Content uploaded by Xuewei Wang
Author content
All content in this area was uploaded by Xuewei Wang on Nov 05, 2023
Content may be subject to copyright.
Ultrasensitive Ionophore-Based Liquid Sensors for Colorimetric Ion
Measurements in Blood
Nasrin Ghanbari Ghalehjoughi, Renjie Wang, Savannah Kelley, and Xuewei Wang*
Cite This: Anal. Chem. 2023, 95, 12557−12564
Read Online
ACCESS Metrics & More Article Recommendations *
sı Supporting Information
ABSTRACT: The self-monitoring of electrolytes using a small
volume of capillary blood is needed for the management of many
chronic diseases. Herein, we report an ionophore-based colorimetric
sensor for electrolyte measurements in a few microliters of blood. The
sensor is a pipet microtip preloaded with a segment of oil (plasticizer)
containing a pH-sensitive chromoionophore, a cation exchanger, and
an ionophore. The analyte is extracted from the sample into the oil via
a mixing protocol controlled by a stepper motor. The oil with an
optimized ratio of sensing chemicals shows an unprecedentedly large
color response for electrolytes in a very narrow concentration range
that is clinically relevant. This ultrahigh sensitivity is based on an
exhaustive response mode with a novel mechanism for defining the lower and higher limits of detection. Compared to previous
optodes and molecular probes for ions, the proposed platform is especially suitable for at-home blood electrolyte measurements
because (1) the oil sensor is interrogated independent of the sample and therefore works for whole blood without requiring plasma
separation; (2) the sensor does not need individual calibration as the consistency between liquid sensors is high compared to solid
sensors, such as ion-selective electrodes and optodes; and (3) the sensing system consisting of a disposable oil sensor, a programmed
stepper motor, and a smartphone is portable, cost-eective, and user-friendly. The accuracy and precision of Ca2+ sensors are
validated in 51 blood samples with varying concentrations of total plasma Ca2+. Oil sensors with an ultrasensitive response can also
be obtained for other ions, such as K+.
■INTRODUCTION
Blood tests are routinely performed for medical diagnostics as
blood carries valuable information on health status. Although
blood sampling is more invasive than other bodily fluids, such
as urine, sweat, saliva, and tears, blood composition is well-
regulated and has confirmed correlations with many diseases.
Blood chemistry analysis has its own challenges due to the
properties of blood, including the dark color, high turbidity,
and high protein and cell contents. In central laboratories,
blood samples usually undergo pretreatments, such as
coagulation, centrifugation, dilution, deproteination, and cell
lysis, before instrumental analysis is performed. Most blood
constituents, such as ions, gases, metabolites, lipids, drugs, and
proteins, can now be readily analyzed. As a complement to
laboratory diagnosis, point-of-care testing (PoCT) has rapidly
grown over the past few decades because it reduces turnaround
times and allows for medical interventions without delay.
Blood chemistry analysis in PoCT devices primarily relies on
sensor technologies that directly read blood composition
without multistep assay procedures. A remarkable example is
benchtop and handheld blood gas/electrolyte/metabolite
analyzers used in hospitals. Most blood chemistry analyzers
use electrochemical sensors such as ion-selective electrodes
and chemically modified electrodes, as these sensors are
intrinsically unsusceptible to the color and turbidity of
blood.
1,2
Optical sensors based on fluorescence or absorbance
allow for direct tests of hemoglobin oxygen saturation in blood
but usually need plasma separation, significant dilution, and/or
inner filter eect correction for other analytes due to the
optical interference of hemoglobin.
3−5
Although optical
detection in the near-infrared range reduces hemoglobin
interference, its application in PoCT for whole blood remains
limited.
6
At-home blood testing is an emerging trend in PoCT as it
allows patients to monitor their health conditions in a timely
manner.
7
Blood tests at home have unique requirements
compared to those performed by healthcare professionals.
First, the blood volume must be low (ideally <10 μL) as
patients can only collect their capillary blood. Second, the
operation needs to be as simple as possible so that patients
with dierent backgrounds can use these devices with minimal
to no training. Third, at-home devices need to be more
Received: July 4, 2023
Accepted: August 1, 2023
Published: August 11, 2023
Articlepubs.acs.org/ac
© 2023 American Chemical Society 12557
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
Downloaded via Xuewei Wang on August 23, 2023 at 01:31:42 (UTC).
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
aordable than the PoCT devices used in clinics and hospitals.
The blood glucose meter, as one of the most successful at-
home sensors, has revolutionized the management of diabetes.
Based on similar electrochemical-sensing principles on
enzyme-modified electrodes, other analytes such as lactate,
uric acid, urea, creatinine, cholesterol, and ketones can also be
detected at home using fingerstick blood. Lateral flow
immunoassay strips are also commercially available for the
qualitative or quantitative analysis of blood components
including antibodies (e.g., autotest VIH for the HIV antibody)
and proteins (e.g., A1CNow+ for hemoglobin H1Ac).
However, many other analytes urgently need to be measurable
at home. Hypoparathyroidism patients suer from a para-
thyroid hormone deficiency and have abnormally low blood
Ca2+ concentrations. To reduce the risk of various acute and
chronic complications, their Ca2+ levels need to be adjusted by
the administration of appropriate dosages of supplemental
calcium, guided by frequent blood Ca2+ assessments.
8
Patients
with chronic kidney disease, heart failure, diabetes mellitus,
and atherosclerotic cardiovascular disease are at an increased
risk of hypokalemia and hyperkalemia.
7,9,10
Because many life-
threatening dyskalemia incidences lack obvious symptoms,
timely and frequent blood K+monitoring is crucial to inform
medical practices and reduce hospitalization and mortality
rates. In addition, adequate monitoring of other ions such as
Na+, Li+, and Mg2+ will improve the management of diseases
such as diabetes insipidus, bipolar disorder, and depres-
sion.
11,12
Electrolyte measurements in whole blood have been
routinely performed in hospitals via ion-selective electrodes
in both benchtop and handheld analyzers. However, all
commercial blood electrolyte analyzers require on-site
calibration via automated fluidic systems, rendering the devices
complicated and costly. More importantly, even the handheld
analyzers using miniaturized electrodes require ∼100 μL of
blood, much larger than the ∼10 μL volume of a blood drop
collected from a fingerstick. As a result, these blood electrolyte
analyzers are not ideal for home use. Ionophore-based ion-
selective optodes are the optical counterpart of ion-selective
electrodes.
13−15
Both methods use a water-immiscible matrix
containing hydrophobic sensing chemicals to detect ions in an
aqueous sample. In contrast to the success of ion-selective
electrodes in blood analyzers, ion-selective optodes have rarely
been explored for whole blood analysis because of optical
interference. The classical formats of ion-selective optodes
include thin films and micro/nanoparticles, neither of which
are suitable for blood analysis since detecting the absorbance
or fluorescence of optodes is prohibited when optodes are
covered or surrounded by highly absorbing and scattering
blood. Ion-selective optodes based on upconverting nano-
particles reduce the background absorbance and autofluor-
escence in blood, but the whole blood test was only
demonstrated for 10-fold diluted blood with a standard
addition protocol.
16
The Xie group designed nanooptode-
loaded hydrogels to detect blood electrolytes using the
diusion distance of electrolytes.
17,18
However, the accurate
identification of the diusion boundary is not easy, and the
storage stability of hydrogel is a concern in real-world
applications.
Ion-selective optodes usually use semisolid matrices, such as
plasticized, PVC-containing sensing chemicals, to extract
analyte ions from aqueous samples. Ions can also be extracted
into a water-immiscible liquid that dissolves sensing chemicals,
which is a liquid−liquid extraction process. In contrast to
conventional 2D optode films, a segment of organic liquid can
serve as a 3D optode that can be optically interrogated from
one side, with no light passing through the sample. The Bakker
group used bulk solutions of dichloroethane as the optodes for
the titrimetric detection of electrolytes in bulk aqueous
samples including serum.
19
However, this solvent seems
incompatible with blood (concerns with hemolysis), and the
titration protocol is hard to apply outside laboratories. Parallel
flow microfluidics was examined for liquid−liquid ion
extraction and optical ion detection, but biological samples
were not tested, probably because the liquid−liquid interface is
unstable with these samples.
20,21
We recently functionalized
the oil stream in droplet microfluidics with hydrophobic
sensing chemicals to create oil segment-based optodes.
22,23
Since the oil segments and aqueous droplets are spatially and
temporally separated, optode interrogation at an angle
perpendicular to the microfluidic channel does not suer
from optical interference from the aqueous sample. As a result,
electrolyte measurements in nanoliters of whole blood have
been realized. However, a microfluidic system consisting of
pumps, chips, microscopes, and laser-induced fluorescence
detection setups is unsuitable for home use.
In this paper, we report a new configuration of organic,
liquid-based optodes for whole blood electrolyte measure-
ments in resource-limited settings, such as patients’ homes. We
formulated liquid sensors that exhibit sharp color responses to
the analyte over its clinically relevant concentration and
designed stepper-motor-based mixing protocols to enhance the
mass transfer between the optode and the sample.
■EXPERIMENTAL SECTION
Reagents and Materials. Calcium ionophore II (CaII,
N,N,N′,N′-tetra[cyclohexyl]diglycolic acid diamide), potassi-
um ionophore I (valinomycin), sodium tetrakis[3,5-bis-
(trifluoromethyl)phenyl]borate (NaTFPB), chromoionophore
III (ChIII, 9-(diethylamino)-5-[(2-octyldecyl)imino]benzo[a]-
phenoxazine), and dioctyl sebacate (DOS), all of Selectophore
grade, as well as all salts and buers were purchased from
MilliporeSigma.
Sensor Preparation and Operation. Sensing oils were
prepared by dissolving sensing chemicals in DOS via
sonication. A 16-channel E1-ClipTip Electronic Pipette was
used to aspirate liquids and perform mixing in a 30 μL ClipTip
pipet tip. The pipet tip was loaded with a specific volume of
sensing oil to create the disposable sensor. A specific volume of
aqueous sample was aspirated into the pipet tip containing the
sensing oil. Repeated aspiration and dispensing were
performed by the electronic pipet to allow for ion transfer
between two immiscible phases. When the aspiration and
dispensing speeds were equal, the automatic “mixing” function
of the pipet was used. When the speeds were unequal, the
button was pressed to repeat the programmed aspiration and
dispensing steps with the loop function. In the first 15 mixing
cycles, the speeds of aspiration and dispensing were set to be 7
and 3, respectively. In the subsequent 15 mixing cycles, the
aspiration and dispensing speeds were 3 and 7, respectively.
The alternation was continued until the equilibrium response
was reached. An iPhone 13 mini was used to take photographs.
The position of the iPhone was fixed by a yAyusi overhead
phone mount purchased from Amazon, and a rectangular white
LED light box was underneath the pipet tip to enhance
brightness. The color of each oil segment was analyzed by an
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12558
iPhone app, Color Mate - Convert and Analyze Colors, and the
hue of each color was extracted for quantitative analysis.
UV−Vis Absorption Spectroscopy. The absorption
spectrum of the sensing oil was collected using a Varioskan
LUX multimode microplate reader equipped with a μDrop
plate that requires a sample volume of 2 μL.
Ca2+ Measurements in Blood Specimens. Anonymized
human blood specimens were obtained from the Blood Gas
Lab at Virginia Commonwealth University (VCU) Medical
Center. They were leftover blood samples submitted to the
Blood Gas Lab for routine medical care purposes which were
not part of this research; therefore, this study is not classified as
human subjects research. The blood was anticoagulated with
lithium heparin. To test plasma Ca2+ in whole blood using the
proposed method, a pipet tip containing 7.5 μL of oil and 5.0
μL of pH 7.4 HEPES as the buer was used as the sensor, and
2.5 μL of blood was aspirated into the pipet tip. The oil color
was analyzed after the pipet-controlled mixing procedure to
calculate the Ca2+ level in the blood. Fourteen blood
specimens were obtained and spiked with extra Ca2+ to make
a total of 51 samples with a broader range of Ca2+
concentrations for sensor assessments. A portion of each
blood sample was centrifuged to collect the plasma for the
determination of the total Ca2+ concentration in plasma using a
Pointe Scientific Calcium (Arsenazo III) Reagent Set. The
hematocrit of each blood specimen was tested with an i-STAT
E3+ cartridge.
■RESULTS AND DISCUSSION
Oil-Based Colorimetric Ca2+ Sensors. DOS is a
commonly used plasticizer in ion-selective optodes. A
chromoionophore (ChIII), an ion exchanger (NaTFPB), and
an ionophore (CaII) are dissolved in DOS to create the
sensing oil. The sensor is a 30 μL pipet tip containing 7.5 μL of
the sensing oil (Figure 1A). The electronic pipet aspirates 7.5
μL of an aqueous solution into the tip to bring the two phases
into contact. After a multistep aspirating and dispensing
protocol (see below), the oil color is analyzed. Two microliters
of each sensing oil is further transferred to the μDrop plate for
spectrophotometric analysis. Figure 1B is a photograph of the
pipet tips after the sensing oils are equilibrated with dierent
aqueous solutions. Figure 1C shows the corresponding
response curve using the hue of the oil phase.
The original oil containing the dissolved ChIII, NaTFPB,
and CaII is yellow because the commercial ChIII molecule is in
its neutral form. There are no changes in the oil composition
and color when the aqueous phase is NaOH. However, the oil
becomes blue after equilibration with a HCl solution, as
protons are extracted into the oil to protonate ChIII.
Concurrently, Na+is released into the aqueous phase to
maintain the electroneutrality of the oil phase. The oil is
equally blue when the aqueous phase is HEPES buer at pH
7.4, indicating that ChIII is also fully protonated due to its high
basicity (pKa= 13.4).
24
As the Ca2+ concentration in the buer
increases, the oil extracts more Ca2+ to bind to the ionophore.
Correspondingly, fewer protons are extracted into the oil to
protonate ChIII because the total positive charge carried by
the cations in the oil is limited by the total negative charge of
TFPB. A decreasing protonation degree of ChIII is responsible
for the color change in Figure 1B, in which the oil becomes less
blue and more yellow with increasing Ca2+. This trend in the
protonation degree is also evidenced by the absorbance of the
sensing oils (Figures 1D and 1E).
Ion sensing based on the protonation/deprotonation of a
pH-sensitive chromoionophore is a classical response principle
of ionophore-based optodes.
13−15
However, traditional opto-
des exhibit dynamic ranges spanning several orders of
magnitude. The chromionophore has a limited window of
total color/absorbance and fluorescence change. When the
change occurs over a very broad concentration range, the
response is small in the physiologically relevant range, usually
≪1 order of magnitude. This low response slope and the
resulting low precision are important reasons why practical
biomedical applications of ion-selective optodes have not been
achieved. Indeed, most Ca2+ optodes exhibit a dynamic range
of 3−7 orders of magnitude,
25−29
while the plasma Ca2+
concentration only spans approximately 0.1 and 0.2 order of
magnitude for healthy people (2.2−2.6 mM) and asymptotic
patients (1.8−3.0 mM), respectively.
30,31
The Bakker group
proposed an exhaustive response mode of nanoparticle
optodes, in which the analyte ions in the sample are depleted
in the sensing process due to the ecient extraction of
Ca2+.
32,33
The degree of protonation of the chromoionophore
decreases by approximately 50% as Ca2+ increases from 1 to 15
μM (1.2 orders of magnitude), representing an improved
response slope. In contrast, the ChIII in our sensing oil
undergoes a 91% change in protonation degree over a Ca2+
range of only 0.4 order of magnitude (Figure S1), much larger
than those of all previous ion-selective optodes.
This extremely steep, gate-like response requires specific
ratios of sensing chemicals. Scheme 1 illustrates how the lower
Figure 1. (A) Photograph of a pipet microtip containing the Ca2+-
sensitive oil only, the oil and the aqueous sample, and two phases after
the mixing procedure. The red and green arrows point to the oil and
aqueous phases, respectively. (B) Photograph of pipet tips containing
7.5 μL of oil equilibrated with 7.5 μL of 0.1 M HEPES buer
containing dierent levels of CaCl2. The oil contains 0.67 mM ChIII,
1.33 mM NaTFPB, and 8.00 mM CaII. The HEPES buer is at pH
7.4, using KOH as the base. The concentration of both the HCl and
NaOH solutions is 0.1 M. (C) Response curve based on the hue of
the oil segment. (D) Absorption spectra of the sensing oil after
equilibration with the aqueous phase. (E) Calibration curve based on
the absorbance of the oil at 648 nm.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12559
and upper limits of detection are defined in Figure 1. The 7.5
μL volume of oil contains 5 nmol of ChIII, 10 nmol of
NaTFPB, and 60 nmol of CaII. When the aqueous phase is 7.5
μL of pH 7.4 HEPES buer without Ca2+, 5 nmol of protons is
extracted into the oil to exchange 50% of the Na+(5 nmol)
from TFPB (Scheme 1A). This process is driven by the high
basicity of ChIII and the hydrophilicity of Na+. These protons
protonate 100% of the 5 nmol of ChIII, making the oil blue.
When the aqueous phase contains 2.5 nmol of Ca2+, Ca2+
could be fully extracted into the oil with an excess ionophore,
displacing 50% of the Na+(5 nmol). There is still 5 nmol of
Na+available in the oil to be displaced by H+, allowing the 5
nmol of ChIII to be protonated. Therefore, the oil remained
blue (Scheme 1B). Using the Arsenazo III-based calcium test
kit, the quantity of Ca2+ in this aqueous sample was
determined to be <0.1 nmol post-sensing, confirming the
exhaustive sensing mode. Because there is a finite equilibrium
constant for the H+−Na+exchange governed by the pKaof
ChIII and the log Pof Na+, it is understandable that not 100%
of the 5 nmol of Na+will be displaced by protons, and
therefore slightly less than 100% of ChIII is protonated.
According to Figures 1E and S1, 87% of ChIII is protonated
for 2.5 nmol of Ca2+, which is reasonably close to complete
protonation. In the classical theory of cation-selective optodes,
the charge carried by the extracted analyte cations is assumed
to be equal to the amount of the chromoionophore that
undergoes deprotonation due to cation extraction.
13−15,32
In
our method, the complete extraction of a considerable amount
of Ca2+ (2.5 nmol in theory according to Scheme 1 and 2.0
nmol in practice according to Figure 1) does not significantly
perturb the full protonation of ChIII due to the excess
NaTFPB. As a result, a negligible response is induced until this
threshold amount of Ca2+ is reached to render the proton
extraction capability of the optode insucient. When the
aqueous phase contains 5 nmol of Ca2+, its extraction requires
the displacement of 100% of the 10 nmol of Na+in the oil.
Since the positively charged Ca2+−ionophore complexes use all
TFPB as the counterion to maintain electroneutrality, no
protons can be extracted, and ChIII remains fully deproto-
nated, marking the upper limit of detection. The amount of
Ca2+ left in this sample after the sensing process is determined
to be ∼0.2 nmol, confirming the exhaustive response mode at
this high-concentration end. Between the lower and upper
limits of detection (2.5−5 nmol), an increasing amount of Ca2+
induces a decreasing protonation degree of ChIII and a gradual
color change from blue to yellow.
This ultrasensitive response was not obtained in previous
ion-selective optodes, probably because of the use of dierent
ratios of sensing chemicals and/or the non-exhaustive response
mode. It was pointed out that there is an excess
chromoionophore over the ion exchanger in previously
reported exhaustive optodes using nanospheres.
32,33
The
lower limit of detection is not well-defined, since any extraction
of analyte ions should decrease the protonation degree of the
chromoionophore in the absence of an excess ion exchanger,
which may be why a dynamic range starting from 0 was
reported. Also, the chromoionophore is only ∼70% protonated
in the absence of the analyte, reducing the response window by
30%. Indeed, when ChIII is increased from 5 to 10 nmol to
match the TFPB in our sensing oil, its protonation degree also
decreases for the pH 7.4 buer, and the dynamic range is
broadened (Figure S2). There are reports on Ca2+ optodes
using an excess ion exchanger, but they operate in the
conventional equilibrium mode because of a low ionophore
concentration and/or a large sample volume.
27−29
The lower
detection limit is in the low micromolar to nanomolar range,
and the dynamic range covers several orders of magnitude.
Scheme 1. Response Principle of the Ultrasensitive Ca2+-Selective Oil Sensor with Well-Defined Lower and Upper Limits of
Detection
a
a
The sample contains (A) no Ca2+, (B) 2.5 nmol of Ca2+, and (C) 5.0 nmol of Ca2+. The number of moles of sensing chemicals and analytes is
labeled based on experimental conditions in Figure 1. Extra ionophores are not shown for clarity.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12560
The response in exhaustive optodes is based on the number
of moles instead of the concentration of analyte ions in the
sample. When the volume of the aqueous phase is reduced
from 7.5 to 2.5 μL (from Figure 1 to Figure 2A), the dynamic
range remains 2.5−5.0 nmol, corresponding to 1.0−2.0 mM in
the 2.5 μL sample as opposed to 0.33−0.67 mM in the 7.5 μL
sample. On the other hand, the number of moles of sensing
chemicals in the oil controls the extraction capacity for Ca2+
and therefore can be used to tune the response range. For
example, when the amount of three sensing chemicals is
reduced three-fold by reducing the oil volume (Figure 2B) or
diluting the oil (Figure 2C), the dynamic range is indeed
shifted to approximately 0.8−1.7 nmol.
Mixing Protocol and Response Time of the Oil
Sensor. The response of the oil sensor relies on the ecient
ion transfer between the two immiscible phases. If two liquids
are static, the complete color change takes at least four days
because of the slow ion diusion (Figure S3). An electronic
pipet pushes and pulls the oil and sample segments multiple
times in the pipet tip. The pressure-driven turbulent flow
enhances the mass transfer within each phase and across the
interface. Moreover, a thin layer of oil is formed on the inner
surface of the hydrophobic polypropylene pipet tip along the
mixing distance (Figure S4 and Supplementary Video). This
oil layer provides an extensive interface for ion transfer with
the moving aqueous sample, accelerating the response. Figures
3A and 3B show the eect of mixing speed and volume on the
response time. The response becomes faster as the speed of
mixing (aspiration and dispensing) increases, probably because
of the more aggressive turbulence flow. A mixing volume of 15
μL or more permits a much faster response than a mixing
volume of 5 μL. With a total volume of 15 μL of liquid in the
pipet tip, a lower mixing volume restricts the generation of a
large area of the oil layer on the pipet surface and reduces the
oil−sample contact. A response time of about 10−15 min
using a mixing speed of at least 7 (a pipet setting) and a mixing
volume of at least 15 μL is acceptable for at-home monitoring.
Since chemical diusion in the viscous oil is likely to be a
limiting step of the colorimetric response, a reduced volume of
oil is expected to accelerate the response further. Indeed, the
equilibrium response is achieved within 3 min for 2.5 μL of oil
(Figure 3C), much shorter than the 15 min response time of
7.5 μL of oil under the same mixing protocol.
We explored another mixing protocol based on unequal
aspiration and dispensing speeds. The first round of mixing
consists of 15 cycles of aspiration at a speed of 7 and
dispensing at a speed of 3. Although the oil segment is initially
closer to the pipet, the oil and the sample switch positions after
mixing (the sample becomes closer to the pipet, Figure S5).
The next round of mixing consists of 15 cycles of aspiration at
a speed of 3 and dispensing at a speed of 7. At the end of this
round, the oil and the sample switch positions again (the oil
becomes closer to the pipet, Figure S5). The equilibrium
response is reached after 8 rounds of mixing in total. As is
shown in Figure 3D, the response is slightly faster when using
this unequal-speed mixing protocol. Moreover, the oil is
primarily in one segment after this mixing protocol instead of
breaking into two segments (Figure S6), which is preferred for
the color analysis. Therefore, the unequal-speed mixing
protocol is used in this work unless otherwise specified.
When the blood after sensing is centrifuged, no reddish color is
observed in the supernatant, suggesting no hemolysis caused
by this mixing protocol.
Oil-Based Colorimetric K+Sensors. We did preliminary
experiments on formulating oil sensors for K+based on the
Figure 2. Dependence of the dynamic range on the volume ratio of
two phases and the amount of sensing chemicals in the oil. The oil
compositions in A and B are the same as those in Figure 1, while the
oil is diluted three-fold in C (0.22 mM ChIII, 0.44 mM NaTFPB, and
2.67 mM CaII).
Figure 3. Response time of the oil sensor under dierent mixing
protocols. Three representative amounts of Ca2+ in the aqueous phase
are selected to induce small, medium, and large color changes in the
oil. (A) A comparison of dierent mixing speeds was performed using
a mixing volume of 15 μL. (B) A comparison of dierent mixing
volumes using a mixing speed of 7. (C) The response time of a sensor
with 2.5 instead of 7.5 μL of oil using a mixing speed of 7 and a
volume of 15 μL. (D) A comparison of mixing protocols based on
equal and unequal aspiration and dispensing speeds. The speed
number is the setting in the E1-ClipTip Electronic Pipette. The
composition of the oil and buer is the same as that in Figure 1.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12561
exhaustive response mode. Each oil sensor in the pipet tip
contains 2 nmol of ChIII, 3.5 nmol of NaTFPB, and 35 nmol
of valinomycin. This oil should have a lower and upper
detection limit of 1.5 and 3.5 nmol, respectively, if it operates
in an ideally exhaustive mode. According to Figure 4, the
sensor indeed shows a large color change from bluish to
yellowish in this narrow range, although the actual dynamic
range is slightly wider. There is no interference from other
cations, including Na+, Ca2+, Mg2+, and Li+(Figure S7).
Considering a blood concentration range of 2−8 mM and a
hematocrit range of 35−50%, this sensing ability will meet the
requirement of blood K+analysis with appropriate dilution. An
increasing concentration of sensing chemicals will induce an
upward shift of the dynamic range to reduce the need for
dilution. However, a higher concentration of valinomycin is
not fully soluble in DOS. The solubility issue may be addressed
using other oils or K+ionophores in future work.
Assessments of the Ca2+ Sensor in Whole Blood. Ion-
selective optodes based on chromoionophores exhibit cross-
sensitivity to the sample pH. The exhaustive response mode
reduces this pH interference, but it is not completely pH-
independent, as evidenced by the change in the absolute
absorbance of nanooptodes at dierent pH values.
32
Our oil
sensor also shows noticeable changes in color and absorbance
when the sample pH varies between 7.2 and 7.6 (Figures S8
and S9), which is a potential source of error in blood Ca2+
measurements. To address the pH interference and attain the
highest possible accuracy, a dilution buer is included in the
pipet tip as a part of the disposable sensor. Five microliters of
0.1 M HEPES buer at pH 7.4 in the pipet tip is used to dilute
2.5 μL of blood. This design is not impractical since the buer
can be sandwiched between two oil segments in the pipet
microtip to prevent its evaporation during storage (Figure
S10), and the mixing protocol with unequal aspiration and
dispensing speeds easily merges the buer and blood as well as
the two oil segments (Figure S11). The buer may also be
replaced by a layer of hydrogel with buer or even dried buer
in the pipet tip in future work.
Calcium ionophore II (ETH 129) has been successfully used
in commercial blood analyzers.
1
It has perhaps the highest
binding constant for Ca2+ in DOS (25.5) among all calcium
ionophores and shows the highest selectivity (e.g., a selectivity
coecient of −6.2 over Na+in DOS) that is sucient for
blood analysis.
34
Therefore, we did not expect any interference
from other electrolytes in the blood. However, as shown in
Figure 5, a color response is clearly observed when 10−140
mM Na+is added to 1 mM Ca2+ in the 2.5 μL sample. The
Na+concentration in blood is about 140 mM and therefore
will significantly aect the color of the Ca2+-sensing oil. The
Na+-induced response also reduces the color window for Ca2+
detection. To address these problems, we use a background of
140 mM NaCl in standard solutions to simulate this blood
background for calibration purposes and reduce ChIII from
0.67 to 0.50 mM to maximize the color response of Ca2+ in the
presence of Na+.Figures 6A and 6B show the response of the
modified oil in standard solutions with the NaCl background
and the corresponding hue-based response curve.
Figure 4. (A) Photograph of pipet tips containing 15 μL of K+-
sensitive oil equilibrated with 3.0 μL of 0.1 M HEPES buer with
varying levels of KCl. The oil contains 0.133 mM ChIII, 0.233 mM
NaTFPB, and 2.33 mM valinomycin. The HEPES buer is at pH 7.4,
using NaOH as the base. (B) Response curve based on the hue of the
oil segment.
Figure 5. Selectivity of the Ca2+ sensor over other cations based on
color (A) and absorbance (B, for Na+only). The concentrations are
those of the 2.5 μL sample, excluding the 5.0 μL of HEPES buer. All
2.5 μL samples contain 1.0 mM Ca2+.
Figure 6. (A) Color response of the oil sensor to dierent amounts
and concentrations of Ca2+ in 2.5 μL of pH 7.4 HEPES buer with
140 mM NaCl. There is 5.0 μL of extra HEPES buer without NaCl
in each pipet tip. The sensing oil is DOS containing 0.50 mM ChIII,
1.33 mM NaTFPB, and 8 mM CaII. The volume of both phases is 7.5
μL. (B) Calibration curve based on the hue of the oil.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12562
For blood testing, 2.5 μL of human blood is introduced into
the pipet tip with 7.5 μL of the sensing oil and 5.0 μL of the
HEPES buer. Figure 7A shows the sensor after equilibration
with whole blood with varying levels of plasma Ca2+. Each
measurement is performed in triplicate to demonstrate the
precision of this analytical method. As can be seen, the oil
sensor exhibits highly consistent color in each triplicate
experiment with a standard deviation of only 0.5−2.1 in hue,
suggesting excellent precision in testing real-world samples.
Unlike the preparation of solid sensors consisting of multiple
layers and interfaces, the precise preparation of the solution is
much easier, and the sensor-to-sensor variation could be
minimal. Moreover, the color response of the oil sensor to an
increasing amount of Ca2+ in blood appears comparable to that
in standard solutions, suggesting no significant matrix eects.
The accuracy of the oil sensor is evaluated quantitatively in
more blood samples. The number of moles of total Ca2+ in the
2.5 μL of each blood sample is determined using the
calibration curve shown in Figure 6B. This nonlinear graph
of a fourth-degree or greater polynomial function is not
practically solvable. Instead of using the equation of the
calibration curve, the quantity of Ca2+ is estimated based on
the graph itself, which is fairly accurate with 11 data points. For
real products, a calibration curve constructed from more
standard solutions within this Ca2+ range (e.g., 51 concen-
trations for 51 data points) will permit an even more accurate
determination of the Ca2+ quantity from the hue without the
need for a mathematical equation. The concentration of total
Ca2+ in plasma is a standard parameter used in medicine for
managing chronic diseases, such as hypoparathyroidism. To
convert the number of moles of Ca2+ in whole blood to its
concentration in plasma, the plasma volume is calculated by
multiplying the total blood volume by (1 −hematocrit). As
shown in Figure 7A, the number of moles of Ca2+ and the
plasma Ca2+ concentration determined by the oil sensor match
those determined by the commercial test kit, with an error of
≤3.7%. Figure 7B shows the correlation between the proposed
oil sensor and the reference method in blood samples with 51
dierent concentrations of plasma Ca2+. The error relative to
the reference method ranges from 0.2% to 8.0%, and 43
samples have an error of <5%. Plasma Ca2+ includes free Ca2+
(ionized Ca2+) and Ca2+ bound or complexed to proteins and
anions. The high accuracy of the oil sensor suggests that the oil
containing the ionophore with an extremely high binding
anity (formation constant of 1025.5) not only extracts ionized
Ca2+ but also frees and extracts bound and complexed Ca2+.
Although a 16-channel pipet was used in this study, a real-
world sensing system will only need a single-channel electronic
pipet (Figure S12), both of which could be portable and cost
less than $1000. The cost of the sensing oil for each Ca2+
measurement is only $0.41, even with the price at
MilliporeSigma. Although the need for hematocrit correction
is a limitation of the method, the hematocrit can be
determined by wearable, non-invasive technologies (e.g., Alio
SmartPatch) or home-use analyzers requiring only 2−10 μL of
capillary blood (e.g., EKF Diagnostics’ Hemo Control Analyzer
and Nova Biomedical’s StatStrip Analyzer). It is also possible
to integrate the hematocrit measurement into this ion-sensing
device.
■CONCLUSIONS
A water-immiscible oil as a liquid sensor allows for the
ultrasensitive optical detection of ions in complex samples,
such as whole blood. This sensing method may work for other
ions such as Mg2+ and Li+, using corresponding ionophores. A
sensing system requiring liquid mixing is more complicated
than solid-state sensors, such as glucose meters, that do not
involve any motors. However, the proposed platform is still
practical for use in homes and small clinics due to its
reasonable portability, low cost, ease of use, and fast
turnaround time. Multibore tubing may allow for simultaneous
measurements of multiple analytes from a drop of blood.
■ASSOCIATED CONTENT
*
sı Supporting Information
The Supporting Information is available free of charge at
https://pubs.acs.org/doi/10.1021/acs.analchem.3c02926.
Response curve of the Ca2+-sensing oil based on the
protonation degree and additional photographs for
colorimetric responses under more conditions, mixing
processes, pH dependence, storage stability, specificity,
and pipettes (PDF)
Video showing the mixing protocol (AVI)
■AUTHOR INFORMATION
Corresponding Author
Xuewei Wang −Department of Chemistry, Virginia
Commonwealth University, Richmond, Virginia 23284,
United States; orcid.org/0000-0001-9604-3045;
Email: wangx11@vcu.edu
Figure 7. (A) Color response of the oil sensor to blood with varying
levels of Ca2+. One blood specimen is spiked with Ca2+ to generate a
range of Ca2+ concentrations for demonstration of the sensor
response. Each concentration is tested in three pipet tips with oil
sensors to show the sensor-to-sensor consistency. (B) Correlation of
the proposed oil sensor and the commercial test kit in determining the
concentration of total plasma Ca2+ in 51 blood samples.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12563
Authors
Nasrin Ghanbari Ghalehjoughi −Department of Chemistry,
Virginia Commonwealth University, Richmond, Virginia
23284, United States
Renjie Wang −Department of Chemistry and Biochemistry,
Florida Atlantic University, Boca Raton, Florida 33431,
United States; orcid.org/0000-0002-2969-0987
Savannah Kelley −Department of Chemistry, Virginia
Commonwealth University, Richmond, Virginia 23284,
United States
Complete contact information is available at:
https://pubs.acs.org/10.1021/acs.analchem.3c02926
Author Contributions
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript.
Notes
The authors declare no competing financial interest.
■ACKNOWLEDGMENTS
This work is financially supported by Virginia Commonwealth
University (startup grant for X.W.), Calcilytix Therapeutics,
and the Orphan Disease Center. We thank Dr. Lorin
Bachmann and the sta of the Blood Gas Lab of the Virginia
Commonwealth University Medical Center for providing
blood specimens.
■REFERENCES
(1) Lewenstam, A. Electroanalysis 2014,26 (6), 1171−1181.
(2) Yoo, E.-H.; Lee, S.-Y. Sensors 2010,10 (5), 4558−4576.
(3) Tusa, J. K.; He, H. J. Mater. Chem. 2005,15 (27−28), 2640−
2647.
(4) Kimball, J.; Chavez, J.; Ceresa, L.; Kitchner, E.; Nurekeyev, Z.;
Doan, H.; Szabelski, M.; Borejdo, J.; Gryczynski, I.; Gryczynski, Z.
Methods Appl. Fluoresc. 2020,8(3), 033002.
(5) Ceresa, L.; Kimball, J.; Chavez, J.; Kitchner, E.; Nurekeyev, Z.;
Doan, H.; Borejdo, J.; Gryczynski, I.; Gryczynski, Z. Methods Appl.
Fluoresc. 2021,9(3), 035005.
(6) Cai, Y.; Wei, Z.; Song, C.; Tang, C.; Han, W.; Dong, X. Chem.
Soc. Rev. 2019,48 (1), 22−37.
(7) Hutter, T.; Collings, T. S.; Kostova, G.; Karet Frankl, F. E. Sens.
Diagn. 2022,1(4), 614−626.
(8) Brandi, M. L.; Bilezikian, J. P.; Shoback, D.; Bouillon, R.; Clarke,
B. L.; Thakker, R. V.; Khan, A. A.; Potts, J. T., Jr J. Clin Endocrinol
Metab 2016,101 (6), 2273−2283.
(9) Chang, A. R.; Sang, Y.; Leddy, J.; Yahya, T.; Kirchner, H. L.;
Inker, L. A.; Matsushita, K.; Ballew, S. H.; Coresh, J.; Grams, M. E.
Hypertension 2016,67 (6), 1181−1188.
(10) Nilsson, E.; Gasparini, A.; A
rnlöv, J.; Xu, H.; Henriksson, K.
M.; Coresh, J.; Grams, M. E.; Carrero, J. J. Int. J. Cardiol 2017,245,
277−284.
(11) Nederlof, M.; Heerdink, E. R.; Egberts, A. C. G.; Wilting, I.;
Stoker, L. J.; Hoekstra, R.; Kupka, R. W. Int. J. Bipolar Disord 2018,6
(1), 12.
(12) Anna, S.; Aleksandra, S.; Ewa, P. Magnes Res. 2016,29 (3),
112−119.
(13) Bakker, E.; Buhlmann, P.; Pretsch, E. Chem. Rev. 1997,97 (8),
3083−3132.
(14) Mistlberger, G.; Crespo, G. A.; Bakker, E. Annual Review of
Analytical Chemistry 2014,7(1), 483−512.
(15) Xie, X.; Bakker, E. Anal Bioanal Chem. 2015,407 (14), 3899−
3910.
(16) Xie, L.; Qin, Y.; Chen, H.-Y. Anal. Chem. 2013,85 (5), 2617−
2622.
(17) Du, X.; Xie, X. ACS Sens 2017,2(10), 1410−1414.
(18) Du, X.; Zhai, J.; Zeng, D.; Chen, F.; Xie, X. Sens Actuators B
Chem. 2020,319, No. 128300.
(19) Zhai, J.; Xie, X.; Cherubini, T.; Bakker, E. ACS Sens 2017,2
(4), 606−612.
(20) Hisamoto, H.; Horiuchi, T.; Uchiyama, K.; Tokeshi, M.;
Hibara, A.; Kitamori, T. Anal. Chem. 2001,73 (22), 5551−5556.
(21) Hisamoto, H.; Horiuchi, T.; Tokeshi, M.; Hibara, A.; Kitamori,
T. Anal. Chem. 2001,73 (6), 1382−1386.
(22) Wang, X.; Sun, M.; Ferguson, S. A.; Hoff, J. D.; Qin, Y.; Bailey,
R. C.; Meyerhoff, M. E. Angew. Chem., Int. Ed. 2019,58 (24), 8092−
8096.
(23) Wang, R.; Zhou, Y.; Ghanbari Ghalehjoughi, N.; Mawaldi, Y.;
Wang, X. Anal. Chem. 2021,93 (40), 13694−13702.
(24) Qin, Y.; Bakker, E. Talanta 2002,58 (5), 909−918.
(25) Morf, W. E.; Seiler, K.; Rusterholz, B.; Simon, W. Anal. Chem.
1990,62 (7), 738−742.
(26) Rong, G.; Kim, E. H.; Poskanzer, K. E.; Clark, H. A. Sci. Rep
2017,7(1), 10819.
(27) Xu, C.; Wygladacz, K.; Qin, Y.; Retter, R.; Bell, M.; Bakker, E.
Anal. Chim. Acta 2005,537 (1), 135−143.
(28) Stelmach, E.; Nazaruk, E.; Maksymiuk, K.; Michalska, A. Anal.
Chem. 2021,93 (39), 13106−13111.
(29) Dailey, A. L.; Greer, M. D.; Sodia, T. Z.; Jewell, M. P.; Kalin, T.
A.; Cash, K. J. Biosensors (Basel) 2020,10 (9), 120.
(30) Carroll, M. F.; Schade, D. S. Am. Fam. Physician. 2003,67 (9),
1959−1966.
(31) Abate, E. G.; Clarke, B. L. Front Endocrinol (Lausanne) 2017,7,
172.
(32) Xie, X.; Zhai, J.; Bakker, E. Anal. Chem. 2014,86 (6), 2853−
2856.
(33) Xie, X.; Zhai, J.; Crespo, G. A.; Bakker, E. Anal. Chem. 2014,86
(17), 8770−8775.
(34) Qin, Y.; Mi, Y.; Bakker, E. Anal. Chim. Acta 2000,421 (2),
207−220.
Analytical Chemistry pubs.acs.org/ac Article
https://doi.org/10.1021/acs.analchem.3c02926
Anal. Chem. 2023, 95, 12557−12564
12564
Recommended by ACS
Dual Ion-Selective Membrane Deposited Ion-Sensitive Field-
Effect Transistor Integrating a Whole Blood Processing
Microchamber for In Situ Blood Ion Testing
Xiao-Wen Chen and Nien-Tsu Huang
JANUARY 19, 2023
ACS SENSORS READ
Rapid and Accurate Measurement of the Na+/K+ Balance in
Urine for Remote Patient Monitoring Using a Symmetric
Electrode Architecture
Guillaume Bouilly.
MARCH 01, 2023
ANALYTICAL CHEMISTRY READ
Three-in-One Detector by 3D Printing: Simultaneous
Contactless Conductivity, Ultraviolet Absorbance, and Laser-
Induced Fluorescence Measurements for Capillary...
Yingchun Wang, Min Zhang,
et al.
JANUARY 15, 2023
ANALYTICAL CHEMISTRY READ
In Vivo Transdermal Multi-Ion Monitoring with a
Potentiometric Microneedle-Based Sensor Patch
Águeda Molinero-Fernández, Gastón A. Crespo,
et al.
DECEMBER 07, 2022
ACS SENSORS READ
Get More Suggestions >
