| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Automation and Analytical Techniques |
1 Department of Metabolic and Endocrine Diseases and Department of Biomedical Genetics, University Medical Center Utrecht, Utrecht, The Netherlands; 2 Department of Metabolic Diseases, University Medical Center Utrecht, Utrecht, The Netherlands.
aAddress correspondence to this author at: Department of Metabolic and Endocrine Diseases, University Medical Center Utrecht, Postbox 85090, 3508 AB Utrecht, The Netherlands. Fax +31887555350; e-mail S.Fuchs{at}umcutrecht.nl.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: We developed 2 chromatographic techniques, a nonchiral derivatization with chiral (chirasil-L-val column) separation in a GC-MS system and a chiral derivatization with Marfey’s reagent and LC- MS analysis. We validated the techniques for D-serine, L-serine, and glycine determination in cerebrospinal fluid (CSF), evaluated several confounders, and determined age-dependent human concentration ranges.
Results: Quantification limits for D-serine, L-serine, and glycine in cerebrospinal fluid were 0.14, 0.44, and 0.14 µmol/L, respectively, for GC-MS and 0.20, 0.41, and 0.14 µmol/L for LC-MS. Within-run imprecision was <3% for both methods, and between-run imprecision was <13%. Comparison of both techniques with Deming regression yielded coefficients of 0.90 (D-serine), 0.92 (L-serine), and 0.96 (glycine). Sample collection, handling, and transport is uncomplicated—there is no rostrocaudal CSF gradient, no effect of storage at 4 °C for 1 week before storage at –80 °C, and no effect of up to 3 freeze/thaw cycles. Conversely, contamination with erythrocytes increased D-serine, L-serine, and glycine concentrations. CSF concentrations for 145 apparently healthy controls demonstrated markedly and specifically increased (5 to 9 times) D-serine concentrations during early central nervous system development.
Conclusions: These 2 clinically applicable analysis techniques will help to unravel pathophysiologic, diagnostic, and therapeutic issues for disorders associated with central nervous system abnormalities, NMDA-receptor dysfunction, and other pathology associated with D-amino acids.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-carbon atom. Incorporation of a particular amino acid enantiomer in proteins or (poly)peptides determines the spatial architecture of these biological polymers and plays a major role in enzymatic specificity and structural interactions (1). Consequently, it was assumed that homochirality evolved in nature, with all living organisms being composed only of L-amino acids (2). With the advance of chromatographic analysis techniques, improved determination of the different amino acid enantiomers was achieved, revealing the unexpected but undeniable presence of small quantities of D-amino acids in lower and higher animals, plants, and foods (1). Research has largely focused on the D-amino acid D-serine, which has been identified in surprisingly high concentrations in the mammalian central nervous system (3). Subsequent studies demonstrated endogenous D-serine metabolism and synthesis from L-serine (4). Like glycine, D-serine functions as a neuromodulator through binding to the N-methyl D-aspartate (NMDA)1 excitatory amino acid receptor (5). NMDA receptors are involved in central nervous system development, brain plasticity, memory, and learning. NMDA-receptor dysfunction has been implicated in various pathological conditions, including schizophrenia, epilepsy, stroke, and neurodegenerative conditions (6). Recently, we reported that D-serine might be essential for human central nervous system development and provided the first example of human D-serine deficiency in patients with 3-phosphoglycerate dehydrogenase deficiency (7). Other studies have implicated altered D-serine concentrations in schizophrenia (8)(9), amyotrophic lateral sclerosis (10), and nociception (11). Together, these studies imply that D-serine is important in human physiology and pathology.
The presence and possible important roles of D-amino acids in higher organisms (12) not only challenge former theories on mammalian physiology, but also contribute to new insights in disease and treatment modalities. Therefore, it is essential to be able to study D-amino acids in biological samples in a clinical setting. Despite technological advances in the analysis of chiral compounds (for a comparison of the different techniques, we refer to reviews on this subject (13)(14)(15)(16)), analysis of D-amino acids in biological samples remains challenging because of interference from high concentrations of L-amino acids and biological substances such as peptides and amines. To enable simultaneous determination of different D- and L-amino acids in small volumes of biological fluids, we optimized 2 stable isotope dilution analysis techniques (GC-MS (17) and LC-MS (18)) for a clinical laboratory setting. We focused on the simultaneous determination of D-serine, L-serine, and glycine concentrations in human cerebrospinal fluid (CSF), because of their mutual involvement in glutamatergic neurotransmission.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-2,4-dinitro-5-fluorophenyl-L-alaninamide (Marfey’s reagent), and ammonium formiate from Sigma-Aldrich Inc.; chloroform, 5-sulfosalicylic acid, L-serine, glycine, disodiumtetraborate.10H2O, formic acid, and hydrochloric acid from Merck; pentafluoropropionic anhydride from Pierce; acetonitrile from Rathburn Chemicals Ltd.; acetone from Fluka Chemicals Ltd.; and labeled stable isotopes 3-13C(99%)DL-serine and 1,2-13C2(99%)glycine from Cambridge Isotope Laboratories, Inc. All chemicals were of guaranteed grade.
csf samples
For validation studies and evaluation of potential confounding factors, we procured remnants of CSF samples from the laboratory of metabolic and endocrine diseases, the laboratory of microbiology, and the neonatal care unit of our university hospital.
D-Serine concentrations were determined in human CSF samples sent to the laboratory of microbiology in our hospital from July 2004 to November 2007 to exclude meningitis. To avoid interference with the normal routine in the laboratory of microbiology, these samples were kept at 4 °C for 1 week before being stored at –80 °C (no difference with our normal routine of immediate storage at –80 °C; see Supplemental Table 1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol54/issue9). We excluded samples with more than 100 erythrocytes/mL, meningitis, HIV, intracranial pathology, perinatal asphyxia, serine biosynthesis disorders, epilepsy, schizophrenia, and neurodegenerative disorders. Sample use for our studies was approved by the medical ethics committee of the University Medical Center Utrecht.
quality control samples
CSF samples from the neonatal care unit were pooled. To simulate different clinically relevant concentrations, sample volumes of 100, 200, and 300 µL (QC1a–c) were derivatized following the normal GC-MS sample preparation procedure described below, which includes drying of the solution after addition of internal standard and derivatization of the complete residue. Compared with the 200-µL sample, the 100- and 300-µL samples simulated concentrations that are 0.5 or 1.5 times the concentrations of the pooled CSF batch, respectively. For recovery analysis, CSF from the laboratory of metabolic and endocrine diseases was pooled (QC2) and spiked with aqueous amino acid solutions, theoretically increasing L-serine concentrations with 33.33 µmol/L and D-serine and glycine concentrations with 6.66 µmol/L, thus doubling the expected concentrations (QC3). Similarly, we prepared QC4 samples by spiking pooled CSF with aqueous solutions of 15 µmol/L D-serine, L-serine, and glycine.
gc-ms analysis
Sample preparation procedure.
CSF samples were thawed at room temperature and derivatized according to the method described by Bruckner et al. (17). After stirring, we transferred 200 µL CSF to 1.5 mL microtubes (Sarstedt) and added 50 µL internal standard solution (aqueous solution of 600 µmol/L 3-13C-DL-serine and 60 µmol/L 1,2-13C2-glycine). The samples were deproteinized by adding 200 µL aqueous 5-sulfosalicylic acid, thoroughly mixing, and centrifuging in a Heraeus Biofuge Pico centrifuge (Dijkstra Vereenigde BV) at 12 354g for 4 min at room temperature. We applied the supernatant to Durapole microfilters (0.22 µm; Millipore) to eliminate remaining (oligo-)peptides and transferred the filtrate to Pyrex culture tubes with screw caps and PTFE-faced rubber lining (Sigma-Aldrich Inc.). The solvents were removed in a nitrogen stream at 40 °C. We added 250 µL of 2.5 mol/L HCl in 2-propanol (acetyl chloride in 2-propanol 1:4 v/v) to the dry residue. After heating for 45 min at 70 °C in heating blocks, reagents were removed in a nitrogen stream at ambient temperature. We added 400 µL chloroform and 100 µL pentafluoropropionic anhydride and heated the mixture at 100 °C for 20 min. Reagents were removed in a nitrogen stream at ambient temperature. We dissolved the residues in 50 µL chloroform and subjected 2 µL to GC-MS. For each analytical run, we prepared a 7-point calibration curve with aqueous solutions of L-serine (0–112.5 µmol/L), D-serine, and glycine (0–26.25 µmol/L).
Chromatographic conditions.
The GC comprised a HP-5890 gas chromatograph and a HP-7673 automatic sampler (Agilent Technologies Netherlands BV). The derivatized amino acids were delivered by automatic injection (split injection port 1:20) over a glass wool liner to the WCOT fused-silica CP chirasil-L-val (N-propionyl-L-valine tert-butylamide polysiloxane) capillary column (25 by 0.25 mm internal diameter; 0.12 µm film thickness; Chrompack). Helium was used as carrier gas (1 mL/min), with automatic pressure adaptation. The temperature program started at 80 °C for 3 min, increased at 3 °C/min to 190 °C, and held at 190 °C for 5 min. The injector and detector temperatures were set at 220 °C.
Mass spectrometric conditions.
In the quadrupole HP-5989B mass spectrometer (Agilent Technologies), the eluted derivatized amino acids were ionized by negative chemical ionization using 5% ammonia in methane (NTG). The ion source and the quadrupole temperatures were set at 250 °C and 150 °C, respectively, according to the manufacturer’s protocol. The MS was run in the selected ion monitoring mode (SIM). The appropriate ion sets were selected, using the following characteristic mass fragments (m/z) of the N(O) pentafluoropropionyl-2-propanol esters of the amino acids: serine (m/z 255), 13C-serine (m/z 256), glycine (m/z 243), and 13C2-glycine (m/z 245). GC-MS control and data processing were performed with HP G1034C and G1710BA MS Chemstation software, respectively (Agilent Technologies).
lc-ms analysis
Sample preparation procedure.
CSF samples were thawed at room temperature and derivatized according to the method described by Goodlett et al. (18) and Berna and Ackermann (19). We added 50 µL internal standard solution (600.5 µmol/L 3-13C-DL-serine and 61 µmol/L 1,2-13C2-glycine in 0.1 mol/L HCl) to 100 µL CSF. On drying in a nitrogen stream, the residues were derivatized with 50 µL 0.5% Marfey’s reagent (wt/vol in acetone) and 100 µL 0.125 mol/L disodiumtetraborate.10H2O for 30 min at 40 °C. The reaction was stopped with 25 µL 4 mol/L HCl. We diluted the resulting solution (1:10) with eluent buffer (250 mg ammonium formiate in 1 L milliQ-water; pH adjusted to 4.6 with formic acid) and subjected 10 µL to LC-MS. For each analytical run, we prepared a 6-point calibration curve with aqueous solutions of L-serine (0–100 µmol/L), D-serine, and glycine (0–20 µmol/L).
Chromatographic conditions.
The derivatized amino acids were resolved on an Alliance 2795 HPLC system (Waters), with separation on an Atlantis dC18 analytical column (3 µm, 3.9 by 150 mm) (Waters), using a linear gradient of 100% mobile phase A (250 mg ammonium formiate in 1 L milliQ-water; pH adjusted to 4.6 with formic acid) to 50% mobile phase A and B (acetonitrile) in 15 min. The flow rate was 0.3 mL/min.
Mass spectrometric conditions.
A Quattro Ultima triple quadrupole mass spectrometer (Waters) was used in the negative electron spray ionization (ESI) mode. The following mass spectrometer settings were used: capillary voltage 3.0 kV, cone voltage 40 V, cone gas flow 185 l/h, desolvation gas flow 677 l/h, collision gas pressure 1.33e–3, and source temperature 150 °C. The appropriate ion sets were selected, using the following characteristic mass fragments (m/z) of the dinitrophenyl-L-alanine-amides of the amino acids: serine (m/z 356.1), 13C-serine (m/z 357.1), glycine (m/z 326.1), and 13C2-glycine (m/z 328.1). Masslynx software, which included Quanlynx (Waters), was used for instrument control, data acquisition, and data processing.
validation studies
For both analysis techniques, we performed validation in accordance with the guidelines from the European Committee for Clinical Laboratory Standards (20), using CSF samples and the lowest standards of the calibration curves to assess the limit of detection (LOD, concentration at a signal-to-noise ratio of 3, n = 10) and limit of quantification (LOQ, concentration at a signal-to-noise ratio of 10, n = 10). We used quality control samples to assess within-run (n = 10 for QC1a–c, QC2, and QC3) and between-run (n = 10 for QC1a–c, QC2, and Q3) imprecision. We analyzed recovery in QC3 and QC4 samples (n = 10), assessed linearity and range of detection in CSF (LOQ to the highest concentration of the calibration curve (r2 > 0.99) + 10%) for D-serine, L-serine, and glycine, and compared the results obtained with the 2 new analysis techniques (n = 68). Stability of samples prepared for LC-MS analysis was assessed up to 13 days.
influence of possible confounding factors
We evaluated the effect of CSF fraction on amino acid concentrations (GC-MS), using different CSF fractions from the same lumbar puncture. These samples were available after diagnostic CSF neurotransmitter analyses, for which CSF was divided in 6 consecutive withdrawal portions: 1) 0.5 mL directly stored at –80 °C; 2) 2 mL stored at 4 °C until analysis, thereafter storage at –80 °C; 3) 1.5 mL directly stored at –80 °C; 4) 1 mL stored at 4 °C until analysis, thereafter storage at –80 °C; 5) 1–2 mL stored at 4 °C until analysis, thereafter storage at –80 °C; and 6) 1–2 mL directly stored at –80 °C. We tested the influence of repetitive freeze/thaw cycles by analyzing (LC-MS) 5 different CSF samples after 1, 2, and 3 freeze/thaw cycles (5 samples are required to detect a 15% difference using a technique with a 7% imprecision with a power of 0.8, = 0.05, β = 0.1). Finally, we assessed the effect of CSF contamination with blood (>200 erythrocytes/mL) (GC-MS) by comparing concentrations within age groups with CSF containing virtually no blood (<100 erythrocytes/mL) (7).
statistical analysis
We calculated amino acid concentrations after linear regression analysis of the calibration curve (SigmaStat 3.0), using peak area ratios of the amino acids to their internal standards. We compared the GC-MS and LC-MS techniques using Deming regression, evaluated the influence of storage at 4 °C before storage at –80 °C with a paired 2-tailed Student t test, evaluated the influence of erythrocyte contamination within the corresponding age group with an unpaired 2-tailed Student t test, and analyzed CSF fraction and freeze-thaw data with a repeated measurements model. The level of significance was set at P = 0.05. CSF concentrations were categorized by age. For each group, the median and CI were determined.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). Whereas GC-MS sample preparation comprised approximately 6–8 h, LC-MS sample preparation took only 60 min. Elution time was similar (GC-MS 6–12 min; LC-MS 12–14 min).
|
validation studies
Both techniques enabled the determination of low D-serine, L-serine, and glycine concentrations in CSF, with good recovery and range of detection (Table 1
) amply covering the physiological range (Table 2
). The calibration curves for both methods were reproducible and linear over the range tested (r2 > 0.99, Table 1
; based on previous GC-MS analyses showing physiological D-serine concentrations varying between 2 and 14 µmol/L, linearity for the LC-MS was tested only up to a concentration of 30 µmol/L). Furthermore, analysis with both methods was reproducible (within-run imprecision <3%, between-run imprecision <13%) (Table 1
). Analysis 13 days after sample preparation for LC-MS did not influence D-serine, L-serine, or glycine concentrations [D-serine 8.0 µmol/L, after 13 days 8.8 µmol/L (equaling 2 times the between-run imprecision of 4.7%), L-serine 46.1 µmol/L, after 13 days 48.0 µmol/L (within 2 times the between-run imprecision of 5.7%), and glycine 7.4 µmol/L, after 13 days 7.3 µmol/L (within 2 times the between-run imprecision of 6.0%)].
|
|
GC-MS analysis yielded on average 10% higher D-serine (concentration range tested, 0.04–39.5 µmol/L), 8% higher L-serine (10.4–129.6 µmol/L), and 4% higher glycine concentrations (4.0–101.5 µmol/L) than LC-MS analysis (all within the imprecision resulting from the between-run reproducibility of both analysis techniques for the respective amino acids).
influence of possible confounding factors
We evaluated possible clinically relevant confounders by GC-MS and LC-MS. The effect of a rostrocaudal gradient was assessed in CSF from 10 patients from whom 2 (n = 8) or 3 (n = 2) fractions from the same lumbar puncture were available. Fraction analysis did not yield significant differences between any fraction for any amino acid of interest (see Supplemental Table 2 in the online Data Supplement). Similarly, up to 3 repetitive freeze/thaw cycles did not affect D-serine, L-serine, or glycine concentrations (freeze/thaw SEM in the same order of magnitude as the reproducibility SEM; see Supplemental Table 3 in the online Data Supplement). We evaluated the effect of traumatic punctures by comparing 25 samples containing >200 erythrocytes/mL with 39 samples containing <100 erythrocytes/mL within the corresponding age group. Contamination with erythrocytes increased D-serine, L-serine, and glycine concentrations significantly in all age groups containing more than 5 samples, except for L-serine concentrations in the age group 3–32 years (Table 3
).
|
csf concentrations in apparently healthy controls
We measured D-serine, L-serine, and glycine concentrations in 145 CSF samples, of which 46 were analyzed by GC-MS and 99 by LC-MS (Table 2
). Amino acid concentrations decreased with age, which was most pronounced for D-serine, decreasing 4.9-fold (GC-MS) or 9.0-fold (LC-MS) during the first 3 years of life and remaining constant thereafter. During this same period, L-serine concentrations decreased 1.6-fold (GC-MS) or 1.9-fold (LC-MS), and neonatal glycine concentrations decreased 1.8-fold (GC-MS) or 1.2-fold (LC-MS) to the lowest concentrations in childhood, to increase 1.4-fold (GC-MS) or 1.5-fold (LC-MS) to values >50 years of age.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because a CSF rostrocaudal concentration gradient appears to exist for monoamine metabolites (23)(24), we evaluated the effect of CSF fraction on amino acid concentrations. These data show that any CSF fraction may be used for D-serine, L-serine, and glycine analysis. This finding replicates the absence of a rostrocaudal glycine CSF gradient in the only other study on amino acid CSF gradients we identified (25). Another possible confounding factor was repetitive freeze/thaw cycles, which might occur in clinical practice, for example upon diagnostic reevaluation, analytical problems, or sample transport. Up to 3 freeze/thaw cycles did not influence D-serine, L-serine, and glycine concentrations. However, caution remains warranted after repetitive freezing and thawing when analyzing less stable amino acids, such as glutamine or asparagine. Finally, as might be expected because serum amino acids usually greatly exceed CSF amino acid concentrations (26), traumatic punctures (>100 erythrocytes/mL) should be either rejected or mathematically corrected for erythrocyte content, as is done for other blood constituents such as leukocytes or protein (27).
The LC-MS technique has several advantages. LC-MS sample preparation was definitely less laborious and time-consuming, and theoretically can be automated and performed in 96-well plates. Separation is achieved on a nonchiral column, concomitantly applicable for a wide variety of polar compounds. Furthermore, the relatively short LC-MS derivatization reaction time in alkaline solution at 40 °C might facilitate simultaneous determination of a variety of chiral amino acids, including glutamate and aspartate. These dicarboxylic amino acids are important excitatory amino acids in human physiology and pathology (28) but are difficult to quantify accurately (29) because glutamine and asparagine, present in high concentrations in CSF, are very prone to conversion to glutamate and aspartate, respectively. This problem was not encompassed by a milder propylation step in GC-MS sample preparation (data not shown). However, in laboratories without LC-MS, the GC-MS method is a very suitable alternative.
CSF D-serine, L-serine, and glycine concentrations were measured in a human population of varying ages using the 2 different analytical techniques. Because of differences in control sample sizes, we were unable to directly compare values between these methods. Nevertheless, in the validation studies, we showed good correlation between the 2 techniques. Because some deviations are inherent to the use of different analysis techniques, it is advisable for clinical practice to refer to values that have been generated by the same technique. The decrease in concentrations with age underscores the importance of age-specific reference intervals. The median values of the D-serine and L-serine concentrations we observed here are somewhat lower than those we reported previously (7). This might be explained by the different and much larger population we studied here. Nevertheless, we were able to replicate our previous finding of 7-fold increased D-serine concentrations during the first few months of life (7). In addition, the values determined by both techniques correlate well with the recently published values of total serine and glycine in CSF of 77 infants (serine, median 52 µmol/L, range 25–105 µmol/L; glycine: median 9 µmol/L, range 3–19 µmol/L) (30) and also with older studies (31). To our knowledge, age-dependent reference intervals for the separate isomers are not available for comparison with our results. In our previous study (7), we showed D-serine deficiency in patients with 3-phosphoglycerate dehydrogenase deficiency, illustrating that our analysis technique differentiates between physiology and pathology.
|
Acknowledgments |
|---|
Financial Disclosures: None declared.
Acknowledgments: We are indebted to A. Uijttewaal (Department of Metabolic and Endocrine Diseases, University Medical Center Utrecht), the Laboratory for Medical Microbiology, the Laboratory of Metabolic and Endocrine Diseases, and the Department of Neonatology of the University Medical Center Utrecht for their collaboration in cerebrospinal fluid collection, and to Dr. E. Struys (Department of Clinical Chemistry, VU University Medical Center, Amsterdam) for directing our attention to the HPLC analysis method using Marfey’s reagent.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  K. Hashimoto CSF Serine Enantiomers and Glycine in the Study of Neurologic and Psychiatric Disorders Clin. Chem., September 1, 2008; 54(9): 1413 - 1414. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
