Abstract

Free aqueous metabolites in meat, fat and egg of a 4-year-old female Siberian sturgeon were analyzed in a metabolomic perspective. A total of 245 metabolites were detected and 91 were quantified. Major common metabolites among meat, fat and egg were lactic acid, creatine and alanine. There were some tissue-specific metabolites but at only minor levels. Carnosine, a dipeptide consisting of β-alanine and histidine and known for its geroprotetive (anti-aging) function, occurred in abundance in meat and fat but was not detected in egg. The same tendency was seen for another major metabolite, betaine (also known as N,N,N-trimethylglycine). These metabolomic profiles well characterized and discriminated meat, fat and egg, which were primarily characterized with creatine, betaine/choline and glutamic acid, respectively. The metabolite profile help developing aquacultural feed for sturgeons as well as utilizing sturgeons for human nutrition.

1. Introduction

Sturgeons are fish species widely distributed in freshwaters and coastal waters in the Northern Hemisphere. Their archaic forms with ganoid scales were called as “living fossils” by Charles Darwin ¹ (p. 107) and place them at the phylogenic base of the ray-finned fishes, being regarded as “frozen in time” ². This is because most ray-finned fishes generally have experienced 3–4 rounds of whole-genome duplications, whereas sturgeons have only 2 rounds, and the ancestral traits of sturgeons are well conserved. In 27 to 28 species of sturgeons, however, diverse and complex chromosome aneuploidy makes genome analyses difficult. In fact, the number of chromosomes in sturgeons varies widely from the lowest (120) in the sterlet sturgeon, Acipenser ruthenus, to the highest (~437) in Siberian sturgeon, A. baerii, and intra-specific variations are wide, too ³. Genomic engineering of sturgeons has thus been hard to challenge.

In addition to the unique biological characteristics, sturgeons have economic importance as fishery resources in North America and Eurasia. In particular, their roe has been traded at high prices as caviar. Their meat was consumed in large quantities as “Albanian beef” in North America in the 18th and 19th centuries and is still consumed mainly in Russia, while it is protected by the Washington Convention because it is in danger of extinction due to overfishing and poaching. Therefore, demands for aquaculture of sturgeons, have been swelling, but their growth and sexual maturation are slow. Quantitative aspects such as artificial hatching and feed improvement have thus been strengthened, and in recent years, genome mining has been attempted, though there are also difficulties due to the genome aneuploidy of sturgeons.

Instead, metabolomics has been increasingly used in fish aquaculture and nutrition studies ^{4, 5, 6, 7} as well as in other taxa. Metabolomics on farmed Chinese perch (Siniperca chuatsi) revealed that farming in recirculating ponds could be more beneficial than non-circulating pond aquaculture systems ⁵. Metabolomic study on female Chinese sturgeon (Acipenser sinensis) cultured in flow-through pools showed that ovarian development is associated with metabolomic changes related to energy dynamics ⁶. This study, conducted in a running-water pond, provides metabolomic outline of aqueous metabolites of a Siberian sturgeon (Acipenser baerii), and offers hint to guesstimate their metabolic pathways contributing to growth promotion and improvement of meat and egg qualities.

2. Materials and Methods

The studied Siberian sturgeon (Acipenser baerii), four-year-old 3.5-kg female, was reared at Hiroshima-tyozame, the sturgeon farm in Hiroshima City, Japan, under the guidelines and regulations relevant to the Sustainable Aquaculture Production Assurance Act (Japan), and was dissected at the farm under the guidelines and regulations relevant to the Act on Welfare and Management of Animals (Japan). It was fed with dry pellets containing fishmeal, flour, soybean oilcake and corn gluten meal. The fish was dissected onsite, and tissue samples (meat, fat and egg) were kept frozen at -20^oC to -70^oC until analysis.

Three sub-samples (n = 3) for each tissue, i.e., a total of nine sub-samples (3×3), were excised on dry ice in the laboratory and shipped with dry ice to Human Metabolome Technologies, Inc. (HMT; Tsuruoka, Japan) for metabolomic analysis of aqueous metabolites. Amounts (mg) of the nine sub-samples were: 36.7, 33.4 and 32.9 for meat; 33.5, 34.2 and 31.4 for fat; and, 35.1, 35.5 and 40.4 for egg, according to the HMT instruction. The following procedures were performed at HMT: each sub-sample was milled in 750 μl of cold 50% (v/v) acetonitrile with 20 μM internal standards, added with 750 μl of 50% (v/v) acetonitrile, and centrifuged at 2300×g for 120 min at 4^oC; Aliquots (400 μl) of the supernatants were transferred into ultrafiltration tubes (Ultra-free MC PLHCC centrifugal filter unit, HMT; cutoff at 5 kD) and centrifuged at 9100×g for 120 min at 4^oC; and, the ultra-filtrates were dried and re-suspended in each 50 μl of Milli-Q water.

Cationic and anionic forms of aqueous low molecular weight (<1000 kD) metabolites were determined by combined capillary electrophoreses and time-of-flight mass spectrometry using the Agilent CE-TOFMS systems (Agilent Technologies, Santa Clara, CA) at a CE voltage of 30 kV and an MS scan range of m/z 50-1000. The peaks having >3 m/z values were extracted using the automatic integration software MasterHands version 2.17.1.11 ⁸. The extracted peaks were analyzed for detailed m/z values (with an allowable error margin of ±10 ppm) and migration times (with an allowable error margin of ±0.5 min) to assign corresponding compounds in the HMT libraries for standard and known-unknown metabolites. Assigned peaks were quantified based on the peak areas compared with the normalized peak areas of internal standards. Quantified peaks were used for principal component analysis (PCA).

Occurrences and distributions of quantified metabolites in tissues were visualized as a chord diagram generated by the Circos algorithm online (https://mkweb.bcgsc.ca/tableviewer/) ⁹. Biomarker metabolites for a tissue were specified by the Linear Discriminant Analysis (LDA) Effect Size algorithm (LEfSe; https://huttenhower.sph.harvard.edu/galaxy/) ¹⁰. Major metabolites were further projected on metabolic pathway maps generated by Visualization and Analysis of Networks containing Experimental Data (VANTED) ¹¹ based on known human metabolic pathways available at Kyoto Encyclopedia of Genes and Genomes (KEGG; https://www.genome.jp/kegg/) ¹².

3. Results and Discussion

Candidate compounds registered in libraries were assigned to 245 peaks (137 cationic and 108 anionic) based on the m/z values and migration times. A total of 101 peaks were commonly seen in the meat, fat and egg samples (Figure 1), in which the most abundant peaks were annotated as lactic acid, creatine and alanine (Table 1 and Table S1). Lactic acid and alanine are chiral and consist of two enantiomers, respectively; these and other enantiomers were not separated by CE-TOFMS, which is a methodological limitation of this study.

Distribution of 91 quantified metabolites (out of 245 detected metabolite) in meat, fat and egg is depicted as a chord diagram (Figure 2); other 154 metabolites were detected at only non-quantifiable levels and not shown in the chord diagram. Greater amounts and wider variety of metabolites were distributed in the order of meat, fat and egg. There were eight metabolites unique to meat in Figure 2. They were 3-hydroxybutyric acid; uridine-5'-triphosphate (UTP); phosphoribosyl pyrophosphate (PRPP); nicotinamide adenine dinucleotide phosphate (NADP+); N,N-dimethylglycine; cytidine-5’-triphosphate (CTD); cis-aconitic acid; and, coenzyme A (CoA). Similarly, six metabolites uniquely detected in fat were: 3-phosphoglyceric acid; phosphoenolpyruvic acid; tyramine; gluconic acid; 2-oxoisovaleric acid; and 2-phosphoglyceric acid. Then, four egg-unique metabolites were deoxythymidine-5’-diphosphate (dTDP), cysteine, thymidine and deoxythymidine-5’-monophosphate (dTMP).

Meat and fat shared 64 peaks, in which representative peaks were annotated as carnosine and betaine (also known as N,N,N-trimethylglycine). Meat and egg shared minor 5 peaks including the ones annotated as thymidine and 2'-deoxythymidine 5'-monophosphate (dTMP). Fat and egg shared 4 minor peaks that occurred only at less than quantifiable levels. On the other hand, 30, 21 and 20 peaks were unshared and seen only in meat, fat and egg, respectively.

PCA score plot resulted in separate clusters of meat, fat and egg of the sturgeon on the two-dimensional plane of PC1 and PC2 coordinates, which accounted for 49.6% and 28.1%, respectively, of the total variation in the dataset (Figure 3; Table S1). The greatest loadings (-0.998 to -0.996 as absolute values) to PC1 scores were of the peaks annotated as methionine sulfoxide, isoleucine, threonine and leucine; and the greatest loadings (0.986 to 0.978) to PC2 scores were of N-acetyl-galactosamine, theobromine, guanine and trans-glutaconic acid. The rest of the total variation, 22.3%, was explained collectively by PC3 to PC8 that accounted for 6.2% to 1.4% sequentially.

Separation of the PCA-derived clusters were also contributed by such metabolites as indicated by the Linear Discriminant Analysis (LDA) Effect Size algorithm (LEfSe; Figure 4). Meat was primarily discriminated by the relative abundances of representative metabolites such as creatine, β-alanine, glycerol-3-phosphate, fructose-1,6-diphosphate, succinic acid and glucose-1-phosphate. Similarly, fat was primarily discriminated by relative abundances of betaine, choline, betaine aldehyde, guanine, hydroxyproline, hypoxanthine and spermidine, though some were minor. Then, egg was discriminated by the relative abundances of glutamic acid, glycine, putrescine, serine, valine and some minor metabolites.

Only free aqueous amino acids were determined by CE-TOFMS in this study. Their enantiomeric properties were not determined but reasonably presumed to be mostly L-forms. The 20 standard proteinogenic amino acids were found, though cysteine was detected only in egg at an unquantifiable level (Table S1). The 21st and 22nd proteinogenic amino acids, L-selenocystein and L-pyrrolysine, were not detected. In contrast, non-proteinogenic amino acids such as creatine, β-alanine and betaine were found relatively abundantly in meat and fat of Siberian sturgeon. Other non-proteinogenic amino acids such as hydroxyproline, γ-aminobutyric acid (GABA), ornithine, citrulline and sarcosine were also found but only at low levels.

Carnosine was the most abundant dipeptide and composed of β-alanine and histidine that were relatively abundant accordingly (Table 1). Carnosine is known for its antioxidant and geroprotective (anti-aging) properties and thus may have relevant biological functions in sturgeon. Another dipeptide, homocarnosine composed of GABA and histidine, was also detected but only at an unquantifiable level (Table S1). Putrescine and creatinine that are breakdown derivatives of amino acids were also detected at the levels higher than those of the proteinogenic amino acids, asparagine and cysteine.

Profiles of the free aqueous amino acids were compared with those of total amino acids (including the ones hydrolyzed from proteins) reported in previous studies (Table 2). Briefly, the proportion of alanine in free aqueous amino acids in this study was higher than that in total amino acids reported in previous studies where glutamine/glutamic acid and aspartic acid are relatively abundant. Ratios of alanine to glutamine/glutamic acid are 0.66–1.6 in this study and 0.03–0.46 in previous studies; and, the ratios of alanine to aspartic acid are 3.0–9.3 in this study and 0.06–0.78 in previous studies. Free aqueous amino acid pools are thus characterized with the relative abundance of alanine.

Representative metabolites such as lactic acid and alanine were projected on the KEGG human metabolic pathways as a model vertebrate metabolic map (Figure 5). High concentrations of lactic acid and alanine in meat (muscle) may be associated with the Cori cycle and the Cahill cycle (alanine-glucose cycle) via which glycolysis (in muscle) and gluconeogenesis (in liver), i.e., consumption and regeneration of glucose, occur, respectively. High concentration of alanine is known to inhibit the last step of glycolysis, i.e., the conversion of phosphoenolpyruvic acid (PEP) to pyruvic acid by pyruvate kinase ¹⁸. Particular (though minor) occurrence of PEP in fat may possibly associated with occurrence of alanine that may limit the PEP-to-pyruvate conversion.

With the expect of more insight from coupling of metabolomic profiling and gene expression analysis as done by Zhu et al. ⁶, metabolomic information reported in this study affords nutritional implications in developing aquacultural feed for sturgeons as well as in utilizing sturgeons from newfangled perspectives for human nutrition.

4. Conclusions

This study produced the first metabolomic profile of Siberian sturgeon. Major common metabolites among meat, fat and egg tissues were lactic acid, creatine and alanine. Predicted pathways based on the abundance of lactic acid and alanine are Cori and Cahill cycles involved in consumption and regeneration of glucose, respectively. The prediction is, however, based only on a single individual (though nine sub-samples were analyzed) and will be deepened with more specimens, as well as with liposoluble metabolites such as fatty acids in addition to aqueous ones, in succeeding studies.

Acknowledgements

Part of this study was supported by Hokkaido Sugar Co., Ltd. (Sapporo, Japan) and JSPS KAKENHI Grant Number 21K05782. Mr. Kazuyoshi Fujimoto of the surgeon farm Hiroshima-tyozame and Ms. Akiko Koi kindly provided technical advice and assistance.

Statement of Competing Interests

The author declares that there is no conflict of interest.

List of Abbreviations

AA, amino acid

CTD, cytidine 5’-triphosphate

dTDP, deoxythymidine-5’-diphosphate

dTMP, 2'-deoxythymidine 5'-monophosphate

HMT, Human Metabolome Technologies, Inc.

KEGG, Kyoto Encyclopedia of Genes and Genomes

LDA, Linear Discriminant Analysis

LEfSe, LDA Effect Size algorithm

PEP, phosphoenolpyruvic acid

PRPP, phosphoribosyl pyrophosphate

UTP, uridine 2'-triphosphate

References

Supplementary Materials