Efficient determination of purine metabolites in brain tissue and serum by high-performance liquid chromatography with electrochemical and UV detection

Abstract

The purine metabolic pathway has been implicated in neurodegeneration and neuroprotection. High-performance liquid chromatography (HPLC) is widely used to determine purines and metabolites. However, methods for analysis of multiple purines in a single analysis have not been standardized, especially in brain tissue. We report the development and validation of a reversed-phase HPLC method combining electrochemical and UV detection after a short gradient run to measure seven purine metabolites (adenosine, guanosine, inosine, guanine, hypoxanthine, xanthine and urate) from the entire purine metabolic pathway. The limit of detection (LoD) for each analyte was determined. The LoD using UV absorption was 0.001 mg/dL for hypoxanthine (Hyp), inosine (Ino), guanosine (Guo) and adenosine (Ado), and those using coulometric electrodes were 0.001 mg/dL for guanine (Gua), 0.0001 mg/dL for urate (UA) and 0.0005 mg/dL for xanthine (Xan). The intra- and inter-day coefficient of variance was generally <8%. Using this method, we determined basal levels of these metabolites in mouse brain and serum, as well as in post-mortem human brain. Peak identities were confirmed by enzyme degradation. Spike recovery was performed to assess accuracy. All recoveries fell within 80–120%. Our HPLC method provides a sensitive, rapid, reproducible and low-cost method for determining multiple purine metabolites in a single analysis in serum and brain specimens.

Introduction

A growing body of evidence supports an important role of the purine metabolic pathway (Fig. 1) in various neurological disorders including brain injury, Parkinson’s disease (PD) and other neurodegenerative diseases (Burnstock, 2008). Adenosine (Ado) is well known to modulate neuronal and synaptic function through its A1 and A2 receptors (Stone, 2005; Schwarzschild et al., 2006). Inosine (Ino) has been shown to be neuroprotective either directly (Irwin et al., 2006) or indirectly through metabolic conversion to downstream metabolites (Gomez and Sitkovsky, 2003). Similarly, guanine (Gua)-based guanosine (Guo) is implicated as a modulator of neural function (Schmidt et al., 2007). Hypoxanthine (Hyp) and xanthine (Xan) have been linked to glutamate-mediated excitotoxicity (Stover et al., 1997) and oxidative stress (Quinlan et al., 1997), and a recent study implicated a potential role of Xan as a biomarker of PD (LeWitt et al., 2011). Remarkably, a convergence of laboratory and epidemiological data has recently identified urate (UA), the enzymatic end product of purine degradation in humans, as a molecular predictor of both risk and progression of PD and as a candidate neuroprotectant for the treatment of PD (Ascherio et al., 2009; Cipriani et al., 2010). Therefore, extensive detection and quantification of the purine degradation pathway metabolites in brain may provide insight into their relevance to different physiological and pathological conditions.

Figure 1.

Purine degradation pathway. Adonosine is converted to inosine through the removal of the amine moiety by adenosine deaminase, and inosine is degraded to hypoxanthine through the removal of phospho-1-ribose by purine nucleoside phophorylase. Guanosine is converted to guanine via the action of purine nucleoside phosphorylase, and guanine is then degraded to xanthine through the action of guanine deaminase. In the presence of xanthine oxidase, hypoxanthine and xanthine are converted to urate. Urate constitutes the end product of purine catabolism in humans owing to lack of urate oxidase activity.

High-performance liquid chromatography (HPLC) has been the prevalent method of measuring nucleotides, nucleosides and major purine bases in different biological samples (Bakay et al., 1978; Nissinen, 1980; Ryba, 1981; Zakaria and Brown, 1981; Iriyama et al., 1984; Wynants and Van Belle, 1985; Smolenski et al., 1990; Liu et al., 1995; Takahashi et al., 2010; Struck et al., 2011). Although many of those HPLC-based protocols are capable of separating and quantifying multiple purines, they often demand a large injection volume and long retention time, and have low throughput and relatively low sensitivity. The ability to measure much of the purine degradation pathway in a single analysis may prove to be a valuable tool in understanding its role in human diseases like PD, as well as in their animal models. A single-run analysis may lead to a better measurement of the purine pathway by eliminating the potential variation inherent in measuring analytes using separate analyses.

Towards this goal, we describe the development of a dual-pump gradient HPLC method using UV and electrochemical detection (ECD). This method achieves suitable separation and sensitivity in a short run time and is capable of measuring seven purine metabolites in tissue and serum with minimal sample preparation and high throughput.

Experimental protocol

Collection of mouse and human tissues

Mice 10 to 12 weeks old C57BL/6 (J) and weighing 29±1.3 g were obtained from Jackson Laboratories (Bar Harbor, ME, USA). They were kept under standard conditions (temperature 21±2°C, humidity 30–70%, 12 h light–dark cycle) and with water and standard pellet feed ad libitum. Mouse whole blood was collected via a lancet (Goldenrod Animal Lancet, Mineola, NY, USA) puncture of the submandibular vein. The mice were killed via cervical dislocation, and the brain was removed. The striatum of each hemisphere were collected separately and all tissue samples were frozen on dry ice. Postmortem human brain samples were obtained from the MassGeneral Aging and Disability Resource Center/Harvard NeuroDiscovery Center neuropathology core B repository in accordance with institutional, state and federal regulations, as well as the wishes of the families of donors. Fresh frozen tissue (stored at −80°C) samples were collected from 10 male control brains, defined as those without evidence of neurodegenerative disease (such as Parkinson, Huntington or Alzheimer’s disease) and with postmortem interval <24 h (mean 19.8±6.0) and age limited to >50 years (mean 82.0 11.2). Tissue samples (~100–200 mg) were dissected on dry ice from striatum. All tissue was kept frozen at −80°C until processed for HPLC analysis.

Instrumentation

The reversed-phase HPLC system comprised two pumps, a model 584 and a model 582 isocratic pump, feeding a high-pressure gradient mixer. Samples were injected using a model 524 autosampler with a 100 μL sample loop and analysis was performed using a model 5600A CoulArray with a 528 UV–vis detector followed by two model 5011A coulometric cells. All equipment was obtained from ESA Biosciences (Chelmsford, MA, USA). Analyte separation was achieved using a batch-tested Varian Microsorb-MV reversed-phase C₁₈ column (150 × 4.6 mm i.d., 5 μm, 100 A; Varian Inc., Palo Alto, CA, USA).

Chemicals and reagents

Acetonitrile (HPLC-grade), potassium phosphate monobasic (HPLC-grade) and EDTA (electrophoresis grade, ≥99%) were supplied by Fisher Chemical (Pittsburgh, PA, USA). Ado, Guo, Gua, Hyp, Ino, UA, Xan and 3,4-dihydroxybenzylamine (DHBA) standards (≥99%) were supplied by Sigma Aldrich (St Louis, MO, USA). Sodium 1-pentanesulfonate (≥99%) and methyl-DOPA sesquihydrate (MD, ≥99%) were obtained from Fluka Analytical (Sigma-Aldrich). Double-distilled water was obtained from a Milli-Q Water System (Millipore, Billerica, MA, USA). All water was subsequently passed through a C₁₈ Maxi-Clean cartridge (Alltech, Deerfield, IL, USA) to remove any potential organic contaminants.

HPLC operating parameters

A dual mobile phase gradient was used to achieve appropriate separation of all analytes of interest. Mobile phase A contained 0.52 mm sodium 1-pentanesulfonate and 0.20 m KH₂PO₄ monobasic at pH 3.5 using 85% phosphoric acid (HPLC-Grade, Fisher Scientific, Pittsburgh, PA, USA). Mobile phase B had the same final concentrations as mobile phase A, except for the addition of 10% acetonitrile (v/v). The gradient composition is shown in Fig. 2. The flow rate was 1.0 mL/min, and the system was allowed to equilibrate at that flow rate for 15 min prior to the first sample injection. The sample injection volume was 12 μL.

Figure 2.

Mobile phase gradient paradigm. Mobile phase A: 0.2 m KH₂PO₄ monobasic, 0.52 mm sodium 1-pentanesulfonate, pH 3.5. Mobile phase B: 0.2 m KH₂PO₄ monobasic, 0.52 mm sodium 1-pentanesulfonate, 10% acetonitrile, pH 3.5. MP: Mobile phase.

The detectors were linked in series, with the Model 528 UV–vis light spectroscopy spectrophotometer upstream of both electrochemical cells. UV–vis detection was set to a wavelength of 254 nm. The first electrode was set to 0.10 V and acted as a conditioning cell. The analytical electrodes 1 and 2 were set at +0.15 and +0.45 V, respectively. Data were collected using CoulArray Data Station 3.0 software (ESA Biosciences) with auto-range gain enabled.

Preparation of stock solutions and standards

Individual purine stock solutions were dissolved in double-distilled water that had been filtered through a C₁₈ Maxi-Clean cartridge to a final concentration of 1.0 mg/mL except for UA, which was made at a stock concentration of 0.5 mg/mL owing to its solubility. Aliquots of the stocks were stored at −80°C until needed. A working mixed purine standard curve was created by serial dilutions of purine stocks in PE buffer containing 50 mm phosphoric acid, 0.1 mm EDTA, 50 μm MD and 1 μM DHBA (internal standards) from 1.0 mg/mL purine stocks. The working standard curve (except in limit of detection experiments) ranged from 1.0 to 0.001 mg/dL for all purines.

Preparation of mouse and human brain samples for purine analysis

Brain samples were weighed at −60°C and immediately homogenized on ice using a Teflon pestle in 20× volume (v:w) of PE buffer. Extracted samples were then centrifuged at 16,000g for 15 min. The supernatant was then removed and filtered through a 0.22 μm Spin-X Cellulose Acetate filter tube (Corning, NY, USA) at 16,000g for 5 min. Resulting filtrate was stored at −80°C until needed.

Preparation of mouse serum for purine analysis

Whole blood was collected and centrifuged at 16,000g for 15 min. The serum was then transferred and stored at −80°C until needed. Serum deproteination was achieved by the addition of 30 μL of 0.4 m perchloric acid to 50 μL of serum and vortexing briefly. The solution was allowed to incubate on ice for 10 min prior to centrifugation at 1400g for 15 min. The resulting supernatant was removed and added to 20 μL 0.2 m potassium phosphate (pH 4.75) with 1 μm DHBA (internal standard). The resulting solution was filtered through a 0.22 μm Spin-X Cellulose Acetate filter tube at 16,000g. The resulting filtrate was stored at −80°C until needed.

Enzyme degradation

To confirm peak identity, enzyme degradation was performed. Mixed purine standards and mouse brain samples were prepared in 0.2 m potassium phosphate monobasic (pH 7.75). Standards and samples were then incubated with the following individual enzymes: adenosine deaminase, purine nucleoside phosphorylase, xanthine oxidase and urate oxidase (all purchased from Sigma-Aldrich, St Louis, MO, USA). Reaction conditions were 25°C overnight for all enzymes, and concentration of each enzyme was predetermined to be sufficient to completely eliminate the target analyte over the overnight incubation period. The resulting mixtures were centrifuged for 15 min at 15,000 rpm, the supernatant was then filtered through a 0.22 μm Spin-X Cellulose Acetate filter tube (Corning, NY, USA) at 16,000g for 5 min. The resulting filtrate was stored at −80°C until needed.

Spike recovery

Spike recovery experiments were performed to validate the accuracy of the method. Purine standards and mouse serum and brain samples were prepared. Baseline values of each analyte per sample were detected. Stock solutions were then made at 5 times the basal concentrations. Each experiment consisted of a sample control, spike control and spiked sample, all of which were individually made to 60 μL to allow for triplicate runs at 20 μL each. Sample plus mobile phase A (in the amount of the spike) constituted the sample control. The spike control had a specified volume of stock that resulted in 5 times the basal analyte levels plus mobile phase A. The spiked sample included the necessary spike amount of stock and sample. Recovery percentage was calculated by comparing the spiked sample analyte values to the analyte values of the sample control plus spike control levels.

Results and discussion

The main goals of this method were to obtain suitable separation and high sensitivity of seven purine metabolites with a single injection and short run time, allowing for high-throughput analysis of biological samples. This reversed-phase chromatographic method was built upon previous isocratic methods using ECD of UA and Xan (Iriyama et al., 1984; Liu et al., 1995) and underwent optimization of pH and an ion-pairing agent parameters to ensure adequate separation of the analytes of interest. We also took advantage of the differential selectivity of UV and electrochemical detectors for the major purines in biological samples to achieve better signal separation than previously observed.

Determination of electrode potentials and UV–vis wavelength

Hydrodynamic voltammograms were obtained for UA, Xan and Gua to determine the optimum oxidizing potentials for each analyte (Fig. 3). The oxidation of UA increased with greater voltages, reaching a plateau near +0.1 V. A slightly higher potential of +0.15 V (P₁) was chosen to ensure that UA was being fully oxidized, while avoiding oxidation of other similarly retained analytes that might have obscured the UA signal. Oxidation of Gua and Xan reached a plateau at +0.45 V (P₂), at which no co-eluting UA peak interfered with the Gua measurement (data not shown), suggesting that the upstream electrode set at 0.15 V potential had fully oxidized UA. Thus, +0.15 and +0.45 V were chosen as the analytical potentials because they provided full oxidation of the analytes, while avoiding co-oxidation of Gua and UA, which have very similar retention times. A pre-analytical electrode was set to −0.1 V (P₀) to oxidize any potential contaminants that are more easily oxidized than UA. The −0.1 V potential was chosen because more positive potentials partially oxidize UA (Fig. 3), which would weaken the measurable signal at the analytical +0.15 V electrode.

Figure 3.

Hydrodynamic voltammogram curves for urate, guanine, and xanthine. Measurements were taken using a model 5011A coulometric cell. An analytical potential of +0.15 V (P₁) was selected for urate, and an analytical potential of +0.45 V (P₂) was selected for guanine and xanthine. A conditioning potential of −0.1 V (P₀) was chosen to minimize contaminant peaks.

Hyp, Ado, Ino and Guo were detected at a UV wavelength of 254 nm. UA, Xan and Gua were also detectable at this UV wavelength, but electrochemical detection provided a considerably lower limit of detection (LoD; Table 1). This advantage becomes apparent when measuring UA in brain tissue, in which UA concentrations are >5-fold lower than in serum (Cipriani et al., 2010). UA and Gua also have very similar retention times, leading to co-elution and considerably overlapping peaks in UV detection that are easily avoided through electrochemical potential manipulation as described above.

Table 1.

Analytical parameters of merit for purine chromatographic peaks

Analytes	Retention time (min)	Method of detection	Limit of detection (mg/dL)	Standard range (mg/dL)	Slope-intercept	R 2
Guanine	4.52	EC (+0.45 V)	0.001	0.001–1.0	y = 178.05x − 0.558	0.9999
Urate	4.78	EC (+0.15 V)	0.0001	0.0001–5.0	y = 131.76x + 0.5998	1
Hypoxanthine	5.46	UV	0.001	0.001–5.0	y = 10.928x − 0.0001	1
Xanthine	6.90	EC (+0.45 V)	0.0005	0.0005–5.0	y = 38.729x + 1.7033	0.9977
Inosine	11.10	UV	0.001	0.001–5.0	y = 7.8419x + 0.0708	1
Guanosine	11.30	UV	0.001	0.005–1.0	y = 5.3857x + 0.0048	1
Adenosine	12.10	UV	0.001	0.001–5.0	y = 9.9976x + 0.0725	1
DHBA (IS)	3.38	EC (+0.15 V)	—	—	—	—
MD (IS)	8.76	EC (+0.15 V)	—	—	—	—

DHBA, 3,4-Dihydroxybenzylamine; IS, internal standard; MD, methyl-DOPA.

Mobile phase and gradient development

Chromatographic baseline resolution of the analytes of interest was achieved through the manipulation of mobile phase composition and a gradient of organic solvent. The original mobile phase was adapted from Iriyama et al. (1984). Determination of the appropriate pH was performed through the measurement of retention times of all the analytes across a range of mobile phase pH values (Table 2). All other components of the mobile phase were kept constant through the pH calibration. pH dependencies of purine retention times were consistent with their respective values of pK_a in the pH range studied. For example, the greatest drop in the retention time of UA (pK_a at 5.4) occurred as pH was increased from 5 to 6, as expected given the increasing likelihood of the anionic urate form, which in contrast to neutral protonated form of urate is not retained on the hydrophobic interaction column. Conversely, Ado (pK_a 3.5) showed a markedly longer retention time as the pH was raised between 3 and 4, consistent with its loss of a proton to become neutral adenosine. Owing to poor separation between UA and Hyp below pH 3, and between Xan and Hyp at the higher pHs tested, a pH of 3.5 was selected for routine use.

Table 2.

Retention times of purine metabolites vs pH and concentration of ion-pairing agent during method development

Chromatographic conditions	Retention time (min)
pHa	Urate	Hypoxanthine	Xanthine	Inosine	Adenosine	MD
2.5	4.0	4.1	4.7	9.1	11.7	—
3.0	3.8	4.1	4.5	9.0	13.3	11.8
4.0	3.8	4.2	4.6	9.0	21.2	5.2
4.5	3.7	4.2	4.6	9.1	24.5	4.7
5.0	3.4	4.2	4.6	9.1	26.1	4.4
6.0	2.7	4.2	4.5	9.1	—	4.3
7.0	2.6	4.1	4.1	8.8	—	4.2
Ion-pairing agentb
0.5 mm 1-Pentanesulfonate	4.3	4.6	5.3	11.8	14.4	—
1.5 mm 1-Pentanesulfonate	3.8	4.0	4.6	9.1	10.8	—
1.5 mm 1-Octanesulfonate	3.9	4.2	4.7	9.8	14.6	—

^aRetention times with varying pH determined using 1.5 mm 1-pentanesulfonate.

^bRetention times with varying ion-pairing agents/concentrations determined at pH 3.5.

MD, Methyl-DOPA.

After the optimal pH was determined, various concentrations of several ion-pairing agents were introduced into the mobile phase to manipulate retention time and individual peak shape. The retention times produced by the various ion-pairing agents and concentrations are shown in Table 2. It was determined that 0.5 mm 1-pentanesufonate produced the best peak symmetry and baseline separation of the variations.

In an attempt to keep analysis times short and throughput high, a gradient was introduced to elute Ado, Guo and Ino more quickly. Under the isocratic conditions these analytes eluted far later than any of the other analytes of interest. Their late elution led to excessive band spreading, contributing to a considerable loss of sensitivity. By using a gradient, these analytes were eluted sooner and with a sharper peak shape than was possible using an isocratic method (Fig. 4a). Optimization of the gradient percentage organic and ramp times was performed to produce the shortest run time possible with a clear chromatographic baseline resolution of the closely eluting Ado, Ino and Guo peaks, without affecting the resolution of earlier analytes.

Figure 4.

Chromatograms of 1 mg/dL standards mixture (a), mouse serum (b), mouse striatum (c) and human striatum (d). Detection of analytes was performed either by ECD at +0.15 V (P1), +0.45 V (P2), or UV–vis at 254 nm. Gua, Guanine; UA, urate; Hyp, hypoxanthine; Xan, xanthine; Ino: inosine; Guo: guanosine; Ado: adenosine. Internal Standards are 3,4-dihydroxybenzylamine (DHBA, 1 μm) and methyl-DOPA (MD, 50 μm).

Method validation

The validity of the method was assessed through determination of the limit of detection, calculation of the linearity and variation between separate standard curves, calculation of inter-/intra-day coefficient of variation (CV). Additionally peak identity and method accuracy were determined and are discussed together with biological sample results.

The LoD was defined as the lowest concentration of each analyte whose peak height exceeded 3 times the height of the average baseline noise. No analyte detected by UV was measurable below 0.001 mg/dL, while Xan and UA measured by electrochemical detection were measurable at 0.0005 and 0.0001 mg/dL, respectively (Table 1). Standard curves containing all the analytes of interest were then run in triplicate and the mean of these three curves was used to determine variation, the slope–intercept formula, and the R² for each analyte (Table 1). All standard curves had very little variation, with the greatest deviation coming from the Xan curve measured by ECD with an R² of 0.998.

The method detection limit (MDL) was also determined for each purine analyte. Eleven sequential runs of freshly prepared 0.005 mg/dL concentration standards were analyzed. The MDL for the Ado values was the highest of the analytes, at 0.0018 mg/dL, with a standard deviation of 0.0006 mg/dL. The MDL for the method is set at that value to ensure that all other analytes can be assayed with at least 99% confidence. Therefore, the limit of quantification (LOQ) of our method is 0.006 mg/dL of analyte, which is 10 times the Ado SD value.

Intra- and interday coefficient of variance percentages (CV) were derived from standard solutions prepared at concentrations of 1.0, 0.1 and 0.005 mg/dL analyte. The mean, standard deviation and CV were calculated (Table 3). The intraday CV experiment was performed by running three standard samples of each concentration at three different time points a day. The intraday variation CV for all analytes was below 10%. Interday CV was assessed by repeating the intraday experiment over the subsequent 2 days, utilizing the same standard solutions and time points as on the first. The first day of CV experiments (intraday assessment) was included in the interday CV calculations, totaling three days of data. Only Xan at 1.0 mg/dL had a CV that exceeded 10%. The increased variation in the Xan measurement is most likely due to variation in the baseline associated with the initiation of the gradient. The gradient begins at approximately the same time that Xan elutes, and causes a small artifact peak that introduces some variability into the Xan measurement that is not present in the measurements of all the other analytes (Table 3).

Table 3.

Intra- and inter-day coefficient of variation

Analytes	Intra-day (the area under the peak)						Inter-day (the area under the peak)
	0.005 mg/dL		0.1 mg/dL		1.0 mg/dL		0.005 mg/dL		0.1 mg/dL		1.0 mg/dL
	x ± SD	CV (%)	x ± SD	CV (%)	x ± SD	CV (%)	x ± SD	CV (%)	x ± SD	CV (%)	x ± SD	CV (%)
Urate	0.48 ± 0.03	6.5	9.41 ± 0.26	2.7	81.51 ± 0.93	1.1	0.48 ± 0.03	6.5	9.41 ± 0.29	3.0	82.77 ± 3.87	4.7
Xanthine	1.08 ± 0.07	6.6	23.27 ± 0.65	2.8	159.01 ± 1.21	0.8	1.08 ± 0.07	6.6	23.32 ± 0.72	3.1	167.57 ± 25.73	15.4
Hypoxanthine	0.11 ± 0.01	7.4	2.25 ± 0.04	1.6	21.79 ± 0.11	0.5	0.11 ± 0.01	7.4	2.25 ± 0.04	1.7	21.83 ± 0.17	0.8
Inosine	0.07 ± 0	2.9	1.41 ± 0.01	0.8	13.14 ± 0.11	0.9	0.07 ± 0	2.9	1.41 ± 0.01	0.8	13.27 ± 0.41	3.1
Adonosine	0.09 ± 0	3.6	1.93 ± 0.02	1.0	19.36 ± 0.09	0.5	0.09 ± 0	3.6	1.93 ± 0.02	0.8	19.29 ± 0.25	1.3
Guanine	0.73 ± 0.07	9.7	20.13 ± 1.96	9.7	218.89 ± 2.02	0.9	0.73 ± 0.07	9.7	19.53 ± 0.39	2.0	218.44 ± 2.31	1.1
Guonosine	0.07 ± 0	5.3	1.57 ± 0.02	1.1	16.86 ± 0.14	0.8	0.07 ± 0	5.3	1.57 ± 0.01	0.8	16.76 ± 0.33	2.0

CV, Coefficient of variation.

Measurement of purine metabolites in tissue and serum

The ultimate goal of this method was to obtain sufficient analyte separation and sensitivity to allow measurement of the purine metabolites of interest in mouse and human tissues, including brain and serum. After appropriate separation was achieved with standard mixtures, this method was applied to mouse serum and brain tissue (Fig. 4 b and c).

C57BL/6 mice were killed, and blood was collected. Serum was then analyzed after deproteination with perchloric acid. Mouse brains were extracted, their striatum were collected and purine analysis was performed. The concentrations were determined from the mean of 10 male mice. Except for guanine, all concentrations of metabolites of interest were considerably above their LoD (Table 1), allowing for accurate measurements of each analyte (Table 4). Routine application of this method later to measurement of purines in brain and serum in mice across different experiments has been consistently producing comparable basal level values, allowing direct comparison and data pooling.

Table 4.

Basal levels of purine metabolites in mouse serum and striatum, and human striatum

Tissue	Urate	Xanthine	Hypoxanthine	Inosine	Adenosine	Guanine	Guanosine
Mouse serum (mg/dL)	1.03 ± 0.001	0.32 ± 0.003	0.13 ± 0.004	0.22 ± 0.001	0.03 ± 0.001	0.003 ± 4.2 × 10−5	0.08 ± 0.002
Mouse striatum (ng/mg wet tissue)	0.46 ± 0.07	1.54 ± 0.19	2.25 ± 0.33	33.98 ± 4.90	190.4 ± 25.63	0.044 ± 0.006	69.62 ± 8.95
Human striatum (ng/mg wet tissue)	7.76 ± 0.89	26.80 ± 2.43	167.9 ± 5.64	89.83 ± 11.18	1.61 ± 0.28	—	—

This method was then applied to the analysis of post-mortem male human striatum (n = 10) processed in a similar manner to the mouse tissue (Table 4). We are aware that more work needs to be done to take postmortem interval into account when analyzing the final values. Nevertheless, the chromatogram of human tissue showed a satisfactory separation and sensitivity (Fig. 4d).

Two strategies were employed to further confirm peaks in biological samples and rule out peak contamination. First, we slightly altered the acetonitrile concentration in both mobile phases to change the analyte retention time in multiple runs of the same sample. Standards and sample analyte retention time changes matched, while analyte concentrations were held constant over different mobile phases (data not shown). Secondly, we performed enzyme degradation studies to eliminate analytes of interest. These studies once again confirmed analyte peaks and negligible underlying contamination. Incubation with urate oxidase, for example, eliminated 91% of urate peak value, and xanthine oxidase eliminated 98% of xanthine and 100% of hypoxanthine peak values.

To further validate the accuracy of our method, spike recovery was performed using mouse serum and brain samples. All recoveries fell within 80–120%. Mouse serum spike recoveries were 80.52% (UA), 103.83% (Xan), 96% (Hyp), 96.51% (Ino), 100.47% (Ado), 118.49% (Gua) and 89.8% (Guo). Mouse striatal spike recoveries were 95% (UA), 99% (Xan), 96% (Hyp), 86% (Ino), 93% (Ado), 84% (Gua) and 99% (Guo).

In conclusion, we have characterized an efficient method of separating seven purine metabolites using HPLC with dual-pump gradient and quantifying them using a combination of electrochemical and UV detection. This method has been validated to provide satisfactory sensitivity, specificity, accuracy and consistency for measurement in biological samples. The power of this method is its short run time and high sensitivity, both of which allow for high-quality and high-throughput analysis of biologically relevant tissue samples, with minimal variation between runs or days. The value of these technical refinements for neuroscience research is increasing with renewed interest in the neurobiology of purines in health and disease. Future efforts will include method development for measurement of allantoin, a nonezymatic oxidation product of UA and therefore an index of oxidative stress in humans (Marklund et al., 2000; Zitnanová et al., 2004) to advance our ability to assess the role of purine metabolic pathway in neurodegeneration and neuroprotection.

Abbreviations used:

Ado

adenosine

DHBA

3,4-dihydroxybenzylamine

Gua

guanine

Guo

guanosine

Hyp

hypoxanthine

Ino

inosine

MD

methyl-DOPA sesquihydrate

PD

Parkinson’s disease

UA

urate

Xan

xanthine