Anal. Chem. 2009, 81, 5961–5965
Determination of Potassium in Red Blood Cells Using Unmeasured Volumes of Whole Blood and Combined Sodium/Potassium-Selective Membrane Electrode Measurements Mariusz Pietrzak and Mark E. Meyerhoff* Department of Chemistry, The University of Michigan, Ann Arbor, Michigan 48109 Recent studies suggest that the measurement of intracellular potassium concentrations in red blood cells (RBCK) can be a marker for assessing the risk, development, and treatment of hypertension. In this work, the combined use of miniature potassium- and sodium-selective membrane electrodes is evaluated as a simple means to determine RBC-K. The proposed method requires two separate sets of electrode measurements: (i) potassium and sodium concentrations in the plasma phase of an unmeasured volume of a whole blood sample, and (ii) determination of potassium and sodium concentrations in the same sample of blood after complete hemolysis by ultrasonic disruption of the RBC membranes. The dilution of sodium concentration after hemolysis can be used to determine hematocrit (Hct) (volume of red cells per unit volume of blood) of the blood. The concentration of potassium within the red blood cells (RBCs) is then calculated using the measured change in potassium levels before and after RBCs lysis and the hematocrit level determined from the sodium electrode measurements and/or a conventional centrifugation method. Good correlation for RBC-K between the proposed method and traditional flame photometry is observed for animal blood samples that possess the range of potassium levels found within human RBCs (80-120 mM). However, when potassium is much lower than that found in human RBCs (known to occur for certain animal species), the Hct measured by the sodium electrode method is falsely low, compared to traditional spun hematocrit values, because of an increased level of sodium within the RBCs, necessitating use of spun Hct levels to assess RBC-K accurately. It is envisioned that this new approach could be further miniaturized into a single-use disposable cartridge type electrode system that would enable rapid point-of-care screening of RBC-K levels in human subjects. Potassium is one of the major elements that are essential for the body’s growth and maintenance of critical physiological functions.1 Based on clinical and epidemiologic studies, one of the many important functions that potassium fulfills in the body * To whom correspondence should be addressed. E-mail: mmeyerho@ umich.edu. (1) He, F. J.; MacGregor, G. A. Physiol. Plant. 2008, 133, 725–735. 10.1021/ac900776d CCC: $40.75 2009 American Chemical Society Published on Web 06/12/2009
is control of heart function and, ultimately, blood pressure regulation.2 Recent research has shown that, in addition to extracellular potassium levels (normally detected in plasma phase of the blood), intracellular potassium concentration is highly associated with blood pressure levels.3,4 Hypertensive patients and those who are at risk for developing hypertension distinguish themselves with relatively low potassium content within red blood cells (RBCs), when compared with normotensive control subjects.4 Therefore, the determination of potassium level in RBCs (RBCK) can be a very important diagnostic tool for clinical analysis. To date, most interest in using potassium measurements as a diagnostic tool has focused on detecting extracellular potassium levels in the plasma phase of blood. Initially, flame photometry was the most popular method for determining the level of potassium in blood; however, over the past 30 years, the use of potassium-selective membrane electrodes that employ valinomycin or bis[(benzo-15-crown-5)-4′-methyl]pimelate (BME 44) as ionophores in polymeric films have completely replaced flame photometry for such measurements. Indeed, all modern electrolyte analyzers used in hospitals employ polymer-membrane-type ionselective electrodes (ISEs) for the rapid detection of not only potassium, but also sodium, calcium, chloride, and magnesium in the extracellular aqueous phase of small volumes of undiluted whole blood.5 Compared to flame photometers, polymer-membrane ISEs are very simple, inexpensive, and robust devices that can be readily prepared in miniaturized form as either simple dipcatheter-type probes, or as planar devices using modern screen printing technology on substrates (plastics, ceramic, etc.) that possess the required electrochemical connections using screenprinted or vapor-deposited metal leads. The direct determination of RBC-K is not possible without first lysing erythrocytes to release potassium and other cell components. At present, a classical flame photometry method is most often used for this determination;6 however, potassium-selective electrodes have already been suggested for this measurement via the use of a similar protocol.6-8 The typical procedure involves (2) He, F. J.; MacGregor, G. A. Am. J. Hypertens. 1999, 12, 849–851. (3) Delgado, M. C.; Delgado-Almeida, A. J. Hum. Hypertens. 2003, 17, 313– 318. (4) Delgado, M. C. Curr. Hypertens. Rep. 2004, 6, 31–35. (5) Amman, D. Ion-selective Microelectrodes; Springer-Verlag: Berlin, 1986. (6) Saulis, G.; Praneviciutea, R. Anal. Biochem. 2005, 345, 340–342. (7) Mangubat, E. A.; Hinds, T. R.; Vincenzi, F. F. Clin. Chem. 1978, 24, 635– 639.
Analytical Chemistry, Vol. 81, No. 14, July 15, 2009
5961
first separating RBCs from the plasma phase of blood by centrifugation and then placing the cells in a given volume of hypotonic solution (water) in which osmotic pressure will rupture the RBCs. The potassium level in the lysate solution (from flame photometry or ISE measurement), in combination with the known volume of cells that are lysed and the known volume of water added, enables the concentration of potassium per unit volume of cells to be reported (in units of mM). The goal of this work was to assess whether the procedure for RBC-K determinations can be greatly simplified using ISE measurements without requiring prior isolation of red blood cells via centrifugation, or any dilution of the original blood sample. For this purpose, miniaturized potassium- and sodium-selective electrodes were fabricated and used to detect RBC-K in undiluted animal (porcine, bovine, sheep) blood. Because the proposed method requires the knowledge of the blood hematocrit level, the sodium electrode was employed to assess Hct levels using the method reported by Kobos et al.,9 in which the change in sodium concentration (decrease) after cell hemolysis (by sonication) was employed to reliably detect the hematocrit of human blood. Concomitantly, the potassium electrode was used to detect the change in potassium concentration after cell lysis. With knowledge of the change in potassium concentration in the undiluted sample, and the volume of red cells from the sodium measurement (and/ or spun hematocrit values), the RBC-K values can be reported. It will be shown that this method works well in animal blood when RBC-K is in the ranges expected for human blood samples. Reagents. Valinomycin (potassium ionophore), sodium ionophore X, potassium tetrakis [3,5-bis-(trifluoromethyl)phenyl]borate (KTFPB), bis(2-ethylhexyl)sebacate (DOS), o-nitrophenyloctyl ether (o-NPOE), poly(vinyl chloride) (PVC), and tetrahydrofuran (THF) were purchased from Fluka (Ronkonkoma, NY). Nonionic surfactants Brij 58 and Triton X-100 were purchased from Aldrich Chemical Co. (Milwaukee, WI) and hydroxyethylcellulose (HEC) from Polyscience, INC (Warnington, PA). All aqueous solutions were prepared with salts of the highest purity available from Fluka using water passed through a Milli-Q (18.2 MΩ cm, Millipore Corp., Billerica, MA) system. All experiments were performed at ambient temperature. Preparation of ISE Electrodes. The potassium-selective membrane cocktail, consisting of 33 wt % of PVC, 66 wt % of DOS, 1 wt % of ionophore (valinomycin), and 30 mol % (relative to the ionophore) of KTFPB was prepared by dissolving appropriate amounts (total of 200 mg of these species) in 1 mL of freshly distilled THF. Similarly, the sodium-selective cocktail was composed of 33 wt % PVC, 66 wt % o-NPOE, 1 wt % of sodium ionophore X, and 30 mol % (relative to the ionophore) of KTFPB dissolved in 1 mL of THF. Two inner electrolyte hydrogels were prepared by dissolving hydroxyethylcellulose (HEC) in 10-2 M KCl or 10-2 M NaCl solution (the final concentration of HEC was 25 g/L), under heating at 60 °C for 5 min. These gels were incorporated into Nalgene plasticized PVC tubings (5 cm in length and an inner diameter (ID) of 1/32 in.), which served as the housing for the miniaturized ISEs. Silver/silver chloride wires inserted into the respective hydrogel internals served as the internal reference electrodes, and these were sealed in
place with parafilm. The tubes filled with the respective hydrogels (and inserted Ag/AgCl electrodes) were then dipcoated at the other ends with the corresponding membrane cocktail solutions. After dipcoating 10 times (with partial drying of the film for 10 min between each coating), the resulting potassium- and sodium-selective electrodes were kept at room temperature for at least 24 h, to allow complete evaporation of the THF solvent. The final dried ISE polymer membranes had thicknesses in the range of 120-150 µm. Potentiometric Measurements. After fabrication, the potentiometric response characteristics of the miniature K+-ISEs and Na+-ISEs were evaluated. The potentiometric selectivity coefficients (log Kpot) were determined using the fixed interference method (FIM).10 The measurements were conducted over a potassium or sodium concentration range of 10-7 M-2 × 10-1 M. Test solutions comprised of 10-1 M interfering cations (Na+, Li+, NH4+, Ca2+, and Mg2+ for K+-selective electrodes and K+, Li+, NH4+, Ca2+, and Mg2+ for Na+-selective electrodes) were used for this purpose. Before potassium and sodium determination in blood samples, a two-point calibration method was employed. The electrodes were calibrated simultaneously by immersing into calibrating solutions A and B (where A is comprised of 1 mmol/L of KCl and 159 mmol/L of NaCl, and B is composed of 150 mmol/L of KCl and 10 mmol/L of NaCl), and if the measured slopes were satisfactory (>50 mV/decade), the electrodes were used subsequently for electrolyte determinations in animal blood samples. Electrochemical potentials of the miniature electrodes were measured with the following galvanic cell: Ag/AgCl(s), KCl (3M)/ sample solution/ion-selective membrane/inner filling solution/ AgCl(s)/Ag using a leak-free miniaturized reference electrode (Warner Instruments, Hamden, CT). Electromotive force (EMF) values for the ISEs vs reference electrode were measured at ambient temperature (ca. 24 °C) via a personal computer (PC) coupled to a high-Z interface (VF-4, World Precision Instruments) and controlled by Labview software (version 7.0, National Instruments). For all calibrations and sample measurements, the K+ISE, Na+-ISE, and reference electrodes were placed in the test solutions or blood simultaneously, and measurements of the potentials of both ISE electrodes versus the reference electrode were recorded at the same time. Measurements in Blood. Approximately 1 mL of fresh animal blood (sheep, bovine, porcine; from Lampire Biological Laboratories, Pipersville, PA), kept at room temperature for 20 min, was transferred to a 2-mL polystyrene beaker cup (Evergreen Scientific, Los Angeles, CA). The three electrodes (K+-selective, Na+selective, and reference) were then immersed in the sample and EMF values that indicated K+ and Na+ in the plasma were recorded. Another ca. 1 mL of the same blood sample was transferred to a 2-mL glass vial (Supelco, Bellefonte, PA). The vial was then closed and the vial cap was sealed with parafilm. The sealed vial was placed in ultrasonic cleaner bath (Branson Model 5510), and sonication was allowed to proceed for 10-15 min, to lyse the RBCs completely. Note that it is not necessary to utilize two aliquots of the same blood sample for the proposed measurements; indeed, one aliquot of blood would
(8) Malon, A.; Maj-Zurawska, M. Sens. Actuators B 2005, 108, 828–831. (9) Kobos, R. K.; Abbott, S. D.; Levin, H. W.; Kilkson, H.; Peterson, D. R.; Dickinson, J. W. Clin. Chem. 1987, 33, 153–158.
(10) Umezawa, Y.; Umezawa, K.; Sato, H. Pure Appl. Chem. 1995, 67, 507– 518.
5962
Analytical Chemistry, Vol. 81, No. 14, July 15, 2009
suffice, with sequential measurements with the electrodes before and after lysis. The sample was then kept at room temperature for 10 min for temperature equilibration. In the current procedure, it was simply more convenient to place the electrodes into the beaker cup with blood to record the extracellular K+ and Na+ levels while simultaneously sonicating another aliquot in a sealed glass vial. After sonication, the same three electrodes were immersed in the sample and EMF values reflective of the changes in the K+ and Na+ levels after lysing of the blood were measured. Another 10 mL of the same blood sample was transferred to a 15 mL centrifuge tube and centrifuged for 30 min at 4000 rpm. Then, the hematocrit level was determined from the volume of the packed cells, and also used for RBC-K calculations and compared with the hematocrit level obtained using the Na+-ISE dilution method. The results of RBC-K determined using the ISE-based method were compared with those obtained by flame photometry, using a Shimadzu Model AA-6200 spectrometer. Emission measurements were conducted at 766.5 nm using an air-acetylene flame and diluted samples of RBCs separated from plasma by centrifugation. RESULTS AND DISCUSSION The goal of this project was to develop a method to determine intracellular potassium in undiluted blood samples using potentiometry. Moreover, we decided to examine whether a simultaneous hematocrit measurement based on the sodium ion dilution approach proposed previously would further simplify the measurement scheme.9 This method relies on the fact that intracellular sodium is very low compared to plasma phase sodium concentrations. Hence, after lysis of the RBCs, which is required to assess intracellular potassium concentration, a concomitant dilution of sodium concentration in the blood sample can be used to determine hematocrit levels. Equation 1, adapted from ref 9, provides Hct as a function of the ratio (R) of sodium ion concentration measured before and after lysing the RBCs:
Hct )
a(1 - R) a + R(b - a)
Table 1. Potentiometric Characteristics of (a) K+-Selective and (b) Na+-Selective Miniature ISEs (a) K+-Selective Electrodes interfering cation +
Na Li+ NH4+ Mg2+ Ca2+
log KK+,X
slopea (mV/decade)
-5.0 -4.5 -2.1 -4.9 -4.3
58.4 ± 0.6 58.2 ± 0.4 57.8 ± 0.2 57.8 ± 0.6 57.5 ± 0.3
(b) Na+-Selective Electrodes interfering cation +
K Li+ NH4+ Mg2+ Ca2+
log KNa+,X
slopea (mV/decade)
-2.5 -3.1 -2.6 -3.7 -3.2
57.4 ± 0.4 58.1 ± 0.5 56.8 ± 0.3 57.7 ± 0.6 57.5 ± 0.4
a Average ± one standard deviation (sd) of n ) 3 electrodes of each type. Slopes reported are observed toward primary ion in presence of 0.1 M interfering ion.
Figure 1. Potentiometric response to changes in potassium or sodium concentration (M) in a background of (9) 10-1 M NaCl for K+-ISE or (b) 10-1 M KCl for Na+-ISE.
(1)
where a is the fraction of water in the plasma (typically, a ) 0.93), and b is the fraction of water inside the erythrocytes (typically b ) 0.65). Therefore, the use of two miniaturized ISEs that exhibit high selectivity toward potassium and sodium under conditions in which the blood can be readily lysed is essential to provide a simplified approach to the RBC-K measurement. POTENTIOMETRIC CHARACTERIZATION OF K+-SELECTIVE AND Na+-SELECTIVE ELECTRODES The miniature K+ and Na+ membrane electrodes prepared were characterized for their potential application RBC-K measurements. Calculated values of the selectivity coefficients and slopes are summarized in Table 1. Figure 1 presents typical calibration curves of the electrodes in a background of high concentration of the main interferent cations in blood (0.1 M NaCl for K+-selective electrode and 0.1 M KCl for Na+-selective electrode). The potassium-selective electrodes exhibited Nernstian calibration slopes (Table 1) over a wide (2 · 10-6 - 2 · 10-1 M) potassium concentration range. In the case of sodium-
selective electrodes, the slopes were also Nernstian, but the log-linear response range was much narrower (6 × 10-4 M-2 × 10-1 M) (see Figure 1), albeit still adequate for the sodium measurements in blood, where sodium levels even after RBCs lysis are >50 mM. Moreover, both types of sensors showed a fast (t95 < 10 s) response to changes in target ion concentration. All response parameters remained stable for at least 1 month for both the K+- and Na+- ISEs. It is well-known that hemolysis occurs when erythrocytes are exposed to either surfactants or ultrasonic energy.11,12 Because the proposed method of potassium determination is based on RBCs lysis, in preliminary studies, both approaches were examined. The sonication method proved to be more attractive, because it was determined that the addition of typical nonionic surfactants to test solutions had a rather significant negative impact on the selectivity of the Na+-ISE (but only very small effects on the K+-ISE). Indeed use of small quantities of Brij 58 or Triton X caused deterioration of the potentiometric selectivity for sodium (11) Miller, M. W. Ultrasound Med. Biol. 2004, 30, 1263–1267. (12) Hutchinson, E.; Sibley, M. J.; Abram, C. Colloid Polym. Sci. 1968, 225, 167–175.
Analytical Chemistry, Vol. 81, No. 14, July 15, 2009
5963
Table 2. Results of Hematocrit Determination Using Na+-Dilution Method and Centrifugation Methods, as Well as the Resulting RBC-K Values Determined Utilizing the Proposed ISE Method and Flame Photometry RBC-K (mM)a
Hematocrit +
b
blood sample
centrifugation
Na -dilution
ISE
ISEc
Flame Photometryb
porcine 1 porcine 2 porcine 3 porcine 4 porcine 5
0.45 0.42 0.42 0.49 0.39
0.45 0.41 0.43 0.48 0.40
104.5 ± 1.0 98.2 ± 1.3 92.3 ± 1.2 96.7 ± 1.0 93.7 ± 1.2
104.5 ± 1.0 100.5 ± 1.4 90.3 ± 1.2 98.6 ± 1.0 91.5 ± 1.2
105.6 ± 1.3 100.0 ± 1.4 93.8 ± 1.2 97.4 ± 1.2 94.7 ± 1.4
sheep 1 sheep 2 sheep 3
0.33 0.38 0.43
0.31 0.21 0.02
78.1 ± 0.8 38.7 ± 1.4 19.5 ± 0.8
82.7 ± 1.0 65.4 ± 1.8 326.0 ± 5.2
77.6 ± 1.6 39.9 ± 1.2 20.7 ± 1.0
bovine 1 bovine 2 bovine 3
0.38 0.45 0.42
0.13 0.05 0.40
32.9 ± 0.6 39.1 ± 0.8 79.4 ± 1.0
87.7 ± 1.2 307.1 ± 4.6 83.2 ± 1.2
33.4 ± 0.8 39.9 ± 1.0 79.8 ± 0.6
a Results reported represent an average value ± one standard deviation (sd) for n ) 3 measurements for each sample. b Results calculated using hematocrit values obtained by centrifugation. c Results calculated using hematocrit values obtained by Na+-dilution.
over potassium cation by over 2 orders of magnitude, completely eliminating the possibility of using the Na+-ISE to detect hematocrit levels via the sodium dilution method. Therefore, the sonication method was chosen for all further experiments with blood. Blood Measurements. The first step for RBC-K measurements was recording the EMF for the K+-ISE, as well as Na+ISE, vs reference electrode in a whole blood sample. Based on prior calibration of the K+- and Na+-sensors with known KCl/NaCl solutions (assembled to have an ionic strength comparable with blood), the voltage measured in the whole blood sample provides a value for the plasma potassium and sodium concentrations. In the second step, the voltages between the K+-ISE, as well as Na+-ISE, and the reference electrode, were measured in the same blood sample (a different aliquot in this case, but this is not a requirement, as sequential measurements in the same aliquot of blood can be made) after RBCs were lysed by sonication. This EMF value provides the final sodium and potassium levels originating from both the plasma phase and within the RBCs. Based on changes in the measured Na+ levels, the Hct value for the given sample was calculated using eq 1. In addition, another aliquot of the same blood sample was centrifuged to determine the definitive Hct value. Changes in measured K+ concentration after lysis can be converted to RBC-K values by dividing this concentration change (in millimolar (mM)) by the volume of red cells per liter of blood (Hct). The entire electrochemical measurement process of single sample requires no more than 2 min, although sonication with current cleaning bath requires much longer times (15 min) than what would likely be needed if optimized sources of ultrasonic energy were employed to disrupt the RBC cell membranes. A comparison of the hematocrit values obtained for animal blood samples by centrifugation method and the Na+-dilution method are presented in Table 2. In addition, Table 2 indicates the RBC-K values determined from the electrode measurements using both approaches for Hct measurements, along with the levels determined by conventional flame photometry method.13 As can be seen, the Hct values correlate very well for samples in (13) Overman, R. R.; Davis, A. K. J. Biol. Chem. 1947, 168, 641–649.
5964
Analytical Chemistry, Vol. 81, No. 14, July 15, 2009
which the RBC-K values fall within the expected ranges for human blood, 80-120 mM of cells,14,15 as measured here by flame photometry. With such correlation, the values calculated for RBC-K when using the Na+ dilution method combined with the K+-ISE determinations have excellent agreement with the flame photometry results. Similarly, the RBC-K values obtained when using the Hct levels measured by centrifugation, along with the K+-ISE measurements, give equivalent results for all samples tested. However, when RBC-K determined by flame photometry was much lower than anticipated, which can occur in certain animal species (bovine, sheep),16-18 the Hct values determined by the Na+ dilution approach exhibit significant negative error compared to centrifugation, resulting in a false high RBC-K value for such samples. This is because when intracellular K+ concentrations are very low (much lower than can be encountered in human samples), the levels of Na+ in the RBCs are much higher than expected, making change in Na+ concentration after lysis much smaller. As shown, for all five samples of porcine blood, all the potentiometric results for Hct determination agreed with those measured using centrifugation, and resulting RBC-K values agreed among all the methods. For all test animals used in this preliminary study, it is known that the physiology of pigs is the most similar to humans.19,20 Indeed, in human samples reported by Delgado and Delgado-Almeida,3 as well as Bugyi et al.,14 for flame photometric determination of RBC-K, the variation in the values ranged only from 81 mM to 111 mM. CONCLUSIONS In conclusion, a very simple new approach for measurement of the intracellular potassium concentration in red blood cells (14) Bugyi, H. I.; Magnier, E.; Joseph, W.; Frank, G. Clin. Chem. 1969, 15, 712–719. (15) Nagaki, J.; Teraoka, M. Clin. Chim. Acta 1976, 66, 453–455. (16) Christinaz, P.; Schatzmann, J. J. Physiol. 1972, 224, 391–406. (17) Soysal, M. I.; Gurcan, E.; Kok, K. S. Trakia J. Sci. 2005, 3, 8–10. (18) Agar, N. S.; Gruca, M. A.; Hellquist, L. N.; Harley, J. D.; Roberts, J. Experientia 1977, 33, 670–671. (19) Musial, F.; Crowell, M. D.; French, A. W. Physiol. Behav. 1992, 51, 643– 646. (20) Zhang, X.; Feng, Z.; Feng, M.; Wang, H.; Liqin, B. Chin. Sci. Bull. 2000, 45, 626–630.
(RBC-K) levels has been developed using miniature potassium and sodium ion-selective electrodes. The method does not require precise pipetting of blood volumes, because no dilution of the blood occurs during the proposed measurement process. The simultaneous measurement of potassium and sodium levels before and after lysing of the blood sample by sonication provides an accurate method to determine RBC-K, provided that the RBC-K levels fall within the range expected for normal and abnormal (hypertensive or prehypertensive) humans. In the future, if desired, further improvement in the reliability of the electrochemical Hct measurement can be readily achieved by adding a simple conductivity measurement of the blood sample via the addition of two small platinum electrodes to the measurement arrangement.21 In combination with the initially determined sodium levels in the plasma phase (used to correct the measured blood
conductance for differences, because of variations in highly conductive Na+ concentrations),21 accurate determination of Hct levels can be made without dependence on Na+ dilution measurements. However, we believe that the current methodology will provide acceptable results for the vast majority of human samples, and we curently await IRB approval to begin testing the proposed approach versus flame photometry in human trials with hypertensive and normotensive subjects.
Received for review April 10, 2009. Accepted May 20, 2009. AC900776D (21) McMahon, D. J.; Carpenter, R. L. Anesth. Analg. 1990, 71, 541–544.
Analytical Chemistry, Vol. 81, No. 14, July 15, 2009
5965
