Anal. Chetn. 1002, 64, 091-094
691
Quantitative Determination of Catecholamines in Individual Bovine Adrenomedullary Cells by Reversed-Phase Microcolumn Liquid Chromatography with Electrochemical Detection Bruce R. Cooper, Jeffrey A. Jankowski, David J. Leszczyszyn, R.M a r k Wightman, and James W. Jorgenson* Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290
A microcolumn liquld chromatographic method for the detennhakn ol “phphrlne and e p h p h r h in 8lngie bovine adrenomedullary cells is described. A single cell le isolated from a cultwe plate, (3,edihydroxybenzyl)amlne Is added as an Internal standard, and the cdl kiysed wtth perchkrlc acid. After centrlfugation, the supernatant Is injected directly into a 42 or 50 pm inner diameter C,, reversed-phase column opwated wlth d.ctroch.mical detection. Detection llmlk for norepinephrine and epinephrine were 46 and 75 amoi, respectively. 01 the 22 cells examined, 8 contained predomlnantly norepinephrine, 10 contained predominantly eplnephrlne, and at least 4 cells contained rlgnlflcant amounts of each. On average, a single cdl contalned approxhatdy 150 hnol of catecholamine.
INTRODUCTION It is traditionally accepted that mammalian adrenomedullary tissue is composed of two types of catecholamine secreting cells. One cell type contains predominantly norepinephrine (NE), while the other type predominantly contains epinephrine (E). Although both cells synthesize NE, it is believed that only E cells contain phenylethanolamine N-methyltransferase(PNMT), the enzyme which converts NE to E. Cells containing PNMT are considered E-storing cells, while those cells without PNMT are considered NE-storing cells. The two cell type theory is based on histochemical studies of medullary tissue. The three most popular staining procedures used were formalin,’ iodate,2and formaldehyde followed by osmium t e t r ~ x i d e .Since ~ all of these stains were selective for NE in vitro, colored cells were believed to be NE-storing cells. Work has also focused on the size, shape, and eledron density of vesicles after fixation. Electron density refers to how darkly stained an object appears when viewed by an electron microscope. It was claimed that small, highly electron dense vesicles contained NE, while E-storing cells contained vesicles which were larger with leas electron density. All of these methods, however, have been criticized. Benedeczky et al. found discrepancies both in vivo and in vitro with the formaldehyde/osmium tetroxide m e t h ~ d . Others ~ ? ~ saw cells with large granules, some with small granules, and some cells with granules of various sizesS6Some researchers concluded that there are not two cell types, but three.’ Those researchers who could not distinguish between NE- and Estoring cells suggested that a transition from one cell type to the other may O C C U T . ~ ~ ~ Past work in this laboratory has used 15-20-pm open tubular liquid chromatography (OTLC) with electrochemical detection to determine quantitatively the neurotransmitters8 and free amino acid poolsQin single neuron cells of the land snail Helix Aspersia. Capillary liquid chromatography is well
* To whom correspondence should be addressed. 0003-2700/92/0364-0691$03.00/0
suited for single-cell analysis because chromatographic resolution is high and injection volumes require only a few nanoliters. This report discusses the w e of packed microcolumns with electrochemical detection to determine quantitatively the amount of NE and E in single bovine adrenomedullary cells. OTLC columns are not used because they do not provide sufficient retention for polar compounds, such as NE. This is the first report using packed microcolumns to analyze single cells, and where mammalian cells with 16-pm diameters are used. All past work on the separation of compounds at the single-cell level has focused on larger invertebrate EXPERIMENTAL SECTION Reagents. Standards, reagents, and mobile-phase constituents were purchased from Sigma Chemical Co. (St. Louis, MO). Ascorbic acid was obtained from Aldrich Chemical Co. (Milwaukee, WI). All chemicals were used as received. A balanced salt solution was composed of 150 mM NaC1,4.2 mM KC1, 1.0 mM NaH2P04,11.2 mM glucose, 0.7 mM MgC12, 2.0 mM CaC12,and 10.0 mM HEPES and adjusted to pH 7.4 with NaOH. The balanced salt solution was made with doubly distilled water. A phosphate-buffered mobile phase was prepared by diluting 85% phosphoric acid to 0.1 M, adjusting to pH 3.0 with NaOH, adding 1mM EDTA, and filtering through a 0.45-pm Nylon filter (Fisher). The mobile phase and all standards were made up in deionized, 0.2 pm filtered water (Barnstead). Chromatographic System. The chromatography columns used were 42 or 50 pm inner diameter fused-silica capillaries, 50 cm long, and slurry packed in our laboratory with YMC ODs-AQ 5-pm spherical C18 particles (YMC, Morris Plains, NJ). The packing procedure is described elsewhere16and was followed with slight modification. A column frit is constructed by sintering a 150 pm long band of glass beads 1 mm from one end of the fused-silica capillary. The 1-mm gap between the frit and end of the capillary allows for the insertion of the working (detection) electrode. The nonfritted end of the fused silica is placed in a high-pressure slurry reservoir containing a slurry of 1:lO (w/v) packing to hexane. An Altex Model llOA pump forces isopropyl alcohol into the reservoir at 3000 psi. The frit allows the hexane to pass while the stationary phase is collected. A column is typically packed in 4-5 h, and the pressure is left on overnight to facilitate settling of the packing. Detection. The detector used was a carbon-fiber electrode which has been described before.I6 The carbon fiber, having a diameter of 9 pm and a length of 1mm, was inserted directly into the end of the capillary column. All of the catecholamines of interest have an oxidation peak potential of +0.6 V vs a Ag/AgCl reference electrode. Amperometric detection was obtained by applying +0.73 V to the working electrode. Currents produced were measured with a Keithley 610C electrometer (Cleveland, OH), and the output voltage was fed directly to a 16-bit analog-to-digitalconverter. Data acquisition and manipulation were performed with an IBM/XT microcomputer. In order to achieve high run-to-run reproducibility, the carbon-fiber working electrode was electrochemically cleaned before each run by applying a triangular waveform from 0 to +1.8 V for 30 s at a rate of 1 V/s.l’ Sample Preparation. The bovine adrenomedullarycells used were cultured following the protocol of Wilson and V i v e r o ~ . ~ ~ ~ ’ ~ 0 1992 American Chemical Society
892
ANALYTICAL CHEMISTRY, VOL. 64, NO. 6, MARCH 15, 1992 Table I. Measured Amounts of Norepinephrine and Epinephrine in Individual Bovine Adrenomedullary Cells cell no.
measd amt. fmol E NE NE-Dominant Cells
28
33
38 43 TIME (MIN)
48
53
Flgurr 1. Chromatogram of a slngle bovlne adrenomedullary cell. The peak labels are as foibws: NE is norepinephrine, AA Is ascorbic acid, E Is eplnephrine, and DHBA is (3,4dlhydroxybenzyl)amine. For chromatographic conditions, see Experimental Section.
Two of the five cell preparations, however, involved a modification to the protocol in order to selectively enrich some culture plates in NE- and some in E-storing cells. This modification is based on that described by Garcia et aLZo During the latter stage of the cell preparation,a hemocytometer was wed to determine the concentration of chromaffim cells. The cells were then diluted to a fmal concentrationof 300000 cells/mL. A 2-mL aliquot was applied to each culture plate in order for each plate to contain approximately 600 000 cells. This corresponds to a cell density of 62000 chromaffin cells/cm2surface area. Only cells which had been in culture for 3-7 days were used. The cell growth medium was replaced with the balanced salt solution prior to analysis. A single cell was removed from the plate and transferred to a 250-nL microvial by the use of a pipet which had a 20 bm inner diameter. Approximately0.5-1 nL of balanced salt solution was transferred along with the cell. M (3,4-diExactly 10.0 nL of a solution of 1.00 X hydroxybenzy1)amine(DHBA), dissolved in 0.8 N HCIOl with 0.5 mM ascorbic acid, was added to the microvial by the use of an in-house constructed microdispenser pipet. The pnuematic microdispenser has been reported to be able to deliver 0.248nL with a relative standard deviation (RSD) of 3.38%.21 For the present work, the microdispenser was set to deliver a volume of 2.0 &/pulse. It is felt that the RSD should be at least equal to, but more likely even lower than, that reported in the literature because of the larger volume being dispensed. DHBA is used as an internal standard, ascorbic acid is added to prevent analyte oxidation, and HCIOI is used to lyse the cell. The microvial was capped with a small piece of parafilm and centrifuged at 12 OOO g for 8 min. The supernatant was injected directly into the chromatography column using a microinjection pipet which has been described before.21 All of the sample preparation and injections were done with the aid of a Wolfe Selectra I1 stereomicroscope (Carolina Biological Supply Co.), a Narishige MM-33 micromanipulator (Medical Systems Corp.), and an Oriel micropositioner (Oriel Corp.).
RESULTS AND DISCUSSION Figure 1displays a chromatogram from a single-cell run, shown after the column dead time. This particular run corresponds to cell number 14 in Table I. P e a h corresponding to NE and E are from a single cell, ascorbic acid is added as an antioxidant, and DHBA is an internal standard. Unlike most reversed-phase systems run with an aqueous mobile phase, this particular stationary phase retains ascorbic acid. In order to confirm the identity of these compounds, runswere performed initially in which the electrode potential was scanned from +0.3 to +1.2 V at a rate of 1V / S . ~This enabled a cyclic voltammogram (CV) to be obtained on each peak. Comparison of the CV for each peak with that of a standard provided additional confiiation of each peaks identity. All of the data shown in this report, however, were performed amperometrically. Quantitation. The entire sample was not injected onto the column because the microinjection pipet never removed
1 2" 3 4" 5" 6 7 8" 9" 10 11
mean f SD'
199 189 169 146 134 130 129 127 78.0 64.7 49.9 129 f 48.4
4.80 34.3 12.5
NDb ND 57.1 36.1 5.02
ND ND
4.45 14.0 f 19.5
E-Dominant Cells 12 13 14 15a 16 17 18 19 20 21 22 mean f SD
26.2 4.24 13.4 42.9 21.9 12.4 12.0 84.1 10.7 7.87 20.5 23.3 f 22.8
196 169 149 137 125 119 113 102 90.2 89.6 73.9 124 f 36.7
a Cell was obtained from a culture plate which had been enriched in NE-dominant cells. *ND is none detected. 'SD is standard
deviation.
all of the liquid from the microvial, and all the liquid in the pipet was never injected onto the column. DHBA permits us to account for the transfer efficiency, as well as the slight difference in the electrochemical response of NE and E. Calibration runs were performed by microdispensing 100 fmol each of DHBA, NE, and E into a microvial and then microinjecting them into the column as previously described. This procedure is repeated using 10 and 500 fmol of both NE and E with 100 fmol of DHBA. A calibration curve is constructed by plotting the ratios of the peak areas of NE and E to the peak area of DHBA for that run vs the number of femtomoles of NE and E originally added to the microvial. The number of moles of catecholamine present in a single cell was determined by ratioing the peak area of each catecholamine to the peak area of DHBA in that run and interpolating the number of moles present from the calibration curves. Data Interpretation. Table I shows the amount of NE and E seen in 22 cells which originated from five independent cell preparations. No correlation between catecholamine content or ratio from cell preparation to cell preparation was seen. Two of the five cell preparations followed a protocol which allowed for the collection of enriched fractions of NE- and E-containing cells. The cells from separated fractions were cultured in individual plates. The normal ratio of NE to E content in a cell plate is 1:4, the same as in an intact gland. After enrichment, the culture plates enriched in NE had a ratio of NE to E of 1.5:l. The other fraction was enriched in E (NE:E, 15). Analyses obtained from cells which were taken from NE-enriched culture plates are labeled in Table I. Since cells are removed randomly from the culture plates, a NE- to E-storing cell ratio of 1:4 was expected when plates from the three nonenriched preparations were used. Table I shows this produced six NE- and ten E-dominant cells. A cell is classified as NE- or E-dominant on the basis of
ANALYTICAL CHEMISTRY, VOL. 64, NO. 6, MARCH 15, 1992
Table 11. Analysis of Quantitative Precision
I
I
-
a
1
I
I
I
0
50
I
100
Norepinephrine
150
200
(fmol)
Flgurr 2. Plot of norepinephrine vs epinephrine contained in individual adrenomedullery cells. ce#s which were found to have an undetectable amount of norepinephrine are given a norepinephrine value of zero. The dashed line represents a catecholamine ratio of unity.
whichever catecholamine is found to be present in a higher amount. The enrichment procedure was performed in hopes of obtaining more data on NE-dominard cells. For experiments in which only NE-enriched culture plates were used, five NE-dominant cells were seen compared to only one Edominant cell. Cultured cells are known to secrete NE and E, even without a stimulus, in order to maintain a basal concentration of these two catecholamines in the surrounding cell bathing fluid.lg A blank run was performed after each cell run to guarantee that none or an insignificant amount of the signal seen in the cell run originated from the balanced salt solution transferred over to the microvial with the cell. This was accomplished by removing approximately 1nL of balanced salt solution from the cell plate immediately after having removed the cell and using this as a blank. In all cases, only one cell was removed from a single culture plate, and the cell was usually removed within 5 min of adding fresh balanced salt solution to the culture plate. If the blank contained more than 0.30 fmol of either NE or E, the data from the previous cell run were discarded. In other words, a cell was removed from a culture plate at a time when the amount of basal secretion was low enough that the error was less than 0.3 fmol. In Table I, cell numbers 1-11 are dominant in NE, while cell numbers 12-22 are dominant in E. These cells could be called NE- and E-storing cells, respectively. It is interesting to note that four out of eleven NE-dominant cells contain no detectable amount of E compared to E-dominant cells which always contained NE. The detection limit for this technique was determined to be 46 amol for NE and 75 amol for E. The detection limit is defined as the amount injected to give a signal to RMS (root mean square) noise ratio (SIN) of 3. It is not surprising to see NE in an E-dominant cell since the former is the immediate precursor to the latter. If a NE cell does not contain PNMT, as is suggested, it is interesting that some of the NE-dominant cells contain E. Some cells, such as numbers 2, 6, 7, and 19, contain significant amounts of both catecholamines. This can best be visualized by plotting the amount of norepinephrine vs epinephrine contained in a single cell, as shown in Figure 2. This finding contradicts the traditional view that there are only two distinct cell typea. It is unclear whether these values truly relate to the in vivo situation. These data may support earlier findings that not all histochemically generated data clearly fall into two cell types: that more than two cell types may exist,' or that cells may be converted from one cell type to the other! Alternatively, these results could represent an error which would occur if more than one cell was transferred from the
-1.43 9.84 -3.47
1
NE E NE
140 3.05 11.9
138 3.35 15.2
E
NDd
ND
NE
27.6 118
27.1 120
E -
diffe
catecholamine
3
I
measd amt, fmol first second inj. injb
sample
2
0
693
%
0.36 1.69
"Obtained from injecting 5 nL of the total 10 nL of sample solution produced after lysing a single cell. bObtained after injecting the remaining half of the sample solution. Second injection relative to the first injection. ND is none detected.
cell plate to the microvial. This could occur either by inadvertently picking up more than one cell with the transfer pipet or by having one or more cells adhere to the outside of the glass transfer pipet and dislodge themselves into the microvial during the cell transfer. These sources of error are unlikely, however, because we anticipated these possibilities and took great care to avoid the problems. Accuracy and Precision. The accuracy of the values given in Table I is difficult to assess because quantitative information on the amount of NE and E in a single cell has never been previously determined. An explanation for the varying amounts of total catecholamine per cell may simply be related to cell size. The cells used had diameters which varied from approximately 13 to 16 pm. This corresponds to some cells having about twice the volume of others, with volumes ranging from 1 to 2 pL, respectively. It was determined using conventional HPLC that if 5 mL of 0.1 N HCIOl was added to a whole culture plate, identical to those used in this study, that the resulting concentration of catecholamine in solution was approximately 18pM. Since there are approximately 600 OOO cells/plate, this corresponds to 159 f 13 fmol of catecholamine/cell. This value is in good agreement with Table I, which indicates an average cell value of 145 f 48.7 fmol. Also, Phillips combined literature values to speculate that a single cell should contain approximately 170 fmol of ~atecholamine.~~ The precision of the analytical procedure was investigated in the following manner. An adrenomedullary cell was isolated and prepared for injection, but for this precision study, the liquid in the microvial was split into two aliquots. Each aliquot was injected and analyzed independently. The amount of catecholamine per cell was determined for both runs as previously described. Table I1 shows the results obtained for performing this on three separate cells and gives an estimate of the overall quantitation error of this procedure. The largest relative error observed was 9.84% and occurred for a cell which had only 3 fmol of E present. Each run in Table I was the result of injecting 10 nL onto the column, whereas the injection volume for the runs in Table I1 was only 5 nL. Since a larger injection volume will provide an increase in SIN, any quantitation error occurring due to instrumental factors for the runs in Table I should be less than those reported in Table 11. We feel that this provides support that the values and uncertainties reported in Table I are biological in origin and not artifacta of the methodology. ACKNOWLEDGMENT This research has been supported by the National Institutea of Health (Grants GM39515 and NS15841). Registry No. NE, 51-41-2; E, 51-43-4. REFERENCES (1) Eranko, 0. Nature 1955, 175, 88. (2) Hiilarp, N.-A.; Hokfelt, B. Acta physld. Sand. 1853, 30, 55-68.
Anal. Chem. IQQ2,64, 694-697 Coupland, R. E.; Pyper, A. S.; Hopwood, D. Nature 1984, 207, 124-242. Benedeczky, A.; Puppi, A.; Tlgyl, A.; Lissak, K. Nature 1964, 204, 591-592. Benedeczky, I.; Puppl, A.; Tigyi, A.; Lissak, K. Nature 1988, 209, 592-594. Hagen, P.; Bannet, R. J. Adrenergic Mechanisms; Little, Brown and Co.: Boston, 1960; pp 83-99. MlcheCBechet. M.; Cotte, 0.; Haon, A.-M. J. Mlcroscop. 1963, 2 , 449-460. Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1988, 67, 436-441. Oates, M. D.; Cooper. B. R.; Jorgenson, J. W. Anal. Chem. 1990, 62, 1573-1577. Kennedy. R. T.: Oates, M. D.; COODW,B. R.; Nickerson, B.; Jorgenson, J. w. sc/ence 1089,.?46,57-63: Okflrowicz. T. M.;Ewing, A. G. Anal. Chem. 1890, 62, 1872-1876. McCaman, R. M.; Welnrlch, D.; Borys, H. J. Neurochem. 1973, 21, 473-476. McAdoo, D. J. Blochemstry of Characterized Neurons; Pergamon: Oxford, U.K., 1978; pp 19-45.
(14) Lent, C. M.; Meuiler, R. L.; Haycock, D. A. J. Neurochem. 1089, 47, 481-490. (15) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1980, 67, 1128-1135. (16) Knecht, L. A.; Guthrie, E. J.; Jorgenson, J. W. Anal. Chem. 1984, 56, 479-482. (17) St. 186-191. Claire, R. L.; Jorgenson, J. W. J. Chromatogr. Scl. 1985, 23, (18) Wilson, S. P.; Viveros, 0. H. €xp. Cell Res. 1861, 733. 159-169. (19) Leszczyszyn, D. J.: Jankowskl, J. A.; Vlveros, 0. H.; Dlliberto, E. J.; Near, J. A.; Wightman, R. M. J. Neufochem. 1001, 56, 1855-1863. (20) Moro, M. A.; Lopez, M. G.; Ganda, L.; Michebna, P.; Garcia, A. G. Anal. Blochem. 1890, 785, 243-248. (21) Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1988, 60, 1521-1524. (22) White. J. G.; St. Claire, R. L.; Jorgenson, J. W. Anal. Chem. 1086, 58, 293-298. (23) Phillips, J. H. Neurosclence 1082, 7, 1595-1609.
RECEIVED for review June 24,1991. Revised manuscript received November 5, 1991. Accepted November 11, 1991.
CORRESPONDENCE Heparin-Responsive Electrochemical Sensor: A Preliminary Study Sir: Polymer membrane type ion-selective electrodes (ISEs) are now used routinely within biomedical instruments to measure clinically important ions (e.g., Na+, Li+, K+, H+,C1-, etc.) in undiluted whole Efforts to develop similar sensors (including immune-based biosensors) suitable for the direct detection of larger biomolecules, such as drugs and specific proteins, have been less successful3 owing to the difficulty in identifving appropriate membrane chemistries that will yield a significant and specific electrochemical response toward the desired analytes. Toward this goal, we describe herein a polymer membrane-based electrode that exhibits significant potentiometric response to heparin in the clinically relevant concentration range. Heparin is an anionic rodlike polysaccharide (copolymer of uronic/iduronic acids alternating with sulfated glucosamine residues; Figure 1)with an average molecular weight of approximately 15000 Daa4 It is the anticoagulant drug used universally during surgical procedures and extracorporeal therapies?" The anticoagulantactivity of thisdrug is believed to be due to its ionic interaction with antithrombin I11 (ATIII, a serine protease inhibitor), causing the formation of heparin-ATIII complexes which potentiate the inhibition of ATIII on enzymes involved in the coagulation cascade.' Because of the potential bleeding risks associated with its use? accurate monitoring of heparin is critical. At present, there is no method suitable for direct and rapid determination of physiological heparin levels. Currently available heparin assays such as the Activated Clotting Time (ACT) are all based on the measurements of blood clotting time. Despite wide clinical use for many years, these clotting time tests are not specific for heparin and lack speed and accuracy, as well as a defiied bio~hemistry.~ We were interested therefore in applying conventionalISE polymer membrane technology to devise a membrane electrode that is capable of detecting directly the concentrations of heparin in blood or plasma samples. Initial studies focused on the use of a quaternary ammonium salt, tridodecylmethylammonium chloride (TDMAC),as the "heparin carrier" incorporated within the polymer membrane phase. TDMAC has structural similarity to polybrene, a highly potent heparin antagonist,l0and is known to possess strong ion-association with heparin." Indeed, TDMAC-heparin complexes have
been incorporated into polymeric materials previously for preparing biocompatible devices,12J3from which heparin is released slowly via an ion-exchange process. In addition, PVC membranes doped with TDMAC or other quaternary ammonium salts have been suggested for fabricating conventional ISEs for small anions including ~ h l o r i d e . ~Nevertheless, ~J~ none of these quaternary ammonium salt-based polymer membrane electrodes have ever been examined in detail with respect to heparin response. Surprisingly, we now find that membranes doped with TDMAC do in fact exhibit significant potentiometric response toward heparin in the presence of normal physiological levels of NaCl. The purpose of this correspondence is to report our initial results aimed at characterizing this response in terms of its dependence on membrane composition and its potential analytical utility for detecting heparin in plasma or whole blood samples. EXPERIMENTAL SECTION Reagents. Poly(viny1 chloride) (PVC) was obtained from Polyscience, Inc. (Warrington, PA). Dioctylsebacate (DOS) and tridodecylmethylammonium chloride (TDMAC) were purchased from Fluka Chemika-Biochemika (Ronkonkoma, NY). Sodium heparin was from Hepar Industries, Inc. (Franklin, OH) and poly(viny1sulfate), chondroitin sulfate A, chondroitin sulfate B, D-glUCOSamine, D-glUCOSamine 2,3-disulfate, D-glUCOSamine 2sulfate, D-gh"ine 3-sulfate,and D-glucosamhe&sulfate were products of Sigma Chemical Co. (St. Louis, MO). Citrated fresh frozen human plasma was obtained from American Red Cross (Southfield, MI). All other reagents were of analytical grade. Solutions were prepared with double-distilled deionized water. Preparation of Polymer Membranes and Apparatus. Polymer membranes were cast using the conventional method for ISE membrane preparation.I6 Small disks (-5 mm i.d.) of polymer membranes were cut and incorporated into Phillips electrode bodies (IS-561, Glasblaserei Moller, Zurich). A 15 mmol/L NaCl solution was used as the internal filling solution throughout the study. Electrodes were soaked in the same NaCl solution overnight prior to their use. The potentiometric response of the membrane electrode was measured relative to an external double junction Ag/AgCl reference electrode at room temperature (-22 "C), with the sample solution being stirred constantly (see Figure 2).
RESULTS AND DISCUSSION Although it was known that tridodecylmethylammonium
0003-2700/92/0364-0694$03.00/00 1992 American Chemical Society
