Blood ¹⁵N:¹³C Enrichment Ratios Are Proportional to the Ingested Quantity of Protein with the Dual-Tracer Approach for Determining Amino Acid Bioavailability in Humans

ABSTRACT

Background

Assessment of amino acid bioavailability is of key importance for the evaluation of protein quality; however, measuring ileal digestibility of dietary proteins in humans is challenging. Therefore, a less-invasive dual stable isotope tracer approach was developed.

Objective

We aimed to test the assumption that the ¹⁵N:¹³C enrichment ratio in the blood increases proportionally to the quantity ingested by applying different quantities of ¹⁵N test protein.

Methods

In a crossover design, 10 healthy adults were given a semi-liquid mixed meal containing 25 g (low protein) or 50 g (high protein) of ¹⁵N-labeled milk protein concentrate simultaneous with 0.4 g of highly ¹³C–enriched spirulina. The meal was distributed over multiple small portions, frequently provided every 20 min during a period of 160 min. For several amino acids, the blood ¹⁵N- related to ¹³C-isotopic enrichment ratio was determined at t = 0, 30, 60, 90, 120, 180, 240, 300, and 360 min and differences between the 2 meals were compared using paired analyses.

Results

No differences in ¹³C AUC for each of the measured amino acids in serum was observed when ingesting a low- or high-protein meal, whereas ¹⁵N AUC of amino acids was ∼2 times larger on the high-protein meal (P < 0.001). Doubling the intake of ¹⁵N-labeled amino acids increased the ¹⁵N:¹³C ratio by a factor of 2.04 ± 0.445 for lysine and a factor between 1.8 and 2.2 for other analyzed amino acids, with only phenylalanine (2.26), methionine (2.48), and tryptophan (3.02) outside this range.

Conclusions

The amino acid ¹⁵N:¹³C enrichment ratio in the peripheral circulation increased proportionally to the quantity of ¹⁵N-labeled milk protein ingested, especially for lysine, in healthy adults. However, when using ¹⁵N-labeled protein, correction for, e.g., α-carbon ¹⁵N atom transamination is advised for determination of bioavailability of individual amino acids. This trial was registered at www.clinicaltrials.gov as NCT02966704.

Introduction

A critical factor in protein quality assessment is amino acid (AA) bioavailability, which represents the fraction of AA available to the organism after digestion and absorption (1, 2). It is usually derived from ileal AA digestibility—that is, digestibility measured at the terminal ileum after ingestion of a protein. As it is assumed that AAs are absorbed exclusively in the small intestine and that the fraction entering the large intestine is not absorbed in nutritionally relevant quantities (3–5). A promising non-, or minimally, invasive approach for measuring AA bioavailability in humans is the dual stable isotope tracer approach (4, 6, 7).

This dual stable isotope tracer method has been applied for measuring the bioavailability of AAs from different intrinsically labeled proteins in humans (8, 9). The method relies on the ingestion of a meal containing a test protein of unknown AA bioavailability, intrinsically labeled with ²H or ¹⁵N, together with a tracer dose of a ¹³C-labeled protein or standard free AA mixture of known AA bioavailability. The comparison of the differential isotopic enrichment (¹⁵N or ²H vs ¹³C) of each AA in the blood and in the meal reflects, when correcting for label transfer, for example, due to transamination, the relative absorption of each AA that originates from the 2 labeled proteins.

Even though the dual-tracer method has been applied in several studies (6, 8, 10) and its results were comparable with bibliographic true ileal or apparent fecal digestibility values in pigs and humans (9, 11), the approach has not been validated. The approach is based on several assumptions. One of these assumptions is that the ¹⁵N:¹³C (or ²H:¹³C) enrichment ratio for each AA in blood is identical to the ratio after absorption and that postabsorptive metabolism, including isotopic fractionation, does not influence this ratio. One can test this assumption by varying the quantity of bioavailable protein. Varying the protein quantity provided or its bioavailability should result in a proportional response of this ratio in peripheral blood.

In this study we applied the dual-tracer approach for AA bioavailability using ¹⁵N-milk protein concentrate together with ¹³C-spirulina as test and standard protein, respectively. These protein sources were chosen as their AA digestibility is known (8, 12). The aim was to test the critical assumption of the method that variation in bioavailability should be proportionally reflected in changes in isotope ratios. Therefore, different quantities of test protein were applied to validate whether the ¹⁵N:¹³C enrichment ratio of AAs in the peripheral circulation increased proportionally to the quantity of ¹⁵N-labeled milk protein concentrate ingested while keeping the quantity of ¹³C spirulina identical.

Methods

The experimental protocol was approved by the Medical Ethical Committee of Wageningen University (registered at clinicaltrials.gov: NCT02966704).

Study population

The study population consisted of 10 healthy subjects (5 males, 5 females), with an age between 20 and 35 y. Habitual dietary intake was measured with a food-frequency questionnaire, and general health was assessed using a basic questionnaire. Height and body weight were measured. To be included in the study, subjects needed to have a BMI (kg/m²) between 18.5 and 25 and a habitual dietary protein intake between 10% and 30% of total energy intake. Subjects were excluded when having a history of medical or surgical events or a disease that might influence study outcomes, using medical drugs except for paracetamol, being allergic or intolerant to milk or lactose, when having an alcohol consumption >14 units for females or >21 units for males per week, smoking, using abusive drugs, performing moderate to intense physical activity for >5 h/wk, reporting weight loss or gain of >3 kg in the last month, reporting a slimming or medical diet, or being pregnant or breastfeeding. All subjects gave their written informed consent. The included subjects were 23.5 ± 3.2 y old, had a BMI of 22.2 ± 0.58, and habitual protein intake of 14.4% ± 2.5% of total energy intake.

Study design

A randomized crossover design was applied with 2 different experimental meals, providing either 25 or 50 g of ¹⁵N-enriched milk proteins. These meals were provided on 2 different test days, which were separated by 1 wk. Order of the meals was randomized. The day prior to each test day subjects were asked to refrain from vigorous exercise and foods naturally enriched in ¹³C, like corn or pineapple. Furthermore, subjects were asked to copy their dietary behavior from the first test day to the second test day. For practical reasons, the study was executed in 2 parts: the first part, executed in August 2016, included 4 participants (2 males, 2 females) and the second part, executed in November 2016, included 6 participants (3 males, 3 females).

Experimental meals

To produce ¹⁵N-labeled milk protein concentrate, 2 Holstein Friesian dairy cows were rumen-infused with ¹⁵N-ammonium sulfate (98% enriched, ∼3.07 g/h per cow; Cambridge Isotope Laboratories, Inc.) and milk protein was extracted. The protein content of the ¹⁵N-labeled milk protein concentrate was 67.9% (N × 6.38) and the total ¹⁵N enrichment was 2.27 atom percent (AP) or 1.90 atom percent excess (APE), as determined using an elemental analyzer coupled to isotope ration mass spectrometer. The 2 experimental meals consisted of either 25 g or 50 g of ¹⁵N-labeled milk protein concentrate providing 18.2 g (low-protein meal) and 35.5 g (high-protein meal) of crude protein, respectively. Both meals contained 400 mg U¹³C-labeled whole spirulina [>98% enriched (>98AP); Cambridge Isotope Laboratories, Inc.]. The commercially obtained ¹³C-labeled whole spirulina (>98% enriched; Cambridge Isotope Laboratories, Inc.) consisted of 39.6% (N × 6.25) protein, 2.83% fat, 23.2% neutral detergent fiber, 4.94% ash, and 2.30% moisture. It was produced under quality standards: ISO Guide 34, ISO/IEC 17025, ISO 13485, current good manufacturing practice. The low-protein and high-protein meal also contained 55 g or 32 g dextrine-maltose (Fantomalt; Nutricia), 65 g concentrated lemonade syrup (Karvan cevitam, Cassis), 5.5 g or 5.0 g sunflower oil, 12 g or 7 g locust bean gum (Johannesbroodpitmeel), and 338 g or 341 g water, respectively. Both mixed meals were similar in energy content (513 kcal for the low-protein meal and 514 kcal for high protein) and total weight; moreover, no visual differences in color or viscosity were noticed. Nutrient composition was different between the meals, with the high-protein meal containing 27.8% of energy (E%) protein and 61.9 E% carbohydrates and the low-protein meal contained 14.3 E% protein and 75.4 E% carbohydrates. Both meals contained 10.3 E% fat.

Details of the test day

On the test days, subjects arrived at the human research facilities of Wageningen University after an overnight fast of ≥10 h. After measurement of body weight, a catheter was placed in the antecubital vein and baseline blood samples were taken. Subsequently, the subjects ingested the semi-liquid meal, divided in 9 portions, every 20 min (t = 0, 20, 40, 60, 80, 100, 120, 140, and 160 min). Postprandial blood samples were taken at t = 30, 60, 90, 120, 180, 240, 300, and 360 min after the first portion of the semi-liquid meal was consumed. Blood serum samples were used to determine isotopic enrichment of free AAs. At each time point, 16 mL of blood was collected in two 8-mL SST II Advance blood collection tubes (BD Biosciences). After 30–60 min of clotting, the tubes were centrifuged (1550 × g, 10 min, room temperature) and subsequently serum was placed into aliquots and stored at −80°C until measurement.

¹⁵N and ¹³C AA enrichments

The meal samples were hydrolyzed (6 M HCl at 110°C for 24 h) and filtered (0.22 μm) prior to the extraction. To obtain a proper estimate for meal enrichment, the meals of all study days were analyzed in duplicate. Serum samples were acidified using 1 M HCl and directly mixed with the resin. Serum free AAs were extracted using cation-exchange Dowex AG50×8 resin conditioned under H⁺-form and eluted with 6 M NH₄OH. The purified AAs were analyzed by Gas chromatography combustion isotope ratio mass spectrometry as N(O)-ethoxycarbonyl ethyl ester derivatives. Briefly, 3.2 mL of an ethanol:pyridine solution (80:20, vol:vol) were added to 4 mL of the aqueous sample, the resulting mixture was vortexed before the addition of 400 μL ethyl chloroformate, and the solution gently shaken several times until no bubbles were formed. Then, 2 mL of dichloromethane:hexane (50:50, vol:vol) with 1% ethyl chloroformate (vol:vol) was added and the solution was vigorously vortexed. The upper organic phase was collected and evaporated under stream N₂. The obtained residue was diluted in 50 μL of ethylacetate and placed in chromatographic vials, and the samples were stored at −20°C until injection. The ¹⁵N and ¹³C enrichments of derivatized AAs were determined using an Agilent 7890B gas chromatograph (Agilent Technologies) coupled to an Isoprime isotope ratio mass spectrometer (Isoprime; GV Instrument) via the GC5 Isoprime interface. The temperature in the interface was regulated at 350°C and combustion oven was maintained at 930°C and 850°C for ¹⁵N and ¹³C analysis, respectively. A 30-m Rxi-17 capillary column (0.25 mm i.d. and 0.5-μm film thickness; Restek) was used in a constant flow mode, and the high-purity helium was used as the carrier gas at a flow rate of 1.2 mL/min. The inlet temperature was set at 270°C. Samples (2 μL) were injected in splitless mode for ¹⁵N analysis and in split mode (10:1) for ¹³C analysis. The temperature program started at 150°C, increased to 200°C by 4°C/min, and then to 270°C by 25°C/min. The temperature was maintained for 20 min at 270°C. The stable isotopic compositions of nitrogen and carbon were measured using the conventional delta per mill notation: the δ¹⁵N and δ¹³C values were expressed relative to the international standards (AIR-N2 and PDB, respectively). In serum, the enrichment (AP) of ¹⁵N and ¹³C was corrected for the baseline (t = 0) enrichment of each individual AA to obtain APE. To obtain APE, in the meal the enrichment of ¹⁵N and ¹³C was corrected for overall background protein ¹⁵N and ¹³C abundance determined using an elemental analyzer coupled to an isotope ration mass spectrometer and measured in the unlabeled milk collected from cows before the start of label infusion.

Calculations and statistical analyses

The postprandial response in plasma AA ¹⁵N and ¹³C enrichment (APE) is given for each of the analyzed AAs. To quantify the postprandial response, the AUC of plasma AA enrichment over the entire study day (t = 0−360 minutes) was calculated using the trapezoid rule. The AUC was chosen as the main response parameter in the absence of a clear steady state in our experiment.

The blood-to-meal AA ¹⁵N- to ¹³C-enrichment ratio (R1) between the 2 protein sources was calculated according to the following formulas below:

(1)

R_meal is the meal enrichment ratio of the i AA, with ¹⁵N AA and ¹³C AA the ¹⁵N-AA and ¹³C-AA enrichments of the i AA (in APE).

(2)

R_blood is the blood enrichment ratio of the i AA, with ¹⁵N_AUC AA and ¹³C_AUC AA the AUC values of isotopic enrichment of a specific AA (in APE). Finally, R₁ is the blood-to-meal ¹⁵N- to ¹³C-enrichment ratio of the i specific AA.

(3)

The R₁ value depends for each AA on the ratio between AA digestibility of milk protein and spirulina.

Because of a potential label transfer (e.g., the exchange of ¹⁵N, but not of ¹³C) between indispensable AAs through transamination of the α-carbon ¹⁵N, digestibility can either be over- or underestimated using R₁. Therefore, the R₁ value for each AA was compared with R₂ as follows:

(4)

R₂ is the relative digestibility of the i AA between milk protein and spirulina of the i AA as obtained from the literature (8, 12).

Comparing R₁ to R₂ allows estimation of a transamination correction factor (TcF):

(5)

When TcF = 1 there is no label transfer due to, for example, transamination; with TcF <1 there is label transfer with a gain in ¹⁵N compared with ¹³C; and with TcF >1 there is label transfer with a reduction in ¹⁵N compared with ¹³C.

All data are expressed as means ± SDs. Differences in enrichment of the 2 meals were tested using a t test with differences being significant when P < 0.05. Within-subject differences in AUC and R_blood values between the 2 test meals for each label and each AA were compared using a paired t test. The 95% CI was used to check if the difference in R_blood between the high- and low-protein meal was 2-fold.

Results

¹⁵N- and ¹³C-AA meal enrichment and R_meal calculation

¹⁵N and ¹³C enrichment (APE) of AAs in the meal and the ratio ¹⁵N:¹³C enrichment of the AAs in the meal (R_meal)are given in Table 1. The ¹⁵N enrichment was numerically similar in both meals and did not differ significantly for all AAs except for tryptophan and phenylalanine (P < 0.05; Table 1). The ¹³C enrichment was lower on the high-protein meal compared with the low-protein meal for all AAs, except for lysine (P = 0.057) and methionine (P = 0.128), due to dilution of the ¹³C-labeled AAs by carbons from the milk protein AAs.

TABLE 1.

Enrichment of individual amino acids in the semi-liquid mixed meals with 25 g (low protein) or 50 g (high protein) of ¹⁵N-labeled milk protein concentrate with 0.4 g of ¹³C-labeled spirulina¹

	15N enrichment (APE)			13C enrichment (APE)			Rmeal = 15N:13C
	Low protein	High protein	P	Low protein	High protein	P	Low protein	High protein
Ala	1.66 ± 0.091	1.83 ± 0.058	>0.05	1.33 ± 0.120	0.79 ± 0.062	<0.001	1.25	2.33
Gly	1.45 ± 0.084	1.58 ± 0.133	>0.05	1.53 ± 0.123	0.97 ± 0.061	<0.001	0.95	1.63
Val	1.61 ± 0.050	1.65 ± 0.087	>0.05	0.59 ± 0.050	0.33 ± 0.022	<0.001	2.70	4.96
Leu	1.53 ± 0.028	1.58 ± 0.024	>0.05	0.53 ± 0.049	0.30 ± 0.019	<0.001	2.87	5.34
Ile	1.71 ± 0.118	1.90 ± 0.060	>0.05	0.62 ± 0.054	0.34 ± 0.020	<0.001	2.78	5.50
Thr	1.28 ± 0.063	1.60 ± 0.232	>0.05	0.81 ± 0.200	0.49 ± 0.133	<0.05	1.58	3.28
Phe	1.53 ± 0.028	1.61 ± 0.014	<0.05	0.54 ± 0.047	0.30 ± 0.019	<0.001	2.82	5.29
Lys	2.03 ± 0.071	2.07 ± 0.022	>0.05	0.49 ± 0.182	0.26 ± 0.068	>0.05	4.16	7.93
Met	1.54 ± 0.083	1.76 ± 0.141	>0.05	0.46 ± 0.161	0.31 ± 0.031	>0.05	3.37	5.67
Trp	0.87 ± 0.027	0.94 ± 0.034	<0.05	0.40 ± 0.064	0.22 ± 0.026	<0.01	2.20	4.20

¹The table shows the mean ± SD of meals provided on different test days, n = 4 for ¹³C and n = 3 for ¹⁵N. APE, atom percent excess.

Appearance of ¹⁵N- and ¹³C-labeled AA in blood and R_blood calculation

The appearance of ¹³C and ¹⁵N enrichment in the blood AA pool (Figure 1) showed that ¹⁵N enrichment was numerically lower after ingestion of 25 g compared with 50 g ¹⁵N-labeled milk protein concentrate, whereas no visual difference was observed on ¹³C enrichment. Enrichment of both ¹³C and ¹⁵N AAs steadily increased during consumption of the multiple small portions and peaked at 180 min. In the period after the ingestion of the multiple small portions, towards the end of the study day (t = 240–360 min), a decrease in labeled AAs in blood was observed for most AAs. The AUC ¹³C and ¹⁵N enrichment and R_blood of the different circulating AAs are reported in Table 2. Comparing both meals, no significant differences in the AUC for ¹³C enrichment for each of the AAs was observed, whereas the AUC of ¹⁵N AAs in serum was significantly larger on the high-protein meal for each of the measured AAs, as expected. Consequently, R_blood (¹⁵N:¹³C) was larger after ingestion of the high-protein meal compared with the low-protein treatment for all AAs except for tryptophan. Doubling the intake of ¹⁵N-labeled AAs increased the ¹⁵N:¹³C ratio by a factor varying between 1.8 and 2.2 for the analyzed AAs, with only phenylalanine (2.26), methionine (2.48), and tryptophan (3.02) outside this range. For lysine, the fold-difference between the meals was numerically 2. Other AAs tended to deviate numerically from the expected 2-fold difference, even though the factor 2 was in the CI of each AA due to high variation between individuals.

FIGURE 1.

Specific amino acid enrichment in blood of healthy subjects receiving a meal containing 25 g (A and C) or 50 g (B and D) of ¹⁵N-labeled milk protein concentrate with 0.4 g of ¹³C-labeled spirulina. The meal was frequently distributed in 9 discrete portions given every 20 min; the time frame is indicated on top of each graph. Values are means ± SDs, n = 10. APE, atom percent excess.

TABLE 2.

AUCs of amino acid enrichment in blood of 10 healthy subjects receiving 25 g (low protein) or 50 g (high protein) of ¹⁵N-labeled milk protein concentrate with 0.4 g of ¹³C-labeled spirulina¹

	15N AUC			13C AUC			Rblood = 15N AUC/13C AUC			High:low protein	Lower and upper 95% CI (high:low protein)
	Low protein	High protein	P	Low protein	High protein	P	Low protein	High protein	P
Ala	44.2 ± 7.65	83.6 ± 13.1	<0.001	19.6 ± 2.87	21.3 ± 4.48	>0.05	2.29 ± 0.471	4.00 ± 0.635	<0.001	1.81 ± 0.448	1.53, 2.09
Gly	37.5 ± 8.70	69.1 ± 14.7	<0.001	18.1 ± 4.37	17.3 ± 5.18	>0.05	2.10 ± 0.411	4.33 ± 0.656	<0.001	2.10 ± 0.580	1.70, 2.50
Val	82.9 ± 14.1	151.6 ± 25.0	<0.001	37.9 ± 7.65	33.3 ± 5.38	>0.05	2.21 ± 0.273	4.57 ± 0.434	<0.001	2.11 ± 0.440	1.84, 2.38
Leu	62.9 ± 11.5	125.0 ± 18.9	<0.001	32.4 ± 4.51	29.8 ± 4.83	>0.05	1.94 ± 0.271	4.22 ± 0.319	<0.001	2.21 ± 0.398	1.97, 2.46
Ile	73.4 ± 15.9	142.3 ± 25.1	<0.001	48.3 ± 7.91	42.7 ± 5.94	>0.05	1.51 ± 0.194	3.33 ± 0.381	<0.001	2.19 ± 0.358	1.95, 2.42
Thr	42.9 ± 10.7	89.6 ± 20.5	<0.001	27.1 ± 5.53	26.6 ± 5.16	>0.05	1.75 ± 0.453	3.46 ± 0.953	<0.001	1.92 ± 0.588	1.45, 2.39
Phe	63.6 ± 12.4	125.7 ± 27.8	<0.001	28.6 ± 4.84	25.8 ± 5.27	>0.05	2.23 ± 0.368	4.91 ± 0.785	<0.001	2.26 ± 0.599	1.89, 2.64
Lys	108 ± 15.9	190 ± 17.1	<0.001	19.9 ± 2.65	17.8 ± 4.05	>0.05	5.33 ± 0.300	11.0 ± 2.02	<0.001	2.04 ± 0.445	1.75, 2.33
Met	72.6 ± 20.2	153 ± 36.2	<0.001	9.3 ± 7.80	14.6 ± 16.0	>0.05	10.0 ± 4.69	19.5 ± 12.3	<0.05	2.48 ± 1.858	1.10, 3.85
Trp	31.6 ± 7.39	62.1 ± 3.55	<0.001	7.91 ± 7.50	7.68 ± 4.50	>0.05	5.26 ± 3.35	10.0 ± 4.51	>0.05	3.02 ± 3.021	0.376, 5.67

¹Values are means ± SDs unless otherwise indicated, n = 10. The meal was frequently distributed in 9 discrete portions given every 20 min. R_blood for both meals and the fold-difference between these 2 meals are shown.

Blood-to-meal ¹⁵N:¹³C enrichment ratio (R₁) and estimation of ¹⁵N transamination for each AA (TcF)

R₁ for AAs was in the range from 0.544 to 3.44 (Table 3). For the different AAs, R₁ depends on the difference in the relative absorption of the ¹⁵N- and ¹³C-labeled AAs from milk and spirulina protein, respectively, but also depends on label transfer (e.g., on the level of transamination of the ¹⁵N atom on the α-carbon). For each circulating AA, the ratio R₁ was compared with the “reference” R₂, which is the milk protein to spirulina ratio from AA digestibility values reported in the literature (8, 12). The R₂ values ranged from 1.00 to 1.23 (Table 3). When AAs do not undergo label transfer such as transamination, as known for lysine and threonine, R₁ is expected to be not different from the ratio R₂. When there is a loss or a gain in ¹⁵N:¹³C atom enrichment due to metabolism such as α-carbon transamination or homocysteine remethylation, deviations of R1 above or below the value of R₂ are expected. This was expressed by calculating transamination correction factors (TcFs = R₁/R₂), which ranged from 0.42 to 2.11 for the different AAs (Table 3). TcFs reported in this study are an estimation and represent the potential correction required for the R₁ values to calculate AA digestibility of the ¹⁵N-labeled test protein using the ¹⁵N:¹³C dual-tracer approach.

TABLE 3.

Comparison of current data with the literature for estimation of TcF¹

	Digestibility, %			R1 = Rblood/Rmeal		TcF = R2/R1
	Milk2	Spirulina3	R2	Low protein	High protein	Low protein	High protein
Ala	95.9	—	—	1.84 ± 0.377	1.72 ± 0.273	—	—
Gly	99.3	—	—	2.21 ± 0.434	2.66 ± 0.403	—	—
Val	93.4	87.1	1.07	0.82 ± 0.101	0.92 ± 0.087	1.32 ± 0.155	1.17 ± 0.100
Leu	95.0	86.0	1.10	0.68 ± 0.094	0.79 ± 0.060	1.65 ± 0.235	1.40 ± 0.100
Ile	95.0	84.2	1.13	0.54 ± 0.070	0.61 ± 0.069	2.11 ± 0.321	1.89 ± 0.204
Thr	95.4	82.5	1.16	1.11 ± 0.287	1.05 ± 0.291	1.10 ± 0.273	1.18 ± 0.313
Phe	94.9	95.3	1.00	0.79 ± 0.131	0.93 ± 0.148	1.29 ± 0.224	1.10 ± 0.159
Lys	95.6	77.5	1.23	1.28 ± 0.072	1.39 ± 0.254	0.96 ± 0.053	0.91 ± 0.162
Met	91.6	84.1	1.09	2.97 ± 1.390	3.44 ± 2.170	0.42 ± 0.145	0.57 ± 0.544
Trp	—	—	—	2.39 ± 1.521	2.37 ± 1.074	—	—

¹Values are means ± SDs unless otherwise indicated, n = 10. The blood-to-meal ¹⁵N/¹³C ratios (R₁) obtained in healthy participants who received 25 g (low protein) or 50 g (high protein) of ¹⁵N-labeled milk protein concentrate with 0.4 g of ¹³C-labeled spirulina are shown. Literature data on milk protein and spirulina amino acid digestibility were used to calculated R₂ and subsequently TcF. TcF, transamination correction factor.

²Data from reference 12

³Data from reference 8.

Discussion

The dual-tracer approach has been proposed to measure food protein AA bioavailability by using a labeled (¹⁵N or ²H) test protein together with a tracer dose of labeled (¹³C) standard protein or AA mixture in a test meal (4). This approach was shown to be a minimal invasive technique to measure dietary AA digestibility in adults, children, and cystic fibrosis patients (6, 10, 13). This dual-tracer approach was applied using ²H-labeled test protein in combination with ¹³C-labeled standard spirulina (8, 9, 11) as well as with ¹⁵N-labeled spirulina protein in combination with ²H-labeled phenylalanine (6). In this context, the present study evaluated the potential of using ¹⁵N-labeling of milk protein with ¹³C spirulina for measuring dietary protein AA bioavailability in humans. Variation in bioavailability should be proportionally reflected in changes in isotope ratios. The validity of this approach is difficult to test with protein sources differing in bioavailability as, by definition, variations in bioavailability between such sources and methodology (dual isotope technique vs. true ileal AA disappearance) have to be combined. To circumvent this problem, we tested this crucial assumption by doubling the provision of a highly digestible protein source.

We evaluated the effect of the quantity of ¹⁵N-labeled test milk protein when applying the dual-tracer method on blood isotopic ¹⁵N and ¹³C-AA enrichment by providing 25 g or 50 g of ¹⁵N-labeled milk protein concentrate with 0.4 g of ¹³C-labeled spirulina. Doubling the quantity of milk protein provided was intended to double the quantity absorbed. Increasing the quantity of protein in a meal does not have detrimental effects on the proportion digested and absorbed, as previous studies showed that increasing protein content up to 400 g/kg dry matter intake did not affect true ileal crude protein digestibility in rats (14) and pigs (15), nor did crude protein content affect true ileal AA digestibility in pigs (16). Moreover, milk protein is an easily digestible protein, with true ileal digestibility values >90% (12). Thus, increasing the quantity provided will also result in proportional AA absorption and thereby a good strategy to test whether the measured blood isotopic ¹⁵N:¹³C ratio is not affected by postabsorptive metabolism.

No difference was observed in the ¹³C enrichment of circulating AAs between the low- and high-protein meal, which both contained an equal quantity of ¹³C-labeled spirulina. As expected, the ¹⁵N AUC was higher after ingestion of the high-protein meal compared with the low-protein meal. Most importantly, the 2-fold difference for R_blood between these meals for lysine confirmed our idea that the ¹⁵N:¹³C ratio in the blood is proportional to the quantity of AAs absorbed. Furthermore, for all other AAs no significant deviation from 2 was observed, although this might be related to the high variation, as phenylalanine, methionine, and tryptophan are numerically very deviating. Thereby the results validate a crucial assumption in the application of the dual-isotope approach that the isotopic appearance of AAs is proportional to the absorbed quantities, especially for lysine. Thus, potential differences in postabsorptive metabolism for ¹⁵N- and ¹³C-labeled AAs are not affected by the quantity absorbed (7). This observation is also critical for the future application of the dual-tracer approach as it was unclear whether the relative digestibility estimated with the dual-tracer approach would be affected by a quantity of protein provided.

In previous dual-tracer studies, similar protein quantities (21–24 g vs 18.2 g in the present study for the low-protein meal) were provided in a meal with higher energy content (714–834 kcal vs this study's meal of 513 kcal) (9, 11). Therefore, our study's protein-to-energy ratio was only slightly higher for the low-protein meal (14.3 E% protein) but markedly higher for the present high-protein meal (27.8 E% protein) compared with other dual-tracer studies. As it is well known that the amount of carbohydrates in a protein meal can influence the retention of AAs in the body (17), this could theoretically have influenced postabsorptive metabolism of labeled AAs and, in the case of isotopic fractionation, result in deviating labeled AA fluxes (18). However, here we showed that postabsorptive metabolism of labeled AAs does not influence the proportional appearance of labeled lysine and potentially other AAs.

Unlike earlier work using the dual-tracer approach (8) we used the combination of ¹⁵N test protein and ¹³C reference protein to test the critical assumption of proportional isotope ratios. Depending on the protein source, labeling with ¹⁵N is more convenient and efficient than ²H labeling, although it comes with additional challenges. In the dual-isotope approach with a ¹³C-labeled reference protein, there is for most AAs no exchange of ¹³C for the ¹³C-AAs absorbed from the reference ¹³C-labeled protein, and the ¹³C enrichment of circulating AAs is only related to its digestibility and absorption, except for methionine. In contrast, for the ¹⁵N-labeled test protein, the ¹⁵N enrichment of circulating AAs is proportional to the digestibility and absorption of the AA from the test protein but also to the level of α-carbon ¹⁵N atom transamination of the AAs. Transamination is a complex process of AA metabolism that is active in different tissues and differently affects AAs (19–21). Assuming that transamination is the main source of discrepancy between predicted (i.e., R₂) and observed relative digestibility (i.e., R₁) between our 2 protein sources, TcFs could be estimated. The TcF was calculated by comparing the relative digestibility measured in this study (R₁) with that calculated from data in the literature (R₂), although these literature values were determined with different methods and in different populations (8, 12). The values of the transamination factors ranged from 0.42 to 2.11 for ¹⁵N atom transamination. It is important to realize that these TcFs are much higher than those calculated for loss of deuterium (²H) from indispensable AAs that ranged from 1.002 to 1.081 (8). This is unavoidable because only 1 hydrogen atom out of multiple hydrogen atoms on the AA molecule is lost during transamination in comparison with the loss (or gain) of the only nitrogen atom in many AAs. The estimated TcFs >1 indicate that for the branched-chain AAs valine, leucine, and isoleucine ¹⁵N compared with ¹³C is lost. This is substantiated by the fact that, even though the transamination of branched-chain AAs is reversible, the likeliness of re-amination of the α-keto-acid with labeled nitrogen is virtually nil because of the rate of the process and the large amino-nitrogen pool size (18, 22). In contrast, for methionine, a transamination factor <1 was estimated from our data, suggesting an increased ¹⁵N relative to ¹³C, which may be caused by homocysteine remethylation (23). From our results it is clear that measuring the AA bioavailability of a test protein using ¹⁵N-labeling of the test protein requires a specific correction factor for most AAs. The rate of label transfer due to metabolism such as transamination and the appropriate correction factors for future studies need to be quantified before the dual-tracer approach can be routinely applied using ¹⁵N-labeled proteins. However, for AAs that do not undergo transamination, lysine and threonine, a correction factor for transamination may be less critical for the right determination of AA bioavailability than for other indispensable AAs such as the branched-chain AAs.

Our results also indicate other aspects that are essential for future applications of the dual-tracer approach. The R_meal was expected to be 2 times larger for the high-protein meal compared with the low-protein meal. The enrichments measured in our meal showed a high level of variation, especially for ¹³C, most likely due to flaws in the homogeneity of the meal and subsampling portions for analysis, as mixing small quantities of highly labeled spirulina within a semi-liquid meal does not necessarily lead to a homogeneous distribution. Moreover, the values differed slightly from our expectations; however, this could also be due to uncertainty in AA enrichment of spirulina and the unknown AA content and enrichment of the thickening agent. As the primary aim of the current study was not to calculate digestibility of the test protein, this uncertainty in our meal data did not influence our observation that the ¹⁵N:¹³C enrichment ratio in the blood increased proportionally to the quantity of ¹⁵N-labeled protein ingested. However, it affected the size of the blood-to-meal ratio (R₁) and the TcF values; therefore, no quantitative comparisons between low- and high-protein meals in R₁ of TcF were made.

The pattern of frequently feeding small meals prevented the potential effect of large differences in the intestinal kinetics (24, 25) between spirulina and milk protein and led to a parallel kinetic response of isotope enrichment in blood AAs. Accordingly, the pattern of ¹³C and ¹⁵N enrichment of circulating AAs over time showed quite similar profiles, although with a slightly faster decrease in ¹⁵N enrichment after the last meal. However, we did not obtain an isotopic plateau, probably because of a too-short meal feeding period or a lack of an initial priming bolus as applied by Devi et al. (8). Due to a lack of plateau, AUC was chosen for calculations. However, it is uncertain whether this issue, analytical variations for the chosen isotopic labels, or day-to-day subject variation resulted in the relatively high variation that was observed in Table 2. Moreover, the high variation for some AAs may have caused a lack in statistical power to show differences due to the quantity of test protein. As a result, the estimated TcFs also show high variability, especially compared with the previously reported <1% variation in ²H TcFs (8).

The calculations with the dual-tracer approach rely on multiple ratios; therefore, small deviations in isotopic enrichment in blood samples may have large effects on the final result of the digestibility calculation. For example, a 5% increase or decrease in ¹⁵N or ¹³C enrichment in the blood alone could result in an ∼10% increase or ∼10% decrease in R₁, and with similar variations in the meal enrichment this will be even larger. A slightly higher SD in mean indispensable AA digestibility value was also shown by Kashyap et al. (9), where digestibility values obtained with the dual-tracer approach were compared with bibliographic true ileal or apparent fecal digestibility values in pigs and humans. This potential variation should be taken into account when designing studies.

In conclusion, this study demonstrated that the ¹⁵N:¹³C enrichment ratio of AAs in the peripheral circulation increased proportionally to the quantity of ¹⁵N-labeled milk protein concentrate ingested, especially for lysine, indicating that postabsorptive metabolism is similar for an AA with either a ¹³C or ¹⁵N label, and that the ratio of labels in the blood is proportional and thereby representative for absorption. This illustrates the validity of the dual-isotope method to measure relevant variation in the bioavailability.

However, applying this dual-tracer approach with ¹⁵N to determine bioavailability of AAs from food proteins requires correction factors that are especially critical for AAs with high transamination rates and less essential for AAs that do not undergo transamination like threonine and lysine. The correction factors estimated in the current study need further quantification and validation before application.

Notes

This work was supported by the Dutch Dairy Association.

Author disclosures: DT is a member of the Journal's Editorial Board. The authors report no conflicts of interest.

The funders had no role in data collection and analysis, decision to publish, or preparation of the manuscript.

Abbreviations used: AA, amino acid; AP, atom %; APE, atom % excess; E%, % of energy; TcF, transamination correction factor.