Measurement of plasma choline in acute coronary syndrome: importance of suitable sampling conditions for this assay

Blood choline has been proposed as a predictor of acute coronary syndrome (ACS), however different testing procedures might affect the choline concentration because the lysophospholipase D activity of autotaxin (ATX) can convert lysophosphatidylcholine to lysophosphatidic acid (LPA) and choline in human blood. Although the influences of ATX on LPA levels are well known in vivo and in vitro, those on choline have not been elucidated. Therefore, we established suitable sampling conditions and evaluated the usefulness of plasma choline concentrations as a biomarker for ACS. Serum LPA and choline concentrations dramatically increased after incubation depending on the presence of ATX, while their concentrations in plasma under several conditions were differently modulated. Plasma choline levels in genetically modified mice and healthy human subjects, however, were not influenced by the ATX level in vivo, while the plasma LPA concentrations were associated with ATX. With strict sample preparation, the plasma choline levels did not increase, but actually decreased in ACS patients. Our study revealed that ATX increased the choline concentrations after blood sampling but was not correlated with the choline concentrations in vivo; therefore, strict sample preparation will be necessary to investigate the possible use of choline as a biomarker.

. Variations in serum choline and LPA levels after incubation. ATX antigen levels were measured in normal human serum (ATX (+)) and ATX-depleted serum (ATX (−)) prepared using biotin-anti-ATX antibody and streptavidin-bound magnetic beads (a). Normal human serum (+)(circle) and ATX-depleted serum (−)(square) were incubated at 37 °C for 1 to 4 hours, and the levels of choline (red) and LPA (blue) were determined (b). Results are given as the mean ± SD (n = 2).
SCiEnTiFiC RepORtS | (2018) 8:4725 | DOI: 10.1038/s41598-018-23009-x fluoride (NaF), or heparin and incubated at 37 °C or 4 °C for 1 to 3 hours; the plasma choline and LPA levels were then compared among the plasma samples. When whole blood plus heparin was incubated at 37 °C for 3 hours, both the plasma LPA and choline concentrations were increased by 2.1 ± 0.2 μmol/L and 3.4 ± 0.7 μmol/L (Fig. 2a,b), respectively. Regarding whole blood in the presence of EDTA-2K, the levels of plasma choline were increased up to 1.6 ± 0.2 μmol/L after 3 hours of incubation, while the plasma LPA concentrations were increased by only 0.3 ± 0.0 μmol/L (Fig. 2a,b). Moreover, when whole blood was prepared with EDTA + NaF, the plasma choline levels were increased by 5.8 ± 0.1 μmol/L, while the plasma LPA concentrations were decreased by 0.5 ± 0.1 μmol/L (Fig. 2a,b). The plasma choline and LPA concentrations were not changed when each whole blood sample was incubated at 4 °C. Furthermore, to investigate whether excessive choline is regulated by blood cells, 80 μmol/L of choline chloride was added to each whole blood sample before incubation. Consequently, the plasma choline levels in whole blood with heparin, EDTA-2K, and EDTA-2K + NaF were respectively decreased by 13.5 ± 5.2 μmol/L, 13.5 ± 5.4 μmol/L, and 8.4 ± 2.1 μmol/L after incubation at 37 °C for 3 hours, whereas these decreases were suppressed when incubated at 4 °C (Fig. 2c). Since the choline level was increased regardless of the tube type when the samples were not kept at 4 °C, blood samples should be immediately placed on ice until the preparation of the plasma sample by centrifugation to measure choline levels in healthy subjects and patients, as described below.
Levels of plasma choline and LPA in conditional ATX knock-out or overexpression mice. Since choline and LPA, both of which are mainly produced by ATX from LPC, were not equally regulated, we investigated the effect of the ATX levels on the choline and LPA levels in vivo using ATX knock-out or overexpression mice. As indicated in Fig. 3a, the ATX activity tended to be reduced depending on each mouse genotype (lox/+ Figure 2. Variations in plasma choline and LPA levels obtained from whole blood with the addition of various anticoagulants. Whole blood obtained from healthy human subjects was mixed with heparin (blue), EDTA-2K (EDTA)(green), or EDTA-2K and NaF (EDTA + NaF)(red), then incubated at 37 °C (closed mark) or 4 °C (open mark) for 1 to 4 hours. After the plasma samples were collected by centrifugation at 4 °C, the concentrations of LPA (a) and choline (b) were measured in each plasma sample. (c) Eighty-micromolar choline chloride was added to whole blood with heparin, EDTA-2K (EDTA), or EDTA-2K and NaF (EDTA + NaF), then incubated at 37 °C or 4 °C for 1 to 4 hours. After the plasma samples were collected by centrifugation at 4 °C, the concentrations of choline were measured in each plasma sample. Results are given as the mean ± SD (n = 2).
Correlation of plasma choline levels with ATX-related metabolites and distribution of plasma choline concentrations in healthy subjects. We next compared the correlation of plasma choline with serum ATX antigen or plasma LPA concentrations in healthy human samples (n = 98). The plasma choline concentrations were weakly but significantly correlated negatively with both the plasma LPA (r = −0.232, P = 0.002) and serum ATX antigen levels (r = −0.235, P = 0.020) (Fig. 4a,b), while the plasma LPA level showed a significantly positive correlation with the serum ATX antigen level (r = 0.524, P < 0.001) (Fig. 4c). Regarding LPC, a precursor of LPA, the plasma choline concentration was correlated with LPC (r = 0.258, P = 0.010) (Fig. 4d).
In healthy subjects (n = 98), the mean ± SD of the plasma choline concentration was 9.6 ± 3.5 μmol/L. Plasma choline was significantly higher in men (10.3 ± 3.8 μmol/L) than in women (8.1 ± 2.2 μmol/L) (P < 0.001), while we previously showed that the plasma LPA and ATX antigen levels were higher among women than among men 26 . Table 1 shows the correlation between plasma choline and other parameters. The plasma choline concentrations were significantly and positively correlated with age, body mass index (BMI), LPC, urea nitrogen, creatinine, uric acid, aspartate aminotransferase, alkaline phosphatase, gamma-glutamyl transferase, hemoglobin, hematocrit, sodium and potassium and were significantly and negatively correlated with ATX activity and antigen and chloride. Choline and LPA levels under fasting and fed states. We next investigated the effect of meal consumption on the levels of choline and LPA. To analyze the modulation of choline and LPA by meal intake, we obtained blood samples serially before and at 30 min after each meal in 5 healthy subjects, who consumed 3 ordinary meals per day, and compared the dynamism of the plasma choline and LPA levels. As shown in Fig. 5, the plasma choline concentrations were significantly higher after first food intake (P = 0.042) and tended to be increased after the second (P = 0.080) and third (P = 0.068) food intakes, while no variation in the LPA levels was observed.
Association of plasma choline concentration with cardiovascular events. We investigated the association between the plasma choline level and CAD. The basic characteristics of the three groups (normal coronary arteries (NCA), stable angina pectoris (SAP) and ACS) according to diagnosis were shown in a previous study 27 . The plasma choline level was significantly higher in men than in women (11.5 ± 3.3 vs. 9.0 ± 3.0 μmol/L, P < 0.001) as was the case in healthy subjects. In men, the plasma choline concentration was significantly decreased in the group with ACS (10.1 ± 2.6 μmol/L), compared with the NCA (13.3 ± 3.7 μmol/L, P = 0.003) and SAP (11.6 ± 3.1 μmol/L, P = 0.032) groups (Fig. 6), whereas no differences among the groups were observed in women. Therefore, we analyzed the association between the plasma choline levels and ACS-related factors in further detail in a men-only group. When all the male patients who had undergone coronary angiography (n = 110) were divided into four equal groups based on their plasma choline values, trends toward significant associations between an increase in the choline quartile and the red blood cell count, high sensitive C-reactive protein (hsCRP) level, and number of ACS subjects were observed (Table 2). Although, the troponin T levels in ACS group were higher than those in NCA and SAP groups as we described previously 27 , the significant correlation between plasma choline and troponin T was not observed. Finally, to further confirm the possible usefulness of plasma choline as a biomarker for the diagnosis of ACS, we constructed a logistic regression analysis examining male subjects only. When explanatory variables (age, BMI, current smoking, diabetes, red blood cell count, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglyceride, hsCRP, choline quartile and LPA tertile) were introduced into a stepwise multivariate model, only hsCRP (P = 0.047), choline quartile (P = 0.002) and LPA tertile (P = 0.004) were selected as significant predictors of ACS after adjustment (Table 3).

Discussion
ATX is an important enzyme determining LPA level in circulation and theoretically produces equal molar amounts of choline. However, under the condition where whole blood was incubated, these two substances showed different kinetics; the present study showed that the serum LPA and choline concentrations dramatically increased after incubation depending on the presence of ATX (Fig. 1), while the plasma LPA and choline concentrations were differently modulated when the samples were incubated under several conditions; in whole blood plus EDTA-2K, the increase in the plasma LPA level was suppressed by the inactivation of ATX with a chelating agent, while no remarkable modulation of the plasma choline levels was observed. In relation to the variations in the plasma choline level, the degree of increase in plasma obtained from whole blood prepared with  The plasma choline (red) and LPA (blue) levels before and after meals (n = 5) were determined. The data show the percentage change from the value measured before the first meal (mean ± SD). The arrows indicate food intake. *P < 0.05, differences between before and after meals. heparin was higher than that in samples prepared with EDTA. These differences in increased plasma choline levels between EDTA and heparin are consistent with those of previous studies 19,28 . The difference in the choline kinetics between serum and whole blood indicates the presence of other factors from blood cells. Regarding the catabolism of choline, excessively added choline decreased in a time-dependent manner in whole blood samples (Fig. 2c). Although our study did not elucidate the exact underlying mechanism, considering that erythrocytes possess a choline influx-transporter, the choline might have been absorbed by the erythrocytes. Erythrocytes also have a choline transporter that regulates both the influx and efflux of choline 29 , and this choline transporter was inhibited by the blockage of Na+/K+ATPase (EC 3.6.1.3) 30 . This speculation was concordant with the results shown in Fig. 2c; the influx of choline into blood cells might be suppressed by NaF, which inhibits enolase resulting in a rapid decrease in the erythrocyte ATP concentration. In addition, the correlation between plasma choline levels and hemoglobin concentrations and hematocrit levels as well as the gender difference (higher in men) in healthy human subjects suggests the regulation of plasma choline by erythrocytes in vivo. Regarding the association between the plasma choline levels and ATX in vivo, while ATX was at least partly involved in the homeostasis of choline in the in vitro experiments, the results from the animal experiments with genetically modified mice revealed that the regulation of the circulating choline level was not influenced by the ATX level in vivo.
Concordant with these results, in healthy human subjects, the plasma choline level was not correlated with ATX, while the plasma LPA level was positively correlated with ATX. Taken together, our results suggest that both ATX-dependent and ATX-independent pathways are involved in the regulation of plasma choline levels, with the later pathway preferred in vivo. However, the former pathway cannot be ignored when performing laboratory medicine tests, since the choline level changes according to the tube type and handling procedure in the presence of ATX as well as the transporters on blood cells in vitro. Therefore, samples should be kept at a low temperature at all times during processing, from blood collection until measurement. To our knowledge, this is the first study to reveal the impact of ATX on the plasma choline level.
Regarding the timing of blood sampling, it might be desirable to consider the influence of meals. Choline is abundant in foods, and the dietary intake of choline is necessary to obtain materials for cell components and others. Actually, choline deficiency causes various problems including hepatic steatosis 31 , muscle damage 32 , and lymphocyte apoptosis 33 . We investigated the influence of meals on the plasma choline and LPA levels and the diurnal variations of these concentrations. Our data indicated that blood collection should ideally be performed before meals to avoid transient increases in choline.
From analytical and preanalytical aspects, our study revealed that the measurement of serum choline levels is not appropriate because of its in vitro production by ATX. Considering the variations in plasma and serum choline levels, measuring whole-blood choline levels seems to be more suitable for reflecting the activity of released phospholipase. However, a mass spectrometry assay, which has been performed for the assessment of whole-blood choline levels in many studies, is difficult to introduce into laboratory testing because of the initial cost and reproducibility concerns. Therefore, the measurement of plasma choline using samples prepared with anti-coagulants, such as EDTA, on ice using an automatable enzymatic assay would be optimal for the clinical introduction of choline as a biomarker for human diseases.
Regarding the clinical use of choline as a biomarker for ACS, the fact that the activations of phospholipase A 2 (EC 3.1.1.4) and phospholipase D 15 in ischemic heart tissue lead to the release of choline into plasma provides a plausible explanation for why the increased choline level in plasma is similar to that in whole blood. Actually, regarding the use of plasma choline as a predictive marker, some studies have demonstrated the predictive value of increased plasma choline levels in patients with ACS 13,15,16 . However, Body et al. suggested that the plasma choline levels did not help to predict a diagnosis of acute myocardial infarction (AMI) (odds ratio, 1.00; 95% confidence interval, 0.91-1.10; P = 0.98) 12 . Unexpectedly, our study also showed that the plasma choline concentrations in the ACS group were rather lower than those in the non-ACS group, although the difference was not large.  kept on ice prior to centrifugation. The reason why the plasma choline levels were lower in our study than in other studies is likely due to the difference in the sample procedure before analysis. Regardless, even with the rigorous sample preparation protocol used in this study, the plasma choline levels were not increased, but rather decreased in the subjects with ACS, although blood sampling was not always performed under fasting conditions in the present study.
In summary, the present study revealed that although ATX did not affect choline levels in vivo, ATX could increase choline levels in vitro, suggesting the need for strict sample preparation. We also observed that the plasma choline levels were decreased in plasma samples suitably prepared from patients with ACS. Further studies to investigate the possible usefulness of the plasma choline level using this assay and optimized procedures are expected to be performed.

Materials and Methods
Ethics statement and guidelines compliance. All human samples were collected with informed and written consent. The Ethics Review Committee at Juntendo University Hospital approved the study (number: 291). This study was also approved by the institutional review boards of both the University of Tokyo (number: 2602 and 11158). and Juntendo University School of Medicine. All the animal experiments were conducted in accordance with the guidelines and regulations for Animal Care and were approved by the animal committee of Tohoku University (2013PhA-034). All methods were performed in accordance with the relevant guidelines and regulations.

Samples.
The samples used to examine the basic profiles of choline were collected from the antecubital vein of healthy adult volunteers. The serum and plasma samples used to determine the reference ranges for choline were residual samples from those obtained for laboratory analyses (for medical checkups). These healthy subjects (n = 98) were defined as those who did not receive any medical treatment at the sampling. To prepare the plasma samples, the whole blood samples were collected into vacuum tubes containing 3 mg/mL of EDTA-2K and then mixed with 10% v/v of citrate-theophylline-adenosine-dipyridamole (CTAD) 36 (BD Biosciences, Tokyo, Japan), as previously described 23 . The tubes were immediately placed in an ice/water bath, followed by 15 min of incubation. The anticoagulated samples were centrifuged at 2500 × g for 30 min to separate the plasma. As soon as centrifugation was completed, the plasma supernatant was carefully collected to avoid contamination with cell components. To obtain the serum samples, whole blood specimens were directly placed into a glass tube and left for 15 min at room temperature to allow blood clot formation; then, the serum was separated by centrifugation at 1500 × g for 5 min. The samples from patients who had undergone coronary angiography were obtained as previously described 27 . Briefly, subjects who underwent coronary angiography at Juntendo University Hospital (J-Bacchus trial) between July and December 2009 were enrolled according to the entry criteria (no previous examination by coronary angiography, no history of coronary intervention or coronary artery bypass grafting, and having precisely evaluable coronary trees). This is a prospective cross-sectional study registered in the UMIN protocol registration system (#UMIN000002103). One hundred fifty eight patients were initially screened for this study and excluded as follows: 1) maintenance dialysis (n = 5), 2) diabetes treated with insulin (n = 4), and 3) acute or chronic infectious (n = 5) or 4) neoplastic (n = 3) diseases. Patients without significant stenosis were placed in NCA group (n = 32; male/female, 20/12), whereas those with significant stenosis were placed in an ACS group (n = 38; male/female, 29/9) or SAP group (n = 71; male/female, 61/10). Patients with AMI and unstable angina were included in the ACS group. Blood samples were directly collected into glass vacutainer tubes with or without EDTA disodium salt (EDTA-2Na) to obtain plasma and serum, respectively. The samples were immediately placed on ice. The anticoagulated samples were centrifuged at 1000 × g for 10 min, and then the plasma was carefully collected. Whole blood samples collected without EDTA-2Na were left to clot, and the serum was separated by centrifugation at 1000 × g for 10 min. All the samples were either subjected to measurement immediately or stored at −80 °C until measurement. All samples were analyzed in the University of Tokyo Hospital.
Preparation of ATX-depleted serum. ATX was depleted from human serum using ATX-binding beads, as previously described 37 . Ten micrograms of biotin-anti-ATX mAb (prepared from a rat anti-ATX antibody named R10.23 38 ) solution and 80 μL of Dynabeads M-280 Streptavidin suspension (Life Technologies Corp., Carlsbad, CA) were added to 500 μL of TBST for 15 min, followed by washing with TBST twice while keeping the tube on a magnet for the collecting beads. After 300 μL of serum was shaken on the magnet for 15 min, the supernatant was used as ATX-depleted serum.  Serum ATX antigen measurement. Anti-human ATX monoclonal antibodies were produced by immunization with recombinant human ATX expressed in a baculovirus system. An automated immunoassay for the quantitative determination of ATX was then established, and human serum samples were assayed using an automated immunoassay analyzer AIA-system (TOSOH Corp., Tokyo, Japan), as previously described 38 .
Choline determination using an enzymatic assay. Serum and plasma choline concentrations were determined using a modified enzymatic assay, as previously reported 39 . A choline molecule catalyzed by choline oxidase produces two hydrogen peroxides. In the presence of peroxidase, 4-AA and Trinder's reagent are oxidized by the hydrogen peroxides to generate quinoneimine dye. The choline level is then determined by measuring the increased absorbance at 546 nm. Actually, 10 μL of sample was added to 180 μL of Reagent 1, containing 4-AA and peroxidase. After the incubation of this mixture for 5 min at 37 °C, 60 μL of Reagent 2, containing choline oxidase and TOOS, was added followed by another 5 min of incubation. The absorbance at 546 nm was measured at 5 and 10 min after the start of the reaction. The actual handling procedures were performed using a biochemical automatic analyzer (Hitachi 7600 [Hitachi, Tokyo, Japan] or JCA-BM8040 [JEOL, Tokyo, Japan]).

Measurement of phospholipids and other analytes.
The concentrations of sphingomyelin in the serum and of LPC and LPA in the plasma were determined using a specific enzymatic assay with sphingomyelinase 40 and lysophospholipase 41,42 , as previously described. The total phospholipids (choline-containing phospholipids) in the samples were measured using a commercial kit (Alfresa Pharma Corp., Osaka, Japan), following the manufacturer's protocol. We determined the serum phosphatidylcholine concentration by subtracting the sphingomyelin and LPC values from the choline-containing phospholipids, as described previously 37,40 . These phospholipid measurements were performed using an automatic analyzer (Hitachi 7600 [Hitachi, Tokyo, Japan] or JCA-BM8040 [JEOL, Tokyo, Japan]). Other ordinary markers were determined using routine laboratory methods.
Genetically modified mice. ATX knock-out mice were prepared using the Cre-loxP system, as previously described 25 . Exons 6 and 7 in the Enpp2 gene, the active center of ATX, are flanked by loxP sites. Using this system, we prepared mice with various genetic backgrounds (wild type, lox/+, loxP/loxP, +/−, loxP/−). ATX transgenic mice (cre/+tg/+) were produced using a previously reported strategy 44 .

Statistical analysis.
The results were expressed as the mean ± SD. All data were statistically analyzed using SPSS ver. 20.0 (Chicago, IL). The Mann-Whitney test or one-way ANOVA followed by a Kruskal-Wallis post-hoc test was used to compare the choline, LPA and ATX levels between genetically modified mice and the choline levels among the NCA, SAP and ACS groups. The Spearman rank correlation test was employed to determine the significance of the correlation among various test items in healthy human subjects. A post-hoc Wilcoxon test after a Friedman test was performed to compare the choline and LPA concentrations under fasting and fed states. Demographic and clinical variables and the number of ACS patients were compared across quartiles of choline levels using the Jonckheere-Terpstra trend test and the Cochran-Armitage Test, respectively. The independent effect of the biomarkers on the risk of ACS adjusting for potential confounders was determined using a multiple logistic regression analysis. The following variables were initially incorporated into the univariable model: age, BMI, current smoking, diabetes, red blood cell count, HDL-C, LDL-C, triglyceride, hsCRP, choline quartile, and LPA tertile. Statistically significant variables in the multivariable regression analysis selected using a forward stepwise procedure were subsequently included in a new model. A P < 0.05 was considered statistically significant.
Data Availability Statement. The datasets generated or analyzed during the current study are available from the corresponding author on reasonable request.