BACKGROUND: Macrophages and related cells are important cellular mediators of the innate immune system and play important roles in wound healing and fibrosis. Flux through different l-arginine metabolic pathways partially defines the functional behavior of macrophages. Methods to measure metabolites within the nitric oxide synthase/arginase pathways could provide insights into local and systemic inflammatory processes.
METHODS: A targeted metabolomics approach was developed by using hydrophilic-interaction liquid chromatography and electrospray ionization–tandem mass spectrometry to simultaneously measure l-arginine, asymmetric dimethylarginine, symmetric dimethylarginine, l-citrulline, l-ornithine, and l-proline in plasma from humans and mice.
RESULTS: All analytes were quantifiable in human and mouse plasma with a small volume (25 μL), minimal sample preparation, and no derivatization. Patients with high plasma concentrations of C-reactive protein and mice with acute inflammation induced by lipopolysaccharide had significant reductions of arginine metabolites in plasma compared with controls.
CONCLUSIONS: This new assay uses plasma metabolomic measurements to help provide new insights into metabolic changes coupled to the innate immune response. We identified significant changes in arginine metabolism in both humans and mice following an inflammatory stimulus. These changes were associated with decreased plasma arginine metabolite concentrations and increased methylated arginine concentrations.
Macrophages play an essential role in host defense and are the major cells involved in the innate immune response. Macrophage function is impaired in many diseases, including, but not limited to, atherosclerosis, obesity, cancer, multiple sclerosis, and Alzheimer disease. The consequences of impaired macrophage function can have large effects on the ability to recover appropriately from injury (1, 2). During acute and chronic inflammatory conditions macrophages use large quantities of the semiessential amino acid l-arginine (arginine) as a substrate for 2 separate enzymes, inducible nitric oxide synthase (iNOS)7 and arginase 1 (Arg1) (see Fig. 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol57/issue5). The appropriate resolution of the inflammatory response depends on the macrophage's ability to balance the activation of iNOS-mediated classical macrophage (M1) activation pathways that induce cell death and arginase-mediated alternative macrophage (M2) activation pathways that typically promote growth and repair (3, 4). When macrophages cease to return to homeostasis and iNOS activation is poorly controlled, this results in a compromised state of nonresolving inflammation (1).
Arginine acts as a central node for regulating macrophage activation pathways (5). Under normal physiological conditions, mammals obtain arginine from dietary sources or endogenous synthesis in the kidney. In contrast, arginine utilization changes dramatically during initiation of the innate immune response upon the activation of M1 and M2 pathways. The mechanisms underlying arginine metabolism are controlled by multiple enzymes that produce several additional metabolites that also play fundamental roles in innate immunity (3, 6). Large numbers of circulating monocytes in the peripheral bloodstream are under continual immune surveillance during host defense (2); thus, plasma and serum are ideal and easily accessible biological fluids to assess monocyte and macrophage activation pathways that affect arginine metabolism.
One group of metabolites is the methylated arginines, asymmetric dimethylarginine (ADMA) and symmetric dimethylarginine (SDMA), which are generated by the activity of protein methyltransferases (PRMTs), PRMT1 and PRMT2, respectively. ADMA is an endogenous inhibitor of all NOS enzymes, including endothelial NOS (eNOS) and neuronal NOS (nNOS), in addition to iNOS. ADMA concentrations are increased in many diseases that are characterized by vascular dysfunction, including atherosclerosis, diabetes mellitus, and renal failure. SDMA, in contrast, is a competitive inhibitor of cellular arginine transport; increased SDMA levels are associated with renal dysfunction (7). Citrulline, ornithine, and proline also play critical roles in arginine metabolism. Citrulline is a primary end product of NOS activity along with nitric oxide, and it is also generated from ADMA via dimethylarginine diaminohydrolase activity. In contrast, ornithine is the primary product of arginase activity. Although ornithine serves a principal role in the urea cycle, it also serves as an important precursor for several molecules involved in growth and repair, including both polyamines and proline. Proline is essential for the production of many structural molecules such as collagen that are important in rebuilding the extracellular matrix and cell membranes (3, 8, 9).
To better elucidate the mechanisms underlying arginine metabolism, we set out to develop a targeted metabolomics method that would simultaneously quantify 6 arginine metabolites in plasma with well-established roles in the arginine/iNOS/Arg1 pathway (see online Supplemental Fig. 2). We used hydrophilic-interaction liquid chromatography (HILIC) with liquid chromatography–tandem mass spectrometry (LC-MS/MS) to demonstrate that the arginine metabolome undergoes dynamic fluctuations in response to systemic inflammatory stimuli in humans and mice. The method is amenable to basic and clinical research laboratories as a tool for detecting alterations in the innate inflammatory response that influence the clinical outcomes of illness and injury.
Estradiol-17β (E2), sesame oil, lipopolysaccharide (LPS) from Escherichia coli 055:B5, l-arginine, l-citrulline, l-ornithine monohydrochloride, l-proline, SDMA [NG, NG-dimethyl-l-arginineinine di(p-hydroxyazobenzene-p'-sulfonate) salt], and BSA were obtained from Sigma-Aldrich. ADMA (NG,NG-dimethylarginine dihydrocloride) was purchased from Calbiochem. An l-arginine internal standard, U-13C6 l-arginine:HCl, was purchased from Cambridge Isotopes Laboratories. All other chromatography-grade chemicals were obtained from Fisher Scientific.
The use of samples from human patients was approved by the Human Subjects Division of the University of Washington. Concentrations of C-reactive protein (CRP) in heparinized plasma samples were measured on a Synchron DxC clinical immunoassay instrument (Beckman Coulter).
All experiments were approved by the University of Washington Institutional Animal Care and Use Committee. Female C57BL/6J mice were bred in University of Washington vivaria. At 9–12 weeks of age, mice were anesthetized with isoflurane, received a bilateral ovariectomy to remove endogenous ovarian hormones, and were simultaneously implanted with a Silastic capsule containing sesame oil vehicle. One week later, mice were randomly assigned to a sham group [(−)-LPS, n = 8] or a group that received an intraperitoneal injection of 5 mg/kg LPS [(+)-LPS, n = 7] as previously described (10). At 0 h (sham) or 6 h following injury, mice were anesthetized (ketamine 130 mg/kg and xylazine 8.8 mg/kg), and plasma and serum were obtained via cardiac puncture. C57BL/6J male mice (n = 4) were used in a separate experiment designed to ascertain differences in analyte concentrations between serum and plasma. Plasma was anticoagulated by addition of EDTA to a final concentration of 50 mmol/L. For serum, blood was collected into microtainer serum separator tubes (BD) and allowed to clot for 15 min. Plasma and serum samples were centrifuged at 16 000g for 10 min and supernatants were stored at −80 °C until sample preparation for LC-MS/MS.
PLASMA AND SERUM SAMPLE PREPARATION AND CALIBRATORS
Preparation of calibrators.
All analyte stock solutions were prepared in water/0.2% formic acid or methanol/0.1% formic acid and stored at −20 °C. A fresh 200-μmol/L solution of all analytes was used to prepare calibrators. All calibrators were prepared in 19.8 g/L BSA prepared in water and subjected to serial dilutions to prepare a standard curve ranging from 0.08 to 6.25 μmol/L (ADMA and SDMA) and 1.625 to 125 μmol/L (arginine, citrulline, ornithine, proline). To account for very low concentrations of certain analytes present in the BSA preparation, a BSA blank was run as a 0 calibrator with every set of calibrators. The endogenous analytes in the BSA blank were subtracted from all other calibrators and the slope of the calibration curve was used to calculate the concentration of all analytes.
A 25-μL volume of each sample or calibrator was spiked with 2.5 μmol/L internal standard (U-13C6 l-arginine) and incubated at room temperature for 10 min. Samples were extracted with 10 volumes of 0.1% formic acid in isopropanol and agitated in an Eppendorf Thermomixer at 1400 rpm for 10 min. Samples were centrifuged at 16000g and supernatants stored at −80 °C until LC-MS/MS analysis.
HPLC AND MS/MS
We performed separation of analytes by using an LC20AD HPLC system (Shimadzu) with 2 solvent pumps, autosampler, and column oven equipped with an Atlantis HILIC Column (3-μm bead size; 3 × 100 mm; Waters). Samples (5 μL) were resolved with a linear gradient over 1.25 min from 80% mobile phase A (75% acetonitrile/25% methanol/0.2% formic acid) to 45% mobile phase B (0.2% formic acid in water) at a flow rate of 0.2 mL/min and a column temperature of 30 °C. Analytes were ionized by use of electrospray ionization–MS/MS in positive ion mode. Molecular ions [M+H] were analyzed by using the multiple-reaction–monitoring mode of a 4000 QTrap LC-MS/MS (ABI Sciex) with a source temperature of 600 °C. The transitions monitored for each analyte (precursor ion/product-fragment ion in m/z) were: l-arginine, 175.1/43.0, 175.1/60.0, 175.1/70.0; l-arginine internal standard (U-13C6 l-arginine:HCl), 181.0/44.0, 181.0/61.1, 181.0/74.1; l-ornithine, 133.1/116.1, 133.1/70.0; l-citrulline, 176.0/113.0, 176.0/70.1; l-proline, 116.1/52.8, 116.1/42.0, 116.1/70.0; ADMA, 203.1/46.0; and SDMA, 203.2/172.1.
Plasma and serum extracts were kept at 8 °C in the autosampler. Calibrators were run in triplicate and samples in duplicate. Endogenous peak areas were normalized to the peak area of the internal standard and averaged across replicate injections. Normalized endogenous analyte areas were then converted to micromole per liter concentrations by using calibration curves made in BSA/water.
To assess interference due to hypertriglyceridemia, 2 human plasma samples were supplemented with Intralipid 20% (Fresenius Kabi) up to 10 g/L (11). Mixing studies of normal plasma samples with hemolyzed and uremic samples [creatinine >265 μmol/L (>3 mg/dL)] were performed at ratios of 2:1, 1:1, and 1:2. The recoveries presented represent the mean of the 3 mixes. Linear regression analysis was performed to detect any concentration dependence on the recoveries, which was never observed. Ion suppression was analyzed with a standard T-experiment (12) by infusing arginine internal standard (25 μmol/L, 10 μL/min) into the eluent from the chromatographic development of a prepared normal human plasma sample (see online Supplemental Fig. 5).
All data were analyzed by use of Microsoft Excel, Graphpad Prism 5.0 (Graphpad), and R statistical software (http://www.R-project.org). Calibration curves were constructed from the peak area ratios of the BSA calibrators. We analyzed differences between treatment groups using the Student t-test or Mann–Whitney test (2-tailed). Linear regression analysis was used to assess the significance of all associations of continuous variables. Normality of the data used for linear regression and of the residuals was measured by using the Shapiro-Wilks test. Analytes that were not normally distributed were log transformed before analysis. All data are presented as mean (SE) with P < 0.05 as significant.
We first developed an assay to measure arginine metabolism in mouse plasma and serum, and subsequently transitioned the assay to human plasma for clinical studies. Chromatography was optimized by using an HILIC column after C18, cyano, and pentafluorophenyl-propyl stationary phases failed to retain all of the target analytes. The optimal mobile phase contained acetonitrile and methanol and resolved the 6 analytes better than other mobile phases tested (acetonitrile, methanol, isopropanol, and other mixtures of the 3 solvents). Protein precipitation with isopropanol was selected as the sample preparation method because it generated the greatest signal-to-noise ratio and the best imprecision compared with other precipitation reagents (acetonitrile, methanol, mixtures of all 3 solvents, and reagents with MgSO4 added). All analytes were detectable in human plasma and mouse plasma or serum.
All analytes exhibited good chromatographic separation in human (Fig. 1) and mouse plasma (see online Supplemental Fig. 3). In previous studies, the importance of adequate chromatographic separation of ADMA and SDMA was stressed (13). Using the conditions described, we were able to separate the molecules in the mobile phase using 2 transitions that lacked crosstalk: 203.1/46.0 for ADMA and 203.1/172.1 for SDMA.
BSA was selected as the matrix for calibration curves. When we compared slopes of analytes in water and BSA dissolved in water, the slopes were nearly identical. Because of concerns about matrix matching, BSA was selected as a better matrix for calibrators than water alone. All BSA calibration curves were linear (see online Supplemental Fig. 4) and all had r2 > 0.99 (see online Supplemental Table 1).
Intraassay and interassay imprecision (14) for all of the analytes was <10% (Table 1). Recovery of each of the analytes in human plasma was 85.1%–123.6% (Table 2). All analytes in BSA were linear from 1.625 to 125 μmol/L (arginine, citrulline, ornithine, proline) and 0.078 to 6.25 μmol/L (ADMA, SDMA) when diluted with water (Fig. 2; online Supplemental Table 2). Dilutions were made in water to test a matrix that differed from the matrix-matched calibrators, for which we previously demonstrated a linear response. Thus, we diluted BSA or plasma to test the effect of matrix complexity on linearity.
The limit of quantification (LOQ) for all analytes ranged from 3.1 μmol/L for arginine, ≥0.07 μmol/L for ADMA and SDMA, and 0.31–0.84 μmol/L for the remaining analytes (Table 1). Interference from uremia or hemolysis was <6% for all analytes except ADMA, which exhibited 12% interference under uremic conditions. All analytes displayed no interference due to hypertriglyceridemia up to 4.6 g/L (see online Supplemental Table 3). Ion suppression was evaluated by using a T-infusion experiment (see online Supplemental Fig. 5), which demonstrated acceptable separation of the analytes from the most severe ionization inhibitors.
Investigators who used only plasma or serum samples have reported discrepant concentrations for various arginine metabolites and have suggested that arginine concentration is higher in serum than in plasma [reviewed in (15)]. To test this more formally, we measured basal concentrations of all analytes in serum and plasma drawn simultaneously from mice (see online Supplemental Table 3). Indeed, arginine concentrations were approximately 34% higher in serum, and ornithine concentrations were about 80% higher in serum than plasma. In contrast, proline was 25% higher in plasma than in serum; serum concentrations were not significantly different for citrulline, ADMA, and SDMA. Although we did not directly compare plasma and serum concentrations in a similar experiment in humans, our normal population means for serum and plasma were similar to those observed in mice (data not shown).
EXPERIMENTAL DETERMINATION OF ANALYTES IN MOUSE AND HUMAN PLASMA
To determine whether arginine metabolism is altered in systemic inflammatory states, we measured arginine and metabolites in the plasma of mice with and without a peripheral injection of a proinflammatory stimulus (i.e., LPS). After a 6-h rest, the 2 groups of mice had different plasma concentrations of arginine and metabolites (Table 3). Plasma concentrations of arginine (P = 0.016) and proline (P = 0.003) were significantly decreased in the LPS-treated group compared to their untreated counterparts. The arginine/ADMA ratio, a measure of arginine bioavailibity (16), was also significantly higher in untreated mice (P < 0.0001).
We then determined whether inflammation in humans was also associated with changes in arginine metabolism. Arginine and metabolite concentrations were determined in 59 human plasma samples divided into 2 groups on the basis of high-sensitivity CRP (hsCRP) concentration: a normal group with concentrations ranging from 0 to 10 mg/L hsCRP, the upper limit of reference range for the assay, and an increased group with concentrations >10 mg/L hsCRP. There were no significant age or sex differences between the populations (see online Supplemental Table 5). Several analyte concentrations were significantly higher in the normal-hsCRP group than the increased-hsCRP group (see online Supplemental Table 6): arginine (P = 0.0006), citrulline (P = 0.0007), ornithine (P = 0.048), and proline (P = 0.0043); in contrast, ADMA (P = 0.0054) and SDMA (P = 0.0005) were significantly higher in the increased-hsCRP group. Arginine, citrulline, ornithine, and proline concentrations exhibited a statistically significant negative association with increasing hsCRP concentration, whereas ADMA and SDMA concentrations displayed a statistically significant positive association with increasing hsCRP (Fig. 3).
We have developed a method for the simultaneous measurement of arginine, citrulline, ornithine, proline, ADMA, and SDMA in plasma from humans and mice. The method employs a small sample volume (25 μL), use of a stable-isotope–labeled l-arginine internal standard, a simple sample preparation method with isopropanol (17), and a chromatographic run time of 9 min without the need for derivatization. Our technique complements existing published methods that use HILIC chromatography and LC-MS/MS (13, 18, 19) and builds on them by multiplexing the analysis with 6 analytes and extending our findings to a human population stratified by hsCRP and a mouse model of inflammation.
Our reported concentrations for arginine, citrulline, ornithine, and proline are consistent with previously published values in mouse and human plasma and serum (13, 18, 20–25). To our knowledge, this is the first report of a method that uses HILIC and multiple reaction monitoring to measure citrulline and proline in plasma. Our method is similar to a previously reported method for the measurement of proline in muscle (26), and our reported concentrations in plasma are similar to previously published values obtained by using other methods (20, 23). Finally, to our knowledge this study is also the first to compare concentrations for arginine metabolites between human and mouse in clinically relevant models of systemic inflammation. Analyte measurement across species is important for making appropriate comparisons between results from animal models and human clinical studies.
The methylated arginine derivatives ADMA and SDMA are particularly important analytes with different biological activities but very similar chemical structures (7). Our method partially resolves the 2 analytes and uses transitions that have no crosstalk. In a subset of human plasma samples with normal concentrations of hsCRP (hsCRP ≤10 mg/L), ADMA and SDMA concentrations ranged from 0.10–1.43 μmol/L and 0.11–2.90 μmol/L, respectively. Although 10 mg/L is the upper end of the reference interval in our assay, hsCRP concentrations between 3–10 mg/L are often found in individuals at high risk for cardiovascular disease (27). Indeed, the highest ADMA and SDMA concentrations reported in the normal range were from patients with hsCRP concentrations ≥2.8 mg/L. Our experimental design did not include careful clinical characterization of the patients from whom plasma samples were obtained for method development and other initial studies. The literature for ADMA and SDMA concentrations in mouse models is less extensive; however, our reported concentrations (0.7–1.3 μmol/L) are consistent with published values (21, 28). In the future, it will be important to determine the influence of local vs systemic, acute vs chronic, and infectious vs noninfectious inflammatory processes on plasma concentrations of the methylated arginines.
There are also important, and potentially biologically meaningful, differences between some analytes in plasma and serum. By comparing analyte concentrations between serum and EDTA-treated plasma samples collected simultaneously from the same mouse, we found that arginine concentrations were consistently higher in serum than in plasma. This finding is consistent with previous reports showing that arginine concentrations are higher in serum than plasma (15). One possible explanation is that higher serum arginine concentrations result from arginine release from erythrocytes, which occurs during the coagulation cascade. Proline concentrations, in contrast, were about 25% higher in plasma than in serum. The results we obtained for arginine and proline can also be attributed to the use of EDTA as an antiocogulant. Because EDTA stabilizes metalloproteases and other labile molecules (29), the activity of enzymes that suppress arginine catabolism and enhance proline catabolism may be increased in serum. The differences between plasma and serum are also not likely to be attributable to nonspecific monocyte or neutrophil activation due to pyrogens in the draw tube, because every effort was taken to observe proper aseptic technique. These findings underscore the important differences between plasma and serum that can make it difficult to compare data across related metabolomics studies (15, 29, 30).
Results of several studies have suggested that increased systemic inflammation is associated with impaired arginine metabolism (31–34). We therefore conducted 2 separate sets of experiments in both humans and mice to determine whether we could use this method to test the hypothesis that increased systemic inflammation alters flux through arginine metabolic pathways by decreasing bioavailability of arginine and its amino acid metabolites and increasing the concentration of methylated arginines. In our human population samples that were stratified according to plasma hsCRP concentration, we found that samples from individuals with increased hsCRP exhibited dramatic declines in plasma arginine, citrulline, ornithine, and proline. Associations between concentrations of plasma hsCRP and these metabolites demonstrated significant inverse correlations with increased plasma hsCRP concentrations. Similar fluxes in arginine metabolism were also observed in our mouse model of acute systemic inflammation, because plasma concentrations of arginine and proline were also suppressed in LPS-treated mice compared to controls.
Because increased plasma concentrations of ADMA, and to a lesser extent SDMA, are also associated with many systemic inflammatory conditions (7), we also hypothesized that plasma methylated arginine concentrations would also be higher in the increased-hsCRP group and in LPS-treated mice. Although there were no significant differences in ADMA or SDMA concentrations in LPS-treated mice, both ADMA and SDMA concentrations were significantly higher in the increased-hsCRP group and, overall, methylated arginine concentrations showed a significant correlation with increased hsCRP. These findings are supported by recent findings from the Scottish Health Survey, which showed that increased hsCRP (>10 mg/L) is a strong prognostic indicator for increased risk of severe cardiovascular events and all-cause mortality (35).
Taken together, these results support our hypothesis and further the concept that flux through arginine metabolic pathways is part of a complex mechanism employed by circulating monocytes, tissue macrophages, and other immune cells to resolve systemic inflammation. Based on our simplified model of arginine metabolism, our results support a mechanism whereby acute systemic inflammation shifts macrophage homeostasis to a predominant M1 phenotype. This phenotype is characterized by diminished arginine bioavailability and decreased arginine catabolism, which in turn impair activation of M2 pathways that require ornithine and proline to support cell growth and repair mechanisms. Increased plasma ADMA and SDMA are also part of this mechanism, although whether these increases are a cause or consequence of increased systemic inflammation is unclear. The observed decrease in citrulline is more complex and may be due to a number of factors, including: decreased activity of diaminodyhydrolase-mediated degradation of ADMA, decreased activity of ornithine transcarboxylase–mediated degradation to citrulline, or alterations in de novo conversion of arginine from citrulline (8, 32, 36). Better elucidation of these mechanisms will require an expanded profile of additional analytes in the arginine metabolome combined with studies at the cellular and molecular levels.
Targeted metabolomics approaches can play an important role in furthering our understanding of the pathophysiological mechanisms in human health and disease. The method described here for simultaneously measuring arginine and 5 metabolites harnesses the analytical sensitivity and specificity of LC-MS/MS to probe systemic fluctuations of the arginine/iNOS/Arg1 metabolic pathway and will be used to assess the importance of arginine metabolism as a regulator of innate immunity.
↵7 Nonstandard abbreviations:
- inducible nitric oxide synthase;
- arginase 1;
- asymmetric dimethylarginine;
- symmetric dimethylarginine;
- protein methyltransferase;
- endothelial nitric oxide synthase;
- neuronal nitric oxide synthase;
- dimethylarginine diaminohydrolase;
- hydrophilic liquid interaction liquid chromatography;
- liquid chromatography–tandem mass spectrometry;
- C-reactive protein;
- limit of quantification;
- high-sensitivity CRP.
Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.
Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the Disclosures of Potential Conflict of Interest form. Potential conflicts of interest:
Employment or Leadership: A.N. Hoofnagle, Clinical Chemistry, AACC.
Consultant or Advisory Role: A.N. Hoofnagle, Thermo Fisher.
Stock Ownership: None declared.
Honoraria: None declared.
Research Funding: Clinical Mass Spectrometry Facility at the University of Washington; C.M. Brown, NIH grant F32AG027614; P.M. Wise, NIH grant RO1AG002224; A.N. Hoofnagle, Waters, Bruker Daltonics, the Clinical Nutrition Research Unit/Nutrition and Obesity Research Center (P30DK035816), and the Seattle Mouse Metabolic Phenotyping Center (U24 DK076126).
Expert Testimony: None declared.
Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.
Acknowledgments: We thank Dr. Carol A. Colton (Duke University Medical Center) for helpful discussions and comments on the manuscript.
- Received for publication September 7, 2010.
- Accepted for publication February 24, 2011.
- © 2011 The American Association for Clinical Chemistry