 40 , necessary to guarantee a physical support for plant growth. Specifically, in each hydroponic unit, 16 lettuce (Lactuca sativa; initial weight: 2.95 ± 0.5 g) and 3 celery (Apium graveolens; initial weight: 20.8 ± 5.0 g) seedlings were planted two days before introduction of the fish (density = 12 plants/m 2 ). Recirculating water flow from the fish tank to the hydroponic unit was regulated by a 1900 L/h pump (Eheim GmbH & Co, Deizisau, Germany) completing 3 water renewals per hour. Specifically, water was pumped from the fish tank to the hydroponic unit, and then returned to the fish unit through a siphon. The siphon was equipped with further synthetic foam for extra mechanical filtration (foam was cleaned once per week). www.nature.com/scientificreports/ Feeding trial. At the beginning of the experiment (t 0 ), the six aquaponic systems were randomly assigned to the experimental groups (Hi0, Hi50) according to an experimental design with triplicate tanks per dietary treatment. Feeding trial duration was 60 days, in which sturgeons almost triplicated their weight and were fed as follows: fish fed on the 0% of BSF meal diet (Hi0 group); fish fed on the diet including 50% of BSF full-fat prepupae meal (Hi50 group). Feed particle were 0.5-1 mm in size. Sturgeons were fed three times a day the experimental diets (3% body weight daily). At the beginning (t 0 ) and at the end of the feeding trial (t 1 ), after a 10-h fasting period, the required fish were sampled, euthanized with a lethal dose of MS222 (0.3 g/L; Merck KGaA, Darmstadt, Germany) and properly stored for further analyses.
Biometry. For growth measurements, 60 sturgeons per dietary group (n = 3) at both t 0 and t 1 were randomly collected from the different tanks. Wet weight was measured with an OHAUS Explorer (OHAUS Europe GmbH, Greifensee, Switzerland) analytical balance (precision 0.1 mg). The specific growth rate (SGR) was calculated as follows: SGR% = [(lnWf -lnWi)/t) × 100, where Wf is the wet weight determined at t 1 , Wi, the wet weight determined at t 0 , and t, the number of days (60). During the trial, dead fish were removed and recorded to estimate the final survival rate.
Fatty acid composition. Lipid content and fatty acid composition of experimental diets (n = 3) and fish fil- Histology. Liver, small intestine and spiral valve from 15 different sturgeons per dietary group (n = 3) were randomly collected at both t 0 and t 1 and processed according to Piccinetti et al. 65 . For details, see Supplementary Information section. In order to evaluate the percentage of fat fraction in the liver (PFF), three sections per fish (15 fish per dietary group; n = 3), at 100 µm intervals, were acquired and analysed by mean of the ImageJ software setting an homogeneous threshold value according to Zarantoniello et al. 38 . Non evaluable areas on sections, such as blood vessels and bile ducts, were not considered. Results were reported as mean ± SD of the area occupied by fat on the total hepatic parenchyma analysed on the section. A semi-quantitative evaluation was performed on small intestine, pyloric caecum and spiral valve morphology based on mucosal folds height, supranuclear vacuolization of enterocytes and abundance of goblet cells as previously described in Urán et al. 66 . Specifically, for the morphometric evaluation of mucosal folds height, ten transversal sections per fish (15 fish per dietary group) of small intestine, pyloric caecum and spiral valve, at 200 μm intervals, were analysed as described in Cardinaletti et al. 28 . All the undamaged and non-oblique folds were measured (at least 150 measurements per fish) using ZEN 2.3 software (Carl Zeiss Microscopy GmbH) and measurements were reported as height mean ± SD (µm) 28 . For the semi-quantitative analysis of supranuclear vacuoles and goblet cells, 3 whole intestine circular transversal sections per fish (15 fish per dietary group), at 200 μm intervals, were analysed. The sections were analysed by experienced staff in two independent blinded evaluations and an arbitrary unit was assigned as described in Panettieri et al. 67 . Scores were assigned as follows: supranuclear vacuoles + = scattered, + + = abundant; goblet cells + = 0/4 per villus, + + > 4 per villus.
FTIR measurements. Samples of liver and small intestine collected at t 1 from 6 different sturgeons per dietary group (n = 3), were quickly dissected and immediately frozen at − 80 °C. Samples were then prepared for infrared spectroscopy (IR) measurements 68 as reported in Supplementary Information section.
Sturgeon gut microbiome. RNA extraction and cDNA synthesis. Gut samples from Hi0 and Hi50 groups were collected at t 0 and t 1 . Specifically, 9 different sturgeons per dietary group (n = 3) were collected and processed as previously described by Zarantoniello et al. 38 . The obtained cell pellets were covered with RNA later Stabilization Solution (Ambion, Foster City, CA, USA) and stored at − 80 °C until the extraction of total microbial RNA performed by Quick-RNA Miniprep kit (Zymo Research, CA, USA). The quantity and purity of the extracted RNA were checked using a Nanodrop ND 1000 (Thermo Fisher Scientific). Moreover, the absence of residual DNA contamination was checked by PCR as described by Garofalo et al. 69 . Each sample RNA (10 μL) was reverse-transcribed in cDNA using oligo (dT) and random hexamer primers from SensiFAST cDNA Synthesis Kit for RT-qPCR (Bioline, London, UK).
16S rRNA gene amplicon target sequencing. The portion of 16S rRNA gene (V3-V4 region) from each sample cDNA was amplified by PCR as previously described by Klindworth et al. 70 . The PCR products were further processed and sequenced by MiSeq Illumina instrument (Illumina, San Diego, California, USA) following the procedure detailed by Osimani et al. 18 . Statistical analysis. All data (except for microbiome) were analysed by t-test, with diet as the explanatory variable and presented as mean ± SD. The statistical software package Prism5 (GraphPad Software) was used. Significant differences between Hi0 and Hi50 were indicated as follows: ns, non-significant; *p < 0.05; **p < 0.01; ***p < 0.001 and ****p < 0.0001. For microbiome bioinformatics analyses, raw reads were first merged with the FLASH software and analysed with the QIIME 1.9.0 software 73 ; the detailed pipeline was described by Ferrocino et al. 74 . OTUs clustering was obtained at 97% of similarity and centroids sequencing were mapped against the Greengenes 16S rRNA gene database. OTU tables generated by QIIME were rarefied at the lowest number of reads and showed the highest reached taxonomic resolution. The vegan package of R was used for the alpha diversity calculation.