Electrophoretic Analysis of Biomarkers using Capillary Modification

Jul 16, 2010 - Innovative Chromatography Group, Irish Separation Science Cluster (ISSC), Department of Chemistry & the ABCRF,. University College Cork...
0 downloads 0 Views 744KB Size
Download PDF
Anal. Chem. 2010, 82, 6895–6903

Electrophoretic Analysis of Biomarkers using Capillary Modification with Gold Nanoparticles Embedded in a Polycation and Boron Doped Diamond Electrode Lin Zhou,† Jeremy D. Glennon,† and John H. T. Luong*,†,‡ Innovative Chromatography Group, Irish Separation Science Cluster (ISSC), Department of Chemistry & the ABCRF, University College Cork, Cork, Ireland and Biotechnology Research Institute, National Research Council Canada, Montreal, Quebec, Canada H4P 2R2 Field-amplified sample stacking using a fused silica capillary coated with gold nanoparticles (AuNPs) embedded in poly(diallyl dimethylammonium) chloride (PDDA) has been investigated for the electrophoretic separation of indoxyl sulfate, homovanillic acid (HVA), and vanillylmandelic acid (VMA). AuNPs (27 nm) exhibit ionic and hydrophobic interactions, as well as hydrogen bonding with the PDDA network to form a stable layer on the internal wall of the capillary. This approach reverses electro-osmotic flow allowing for fast migration of the analytes while retarding other endogenous compounds including ascorbic acid, uric acid, catecholamines, and indoleamines. Notably, the two closely related biomarkers of clinical significance, HVA and VMA, displayed differential interaction with PDDA-AuNPs which enabled the separation of this pair. The detection limit of the three analytes obtained by using a boron doped diamond electrode was ∼75 nM, which was significantly below their normal physiological levels in biological fluids. This combined separation and detection scheme was applied to the direct analysis of these analytes and other interfering chemicals including uric and ascorbic acids in urine samples without off-line sample treatment or preconcentration. Indoxyl sulfate (IXS), a metabolite of tryptophan (TRP) and a dietary protein, is derived from intestinal metabolism and liver conjugation1 and excreted in urine in high concentration. It is also an endogenous compound in mammals, in mouse plasma and brain samples, as detected by liquid chromatography/tandem mass spectrometry.2 This circulating protein-bound uremic toxin stimulates glomerular sclerosis, interstitial fibrosis, and the progression rate of renal failure. IXS induces endothelial dysfunction by inhibiting endothelial proliferation and migration in vitro * To whom correspondence should be addressed. † Innovative Chromatography Group, Irish Separation Science Cluster (ISSC), Department of Chemistry & the ABCRF, University College Cork. ‡ Biotechnology Research Institute, National Research Council Canada. (1) Dealler, S. F.; Hawkey, P. M.; Millar, M. R. J. Clin. Microbiol. 1988, 26 (10), 2152–2156. (2) Wang, G.-F.; Korfmacher, W. A. Rapid Commun. Mass Spectrom. 2008, 23 (13), 2061–2069. 10.1021/ac101105q  2010 American Chemical Society Published on Web 07/16/2010

and its role in oxidative stress is implicated.3 Homovanillic acid (HVA) is a major catecholamine metabolite, associated with the brain dopamine level. As a biomarker of metabolic stress of 2-deoxy-D-glucose, the HVA level in the brain and the cerebrospinal fluid is indicative of pheochromocytoma and neuoroblastoma.4 Metanephrine, one of the hormones produced by the adrenal glands, breaks down to normetanephrine and vanillylmandelic acid (VMA) via the intermediate 4-hydroxy-3-methoxy-phenylglycol. Thus, VMA is always detected in the urine together with HVA and other catecholamine metabolites from pheochromocytoma or catecholamine-secreting chromaffin tumor cells.5,6 Catecholamine levels can be found in blood samples; however, the urine test reflects the production rate of the catecholamines over the collection period. Even at abnormal levels, the elevated quantities of these catecholamines are still very low, i.e., highly selective and sensitive analytical methods are needed for such important biomarkers. The urinary VMA and HVA values are most useful and can be correlated with the stage of disease, management, any maturation of tumor, and prognosis from children with neuroblastoma.4 The urine HVA/VMA ratio could be a screening tool to support earlier detection of Menkes disease, a disorder that affects copper levels in the body, leading to copper deficiency.7a The mechanism that removes HVA from the brain is still poorly understood, however, the efflux transport of HVA from the brain plays an important role in controlling the HVA level in the brain. This HVA efflux transport system is inhibited by several organic anions including IXS, and metabolites of monoamine neurotransmitters but not neurotransmitters per se.7b Thus, it is of clinical importance to develop a rapid and sensitive method for simultaneous analysis of HVA, VMA, and IXS in urine and other biological samples. (3) Tumur, Z.; Niwa, T. Am. J. Nephrol. 2009, 29, 551–557. (4) Liebner, E. J.; Rosenthal, I. M. Cancer 2006, 32 (3), 623–633. (5) Eisenhofer, G.; Lenders, J. W. M.; Linehan, W. M.; Walther, M. M.; Goldstein, D. S.; Keiser, H. R. New Engl. J. Med. 1999, 340, 1872–1879. (6) Lenders, J. W. M.; Keiser, H. R.; Goldstein, D. S.; Willemsen, J. J.; Friberg, P.; Jacobs, M.-C.; Kloppenborg, P. W. C.; Thien, T.; Eisenhofer, G. Ann. Intern. Med. 1995, 123, 101–109. (7) (a) Menkes, J. H.; Alter, M.; Steigleder, G. K.; Weakley, D. R.; Sung, J. H. Pediatrics 1962, 29, 764–779. (b) Mori, S.; Takanaga, H.; Ohtsuki, S.; Deguchi, T.; Kang, Y.-S.; Hosoya, K.-I.; Terasaki, T. J. Cereb. Blood Flow Metab. 2003, 23, 432–440. (8) (a) Issaq, H. J.; Delviks, K.; Janini, G. M.; Muschik, G. M. J. Liquid Chromatogr. Relat. Technol. 1992, 15/18, 3193–3201.

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6895

HVA and VMA in infant urine has been analyzed by capillary electrophoresis (CE) with UV detection with a detection limit (LOD) of 3.5 × 10-4 M for HVA and 1.8 × 10-4 M for VMA.8a A simultaneous determination of VMA, HVA, creatinine, and uric acid uses capillary electrophoresis and a 30 mM phosphate buffer (pH 7.0) containing 150 mM sodium dodecyl sulfate (SDS). The detection is by UV absorbance at 245 nm and the run is rather lengthy (15 min).9 Although the authors claimed lower LODs for both HVA and VMA (5.5 × 10-5 M and 5.0 × 10-5 M), such LODs are still significantly above the normal levels found in healthy subjects (8.2 to 41 µM for HVA and 11.6 to 28.7 µM for VMA).10 Besides high LOD, UV detection is also problematic as biological samples often consist of several compounds with strong absorption at low wavelengths. The neurotransmitters can be conjugated with a strong fluorophore and analyzed by MEKC with fluorescence detection to achieve high sensitivity.11a This approach might be feasible for routine analysis of a few compounds but becomes impractical and timeconsuming for multiple analytes. The use of CE-MS for analysis of VMA, HVA, and other biomarkers for metabolic disorders in newborns is available elsewhere.11b CE equipped electrochemical detection (ECD) using a bare silica capillary can be used to detect HVA and VMA after electrophoretic separation at pH 5.12a,b However, HVA and VMA are not baseline separated and other catecholamines including IXS are not included.12a HVA and VMA also become less electroactive at pH > 5, the minimal required pH for electrophoretic separation of HVA and VMA. In general, bare fused silica capillaries at high pH are used in such studies to resolve IXS, HVA, and VMA. The separation is lengthy since negatively charged IXS, HVA, and VMA migrate to the anode, i.e., opposite flow direction to the electroosmotic flow (EOF). In addition, HPLC with isocratic elution and spectrophotometric detection are often used for analysis of tryptophan metabolites including IXS.12c HPLC is also coupled with mass spectrometry to detect compounds associated with purple urinary bag syndrome (PUBS) and IXS is one of these toxic compounds.12d Micellar electrokinetic chromatography (MEKC) using a bare fused silica capillary and laser induced fluorescence detection (a KrF excimer laser, λ ) 248 nm) has been used to detect HVA, VMA, and IXS. Under the best condition with this expensive instrumentation, the detection limit for HVA and VMA is 170 nM and 150 nM, respectively.12e This work describes a novel scheme for the analysis of IXS, HVA, and VMA in the presence of tryptophan and other important catecholamines and indoleamines (Table 1). The fused silica capillary is coated with a thin layer of poly(diallyl dimethylammonium) chloride (PDDA) or gold nanoparticles (AuNPs) embed(9) Shirao, M. K.; Suzuki, S.; Kobayashi, J.; Nakazawa, H.; Mochizuki, E. J. Chromatogr. B 1997, 693, 463–467. (10) Garcia, A.; Heinanen, M.; Jimenez, L. M.; Barbas, C. J. Chromatogr. A 2000, 871, 341–350. (11) (a) Caslavska, J.; Gassmann, E.; Thormann, W. J. Chromatogr. A 1995, 709, 147–156. (b) Senk, P.; Kozak, L.; Foret, F. Electrophoresis 2004, 25, 1447–1456. (12) (a) Li, X. J.; Jin, W. R. Chin. Chem. Lett. 2002, 13 (9), 874–876. (b) Li, X. J.; Jin, W. R.; Weng, Q. F. Anal. Chim. Acta 2002, 461 (1), 123–130. (c) Marklova, E.; Makovickova, I.; Krakorova, I. J. Chromatogr. A 2000, 870, 289–293. (d) Bar-Or, D.; Rael, L. T.; Bar-Or, R.; Craun, M. L.; Statz, J.; Garrett, R. E. Clin. Chim. Acta 2007, 378, 216–218. (e) Paquette, D. M.; Sing, R.; Banks, P. R.; Waldron, K. C. J. Chromatogr. B: Biomed. Sci. Appl. 1998, 714 (1), 47–57.

6896

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

ded in PDDA to reverse the EOF, allowing fast migration of IXS, VMA, HVA, and tryptophan. In contrast, catecholamines and indoleamines migrate against the EOF and emerge very late in the electropherogram, i.e., they do not interfere with the analysis of IXS, VMA, HVA, and tryptophan. The presence of AuNPs plays an important role in baseline separation of several compounds and a mechanism is given to decipher the interaction between AuNPs and the analytes. Although the composite consisting of AuNPs embedded in PDDA has been reported by Chen et al.,13 this is the first systematic application of an AuNP-PDDA coated capillary for simultaneous analysis of IXS, HMA, and VMA in urine samples. Together with sample stacking, the applicability of this approach for analysis of such important biomarkers in urine samples with improved detection sensitivity is also presented and discussed in detail. EXPERIMENTAL SECTION Chemicals. Poly(diallyl dimethylammonium) chloride (PDDA, MW ) 200 000-350 000, 20 wt % in water), hydrogen tetrachloroaurate tetrahydrate (HAuCl4 · 4H2O), Tris (hydroxymethyl)aminomethane, phosphoric acid (H3PO4), and other chemicals were purchased from Sigma (Dublin, Ireland). Unless otherwise stated, a 50 mM H3PO4 solution was adjusted to pH 3.0 with 0.5 M Tris buffer and used as the separation buffer. The standard stock solutions (5.0 mM) of the analytes were prepared daily in deionized water. All solutions were prepared in Milli-Q ultrapure water and filtered through a 0.22 µm pore size membrane followed by sonication for 5 min prior to use. Synthesis of PDDA-Gold Nanoparticle (AuNPs) Composite. The PDDA-AuNP composite was prepared as described by Chen et al.13 PDDA (250 µL, 4% wt. in H2O), 40 mL of H2O, 200 µL of 0.5 M NaOH, and 300 µL of HAuCl4 (10 mg/mL) were thoroughly mixed in a beaker, covered with an inverted culture dish for 2 min. The mixture was then maintained at 100 °C for 30 min, resulting in a ruby red solution. The UV-visible spectrum of the AuNP colloid was recorded on a HP 8453 UV-visible spectrophotometer in a 1 cm optical path quartz cuvette. The size distribution of AuNPs in PDDA was measured using a zetasizer Nano ZS system (Malvern Instruments, MA) which is based on a dynamic light scattering (DLS) technique. TEM micrographs were obtained by a Delong LVEM (Soquelec, Montreal, QC, Canada) low-voltage TEM at 5 kV. A small amount of PDDA-AuNPs was sonicated to disperse the material. A 20 µL sample of well dispersed suspension was then dried on a Formvar-carbon coated grid and analyzed. Preparation of Coated Capillaries. A fused-silica capillary (50 µm id and 365 µm od) purchased from Polymicro Technologies (Phoenix, AZ, USA) was cut to 45 cm as the effective capillary length. In order to expose the maximum number of silanol groups on the silica surface, the fused-silica capillary was rinsed with 1.0 M NaOH and deionized water for 15 min each. The preconditioned capillary was then rinsed with the PDDA or the PDDA-AuNP (13) Chen, H.-J.; Wang, Y.-L.; Wang, Y.-H.; Dong, S.-J.; Wang, E. Polymer 2006, 47 (2), 763–766. (14) (a) Luong, J. H. T.; Male, K. B.; Glennon, J. D. Analyst 2009, 134 (10), 1965–1979. (b) Kraft, A. Int. J. Electrochem. Sci. 2007, 2, 355–385. (c) Swain, G. M.; Ramesham, R. Anal. Chem. 1993, 65, 345–351. (d) Xi, J.; Granger, M.; Chen, Q.; Strojek, K.; Lister, T.; Swain, G. Anal. Chem. 1997, 69, 591A– 597A. (e) Tenne, R.; Patek, K.; Hashimoto, K.; Fujishima, A. J. Electroanal. Chem. 1993, 347, 409–415.

Table 1. Chemical Structure, pKa Values and Aqueous Solubilities (A, g/L) of Various Analytes

solution for 15 min followed by an incubation period of 15 min. The coated capillary was gently rinsed with deionized water for 3 min to flush out unadsorbed coating materials. Before the first run, the coated capillary was equilibrated with the running buffer for 15 min but only for 3 min between the runs. All of these procedures were performed at 25 °C. For overnight or prolonged

storage, the capillary was rinsed with deionized water for 15 min and then stored with the capillary ends dipped in deionized water. Capillary Electrophoresis with Electrochemical Detection. The capillary outlet was epoxy sealed into a pipet tip so that only ∼1 cm protruded. The pipet tip was firmly attached vertically into a micromanipulator (HS6, World Precision Instruments, Sarasota, Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6897

FL, USA) with three-dimensional adjustment capabilities. A cylindrical cathodic/detection reservoir (2 cm diameter ×1 cm height) contained Pt wires (1 mm in diameter, 99.9% purity), serving as the counter electrode for amperometric detection and the cathode for electrophoresis. An Ag/AgCl (3 M NaCl) reference electrode was placed vertically into the reservoir, whereas the BDD electrode was inserted upward from the reservoir’s bottom and sealed with epoxy (the working reservoir volume was ∼3 mL). The micromanipulator and a laboratory jack (to which the reservoir was solidly mounted) were attached to a solid breadboard to prevent movement during alignment. The capillary outlet was aligned to the detecting electrode using the micromanipulator with the aid of a surgical microscope (World Precision Instruments). The capillary outlet was adjusted until it touched the electrode surface (evident by a slight bend in the capillary observed by microscopic inspection) and it was then backed off 25-30 µm using the micromanipulator’s z-control. The BDD electrode was connected to an electrochemical workstation (CHI660C, CH Instruments, Austin, TX, USA) consisting also of a platinum wire (1 mm in diameter) as counter electrode and an Ag/AgCl (3 M NaCl) electrode as reference electrode. The BDD electrode (3 mm in diameter, 0.1% doped diamond) was purchased from Windsor Scientific (Slough, Berkshire, U. K.). The analytes are detected by a boron doped diamond (BDD) electrode which is positioned close to the capillary outlet. The BDD electrode is advocated in this work due to its stable low background current and a wide applied potential window. BDD is also resistant to fouling due to the hydrogen surface termination and sp3 carbon bonding (no extended pi-electron system).14a,b BDD can be considered as general-purpose working electrodes with a broad array of applications for use with HPLC and CE. Pioneering work in the early 1990s was conducted by Swain and co-workers14c,d and the Fujishima group.14e The electrophoretic separation was conducted at -10 kV (reversed polarity) unless otherwise stated. A plastic cap with a central hole of ∼1 mm was firmly attached to the surface of the BDD electrode to reduce the active sensing area. The analyte sample was injected electrokinetically for 5 s at -10 kV. Peak identification was based on the migration time of a single standard with that of unknown peaks. However, if the resolution between any peak pair was low, then peak identification was performed by spiking both solutes individually. The pKa values and aqueous solubility of the analytes were obtained using the ACD/ Structure Designer software (Advanced Chemistry Development, Toronto, ON, Canada). The degree of ionization was estimated as pH ) pKa+ log [A-/AH] for acid and pH ) pKb+log [BH+/B] for base (the Henderson-Hasselbalch equation). RESULTS AND DISCUSSION Performance of the PDDA Coated Capillary. PDDA was firmly adsorbed on the inner walls of the capillary via ionic interactions between the negatively charged SiO- of fused silica and the quaternary ammonium groups of the polymer. Indeed, immersion of a substrate (glass, quartz, silica wafer, gold, silver, and even Teflon) into an aqueous 1% solution of this positively charged polymer results in the strong adsorption of a mono6898

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

Figure 1. Electropherograms obtained using a PDDA coating capillary (50 µm id and 45 cm effective length) for the separation of 20 µM IXS, 25 µM VMA, 25 µM HVA, 25 µM TRP, 100 µM isoproterenol (ISP), 100 µM normetanephrine (NMN), 100 µM epinephrine (EP), 50 µM 5-hydroxytryptamine (5-HT), 125 µM 4-hydroxy-3-methoxybenzylamine (HMBA), and 75 µM tryptamine (TA). The running buffer consisted of 50 mM H3PO4-Tris, (a) pH 3, (b) pH 4, and (c) pH 5. The separation voltage was applied at -10 kV with an injection time of 5s at -10 kV. BDD at +1.0 V vs Ag/AgCl, 3 M NaCl.

layer (1.6 nm) of PDDA on the substrate.15,16 The adsorption of a PDDA thin film on a glass substrate was also reported elsewhere.17 Notice also that the charge of PDDA is not pH dependent, as reflected by constant EOF in the range pH 2-8 provided the capillary is coated with high-molecular weight PDDA.18 Thus, PDDA with MW of 200 000-350 000 was used in this study for coating the capillary. At -10 kV, except for the epinephrine (EP) and normetanephrine (NMN) pair, all analytes were baseline resolved when 50 mM H3PO4-Tris pH 3.0 was used as the running buffer (Figure 1, curve a). On the basis of the calculated pKa values for the analytes (Table 1), fully deprotonated and highly negatively charged IXS exhibited high electrophoretic mobility and migrated concomitantly with EOF as the first peak in the electropherogram. The carboxylate/carboxyl ratio, estimated as 10(pH-pKa), is ∼0.16 and 0.04 for VMA and HVA, respectively. Thus, VMA should emerge before HVA and slightly ahead of EOF. At pH 3, the neutral form of TRP should be predominant (neutral TRP/TRP+ ) 4.2);therefore, it should migrate very closely to EOF and trail behind both VMA and HMA. The EP-NMN pair was slightly split further by conducting the separation at pH 4 with improved detection sensitivity but the running time was also slightly longer and VMA emerged very close to IXS (Figure 1, curve b). The run was lengthier at pH 5, with only 4 discernible peaks emerging in the electropherogram as HVA comigrated with VMA and TRP became more neutral at this pH and emerged far behind the HVA peak. The remaining analytes acquired more negative charges and interacted strongly with positively charged PDDA and were not eluted after 1200 s into the experiment. The detection sensitivity was also greatly compromised at this running (15) Kotov, N. A.; Harazsti, T.; Turi, L.; Zavala, G.; Geer, R. E.; Dekany, I.; Fendler, J. H. J. Am. Chem. Soc. 1997, 119, 6821–6832. (16) Moriguchi, I.; Teraoka, Y.; Kagawa, S.; Fendler, J. H. Chem. Mater. 1999, 1, 1603–1608. (17) Hrapovic, S.; Liu, Y.; Enright, G.; Bensabaa, F.; Luong, J. H. T. Langmuir 2003, 19, 3958–3965. (18) Wang, Y.; Dubin, P. L. Anal. Chem. 1999, 71, 3463–3468.

pH with a tilted baseline (Figure 1, curve c). It is of interest to compare the migration order of these peaks with MEKC performed at normal electrophoretic separation conditions by Caslavska et al.11a In such a study, a buffer composed of 75 mM sodium dodecyl sulfate (SDS), 6 mM Na2B4O7, and 10 mM Na2HPO4 (pH 9.2) is used with the separation potential set at +20 kV. The migration sequence is TRP, HVA, VMA, and IXS, which is opposite to the migration order shown in Figure 1. The remaining 7 catecholamines and indoleamines with a positive charge from the ammonium ion (NH3+) and/or the secondary amino group (-NH+-) were completely ionized at pH 3 and migrated to the cathode with high mobilities, i.e., countercurrent to the EOF. However, they were eventually driven out the capillary by EOF with higher mobility. Notice that a buffer composed of 200 mM boric acid, 100 mM potassium hydroxide, and 0.1% hydroxyethylcellulose, pH 9.2, has been used to separate VMA from various indoleamines under normal separation.19 In this case, VMA trailed far behind other indoleamines, as expected from its negative charge at this pH. PDDA can also be added as a buffer additive as described by Tseng et al.,20 resulting in high and reversed EOF. The mobility of indolamines and catecholamines decreases as the PDDA concentration increases. The separation of 14 analytes including indolamines, catecholamines, and metanephrines is achieved within 33 min under optimal separation conditions (1.2% PDDA and 5 mM formic acid at pH 4.0). Indeed, PDDA has been used to coat fused silica capillary to form a thin film for the subsequent absorption of carbon nanotubes.21 Besides adsorption, PDDA can be chemically bonded onto the interior capillary wall with an anodal EOF independent of pH ranging from 2.2 to 8.8.22 The lifetimes of both the bonded and physically coated capillaries exceeded 40 h of continuous use at 240 V/cm at pH 4. Our experimental data confirmed that the PDDA could be reused for several repeated runs and the capillary was easily reconditioned as described earlier. The coated capillary also exhibited good tolerance to methanol and 0.1 HCl. Separation on Capillary Coated with PDDA-Gold Nanoparticles. Our next strategy was to form gold nanoparticles (AuNPs) on the capillary wall, since they have been known to interact with several compounds with amino, hydroxyl, and carboxylic groups.23 In principle, AuNPs can be prepared separately and adsorbed on the PDDA layer. This concept has been used to coat the glass microchip channel with citrate-stabilized AuNPs for the separation of phenols.24 Another approach is to form AuNPs by electroless plating via hydroxylamine-mediated reduction.17 However, the charge of the PDDA-AuNP layer returns to negative and the separation must be performed under normal separation. Indeed, covalent attachment can be used to immobilize AuNPs onto the capillary wall, for instance, the preparation of dodecanethiol AuNPs on prederivatized 3-aminopropyl-trimethox(19) Stocking, C. J.; Slater, J. M.; Simpson, C. F. Exp. Nephrol. 1998, 6, 415– 420. (20) Tseng, W.-L.; Chen, S.-M.; Hsu, C.-Y.; Hsieh, M.-M. Anal. Chim. Acta 2008, 613 (1), 108–115. (21) Luong, J. H.T.; Bouvrette, P.; Liu, Y.; Yang, D.-Q.; Sacher, E. J. Chromatogr. A 2005, 1 (2), 187–194. (22) Liu, Q.; Lin, F.; Hartwick, R. A. J. Chromatogr. Sci. 1997, 35 (3), 126–130. (23) Zhong, Z. Y.; Patskovskyy, S.; Bouvrette, P.; Luong, J. H. T.; Gendanken, A. J. Phys. Chem. B 2004, 108 (13), 4046–4052. (24) Pumera, M.; Wang, J.; Grushka, E.; Polsky, R. Anal. Chem. 2001, 73, 5625– 5628.

Figure 2. Electropherograms obtained using a PDDA-AuNPs coated capillary at pH 3, 4, and 5. Other conditions were same as Figure 1.

ysilane or 3-mercaptopropyl-trimethoxysilane fused-silica capillaries.25 Again; several steps are required for the preparation of such modified capillaries. In order to maintain the positive charge for the coating layer, a one step procedure described by Chen et al.13 was used to prepare the PDDA-AuNP composite since PDDA acts as both reducing and stabilizing agents for AuNPs. The PDDA-gold colloid exhibited a surface plasmon resonance (SPR) around 527-528 nm, which could be related to AuNPs with the mean diameter of 27-28 nm, in agreement with the report on the SPR peak obtained for AuNPs with an average diameter of 27 nm.26 Dynamic light scattering confirmed that AuNPs should have a mean diameter of 28 nm with narrow particle size distribution. A TEM micrograph of PDDA-AuNPs indicated that AuNPs exhibited an average size of 25 nm (figure not shown). Notice that the excellent stability of AuNPs arises from the electrosteric effect of PDDA, in agreement with Mayer et al.27 where PDDA is simply used as a protecting and stabilizing agent for AuNPs. Under the same electrophoretic and detection condition for the PDDA coated capillary, the electropherogram obtained for the 10 analytes using the PDDAAuNP coated capillary is shown in Figure 2. At pH 3, EP was satisfactorily separated from NMN and the peaks were significantly sharper with low background current, resulting in significantly improved detection sensitivity. For the microchip channel with coated PDDA/AuNPs layer-by-layer, improved resolution and detection sensitivity has been reported for the three aminophenols, although the rationale behind such behavior is not known.24 Our experimental data also confirmed that the run was longer at pH 4 without any improvement in the separation of the NMN-EP pair with slightly reduced detection sensitivity. At pH 5, only four peaks (IXS, VMA, HVA and TRP) emerged in the electropherogram with the TRP peak far behind the HVA peak. The circumstances determining the effect of the PDDA-AuNP coated capillary are complex. The viscosity of the PDDA-AuNP solution is about 90% of the PDDA solution, i.e., it is easier to pump the PDDA-AuNP solution through the capillary to form a (25) (a) O’Mahony, T.; Owens, V. P.; Murrihy, J. P.; Guihen, E.; Holmes, J. D.; Glennon, J. D J. Chromatogr. A 2003, 1004 (1-2), 181–193. (b) Yang, L.; Guihen, E.; Holmes, J. D.; Loughran, M.; O’Sullivan, G. P.; Glennon, J. D. Anal. Chem. 2005, 77 (6), 1840–1846. (26) Nasir, S. M.; Nur, H. J. Fundam. Sci. 2008, 4, 245–252. (27) Mayer, A. B. R.; Hausner, S. H.; Mark, J. E. Polym. J. 2000, 32, 15–22.

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6899

Figure 3. Electropherograms obtained using a PDDA-AuNP coated capillary with the BDD electrode for the study of (A) buffer concentration effect: 25, 50, 75, and 100 mM H3PO4-Tris, pH 3, (B) injection time effect (3, 5, 7, and 10 s), (C) effect of detection potential (+0.6, +0.8, +1.0,+ and 1.2 V vs Ag/AgCl, 3 M NaCl), (D) effect of separation voltage (-5, -10, -15, and -20 kV). The sample mixture consisted of 20 µM IXS, 25 µM (each) VMA, HVA, TRP, 100 µM ISP, 100 µM NMN, 100 µM EP, 50 µM 5-HT, 125 µM HMBA, 125 µM DHBA, and 75 µM TA.

homogeneous layer.28 The incorporation of AuNPs in the polymer network could be attributed to the structural change in pristine PDDA after the synthesis of AuNPs. It has been shown that CdC and C-N are produced after PDDA reacts with HAuCl4 by (28) Zhang, Z.-X.; Yan, B.; Liu, K.; Liao, Y.-P.; Liu, H.-W. Electrophoresis 2009, 30, 379–387.

6900

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

comparing the FTIR spectra of PDDA and PDDA-AuNPs.13 For the latter, a new peak appears at 1578 cm-1, which can be assigned for CdC and CdN in-plane vibration. In aqueous milieu, gold nanoparticles are negatively charged and the absolute value of the ζ-potential decreases as the pH decreases, whereas above pH 5.8, the absolute value of the ζ-potential

Table 2. Linearity and Detection Limit of the Analytesa,b analytes

linear range (µM)

calibration equation

R2

detection limit (µM)

indoxyl sulfate (IXS) vanillylmandelic acid (VMA) homovanillic acid (HVA) tryptophan (TRP) isoproterenol (ISP) normetanephrine (NMN) epinephrine (EP) 5-hydroxytryptamine (5-HT) 4-hydroxy-3-methoxybenzylamine (HMBA) 3,4- dihydroxybenzylamine (DHBA) tryptamine (TA)

1.00-40 1.25-50 1.25-50 1.25-50 5.00-200 5.00-200 5.00-200 2.50-100 6.25-250 6.25-250 3.75-150

I ) 0.4839X+0.1490 I ) 0.3635X+0.3036 I ) 0.3034X+0.5667 I ) 0.3025X+1.0870 I ) 0.0871X+0.7481 I ) 0.0865X+0.6429 I ) 0.0945X+0.3850 I ) 0.1686X+0.9364 I ) 0.0822X+0.0446 I ) 0.0866X+0.3842 I ) 0.1476X+0.5307

0.9998 0.9998 0.9998 0.9998 0.9995 0.9995 0.9995 0.9995 0.9991 0.9991 0.9995

0.3 0.4 0.4 0.4 1.7 1.7 1.7 0.8 2.1 2.1 1.3

a I ) peak current (nA), X ) analyte concentration (µM). b Electrophoretic separation was carried out using a PDDA-AuNP coated capillary (50 µm id, 365 µm od) with an effective length of 45 cm, running buffer, 50 mM H3PO4-Tris, pH 3.0, separation voltage, -10 kV, injection for 5s at -10 kV. Detection: BDD electrode at +1.0 V vs. Ag/AgCl, 3M NaCl.

Figure 4. The comparison electropherograms for (a) nonsample stacking using a running buffer of 50 mM H3PO4-Tris, pH 4 with an injection time of 5 s (b) sample stacking, injection buffer: 10 mM H3PO4-Tris, pH 2, and the running buffer of 50 mM H3PO4-Tris, pH 4 with an injection time of 10 s. (c) sample stacking, injection buffer: 10 mM H3PO4-Tris, pH 2, and the running buffer was similar to the injection buffer with an injection time of 10 s. Concentration of analytes: 50 µM (each) IXS, VMA, AA, HVA, UA, TRP. Separation voltage: -10 kV. BDD at +1.0 V vs Ag/AgCl, 3 M NaCl.

becomes almost constant (∼-45 to -50 mV). The negative charge of the gold surface may be explained by the fact that the Au-OH present on the gold surface can lose protons as the pH increases to form Au-O-groups on the nanoparticle surface.29 AuNPs have a partially hydroxylated surface with a pKa value of 3.2 (Au-OH) in the presence of water.29 At pH 3, around the pKa value of AuNPs, -OH groups are predominant (-OH groups/O- groups ) 1.6). The rest of the gold surface should be metallic, i.e., essentially hydrophobic. Therefore, AuNPs would display ionic and hydrophobic interaction, as well as hydrogen bonding with the PDDA network. The synthesis and incorporation of AuNPs in the polymer network would affect the PDDA structural change, which in turn increased the charge density and coverage efficiency of the coating. At pH 4 or 5, Au-O- groups were predominant, resulting in a (29) Sylvestre, J.-P.; Kabashin, A. V.; Sacher, E.; Meunier, M.; Luong, J. H. T. J. Am. Chem. Soc. 2004, 126, 7176–7177.

decrease of the ζ-potential of the PDDA-AuNPs composite, i.e., the EOF is reduced. Notice that TRP also becomes more neutral at pH 5, whereas the charges of the catechoamines and indolamines were highly negative and such analytes were expected to interact strongly with PDDA. Indeed, all catecholamines and indoleamines were not eluted after 24 min into the experiment. In contrast, Zhang et al.28 reported that the separation using the PDDA coated capillary is significantly longer than the one modified with PDDA-AuNPs for analysis of heroin and impurities. In such a study, the separation is performed at pH 5.2 with addition of 3% methanol into the running buffer consisting of 120 mM ammonium acetate. Therefore, it was somewhat difficult to compare the result obtained in this work with that of Zhang et al.28 Other Optimal Conditions and Sample Stacking. The 50 mM phosphoric-Tris buffer pH 3 appeared to provide the highest detection sensitivity with good separation resolution. Indeed, 4-hydroxy-3-methoxybenzylamine comigrated with 3,4-dihydroxybenzylamine when the buffer strength increased to 100 mM and the run was longer with lower detection sensitivity (Figure 3A). This pair was coeluted as one single peak if the separation was carried out with the PDDA coated capillary. Thus, a running buffer consisting of 50 mM phosphoric-Tris, pH 3 was used to optimize the separation potential and the detection potential of the BDD electrode. With respect also to detection sensitivity and separation efficiency (N/cm), the electrokinetic injection of the sample at -10 kV for 5 s was optimal compared to the results obtained at shorter (3 s) or longer (7 and 10 s) injection times (Figure 3B). Beyond 5 s, the resulting peaks were very broad with compromised detection sensitivity and separation efficiency. In contrast, 3 s was not sufficient, as reflected by considerably lower peaks. The BDD detection potential, poised at +1 V vs 3 M Ag/AgCl provided highest detection sensitivity compared to the detection performed at lower or higher applied potentials (Figure 3C). Both resolution and detection sensitivity were severely deteriorated when the separation was conducted at -15 kV and -20 kV, compared to the run at -10 kV (Figure 3D). Linearity and LOD obtained for 11 analytes are summarized in Table 2. The four fast migrating analytes, IXS, VMA, HVA, and tryptophan exhibited a LOD of 0.3 µM, whereas a higher LOD, ∼1 µM was obtained for the slower migrating group. Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6901

Figure 5. Analysis of urine sample using a running buffer consisting of 50 mM H3PO4-Tris, pH 4. Separation voltage: -10 kV with an injection time of 5 s at -10 kV. Detection: BDD at +1.0 V vs Ag/ AgCl, 3 M NaCl. The urine samples were filtered using a 0.22 µm Millipore filter and diluted in the injection buffer with different dilution. (A) 10-fold dilution, pristine and spiked urine samples were presented as the lower and upper curve, respectively. Peak identification: (1) IXS; (2) VMA; (3) AA; (4) HVA; (5) UA; and (6) TRP. The diluted urine sample was spiked with 20 µM IXS, 20 µM VMA, 100 µM AA, 20 µM HVA, 100 µM UA, and 20 µM TRP. (B) 15-fold dilution, pristine and spiked urine samples were presented as the lower and upper curve, respectively. Peak identification: (1) IXS; (2) VMA; (3) AA; (4) HVA; (5) UA; and (6) TRP. The diluted urine sample was spiked with 20 µM IXS, 20 µM VMA, 100 µM AA, 20 µM HVA, 100 µM UA, and 20 µM TRP.

For further LOD improvement, a series of experiments based on field-amplified sample stacking (FASS)30 was performed by exploiting the conductivity difference between the sample zone and the running buffer to effect preconcentration. A standard solution of 6 analytes including ascorbic and uric acids, two endogenous urinary compounds, was prepared in 10 mM H3PO4 pH 2 and injected at -10 kVfor 10 s with the separation performed at -10 kV using 50 mM H3PO4 pH 4 as the running buffer. It should be noted that at pH 3, ascorbic acid migrated very closely with HVA, whereas uric acid was not baseline separated from TRP. In principle, the amount of stacking is proportional to the conductivity difference between the running (30) Weng, Q.-F.; Xu, G.-W.; Yuan, K.-L; Tang, P J. Chromatogr. B 2006, 835, 55–61.

6902

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

Figure 6. Analysis of an urine sample using a running buffer consisting of 50 mM H3PO4-Tris, pH 4. Separation voltage: -10 kV with an injection time of 5 s at -10 kV. Detection: BDD at +1.0 V vs Ag/AgCl, 3 M NaCl. The urine sample was filtered using a 0.22 µm Millipore filter and diluted 8-fold in the injection buffer. (A) with sample stacking and (B) without sample stacking.

buffer and the sample solution. A preconcentration factor of 4 was observed for IXS, AA, and HVA compared to 3 for VMA and UA and only 1.25 for TRP (Figure 4, curve b). With sample stacking, the detection limit of IXS, AA, and HVA was 75 nM compared with nonstacking (300 nM, Figure 4, curve a), significantly lower than the physiological levels of these biomarkers in urine as discussed later. Notice that peak-width was broadened when the injection time was greater than 10 s (data not shown). In another attempt, the capillary was first filled with 50 mM H3PO4 pH 4 followed by a sample injection as described above. The cathodic running buffer was changed to 10 mM H3PO4, pH 2 instead of 50 mM H3PO4, pH 4. In this case, the separation window was narrower, resulting in the comigration of three pairs: IXS/VMA, AA/HVA, and UA/TRP (Figure 4, curve c). No further improvement in separation resolution was attained by adding several organic modifiers including 10% methanol, 10% acetonitrile, 5 mM methyl β-cyclodextrin, and 5 mM cetyl trimethylammonium bromide in the running buffer. Analysis of Biomarkers in Urine. The electropherogram of an authentic urine sample obtained from a healthy female shows 5 low peaks and one very high peak. All biomarkers were detected even if the urine sample was diluted to 10- and 15-fold, respectively in the injection buffer (Figure 5). Standard IXS, HVA, VMA,

ascorbic acid (AA), uric acid (UA), and TRP were spiked into the urine sample for peak identification. Dilution of the urine sample also improved peak to peak separation and peak identification. All peaks were identified and the sequence was assigned as IXS, VMA, AA, HVA, UA, and TRP. The highest peak was attributed to UA as expected since its normal concentration in urine ranges from 1.23 to 3.7 mM (1.48-4.43 mmol/day with an average urine volume of 1200 mL).31 The UA peak was confirmed by matching the migration time and spiking the urine sample with 30 µM uric acid. Peak deconvolution and peak area were obtained using WinPLOTR (Version: May, 2009, http://www.cdifx.univ-rennes1.fr/ winplotr/winplotr.htm). The concentration for each analyte was estimated by comparing the peak areas of the authentic and spiked urine samples. The average concentration of IXS, VMA, HVA, AA, and UA for two different urine samples obtained from the same healthy female was determined to be 170, 55, 87, 142, and 1,075 µM, respectively. Normal levels of such analytes in urine are 100-1000 µM for IXS,32 50-1000 µM for VMA,10 14-125 µM for HVA,10 and 150-200 µM for AA,33 i.e., the values estimated by CE-ECD fall in the normal range for healthy subjects. Therefore, sample staking was not mandatory for the simultaneous analysis of IXS, HVA, and VMA in the presence of UA, AA, and TRP since the LOD of such analytes was significantly lower than their normal physiological levels. Sample stacking of the urine samples at low dilution (below 5-fold) did not improve the detection limits (figure not shown), very likely due to a high level of salts in the urine sample. With an 8-fold dilution, sample stacking provided a significant improvement in the LOD with a sharper corresponding peak for each analyte as shown in Figure 6A. In particular, sample stacking also allowed for the detection of TRP, the last peak that migrated very (31) Tietz, N. W. Fundamentals of Clinical Chemistry: W.B. Saunders Co.: Philadelphia, 1987. (32) Harlit, H. J. Biol. Chem. 1933, 537–545. (33) Ridi, E.; Moubasher, M. S. R.; Hassan, Z. F. Biochem. J. 1951, 49, 246– 251. (34) Matsuo, M.; Tasaki, R.; Kodama, H.; Hamasaki, Y. J. Inherit. Metab. Dis. 2005, 28 (1), 89–93. (35) Luo, D.; Wu, L.; Zhi, J. ACS Nano 2009, 3 (8), 2121–2128.

closely to the uric acid peak. In addition, both HVA and VMA were positively detected and identified compared with nonsample stacking (Figure 6B). Thus, the method became useful for estimation of the urine HVA/VMA ratio, a useful screening method for Menkes disease. This ratio could range from 4.1 to 69.7 owing to impaired activity of dopamine β-hydroxylase, a copper-dependent enzyme.34 CONCLUSIONS In brief, a novel scheme was described for electrophoretic separation and detection of several important biomarkers in urine. A fused silica capillary with a coating layer of PDDA and AuNPs reversed the electroosmotic flow and served as a stable layer for resolving the analytes. No electrode fouling was observed during repeated analysis and even with the urine sample. The detection limit obtained for IXS, HVA, and VMA was considerably below their normal physiological levels in biological samples. The method was simple and capable of measuring several important biomarkers in urine samples without sample pretreatment with excellent selectivity and detection sensitivity. Sample stacking could be easily performed to improve detection limits of these analytes in urine samples provided such samples were diluted properly to reduce the level of uric acid and salts. Furthermore, many other basic neurotransmitters and their acidic metabolites could also be detected. Notice also that a BDD nanoforest electrode (BDDNF) can be fabricated by hot filament chemical vapor deposition.35 This type of electrode exhibits improved detection sensitivity compared to conventional planar BDD electrodes. Integration of BDDSNF with capillary electrophoresis is a subject of future endeavor. ACKNOWLEDGMENT The authors thank the Science Foundation Ireland (SFI) for an SFI Walton Fellowship (JHTL), an IRCSET Embark Award (LZ), and an SFI-SRC Grant for the Irish Separation Science Cluster (ISSC). Received for review April 27, 2010. Accepted July 5, 2010. AC101105Q

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6903