Content uploaded by Jia Li
Author content
All content in this area was uploaded by Jia Li on Jul 05, 2019
Content may be subject to copyright.
Registered charity number: 207890
Featuring work from the Molecular Engineering Research
Group in the School of Science and Engineering at the
University of the Sunshine Coast, Queensland, Australia.
Title: Multiplex lateral fl ow detection and binary encoding
enables a molecular colorimetric 7-segment display
Binary and molecular encoding enables a digital-like 7 segment
display on a paper-based biosensor. The display operates without
batteries and wires, providing an e cient, compact, and intuitive
output for multiplexing applications.
As featured in:
See Jia Li and Joanne Macdonald,
Lab Chip, 2016, 16, 242.
www.rsc.org/loc
Lab on a Chip
COMMUNICATION
Cite this: Lab Chip,2016,16,242
Received 25th October 2015,
Accepted 24th November 2015
DOI: 10.1039/c5lc01323b
www.rsc.org/loc
Multiplex lateral flow detection and binary
encoding enables a molecular colorimetric
7-segment display†
Jia Li
a
and Joanne Macdonald*
ab
Multiplex expansion in point-of-care diagnostics usually requires a
linear increase of premium commodities such as reagents or
space. Here we demonstrate the power of binary and molecular
encoding to compress device operations. We describe the first
colorimetric 7-segment display on a paper-based biosensor, pro-
viding compact and intuitive read-outs for multiplex detections.
Multiplexing is a critical parameter for increasing diagnostic
efficiency. The strategies that enable simultaneous analysis of
multiple samples are largely dependent on the underlying
diagnostic technology. Multiplex detection using enzyme-
linked immunosorbent assays, real-time polymerase chain
reactions, microarrays and/or bead-based methods (such as
the Bio-Rad Bio-Plex®Systems
1
and Luminex MagPix®
2
)
enable high-throughput and low volume processing, but
require non-portable equipment and trained personnel for
operation. For point-of-care settings, lateral flow devices
(LFDs) are ideal candidates, but few commercial devices
employ multiplexing. This is due to issues with specificity
and reproducibility, as well as expansion limits, as described
by Washburn, because the flow rate decreases with distance
from the conjugate pad.
3
Here we propose a novel solution
for expanding LFD multiplex detection that increases the effi-
ciency of detection without expanding the device dimensions
and consuming excess reagents. Our generic solution com-
presses multiplex LFD data by borrowing from computational
science and applying binary encoding to create signature pat-
terns of test dots. Thus, a sample is diagnosed based on the
set of test dots that appear on the device. By judicious
arrangement of test dots, we create a 7-segment LFD display
that simulates digital display of information for easy interpre-
tation by the end-user.
To develop a generic multiplex LFD, we first identified
twelve individual antigen–antibody single-plex assays able to
operate using a sandwich immunoassay format. In this format
an analyte is sandwiched between a capture and detection anti-
body via tagged antigens (Fig. 1A). Gold nanoparticles (AuNPs)
conjugated to the capture antibody transform analyte detection
into a colorimetric signal. We chose fluorescein and anti-
fluorescein as the common AuNP conjugated capture antibody,
and identified an additional twelve commercially available anti-
gen–antibody combinations that could potentially be used as
detection antibodies (Fig. 1B). The analyte was a single-
stranded DNA corresponding to a segment of the L gene of
Rift Valley fever virus (RVFV).
4
Importantly, we kept the RVFV
DNA analyte sequence identical in all tests (apart from the 3′
antigen labels) to eliminate any differences in behavior due
to subtle changes in nucleotide sequences.
5
In addition, use
of a dual-labeled synthetic analyte reduced the complexity of
interactions for our proof-of-concept demonstration (applied
implementation would utilize labeled oligonucleotide, anti-
body, or aptamer binding partners to capture an unlabeled
analyte). All twelve antigen–antibody pairs were effective as
detection entities in the single-plex LFDs (Fig. 1 & S1†), with
standard LFD sensitivities as low as 5 nM.
6
We note in partic-
ular the success of labels Alexa Fluor®488, Cascade Blue®,
Lucifer yellow, benzopyrene, BODIPY®FL, and dansyl, which
have not previously been applied in lateral flow detection.
After demonstrating successful operation of all twelve
detection antigen–antibody combinations in a single-plex
LFD, we then attempted to combine these into a multiplex
LFD array using a dot-matrix (3 ×4) format. For the arrange-
ment of detection antibodies on the membrane, we hypothe-
sized that the antibodies furthest from the conjugate pad
would suffer the most from loss of reagents, and thus placed
the least sensitive detection antibodies closest to the conju-
gate pad (Fig. S2A†). Specificity testing indicated three non-
specific reactions (Fig. S2B†). To investigate the lack of
242 |Lab Chip,2016,16,242–245 This journal is © The Royal Society of Chemistry 2016
a
Inflammation and Healing Research Cluster, Genecology Research Center, School
of Science and Engineering, University of the Sunshine Coast, Locked Bag 4,
Maroochydore DC, Queensland 4558, Australia. E-mail: jmacdon1@usc.edu.au
b
Division of Experimental Therapeutics, Department of Medicine, Columbia
University, New York, NY 10032, USA. E-mail: jm2236@columbia.edu
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c5lc01323b
Lab Chip,2016,16,242–245 | 243This journal is © The Royal Society of Chemistry 2016
binding for BODIPY®FL and dansyl antigen–antibody pairs,
we reduced the LFD to a 7-dot array containing six detection
antibodies (chosen due to their superior specificity; see Fig.
S2†), with a seventh detection antibody of either anti-
BodipyFL, anti-dansyl, or anti-Cy5 detection antibody (Fig. 2A;
anti-Cy5 was included as a comparative control). These
were tested with a single dual-labeled analyte containing
either 3′BODIPY®FL, dansyl, or Cy®5 antigens, or with
an analyte reaction mix containing the corresponding seven
antigens. Interestingly, the BODIPY®FL antigen still did not
produce any observable test dots, whereas the dansyl test dots
appeared and showed specificity in the seven antigen-
antibody pairs system (Fig. 2). The lack of BODIPY®FL binding
(Fig. 2 & S2†) is most likely due to the larger surface area
ablating detection, since BODIPY®FL was the least sensitive
in the single-plex assays (Fig. 1). However, the lack of dansyl
binding in the 3 ×4 array but successful binding in the 7-dot
array was consistent in multiple tests. This suggests the pres-
ence of other detection antibodies precluded dansyl binding,
despite it being placed in the first row of the 3 ×4 array. The
7-dot array that incorporated dansyl (Fig. 2), demonstrated
significant relative higher dot intensity (p<0.05) compared
to the Cy®5 (Fig. 1 and 2), and was used for all subsequent
demonstrations. These results support the notion that
specificity is a hindrance in expansion into multiplexing, but
with judicious design and repeated testing, compatible com-
binations can be identified that do not cross-react but
instead reproducibly give selective responses.
On a linear scale, our 7-dot LFD array is an improvement
on the current maximum multiplex detection that employs
antigen–antibody reactions, since only penta-plex LFD has
been previously reported.
7
However, stacking the test lines
(or dots) along the flow path of the lateral flow device does
not provide an intuitive result output.
8
Inspired by the
electronic information display systems used in digital
watches and elevators, we considered development of a
proof-of-concept 7-segment LFD display. We devised a detec-
tion system to represent each segment of a 7-segment display
with one detection antibody (from the seven) (Fig. 3A and S3A†).
Fig. 2 Hepta-plex lateral flow detection results for combinations of
seven antigen–antibody pairs. A: Positioning of detection antibodies
deposited upon lateral flow strips. B: Hepta-plex lateral flow detection of
1μM dual-labeled analyte alone (containing either 3′BODIPY®FL, Cy®5,
or dansyl), or combinations of dual-labeled analyte (1 μM) containing 3′
biotin, Cascade Blue®, digoxigenin, dintrophenyl, tetramethylrhodamine
(TAMRA), and Texas Red®and either BODIPY®FL, Cy®5 or dansyl, as
indicated. Assay was repeated three times and a representative photo-
graph is shown. Experimental details are described in the ESI.†
Fig. 3 7-Segment display of numbers on lateral flow strips. A:
Positioning of the detection antibodies to form the 7-segments of the
display. B: Addition of labeled analyte signature mixtures (1 μM each)
resulted in the successful appearance of numbers (0 to 9). The assay
was performed three times with similar results; a photograph of one
test is shown. Experimental details are described in the ESI.†
Fig. 1 (A) Design of single-plex LFDs: a dual-labeled analyte (RVFV
DNA) was sandwiched between capture and detection antibodies due
to antigen–antibody binding; AuNPs conjugated to the capture anti-
body (mouse anti-fluorescein antibody) enabled visualization of bind-
ing through the appearance of a red color at the test dot; rabbit anti-
mouse antibody, which can directly bind the capture antibody (mouse
anti-fluorescein), was deposited in parallel as a control. (B) Determina-
tion of detection limits for each single-plex assay: detection antibodies
corresponding to the 3′antigen of dual-labeled RVFV were deposited
at the test dot; resultant assay color intensity (average and standard
deviation of 4 individual tests) was quantified using ImageJ software
and plotted against DNA concentration; a cut-off defined as three
standard deviations above the average was used to determine the limit
of detection for each assay. Experimental details are described in the
ESI.†
Lab on a Chip Communication
244 |Lab Chip,2016,16,242–245 This journal is © The Royal Society of Chemistry 2016
Initially we used duplicate test dots to represent each seg-
ment, however, this resulted in position effects where the
morphology of the second test dot was influenced by the first
test dot in the vertical segments (Fig. S3†). We thus reduced
the vertical segments to only incorporate one test dot
(Fig. 3A). By addition of appropriate label mixtures, we con-
sistently demonstrated clear numbers (0–9) on the 7-segment
LFD display (Fig. 3B).
Our successful demonstration of the numbers 0–9 using
seven antigen–antibody pairs on a LFD is the first digital-like
display of numbers on a paper-based biosensor. It operates
as a single-use static output that employs a pre-defined
molecular encoding strategy for information transfer. The
work is a significant advance on our previous molecular
7-segment display, based on molecular logic gates, which
required multiple additions of samples to each segment.
5
Here the solid-phase LFD enables a single addition of sample
to create a display, made possible through the novel combi-
nation of both molecular and binary encoding. This is nota-
bly different from other LFD computational strategies, which
showcase embedded molecular logic gates.
9,10
Our method
also advances previous multiplex LFDs, which use multiple
lines,
7,11
bi-directional,
12,13
parallel,
14,15
and multi-directional
systems.
16,17
Notably, intuitive read-outs were previously pro-
posed by Shen et al., where blood types on a paper-based bio-
sensor were displayed using letters deposited in different sec-
tions of the biosensor (not overlayed).
18,19
However,
expansion of these multiplex LFDs is hampered by either
space limitations within the original device dimension or the
requirement to consume more reagents as the device size
increases. In contrast, our novel combination of both binary
and molecular encoding demonstrates improved multiplexing
efficiency while minimizing excess reagent requirements.
The most compact LFD system to date is the microarray
LFD,
20,21
which similarly minimizes both reagent consump-
tion and device dimension. The largest demonstration to
date is detection of 384 recombinant protein antigens for
analysis of human protein atlas antibody cross-reactivities.
21
Here we provide further compaction by applying binary
encoding. Our generic antigen–antibody method could theo-
retically detect 127 (2
7
−1) discrete analytes using only 7
antigen–antibody combinations. Noticeably, our system would
only apply to differentiation of discrete analytes: different
encoding schemes would need to be employed to differenti-
ate analyte mixtures. However, implementation of both
binary-encoded and traditional microarray LFD requires
external readers to interpret results. Our 7-segment LFD
intentionally restrict outputs to a numerical system to dem-
onstrate an intuitive read-out applicable to point-of-care diag-
nostics that does not require external reading devices.
In this study we used an artificial ideal analyte for proof-
of-concept demonstration of multiplex LFD displays. Impor-
tantly, these displays are generic and can be applied for the
detection of any candidate analyte (e.g. nucleic acids, proteins,
lipids, or small molecules) through the use of labeled binding
partners. Practically, this could be implemented via labeled
antibodies or aptamers for detection of their corresponding
antigens. In addition, we are also considering labeled
primers and probes for multiplex nucleic acid amplification,
which would additionally assist with detection of low copy
numbers. In this context, we note the advancement of iso-
thermal nucleic acid amplification technologies that offer
prospects for rapid point-of-care detection.
22,23
Determina-
tion of analyte concentration could be performed using
electronic reading devices (that assess intensity within a
defined test area regardless of dot morphology) and by pre-
training with known concentrations of analyte. Notably,
implementation of pattern displays requires the use of mix-
tures of labeled binding partners, such that an analyte must
combine with a unique subset of labels. This means that
multiple individual sub-reactions must work synchronously,
which ultimately increases confidence that can be drawn
from a successful display.
Conclusions
Our novel 7-segment LFD enables compaction of
multiplexing by borrowing from computational science and
employing a binary encoding scheme that moves beyond
space limitations. The easy-to-interpret results via a 7-seg-
ment display format are highly relevant to point-of-care appli-
cations. These displays are generic and can be applied for
multiplex detection of any candidates (e.g. nucleic acids, pro-
tein, lipid, or small molecules) if capture reagents incorpo-
rate our specific recognition labels. This is particularly ame-
nable for nucleic acid lateral flow detection, if upstream
nucleic acid amplification uses primers and probes that
incorporate the recognition labels.
22
Further improvements
could focus on employing advanced deposition techniques to
improve the performance and presentation of the outputs.
24
Acknowledgements
We thank John Bartlett, David McMillan, and Fabrice
Rossignol for support and advice, and Richard Burns for care-
ful reading of this manuscript. This work was funded by the
Queensland Government, Department of Science, Information
Technology, Innovation and the Arts (DSITIA, Australia), and
an internal HDR grant (University of the Sunshine Coast,
Australia).
Notes and references
1 I. Bio-Rad Laboratories, Bio-Rad Bio-Plex®Systems, http://
info.bio-rad.com/ww-Bio-Plex-Luminex-lp1.html?WT.srch=
1&WT.mc_id=aw-pqd-AU-bioplex_luminex_gold&WT.knsh_
id=631f362c-bf07-4e69-b01a-00fb1c8cb35f&gclid=COKLk_
mOjcgCFYiCvQodPpEJWA, Accessed September 23rd, 2015,
2015.
2 L. Corporation, Luminex MagPix®, http://www.luminexcorp.
com/clinical/instruments/magpix/, Accessed September 23rd,
2015, 2015.
3 E. W. Washburn, Phys. Rev., 1921, 17, 273–283.
Lab on a ChipCommunication
Lab Chip,2016,16,242–245 | 245This journal is © The Royal Society of Chemistry 2016
4 M. Euler, Y. Wang, O. Nentwich, O. Piepenburg, F. T. Hufert
and M. Weidmann, J. Clin. Virol., 2012, 54, 308–312.
5 J. Macdonald, Y. Li, M. Sutovic, H. Lederman, K. Pendri, W.
Lu, B. L. Andrews, D. Stefanovic and M. N. Stojanovic, Nano
Lett., 2006, 6, 2598–2603.
6 J. Li, D. McMillan and J. Macdonald, Sens. Mater., 2015, 27,
549–561.
7 P. Noguera, G. A. Posthuma-Trumpie, M. van Tuil, F. J. van
der Wal, A. de Boer, A. P. H. A. Moers and A. van
Amerongen, Anal. Bioanal. Chem., 2011, 399, 831–838.
8 J. E. Poje, T. Kastratovic, A. R. Macdonald, A. C. Guillermo,
S. E. Troetti, O. J. Jabado, M. L. Fanning, D. Stefanovic and
J. Macdonald, Angew. Chem., Int. Ed., 2014, 53, 9222–9225.
9 J. Chen, Z. Fang, P. Lie and L. Zeng, Anal. Chem., 2012, 84,
6321–6325.
10 Y. Huang, W. Wen, D. Du, X. Zhang, S. Wang and Y. Lin,
Biosens. Bioelectron., 2014, 61, 598–604.
11 L. Tian, T. Sato, K. Niwa, M. Kawase, A. C. Tanner and N.
Takahashi, BioMed Res. Int., 2014, 2014, 180323.
12 S. M. Hossain, C. Ozimok, C. Sicard, S. D. Aguirre, M. M. Ali,
Y. Li and J. D. Brennan, Anal. Bioanal. Chem., 2012, 403,
1567–1576.
13 S. M. Hossain, R. E. Luckham, M. J. McFadden and J. D.
Brennan, Anal. Chem., 2009, 81, 9055–9064.
14 D. a. C. Becton, BD™EZ Flu A + B Test, http://www.bd.com/
ds/productCenter/256050.asp, Accessed January 12th, 2015.
15 A. Technologies, RAID, http://www.alexeter.com/biow/Products/
products/strips/RAID_DX.asp, Accessed January 12th, 2015.
16 C. Z. Li, K. Vandenberg, S. Prabhulkar, X. Zhu, L. Schneper,
K. Methee, C. J. Rosser and E. Almeide, Biosens. Bioelectron.,
2011, 26, 4342–4348.
17 K. L. Peters, I. Corbin, L. M. Kaufman, K. Zreibe, L. Blanes
and B. R. McCord, Anal. Methods, 2015, 7,63–70.
18 M. Li, J. Tian, M. Al-Tamimi and W. Shen, Angew. Chem., Int.
Ed., 2012, 51, 5497–5501.
19 M. Li, W. L. Then, L. Li and W. Shen, Anal. Bioanal. Chem.,
2014, 406, 669–677.
20 J. Gantelius, C. Hamsten, M. Neiman, J. M. Schwenk, A.
Persson and H. Andersson-Svahn, J. Microbiol. Methods,
2010, 82,11–18.
21 J. Gantelius, T. Bass, R. Sjoberg, P. Nilsson and H.
Andersson-Svahn, Int. J. Mol. Sci., 2011, 12, 7748–7759.
22 J. Li and J. Macdonald, Biosens. Bioelectron., 2014, 64C,
196–211.
23 A. James and J. Macdonald, Expert Rev. Mol. Diagn.,
2015, 15, 1475–1489.
24 J. Li, F. Rossignol and J. Macdonald, Lab Chip, 2015, 15,
2538–2558.
Lab on a Chip Communication