Anal. Chem. 2001, 73, 3199-3205
Articles
Influence of Natural, Electrically Neutral Lipids on the Potentiometric Responses of Cation-Selective Polymeric Membrane Electrodes Philippe Bu 1 hlmann,*,† Masakazu Hayakawa, Takahito Ohshiro, Shigeru Amemiya, and Yoshio Umezawa*
Department of Chemistry, School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan
Ionophore-free ion exchanger electrodes were found to exhibit quite a high selectivity for the creatininium ion; however, measurements in diluted urine samples revealed large emf drifts. Potentiometric, chromatographic, NMR, and mass spectrometric evidence did not reveal any major cationic interfering agents, and anionic interfering agents cannot trivially explain the consistently positive emf drifts. Ultrafiltration of urine samples showed that the interfering agents have molecular weights below 1000 u. The drifts are apparently caused by electrically neutral lipophilic compounds of low molecular weight that are easily extracted into organic phases. Follow-up experiments showed that p-cresol and cholesterol cause no significant emf responses but that coproporphyrin, phosphatidylserine, taurocholic acid, cholic acid, phosphatidylethanolamine, and octanoic acid cause positive emf drifts of the type that was observed with the urine samples. The extent of the responses and the response time depend not only on the specific compound but also on the cation in the sample solution. These results suggest that the emf drifts are due to extraction of such natural lipids into the organic membrane phase where they interact in an ionophore-like fashion with the analyte and interfering ions. Changes in the potentiometric selectivities after contact with natural lipids support this interpretation. The same effect of natural lipids is also expected for ionophorebased electrodes. Indeed, exposure of a valinomycinbased electrode to a methylene chloride extract of urine resulted in a significant reduction of the Na+ discrimination, increasing log Kpot K,Na from -3.9 to -3.1. Polymer membrane-based ion-selective electrodes (ISEs) for many analytes have been developed1-3 and are widely used for various practical applications. A typical ISE responds to the target 10.1021/ac0015016 CCC: $20.00 Published on Web 06/15/2001
© 2001 American Chemical Society
ion and to interfering ions that have the same charge sign as the target ion.4 Interferences from ions of the opposite charge are usually limited to samples with high activities of lipophilic coions.5 Electrically neutral species are ordinarily not considered as interfering agents; however, it is well-known that certain nonionic species can affect emf responses significantly.2 Although previous work focused on synthetic nonionic surfactants and alcohols, effects of lipids that naturally occur in biological samples have to date not been reported. Here we report that certain natural ubiquitous lipids may lead not only to emf drifts but also to substantial selectivity changes. The cationic responses to aqueous solutions of methanol, ethanol, and propanol, as obtained with potassium stearate gel membranes, were some of the earliest reported potentiometric responses to nonionic compounds.6 They were qualitatively explained by the influence of these alcohols on the structure of the stearate gel that was used as sensor membrane and on the activity of the potential-determining K+ in the sample solution. The effect of nonionic analytes on the activity of a sample ion was later used to determine, for example, water in methanol, ethanol, acetic acid, or acetonitrile,7,8 or ethanol in alcoholic beverages.9 Because such measurements rely on changes in the activity of an ionic species in the sample solution, they can be † Present address: Department of Chemistry, University of Minnesota, 207 Pleasant Street Southeast, Minneapolis, MN 55455. (1) Umezawa, Y. CRC Handbook of Ion-Selective Electrodes: Selectivity Coefficients; CRC Press: Boca Raton, FL, 1990. (2) Bu ¨ hlmann, P.; Pretsch, E.; Bakker, E. Chem. Rev. 1998, 98, 1593. (3) Bakker, E.; Bu ¨ hlmann, P.; Pretsch, E. Electroanalysis 1999, 11, 915. (4) Bakker, E.; Bu ¨ hlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083. (5) Bu ¨ hlmann, P.; Amemiya, S.; Yajima, S.; Umezawa, Y. Anal. Chem. 1998, 70, 4291. (6) Botre, C.; Mascini, M.; Memoli, A. Anal. Chem. 1972, 44, 1371. (7) Kakabadse, G. J.; Al-Aziz, M. S.; Hamilton, I. C.; Olatoye, E. O.; Perry, R.; Tipping, A. E.; Vaudrey, S.; Al-Yawer, N. F. N. Analyst 1988, 113, 1365. (8) Karim, M. R. O.; Karadaghi, T. M. Analyst 1988, 113, 1593. (9) Kokovkin, V. V.; Smolyakov, B. S. J. Anal. Chem. 1995, 50, 519.
Analytical Chemistry, Vol. 73, No. 14, July 15, 2001 3199
performed not only with polymeric membranes but also with solidstate electrodes. A second type of effects of nonionic compounds on emf responses occurs only in the case of liquid and polymeric membrane electrodes. These effects arise when a nonionic compound partitions into an organic sensing membrane and, directly or indirectly, affects the membrane activity of the potentialdetermining ions, which are distributed across the samplemembrane interface. For example, plasticized poly(vinyl chloride) (PVC) membranes containing tetraphenylborate salts of barium complexes with polyethoxylates, originally developed as Ba2+selective electrodes, respond to various acyclic polyether-type nonionic surfactants,10-12 valinomycin- and crown ether-based electrodes respond to higher alcohols13 and dinitrophenol,14 and a number of anion-selective electrodes were found to respond to phenols.15-19 A particular interest has focused on emf interferences of nonionic surfactants, because automated commercial blood gas/ electrolyte analyzers commonly operate with calibration and wash solutions that contain nonionic surfactants (e.g., polyether-based compounds).20,21 Meyerhoff and co-workers showed that concentrations of ionic surfactants as low as 5.5 × 10-5 M affected emf responses of cation selective electrodes. They modeled their response function, discussed changes in selectivities, and reported optical results confirming that nonionic surfactants affect the emf response by partitioning into the sensor membrane and directly interacting with the cationic analytes and the ionophore. Our interest in emf responses to electrically neutral species was prompted by an unexpected interference that we observed in measurements of creatinine, which is a major analyte in clinical chemistry.22-30 Ionophore-based ISEs for creatininium were reported28 but have so far not been used for real samples. To selectively bind creatininium, we synthesized the hydrogen bond(10) Jones, D. L.; Moody, G. J.; Thomas, J. D. R. Analyst 1981, 106, 974. (11) Sugawara, M.; Nagasawa, S.; Ohashi, N. J. Electroanal. Chem. 1984, 176, 183. (12) Chernova, R. K.; Kulapina, E. G.; Materova, E. A.; Tret’yachenko, E. V.; Novikov, A. P. Zh. Anal. Khim. 1992, 47, 1464. (13) Anzai, J.; Liu, C. C. Anal. Chim. Acta 1991, 248, 323. (14) Fitzgerald, E. M.; Djamgoz, M. B. A. J. Neurosci. Methods 1995, 59, 273. (15) Mokrov, S. B.; Stefanova, O. K.; Ivankov, V. M.; Karavan, V. S. Russ. J. Electrochem. 1995, 31, 150. (16) Lunsford, S. K.; Ma, Y.-L.; Galal, A.; Striley, C.; Zimmer, H.; Mark, H. B., Jr. Electroanalysis 1995, 7, 420. (17) Ito, T.; Radecka, H.; Umezawa, K.; Kimura, T.; Yashiro, A.; Lin, X. M.; Kataoka, M.; Kimura, E.; Sessler, J. L.; Odashima, K.; Umezawa, Y. Anal. Sci. 1998, 14, 89. (18) Ito, T.; Radecka, H.; Tohda, K.; Odashima, K.; Umezawa, Y. J. Am. Chem. Soc. 1998, 120, 3049. (19) Odashima, K.; Ito, T.; Tohda, T.; Umezawa, Y. Chem. Pharm. Bull. 1998, 46, 1248. (20) Espadas-Torre, C.; Bakker, E.; Barker, S.; Meyerhoff, M. E. Anal. Chem. 1996, 68, 1623. (21) Malinowska, E.; Meyerhoff, M. E. Anal. Chem. 1998, 70, 1477. (22) Lawson, P. L. Arch. Pathol. Lab. Med. 1995, 119, 312. (23) Bu ¨ hlmann, P., Ph.D. thesis, Swiss Federal Institute of Technology (ETH), Zu ¨ rich, Switzerland, 1993. (24) Bu ¨ hlmann, P.; Simon, W. Tetrahedron 1993, 49, 7627. (25) Bell, T. W.; Hou, Z.; Luo, Y.; Drew, M. G. B.; Chapoteau, E.; Czech, B. P.; Kumar, A. Science 1995, 269, 671. (26) Beckles, D. L.; Maiorello, J.; Santora, V. J.; Bell, T. W.; Chapoteau, E.; Czech, B. P.; Kumar, A. Tetrahedron 1995, 51, 363. (27) Osborne, M. D.; Girault, H. H. Mikrochim. Acta 1995, 117, 175. (28) Kelly, P. M.; Kataky, R.; Parker, D.; Patti, A. F. J. Chem. Soc., Perkin Trans. 2 1995, 1955. (29) Bochenska, M.; Biernat, J. F. Proceedings of the 2nd Bioelectroanalytical Symposium, Ma´trafu ¨ red, 1992; Akade´miai Kiado´, Budapest, 1993; p 255. (30) Meyerhoff, M.; Rechnitz, G. A. Anal. Chim. Acta 1976, 85, 277.
3200
Analytical Chemistry, Vol. 73, No. 14, July 15, 2001
forming ionophores 1-{5,7,7-trimethyl-2-(1,3,3-trimethylbutyl)octyl}isocytosine, N′-methylureidonaphthoic acid 3,6,9,12-tetraoxadocosyl amide, and 2,7-di-tert-butyl-9,9-dimethyl-4,5-xanthenepot ) -0.4, diylbiscarbamic acid dibenzyl ester31 (log Kcreatininium,K 0.0, and -1.5, respectively). Unfortunately, we found an unsatisfactory K+ discrimination of ISEs based on these ionophores, probably because of dipole-ion interactions between K+ and ionophore oxygens, and ionophore self-association due to the presence of hydrogen bond-accepting and donating groups.32 To avoid these complications, we tested32 ISEs based on R,R,′-bis(N′-butylthioureylene)-m-xylene31 and R,R,R,′R,′-tetrakis(trifluoromethyl)-1,3-benzene dimethanol,33 which only contain poor hydrogen bond acceptors. The negligible response for the former pot and the log Kcreatininium,K of -1.8 for the latter suggested that creatininium ionophores that bind only to the carbonyl group are not effective, likely because of the charge delocalization in creatininium. Apparently, better ionophores must contain multiple preorganized interaction sites and structural elements that prevent self-association. However, when we optimized ionophore-free ion-exchanger PVC membrane electrodes34 containing tetraphenylborate salts for a comparison with these ionophore-based ISEs, we were surprised to find that ionophore-free membranes can exhibit selectivities that were comparable or even superior to previously reported28 ionophore-based creatininium-sensitive electrodes. Although the selectivities of our optimized ionophore-free membranes appeared sufficient for measurements in urine, applications in diluted urine samples resulted in severe emf drifts. We report here on potentiometric, chromatographic, spectroscopic, and extraction experiments that let us conclude that these drifts are not simply due to interfering cations but rather result from a combination of several natural nonionic interfering agents. Followup experiments with valinomycin-based and ionophore-free ionexchanger electrodes allowed the identification of some of the interfering agents and showed that lipid interferences are not limited to ionophore-free electrodes. EXPERIMENTAL SECTION Reagents. All reagents were of the highest commercially available grade and were used without further purification. Deionized and charcoal-treated water (18.2 MΩ‚cm specific resistance) obtained with a Milli-Q PLUS or Elix 3/A10 reagentgrade water system (Millipore, Bedford, MA) was used for all sample solutions. Potassium tetrakis[3,5-bis(trifluoromethyl)phenyl]borate (KTFPB), potassium tetrakis(p-chlorophenyl)borate (KTClPB), and 2-nitrophenyl octyl ether (oNPOE) were purchased from Dojindo Laboratories (Kumamoto, Japan); bis(2-ethylhexyl) phthalate (DOP) and chloroparaffin (60% chlorine, d20 4 ) 1.387, Selectophore), from Fluka (Buchs, Switzerland); L-R-phosphatidylserine (from bovine spinal cord) and tyrosine, from Wako (Osaka, Japan); N-methyl-L-lysine and coproporphyrin I, from Aldrich (Milwaukee, WI); phosphatidylethanolamine, from Sigma (St. Louis, MO); and all other natural lipids and organic com(31) Bu ¨ hlmann, P.; Nishizawa, S.; Xiao, K. P.; Umezawa, Y. Tetrahedron 1997, 53, 1647. (32) Hayakawa, M. MS thesis, The University of Tokyo, Tokyo, Japan, 1997. (33) Krespan, C. G. J. Org. Chem. 1979, 44, 4924. (34) For a critical description of “ionophore-free ion-exchanger electrodes” see ref 2, p 1595.
pounds, from Tokyo Kasei Kogyo (Tokyo, Japan). Urine samples were provided by healthy nonhospitalized volunteers. ISE Membranes. Ionophore-free solvent polymeric membranes containing KTFPB or KTpClPB (3.5 wt %), and plasticizers: PVC (2:1 w/w) were prepared as reported previously.35,36 Valinomycin-based membranes contained 1.4 wt % ionophore, 64 mol % KTClPB, and bis(2-ethylhexyl) sebacate:PVC (2:1 w/w). Circles ∼7 mm diameter were cut out from the thus prepared master membranes and mounted on Philips-type electrode bodies (Glasbla¨serei Mo¨ller, Zu¨rich, Switzerland) equipped with tubular glass assemblies incorporating an internal Ag/AgCl reference electrode (Mo¨ller). Emf Measurements. Double-junction-type Ag/AgCl reference electrodes (Denki Kagaku Keiki (DKK), Tokyo, Japan) and ion meters (models IOL 30, 40, or 50, DKK) were used for all emf measurements. The cell assembly for the potentiometric measurements was as follows: Ag | AgCl | 3 M KCl || outer filling solution || sample | membrane | inner filling solution | AgCl | Ag Unless mentioned otherwise, a 0.1 M acetate buffer containing 10 mM creatinine hydrochloride was used as the inner solution of the ISEs, and the electrodes were conditioned overnight in the same buffer. Acetate buffer (1.35 M in the preliminary experiments and 0.1 M for selectivity and drift experiments) containing 10 mM creatinine hydrochloride, and 1 M CH3COOLi solutions were used as the outer filling solutions for the ionophore-free ion exchanger and valinomycin electrodes, respectively. Acetate buffer solutions were prepared from Mg(OAc)2 and acetic acid. All experiments were performed at 26 ( 1 °C. With the exception of preliminary experiments, calibration curves were obtained by measuring the emf in a concentrated analyte solution and the repeated removal of an aliquot of the sample and addition of analyte-free buffer solution. Measurements of responses to natural lipids and lipid exposure of membranes between selectivity determinations were performed using 1-L samples to avoid significant lipid depletion in the aqueous solution. Activity coefficients were calculated according to a twoparameter Debye-Hu¨ckel approximation.37 The LiCl parameters were used as an approximation for those of creatininium hydrochloride. A value for the latter is apparently not known, but the mobility of the piperidinium ion, which is comparable in size to the creatininium ion, differs from the value of the Li+ mobility by only 4%.38 Liquid junction potentials at the reference electrodes were calculated using the Henderson approximation,39 using the value of the Li+ mobility as an approximation for that of the creatininium ion. Selectivity coefficients were determined using the separate solution method (SSM)40,41 by linear regression of emf responses from the range of Nernstian responses. Average response slopes of valinomycin-based electrodes were 58.8, 59.9, and 59.7 mV/decade for Na+, NH4+, and K+, respectively. Average (35) Amemiya, S.; Bu ¨hlmann, P.; Tohda, K.; Umezawa, Y. Anal. Chim. Acta 1997, 341, 129. (36) Amemiya, S.; Bu ¨hlmann, P.; Tohda, K.; Umezawa, Y. Anal. Chim. Acta 1997, 351, 407. (37) Meier, P. C. Anal. Chim. Acta 1982, 136, 363. (38) Lide, D. R. CRC Handbook of Chemistry and Physics; CRC Press: Boca Raton, 1999. (39) Morf, W. E. The Principles of Ion-Selective Electrodes and of Membrane Transport; Elsevier: New York, 1981. (40) Buck, R. P.; Lindner, E. Pure Appl. Chem. 1995, 66, 2527. (41) Bakker, E.; Pretsch, E.; Bu ¨ hlmann, P. Anal. Chem. 2000, 72, 1127.
response slopes of ionophore-free ion-exchanger electrodes were 61.5, 59.8, and 51.8 mV/decade for Na+, NH4+, and K+, respectively. These electrodes also gave Nernstian responses (57.3 mV/ decade) to the creatininium ion activity in the range from 0.1 to 1 mM. (The theoretical response slope at 26 °C is 59.4 mV/ decade.) At higher concentrations, super-Nernstian slopes up to 72 mV/decade were observed in experiments when the creatininium activity was stepwise decreased, and slightly sub-Nernstian responses were observed when creatininium hydrochloride was added to the solution. The reason for this response behavior is not fully clear but may be related to the extraction of neutral creatinine into the membrane. A similar response behavior has been previously discussed42,43 and was observed, for example, for salicylate ISEs.42 Extractions. Extractions of undiluted urine samples were performed by very gentle stirring of a typically 100-mL sample and 100 mL of methylene chloride or chloroform over at least 6 h. The organic phase was then dried over sodium sulfate, and the solvent was removed using a rotary evaporator. FAB-MS. Positive mode fast atom bombardment mass spectra were recorded on a JEOL spectrometer using Xe as ionization gas. Because the urine extracts were viscous liquids, addition of a sample matrix was not necessary. Ion Chromatography (IC). IC was performed with a cation exchanger column IC-C1, an ion chromatograph HIC 6A of Shimadzu (Kyoto, Japan), and 5 mM HNO3 as elution buffer. Filtration. Urine samples were filtered through ultrafiltration membranes (YM1 ultrafilter, Amicon, Beverly, MA) with a cutoff value at 1000 u. RESULTS AND DISCUSSION To obtain a strong response to the creatininium cation, preliminary tests with various ionophore-free ion-exchanger membranes were performed using 1.35 M acetate buffer solutions of pH 3.8. At this pH, creatinine, 1 (see Figure 1), occurs predominantly in its singly protonated form (pKa ) 4.88).44 Membranes with DOP as membrane plasticizer responded nearly equally strongly to K+, and creatininium, and membranes with oNPOE as membrane plasticizer exhibited a significant K+ pot interference, too (log Kcreatininium,K > -1). In contrast, all of the membranes that were prepared using chloroparaffin as membrane plasticizer suffered from a much smaller K+ interference (log pot Kcreatininium,K > -2.0). This dependence of the selectivity on the membrane plasticizer is not very surprising, because chloroparaffin lacks functional groups that can strongly interact with K+, but DOP and oNPOE have polar groups that can solvate K+ by ion-dipole interactions. Chloroparaffin-PVC membranes with tetrakis[3,5-bis(trifluoromethyl)phenyl]borate as ion-exchange sites regularly gave Nernstian response slopes (57 ( 2 mV/ decade) in the creatininium activity range from 0.1 to 1 mM, but in the same activity range, chloroparaffin-PVC membranes with tetrakis(p-chlorophenyl)borate frequently gave response slopes as small as 36 mV/decade. In view of the high K+ concentrations in urine, all subsequent experiments were performed with chloro(42) Egorov, V. V.; Borisenko, N. D.; Rakhman’ko, E. M. J. Anal. Chem. 1998, 53, 750. (43) Bu ¨ hlmann, P.; Umezawa, Y. Electroanalysis 1999, 11, 687 and references therein. (44) Grzybowski, A. K.; Datta, S. P. J. Chem. Soc. 1964, 187.
Analytical Chemistry, Vol. 73, No. 14, July 15, 2001
3201
Figure 2. Emf drifts observed using a chloroparaffin-PVC-KTFPB membrane in three consecutive measurements performed for 10 different urine samples diluted with 1.35 M acetate buffer (pH ) 3.8). To make the drifts readily apparent, the emf values of the first measurements of each sample are normalized to 0 mV. Emf values measured for the same sample are connected by a solid line.
Figure 1. Structure formulas of creatinine, 1, and the natural lipophilic compounds cholesterol, 2; cholic acid, 3; taurocholic acid, 4; phosphatidylethanolamine, 5; phosphatidylserine, 6; p-cresol, 7; and coproporphyrin I, 8; which were tested as possible electrically neutral interfering agents.
paraffin as membrane plasticizer and KTFPB to provide for ionic sites. The responses of chloroparaffin- PVC-KTFPB membranes to NH4+, Na+, and K+ were well-reproducible from electrode to electrode, as can be seen from the small standard deviations of pot pot the selectivity coefficients (SSM; log KNH and KNH ). The 4,Na 4,K range of experimentally observed creatinine selectivity values pot was wider (log Kcreatininium,Na ) -3.2 ( 0.7). Electrodes with high pot selectivity were characterized by log Kcreatininium,Na ) -4.1, log pot pot Kcreatininium,NH4 ) -2.9, and log Kcreatininium,K ) -3.1. Even though these selectivities are only governed by the lipophilicity of creatininium and the interfering ions, they seemed high enough for measurements in urine, which typically contains ∼13 mM creatinine, 160 mM Na+, 35 mM NH4+, and 60 mM K+.45 For the analysis of real urine samples, fresh morning samples of 20 volunteers were diluted by a factor of 10 with 1.35 M acetate buffer. Measurements were then performed by measuring the diluted sample itself, adding aliquots of a concentrated creatinine solution to raise the creatinine concentration to 10-4, 10-3, and 10-2 M, and finally, by rinsing the electrode in acetate buffer. After each step, the emf value was recorded when the measured value was stable within 0.1 mV per 30 s.46 This sequence of standard additions and rinsing was performed three times and was then followed by a change-over to the next sample. The evaluation of these standard addition experiments gave creatinine concentrations that fell in the order of magnitude in the same range as the creatinine concentrations determined independently using the classical Jaffe´ method, but did not allow accurate creatinine determinations. Instead, the measurements revealed significant (45) Geigy Scientific Tables: Units of Measurement, Body Fluids, Composition of the Body, Nutrition; Lentne, C., Ed.; Ciba-Geigy: Basel, 1981. (46) Uemasu, I.; Umezawa, Y. Anal. Chem. 1982, 54, 1198.
3202 Analytical Chemistry, Vol. 73, No. 14, July 15, 2001
emf drifts that had not been similarly observed for pure creatinine, Na+, NH4+, and K+ solutions. This drift is illustrated by Figure 2, which shows the three consecutively measured emf values for each of the first 10 of the 20 diluted urine samples. Instead of three identical emf values for each sample, steady increases of the measured emf from the first to the third measurement were observed. Although the extent of the drift depended evidently on the actual composition of each sample, the same kind of drifts was also observed for the other set of 10 samples. To elucidate the origin of this emf drift, we at first measured emf responses to various cations that were considered to be possible interfering agents. Urine is a highly complex liquid, but because of the clinical relevance of urine analysis, its natural composition is well-known. The appearance of a synthetic compound in samples of 20 healthy volunteers seemed possible, but is very unlikely the explanation of more than a minor section of the whole observed emf drift shown in Figure 2. Therefore, extensive tables45 describing the composition of urine seemed to offer a reasonable chance of revealing the nature of a possible cationic interfering agent. In addition to interfering agents that are cationic independent of pH, compounds that can be protonated at pH 3.8 were considered; however, testing the emf responses to guanidine, carnitine, dimethylamine, tyrosine, threonine, 2-phenylethylamine, alanine, serine, histidine, arginine, glycine, glycocyamine, lysine, N -methyl-L-lysine, and calcium in physiological concentrations gave responses of a few mV at most and cannot explain these drifts in the presence of much more strongly responding ions. This suggested that no single cation causes the unexpected responses. Importantly, the electrodes responded very strongly to undiluted urine samples of neutral or slightly alkaline pH. Within a few minutes, the emf rose to values several tens of mV larger than the one expected for a typical creatininium concentration in urine, even though 50 mV. Changing to KCl-free solutions of these compounds at the end of the experiments resulted in potential drops of 104, 98, and 136 mV for cholic acid, phosphatidylethanolamine, and octanoic acid, respectively, clearly showing that the emf responses to these compounds are not simply artifacts from a hydrogen ion response. For comparison, Figure 5 shows the responses of electrodes dipped into 4 mM NH4Cl. Again, the addition of p-cresol (7) has little effect on the emf. The response to taurocholic acid (4) is very small, which is apparently the result of its fairly low lipophilicity (cmc ) 10 mM).51 The emf responses to phosphatidylethanolamine (5) and cholic acid (3) are similar, as in the case for K+. In contrast to the K+ case, coproporphyrin I (8) has no discernible effect and the emf response to octanoic acid (9) is quite small, but a small effect of cholesterol (2) is discernible. By far, the biggest difference in the responses for the K+ and NH4+ cases is found for phosphatidylserine (6). The slowness of this response can be explained by the very low solubility of phosphatidylserine in the aqueous sample solution. Although enough compound was added to the aqueous solution to give a 10-5 M total concentration, a significant amount, if not the majority of the compound, did not dissolve but, instead, could be seen sticking to the walls of the sample beaker. At these very low concentrations of effectively dissolved phosphatidylserine, the 3204 Analytical Chemistry, Vol. 73, No. 14, July 15, 2001
Figure 5. Emf response of an electrode to a chloroparaffin-PVCKTFPB membrane dipped into 4 mM NH4Cl solution upon addition of various natural lipophilic compounds: cholesterol, 2 (10-5.0 M); cholic acid, 3 (10-5.0 M); taurocholic acid, 4 (10-5.0 M); phosphatidylethanolamine, 5 (10-5.0 M); phosphatidylserine, 6 (10-5.0 M); p-cresol, 7 (10-4.0 M); coproporphyrin I, 8 (10-5.0 M); and octanoic acid, 9 (10-4.0). All of the concentrations indicate total concentrations in the sample and do not necessarily mean that all of the compound dissolved homogeneously in the sample solution. Increases of the lipid concentration during the course of the experiment are indicated by arrows.
mass transport of phosphatidylserine into the ISE membrane becomes very slow, even if the sample is vigorously stirred. The different drift behaviors in the K+ and NH4+ cases suggest that some of the lipids are extracted into the membrane, where they interact to a different extent with the K+ or NH4+ ions. The drifts caused by these lipids appear to have an origin that is very similar to those previously observed for synthetic surfactants,20 that is, extraction of the electrically neutral interfering agent into the sensor membrane, where it interacts ionophore-like with analyte and interfering ions. The rates with which the emf values change appear to be determined by mass transport, and the ultimate extent of the emf drift appears to reflect the strength of the surfactant-interfering agent and surfactant-analyte interactions. In accordance with this explanation, Figures 4 and 5 show that quick emf drifts are not necessarily small and large emf drifts not necessarily slow. Interestingly, the effect of 6, which is much stronger on the NH4+ response than than it is on the K+ response (Figures 6 and 5, respectively), indicates a different extent of interaction between 6 and these two ions, which must be related to the structure of this lipid. It appears likely that strong hydrogen bonds between NH4+ and the phosphate ester group affects the observed drifts. Evidently, the lipid extraction into the membrane must also affect pot the potentiometric selectivities. To demonstrate this, KNH and 4,Na pot KNH4,K selectivity coefficients of electrodes before and after exposure of the membranes to natural lipids were determined. Because of the high lipophilicity of these lipids, their backextraction from the membrane into the aqueous sample during the relatively brief period of selectivity determination is not a significant problem.52 As a result, the selectivity measurements after the lipid exposure with lipid-free solutions give a good
Table 1. Selectivity Coefficients of a Chloroparaffin-PVC-KTFPB Membrane before and after Exposure to Phosphatidylserine (10-5 M,a 6), Phosphatidylethanolamine (10-5 M,a 5), and Cholic Acid (10-4 M, 3)
pot log KNH 4,Na pot log KNH4,K
before
6
5
3
-1.0 -0.2
-0.6 -0.1
-0.5 0.0
-0.4 -0.1
a
Not fully soluble at this concentration; value gives the total amount of lipid in the sample (1 L).
indication of the selectivity of lipid-contaminated membranes. pot Table 1 shows that the value of KNH is more affected by the 4,Na exposure of the membranes to phosphatidylserine, 6, phosphatipot dylethanolamine, 5, and cholic acid, 3, than is that of KNH . 4,K From a molecular point of view, this is not very surprising. The Na+ ion is much smaller than the K+ ion and, because of a higher charge density, is well-known to interact more strongly with dipolar functional groups,39 as they are part of an extracted lipid. It should also be noted that these results cannot be explained by an unaccounted effect of lipids on the liquid junction potential at the reference electrode. Importantly, extraction of natural lipids affects not only ionophore-free ion-exchanger electrodes but also ionophore-based ISEs. An ideally selective ionophore interacts with only the target analyte, not with an interfering ion, thus avoiding any stabilization of the latter ion in the membrane.3 The lack of any competition from an ionophore for interaction with the interfering ion should make an ISE based on such an ionophore particularly sensitive to selectivity changes induced by lipid extraction. On the other hand, an ISE based on an ionophore that interacts with both the analyte and interfering ions is expected to exhibit a more robust selectivity. Knowing that valinomycin interacts strongly with K+, NH4+, and Na+,53 we expected only small, if any, changes in the selectivity of a K+-selective electrode based on valinomycin. Indeed, Table 2 shows that phosphatidylserine and cholic acid pot pot had no major effect on KK,NH and KK,Na ; however, when a urine 4 sample was extracted with methylene chloride, the organic solvent was evaporated, the residue was redissolved in water to give a concentration approximately identical to the concentration of the methylene chloride-extracted compounds in undiluted urine, and pot pot KK,NH and KK,Na selectivities were determined after exposure of 4 the electrode to this solution for 10 h, the observed sodium interference was significantly larger than before the exposure (see Table 2). This shows that losses in selectivity due to exposure of the sensor membrane to natural lipids can occur even for the widely used and generally well-behaved valinomycin-based elec(52) Dinten, E.; Spichiger, U. E.; Chaniotakis, N.; Gehrig, P.; Rusterholz, B.; Morf, W. E.; Simon, W. Anal. Chem. 1991, 63, 596. (53) Bakker, E.; Willer, M.; Lerchi, M.; Seiler, K.; Pretsch, E. Anal. Chem. 1994, 66, 516.
Table 2. Selectivity Coefficients of a Valinomycin-Based ISEa before and after Exposure to Phosphatidylserine (10-5 M,b 6), Cholic Acid (10-4 M, 3), and a Methylene Chloride Extract of Urinec
pot log KK,Na pot log KK,NH 4
before
6
3
urine extract
-3.9 -1.9
-4.1 -1.8
-4.2 -1.7
-3.1 -1.9
a Membrane composition: 1.4 wt % ionophore, 64 mol % KTClPB, and bis(2-ethylhexyl) sebacate-PVC (2:1 w/w). b Not fully soluble at this concentration; value gives the total amount of lipid in the sample (1 L). c See text.
trode. The large selectivity change in the case of the urine extract sample may result from a combination of the interferences from several of the compounds investigated here. Alternatively, electrically neutral interfering agents not considered so far may exist. CONCLUSIONS In summary, we have shown that highly lipophilic compounds that naturally occur in relatively low concentrations in urine can cause significant emf drifts and changes in selectivities of ionophore-free and valinomycin-based membrane electrodes. Similar interferences for ISEs based on other ionophores for cations or anions seem very probable. It appears likely that such drifts and selectivity changes also occur in other samples, such as blood serum or plasma or environmental samples. The mechanism of the interference from lipids appears to be the same as the one previously suggested for the drifts due to synthetic surfactants,20 that is, extraction of the electrically neutral interfering agent into the sensor membrane, where it interacts ionophore-like with analyte and interfering ions. Importantly, a correct understanding of the lipid interferences reported here may eventually also open ways to overcome them. A relatively simple way to prevent the accumulation of electrically neutral lipids in ISE membranes seems, for example, to be given by the use of only brief exposures to samples and extended washing cycles that deplete the ISE membrane of these lipids. Simple frequent recalibrations may be another solution to the drift problem if possible interfering agents occur in sufficiently low concentrations that do not change the selectivity significantly. ACKNOWLEDGMENT This work was supported by the Ministry of Education, Science, and Culture, Japan. We gratefully acknowledge Tatsuyuki Harita of Iatron Laboratories, Tokyo, Japan, for his assistance with the analysis of creatinine in urine with the Jaffe´ and an enzymatic reaction. S.A. thanks the Japan Society for the Promotion of Science (JSPS) for a postdoctoral fellowship. Received for review December 19, 2000. Accepted May 8, 2001. AC0015016
Analytical Chemistry, Vol. 73, No. 14, July 15, 2001
3205