Metabolite Profiles from Dried Biofluid Spots for Metabonomic

Nicola Gray , Matthew R. Lewis , Robert S. Plumb , Ian D. Wilson , and Jeremy K. Nicholson. Journal of Proteome Research 2015 14 (6), 2714-2721...
0 downloads 0 Views 665KB Size
Metabolite Profiles from Dried Biofluid Spots for Metabonomic Studies using UPLC Combined with oaToF-MS Filippos Michopoulos,†,‡ Georgios Theodoridis,‡ Christopher J. Smith,† and Ian D. Wilson*,† Department of Clinical Pharmacology, Drug Metabolism and Pharmacokinetics, AstraZeneca, Mereside, Alderley Park, Macclesfield, Cheshire SK10 4TG, United Kingdom, and Laboratory of Analytical Chemistry, Aristotle University of Thessaloniki, 541 24 Greece Received February 9, 2010

Abstract: The potential of dried blood, plasma and urine spots, deposited on a paper substrate, combined with reversed-phase ultra performance liquid chromatography and orthogonal acceleration time-of-flight mass spectrometry (UPLC-oaToFMS), with electrospray ionization (ESI), has been investigated for global metabolic profiling. When compared to analysis using protein precipitated plasma, both blood and plasma spots gave comparable profiles and numbers of ions etc., using both positive and negative ESI. In the case of urine the results for spots obtained in positive ESI were also comparable to those of the untreated sample, but with negative ESI, a significant reduction in ions was noted for spotted samples. These preliminary data suggest that, with further optimization, biofluid spotting may provide an alternative to conventional methods for this type of work in suitable applications.

of proteins, and this is most often performed via solvent precipitation7-11 or, less frequently, solid phase extraction (SPE).12 An alternative method, which has been used for many years in clinical practice and is becoming increasingly popular for drug analysis (e.g.,13-17), is to collect blood samples on to special types of absorbent paper. The “blood spots” collected in this way are then dried, ready for subsequent extraction and analysis. Such an approach may be particularly useful for metabonomic studies involving multiple sample collection from small animals such as mice or rats, where large serial samples are neither practical nor ethical. The use of dried blood spots may also have advantages for studies in human populations, either for pediatric investigations (where “finger prick” samples are usual) or for large scale epidemiological work, where the preparation of blood plasma or serum, and its subsequent storage and transport, poses logistical difficulties. Here we have conducted preliminary studies on the potential of dried blood spot cards for use in the storage/sample preparation of whole blood, plasma, and urine in metabolite profiling studies.

Keywords: metabonomics • metabolomics • ultra performance liquid chromatography • mass spectrometry • dried blood • plasma and urine spots • biofluid analysis

Experimental Section

Introduction The use of HPLC and UPLC-MS to obtain the “global” metabolic profiles of biofluids and tissue extracts is becoming an increasingly important methodology for metabonomic and metabolomic studies.1-3 An important feature in obtaining these metabolite profiles is that sample preparation should be minimized, thereby avoiding biasing the result through losses, or selective enrichment, of particular compounds or classes of metabolites. In addition, any form of sample manipulation carries with it the potential for introducing variability. However, in the case of samples such as blood or blood plasma/serum, it is essential to remove cells and proteins before analysis by LC methods to avoid the otherwise rapid loss of chromatographic performance that would result. Samples such as urine or bile can be analyzed by LC-MS using, for example, reversedphase gradient separations on C-18 bonded stationary phases with little pretreatment other than centrifugation and dilution.4-6 However, the analysis of plasma or serum requires the removal * To whom correspondence should be addressed. Ian Wilson, e-mail [email protected], Tel 00 44 1625 513424, Fax 00 44 1625 516962. † AstraZeneca. ‡ Aristotle University of Thessaloniki.

3328 Journal of Proteome Research 2010, 9, 3328–3334 Published on Web 04/26/2010

Solvents and Reagents. The solvents used for LC-MS analysis (ACN, MeOH), of HPLC grade, and formic acid, of analytical grade, were purchased from Fisher (Loughborough, Leicestershire, U.K). Standard leucine enkephaline was purchased from Sigma-Aldrich (Gillingham Dorset, U.K.). Water (18.2MΩ) was obtained from a Purelab Ultra system from Elga (Bucks, U.K.). The test mixture, used to provide a preliminary check on column performance, containing theophylline, caffeine, hippuric acid and nortriptyline, was from Waters (Waters Corp, Milford, MA). Sample Preparation. All biofluid samples were spotted onto non treated paper cards (ID Biological Systems 226, Abbots Langley, U.K.), allowed to dry at room temperature and cored with a BDS600 DUET II semi automated punch system (BDS ROBOTICS, Abbots Langley, U.K.). In the case of blood samples, 5 mL of whole blood were obtained from a healthy male Wistar-derived rat and placed into a lithium heparin blood tube. For analysis multiple 20 µL aliquots of blood were spotted onto the paper resulting in spots of 7 mm diameter, which were left to dry at room temperature prior to extraction. For analysis, sets of five 3.2 mm cores (equivalent to 20 µL of whole blood) were obtained, with each core obtained from the center of a separate blood spot (all five cores used to prepare each sample were obtained from samples 10.1021/pr100124b

 2010 American Chemical Society

technical notes

Metabolite Profiles from Dried Biofluid Spots spotted onto the same paper card). Each set of five cores was then placed into an Ependorff tube and extracted at vortex for 20 min using 160 µL of 25% (v/v) aqueous MeOH. The extracts were centrifuged for 10 min at 20 800× g (Centrifuge 5417C, Eppendorf, Hamburg, Germany), and then 120 µL of each clear supernatant were placed one by one into a 2 mL HPLC vial with a 0.1 mL glass insert (Kinesis, Cambs, U.K.) and then reduced to dryness at 40 °C under a stream of dry nitrogen. For analysis, samples were resuspended in 240 µL of 50% aqueous MeOH (v/v) and centrifuged at 3500× g for 20 min immediately prior to LC-MS analysis. Plasma was obtained after centrifugation (3000× g for 20 min) of 1 mL of the rat blood sample used above for blood spotting. Multiple aliquots of 20 µL plasma were spotted onto the blood spot paper, giving spots of 1 cm diameter, which were then left to dry at room temperature prior to extraction. For extraction, five cores, equivalent to 20 µL of plasma (4 × 4.7 mm +1 × 3.2 mm), were obtained from the center of five individual plasma spots (all spotted on to the same piece of paper), and two sets of these multiple cores were then placed in individual Ependorff tubes and extracted using the blood spot protocol described above. Blood/plasma spot samples were also compared with protein precipitated plasma whereby aliquots (50 µL) of the rodent plasma samples were prepared by protein precipitation with 150 µL of cold methanol (-20 °C). Precipitated proteins were removed by centrifugation at 20 800× g for 10 min. Subsequently 120 µL of the supernatant was diluted with 60 µL of water, vortexed and centrifuged at 3500× g for 20 min prior to LC-MS analysis. A pooled human urine sample was used for these studies, prepared by mixing aliquots of 150 µL each from a set of 15 male and 15 female urines samples obtained from normal, healthy volunteers. For analysis, multiple 20 µL aliquots of this pooled urine sample were spotted onto the paper, resulting in spots of 1.3 cm diameter, which were left to dry at room temperature prior to extraction. For analysis, the extraction of a total of 8 × 4.7 mm cores was performed, with multiple cores obtained from each of three separate urine spots (maximum of three cores per spot) all spotted on to the same piece of paper. Five sets of these multiple urine spot cores were prepared, placed into Ependorff tubes and then extracted using 200 µL of 25% aqueous MeOH, with vortexing for 20 min. The extracts were centrifuged for 10 min at 20 800× g and then 150 µL of each of the clear supernatants were combined in one HPLC vial, with a glass insert, and evaporated one by one at 40 °C under a stream of dry nitrogen. The samples were then resuspended in 200 µL of H2O and then centrifuged at 3500× g for 10 min prior to UPLC-MS analysis. Urine samples extracted from spots were compared to untreated pooled urine samples prepared for analysis by centrifugation at 3500× g for 10 min and diluted 1/1 (v/v) with water. UPLC-MS. Sample analysis was performed by UPLC-MS using an ACQUITY UPLC System coupled to Q-TOF Micro mass spectrometer (Waters, Milford, MA). The reversed-phase gradient separations were performed on a 2.1 × 100 mm, 1.7 µm, ACQUITY UPLC BEH, column fitted with a 0.2 µm ACQUITY UPLC column in-line filter for urine analysis and a 2.1 × 5 mm Van Guard precolumn for plasma/blood analysis. All samples (10 µL) were analyzed with gradient elution (see below) and in a random order. The samples were maintained at 4 °C in the autosampler.

Plasma Analysis. For plasma/blood analysis, the column temperature was maintained at 50 °C. The binary solvent system was composed of acidified H2O (solvent A) and MeOH (solvent B) with 0.1% formic acid (v/v). A series of linear gradients, at a flow rate of 0.4 mL/min, were applied starting with 95% A from 0 to 0.5 min, changing to 60% A at 2.5 min, 30% A at 4.5 and finally rising to 100% B at 10 min. This solvent composition was held for 2 min before the column was returned to the starting condition and held for 2.5 min before the next injection. Prior to the start of the analysis a series of 10 “conditioning injections” of 20 µL protein precipitated plasma were performed (see refs 4, 5, and 12). Urine Analysis. For urine analysis, a column temperature of 40 °C was used. The binary solvent system was composed of acidified H2O (solvent A) and ACN (solvent B) with 0.1% formic acid. A series of linear gradients, at a flow rate of 0.4 mL/min, was applied starting with 95% A from 0 to 1.0 min, changing to 80% A at 5 min, 55% A at 8 min and finally 100% B at 9.8 min. This solvent composition was held for 1.7 min before the column was returned to the starting condition and held for 2.5 min before the next injection. Before the analysis of the samples, a series of five “conditioning injections” of 10 µL of the pooled urine sample were performed to equilibrate the system. ESI Mass Spectrometry. Mass spectrometry was performed using a Waters Micromass Q-TOF Micro (Milford, MA) operating in positive/negative ion electrospray mode. For positive acquisition, capillary and cone voltages were set 3.5 kV and 2.5 kV respectively and 28 V and 35 V respectively. The desolvation temperature was set to 370 °C and the source temperature to 120 °C. The cone gas was set to a flow rate of 0 L/h and the desolvation gas flow was maintained at 600 L/h. For mass accuracy, a LockSpray interface was used with leucine-enkephalin (556.2771/554.2615 amu) solution 2 mg/L was used, at 30 µL/min, as the lock mass. Full scan data were collected from 100 to 800 m/z over a period of 13 and 12 min for plasma/blood and urine, respectively. Data were acquired with MassLynx operating software (Waters) with a scan time of 0.3 s and an interscan delay of 0.1 s. Data Processing. The raw spectrometric data acquired were processed by MarkerLynx application manager (Waters). MarkerLynx uses ApexTrack peak integration to detect chromatographic peaks. The track peak parameters were set as follows: Peak width and peak-to-peak baseline noise were automatically calculated, minimum intensity 80, mass window 0.05 (Da), retention time window 0.2 min, noise elimination level 6, mass tolerance 0.05 (Da), mass range 100-800 amu and with exclusion of deisotopic data. The peak list data obtained by MarkerLynx were further processed by Simca-P 11 (Umetrics, Windsor, UK) for multivariate data analysis such as, for example, principal components analysis (PCA).

Results and Discussion As indicated, a range of biofluids including blood, blood plasma and urine were spotted on to the cards and then compared with either the untreated biofluid in the case of urine or protein-precipitated samples or plasma spots for blood. UPLC-MS in each case was performed using both positive and negative modes of electrospray ionization (ESI). The resulting Journal of Proteome Research • Vol. 9, No. 6, 2010 3329

technical notes

Michopoulos et al.

Figure 1. Representative BPI chromatograms of a dried plasma spot extract in positive (A) and negative (D) ESI, a dried blood spot extract in positive (B) and negative (E) ESI and protein precipitated plasma in positive (C) and negative (F) ESI.

data were then compared using multivariate statistical analysis using principal component analysis (PCA) as described below. Blood and Plasma Spots: Comparison with Proteinprecipitated Plasma. A typical base peak ion (BPI) trace for the analysis of protein precipitated plasma, analyzed using gradient reversed-phase UPLC in positive ESI, is shown in Figure 1C, while the equivalent traces for plasma and blood spots are shown in Figure 1A and B, respectively. A number of features are apparent from this comparison. Thus, all 3 sets of data are clearly rich in peaks with both blood spots and plasma spots showing many of the features visible in the protein precipitated plasma. In all 3 cases, the peaks eluting in the 3330

Journal of Proteome Research • Vol. 9, No. 6, 2010

range ca. 2-4 min represent polyethylene glycol contamination from the vials used to collect the blood, while examination of the blank extracts also revealed a number of peaks (e.g., peaks eluting between 4.5 and 6.5 min) as polypropylene glycol contaminants introduced to the spotted samples during the extraction process and chromatographic analysis. Nevertheless, the fingerprint of peaks corresponding to lipid ions, with retention times between, for example, 8 and 10 min (e.g., see peaks eluting at 8.35. 8.67, 8.84, 9.39 min etc.), showed considerable similarity across the sample types. In terms of repeatability, when the data for all five replicate analyses were examined, 62% of the 1239 ions detected for the blood spots

Metabolite Profiles from Dried Biofluid Spots

technical notes

Figure 2. PCA scores plot of samples derived from blood (A, C) and urine (B, D). Left hand side plots (A, B) are based on positive ESI data and right hand side plots (C, D) are for negative ESI data respectively.

using positive ESI showed a CV of less than 15%, compared to 20% for the 1039 ions detected in the extracted plasma spots, and 60% of the 954 ions observed in the protein precipitates. If a CV of 30% is used, the equivalent values are 75% for blood spots, 52% for plasma spots and 73% for protein-precipitated plasma. The data for these peaks are provided in supplementary Tables S1-3 (Supporting Information) for all three sample types (in constructing these tables, ions with two or more zero values among the 5 replicates were excluded). Analysis of these extracts in negative ESI again showed broadly similar profiles for all three sample types when examined as BPI traces (Figure 1D-F) with generally much lower levels of background interferences. Analysis in negative ion mode revealed many fewer ions than seen with positive ion mode; however, in terms of repeatability, analysis of the negative ESI data for all five replicate analyses showed that 66% of the 257 ions detected for the blood spots showed a CV of less than 15%, compared to 87% for the 122 ions detected in the extracted plasma spots and 75% of the 162 ions observed in the protein precipitated samples. If a CV of 30% is used, the equivalent values are 83% for blood spots, 95% for plasma spots and 86% for proteinprecipitated plasma (see supplementary Tables S4-6 (Supporting Information); as above, ions with two or more zero values among the 5 replicates were excluded). When the UPLC-MS results for the blood/plasma spots and protein precipitated samples were analyzed via PCA the sample types were seen to cluster together in groups for both the positive and negative ESI data (Figure 2A and C). Group differences for the positive ion UPLC-MS data were seen, influenced by the background interference peaks that were seen in the spotted samples (e.g., see peaks at 6.57 and 9.56 min Figure 1A and B) and relative differences in lipid-related ions assigned to peaks with retention times of 8.35 (m/z 520.34), 8.66 (m/z 496.33), 8.84 (m/z 522.35), 9.39 (m/z 524.37) respec-

tively. Representative data for a number of differentiating ions in positive ESI are presented in Table 1 (top), where ions with m/z 389.25, 365.13, 568.356, 544.34, 520.34, 518.30, 522.35, and 538.40 seem to differentiate between sample groups. In addition to these ions, the positive ESI data were carefully examined for differences related to phospholipids ions (e.g., 184.07, 496.33, 524.37, 703.58, 758.56, 806.56) between spotted and protein-precipitated samples and the results are given in Table 2. Most of these ions gave significantly higher peak area values for protein precipitated samples and dried blood spot derived samples compared to the equivalent plasma spots. Similar PCA of the negative ESI data gave the result shown in Figure 2C, with good grouping of the data by sample type. As mentioned above, background interferences were minor compared to those seen in positive ESI (see the ion eluting at ca. 9.6 min in Figure 1D and E). Representative data for a number of differentiating ions are given in Table 1 (bottom). Urine Spots: Comparison with Untreated Urine. As can be seen from the BPI traces of urine shown in Figure 3, both urine spots (Figure 3A) and untreated urine (Figure 3B) gave very similar profiles in positive ESI. The extracts contained similar numbers of ions compared to the untreated urines with 1109 versus 1280 detected respectively, with greater than 60% common to both. When the replicates for the positive ESI analysis of the urine were examined, some 9% of the 984 ions detected for the urine spots showed a CV of less than 15%. In the case of the untreated urine, 60% of the 1166 ions showed a CV of less than 15% (the corresponding values for CVs of 30% are 43% and 81% respectively). Analysis of these extracts in negative ESI (Figure 3C) however, showed considerable differences in the BPI traces compared to control urine (Figure 3D). Thus, for urine spots, most of peaks observed showed a significant loss of intensity compared to the control urine samples (Figure 3D). For example, the ions eluting at 2.99, 4.22, Journal of Proteome Research • Vol. 9, No. 6, 2010 3331

technical notes

Michopoulos et al.

Table 1. Retention, Average Peak Area and CV Values for Differentiating Ions (in Elution Order) from Positive and Negative ESI Blood Derived Dataa Positive ESI blood derived data m/z 389.25

PPT DBS DPS

m/z 365.13

m/z 544.34

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

4.93 4.93 4.93

196.19 435.37 537.14

5.93 3.13 31.37

6.57 6.57 6.57

3.27 1019.40 854.36

38.95 6.25 33.66

8.30 8.30 8.30

185.19 94.15 55.69

3.89 6.55 8.87

8.33 8.33 8.33

919.31 722.63 333.28

6.86 6.36 39.28

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

8.35 8.35 8.35

1331.12 951.19 440.55

0.59 10.83 4.58

8.70 8.70 8.70

902.76 671.99 474.59

3.71 6.08 4.03

8.84 8.84 8.84

770.19 821.13 333.48

3.74 7.46 45.89

9.60 9.60 9.60

375.81 203.27 78.24

18.31 2.43 33.21

m/z 520.34

PPT DBS DPS

m/z 568.35

m/z 518.30

m/z 522.35

m/z 538.40

Negative ESI blood derived data m/z 552.31

PPT DBS DPS

m/z 528.30

a

m/z 480.30

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

8.30 8.30 8.30

54.57 16.16 12.05

2.44 8.25 10.04

8.32 8.32 8.32

261.26 198.06 123.91

2.20 9.48 3.30

8.36 8.36 8.36

305.50 452.87 256.09

41.66 4.73 3.83

8.70 8.70 8.70

630.33 645.06 420.39

0.72 6.35 3.47

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

8.86 8.86 8.86

112.57 80.45 58.33

2.11 8.37 4.83

9.40 9.40 9.40

149.03 130.43 70.17

2.22 8.53 6.08

9.58 9.58 9.58

1.61 733.87 177.42

58.26 10.98 19.42

9.71 9.71 9.71

96.48 67.05 46.55

9.15 12.85 8.43

m/z 566.34

PPT DBS DPS

m/z 504.30

m/z 568.36

m/z 355.31

m/z 281.24

PPT, protein precipitated samples; DBS, dried blood spot; DPS, dried plasma spot. See also Tables S1-6, Supporting Information.

Table 2. Retention, Average Peak Area and CV Values for Representative Phospholipids Ions (in Elution Order) from Positive ESI Blood-Derived Dataa m/z 184.07

PPT DBS DPS

m/z 496.33

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

8.30 8.30 8.30

1683.33 1934.01 773.16

3.25 3.44 19.61

8.68 8.68 8.68

1853.15 1446.06 878.11

3.76 7.74 6.61

9.39 9.39 9.39

1424.04 1507.64 656.30

1.72 2.02 46.44

RT

average peak area

CV

RT

average peak area

CV

RT

average peak area

CV

11.25 11.25 11.25

98.16 336.63 96.77

30.36 3.08 15.07

11.40 11.40 11.40

855.67 769.88 420.19

13.31 7.26 27.87

11.40 11.40 11.40

922.10 682.59 311.98

1.48 6.52 14.19

m/z 703.58

PPT DBS DPS a

m/z 758.56

m/z 806.56

PPT, protein precipitated samples; DBS, dried blood spot; DPS, dried plasma spot.

3.63, 6.99, and 8.21 min were present at much reduced intensity in the spotted sample extracts. Peak area and CV values for these ions are given in Table 3 and show between 2 and 10 times higher values for the control urine samples. These results for negative ESI show that acidic compounds such as, for example, adipic acid (m/z 177.90, RT 3.63 min) appear to be poorly recovered by the current methodology. This could result from a number of causes including instability, leading to the loss of the compounds or their interacting strongly with the paper substrate such that they were not efficiently recovered by the present extraction protocol. In terms of repeatability, analysis of the negative ESI data for all five replicate samples showed that 70% of the 185 ions detected for the urine spots showed a CV of less than 15%, compared to 32% for the 584 ions detected in the control urine (the corresponding values for CVs of 30% are 8% and 76%, respectively). However, due 3332

m/z 524.37

Journal of Proteome Research • Vol. 9, No. 6, 2010

to low peak intensity, many of the features present in the spotted samples were close to the background noise level and were not found by the peak finding algorithm. As a result, only 30% of the total control urine ions were detected in the spotted urine samples. The CV data for the peaks present in control urine and spots for positive and negative ESI data are given in supplementary Tables S7 to S10 (Supporting Information; as with the blood/plasma-derived data, ions with two or more zero values among the 5 replicates were excluded). PCA of the urine-derived data gave the results shown in Figure 2B and D, with untreated and spotted urines clustering into distinct groups. However close examination of the loadings and S-plot (PLS-DA model) of the positive ESI data for urine did not reveal any particular ion as been important for differentiation between sample groups. As would be expected on the basis of the lower CVs seen for the negative ESI data

technical notes

Metabolite Profiles from Dried Biofluid Spots

Figure 3. Representative BPI chromatograms of dried urine spots in positive (A) and negative (C) ESI and untreated urine samples in positive (B) and negative (D) ESI modes.

Table 3. Retention, Average Peak Area and CV Values for Differentiating Ions (in Elution Order) from Negative ESI Urine-Derived Data RT

control urine 2.39 DUSa 2.39

m/z 149.90 average peak area

250.40 47.46

CV

RT

12.39 3.00 5.78 3.03

m/z 177.90 RT

control urine 4.81 4.90 DUSa a

average peak area

2960.66 657.81

m/z 172.85 average peak area

134.00 19.10

CV

RT

18.21 3.66 3.69 3.66

m/z 268.95 CV

RT

8.17 6.51 4.46 6.51

average peak area

148.16 15.09

m/z 177.90 average peak area

670.83 174.49

CV

RT

8.89 4.21 4.98 4.26

m/z 540.85 CV

RT

17.04 7.37 10.55 7.37

average peak area

320.34 62.63

m/z 286.80 average peak area

CV

20.07 2.99

22.65 9.97

m/z 268.95 CV

RT

45.73 8.21 59.61 8.21

average peak area

CV

462.67 70.05

13.37 6.08

DUS, dried urine spot. See also Tables S7-10 (Supporting Information).

Journal of Proteome Research • Vol. 9, No. 6, 2010 3333

technical notes for urine spots discussed above, when PCA analysis of these was performed the spots were rather more tightly clustered than the untreated urines (Figure 2D).

Discussion Our normal protocol for UPLC-MS of plasma uses 50 µL of plasma12 and results in the effective injection of ca. 1.6 µL of sample on column. When 20 µL of blood were spotted onto the paper the resulting spots were ca. 7 mm diameter compared to 1 cm spots when the same amount of plasma was spotted. The increased viscosity of the blood therefore resulted in more concentrated spots than those obtained for plasma. Similarly, urine gave spots of ca. 1.3 cm in diameter. In all of these cases, to obtain sufficient sensitivity, we combined the extracts of a number of cores for each sample and the generally reduced repeatability of the data for the spotted samples may reflect this rather than an inherent irreproducibility for biofluids spots. The somewhat better results seen for whole blood compared to plasma may reflect the more compact, and therefore more concentrated, nature of the spots obtained for the different sample types. However, as these data show, the results for blood spots compared well with those for protein precipitated plasma, for both positive and negative ESI, in terms of the profiles obtained and the number of ions detected. While the repeatability seen for the blood spots was not as good as that for the protein precipitated plasma it was overall acceptable, with 62% of the detected ions having CVs of less than 30% in positive ESI and 76% in negative mode. Similarly the plasma spots, while not having as good a repeatability as protein precipitated plasma or blood spots, also gave acceptable repeatability. While a number of areas, such as, for example, sample stability and appropriate storage conditions etc., remain to be investigated, the use of blood (or plasma) spots for metabonomics analysis seems, on the basis of these preliminary results, worthy of further investigation. The results for urine are also promising and appear particularly interesting with respect to highlighting the potential of the paper to interact with the sample. So, while the positive ion data for urine spots were very comparable to that of the untreated urine (albeit with poorer repeatability) there was a significant loss of analyte peaks when the negative ion data were compared with that obtained from the raw urine. Thus as well as the need to investigate sample stability and sample storage conditions for urine spots there may also be a need to improve the recovery of acidic analytes (assuming that the poor recoveries observed here are not due to analyte instability) if this methodology is to be used for urine. Obviously the application of this type of technology is not limited to only blood, plasma, and urine but might equally well be applied to other biofluids/tissue extracts. Indeed, further preliminary studies on rat bile suggest that biofluid spotting can equally well be applied to this sample type (data not shown). Because of sensitivity issues, the preliminary work here was based on the combination of multiple cores but with the application of more sensitive instrumentation; perhaps in

3334

Journal of Proteome Research • Vol. 9, No. 6, 2010

Michopoulos et al. combination with miniaturized separation systems (capillary or “chip”-based) such procedures would not be necessary.

Conclusions These preliminary results show that the use of dried blood, and other biofluid, spots represents an interesting alternative to conventional liquid samples for generating metabolite profiles. The use of dried blood spots in particular may represent an advantage in situations where samples are limited in volume (e.g., rodent samples), are already used in clinical practice (e.g., heel prick samples for young children), or in situations where the logistical problems involved in taking and transporting large volumes of samples could be problematic (e.g., epidemiological samples etc.).

Acknowledgment. We thank Mr. John Dinan of ID Biological Systems LLC for providing the untreated paper used in the study together with the loan of the BDS600 Duet 11. Supporting Information Available: Supplementary tables. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Lenz, E. M.; Wilson, I. D. J. Proteome Res. 2007, 6, 443–458. (2) Theodoridis, G.; Gika, H. G.; Wilson, I. D. TrAC, Trends Anal. Chem. 2008, 27, 251–260. (3) Wu, Z.; Huang, Z.; Lehman, R.; Zhao, C.; Xu, G. Chromatographia 2009, 69, S23–S32. (4) Gika, H. G.; Theodoridis, G. A.; Wingate, J. E.; Wilson, I. D. J. Proteome Res. 2007, 6, 3291–3303. (5) Gika, H. G.; Macpherson, E.; Theodoridis, G.; Wilson, I. D. J. Chromatogr., B 2008, 871, 299–305. (6) Want, E. J.; Wilson, I. D.; Gika, H.; Theodoridis, G.; Plumb, R. S.; Shockcor, J.; Holmes, E.; Nicholson, J. K. Nat. Protoc. 2010, 5, 1005-1018. (7) Sabatine, M. S.; Liu, E.; Morrow, D. A.; Heller, E.; McCarroll, R.; Wiegand, R.; Berriz, G. F.; Roth, F. P.; Gerszten, R. E. Circulation 2005, 112, 3868–3875. (8) Want, E. J.; O’Maille, G.; Smith, C. A.; Brabdon, T. R.; Uritboonthai, W.; Qui, C.; Trauger, S. A.; Siuzdak, G. Anal. Chem. 2006, 78, 743– 752. (9) Zelena, E.; Dunn, W.; Broadhurst, D.; Francis-McIntyre, S.; Carroll, K.; Begley, P.; O’Hagan, S.; Knowles, J. D.; Halsall, A.; HUSERMET Consortium, I. D.; Wilson, D.; Kell, D. Anal. Chem. 2009, 81, 1357– 1364. (10) Dunn, W. B.; Broadhurst, D.; Brown, M.; Baker, P. N.; Redman, C. W. G.; Kenny, L. C.; Kell, D. B. J. Chromatogr., B. 2008, 871, 288–298. (11) Bruce, S. J.; Jonsson, P.; Clorarec, C.; Trygg, J.; Marklund, S. L.; Moritz, T. Anal. Biochem. 2008, 372, 237–249. (12) Michopolous, F.; Lai, L.; Gika, H.; Theodoridis, G.; Wilson, I. J. Proteome. Res. 2009, 8, 2114–2121. (13) Barfield, M.; Spooner, N.; Lad, R.; Parry, S.; Fowles, S. J. Chromatogr., B 2008, 870, 32–37. (14) la Marca, G.; Malvagia, S.; Filipa, L.; Fiorini, P.; Innocenti, M.; Luceri, F.; Pieraccini, G.; moneti, G.; Frances, S.; Dani, F. R.; Guerrini, R. J. Pharm. Biomed. Anal. 2008, 48, 1302–1396. (15) Spooner, N.; Lad, R.; Barfield, M. Anal. Chem. 2009, 81, 1557–1563. (16) Abu-Rabie, P.; Spooner, N. Anal. Chem. 2009, 81, 10275–10284. (17) Van Berkel, G. J. Anal. Chem. 2009, 81, 9146–9152.

PR100124B