Using smell to triage samples in point-of-care assays

Article abstract of DOI:10.1002/anie.201207008

Smell of success: Reagent 1 provides the dual readouts of odor (ethanethiol) and fluorescence (derivative of 7-hydroxycoumarin) and can be used in down-selection assays based on smell and quantitative fluorescence assays of the samples that give a positive result. An important feature of 1 is the matched sensitivity of the two outputs. This reagent is designed for use in resource-limited settings and is demonstrated in assays that detect enzymes. Copyright

Full text of DOI:10.1002/anie.201207008

Angewandte
Chemie
DOI: 10.1002/anie.201207008
Assay Development
Using Smell To Triage Samples in Point-of-Care Assays**
Hemakesh Mohapatra and Scott T. Phillips*
Quantification of an analyte in point-of-care assays is more
time consuming, resource intensive, and expensive than
conducting simple yes/no qualitative assays. Quantitative
assays require highly engineered assays, electronic readers,
and knowledgeable operators, whereas qualitative assays
often involve simple colorimetric responses on a strip of
paper.[1] In resource-limited environments, such as remote
regions in the developing world, the challenges in conducting
quantitative point-of-care assays become even more pro-
nounced since many individuals must be tested quickly for the
presence and (if appropriate) the quantity of a biomarker to
determine the necessary treatment and dosage. To minimize
the time commitment associated with quantitative point-of-
care assays in these regions, we now describe a reagent and
assay strategy that enables rapid triaging of samples followed
by quantitative measurements of positive tests. These require-
ments are achieved by using the same single-step assay and
does not require a quantitative measurement to be obtained
on all samples (which is the standard approach). The triaging
step is accomplished by smell (a process that requires little
time), and the quantitative measurements are obtained by
using fluorescence spectroscopy (Scheme 1). With this
approach, the absence of the analyte in the triaging step
means that the quantitative measurement can be omitted,
thus saving time and effort.
This type of rapid down-selection process is made possible
by small-molecule 1, which enables both the qualitative and
Scheme 1. A strategy for triaging samples in point-of-care assays by using
reagent 1. a) Depiction of the concept. b) Reaction of reagent 1 with
hydrogen peroxide (a signal-transduction reagent) to form the dual readouts
of smell (ethanethiol) and fluorescence (2). c) An assay strategy that
involves the use of 1 and detection reagent (3) to detect enzyme analytes.
quantitative readouts in a single assay (Scheme 1a).[2] The
qualitative readout (i.e., the odor of released ethanethiol)[3]
matches the sensitivity of the quantitative readout (fluores-
cence); this feature is crucial for an effective down-selection
process, particularly when detecting and quantifying analytes
that are present in low concentrations.
Reagent 1 functions by reacting selectively with hydrogen
peroxide[4] by oxidative cleavage of the arylboronate on 1 to
generate an unstable intermediate phenol, which then under-
goes a rapid 6-exo-trig cyclization to release ethanethiol
(odor) and generate a 7-hydroxycoumarin derivative (2;
fluorescent readout; Scheme 1b).
Hydrogen peroxide, however, is not the target analyte in
this system. Instead, it is a signal-transduction reagent that
enables a detection reagent (3) and the output reagent (1)
(Scheme 1c) to operate in tandem in a single assay. With this
approach a variety of analytes can be detected (and down-
selected for) using essentially the same reagents and mea-
surement procedures in all assays (only reagent 3 would need
to be modified from one assay to the next). For example,
hydrogen peroxide can be derived from specific activity-based
detection events with enzyme biomarkers (Scheme 1c) when
the presence of the enzyme causes release of glucose from the
detection reagent (3). This glucose is then converted into
hydrogen peroxide by glucose oxidase, which also is present
as a reagent in the assay.
[*] H. Mohapatra, Prof. S. T. Phillips
Department of Chemistry, The Pennsylvania State University
University Park, PA 16802 (USA)
E-mail: sphillips@psu.edu
Homepage: http://research.chem.psu.edu/stpgroup/
[**] This work was supported by the Arnold and Mabel Beckman
Foundation, the Camille and Henry Dreyfus Foundation, and Louis
Martarano. S.T.P acknowledges support from the Alfred P. Sloan
Research Fellows program. We appreciate the contributions of Kyle
Schmid, Kristin Beiswenger, Travis Cordes, Nicole Thom, Marley
Pillion, Kimy Yeung, Henry Kaweesi, Anthony DiLauro, Jessica
Robbins, Matthew Baker, Greg Lewis, and Michael Olah in providing
independent evaluation of the qualitative portions of the assay.
In point-of-care settings, the quantitative fluorescence
assay would be accomplished by using a handheld fluores-
cence spectrometer (a variety of relatively inexpensive
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201207008.
Angew. Chem. Int. Ed. 2012, 51, 11145 –11148
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11145

.
Angewandte
Communications
commercial versions are available), or by using an even less
expensive paper-based microfluidic device that contains
internal fluidic batteries to power on-chip fluorescence
assays.[5] In this study, we demonstrate the performance of
the reagent by using a laboratory fluorescence spectrometer.
Because of the key role of 1 in the down-selection assay, the
focus herein is on the design, synthesis, and performance of 1.
We also demonstrate model quantitative assays for the
enzymes b-d-galactosidase[6] and alkaline phosphatase[7] to
illustrate that multiple classes of enzymes can be detected and
measured with this reagent. Future reports will describe fully
developed down-selection assays conducted in settings out-
side of the laboratory.
1% MeOH, pH 7.4) was monitored by using fluorescence
spectroscopy; the emission intensity between 500 nm and
570 nm increased over a period of 12 minutes as 1 was
converted into 2 (Figure 2). The rapid increase in fluores-
cence emission in this spectral region further confirms the
formation of 2.
We prepared reagent 1 in two steps from commercially
available reagents (Scheme 2). The sequence of synthetic
Figure 2. Time-dependent fluorescence emission (I) when
1 ([1]final =20 mm in 10 mm phosphate buffer containing 1% MeOH (v/
v), pH 7.4) was treated with hydrogen peroxide ([H2O2]final =22 mm) to
form 2. The excitation wavelength (lex) was 470 nm and the experiment
was conducted at 208C. Wavelength scans were acquired every 2
minutes. The initial scan and the data were acquired at 2 minutes
overlap.
Scheme 2. Synthesis of sensor reagent 1. Reaction conditions: a) Bis-
(pinacolato)diboron, KOAc, [PdCl2dppf]·CH2Cl2, 1,4-dioxane, 808C
(84%). b) CH2(COSEt)2, DABCO, molecular sieves (3 ꢀ), THF (35%).
steps involved a Miyaura borylation reaction[8] of 4 followed
by a base-catalyzed Knovenagel condensation[9] to transform
5 into 1. Although this Knovenagel condensation proceeded
in only modest yield (35%), the remainder of the material
consisted primarily of recovered starting material.
When treated with 5 equivalents of hydrogen peroxide,
1 (0.5 mm in 1:1 MeOH/10 mm phosphate buffer, pH 7.4,
208C) was consumed within minutes and 2 was formed
immediately (Figure 1). In fact, the cyclization reaction to
When 1 is exposed to different concentrations of hydro-
gen peroxide in a fixed-time assay, the intensity of the
fluorescence emission signal obtained is proportional to the
concentration of hydrogen peroxide (Figure 3). This correla-
tion between intensity of fluorescence signal and concentra-
tion of hydrogen peroxide suggests that hydrogen peroxide
should operate effectively as a signal-transduction reagent in
a two-reagent assay for detecting enzyme biomarkers, as
depicted in Scheme 1c.
To test this idea, we set up two separate two-component
assays for selectively detecting either b-d-galactosidase or
alkaline phosphatase (Scheme 3).The goal was to demon-
strate that: 1) hydrogen peroxide does in fact serve as an
effective signal-transduction reagent, 2) by using reagent
1 enzyme analytes can be detected qualitatively at levels that
are relevant to the quantitative assay, and 3) the assay strategy
is amenable to a variety of classes of enzyme analytes.
Figure 1. Overlayed LC–MS spectra for three sequential injections of
aliquots taken from the reaction of 1 (0.5 mm) with 5 equivalents of
H2O2 in 1:1 MeOH/10 mm phosphate buffer, pH 7.4 at 208C.
form 2 (Scheme 1b) is sufficiently rapid that none of the
oxidative cleavage product (i.e., the phenol) was detectable
by LC–MS. The presence of a characteristic thiol odor during
these reactions confirmed the release of ethanethiol.
The progress of the reaction of hydrogen peroxide
(1.1 equiv) with 1 (20 mm, 10 mm phosphate buffer containing
Scheme 3. Specific detection reagents 3a and 3b that are used in
combination with 1, as shown in Scheme 1c, to detect and quantify
b-d-galactosidase and alkaline phosphatase, respectively.
11146 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 11145 –11148

Angewandte
Chemie
Figure 3. Calibration curve for the detection of hydrogen peroxide by
using 1 in a 10 minute assay. The final concentration of 1 in the assay
was 20 mm in 10 mm phosphate buffer, pH 7.4 containing 1% (v/v)
MeOH. The fluorescence emission intensity (I) was measured at
510 nm with lex =470 nm. I0 denotes the fluorescence intensity
obtained for [H2O2]=0 mm. The inset graph reveals the linear region
defined by the dotted rectangle. All data points represent the average
of three independent measurements. The limit-of-detection (LOD)
(defined as 3ꢁstandard deviation of 1 mm H2O2/slope of the line in
the inset) is 0.3 mm.
We prepared two calibration curves, one for b-d-galacto-
sidase and one for alkaline phosphatase, (Figure 4) from
results obtained from 1 h assays. The calibration curve for b-
d-galactosidase (Figure 4a) reveals
a limit-of-detection
(LOD) of 21 nm (enzyme activity of 2.2 UmLÀ1), with
a linear dynamic range extending up to 500 nm b-d-galacto-
sidase.
Figure 4. Calibration curves for quantitative detection of b-d-galactosi-
dase (a) and alkaline phosphatase (b) by using reagents 1 and 3
(Scheme 3). a) The final concentration of 1 was 20 mm in 10 mm
phosphate buffer, pH 7.4 containing 1% (v/v) MeOH. The concentra-
tion of lactose (3a) and glucose oxidase were 10 mm and 1000 ULÀ1
respectively. b) The final concentration of 1 in the assay was 20 mm in
50 mm Tris buffer, pH 7.4 containing 1% (v/v) MeOH. The concen-
tration of d-glucose-6-phosphate (3b) and glucose oxidase were
10 mm and 1000 ULÀ1 respectively. In both (a) and (b), the fluores-
cence emission intensity (I) was measured at 510 nm using
To demonstrate the effectiveness of the smell-based
down-selection process, we conducted double-blind tests as
follows: ten individuals each were given four different
samples, two of which were control assays. The other two
assays were chosen randomly from a selection that contained
either 0 nm or 200 nm b-d-galactosidase. By using only smell
as an indicator, 80% of samples were identified correctly
either as being the same as the control or as containing b-d-
galactosidase based on a distinct thiol odor. This level of
sensitivity in the qualitative assay is perfectly matched with
the sensitivity of the quantitative assay, which indicates that
the former is an effective down-selection tool for the latter.
The calibration curve for alkaline phosphatase (Fig-
ure 4b) revealed a LOD of 10 ULÀ1 alkaline phosphatase,
with a dynamic range extending to 1000 ULÀ1. In this case,
when assays containing either 0 ULÀ1 or 100 ULÀ1 alkaline
phosphatase were tested, 70% of samples were identified
correctly by smell either as being the same as the control or as
containing a distinct thiol odor. This result further confirms
the match in sensitivity of the qualitative and quantitative
readouts for the assay.
lex =470 nm, and the assay time was 1 h. I0 denotes the fluorescence
intensity obtained in the absence of the enzyme analyte. All data
points represent the average of three independent measurements.
odor will be as apparent in real samples under typical
conditions encountered in resource-limited settings. Likewise,
a larger sample pool will be used to determine the limit of
sensitivity for the qualitative portion of the assay. The initial
results outlined herein demonstrate that appropriate reagent
design enables rapid down-selection before quantitative
fluorescence assays are conducted. This strategy should
facilitate the time-consuming process of performing quanti-
tative assays in resource-limited environments.
In conclusion, we have developed a new, dual-readout-
assay strategy that enables triaging of samples based on smell,
followed by quantitative fluorescent measurements of the
positive samples. Future studies will ascertain whether the
Received: August 29, 2012
Published online: October 4, 2012
Angew. Chem. Int. Ed. 2012, 51, 11145 –11148
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 11147

.
Angewandte
Communications
[4] E. W. Miller, A. E. Albers, A. Pralle, E. Y. Isacoff, C. J. Chang, J.
Am. Chem. Soc. 2005, 127, 16652 – 16659.
[5] N. K. Thom, K. Yeung, M. B. Pillion, S. T. Phillips, Lab Chip 2012,
12, 1768 – 1770.
[6] The enzyme b-d-galactosidase is a general biomarker of fecal
contamination in drinking water, see: C. M. Davies, S. C. Apte,
Environ. Toxicol. 1999, 14, 355 – 359.
[7] The enzyme alkaline phosphatase (ALP) is commonly used as an
indicator of bacterial contamination in dairy products, see: R.
Aschaffenburg, J. E. C. Mullen, J. Dairy Res. 1949, 16, 58 – 67.
[8] T. Ishiyama, M. Murata, N. Miyaura, J. Org. Chem. 1995, 60,
7508 – 7510.
Keywords: fluorescence spectroscopy · hydrogen peroxide ·
point-of-care · sensors · thiols
.
[1] a) A. W. Martinez, S. T. Phillips, G. M. Whitesides, E. Carrilho,
Anal. Chem. 2010, 82, 3 – 10; b) P. Yager, T. Edwards, E. Fu, K.
Helton, K. Nelson, M. R. Tam, B. H. Weigl, Nature 2006, 442,
412 – 418.
[2] The following reference shows a different type of down-selection
process that was used as a tool in organic synthesis: S. H. Shabbir,
C. J. Regan, E. V. Anslyn, Proc. Natl. Acad. Sci. USA 2009, 106,
10487 – 10492.
[3] The human nose is particularly sensitive to low-molecular-weight
thiols and is capable of detecting concentrations as low as 2 ppb,
see: J. Greenman, M. El-Maaytah, J. Duffield, P. Spencer, M.
Rosenberg, D. Corry, S. Saad, P. Lenton, G. Majerus, S. Nachnani,
J. Am. Dent. Assoc. JADA 2005, 136, 749 – 757.
[9] a) T.-H. Kim, T. M. Swager, Angew. Chem. 2003, 115, 4951 – 4954;
Angew. Chem. Int. Ed. 2003, 42, 4803 – 4806; b) P. A. Rose, H. J.
Liu, Synth. Commun. 1991, 21, 2089 – 2095.
11148 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 11145 –11148