Natural Abundance Stable Carbon Isotope Evidence for the Routing and de novo Synthesis of Bone FA and Cholesterol Susan Jima, Stanley H. Ambroseb, and Richard P. Eversheda,* a
Organic Geochemistry Unit, Biogeochemistry Research Centre, School of Chemistry, University of Bristol, Bristol BS8 1TS, United Kingdom, and bDepartment of Anthropology, University of Illinois, Urbana-Champaign, Urbana Illinois 61801
ABSTRACT: This research reported in this paper investigated the relationship between diet and bone FA and cholesterol in rats raised on a variety of isotopically controlled diets comprising 20% C3 or C4 protein (casein) and C3 and/or C4 nonprotein or energy (sucrose, starch, and oil) macronutrients. Compoundspecific stable carbon isotope analysis (δ13C) was performed on the FA (16:0, 18:0, 18:1, and 18:2) and cholesterol isolated from the diet (n = 4) and bone (n = 8) of these animals. The dietary signals reflected by the bone lipids were investigated using linear regression analysis. δ13C values of bone cholesterol and stearic (18:0) acid were shown to reflect whole-diet δ13C values, whereas the δ13C values of bone palmitic (16:0), oleic (18:1), and linoleic (18:2) acids reflected dietary FA δ13C values. Dietary signal differences are a result of the balance between direct incorporation (or routing) and de novo synthesis of each of these bone lipids. Estimates of the degree of routing of these bone lipids gleaned from correlations between ∆13Cdlipid − wdiet (= δ13Cdiet lipid − δ13Cwhole diet) spacings and ∆13Cblipid − wdiet (= δ13Cbone lipid − δ13Cwhole diet) fractionations demonstrated that the extent of routing, where 18:2 > 16:0 > 18:1 > 18:0 > cholesterol, reflected the relative abundances of these lipids in the diet. These findings provide the basis for more accurate insights into diet when the δ13C analysis of bone fatty FA or cholesterol is employed. Paper no. L9117 in Lipids 38, 179–186 (February 2003).
Since the 1970s, stable isotope analysis (13C/12C and 15 14 N/ N) has provided a direct method with which to explore trophic interactions in modern and ancient food webs. The rationale behind using stable isotope analysis for dietary reconstruction is based on two well-established observations: (i) different food groups have characteristically different isotope ratios, and (ii) when these food groups are consumed by an organism, they influence the isotopic composition of its tissues. Hence, measured isotope values of a consumer’s tissues serve as a natural tracer for its dietary intake. The dietary information or dietary signal obtained from the δ13C analysis of different consumer tissues reflects different aspects of the diet (1–5). The relationship between dietary macronutrient components and consumer tissue types is complex and is *To whom correspondence should be addressed at Organic Geochemistry Unit, Biogeochemistry Research Centre, University of Bristol, School of Chemistry, Cantock’s Close, Bristol BS8 1TS, United Kingdom. E-mail:
[email protected] Abbreviations: BSTFA, bis(trimethylsilyl)trifluoroacetamide; GC/C/IRMS, gas chromatography/combustion/isotope ratio mass spectrometry; TLE, total lipid extract; TMS, trimethylsilyl. Copyright © 2003 by AOCS Press
thought to depend on several factors, including the nutritional status and digestive physiology of the animal, the turnover rate of the tissue, and its biosynthetic pathway (6). Insights into the relationship between the isotopic composition of specific dietary macronutrients and body tissues have been gleaned from isotopically controlled animal feeding experiments (7–11). Important findings from these studies include: (i) an enrichment of 0.8 ± 1.1‰ is observed between the carbon isotopic composition of whole animals (nematodes, insects, shrimps, snails) and that of their respective diets (7); (ii) the isotopic relationships among the dietary biochemical components of foodstuffs, namely, δ13Ctotal organic matter > δ13Clipid, δ13Ccarbohydrates > δ13Clipid, and δ13Cprotein > δ13Clipid, is inherited by the tissues of animals raised on them (7); (iii) bone collagen δ13C values are biased toward that of the dietary protein (10,11); and (iv) bone apatite δ13C values reflect that of the whole diet (10,11). On the molecular level, Hare et al. (9) measured the δ13C and δ15N values of individual collagenous amino acids isolated from modern (including laboratory-raised) pigs and archaeological bone using preparative ion-exchange HPLC, followed by off-line combustion and isotope ratio mass spectrometry (IRMS). This study showed that a characteristic pattern existed among the carbon and nitrogen isotope values of the amino acids. Moreover, the comparison of the stable isotope values of essential and nonessential amino acids derived from collagen to those present in the diet of the pigs gave insights into the metabolic pathways that govern these amino acids. The technique of gas chromatography/combustion/isotope ratio mass spectrometry (GC/C/IRMS), originally reported by Matthews and Hayes (12), allows the separation and measurement of the stable isotope ratios of individual compounds in a sample mixture. GC/C/IRMS requires only nanograms of individual compounds to be introduced to achieve a precision of ±0.3‰ for carbon isotope determinations. It is therefore a much more sensitive and less laborious technique to use than preparative HPLC for tracing dietary carbon into consumer tissues at the molecular level. Adopting a molecular approach not only increases the specificity of dietary investigations but also can circumvent many of the problems associated with the effects of contamination encountered in bulk isotope determinations. GC/C/IRMS has allowed dietary insights to be gleaned using individual amino acids (13,14), FA (15–17), and cholesterol (16,18–20; Jim, S., Evershed, R.P., and Ambrose, S.H., unpublished data) as indicators of diet.
179
Lipids, Vol. 38, no. 2 (2003)
180
S. JIM ET AL.
The aim of this paper was to use GC/C/IRMS to assess the relative importance of routing and synthesis de novo for each bone lipid to gain a better understanding of the dietary signal reflected in bone FA and cholesterol δ13C values. To this end, it is necessary to consider the different metabolic pathways that affect their occurrence in bone. Linoleic acid (18:2) is an EFA that cannot be synthesized de novo in higher mammals (21,22). It must therefore be directly incorporated or routed from the diet, and thus bone linoleic acid δ13C values are expected to reflect dietary values. Non-EFA (16:0, 18:0, and 18:1) and cholesterol can be both absorbed directly from the diet and synthesized de novo in the body from acetyl-CoA. Acetyl-CoA is the common metabolite formed from the catabolism of dietary lipids, carbohydrates, and proteins (or from tissue glycogen and fat stores). All the carbon atoms of the common FA, two-thirds of the carbon in carbohydrates, and approximately half of the carbon skeleton of amino acids contribute to the acetyl-CoA pool (23). Hence, δ13C values of synthesized FA and cholesterol are expected to reflect wholediet δ13C values with a bias toward dietary lipid and carbohydrate values. However, the overall dietary signal of bone nonEFA and cholesterol is dependent on the relative importance of the processes of routing vs. de novo synthesis of these lipids. In humans, it has been estimated that the amount of cholesterol synthesized per day (typically 1.0 to 1.5 g) is at least twice that of the daily dietary intake for an average Western diet (24). Approximately half of the dietary cholesterol will be absorbed by the intestine and the other half excreted; thus, dietary cholesterol can be estimated to contribute ca. 20% of total body cholesterol. Dietary FA compositions have been shown to greatly influence the FA compositions in rat bone marrow (25), human serum and plasma lipids (26,27), and human bone and blood phospholipids (28,29), providing semiquantitative evidence for the direct incorporation of dietary FA into consumer tissues. Stable isotopic evidence for the influence of dietary FA on consumer tissues has been reported by Rhee et al. (30) and Stott et al. (16). These two studies showed that the δ13C values of non-EFA in human serum and in pig bone are enriched with respect to their dietary sources. Stott et al. (16) also demonstrated that bone linoleic acid δ13C values, as expected, were highly consistent with dietary values. However, Rhee et al. (30) observed a 3‰ depletion in serum linoleic acid δ13C values with respect to dietary values.
MATERIALS AND METHODS Sample description. Holtzman albino rats were raised on a variety of purified and pelletized diets comprising 20.0% protein, 50.2% sucrose, 15.5% starch, 5.0% oil, 5% fiber, 3.5% minerals, and 1% vitamins. One day after insemination, sperm-positive, 90-d-old female rats were placed on diets that their offspring would consume. Birth occurred 21 d after insemination, and weaning occurred 21 to 23 d later. The sexes were separated prior to sexual maturity. Normal room temperature (20°C) was maintained, and food and water were provided ad libitum. Male and female pairs were sacrificed at 91, 131, and 171 d after birth. Eight rat forelimbs from four distinct diets were sampled for lipid analysis, in addition to the whole diets and dietary oils. The dietary compositions and the δ13C values of individual dietary components comprising the diets are shown in Table 1. Table 2 lists all the animals that were studied. Bulk δ13C analysis of whole diet, dietary macronutrients, bone collagen, and bone apatite. Bone collagen and apatite extraction procedures and δ13C measurements summarized here are described in detail in Ambrose and Norr (10). Lipids were extracted from clean ground bone using petroleum ether. Collagen was extracted by demineralization with 0.1 M HCl, treated with 0.125 M NaOH, solubilized at 95°C, and freezedried. Whole diet, dietary macronutrients, and bone collagen were combusted at >800°C with Cu, CuO, and Ag foil in evacuated sealed quartz tubes. Bone apatite carbonate was prepared by deproteinization with NaOCl, treated with 1 M acetic acid to remove adsorbed carbonate, freeze-dried, and reacted under vacuum with 100% H3PO4 at 25°C. CO2 was cryogenically distilled off-line and analyzed on dual inlet isotope ratio mass spectrometers at the Anthropology Department, University of Illinois (Nuclide 6-60 RMS), or the Illinois State Geological Survey (Finnegan MAT Delta E). Chemicals and precautions. All solvents used were of HPLC grade and purchased from Rathburn Chemicals (Walkerburn, Scotland). The internal standard n-tetratriacontane, cholesterol standard, sodium hydroxide pellets, and derivatizing agent N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) containing 1% vol/vol trimethylchlorosilane were purchased from the Sigma Chemical Company (Dorset, England). All glassware and ceramics employed were washed with Decon 90, dried in an oven, and rinsed with chloroform/
TABLE 1 Rat Dietary Compositions and Their Macronutrient Componentsa,b Components and δ13C values (‰)
Composition
Diet code
Protein
Energy
D2A4 D3G D4H D5I
C3 C3 C4 C4
C3 C4 C4 C3
a
c
Protein 20.0%
Sucrose 50.2%
Starch 15.5%
Oil 5.0%
Milk casein (−24.5) Milk casein (−26.3) Milk casein (−14.6) Milk casein (−14.6)
Beet (−24.2) Cane (−11.0) Cane (−11.4) Beet (−24.2)
Rice (−26.4) Corn (−10.3) Corn (−10.6) Rice (−26.4)
Cottonseed (−27.9) Corn (−14.9) Corn (−14.9) Cottonseed (−27.9)
Reference 9; Ambrose, S.H., unpublished data. One percent of vitamins, 3.5% of minerals, and 5% cellulose (wood and/or corn) were added to each diet. c Energy = ∑ (sucrose, starch, and oil components). b
Lipids, Vol. 38, no. 2 (2003)
ROUTING AND DE NOVO SYNTHESIS OF BONE LIPIDS TABLE 2 Dietary Compositions, Sample Codes, Sex, and Pair of Individual Animalsa Dietary composition Protein C3 C3 C3 C3 C4 C4 C4 C4
Sample
Energyb
Code
Sex
Pair
C3 C3 C4 C4 C4 C4 C3 C3
C3 C3 C3P/C4 C3P/C4 C4 C4 C4P/C3 C4P/C3
M F M F F M F M
2 2 3 3 1 2 2 3
a First, second, and third pair animals were sacrificed at 91, 131, and 171 d old, respectively. P = protein; M = male and F = female. b Energy = ∑ (sucrose, starch, and oil components).
methanol (2:1 vol/vol) prior to use. Disposable rubber gloves were worn throughout the whole experimental procedure, from sample preparation to instrumental analysis, to prevent the contamination of samples with finger lipids. Extraction, filtration, and saponification (neutral and acid) blanks were used to monitor and locate any contamination that might be introduced during the experimental procedure. Preparation of diet and bone samples for lipid extraction. Rat forelimbs were dissected into three different tissue types: bone (ulna and radius), skin, and flesh. Bones were manually cleaned of adhering flesh, cartilage, and tendons by scraping with a scalpel. Half a bone was used for the extraction procedure, typically the upper ulna. Rat bone and diet pellets were ground in a pestle and mortar prior to extraction. Liquid nitrogen was employed to aid the bone crushing process. Ranges of rat sample weights are as follows: 0.03 to 0.13 g of powdered bone and 0.83 to 1.41 g of powdered diet pellets. Extraction of lipids from diet and bone. Samples were transferred into large screw-capped vials and known quantities of n-tetratriacontane (1 mg mL−1 in chloroform) added as an internal standard. The samples were extracted with chloroform/methanol (2:1 vol/vol, 5 to 10 mL) by ultrasonication (3 × 1 h, Decon F5200b), where the supernatant was removed and replaced intermittently. The total lipid extract (TLE) was then concentrated to ca. 5 mL under a gentle stream of nitrogen (5 psi) in an evaporation unit (TurboVap LV; Zymarck Corporation, Hopkinton, MA) with the thermostatic bath set at 40°C. Suspended particulates were removed from the TLE by centrifugation (1800 rpm, 20 min; MSE Mistral 1000) and filtration through a short pipette column packed with activated alumina. Saponification of TLE. Aliquots of TLE were transferred into screw-capped test tubes, blown down to dryness under nitrogen gas, and hydrolyzed with 0.5 M methanolic NaOH (2 mL) at 70°C in a water bath for 1 h. After cooling, the mixture was extracted using hexane (3 × 2 mL), yielding the neutral cholesterol-containing fraction. The mixture was then acidified to pH 3 with 1 M HCl and extracted again using hexane (3 × 2 mL) to yield the FA fraction. Neutral fractions were converted to their trimethylsilyl (TMS) ether derivatives using BSTFA
181
containing 1% vol/vol trimethylchlorosilane. Twenty microliters of BSTFA were added to each sample followed by heating at 70°C (Multiblok Lab-Line) for ca. 1 h. Excess BSTFA was removed under a gentle stream of nitrogen. FA were converted to their methyl ester derivatives. FA fractions were transferred into fresh screw-capped test tubes and blown down to dryness under nitrogen gas. The samples were methylated by adding 100 µL of 14% wt/vol boron trifluoride/methanol complex and heating in a water bath at 70°C for 1 h. After cooling, 2 mL of double-distilled water was added and the FAME extracted using diethyl ether (3 × 2 mL). GC. High-temperature GC analyses of the TLE and neutral fractions were carried out using a Hewlett-Packard (HP) 5890 Series II gas chromatograph fitted with a fused-silica capillary column (15 m × 0.32 mm i.d.) coated with a dimethyl polysiloxane stationary phase (DB-1, J&W Scientific, Folsom, CA; Agilent Technologies, Palo Alto, CA; 0.1 µm film thickness). The temperature of the oven was held isothermally at 50°C (2 min) and then increased to 350°C (20 min) at a rate of 10°C min−1. FAME GC analyses were performed using the chromatograph described above fitted with a fusedsilica capillary column (50 m × 0.32 mm i.d.) coated with a polyethylene glycol stationary phase (CP-WAX 52 CB, Varian; Chrompack, Middleberg, The Netherlands; 0.25 µm film thickness). The temperature of the oven was held isothermally at 40°C (2 min) and then increased to 200°C (10 min) at a rate of 5°C min−1. Hydrogen was used as the carrier gas, and FID was used to monitor the column effluent. Data were acquired and analyzed using HP Chemstation software. GC/MS. GC/MS analyses were performed using a Finnigan 4500 quadrupole mass spectrometer (source temperature, 280°C; electron voltage, 35 eV) interfaced to a Carlo Erba HRGC 5160 Mega series gas chromatograph. The same columns and temperature programs as described above were used for the TLE, neutral, and FAME fractions. Hydrogen was used as the carrier gas. Data were acquired using an INCOS data system and processed using Interactive Chemical Information Software (ICIS) package. GC/C/IRMS. GC/C/IRMS analyses were carried out using a Varian 3500 gas chromatograph coupled to a Finnigan MAT Delta-S isotope ratio mass spectrometer via a Finnigan MAT combustion interface (Pt/CuO) maintained at 850°C. For the neutral fractions, the GC was fitted with a fused-silica capillary column (50 m × 0.32 mm i.d.) coated with a dimethyl polysiloxane stationary phase (CP-SIL 5 CB, 0.25 µm film thickness). The temperature of the oven was held at 50°C (2 min) and then increased to 250°C at a rate of 10°C min−1, then to 300°C (20 min) at 4°C min−1. For the FAME fractions, the same column and temperature program as described above for GC and GC/MS analyses were used. Helium was used as the carrier gas and the mass spectrometer source pressure was maintained at 9 × 10−5 Pa. Data were collected and processed using Finnigan MAT Isobase software. Secondary standards (C19 n-alkane or 18:0 FAME) of known δ13C value were measured every seventh run to monitor any fluctuations in instrumental measurements with time. The precision of triplicate Lipids, Vol. 38, no. 2 (2003)
182
S. JIM ET AL.
GC/C/IRMS analyses carried out on each sample was shown to be ±0.3‰. FA and cholesterol δ13C values were corrected for the addition of derivatizing carbon. RESULTS AND DISCUSSION Diet and bone lipid compositions. Figure 1 presents partial gas chromatograms of the TLE, neutral, and FAME fractions of the four diets and of bone sample C4P/C3 (C4 protein with C3 energy; see Table 2 for sample codes), female, third pair. Dietary lipid compositions reflect that of the oil (cottonseed or corn oil) and casein constituting the diets. The cottonseed and corn oil exhibited very similar lipid distributions characterized by the presence of C32 and C36 DAG and C48 to C54 TAG. Milk fat displays a TAG distribution in the range of C28 to C54 but is characterized by a greater abundance of the C36 to C44 TAG (31). Differences in dietary lipid distributions can be explained by the differential lipid extraction of the C3 and
FIG. 1. Partial gas chromatograms of the total lipid extracts (a, d, g, j, and m), saponified trimethylsilylated neutral (b, e, h, k, and n), and FAME fractions (c, f, i, l, and o) of diets D2A4 (a, b, and c), D3G (d, e, and f), D4H (g, h, and i), D5I (j, k, and l) and bone bC4P/C3 (m, n, and o). 1 = FFA; 2 = sugar; 3 = cholesterol; 4 = β-sitosterol; 5 = internal standard; 6 = C32 and C36 DAG; 7a = C30 to C44 TAG; 7b = C46 to C54 TAG; 8 = C10:0 FAME; 9 = C12:0 FAME; 10 = C14:0 FAME; 11 = C16:0 FAME; 12 = C16:1 FAME; 13 = C18:0 FAME; 14 = C18:1 FAME; and 15 = C18:2 FAME.
Lipids, Vol. 38, no. 2 (2003)
C4 caseins during processing, prior to the formulation of the diets. The C4 casein retained a significant proportion of its lipids and this resulted in the presence of C30 to C44 TAG, and 10:0 and 12:0 FA in the TLE and FAME, respectively, of diets D4H (Figs. 1g and 1i) and D5I (Figs. 1j and 1l). Bone TLE (Fig. 1m) were characterized by the presence of FFA, cholesterol, and C48 to C54 TAG. Trends in the δ13C values of diet and bone components. Figure 2a compares whole diet and dietary lipid with bone lipid, collagen, and apatite δ13C values for the C3 and C4 animals. All bone lipid δ13C values are depleted with respect to collagen or apatite values, and this finding is consistent with the relative depletion in 13C that occurs during lipid biosynthesis with respect to other biochemical pathways (32,33). The majority of bone lipid δ13C values are highly consistent with dietary lipid values. Small, positive diet-to-bone fractionations (∆13Cblipid − dlipid = δ13Cbone lipid − δ13Cdiet lipid) are observed for 16:0, 18:1, 18:2, and cholesterol, where mean ∆13Cblipid − dlipid fractionations are +0.9, +0.7, +1.5, and −0.5‰, respectively. For 18:0, a larger ∆13Cblipid − dlipid fractionation of +3.6‰ toward whole-diet δ13C values is observed, which suggests that a greater degree of de novo synthesis may have occurred with 18:0 when compared to the other non-EFA or cholesterol. However, a true assessment of
FIG. 2. Whole diet, diet and bone lipid, collagen, and apatite δ13C values for: (a) C3 and C4 animals, and (b) C3P/C4 and C4P/C3 animals. Each bone data point represents the mean value for the analyses of two animals. WDIET = whole diet; CHOL = cholesterol; COLL = collagen; APAT = apatite; d = diet and b = bone.
ROUTING AND DE NOVO SYNTHESIS OF BONE LIPIDS
the extent of routing or de novo synthesis can only be gleaned when the δ13C values from the C3P/C4 and C4P/C3 animals are also taken into consideration. The small but significant ∆13Cblipid − dlipid fractionation of +1.5‰ shown for the EFA 18:2 must be caused by isotopic fractionation occurring during assimilation, transport, or catabolism. Figure 2b presents the δ13C values for the C3P/C4 and C4P/C3 animals. Here, bone lipid δ13C values are depleted with respect to collagen or apatite values for the C4P/C3 animals only; for the C3P/C4 animals, bone collagen displayed a comparable δ13C value to the bone FA. The reasons for this will become apparent when we consider the diet-to-bone lipid isotopic relationships in more detail. For both diets and in accordance with the C3 and C4 animals, small although not necessarily positive ∆13Cblipid − dlipid fractionations are seen for 16:0, 18:1, 18:2 where absolute mean values are 0.1, 1.2, and 1.9‰, respectively. In contrast to the C3 and C4 animals, a small absolute mean ∆13Cblipid − dlipid fractionation of 1.1‰ is observed for 18:0, and a large absolute mean ∆13Cblipid − dlipid fractionation of 7.9‰ is observed for cholesterol. When these findings are considered alongside those shown for the C3 and C4 animals, a clearer assessment of the relative importance of routing and de novo synthesis for each of the bone lipids can be gained. Comparable ∆13Cblipid − dlipid fractionations are shown for 18:2 in all four diets, demonstrating that bone 18:2 δ13C values are independent of macronutrient isotopic differences. This finding is expected for linoleic acid, which must be derived solely from the diet. Comparable ∆13Cblipid − dlipid fractionations are also shown for 16:0 and 18:1, suggesting that the majority of these nonessential bone lipids are also directly incorporated from the diet. Conversely, ∆13Cblipid −dlipid fractionations for 18:0 and cholesterol have been shown to be influenced by macronutrient isotopic differences, suggesting that a major proportion of these lipids are derived from de novo synthesis. For all four diets, absolute mean ∆13Cblipid − dlipid fractionations for 16:0, 18:0, 18:1, 18:2, and cholesterol are 0.5, 2.3, 0.9, 1.7, and 4.2‰, respectively. If the magnitudes of these spacings are considered to be a measure of the extent of routing of the non-EFA and cholesterol, then this can be estimated to be of the order 16:0 > 18:1 > 18:0 > cholesterol. The ∆13Cblipid − dlipid fractionation observed for linoleic acid is significant, demonstrating that bone 18:0 δ13C values are consistently more enriched by 1.7‰ with respect to dietary values. This fractionation obviously cannot be taken as a measure of the extent of routing and must arise due to isotopic fractionation occurring during metabolic processes other than de novo synthesis. Linoleic acid is the most abundant FA in these diets but is one of the minor FA observed in bone (Figs. 1c, 1f, 1i, 1l, 1o). If we postulate that the diet provides more than enough linoleic acid for these animals, then the enrichment observed in bone linoleic δ13C acid values is consistent with a kinetic isotope effect occuring during its catabolism or conversion to other metabolites, e.g., prostaglandins and eicosanoids. Certainly, the formation of prostaglandins and eicosanoids from linoleic acid involves enzyme-catalyzed chain elongation and desaturation reac-
183
tions (21,22) where isotopic fractionation may be introduced. Correlations between diet and bone lipid δ13C values. The relationships between diet (whole diet, protein, energy, FA, and cholesterol) and bone (collagen, apatite, FA, and cholesterol) δ13C values were investigated using linear regression. Table 3 summarizes the R 2 values observed between each diet and bone component, and the most significant correlations are highlighted in bold. Collagen and apatite δ13C values were shown to correlate best with dietary protein (R 2 = 0.77, P ≤ 0.01) and whole diet (R 2 = 0.99, P = 0.001) values, respectively [as was also demonstrated for the data presented in Ambrose and Norr (10)], and these findings are consistent with those from the Tieszen and Fagre (11) study. The most significant correlations observed for the bone lipids are consistent with the interpretations of the ∆13Cblipid − dlipid fractionations above and are shown in Figure 3. As expected, bone linoleic δ13C values correlated extremely well with dietary values (R 2 > 0.99, P ≤ 0.001), providing further evidence for its direct incorporation from the diet. It was postulated that the majority of bone 16:0 and 18:1 FA were also routed directly from the diet, and indeed, the highest R 2 values of 0.97 (P ≤ 0.001) and 0.95 (P = 0.001), respectively, were shown with respect to their corresponding dietary FA. Conversely, the majority of bone 18:0 FA and cholesterol was postulated to result from de novo synthesis, and this hypothesis is corroborated by a higher degree of correlation with whole diet δ13C values rather than with their corresponding dietary FA values. Certainly, the low R 2 of 0.21 (P > 0.05) observed between diet and bone cholesterol indicates that the direct incorporation of cholesterol from the diet is not the dominant pathway. Correlations between ∆13Cblipid − wdiet fractionations and 13 ∆ Cdlipid − wdiet spacings. Greater insight into the extent of routing of dietary lipids into bone lipids can be gleaned from plotting ∆13Cblipid−wdiet fractionations against their corresponding ∆13Cdlipid−wdiet spacings (= δ13Cdiet lipid − δ13Cwhole diet). We focus here on the R 2 values and gradients (m) of the regression equations in Figure 4 to interpret diet-to-bone lipid relationships. R 2 values ranged from 0.67 to 0.96 (P ≤ 0.05), showing that these relationships correlated well/very well with each other. Significant differences in the gradients of these regression lines are observed, demonstrating the varying degrees to TABLE 3 R 2 Values from Linear Correlations of Diet and Bone δ13C Valuesa R 2 values Diet component Bone component 16:0 18:0 18:1 18:2 Cholesterol Collagen Apatite a
Whole diet
Protein
Energy
Lipidb
0.92 0.95 0.93 0.91 0.91 0.38 0.99
0.07 0.04 0.01 0.02 0.18 0.77 0.02
0.80 0.87 0.90 0.99 0.73 0.15 0.94
0.97 0.84 0.95 0.996 0.21 NA NA
Boldfaced values indicate the most significant correlations. Lipid refers to corresponding lipid in the diet. NA, not applicable.
b
Lipids, Vol. 38, no. 2 (2003)
184
S. JIM ET AL.
FIG. 3. Linear correlations of: (a) diet and bone 16:0 FA, (b) whole diet and bone 18:0 FA, (c) diet and bone 18:1 FA, (d) diet and bone 18:2 FA, and (e) whole diet and bone cholesterol δ13C values. WDIET = whole diet; CHOL = cholesterol; d = diet; b = bone; and dotted line is where x = y.
which ∆13Cblipid −wdiet fractionations depend on the difference between dietary lipid and whole-diet δ13C values. The largest gradient is observed for linoleic acid where m = 0.97, and this can be interpreted as indicating that 97% of the carbon atoms in bone 18:2 are derived from the diet, reflecting the direct incorporation of this EFA from the diet. Thus, from Figure 4, the degree of routing of diet-to-bone lipids can be estimated to be of the order 18:2 (97%) > 16:0 (80%) > 18:1 (64%) > 18:0 (48%) > cholesterol (21%). Again, these estimates are entirely consistent with previous interpretations of routing as discussed above and, moreover, reflect the relative abundances of these lipids in the diet. The estimated percentage of routing for cholesterol corresponds to the predicted 20% discussed above and compares well with an estimate of 16% (34) where a further set of animals raised on marine protein were also considered. In summary, the δ13C values of the major lipids found in bone were shown to reflect different dietary signals. δ13C values of bone cholesterol and stearic (18:0) acid reflected whole-diet δ13C values, whereas the δ13C values of bone palmitic (16:0), oleic (18:1), and linoleic (18:2) acids reflected dietary FA δ13C values. Differences in their dietary signals were attributed to the different extents to which dietary lipids were routed to bone. Linoleic acid is an EFA, and the routing estimate of 97% confirms its direct incorporation Lipids, Vol. 38, no. 2 (2003)
from the diet. Significantly, the extent of routing of the nonEFA and cholesterol reflected their relative abundances in the diet. Although tissue lipid compositions have been shown to be greatly influenced by dietary lipids (24–28), this is the first stable isotopic experimental evidence at natural abundance to show that FA are absorbed proportionately from the diet. The potential of using bone FA and cholesterol δ13C values as indicators of diet has therefore been shown in this study. However, the interpretation of lipid δ13C values from ecological and archaeological studies, where there are likely to have been temporal changes in diet, requires knowledge of their turnover rates. Bone FA and cholesterol turnover rates have been estimated from the study of rats that were subjected to diet-switch experiments (Jim, S., Evershed, R.P., and Ambrose, S.H., unpublished data). ACKNOWLEDGMENTS We thank the Wellcome Trust for providing the Bioarchaeology Studentship (047442/Z/96/Z) and Fellowship (057166/Z/99/Z) for this research. NERC is thanked for financial support for mass spectrometry facilities (GR 3/2951, GR 3/3758, and FG 6/36/01). We also thank Jim Carter and Andrew Gledhill for technical assistance with GC/MS and GC/C/IRMS. Controlled diet experiments were supported by the National Science Foundation, USA (BNS 9010937 and SBR 9212466), and the University of Illinois Research Board. Con-
ROUTING AND DE NOVO SYNTHESIS OF BONE LIPIDS
185
FIG. 4. Double ∆ correlations of ∆13Cdlipid −wdiet spacings vs. corresponding ∆13Cblipid −wdiet fractionations: (a) 16:0 FA, (b) 18:0 FA, (c) 18:1 FA, (d) 18:2 FA, and (e) cholesterol. WDIET = whole diet; CHOL = cholesterol; d = diet; and b = bone.
trolled diet experiment protocols were approved by the Office of Laboratory Animal Care (OLAC), University of Illinois, UrbanaChampaign.
REFERENCES 1. van der Merwe, N.J. (1982) Carbon Isotopes, Photosynthesis, and Archaeology, Am. Sci. 70, 596–606. 2. Chisholm, B.S. (1989) Variation in Diet Reconstructions Based on Stable Carbon Isotopic Evidence, in The Chemistry of Prehistoric Human Bone (Price, T.D., ed.), pp. 10–37, Cambridge University Press, Cambridge. 3. Lee-Thorp, J.A., Sealy, J.C., and van der Merwe, N.J. (1989) Stable Carbon Isotope Ratio Differences Between Bone Collagen and Bone Apatite, and Their Relationship to Diet, J. Archaeol. Sci. 16, 585–599. 4. Schwarcz, H.P. (1991) Some Theoretical Aspects of Isotope Paleodiet Studies, J. Archaeol. Sci. 18, 261–275. 5. Schwarcz, H.P. (2000) Some Biochemical Aspects of Carbon Isotopic Paleodiet Studies, in Biogeochemical Approaches to Paleodietary Analysis (Ambrose, S.H., and Katzenberg, M.A., eds.), pp. 189–209, Kluwer Academic/Plenum, New York. 6. Koch, P.L., Fogel, M.L., and Tuross, N. (1994) Tracing the Diets of Fossil Animals Using Stable Isotopes, in Methods in Ecology, Stable Isotopes in Ecology and Environmental Science (Lajtha, K., and Michener, R.H., eds.), pp. 63–92, Blackwell Scientific Publications, Oxford. 7. DeNiro, M.J., and Epstein, S. (1978) Influence of Diet on the Distribution of Carbon Isotopes in Animals, Geochim. Cosmochim. Acta 42, 495–506.
8. DeNiro, M.J., and Epstein, S. (1981) Influence of Diet on the Distribution of Nitrogen Isotopes in Animals, Geochim. Cosmochim. Acta 45, 341–351. 9. Hare P.E., Fogel, M.L., Stafford, T.W., Jr., Mitchell, A.D., and Hoering, T.C. (1991) The Isotopic Composition of Carbon and Nitrogen in Individual Amino Acids Isolated from Modern and Fossil Proteins, J. Archaeol. Sci. 18, 277–292. 10. Ambrose, S.H., and Norr, L. (1993) Experimental Evidence for the Relationship of the Carbon Isotope Ratios of Whole Diet and Dietary Protein to Those of Bone Collagen and Carbonate, in Prehistoric Human Bone—Archaeology at the Molecular Level (Lambert, J.B., and Grupe, G., eds.), pp. 1–37, Springer-Verlag, Berlin. 11. Tieszen, L.L., and Fagre, T. (1993) Effect of Diet Quality and Composition on the Isotopic Composition of Respiratory CO2, Bone Collagen, Bioapatite and Soft Tissues, in Prehistoric Human Bone—Archaeology at the Molecular Level (Lambert, J.B., and Grupe, G., eds.), pp. 121–155, Springer-Verlag, Berlin. 12. Matthews, D.E., and Hayes, J.M. (1978) Isotope-Ratio-Monitoring Gas Chromatography–Mass Spectrometry, Anal. Chem. 50, 1465–1473. 13. Fantle, M.S., Dittel, A.I., Schwalm, S.M., Epifanio, C.E., and Fogel, M.L. (1999) A Food Web Analysis of the Juvenile Crab, Callinectes sapidus, Using Stable Isotopes in Whole Animals and Individual Amino Acids, Oecologia 120, 416–426. 14. O’Brien, D.M., Fogel, M.L., and Boggs, C.L. (2002) Renewable and Nonrenewable Resources: Amino Acid Turnover and Allocation to Reproduction in Lepidoptera, Proc. Nat. Acad. Sci. USA 99, 4413–4418. 15. Pond, C.M., and Gilmour, I. (1997) Stable Isotopes in Adipose
Lipids, Vol. 38, no. 2 (2003)
186
16.
17. 18.
19.
20.
21. 22.
23. 24. 25. 26.
S. JIM ET AL.
Tissue Fatty Acids as Indicators of Diet in Arctic Foxes (Alopex lagopus), Proc. Nutr. Soc. 56, 1067–1081. Stott, A.W., Davies, E., Tuross, N., and Evershed, R.P. (1997) Monitoring the Routing of Dietary and Biosynthesized Lipids through Compound-Specific Stable Isotope (δ13C) Measurements at Natural Abundance, Naturwissenschaften 84, 82–86. Hammer, B.T., Fogel, M.L., and Hoering, T.C. (1998) Stable Carbon Isotope Ratios of Fatty Acids in Seagrass and Redhead Ducks, Chem. Geol. 152, 29–41. Stott, A.W., Evershed, R.P., Jim, S., Jones, V., Rogers, J.M., Tuross, N., and Ambrose, S.H. (1999) Cholesterol as a New Source of Palaeodietary Information: Experimental Approaches and Archaeological Applications, J. Archaeol. Sci. 26, 705–716. Jim, S. (2000) The Development of Bone Cholesterol δ13C Values as a New Source of Palaeodietary Information: Models of Its Use in Conjunction with Bone Collagen and Apatite δ13C Values, Ph.D. Thesis, University of Bristol, England, pp. 117–123. Jim, S., Stott, A.W., Evershed, R.P., Rogers, J.M., and Ambrose, S.H. (2001) Animal Feeding Experiments in the Development of Cholesterol as a Palaeodietary Indicator, in Archaeological Sciences ’97 (Millard, A., ed.), pp. 68–77, BAR International Series 939, Oxford. Voet, D., and Voet, J.G. (1995) Biochemistry, 2nd edn., pp. 687–688, John Wiley & Sons, New York. McGarry, J.D. (1997) Lipid Metabolism I: Utilization and Storage of Energy in Lipid Form, in Textbook of Biochemistry with Clinical Correlations (Devlin, T.M., ed.), pp. 361–393, WileyLiss, New York. Krebs, H.A., and Lowenstein, J.M. (1960) The Tricarboxylic Acid Cycle, in Metabolic Pathways (Greenberg, D.M., ed.), Vol. 1, pp. 129–203, Academic Press, New York. Sabine, J.R. (1977) Cholesterol, p. 81, Marcel Dekker, New York. Das, S.K., Scott, M.T., and Adhikary, P.K. (1975) Effect of the Nature and Amount of Dietary Energy on Lipid Composition of Rat Bone Marrow, Lipids 10, 584–590. Pauletto, P., Puato, M., Angeli, M.T., Pessina, A.C., Munhambo, A., Bittolo-Bon, G., and Galli, C. (1996) Blood
Lipids, Vol. 38, no. 2 (2003)
27.
28.
29.
30. 31.
32. 33. 34.
Pressure, Serum Lipids, and Fatty Acids in Populations on a Lake-Fish Diet or on a Vegetarian Diet in Tanzania, Lipids 31, S309–S312. Vidgren, H.M., Ågren, J.J., Schwab, U., Rissanen, T., Hänninen, O., and Uusitupa, M.I.J. (1997) Incorporation of n-3 Fatty Acids into Plasma Lipid Fractions, and Erythrocyte Membranes and Platelets During Dietary Supplementation with Fish, Fish Oil, and Docosahexaenoic Acid-Rich Oil Among Healthy Young Men, Lipids 32, 697–705. Alam, S.Q., Kokkinos, P.P., and Alam, B.S. (1993) Fatty Acid Composition and Arachidonic Acid Concentrations in Alveolar Bone of Rats Fed Diets with Different Lipids, Calcified Tissue Int. 53, 330–332. Rambjør, G.S., Wålen, A.I., Winsor, S.L., and Harris, W.S. (1996) Eicosapentaenoic Acid Is Primarily Responsible for Hypotriglyceridemic Effect of Fish Oil in Humans, Lipids 31, S45–S49. Rhee, S.K., Reed, R.G., and Brenna, J.T. (1997) Fatty Acid Carbon Isotope Ratios in Humans on Controlled Diets, Lipids 32, 1257–1263. Mottram, H.R., and Evershed, R.P. (2001) Elucidation of the Composition of Bovine Milk Fat Triacylglycerols Using HighPerformance Liquid Chromatography–Atmospheric Pressure Chemical Ionisation Mass Spectrometry, J. Chromatogr. A 926, 239–253. DeNiro, M.J., and Epstein, S. (1977) Mechanisms of Carbon Isotope Fractionation Associated with Lipid Synthesis, Science 197, 261–263. Hayes, J.M. (1993) Factors Controlling 13C Contents of Sedimentary Organic Compounds: Principles and Evidence, Mar. Geol. 113, 111–125. Jim, S., Ambrose, S.H., and Evershed, R.P. (in press) Stable Carbon Isotopic Evidence for Differences in the Biosynthetic Origin of Bone Cholesterol, Collagen and Apatite: Implications for Their Use in Palaeodietary Reconstruction, Geochim. Cosmochim. Acta.
[Received July 12, 2002, and in revised form February 24, 2003; accepted February 27, 2003]
