- Email: [email protected]
A microtitre format assay for proline in human serum or plasma
Clinica Chimica Acta 343 (2004) 113 – 118 www.elsevier.com/locate/clinchim
A microtitre format assay for proline in human serum or plasma David J. Grainger *, Sri Aitken Department of Medicine, University of Cambridge, Box 157, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK Received 12 September 2003; accepted 12 December 2003
Abstract Background: Recent studies have suggested that low serum proline concentration may be associated with low bone mineral density. However, further investigation of this association has been hampered by the lack of a relatively high throughput assay for proline in biological fluids. Here we report a sensitive and specific microtitre plate format assay for proline which exploits the chemical interaction between proline and isatin. Methods: Human serum or plasma is deproteinised by incubation with sodium citrate buffer pH4.1 at 95 jC, and the supernatant is reacted with isatin at 95 jC for 3 h. The resultant blue coloured product is quantitated sprectrophometrically. Results: This assay yields a linear standard curve in the range 15 Amol/l to 1 mmol/ l (r = 0.998 F 0.002; n = 8 determinations) with a sensitivity of 31 F 11 Amol/l. None of the other proteogenic amino acids are detected (< 0.3% detection at 10 mmol/l) and the closely related metabolite hydroxyproline is only very weakly detected (3% detection at 10 mmol/l). Using human serum, the assay has linear dilution characteristics and a mean spike recovery of 107 F 5%. Repeated re-measurement of the same serum sample yields an intra-assay coefficient of variation (CV) of 4.8% and an inter-assay CV of 6.1%. Conclusions: This method provides the first reliable micro-titre format assay for proline in human serum. D 2004 Elsevier B.V. All rights reserved. Keywords: Amino acid; Osteoporosis; Collagen; Diet; Blood
1. Introduction It has been known since 1912 that significant concentrations of the proteogenic amino acid monomers are found in circulating blood [1]. In healthy individuals, the levels of various amino acids range from 25 Amol/l for methionine to 500 Amol/l for alanine and valine [2]. Generally, the levels of the amino acids in blood are tightly regulated and do not vary to a great * Corresponding author. Tel.: +44-01223-336812; fax: +4401223-762770. E-mail address: [email protected] (D.J. Grainger). URL: http://www.graingerlab.org. 0009-8981/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cccn.2003.12.022
extent between individuals and as a result they are not routinely measured in clinical studies. Several of the unconjugated proteogenic amino acids (such as glutamate and glycine) are now known to possess potent biological function, particular as neurotransmitters. This has resulted in more extensive studies of the control of the levels of these amino acids in blood. For example, it is known that plasma glutamate concentration remains relatively stable in response to substantial dietary glutamate loading due to the consumption of soy fermentation products [3]. Even under conditions of dramatically altered renal function, such as in Wilson’s disease, the level of the amino acids in plasma remain unchanged [4].
114
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Recently, however, a possible association between the concentration of proline in the blood and the presence of osteoporosis has been uncovered (DJG, J. Nicholson and E. Holmes, unpublished observations). Serum samples from subjects with osteoporosis were compared to samples from control subjects with normal bone mineral density, using an NMR spectrometer to obtain a fingerprint of the low molecular weight metabolites present. Application of pattern recognition software to the 1H-NMR spectra from the serum samples [5], revealed a pattern consistent with low circulating concentrations of proline among subjects with osteoporosis. Such an association is potentially interesting because proline is a major constituent of collagen in bone, and lack of available proline could plausibly impact on bone mineralisation rates. Unfortunately, it is difficult to draw an unequivocal conclusion with regard to the proline concentration from the 1H-NMR spectra of serum samples. Even using 2D NMR techniques (such as TOCSY) to assign the peaks in the spectrum, it is difficult to determine what proportion of the variance in the integral of each peak is due to a single metabolite, and what proportion might be contributed by closely related metabolites (such as hydroxyproline). The interpretation of the spectrum is further hindered by the possibility that subtle variations in the pH of the serum altered the equilibrium between free amino acid and the carbamate formed through reaction with carbon dioxide. A definitive statement that low serum proline is associated with osteoporosis required a conventional biochemical assay for serum proline. There have only been a small number of previous reports of the measurement of serum proline, mostly using chromatographic separation [7– 9]. These studies reported that the normal range of serum proline concentrations in several different populations was 200– 300 Amol/l [6– 8] and that no alteration in serum proline levels were associated with the progress of various fibrotic diseases, such as liver cirrhosis [8,9]. While being accurate, such techniques are not well suited to population comparisons in cohorts with several hundred individuals, as each sample must be analysed separately. In one study, however, Boctor [6] reported a colorimteric determination of serum proline concentration which exploited the specific chemical interaction
between proline and isatin (2,3-indolinedione) to form a blue-coloured insoluble product with an absorption maximum near 595 nm. Here we adapt this method (which was previously performed in 5 ml tubes) for use in a microtitre plate, and circumvent the requirement for the lengthy extraction of the blue precipitated product with acetone. Our improved methods offers a number of advantages over that of Boctor: it is substantially simpler to carry out with only two steps (deproteinisation then colorimetric determination) compared to four steps previously (deproteinisation, removal of picric acid with ion-exchange resin, boiling with alcohol and extraction of the precipitate with acetone). Our assay has a coefficient of variation between replicate measures comparable with HPLC determinations, whereas it is difficult to achieve a coefficient of variation below 20% using the method of Boctor.
2. Experimental 2.1. Preparation of serum and plasma Serum and plasma were prepared from blood withdrawn from the cubital vein using a 19-gauge butterfly needle without the application of a tourniquet. For serum, the blood was allowed to clot in a polypropylene tube for 2 h at room temperature, then cells and the clot were spun out at 1000 g for 5 min and the supernatant (serum) removed. For plasma, the blood was immediately mixed with anticoagulant (CTAD tubes, Beckton Dickinson) and incubated on ice for 15 min. Cells were then spun out at 2500 g for 30 min at 2– 8 jC and the central one-third of the supernatant taken. All samples were aliquoted and stored at 80 jC until assayed. 2.2. Deproteinisation Protein does not interfere with the assay directly (proline contained in proteins does not react with isatin) but it precipitates under the highly denaturing conditions of the assay preventing spectrophotometric quantitation. Traditional methods of precipitating protein (e.g. treatment with 15% trichloroacetic acid or picric acid) do not remove sufficient protein to prevent a further precipitate forming at 95 jC.
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Deproteinisation was therefore carried out as follows: an equal volume of 500 mmol/l sodium citrate buffer pH 4.1 is added to serum, mixed and incubated at 95 jC in an oven for 1 h. Precipitated protein is then spun out (25,000 g for 10 min) and the supernatant retained for proline assay. This method removes 99.8 F 0.1% of the protein present in serum. Note that the supernatant must be removed very carefully, since transfer of even a small amount of precipitated protein results in over-estimation of the serum proline concentration. For each assay, a 10 mmol/l standard solution of Lproline (99% purity; Sigma) in water was prepared (weighing out at least 100 mg of solid to ensure accuracy) and discarded after use. This solution was diluted into phosphate buffered saline containing 70 mg/ml bovine serum albumin to prepare a series of standard solutions ranging from 1 mmol/l to 15.8 Amol/l proline concentration. These standards were subject to the deproteinisation procedure along side the serum samples. 2.3. Proline assay A 10% (w/v) stock solution of isatin (99% purity; Aldrich Chemical) in DMSO was prepared and stored in the dark at room temperature for up to 1 week. 150 Al of each deproteinised standard or serum sample was then dispensed into a half-area 96-well microtitre plate (well volume f 200 Al; Code #3697, Corning). To each sample, 3 Al of isatin stock solution was added with mixing, generating a final isatin concentration of 0.2% (w/v) which is just below the limit of solubility of isatin in aqueous solution at room temperature. The spacer volumes between the wells of the plate were then filled with water and an adhesive plate sealer was firmly applied to prevent any possibility of evaporation during the subsequent incubation. The plate was then incubated at 95 jC in an oven for 3 h. Formation of a blue suspension/precipitate in wells containing proline is visible by the end of the incubation. This suspension was fully dissolved by addition of 50 Al DMSO to each well (25% v/v final concentration) and mixed thoroughly on an orbital shaker, then incubated at room temperature for 15 min. After further orbital mixing, the plate was read at 595 nm and proline concentrations in the unknown
115
samples calculated by interpolation of a linear – linear plot of the standard curve. Standards and unknown samples were routinely assayed in duplicate. All incubations were performed in the dark.
3. Results The kinetics of formation of the blue product were first investigated at room temperature, 45 and 95 jC. The reaction proceeds with complex kinetics which are not adequately described by any simple bimolecular models (Fig. 1a), but equilibrium was apparently achieved after 2 h at 95 jC. There was no appreciable reaction at room temperature even after 8 h and only a partial reaction at 45 jC. The initial reaction rates (Fig. 1b), the Vmax rates (Fig. 1c) and the equilibrium absorbance (Fig. 1d) were all correlated with proline concentration in the standards, but the end-point determination of absorbance (Fig. 1d) showed the most reproducible correlation (r =0.998 F 0.001; n = 8 determinations) and was therefore used throughout. The other 19 proteogenic amino acids, plus taurine, citrulline and hydroxyproline were tested for crossreactivity in this assay. With the exception of hydroxyproline, all of these amino acids read at < 30 Amol/ l apparent proline concentration when tested at 10 mmol/l, representing a cross-reactivity of < 0.3%. The reading was only statistically significantly above background for cysteine (0.26%) and tryptophan (0.25%). Hydroxyproline showed a more significant cross-reactivity, with an apparent proline concentration of 312 F 14 Amol/l when tested at 10 mmol/l (3% cross-reactivity). Although this cross-reaction is statistically significant, it is unlikely to have any practical significance because circulating levels of hydroxyproline above 200 Amol/l have not been reported, even among individuals with bone disease who have elevated levels of this metabolite [9]. At 200 Amol/l, a 3% cross reactivity would result in an artefactual increase in serum proline concentration of approximately 2%. The sensitivity of the assay, defined as the proline concentration equivalent to twice the standard deviation of eight sample blanks, was determined on three separate occasions. The detection threshold was 31 F 11 Amol/l, suggesting the assay is suitable for the detection of proline in serum or plasma which is in the range 200 – 300 Amol/l.
116
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Fig. 1. Analysis of the reaction between proline and isatin. (a) Kinetics of the formation of the blue reaction product, monitored at 595 nm, at room temperature (filled circles), 45 jC (open circles) and 95 jC (open squares). The reaction kinetics were apparently chaotic within the first 60 min but reached an equilibrium by 3 h which was stable for at least 8 h. The initial reaction rate (b), Vmax rate (c) and equilibrium absorbance (d) all correlated with proline concentration among the standard solutions, with equilibrium absorbance giving the most reproducible correlation. Panel (d) therefore represents a typical standard curve used for interpolation of unknown samples.
To determine whether the assay shows linear dilution characteristics, two samples were prepared: serum from an individual (identified in a preliminary screen of healthy laboratory workers) with a relatively high level of proline (f 400 Amol/l), and serum from this individual with additional proline spiked into it, raising the final concentration by 500 Amol/l. These samples were then assayed neat, and after serial twofold dilution in PBS prior to deproteinisation. The assay showed excellent linear dilution characteristics (Fig. 2) for both samples, right down to the sensitivity threshold of the assay. The mean recovery of the spiked proline was 107 F 5% across the range of dilutions tested.
The reproducibility of the assay was characterised by measuring eight replicate aliquots of the same serum preparation (containing 323 Amol/l proline) on each of 3 days. All the assays were performed by the same operator who had considerable practice at removing the supernatant following deproteinisation without disturbing the precipitated protein. The intraassay coefficient of variation (CV) was 4.8% and the intra-day CV was 6.1%. We conclude that the assay has similar reproducibility characteristics to many immunological or enzymatic assays currently used in biochemical laboratories. The assay was then used to measure the proline concentration in serum samples from 80 apparently
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Fig. 2. Effect of dilution on apparent proline concentration in serum. Human serum from a healthy laboratory volunteer (open circles), and the same serum sample spiked with 500 Amol/l additional proline (filled squares), were subjected to serial twofold dilution in PBS prior to deproteinisation and assay. The assay shows excellent linear dilution characteristics and spike-recovery of 96 – 110% across the entire range of proline concentrations that were detectable.
healthy individuals (age range 30 to 76; mean 56 F 9 years). Serum proline concentration was approximately normally distributed in this population with a mean of 258 Amol/l and a standard deviation of 55 Amol/l. There was no statistically significant difference between the sexes (males: 260 F 48 Amol/ l (n = 38); females 252 F 58 Amol/l (n = 42)). Serum proline concentration was not significantly correlated with age, although there was a trend to decreasing serum proline levels in the oldest subjects. In addition, there was no correlation with serum creatinine concentration, a marker for variations in renal function, consistent with the absence of proline in urine.
4. Conclusions The assay presented here has significant advantages over those described previously [6– 8], particularly in terms of the numbers of samples which can be measured simultaneously. Using this well validated method, it will be possible to further investigate the proposed association between serum proline concen-
117
tration and low bone mineral density (DJG, J. Nicholson and E. Holmes, unpublished observations). Furthermore, it should be possible to begin to assess the mechanisms which regulate serum proline concentration. Of the stable proteogenic amino acids, proline is unique in its absence from urine [4], suggesting that renal clearance does not contribute significantly to the regulation of circulating proline levels. Using the assay described here it will also be possible to investigate the relative importance of dietary sources versus endogenous synthesis in the control of circulating proline levels and whole body proline content. In early studies, it was established that serum proline levels are elevated by more than 50% 2 h following ingestion of 50 g of protein [4]. This suggests that dietary sources of proline can markedly influence circulating proline levels, at least in the short term, but further studies will be required to determine whether dietary composition contributes to the regulation of circulating proline levels. Although the method described here represents a significant improvement over existing methods, it still requires efficient deproteinisation of the serum. This is technically demanding, since even a small contamination of the supernatant with precipitated protein can lead to significant over-estimate of the proline concentration. If application of the assay described here provides sufficient information to suggest that measurement of serum proline were useful, for example, for routine clinical monitoring of individuals with, or at risk from, osteoporosis, then alternative assays would have to be devised and validated. For example, it is likely that an assay based on the recombinant NADP-dependent enzyme DV-pyrroline-5-carboxylate reductase, which was recently cloned from Bacillus subtilis [10], could be developed to allow the rapid and accurate determination of proline concentration in untreated serum. Such an assay could be used on many of the autoanalysers currently in use in clinical laboratories.
Acknowledgements David Grainger is a Royal Society University Research Fellow.
118
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
References [1] Van Slyke DD, Meyer GM. J Biol Chem 1912;12:399 – 412. [2] Stein WH, Moore S. The free amino acids of human blood plasma. J Biol Chem 1954;211:915 – 26. [3] Stegink LD, Filer Jr LJ, Baker GL. Plasma amino acid concentrations in normal adults fed meals with added monosodium L-glutamate and aspartame. J Nutr 1983;113:1851 – 60. [4] Stein WH, Bearn AG, Moore S. The amino acid content of the blood and urine in Wilson’s disease. J Clin Invest 1954; 33:410 – 9. [5] Brindle JT, Antti H, Holmes EC, Tranter G, Nicholson JK, Bethell HWL, et al. Rapid and non-invasive diagnosis of the presence and severity of coronary heart disease using 1HNMR based metabonomics. Nat Med 2002;8:1439 – 45.
[6] Boctor FN. An improved method for colorimetric determination of proline with isatin. Anal Biochem 1971;43:66 – 70. [7] Tanaka Y, Minato Y, Hasumura Y, Takeuchi J. Evaluation of hepatic fibrosis by serum proline and amino-terminal type III procollagen peptide levels in alcoholic patients. Dig Dis Sci 1986;31:712 – 7. [8] Vargas-Tank L, Wolff C, Castillo T, Carrion Y, Hoppe A, Martinez V, et al. Serum proline and blood lactate levels in alcoholic patients without hepatic failure: Relationship with alcohol ingestion and histological activity. Liver 1988;8:58 – 62. [9] Gasser AB, Wootton R, Beyeler S, Celada A, Depierre D. Biological variation in free serum hydroxyproline concentration. Clin Chim Acta 1980;106:39 – 43. [10] Belitsky BR, Brill J, Bremer E, Sonenshein AL. Multiple genes for the last step of proline biosynthesis in bacillus subtilis. J Bacteriol 2001;183:4389 – 92.
A microtitre format assay for proline in human serum or plasma David J. Grainger *, Sri Aitken Department of Medicine, University of Cambridge, Box 157, Addenbrooke’s Hospital, Cambridge CB2 2QQ, UK Received 12 September 2003; accepted 12 December 2003
Abstract Background: Recent studies have suggested that low serum proline concentration may be associated with low bone mineral density. However, further investigation of this association has been hampered by the lack of a relatively high throughput assay for proline in biological fluids. Here we report a sensitive and specific microtitre plate format assay for proline which exploits the chemical interaction between proline and isatin. Methods: Human serum or plasma is deproteinised by incubation with sodium citrate buffer pH4.1 at 95 jC, and the supernatant is reacted with isatin at 95 jC for 3 h. The resultant blue coloured product is quantitated sprectrophometrically. Results: This assay yields a linear standard curve in the range 15 Amol/l to 1 mmol/ l (r = 0.998 F 0.002; n = 8 determinations) with a sensitivity of 31 F 11 Amol/l. None of the other proteogenic amino acids are detected (< 0.3% detection at 10 mmol/l) and the closely related metabolite hydroxyproline is only very weakly detected (3% detection at 10 mmol/l). Using human serum, the assay has linear dilution characteristics and a mean spike recovery of 107 F 5%. Repeated re-measurement of the same serum sample yields an intra-assay coefficient of variation (CV) of 4.8% and an inter-assay CV of 6.1%. Conclusions: This method provides the first reliable micro-titre format assay for proline in human serum. D 2004 Elsevier B.V. All rights reserved. Keywords: Amino acid; Osteoporosis; Collagen; Diet; Blood
1. Introduction It has been known since 1912 that significant concentrations of the proteogenic amino acid monomers are found in circulating blood [1]. In healthy individuals, the levels of various amino acids range from 25 Amol/l for methionine to 500 Amol/l for alanine and valine [2]. Generally, the levels of the amino acids in blood are tightly regulated and do not vary to a great * Corresponding author. Tel.: +44-01223-336812; fax: +4401223-762770. E-mail address: [email protected] (D.J. Grainger). URL: http://www.graingerlab.org. 0009-8981/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.cccn.2003.12.022
extent between individuals and as a result they are not routinely measured in clinical studies. Several of the unconjugated proteogenic amino acids (such as glutamate and glycine) are now known to possess potent biological function, particular as neurotransmitters. This has resulted in more extensive studies of the control of the levels of these amino acids in blood. For example, it is known that plasma glutamate concentration remains relatively stable in response to substantial dietary glutamate loading due to the consumption of soy fermentation products [3]. Even under conditions of dramatically altered renal function, such as in Wilson’s disease, the level of the amino acids in plasma remain unchanged [4].
114
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Recently, however, a possible association between the concentration of proline in the blood and the presence of osteoporosis has been uncovered (DJG, J. Nicholson and E. Holmes, unpublished observations). Serum samples from subjects with osteoporosis were compared to samples from control subjects with normal bone mineral density, using an NMR spectrometer to obtain a fingerprint of the low molecular weight metabolites present. Application of pattern recognition software to the 1H-NMR spectra from the serum samples [5], revealed a pattern consistent with low circulating concentrations of proline among subjects with osteoporosis. Such an association is potentially interesting because proline is a major constituent of collagen in bone, and lack of available proline could plausibly impact on bone mineralisation rates. Unfortunately, it is difficult to draw an unequivocal conclusion with regard to the proline concentration from the 1H-NMR spectra of serum samples. Even using 2D NMR techniques (such as TOCSY) to assign the peaks in the spectrum, it is difficult to determine what proportion of the variance in the integral of each peak is due to a single metabolite, and what proportion might be contributed by closely related metabolites (such as hydroxyproline). The interpretation of the spectrum is further hindered by the possibility that subtle variations in the pH of the serum altered the equilibrium between free amino acid and the carbamate formed through reaction with carbon dioxide. A definitive statement that low serum proline is associated with osteoporosis required a conventional biochemical assay for serum proline. There have only been a small number of previous reports of the measurement of serum proline, mostly using chromatographic separation [7– 9]. These studies reported that the normal range of serum proline concentrations in several different populations was 200– 300 Amol/l [6– 8] and that no alteration in serum proline levels were associated with the progress of various fibrotic diseases, such as liver cirrhosis [8,9]. While being accurate, such techniques are not well suited to population comparisons in cohorts with several hundred individuals, as each sample must be analysed separately. In one study, however, Boctor [6] reported a colorimteric determination of serum proline concentration which exploited the specific chemical interaction
between proline and isatin (2,3-indolinedione) to form a blue-coloured insoluble product with an absorption maximum near 595 nm. Here we adapt this method (which was previously performed in 5 ml tubes) for use in a microtitre plate, and circumvent the requirement for the lengthy extraction of the blue precipitated product with acetone. Our improved methods offers a number of advantages over that of Boctor: it is substantially simpler to carry out with only two steps (deproteinisation then colorimetric determination) compared to four steps previously (deproteinisation, removal of picric acid with ion-exchange resin, boiling with alcohol and extraction of the precipitate with acetone). Our assay has a coefficient of variation between replicate measures comparable with HPLC determinations, whereas it is difficult to achieve a coefficient of variation below 20% using the method of Boctor.
2. Experimental 2.1. Preparation of serum and plasma Serum and plasma were prepared from blood withdrawn from the cubital vein using a 19-gauge butterfly needle without the application of a tourniquet. For serum, the blood was allowed to clot in a polypropylene tube for 2 h at room temperature, then cells and the clot were spun out at 1000 g for 5 min and the supernatant (serum) removed. For plasma, the blood was immediately mixed with anticoagulant (CTAD tubes, Beckton Dickinson) and incubated on ice for 15 min. Cells were then spun out at 2500 g for 30 min at 2– 8 jC and the central one-third of the supernatant taken. All samples were aliquoted and stored at 80 jC until assayed. 2.2. Deproteinisation Protein does not interfere with the assay directly (proline contained in proteins does not react with isatin) but it precipitates under the highly denaturing conditions of the assay preventing spectrophotometric quantitation. Traditional methods of precipitating protein (e.g. treatment with 15% trichloroacetic acid or picric acid) do not remove sufficient protein to prevent a further precipitate forming at 95 jC.
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Deproteinisation was therefore carried out as follows: an equal volume of 500 mmol/l sodium citrate buffer pH 4.1 is added to serum, mixed and incubated at 95 jC in an oven for 1 h. Precipitated protein is then spun out (25,000 g for 10 min) and the supernatant retained for proline assay. This method removes 99.8 F 0.1% of the protein present in serum. Note that the supernatant must be removed very carefully, since transfer of even a small amount of precipitated protein results in over-estimation of the serum proline concentration. For each assay, a 10 mmol/l standard solution of Lproline (99% purity; Sigma) in water was prepared (weighing out at least 100 mg of solid to ensure accuracy) and discarded after use. This solution was diluted into phosphate buffered saline containing 70 mg/ml bovine serum albumin to prepare a series of standard solutions ranging from 1 mmol/l to 15.8 Amol/l proline concentration. These standards were subject to the deproteinisation procedure along side the serum samples. 2.3. Proline assay A 10% (w/v) stock solution of isatin (99% purity; Aldrich Chemical) in DMSO was prepared and stored in the dark at room temperature for up to 1 week. 150 Al of each deproteinised standard or serum sample was then dispensed into a half-area 96-well microtitre plate (well volume f 200 Al; Code #3697, Corning). To each sample, 3 Al of isatin stock solution was added with mixing, generating a final isatin concentration of 0.2% (w/v) which is just below the limit of solubility of isatin in aqueous solution at room temperature. The spacer volumes between the wells of the plate were then filled with water and an adhesive plate sealer was firmly applied to prevent any possibility of evaporation during the subsequent incubation. The plate was then incubated at 95 jC in an oven for 3 h. Formation of a blue suspension/precipitate in wells containing proline is visible by the end of the incubation. This suspension was fully dissolved by addition of 50 Al DMSO to each well (25% v/v final concentration) and mixed thoroughly on an orbital shaker, then incubated at room temperature for 15 min. After further orbital mixing, the plate was read at 595 nm and proline concentrations in the unknown
115
samples calculated by interpolation of a linear – linear plot of the standard curve. Standards and unknown samples were routinely assayed in duplicate. All incubations were performed in the dark.
3. Results The kinetics of formation of the blue product were first investigated at room temperature, 45 and 95 jC. The reaction proceeds with complex kinetics which are not adequately described by any simple bimolecular models (Fig. 1a), but equilibrium was apparently achieved after 2 h at 95 jC. There was no appreciable reaction at room temperature even after 8 h and only a partial reaction at 45 jC. The initial reaction rates (Fig. 1b), the Vmax rates (Fig. 1c) and the equilibrium absorbance (Fig. 1d) were all correlated with proline concentration in the standards, but the end-point determination of absorbance (Fig. 1d) showed the most reproducible correlation (r =0.998 F 0.001; n = 8 determinations) and was therefore used throughout. The other 19 proteogenic amino acids, plus taurine, citrulline and hydroxyproline were tested for crossreactivity in this assay. With the exception of hydroxyproline, all of these amino acids read at < 30 Amol/ l apparent proline concentration when tested at 10 mmol/l, representing a cross-reactivity of < 0.3%. The reading was only statistically significantly above background for cysteine (0.26%) and tryptophan (0.25%). Hydroxyproline showed a more significant cross-reactivity, with an apparent proline concentration of 312 F 14 Amol/l when tested at 10 mmol/l (3% cross-reactivity). Although this cross-reaction is statistically significant, it is unlikely to have any practical significance because circulating levels of hydroxyproline above 200 Amol/l have not been reported, even among individuals with bone disease who have elevated levels of this metabolite [9]. At 200 Amol/l, a 3% cross reactivity would result in an artefactual increase in serum proline concentration of approximately 2%. The sensitivity of the assay, defined as the proline concentration equivalent to twice the standard deviation of eight sample blanks, was determined on three separate occasions. The detection threshold was 31 F 11 Amol/l, suggesting the assay is suitable for the detection of proline in serum or plasma which is in the range 200 – 300 Amol/l.
116
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Fig. 1. Analysis of the reaction between proline and isatin. (a) Kinetics of the formation of the blue reaction product, monitored at 595 nm, at room temperature (filled circles), 45 jC (open circles) and 95 jC (open squares). The reaction kinetics were apparently chaotic within the first 60 min but reached an equilibrium by 3 h which was stable for at least 8 h. The initial reaction rate (b), Vmax rate (c) and equilibrium absorbance (d) all correlated with proline concentration among the standard solutions, with equilibrium absorbance giving the most reproducible correlation. Panel (d) therefore represents a typical standard curve used for interpolation of unknown samples.
To determine whether the assay shows linear dilution characteristics, two samples were prepared: serum from an individual (identified in a preliminary screen of healthy laboratory workers) with a relatively high level of proline (f 400 Amol/l), and serum from this individual with additional proline spiked into it, raising the final concentration by 500 Amol/l. These samples were then assayed neat, and after serial twofold dilution in PBS prior to deproteinisation. The assay showed excellent linear dilution characteristics (Fig. 2) for both samples, right down to the sensitivity threshold of the assay. The mean recovery of the spiked proline was 107 F 5% across the range of dilutions tested.
The reproducibility of the assay was characterised by measuring eight replicate aliquots of the same serum preparation (containing 323 Amol/l proline) on each of 3 days. All the assays were performed by the same operator who had considerable practice at removing the supernatant following deproteinisation without disturbing the precipitated protein. The intraassay coefficient of variation (CV) was 4.8% and the intra-day CV was 6.1%. We conclude that the assay has similar reproducibility characteristics to many immunological or enzymatic assays currently used in biochemical laboratories. The assay was then used to measure the proline concentration in serum samples from 80 apparently
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
Fig. 2. Effect of dilution on apparent proline concentration in serum. Human serum from a healthy laboratory volunteer (open circles), and the same serum sample spiked with 500 Amol/l additional proline (filled squares), were subjected to serial twofold dilution in PBS prior to deproteinisation and assay. The assay shows excellent linear dilution characteristics and spike-recovery of 96 – 110% across the entire range of proline concentrations that were detectable.
healthy individuals (age range 30 to 76; mean 56 F 9 years). Serum proline concentration was approximately normally distributed in this population with a mean of 258 Amol/l and a standard deviation of 55 Amol/l. There was no statistically significant difference between the sexes (males: 260 F 48 Amol/ l (n = 38); females 252 F 58 Amol/l (n = 42)). Serum proline concentration was not significantly correlated with age, although there was a trend to decreasing serum proline levels in the oldest subjects. In addition, there was no correlation with serum creatinine concentration, a marker for variations in renal function, consistent with the absence of proline in urine.
4. Conclusions The assay presented here has significant advantages over those described previously [6– 8], particularly in terms of the numbers of samples which can be measured simultaneously. Using this well validated method, it will be possible to further investigate the proposed association between serum proline concen-
117
tration and low bone mineral density (DJG, J. Nicholson and E. Holmes, unpublished observations). Furthermore, it should be possible to begin to assess the mechanisms which regulate serum proline concentration. Of the stable proteogenic amino acids, proline is unique in its absence from urine [4], suggesting that renal clearance does not contribute significantly to the regulation of circulating proline levels. Using the assay described here it will also be possible to investigate the relative importance of dietary sources versus endogenous synthesis in the control of circulating proline levels and whole body proline content. In early studies, it was established that serum proline levels are elevated by more than 50% 2 h following ingestion of 50 g of protein [4]. This suggests that dietary sources of proline can markedly influence circulating proline levels, at least in the short term, but further studies will be required to determine whether dietary composition contributes to the regulation of circulating proline levels. Although the method described here represents a significant improvement over existing methods, it still requires efficient deproteinisation of the serum. This is technically demanding, since even a small contamination of the supernatant with precipitated protein can lead to significant over-estimate of the proline concentration. If application of the assay described here provides sufficient information to suggest that measurement of serum proline were useful, for example, for routine clinical monitoring of individuals with, or at risk from, osteoporosis, then alternative assays would have to be devised and validated. For example, it is likely that an assay based on the recombinant NADP-dependent enzyme DV-pyrroline-5-carboxylate reductase, which was recently cloned from Bacillus subtilis [10], could be developed to allow the rapid and accurate determination of proline concentration in untreated serum. Such an assay could be used on many of the autoanalysers currently in use in clinical laboratories.
Acknowledgements David Grainger is a Royal Society University Research Fellow.
118
D.J. Grainger, S. Aitken / Clinica Chimica Acta 343 (2004) 113–118
References [1] Van Slyke DD, Meyer GM. J Biol Chem 1912;12:399 – 412. [2] Stein WH, Moore S. The free amino acids of human blood plasma. J Biol Chem 1954;211:915 – 26. [3] Stegink LD, Filer Jr LJ, Baker GL. Plasma amino acid concentrations in normal adults fed meals with added monosodium L-glutamate and aspartame. J Nutr 1983;113:1851 – 60. [4] Stein WH, Bearn AG, Moore S. The amino acid content of the blood and urine in Wilson’s disease. J Clin Invest 1954; 33:410 – 9. [5] Brindle JT, Antti H, Holmes EC, Tranter G, Nicholson JK, Bethell HWL, et al. Rapid and non-invasive diagnosis of the presence and severity of coronary heart disease using 1HNMR based metabonomics. Nat Med 2002;8:1439 – 45.
[6] Boctor FN. An improved method for colorimetric determination of proline with isatin. Anal Biochem 1971;43:66 – 70. [7] Tanaka Y, Minato Y, Hasumura Y, Takeuchi J. Evaluation of hepatic fibrosis by serum proline and amino-terminal type III procollagen peptide levels in alcoholic patients. Dig Dis Sci 1986;31:712 – 7. [8] Vargas-Tank L, Wolff C, Castillo T, Carrion Y, Hoppe A, Martinez V, et al. Serum proline and blood lactate levels in alcoholic patients without hepatic failure: Relationship with alcohol ingestion and histological activity. Liver 1988;8:58 – 62. [9] Gasser AB, Wootton R, Beyeler S, Celada A, Depierre D. Biological variation in free serum hydroxyproline concentration. Clin Chim Acta 1980;106:39 – 43. [10] Belitsky BR, Brill J, Bremer E, Sonenshein AL. Multiple genes for the last step of proline biosynthesis in bacillus subtilis. J Bacteriol 2001;183:4389 – 92.