Solid-Phase Colorimetric Determination of Potassium

CLIN. CHEM. 28/9, 1857-1861 (1982)
Solid-Phase Colorimetric Determination of Potassium
Steven C. Chariton, Roger L. Fleming, and Adam Zipp
A nonpolar organic film (plasticized polyvinyl chloride)
containing the ionophore valinomycin was incubated with
an aqueous solution containing potassium ion and a detectable anion (erythrosin B). The amount of erythrosin
B retained by the film after washing could be measured by
absorbance or reflectance, and was directly related to the
potassium concentration.
This dye-binding method is
quantitative for potassium and is suitable for both aqueous
and serum-based solutions. There was no interference by
sodium in the range found in serum. Several polyvinyl
chloride plasticizers and anionic dyes and some other ionophores were found to be useful. The anion binding is
thought to be restricted to the surface.
Additional Keyphrases: polyvinyl chloride membrane
valinomycin
lonophores - etythrosin
film reflectance
K-sensitive film
Langmuir binding isotherm
-
-
.
Monitoring renal function or secondary manifestations
of
other disease states frequently necessitates determination
of
serum potassium concentration.
A normal range of only 3.6
to 5.5 mmol/L in healthy individuals (1) is indicative of how
precisely the concentration
of this ion is regulated. A plasma
potassium concentration
<1.5 mmol/L is often fatal, owing
to respiratory
depression and cardiac arrhythmia.
Conversely,
characteristic
electrocardiographic
changes and even cardiac
arrest occur when potassium exceeds 10 mmolfL (2). Because
of the limited normal range and the effects of relatively small
excursions
from it, assays for potassium
must be highly
accurate.
In the clinical laboratory, potassium is usually determined
by flame
photometry
or ion-specific
potentiometry.
In ion-
specific potentiometry
the electrical potential developed
across a special membrane separating the sample and a reference solution is measured. The special membrane, which
responds only to potassium, contains a catalytic amount of an
ionophore, a molecule that specifically binds the ion under
examination
and transports
it across a hydrophobic
membrane. Thus a conductive path is produced only for this one
ion and the membrane effectively ignores concentration
gradients of all other ions. Valinomycin is commonly used as the
ionophore
(3, 4). The conformation
of this macrocyclic
compound
is such that it forms a three-dimensional
cavity,
lined with oxygen and nitrogen atoms and giving a polarizable
interior suitable for the ligand-solvation
requirements
of
potassium ions (5, 6). Valinomycin shows a very high specificity for potassium binding over sodium and divalent cations.
The methyl and methylene groups forming the hydrophobic
exterior of the molecule allow it to be freely soluble in solvents
of low polarity. When potassium from aqueous solution binds
to valinomycin at the membrane interface, the complex diffuses across the membrane and the trapped potassium is released on the other side (6), thus establishing a conductive
path across the membrane.
A similar mechanism is thought to be involved in the partitioning of metal cations between organic and aqueous phases
Division, Miles Laboratories,
Inc., Elkhart,
Received Apr. 29, 1982; accepted June 23, 1982.
Ames
IN 46515.
in the presence of an ionophore (here acting as a phasetransfer catalyst) (7). In the presence of such a compound,
typically a cyclic polyether, a metal ion can be transported
into
the organic phase. By suitable choice of anions, some of the
anion may accompany the metal-ionophore
complex into the
organic liquid (8), presumably so that buildup of the charge
there from the metal ion would be neutralized. The anions
expected to partition easily would be “soft”-.polarizable
molecules able to delocalize the negative charge. Theoretical
expressions have been developed for the concentration
of
anion expected in the organic phase as a function of cation and
the ionophore concentrations
(9).
The basis for the solid-phase colorimetric potassium test
described
here is our surprising observation
that partitioning
of an organic anion into a valinomycin-containing
solid phase
can proceed readily and quantitatively
as a function of p0tassium concentration.
Specifically, the solid nonpolar phase
is a plasticized polyvinyl chloride (PVC) membrane containing
the potassium-sensitive
ionophore, valinomycin. A drop of the
aqueous phase, which contains potassium ion and a detectable
anion, erythrosin B, is placed on the organic phase and allowed to incubate for several minutes. During this time, both
potassium and erythrosin B partition between the aqueous
and organic phases. After the aqueous drop is removed, the
amount of erythrosin B retained by the solid phase is measured spectrophotometrically.
Others have shown (10) that
when radioactive chloride ion (36Clj was incubated with a
similar membrane, uptake of this added anion was minimal;
instead,
membrane-bound
water deprotonated
to supply a
balancing negative charge to the film. In contrast, we will show
that by suitable choice of anion, this uptake can be made rapid
and quantitative
and is a precise measure of potassium concentration
of the solution in contact with the membrane.
Materials and Methods
Materials. Valinomycin,
18-crown-6
cyclooctadecane),
polyvinyl chloride
mass),
2,6-dichloroindophenol,
(1,4,7,10,13,16-hexa-
(very high molecular
dimethylphthalate,
and
di-
octylphthalate
were from Aldrich Chemical Co., Milwaukee,
WI 53233. Tetrahydrofuran,
1,4-dioxane (“Photrex”; reagent
grade),
acetone (HPLC reagent grade), potassium
chloride,
and phloxine B (reagent grade), were from J. T. Baker
Chemical Co., Phillipsburg, NJ 08865. Erythrosin B, orange
G, phenolphthalein,
hematein,
and dipentyl-, dibutyl-, diethyl-, and dinonylphthalates
were from Eastman Kodak Co.,
Rochester, NY 14650. From MCB Manufacturing
Chemists,
Inc., Cincinnati, OH 45212 we obtained the dyes eosin Y,
phenol red, and Kryptofix K-221 and K-222. Orange IV and
Alizarin Red S came from Allied Chemical Corp., Morristown,
NJ 07960, and 3-(N-morpholino)propanesulfonic
acid, sodium
salt, from Calbiochem-Behring
Corp., La Jolla, CA 92037.
Bis-2-ethylhexylsebacate
and polyvinyl fluoride were from
Scientific Polymer Products, Ontario, NY 14519, and tris2-ethylhexyl
phosphate
was from ChemServices,
West
Chester, PA 19380. The polyester film base (Scotchpar) was
obtained from the Film and Allied Products Division of 3M
Co., St. Paul, MN 55144.
Reflectance
measurements
were taken with a Seralyzer
CLINICALCHEMISTRY,Vol. 28, No. 9, 1982 1857
We used the simplified form of the Kubelka-Munk
reflectance
spectrophotometer
(Ames Division, Miles Laboratories,
Inc., Elkhart,
IN 46515).
We used a flame photometer
(Model IL 443; Instrumenta-
tion Laboratory,
determinations
tassium
and
Inc., Lexington,
equation
K/S
MA 02173) for comparison
of potassium,
calibrating
at 5 mmol
140 mmol of sodium per liter. Nonpolar
of po-
such as erythrosin
All samples were diluted ninefold before assay, and all
cation concentrations
shown correspond to the concentration
centration
was 10 mmol/L. We incubated
a drop of this solution (140 1iL) for 4 mm on a piece of prepared film (0.5 X 1 cm),
then washed the surface for 5 s with a jet of water. After
blotting the film dry, its absorbance or reflectance
could be
before
mounting
at 526 nm of erythrosin
3.0
dilution.
Results
device if required.
B in water
shifted to 552 nm when retained in the film. We usually
measured absorbance at 550 nm. Reflectance measurements
(R) were made with a Seralyzer reflectance spectrophotometer
also at 550 nm.
maximum
R)2/2R
transform
the measured
absorbance
of a test solution to potassium concentration.
The slope of the calibration
lines could also be used to express the film reactivity-in
terms of absorbance units or K/S
per mole of potassium
per liter.
dilution, the buffer
concentration
was 125 mmol of the sodium salt of 3-(Nmorpholino)propanesulfonic
acid per liter and the dye con-
with use of a suitable
-
(K/S) that is proK/S is the ratio
of the absorption (K) to the scattering (S) coefficient.
To convert reflectance measurements
into potassium concentration, we used a two-point calibration, measuring two
samples with known potassium concentrations bracketing the
normal range. The reflectances
were measured and converted
to K/S values, which were used to generate a calibration
line
of K/S vs potassium concentration.
Ordinarily, these computations and other functions are done automatically
by the
Seralyzer instrument.
Absorbance
measurements
were made by suitably mounting
an analytical element in the light path of a spectrophotometer.
As before, we used a two-point calibration
procedure to
the polyester support with double-sided adhesive tape.
Procedure. A general assay procedure was to dilute the
sample such that the potassium concentration was in the range
0.2 to 1.1 mmol/L (equivalent to a ninefold dilution of serum),
The absorption
(1
reflectance
data to a function
to the chromophore
concentration.
portional
of air for 5 mm. As prepared,
this material
was suitable for
absorbance
measurements
because both support and film were
transparent.
For reflectance
measurements
we laminated
a
uniformly
scattering
white material
to the uncoated
side of
measured
=
to transform
solid
phases were prepared by dissolving PVC, dipentylphthalate
(DPP) plasticizer, and ionophore (typical concentrations
of
each were 40, 100, and 1.5 gIL, respectively) in a suitable organic solvent (tetrahydrofuran
or a 1,4-dioxane/acetone
solution, 810 g of dioxane per liter of acetone), and spreading a
uniform layer of this solution onto an inert transparent
polyester support. To dry this film, we exposed it to a stream
with a buffered
solution
of a counterion
B. Unless mentioned
otherwise,
after
(11,12)
Response to aqueous potassium. Test solutions having
nominal potassium concentrations
ranging from 1.8 to 10
mmol/L
were prepared
by adding potassium
chloride
to a
buffer-dye
solution. After reaction, we measured the reflectance of the film at 550 nm. Potassium concentrations
were
A
2.5
2.0
Aqueous
E
C
0
in
U,
B
11
9
1.5
0
-S
a,
U
C
a,
1.0
7
.0
0
.0
5
4
0.5
3
0
I
2
I
4
Potassium
Fig. 1. Response of analytical
6
Concentration
8
10
.2
(mM)
.3
(Potassium,
elements to a rangeof potassium concentration
.4
.5
.6
mM)’
(mmol/L) as meastred by reflectance (A) or absorbance
(B)
Reflectance was transformed to K/S; absorbance was plotted In dot1ie-reciprocal
of potassium solutions at 3.3 and 6.2 mmol/L
1858
CLINICAL CHEMISTRY, Vol. 28, No. 9, 1982
form. Each value was determined after calibrationwith duplicate measwements
compared with those measured by the flame photometer.
Figure 1A shows that the K/S response to potassium is linear.
Conversion of K/S units to those of potassium concentration
gave correlation
statistics
vs the flame
photometer
of slope
1.058, intercept -0.222, correlation coefficient 0.9988, and
standard error 0.128.
In another experiment
we looked at the absorbance response to aqueous potassium solutions, measuring the film
absorbance
after treatment
with potassium/erythrosin
B
solutions. The dependence of the absorbance on potassium
was best described by a hyperbola
K=S-A/(1-I.A)
where K is the potassium
concentration,
A is the measured
absorbance, and S and I are constants of the hyperbola. The
double-reciprocal
plot of Figure lB is therefore linear, and
constants S and I are determined by two-point calibration.
Against the comparison method, the strip gave a slope of 0.987,
intercept
of 0.107, correlation
coefficient
of 0.9997, and
standard error 0.11.
Response
to serum solutions.
To a serum pool with low
potassium concentration we added potassium to give a range
of concentrations up to 9 mmol/L, then applied samples to the
film and measured the film reflectance
at 550 nm. As shown
in Figure 1A, the transformed
reflectance was linearly dependent on the analyte concentration as determined by flame
photometry.
The positive intercept at zero potassium concentration
corresponds
to a small degree of nonspecific
binding. After conversion to potassium concentration
units,
the correlation statistics were: slope 1.058, intercept -0.222,
correlation coefficient 0.9988, and standard error 0.13.
Effect of sodium. A viable potassium test must be free of
interference
by sodium at the 140 mmol/L concentration
found in plasma, 30-fold that of potassium. We incubated the
potassium test strips with solutions corresponding to sample
sodium concentrations ranging from 0 to 900 mmolfL and then
measured the absorbance. At zero potassium concentration,
there was a blank reaction corresponding to 0.2 mmol/L potassium. Sodium-containing
samples gave a small increase in
apparent
potassium
concentration,
but the increase was
random and not dependent on the metal cation concentration.
Effect of lithium. When strips were incubated with lithium
solutions equivalent
to sample concentrations
up to 10
mmol/L, the apparent potassium ion concentration
was less
than 0.1 mmol/L more than the blank.
Reaction conditions. The amount of color retained by the
film is a function of the time during which test liquid and solid
are in contact. Solutions containing two concentrations
of
potassium were incubated on potassium-sensitive
film for 0.5
to 8 mm. As shown in Figure 2A, the amount of dye retained
increases with time. We chose to use a reaction time of 4 mm,
to give a balance between speed and the imprecision that
arises from small discrepancies in test timing. Increasing the
washing time decreased the amount of color somewhat (Figure
2B). A wash time of 5 s was found to be adequate. Acceptable
reactivity was found with a dye concentration
of 10 mmol/L,
although the reactivity did increase with higher dye concentration.
Film composition.
We examined several different film
compositions in order to optimize the formulation. With fixed
amounts of PVC and DPP and varied valinomycin concentration, reactivity was proportional to the valinomycin:plasticizer ratio (Figure 3A). We used a ratio of 14.4 mg/g to give
a balance between reactivity and economy.
Plotting reactivity against DPP/PVC ratio (and adjusting
reactivities
for changes in valinomycin:plasticizer
ratios)
showed that reactivity increased with an increasing DPPIPVC
ratio, but became saturated at ratios greater than 4 (Figure
3B). At these high ratios, however, the film deteriorates, becoming sticky and quite fragile; again, a compromise was in
order and we chose a DPP/PVC ratio of 2.6.
Film thickness. Coatings of fixed composition but variable
thicknesses were prepared by adjusting the solids content of
the spreading solution or the thickness of the wet layer.
Evaluations
of film performance
at two concentrations
of
potassium (Figure 4) show clearly that the reactivity is almost
independent
of film thickness in the range 24 to 260 m.
Alternative
plasticizers.
Compounds
generally used to
plasticize PVC are the high-boiling-point
members of the
phosphate, sebacate, and phthalate families (13). Films prepared by substituting
tris-2-ethylhexylphosphate
or bis-2ethylhexylsebacate
for DPP were both responsive to potassium (Figure 5). Homologs to DPP, substituted at fixed weight
percentage, were also reactive, in the order: dimethyl > diethyl
A
3.0
3.5
9.9
mM
E
cE
o
0
0
U)
0
U)
2.0
2.5
(I)
-
CI)
1.5
9.9
2
4
6
Incubation Time (mm)
8
10
20
mM
30
40
Wash Time (s)
Fig. 2. (A) Effect of varying incubation time of the potassium-dye reagent and a 5-s wash, and (B) effect of varying wash time after
incubation for 4 mm, at two concentrations of potassium
CLINICALCHEMISTRY,Vol. 28, No. 9, 1982 1859
20
A
10
16
8
U
12
>1
B
12
a,
5
a,
U
8
a,
I-
0
4
U
4
2
5
10
15
1
lonophore/PlasticizerRatio (X103)
2
3
4
5
Ptasticizer/PolymerRatio
Fig. 3. Effect on reactivity of varying (A) valinomycin at a fixed PVC/DPP ratio, or (B) the DPP/PVC ratio
A, Units of reactivity are absorbance units per mmoi of potassium per liter X iO. Reactivity adjusted for changes in vaiinomycin/DPP shown in B
.90
10
#{163}
,
.80k
,
8
.70
,
TrI-(2.Ethylhexyl)Phosphate
A
1.8mM
E
.60
E
U)
3
6
a,
a,
a,
0
.50
a.5
a,
3
a,
a,
.30
2
9.9 mM
I
p
I
60
I
I
100
I
‘!
I
140
I
I
I
180
I
220
I
-“
I
260
2
Layer Thickness (pM)
Fig. 4. Potassium-dependent dye binding as a function of the
Besides
erythrosin
B, a fluorescein
rivative, we screened
for use in the assay.
quantitative
binding
of strips reacted with
tassium concentration
correlation coefficient
prisingly, the parent
de-
several other substituted
fluoresceins
Phloxine B and eosin Y both showed
in response to potassium. Correlation
phloxine B against the gravimetric p0gave: slope 0.968, intercept 0.360,
0.992, and standard error 0.49. Surcompound, fluorescein, was inactive.
2,6-Dichloroindophenol
and Orange IV (molecules quite different from each other and from the fluorescein derivatives)
were also useful in the assay method; in a correlation study the
former compound gave a slope of 1.101, intercept -1.02, cor1860
CLINICAL CHEMISTRY,
6
Potassium
8
10
(mM)
x-axis, wei,ed potassium (mmol) in solution; yaxis. potassIum calculated from
reflectance of dye bound
> dibutyl > dipentyl>
dioctyl > dinonyl, i.e., in order of
molecular mass. This is also the order of decreasing mole
fraction, which may be the basis of the effect. We also found
(data not shown) that polyvinyl fluoride could be substituted
for PVC as the structural polymer.
dyes.
4
Gravimetric
Fig. 5. Results obtained with alternative plasticizers
thickness (pm) of the nonpolar phase
Alternative
4
I..
.40
Vol. 28, No. 9, 1982
relation coefficient 0.986, and standard error 0.78. Phenol red,
Alizarin RedS, Orange G, phenolphthalein,
and hematein were
inactive.
Alternative
ionophores.
Separate DPP-plasticized
coatings
were prepared in which we substituted
other ionophores for
valinomycin: a monocyclic crown ether, 18-crown-6, and the
bicycic cryptates, Kryptofix K-221 and K-222. Each film was
treated with an erythrosin B solution containing a range of
sodium or potassium concentrations.
As Table 1 illustrates,
incorporation
of K-222 gives a film that is responsive to potassium but not to sodium, whereas incorporation
of K-221
produces a sodium-sensitive
film that is insensitive to potassium. This selectivity is as expected (14). Films containing
l8-crown-6 showed no dye binding in response to sodium or
potassium.
Table 1. Reflectances of Films Containing the
lonophores K-222 or K-221 after Treatment with
Potassium or Sodium and Dye-Reagent Solutions
Cation
R.fl.ctanc.,
coecn,
K-222
mmol/L
Potassium
Sodium
%R
K-221
Potassium
so small.
Sodium
0
56.1
54.6
54.7
56.4
0.9
50.6
53.4
56.3
2.7
46.1
53.3
54.2
9.0
28.1
9.97
54.3
54.7
56.7
52.5
47.8
51.5
57.9
30.2
ND
ND
54.4
28.0
ND
ND
53.4
18.6
27.0
90.0
270.0
Independence
of the reactivity
to the film thickness
suggests that the interaction
is restricted to the surface.
Surface reactions (and heterogeneous
catalysis) have frequently been expressed in terms of the Langmuir binding
isotherm, where the amount of a molecule L bound to a surface
(LB) is hyperbolically dependent on the free concentration
of
that molecule (LF), e.g.:
LB = B X LF/(C + Lp)
where B is the number of binding sites and C is a constant.
A hyperbolic
dependence of the amount bound is indeed
seen (Figure 1B), as predicted by this model. In addition, the
proportionality
of reactivity with the ionophore/plasticizer
ratio and the plasticizer/polymer
ratio is also consistent. If
potassium and its counterion
partition
mainly into the plasticizer part of the solid phase under the influence of the
ionophore, then increasing either the fraction of plasticizer
or ionophore would have the effect of increasing the number
of potential binding sites.
From the studies of solid phases containing a representative
of three plasticizer classes of various ionophores, or the different dyes that could be bound to the film, evidently there
is no specific matrix-ionophore-dye
molecule
combination
involved, and this kind of test procedure is suitable for any
analyte for which a specific ionophore is available.
The amount of dye retained by the film can be measured
by absorbance or reflectance (or fluorescence, if a suitable
is chosen).
The
measured
absorbance
has a hy-
perbolic dependence on potassium
concentration.
The K/S
transformation,
which is used merely as a convenient algorithm, does, however, provide a function linearly dependent
on potassium
Reactivity
is decreased
somewhat
for serum,
as
compared with aqueous solutions, but the use of serum-based
calibrators obviates this problem. Other systems have been
described for analysis of potassium
(15) and sodium (16) by
partitioning
of a detectable counterion between two liquid
phases, but are less convenient because of the number of sample
manipulations
required,
including
deproteinzation.
The
method described here is considerably more rapid and convenient and has all the advantages associated with use of a
solid-phase format.
References
Discussion
counterion
contributes 125 mmol and the dye 20 mmol of sodium per liter.
Under these conditions, an increase from 120 to 170 mmol/L
produces an increase of only 3.5% in the sodium concentration
of the diluted sample. It is a measure of the specificity of the
ionophore that at zero potassium concentration
the blank is
concentration.
The specificity of the film response is excellent and sodium
ion is not expected to be an interference in serum measurements for two reasons: principally the excellent selectivity
of the ionophore valinomycin but also the pre-existing high
concentration of sodium in the dye-buffer solution. The buffer
1. Henry, R. J., Cannon, D. C., and Winkelman, J. W., Eds., Clinical
Chemistry,
Principals and Technics,
2nd ed., Harper & Row,
Hagerstown, MD, 1974, Chapter 19.
2. Barron, D. N., A Short Textbook of Clinical Chemistry, J. B.
Lippincott Co., Philadelphia, PA, 1973, pp 24-29.
3. Freiser, H., Ed., Ion Selective Electrodes in Analytical Chemistry,
1, Plenum Press, New York, NY, 1978, p 30.
4. Kolthoff, I. M., Applications of macrocyclic compounds in chemical
analyses. Anal. Chem. 51, 1-22R (1979).
5. Truter, M. R., Structure of organic complexes with alkali metal
ions. Struct. Bonding (Berlin) 16,71(1973).
6. Ovchinnikov, Y. A., Physico-chemical basis of ion transport
through biological membranes: lonophores and ion channels. Eur.
J. Biochem. 94, 321-336 (1979).
7. Dix, J. P., and Vogtle, F., Ion selective crown ether dyes. Angew.
Chem. mt. Ed. EngI. 17, 857-859 (1978).
8. Pederson, C. J., Ionic complexes of macrocyclic polyethers. Fed.
Proc. Fed. Am. Soc. Exp. Biol. 27, 1305 (1968).
9. Eisenman, G., Ciani, S., and Szabo, G., The effects of macrotetralide action antibiotics on the equilibrium extraction of alkali metal
salts into organic solvents. J. Membr. Biol. 1, 294-345 (1969).
10. Thoma, A. Z., Viviana-Nauer,
A., Morf, W. E., and Simon, W.,
Mechanism of neutral carrier mediated ion transport through ion
selective bulk membranes. Anal. Chem. 49, 1567-1572 (1977).
11. Kubelka, P., and Munk, F., Em Beitrag zur Optik der Farbenstriche. Z. Tech. Phys. 12,593-598 (1931).
12. Kortum, G., Reflectance Spectroscopy,
Springer-Verlag,
New
York, NY, 1969, p 111.
13. Roff, W. J., and Scott, J. R., Handbook of Common Polymers,
CRC Press, Cleveland, OH, 1971, p 110.
14. Lehn, J. M., and Sauvage, J. P., [21-Cryptates: Stability and selectivity of alkali and alkaline earth macrobicyclic complexes. J. Am.
Chem. Soc. 97,6700-6707
(1975).
15. Hideo, S., Nakahara, K., and Ueno, K., New convenient colonmetric determination
of potassium in blood serum. Talanta 24,
763-765 (1977).
16. Takagi, M., Nakamura, H., Sanui, Y., and Ueno, K., Spectrophotometric
determination
of sodium by ion-pair extraction with
crown ether complexes and monoanionic dyes. Anal. Chim. Acta 126,
185-190(1981).
CLINICALCHEMISTRY,Vol. 28, No. 9, 1982
1861