Choline is a precursor of phosphatidylcholine and a methyl donor (via irreversible mitochondrial oxidation to betaine). Betaine is provided by the diet or from choline oxidation. Betaine is an important osmolyte and a methyl donor for homocysteine. The methylation reaction mediated by betaine-homocysteine methyltransferase (BHMT) gives dimethylglycine and methionine that is in turn the precursor of the universal methyl donor, S-adenosylmethionine (SAM).
A possible role of choline in cardiometabolic diseases has gained attention in recent years. Postprandial hypertriglyceridemia is associated with cardiovascular risk factors. Choline is required for synthesis of phosphatidylcholine that constitutes an integral part of lipoprotein particles and is thus involved in clearance of fats from blood stream and tissues. Accordingly, high choline intake can abolish the effect of high fat diet on liver steatosis. Plasma concentrations of free choline, betaine, and dimethylglycine show diverse associations with cardiometabolic risk factors [1
]. Low fasting plasma betaine and dimethylglycine (i.e.; a betaine demethylation product) predict incident diabetes [4
]. Trimethylamine N-oxide (TMAO), a gut bacteria choline-derived metabolite, is positively associated with cardiovascular risk factors [2
], vascular diseases [5
], and is elevated in patients with type 2 diabetes [2
]. In individuals at risk of diabetes, fasting plasma TMAO concentrations show positive correlation with fasting plasma glucose and total- and low-density lipoprotein (LDL)-cholesterol, but not with 2 h-plasma glucose or fasting triglycerides after oral glucose tolerance test (OGTT) [7
When individuals at risk for diabetes had modified their diet and lifestyle, the changes of plasma TMAO concentration were heterogeneous. A raise of plasma TMAO was associated with improved insulin sensitivity and lowering of plasma lipids [7
], suggesting that TMAO elevation in plasma could parallel glucose and/or lipid metabolisms. TMAO is produced from trimethylamine by flavin-containing monooxygenase 3 (FMO3) mainly in the liver. The expression of FMO3 in adipose tissues (but not in the liver) is positively associated with obesity and inversely associated with insulin sensitivity index [8
]. Thus, adipose tissues could be a source of TMAO that is associated with enhanced insulin sensitivity.
Plasma betaine, choline, and dimethylglycine are rather stable upon retesting [9
], whereas plasma TMAO is generally highly variable [9
]. Ingestion of fish (within 15 min) or beef and egg (after 2–4 h) raises plasma TMAO, but the effect appears to level off 6–12 h later [11
], suggesting that ≥8 h fasting time before testing TMAO could be sufficient. After six hours of oral ingestion of 50 mg d9-TMAO, only 5.5% of the dose retained in plasma, while 19% was detected in muscle and 42% was excreted in urine [13
]. This is in line with studies showing that the major part of the between-individual variations in plasma TMAO in healthy subjects was not explained by possible sources of TMAO in the diet [14
]. In general, plasma choline metabolites can be argued to be influenced by liver, kidney, muscle, and adipose tissues metabolism.
Glucose transport across the brush border occurs by a Na+
/glucose cotransporter. The exit of glucose of the cell across the basolateral membrane occurs through another transporter, GLUT2. In contrast, triglycerides are digested in the stomach, while most fats are hydrolyzed in the duodenum and jejunum and converted into lipid micelles that are partly taken up in the jejunum. A diet that is rich in carbohydrate or fiber can differentially change lipid metabolism by altering intestinal lipase (carbohydrate increases and fiber decreases lipase). Lipid uptake mechanisms (simple passive diffusion or protein-facilitated transport) take place in enterocyte brush border membrane with participation of several receptors and ATP-dependent sterol transporters across the membrane [15
]. Postprandial hypertriglyceridemia is ameliorated by several dietary components such as fibers, polyphenols, medium-chain fatty acids, and long-chain n-3 polyunsaturated fatty acids [16
]. High fat or glucose intake may impact choline uptake or metabolism.
The objective of the present study was to investigate changes of plasma concentrations of choline metabolites after oral fat tolerance test (OFTT) and OGTT among healthy individuals. We hypothesized that acute consumption of fat or glucose (exposures) affects choline metabolism as indicated by plasma concentrations of choline-derived metabolites (outcome). The oral tolerance tests were applied subsequently within one week apart and within-subject changes of plasma choline metabolites were investigated and compared between the tests.
2. Materials and Methods
2.1. Study Population
This intervention study investigated the intra-individual changes of plasma concentrations of choline metabolites after ingesting standardized amounts of fat and glucose applied on two independent days with one week in between. The study was conducted at the Department of Clinical Chemistry and Laboratory Medicine, Saarland University Hospital, Germany between May 2016 and January 2017 (ClinicalTrial.gov.ID: NCT02603237).
The inclusion criteria were: adults, men and women, age between 40 and 66 years, and a stable dietary pattern and body weight over the last 6 months. The exclusion criteria were; pregnant women, BMI > 25 kg/m2, using multivitamin supplements, diseases known to influence choline metabolism; diabetes (type 1 or 2), hypertension, malabsorption disorders, liver or renal disorders, and drugs (i.e.; anti-depressants, anti-epileptics, anti-folate, fibrate, statins, sex hormones, or metformin). Among the 20 subjects who were initially interested in participation; one subject was older than 66 and had multi-morbidities (cancer, vascular disease) and another one refused to participate in the OFTT test. Eighteen individuals (8 men; 10 women) took part in the study and attended the 2 visits.
The protocol was reviewed and approved by the Ethical Commission of the Saarland Region (approval Nr. 256/15). The study was conducted according to the ethical principles documented in Helsinki Declaration and all subjects provided signed informed consents to the study.
2.2. Study Design
The study consisted of two visits with a washout phase of approximately 7 days between the two tests (Supplemental Figure S1
). All participants received OFTT in the first visit in order to determine their fasting glucose and glycated haemoglobin (HbA1c) before applying OGTT in the second visit. Subjects who had undiagnosed type 2 diabetes had to be excluded before applying OGTT in the second visit. Thus, the study was not planned to follow a classical randomized design.
During visit 1, blood and urine samples were collected in the morning after an overnight fasting (≥8 h). Then, participants ingested 250 g milk cream that provided 80 g fat (32% fat, Landliebe®
) within 5 min. The nutritional composition of the milk cream is shown in Supplemental Table S1
. Blood and urine samples were collected again 4 h after OFTT. The participants were asked not to consume any solid foods during the 4 h, while drinking water was allowed during this period. During the second visit, subjects with both normal fasting plasma glucose and HbA1c, as a result of first visit, underwent OGTT. Fasting blood samples were collected before applying OGTT and a second blood collection was conducted 2 h after drinking 300 mL syrup containing 75 g glucose (ACCU CHEK, Dextro O.G-T.; Roche®
, Basel, Switzerland).
Information was collected on health history, weight, height, and waist-to-hip ratio (measured). The participants (n = 18) were all found to be free of diabetes during visit 1 and were thus able to take part in the OGTT test.
2.3. Blood and Urine Collection
Venous blood was collected into heparin- and EDTA-containing tubes before applying OFTT and OGTT. Those samples are referred to as baseline samples, which were collected after an overnight fasting (≥8 h). Additional blood samples were collected 4 h after OFTT and 2 h after OGTT. Blood count was directly conducted in EDTA whole blood. Blood samples were centrifuged and immediately separated into aliquots and stored at −70 °C. After centrifugation, EDTA-plasma (500 µL) was immediately acidified by adding 50 µL acetic acid (1N) and acidified samples were stored at −70 °C for measurements of SAM and S-adenosylhomocysteine (SAH).
Urinary samples (morning urine at baseline and after 4 or 2 h of applying the tests) were collected and separated into aliquots that were stored at −70 °C until analysis. Concentrations of all markers were measured within 8 months after sample collection. Concentrations of choline biomarkers were measured in all samples (4 plasma or urine samples for each participant) from the same person in the same run to reduce analytical variations.
2.4. Measurements of Choline and Its Metabolites
Urine samples were thawed and centrifuged at 10,000× g
for 5 min before measurement of creatinine or choline biomarkers. Plasma and urinary concentrations of TMAO and trimethylamine (a direct gut bacteria metabolite and a TMAO precursor) were measured using ultraperformance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS) with isotope labeled internal standards (d9
-TMAO and d9
-trimethylamine) as described before [17
]. Briefly, methanol/acetonitrile mix (15:85) containing 0.2% formic acid was added to the sample. The internal standards were added and samples were mixed, centrifuged for 5 min at 10,000× g
, and then separated on an Acquity UPLC BEH HILIC column (100 × 2.1 mm) with an Acquity HILIC VanGuard pre-column (5 × 2.1 mm) (Waters Corporation, Milford, MA, USA). The column temperature was set to 30 °C and the flow rate was 0.4 mL/min. The gradient elution consisted of 15 mmol/L ammonium formate (solvent A, pH = 3.5) and acetonitrile (solvent B). Positive multiple reaction monitoring mode was used and TMAO and TMA were monitored using the following ion transitions: m
75.90→58.10 and 59.98→44.02, respectively. d9
-TMAO and d9
-TMA were monitored at m
84.90→66.05 and 129.77→48.09, respectively. Urine TMA and TMAO were prepared using the same method after diluting the sample 1:5 in distilled water. The between-day coefficients of variation (CV) of TMAO and trimethylamine were <8% and <14%, respectively. Plasma trimethylamine levels were below the assay low detection limit (LOD = 0.12 µmol/L) in all 71 samples out of the 72 sample tested (1 sample had 0.19 µmol/L). Thus, the results of plasma trimethylamine are not presented as numerical values.
Plasma and urine concentrations of choline, betaine, and dimethylglycine in addition to plasma SAM, and SAH were measured using UPLC-MS/MS technique and established methods [18
]. Choline, betaine, and dimethylglycine were separated on an Acquity UPLC BEH HILIC column (100 mm × 2.1 mm (i.d.); 1.7 µm particle size) with an Acquity HILIC VanGuard pre-column (5 mm × 2.1 mm (i.d.); 1.7 µm particle size) and a 0.2 µm in-line filter (Waters Corporation). The column temperature was 30 °C and the flow rate was 0.6 mL/min. The solvent A and B were 15 mmol/L ammonium formate (pH 3.5) and acetonitrile, respectively. d9
-Betaine chloride, d9
-choline chloride (Isotec, Sigma-Aldrich, Munich, Germany), and d6
-dimethylglycine HCl (CDN isotopes, Simport, QC, Canada) were used as internal standards. Plasma, or urine (diluted in H2
O 1:5) samples were added to the internal standard mix in acetonitrile. After protein precipitation and vortexing, the samples were centrifuged and 1 µL was injected into the UPLC-MS/MS.
Concentrations of SAH and SAM were measured using labelled isotopes as internal standards (13C5-SAH and d3-SAM). Sample cleanup was conducted using solid-phase extraction (SPE) columns containing phenylboronic acid (Varian Bond Elut PBA columns, Varian Inc., Palo Alto, CA, USA). Aqueous acetic acid, pH 2.636 was used to elute the analytes. The chromatography separation was conducted using ACQUITY UPLC BEH C18 Column, 130 Å, 1.7 µm, 2.1 mm × 50 mm and ACQUITY UPLC BEH C18 VanGuard Pre-column, 130 Å, 1.7 µm, 2.1 mm × 5 mm. The column temperature was maintained at 30 °C, with a mobile phase flow rate of 0.35 mL/min. The mobile phase consisted of 100% aqueous acetic acid (glacial), pH 2.636. The injected volume was 10 µL.
EDTA whole blood HbA1c concentrations were measured using an automated VARIANT™ II TURBO system (Bio-Rad Laboratories, Munich, Germany). Plasma and urinary creatinine and plasma glucose, lipids (high density lipoprotein, low density lipoprotein, triglyceride and total cholesterol), and liver markers such as aspartate-amino-transferase (AST) and alanine-amino-transferase (ALT) were measured using automated methods (COBAS INTEGRA System, Roche Diagnostics, Germany) at the central laboratory of the Saarland University Hospital. Plasma lipids were measured only before and after OFTT (in the first visit).
2.5. Statistical Analyses
Sample size calculation was based on potential changes in plasma concentrations of TMAO (pre-post levels compared with paired t
-test). We considered a change (increase or decrease) of approximately 50% from baseline (approximately 2 µmol/L change compared with zero change) as metabolically relevant. Assuming a standard deviation of the change of (approximately 2 µmol/L), a power of 0.95, and α = 0.05, we estimated a sample size of 13 participants undergoing each test would be sufficient. Changes in betaine, choline, dimethylglycine, and SAM were not used for sample size estimation because the extent of the change could not be postulated based on previous studies. Stratification according to sex was not initially planned because we found no differences in TMAO between men and women in an earlier study [2
]. However, posthoc explorative analyses of the results according to sex are shown in Supplemental Tables S4 and S5
Continuous variables were tested for normal distribution using Kolmogorov-Smirnov test and Lilliefors Significance Correction. The following variables showed skewed distribution: TMAO before OFTT and dimethylglycine before OGTT. TMAO values (both before and after OFTT) were log-transformed and were both found to be normally distributed after the transformation. Log-transformed TMAO was used for paired comparisons (before and after OFTT). The log-transformed dimethylglycine before OGTT was not normally distributed and the non-transformed values were compared. Results of continuous variables and changes over time are shown as mean ± SD (standard deviation). The within-individual differences in concentrations of the biomarkers (pre versus post) were analyzed by using paired-t-test. Changes of the choline metabolites were calculated as; deltaOFTT after OFTT = post-OFTT levels − pre-OFTT levels and deltaOGTT after OGTT = post-OGTT levels − pre-OGTT levels. DeltaOFTT% is the post-OFTT levels − pre-OFTT levels*100/pre-OFTT levels and deltaOGTT % is the post-OGTT levels − pre-OGTT levels × 100/pre-OGTT levels. Paired-t-test was used to compare changes after OFTT (deltaOFTT) with those after OGTT (deltaOGTT). Fasting concentrations of the metabolites before OFTT were compared with those before OGTT (measured one week later) by using paired-t-test.
One-way analysis of variance (ANOVA) test was used to compare all continuous variables between men and women. Stepwise multiple linear regression analyses were applied to study the predictors of plasma TMAO, glucose, and triglycerides after OFTT and plasma TMAO and glucose after OGTT.
Statistical analyses were conducted using IBM SPSS 24.0 (IBM Corp.; Armonk, NY, USA). p values below 0.05 were considered statistically significant and those between 0.05 and 0.10 were interpreted as showing a tendency.
We reported significant within-individual changes in plasma concentrations of choline metabolites after OFTT and OGTT in healthy individuals (Supplemental Figure S5
). Changes of plasma choline and TMAO from baseline were in opposite directions; plasma choline declined and TMAO increased after OFTT, while choline increased and TMAO declined after OGTT. Men and women had different betaine and choline concentrations, but the changes of choline metabolites after the tests were not related to baseline concentrations and not explained by sex. These results are likely to represent a metabolic adaptation to acute ingestion of glucose or fat (postprandial hypertriglyceridemia).
Chronic consumption of fats and/or glucose is associated with multi-components of the metabolic syndrome such as, diabetes, abdominal obesity, and hypercholesterolemia [20
]. Healthy none-obese subjects responded to OGTT by sparing choline and betaine as methyl donors (increased urinary excretion of betaine and choline and increased plasma choline), maintaining glucose, while plasma concentrations of dimethylglycine and SAM declined. Therefore, OGTT reduced the flow of methyl groups via BHMT that transfers a methyl group to produce dimethylglycine and methionine. Urinary excretion of betaine and choline were increased after OGTT, thus supporting the downregulation of the methyl flow through the BHMT pathway. Downregulation of methyl groups has been reported after experimental exposure to high concentrations of glucose [21
]. Chen et al. have shown that intracellular concentrations of choline are elevated and those of dimethylglycine are lowered in neural PC12 cells exposed to high glucose media [21
]. Dimethylglycine is metabolized to glycine. Plasma levels of dimethylglycine and glycine show an inverse correlation with fasting glycemia and insulin resistance in human studies [22
]. Ho et al. performed metabolic profiling in 377 subjects without diabetes before and after OGTT [24
]. Plasma levels of methionine (−1.44 fold), and glycine (−1.15 fold) declined after OGTT [24
]. As in our study, dimethylglycine declined after OGTT (−1.06 fold) [24
]. Combined with our results, lowering methionine, dimethylglycine, glycine, adenosine mono- and di-phosphate and several purines after OGTT in the study of Ho et al.; strongly suggests that the methylation pathway is down regulated in healthy people under glucose challenge [24
]. It is possible that glycine (and formate) pathway links methyl groups and glucose oxidation by regulating mitochondrial flavin-adenine dinucleotide (FAD) cofactors. FMO3 (UniProt P31513) interacts with glucose metabolism [25
]. Flavin mononucleotide is down regulated after OGTT [24
] which may lead to a transient reduction in FMO3 activity and may explain our results on lowering TMAO after OGTT in healthy subjects. In contrast, TMAO is elevated in manifested diabetes which could reflect failure in controlling blood glucose, adaptation to chronically elevated blood glucose, or enhanced FMO3 expression in adipose tissue of patients with diabetes [8
Concentrations of dimethylglycine and SAM remained unchanged 4 h after OFTT, suggesting that betaine (declined after OFTT) had been utilized to maintain sufficient methyl groups as suggested by unchanged dimethylglycine and SAM in plasma. After OFTT, plasma choline declined, while triglycerides increased. Postprandial hypertriglyceridemia may enhance choline conversion into phosphatidylcholine that is required to solubilize chylomicrons and very low density lipoprotein-(VLDL) particles. An SAM-dependent phosphatidylethanolamine methyl transferase (PEMT) is an important source of phosphatidylcholine in the liver. PEMT regulates lipid homeostasis under high fat diet [26
] and it could be theoretically activated under a high fat diet, thus leading to higher utilization of betaine and choline as methyl donors.
Eating egg yolk, fish, or beef is associated with an increase in plasma TMAO in healthy subjects [11
]. Fruits caused a reduction in plasma TMAO by 0.4 µmol/L. According to our results, this could be related to a high glycemic index [11
]. Divergent changes of plasma TMAO have been previously reported among healthy volunteers (TMAO-producers and non-producers) [11
]. Although, the overall effect in our study was a trend towards an increase in plasma TMAO after OFTT, however, TMAO declined after OFTT in 9 out of 18 participants, while it increased in 9 participants. These opposite changes remained unexplained, but might be due to differences in gut bacteria [11
]. In contrast, the increase of TMAO after OGTT occurred in 15 out of 18 subjects. Suppression of FMO3 expression in animals caused weight reduction, especially due to lowering white adipose tissue weight and increasing lean mass [8
]. Moreover, administration of TMAO to mice fed a high fat diet improved glucose tolerance [29
]. In our study, the diverse response of TMAO after OFTT could be related to lean body mass that is not reflected by BMI or waist-to-hip ratio.
Long term effects of dietary patterns and controlled dietary intakes of fat and sugars on choline metabolism are not well studied. A positive association between dietary fat intake and plasma betaine has been reported in a study of middle aged and elderly men and women [30
]. In young none-obese men consuming a high-fat diet (55% of the calories as fat) for 5 days, intervention with a high fat meal (63% of the calories as fat) caused an increase in plasma TMAO [31
]. Boutagy et al. reported an increase in plasma TMAO [31
] similar to that found in the present study after OFTT (93% of the calories as fat). Additionally, an increase in plasma TMAO after 9 months of lifestyle modifications has been related to improved insulin sensitivity and lipid profile in an interventional study [7
]. A raise in plasma TMAO after OFTT in 9 participants could reflect a high fat dietary pattern while, a decline in TMAO after OFTT in the other 9 participants could reflect a dietary pattern that is rather low in fat. The influence of the diet was not verified in the present study.
Alternative explanations of the current results deserve consideration. It is possible that OGTT and OFTT cause differential alterations in nutrient transport through the gut. Changes in plasma metabolites could be due to excretion of choline and TMAO in the bile and the re-absorption of these compounds or could depend on the osmotic pressure in the gut or differ between fat and sugar milieus. Since urine samples were collected 2 h after OGTT and 4 h after OFTT, we could have missed the time window for collecting urine samples that show significant changes. Similarly, we could have missed an optimal time windows (maximal exposure effect) to collect blood samples. This issue could have influenced the comparisons between concentrations of the biomarkers 2 h after OGTT versus those of 4 h after OFTT. However, this explanation is unlikely because our results after OGTT and OFTT are in agreement with previous studies [24
]. Limitations of the present study include lack of information on dietary intake or stool microbiome pattern and the lack of standardization of the meals before the tests and the water drinking volume during the test. Also, the limited number of participants could have hided some effects. A post-hoc sample size calculation based on the observed values showed that 33 and 13 subjects would be sufficient to show changes in TMAO after OFTT and OGTT, respectively (Supplemental Table S6
In summary, the study has shown that OGTT caused a reduction in plasma TMAO and the methylation markers, dimethylglycine and SAM, increased plasma choline, and caused no change in plasma betaine. In contrast, OFTT caused heterogeneous changes in plasma TMAO, caused no change in plasma methylation markers, lowered plasma levels of choline, betaine, and glucose, and increased postprandial triglycerides.