Article pubs.acs.org/JAFC
Natural 15N Abundance in Key Amino Acids from Lamb Muscle: Exploring a New Horizon in Diet Authentication and Assessment of Feed Efficiency in Ruminants Gonzalo Cantalapiedra-Hijar,*,†,§ Isabelle Ortigues-Marty,†,§ Anne-Marie Schiphorst,# Richard J. Robins,# Illa Tea,# and Sophie Prache†,§ †
UMR 1213 Herbivores, INRA, F-63122 Saint-Genès Champanelle, France VetAgro Sup, UMR 1213 Herbivores, Clermont Université, B.P. 10448, F-63000 Clermont-Ferrand, France # Elucidation of Biosynthesis by Isotopic Spectrometry Group, CEISAM, UMR6230, CNRS−University of Nantes, B.P. 92208, F-44322 Nantes, France §
ABSTRACT: Natural 15N abundance (δ15N) varies between individual amino acids (AAs). We hypothesized that δ15N of nontransaminating and essential AAs (“source” AAs, such as phenylalanine) present in animal tissues could be used as a marker of dietary origin, whereas δ15N of transaminating AAs (“trophic” AAs, such as glutamic acid) could give more detailed insights into animal feed efficiency. Two diets based on dehydrated Lucerne pellets were tested in growing lambs, which promoted different feed efficiencies. No dietary effects were noted on δ15N of any AAs analyzed in lamb muscle. In addition, δ15N of phenylalanine was unexpectedly similar to that of glutamic acid, suggesting that δ15N of AAs is significantly derived from the metabolism of the rumen microbiota and, thus, are not suited for diet authentication in ruminants. In contrast, the δ15N of transaminating AAs facilitates an improved prediction of animal feed efficiency compared to the classical isotopic bulk N analysis. KEYWORDS: 15N, irm-GC/MS, authentication, feed efficiency, ruminants
■
lower δ15Ndiet value (1.1 vs 2.6‰). This may be explained by recent results from our laboratory showing that variations between animal (ruminant) tissues and diets in terms of isotopic N composition (Δ15Nanimal−diet) are dependent on differences in the efficiency of N utilization.6 Both isotopic approaches are thus interconnected. First, dietary authentication reliability may be impaired by differences in feed (N) efficiency: that is, δ15Nanimal becomes different from δ15Ndiet as feed N efficiency use decreases.7 Second, feed efficiency assessment through Δ15Nanimal−diet relies heavily on an accurate knowledge of diet isotopic composition, which remains unknown in most field cases. Therefore, measurements of bulk N isotopic analysis have limitations for both diet authentication when the isotopic N compositions of potential feed resources are close and feed efficiency prediction when diet composition is not known. Natural 15N abundance differs substantially among individual amino acids8−10 according mainly to their ability to be transaminated and to be synthesized de novo.11 Thus, individual amino acids (AAs) carry different isotopic information according to the metabolic pathway(s) in which they are involved.11 In this regard, the δ15N values of some AAs from animal tissues, called “source” AAs according to Nielsen et al.12 and corresponding mainly to nontransaminating and essential AAs as described in Braun et al.,11 have been shown to be quite similar to those found in the diet because of the small
INTRODUCTION Livestock production systems need to evolve toward improving the efficiency with which ruminants transform feedstock into food for human consumption. This transition needs to take into account the unique potential of ruminants to use resources not edible for humans, such as cellulose-rich feedstuffs.1 Moreover, information on livestock feeding conditions is currently being sought by consumers,2,3 who are demanding a “green image” of animal food products.3 Assessment of feed efficiency, that is, the ability to transform feed into animal products, as well as authentication of feeding practices, is thus a current concern for the livestock industry. Natural 15N abundance (δ15N expressed in ‰; i.e., the 15 N/14N ratio referenced to an international standard) of animal tissue (δ15Nanimal) reflects that of animal diet (δ15Ndiet) plus a small and positive diet−animal isotopic fractionation factor (Δ15Nanimal‑diet = δ15Nanimal − δ15Ndiet), called the trophic effect. In an attempt to authenticate meat produced from lowinput feeding systems, we recently demonstrated that the bulk N isotopic analysis of ruminant meat (δ15Nanimal) made it possible to discriminate between feeding practices in animals receiving or not receiving legume forage-rich diets.4 However, dietary isotopic signatures in animal tissues can be affected and disturbed by changes in the isotopic N fractionation (δ15Nanimal = δ15Ndiet + Δ15Nanimal‑diet), which may make diet authentication difficult in some feeding conditions. For instance, in an attempt to authenticate exclusively pasture-fed lambs from those receiving a concentrate finishing diet, Biondi et al.5 found that δ15N values of lambs were not ranked according to the isotopic composition of their respective diets: higher δ15Nanimal values (6.3 vs 5.7‰) were found in lambs fed the diet with © XXXX American Chemical Society
Received: February 29, 2016 Revised: May 2, 2016 Accepted: May 5, 2016
A
DOI: 10.1021/acs.jafc.6b00967 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
three tin capsules (solids “light” 5 × 9 mm, Thermo Fisher Scientific, www.thermo.com) for isotope analysis. Bulk isotope δ15N (‰) values were measured using an Integra2 continuous flow isotope ratio monitoring mass spectrometer (Sercon Ltd., Cheshire, UK) following total combustion in an elemental analyzer (Sercon Ltd., Cheshire, UK) (irm-EA/MS). Results were expressed using the delta notation according to the equation
magnitude of isotopic fractionation occurring during their metabolism.13 On the other hand, other AAs, called “trophic” AAs12 and representing mainly transaminating AAs,11 become enriched in 15N relative to diet during their deamination and transamination14 and so contribute to increase Δ15Nanimal−diet.9,12 Glutamic acid and phenylalanine are usually considered as representatives of trophic (animal) and source (diet) AA, respectively.12 The differences between the δ15N values of AAs representing these groups have been demonstrated to provide information similar to the Δ15Nanimal−diet,8,13 but is a more precise estimate of the relationship between an organism and its diet in the trophic chain.9 With the aim of establishing a more rigorous method for determining the relationship between the ruminant and its diet in terms of N isotope composition, we have focused attention on the δ15N (‰) values of the AAs in the protein of longissimus thoracis muscle of lambs. We have therefore conducted compound-specific isotopic analysis of the δ15N (‰) values for individual amino acids (AA-CSIA) from muscle taken from lambs after 75 days feeding on defined diets to test the hypotheses that both feed efficiency and dietary authentication reliability can be improved and independently evaluated, when compared to the classical approach of bulk N analyses. We have paid special attention to AAs considered to reflect the dietary isotopic N composition for diet authentication and to the difference between trophic (or transaminating) AAs and source (or nontransaminating and essential) AAs for feed efficiency assessment.
■
⎡⎛ R sample ⎞ ⎤ δ15 N = ⎢⎜ ⎟ − 1⎥ × 1000 ⎢⎣⎝ R standard ⎠ ⎥⎦ where Rsample and Rstandard are the 15N/14N isotope ratio for the sample being analyzed and the internationally defined standard (atmospheric N2, Rstandard = 0.0036765), respectively, and δ the delta notation in parts per 1000 (‰) relative to the standard, defined at δ15N = 0.00‰. Sample Preparation for Isotopic Analysis of Individual Amino Acids. Because AA-CSIA is a long and resource-consuming process, this analysis was limited to eight animals from each group (L and H). Samples (n total =16) were randomly chosen from the 48 animals. To an aliquot of muscle sample of ∼20 mg was added 3.0 mL of 6 M HCl in a 10 mL closed screw-capped Pyrex hydrolysis tube and the protein hydrolyzed for 16 h at 110 °C. After cooling and evaporation to dryness, the sample was stored at −20 °C until required. For gas chromatographic (GC) analysis, free AAs were derivatized as their N-pivaloyl-O-isopropyl esters as described previously.16 Briefly, an aliquot of norleucine (Nleu) solution was added to the sample tube to give a concentration of 2 mg/mL in the final sample. Following thorough drying under a stream of N2 gas at 50 °C, 2.5 mL of a solution of acetyl chloride in isopropanol (1:4 v/v) was added and the tube heated at 100 °C for 60 min. Following cooling, drying with N2 gas, and washing with 3 × 0.5 mL of CH2CH2, 200 μL of dry pyridine followed by 200 μL of pivaloyl chloride was added. The tube was heated at 60 °C for 30 min and cooled, and 2.5 mL of CH2CH2 was added. Impurities were removed by passage through a small column of silica gel 60 (40−63 mm, 250−325 mesh) and, following drying with N2 gas, the dry residue was stored at −20 °C until required. Isotopic Analysis of Individual Amino Acids. The compoundspecific isotopic analysis of individual AAs was carried out as described previously.16 Sample was taken into 200 μL of ethyl acetate and transferred to a sealed 1 mL GC vial. Nitrogen isotope ratios were measured using a Delta V Advantage isotope ratio mass spectrometer coupled to a Trace Ultra gas chromatograph via a GC combustion Interface III (all Thermo Fisher Scientific, Bremen, Germany). The mass spectrometer was calibrated for δ15N (‰) values by reference to pulses of pure N2 gas from a cylinder calibrated by elemental analyzer (irm-EA/MS) against international reference materials IAEA-N1 or IAEA-N2 (IAEA, Vienna, Austria). To relate the data sets one to another, values of δ15NGC‑A for analytes were normalized with a correction factor (F) that relates the measured value (δ15NGC‑N) of the internal reference standard amino acid norleucine (Nleu) in the given sample to that measured by irm-EA/MS (δ15NEA‑N). This correction factor is described by F = δ15NGC‑N − δ15NEA‑N, where δ15NGC‑N is the measured value for Nleu in the sample and δ15NEA‑N is the mean value for nonderivatized Nleu measured by irm-EA/MS. Each amino acid is then corrected by δ15NA = F + δ15NGC‑A. Calculation and Statistical Analysis. The isotopic N fractionation between the animal muscle and its diet (Δ15Nanimal−diet) was calculated as the δ15N (‰) values of muscle minus the δ15N values of the diet, the latter calculated as the average of δ15N values of each ingredient weighted by the percentage of N the ingredient represents in the diet. Feed conversion efficiency (FCE; %) was calculated as the ADG (kg/day) divided by dry matter intake (kg/day) and multiplied by 100. Because the concentration of individual AAs was not determined in this study, δ15N values of groups of AAs were calculated as the sum of individual δ15N values rather than as weighted averages. The δ15N value of transaminating AAs was calculated from alanine (Ala), glutamic acid + glutamine (Glx), leucine + isoleucine (Ilx), serine (Ser), and valine (Val), whereas that of nontransaminating
MATERIALS AND METHODS
The experiment took place at the Herbivore Research Unit of INRA’s Clermont-Ferrand-Theix Research Centre, France. The animals were handled by specialized staff who ensured their welfare in accordance with EU Directive 609/1986. Isotopic analyses were carried out at the CEISAM unit (UMR6230, CNRS-University of Nantes, France). Animals and Samples. Samples and individual animal performance data from a previously published experiment15 were used in the present study. Briefly, 48 weaned male Romane lambs individually penned indoors were fed diets based on dehydrated Lucerne (Medicago sativa L.) pellets (400 g/lamb/day) supplemented with either a low (L; 100 g, n = 24) or a high (H; 400 g n = 24) amount of barley grain for 75 days prior to slaughter at 189 ± 7.3 days. A limited amount of pelleted wheat straw was also offered throughout the experiment to maintain adequate rumen function, and water and salt blocks were available ad libitum. Feeds were offered half in the morning at 9 a.m. and half in the afternoon at 4 p.m., and refusals were weighed, recorded, and discarded. Samples of offered and refused dehydrated Lucerne pellets, barley grain, and pelleted wheat straw were collected twice weekly for estimations of DM. Samples of offered dehydrated Lucerne pellets, barley grain, and pelleted wheat straw were collected twice weekly and pooled for bulk N isotopic analysis. On the basis of diet intake, differences in the proportion of feeds consumed between H and L lambs were on average 48 vs 37% for dehydrated Lucerne pellets, 12 vs 35% for barley grain, and 40 vs 28% for pelleted wheat straw, respectively.18 Lambs were weighed at the beginning of the experiment and then once per week. Live body weight was recorded just before slaughter, and average daily gain (ADG) determined as the slope of the regression of live body weight against time. A sample of the left longissimus thoracis muscle was taken from the last thoracic rib 24 h post mortem, vacuum packed, frozen at −20 °C, freeze-dried, and milled in a CYCLOTEC grinder (1093 Sample Mill, Tecator, Hoganas, Sweden) with an outlet grid of 200 μm. Pooled feed samples were dried at 60 °C for 72 h and then milled with a 200 μm outlet mill to a fine powder. Isotopic Analysis of Bulk Samples. Samples of feed and meat (giving ∼80 μg of N) were weighed with a 10−6 g precision balance (Ohaus Discovery DV215CD, Pine Brook, NJ, USA) into each of B
DOI: 10.1021/acs.jafc.6b00967 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry
Average δ15N values of experimental ingredients were 1.64, −0.57, and −1.99‰ for barley grain, dehydrated Lucerne pellets, and pelleted wheat straw, respectively. The experimental diets both had a negative value for δ15N (−0.55 vs −0.30‰), with values slightly higher (less negative) as the dietary proportion of barley grain increased (−0.55 vs −0.30‰ for the L and H diets, respectively). The bulk δ15N values in muscle were similar regardless of the proportion of barley grain in the diet when using either all available lambs (P = 0.09; Figure 1a; n = 48) or those further subjected to AA-CSIA (P = 0.14; Figure 1b; n = 16). The isotopic N fractionation (Δ15Nanimal−diet) decreased (P < 0.001) as the barley grain supplementation increased in the diet, but this effect was significant only when all animals (n = 48) were taken into account. No differences (P > 0.10) between diets were apparent concerning the δ15N values of any individual proteinderived AA, with only a close to significant dietary effect on Glx (P = 0.13). Although numerically higher values of δ15N in transaminating (+11%) and total (+8%) AAs in lambs fed L versus H diets were observed, none of the studied group of AAs were significantly affected by diets in terms of δ15N: no difference was seen between Glx and Phe (P = 0.38) or between trophic and source AA (P = 0.99) or between transaminating and nontransaminating AAs (P = 0.19). As expected, growing lambs showed higher FCE as the dietary proportion of barley grain increased (Table 2), and this was true whether the whole data set of individual lambs (P < 0.001, n = 48) or those further subjected to AA-CSIA (P = 0.03, n = 16) were considered. The relationship between the Δ15Nanimal−diet and the difference between different groups of AAs in terms of δ15N is shown in Figure 2. Only the difference between transaminating and nontransaminating AA in terms of natural 15N abundance (δ15NTAA−NTAA) was significantly correlated to Δ15Nanimal−diet (r2 = 0.59; P = 0.001; Figure 2c), with no relationship noted with either δ15NGlx−Phe or δ15Ntrophic AA−source AA (P = 0.28 and 0.99, respectively). As expected, a negative and significant (P < 0.001) linear relationship was observed between Δ15Nanimal−diet and FCE for both the whole data set of individual lambs (Figure 3a, r2 = 0.53) and those further subjected to AA−CSIA (Figure 3b, r2 = 0.73; RMSE = 2.13). Likewise, as shown in Figure 4, a negative significant (P < 0.001) linear relationship was obtained when FCE was regressed against either natural 15N abundance of total AAs (δ15Ntotal AA; r2 = 0.79; P < 0.001) or natural 15N abundance of transaminating AAs (δ15Ntransaminating AA; r2 = 0.80; P < 0.001) or the difference between transaminating and nontransaminating AAs in terms of natural 15N abundance (δ15NtransaminatingAA−nontransaminatingAA; r2 = 0.71; P < 0.001). It should be noted that the power to predict FCE was improved with the shift from bulk N (RMSE = 2.13; Figure 3b) to transaminating AAs (RMSE = 1.85; Figure 4) or total AAs (RMSE = 1.91; Figure 4) compound-specific isotopic analysis. When Δ15Nanimal‑diet values were complemented with the isotopic information given by δ15N values of transaminating AAs, the resulting partial least-squares model was better adjusted (r2 = 0.83) to the observed FCE data and the resulting predictive power was higher (RMSE = 1.61; Figure 5) than any other tested regression model. A PLS-DA was finally carried out to evaluate to what extent the combination of δ15N information on bulk N and protein-derived AAs could discriminate lambs fed the two experimental diets (Table 3). When the isotopic N composition of bulk N was complemented by δ15N values from all available AAs, the PLS-DA model improved the proportion of correctly classified meat
AAs was calculated from phenylalanine (Phe) and proline (Pro) according to the classification proposed in Braun et al.13 A similar grouping of AAs regarding their isotopic N composition was carried out according to Nielsen et al.:12 the δ15N value of trophic AAs (reflecting isotopic bulk N values of animal tissues) was calculated as the sum of Ala + Val + Ilx + Pro + Asx + Glx, whereas that of source AAs (reflecting isotopic bulk N values of diets) was calculated as the sum of Ser + Phe. All statistical analyses were performed using XLStat software (v2015.2.02; Addinsoft, New York, NY, USA). One-way analysis of variance (ANOVA) was used for statistical comparison of each variable between both groups (L vs H), whereas linear regressions were conducted to model isotopic natural abundances from FCE values. A partial least-squares (PLS) regression model was proposed to handle the problem of multicollinearity when using biological variables highly correlated to each other (individual AAs). The PLS model included 1 dependent variable (FCE) and 10 independent variables (δ15N values for each amino acid and δ15N values for bulk N). For the determination of the number of components to keep in the PLS model the cross-validation criterion Q2 was considered. All isotopic variables were initially included in the PLS model and then removed when their variable importance of projection (VIP) values were 0.8). When using bulk N data and δ15N values of Glx, the prediction model was acceptable (Q2 = 0.136) but with a modest proportion of correctly classified meat samples (68.8%) and a low proportion of variance explained (r2 = 0.23).
■
DISCUSSION The phenomenon of dietary isotopic signatures in animal tissues, that is, the transfer of the isotopic N composition from dietary components to animal tissues, has served to trace diets in both animals17 and humans,18 provided that their potential food sources have different δ15N values17 and that the isotopic N fractionation (Δ15Nanimal−diet) can be determined.19 However, diet reconstruction or authentication from isotopic N composition of animal tissues can be challenging when potential diets are close in terms of isotopic N composition and/or the Δ15Nanimal−diet is unknown and variable between different feeding practices and individuals.20 The Δ15Nanimal−diet increases approximately 3.0−3.4‰ on average with each trophic level across terrestrial and aquatic systems,21−23 but it may be quite variable, from 1 to 6‰, even D
DOI: 10.1021/acs.jafc.6b00967 J. Agric. Food Chem. XXXX, XXX, XXX−XXX
Article
Journal of Agricultural and Food Chemistry Table 2. Natural 15N Abundance of Diet, Total longissimus thoracis Muscle (Bulk N), and Individual Amino Acids of the Protein of longissimus thoracis Muscle and Feed Conversion Efficiency in Growing Lambs Fed Diets Based on Dehydrated Lucerne Pellets Supplemented with either Low (100 g) or High (400 g) Amounts of Barley Grain barley grain supplementation low diet δ15N, ‰ −0.55 animal δ15N, ‰ n = 48 3.58 n = 16 3.79 Δ15Nanimal−dieta, ‰ n = 48 4.14 n = 16 4.34 individual protein-derived AAs δ15N, ‰ Ala 4.28 Val 4.12 Ile + Leu 4.68 Pro 6.17 Ser 15.5 Asx 7.14 Glx 9.06 Phe 9.12 groups of protein-derived AAs δ15N, ‰ transaminating AAb 44.8 nontransaminating AAb 15.3 transaminating − 29.5 nontransaminating AA 35.5 trophic AAc source AAc 24.7 trophic − source AAc 10.8 Glx-Phe −0.055 total AAd 60.1 feed conversion efficiency, %e n = 48 19.5 n = 16 19.8
SEM
P value
3.30 3.35
0.12 0.20
0.09 0.14
3.60 3.64
0.12 0.20
