Two Orders of Magnitude Improvement in Detection Limit of Lateral

Subscriber access provided by BRITISH COLUMBIA INST OF TECH

Article

Two orders of magnitude improvement in detection limit of lateral flow assays using isotachophoresis Babak Yaghoubi Moghadam, Kelly Theresa Connelly, and Jonathan D Posner Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/ac504552r • Publication Date (Web): 15 Dec 2014 Downloaded from http://pubs.acs.org on January 5, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

2

Two orders of magnitude improvement in detection limit of lateral flow assays using isotachophoresis

3

Babak Y. Moghadam†, Kelly T. Connelly‡, Jonathan D. Posner†*

1

4 5

†University of Washington, Mechanical Engineering Department, Seattle, WA 98195, USA

6 7

‡University of California, Los Angeles, Mechanical and Aerospace Engineering Department, Los Angeles, CA 90095, USA

8 9

Abstract

10

Lateral flow immunoassays (LFA) are one of the most prevalent point-of-care (POC)

11

diagnostics due to their simplicity, low cost, and robust operation. A common criticism of LFA

12

tests is that they have poor detection limits compared to analytical techniques, like ELISA, which

13

confines their application as a diagnostic tool. The low detection limit of LFA and associated

14

long equilibration times is due to kinetically limited surface reactions that result from low target

15

concentration. Here we use isotachophoresis (ITP), a powerful electrokinetic preconcentration

16

and separation technique, to focus target analytes into a thin band and transport them to the LFA

17

capture line resulting is a dramatic increase in the surface reaction rate and equilibrium binding.

18

We show that ITP is able to improve limit of detection (LoD) of LFA by 400-fold for 90 second

19

assay time and by 160-fold for a longer 5 minutes time scale. ITP-enhanced LFA (ITP-LF) also

20

shows up to 30% target extraction from 100 µL of the sample, while conventional LFA captures

21

less than 1% of the target. ITP improves LoD of LFA to the level of some lab based

22

immunoassays, such as ELISA, and may provide sufficient analytical sensitivity for application

23

to a broader range of analytes and diseases that require higher sensitivity and lower detection

24

limits.

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

1. Introduction

2

The emphasis of medical care is shifting toward prevention and early detection of disease and

3

management of multiple chronic conditions resulting in higher prevalence of point-of-care

4

(POC) testing.1 Introduced in 1987, lateral flow immunoassay (LFA) are one of the most

5

commercial available rapid POC tests.2–5 POC lateral flow-based tests have enjoyed commercial

6

success because of their simplicity, low cost, and robust operation. LFA tests use a paper

7

substrate, e.g. porous membranes, to indicate presence of target analytes. In a conventional LFA,

8

a target analyte binds to a labeled detection protein, e.g. antibody labeled with colloidal gold, and

9

travels across the membrane by capillary action, without user’s manipulation. At the test zone,

10

the target binds with a specific capture molecule immobilized on the surface which accumulates

11

the captured target and generates a visually detectable, i.e. colorimetric, signal.

12

LFA tests have been indicated as a suitable medical diagnostic tool for remote or impoverished

13

settings and are also used as an appropriate technology for a wide variety of field applications.6

14

However, standard LFAs, which are typically simple colorimetric assays, have relatively poor

15

detection limits (~10-11 M)7,8 compared to fluorescence-based LFAs (~10-10 – 10-14 M)9 and

16

analytical methods like ELISA and SPR (~10-12 – 10-18 M).10 In many diagnostic situations,

17

especially at the early stages of a disease, protein markers, such as pathogen antigens or host

18

antibodies, are present at very low concentrations. The serum concentrations of the majority of

19

proteins important in cancer,11 neurological disorders12,13 and the early stages of infection

20

disease14 are reported to range from 10−12 to 10−16 M. LFA tests typically fail to provide a

21

reliable and consistent detection at these target concentrations while lab-based analytical

22

techniques, like ELISA, can routinely detect at these levels and have been shown to operate at

23

concentrations as low as 10-19 M.15 Moreover, LFA tests are often single step assays and are


Page 2 of 33

Page 3 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

specific to antibody-antigen interaction which limits their application for integrated sample

2

preparation (e.g. such is required for nucleic acid purification) and a narrow group of samples.

3

The vast majority of work toward improving detection limits of LFAs has been focused on

4

optimizing affinity of the antibodies used in the test,16,17 and introducing new labeling

5

techniques.18–20 In optimal physiochemical conditions (pH, ionic strength, and temperature) the

6

affinity, Kd, of a specific antibody-antigen pair has a value that is constant and to a large extent

7

influences the analytical sensitivity of the test.6 One useful metric for comparing detection limit

8

of different immunoassays is the initial target concentration CT0 normalized by the affinity

9

constant Kd = koff/kon, where koff and kon are respectively the off- and on- rate constants of the

10

binding reaction. Using normalized detection limit (CT0/Kd) to compare analytical sensitivity of

11

two LFA tests provides a comparison that is more independent of target-capture reagent

12

biochemistry. Fernandez et al.21 reported the lowest detection limit for standards LFAs for

13

prostate specific antibody (PSA) detection as 1 µg/l or ~10-11 M.8 Normalized detection limit

14

concentration of this assay therefore can be calculated as ~10-2 based on the reported KD value of

15

PSA antibodies in the range of 10-9-10-10.22 For analytical detection methods like ELISA the

16

normalized detection limit drops to the range of 10-5 to 10-10 which suggests lower concentration

17

detection limits.

18

Much effort has been focused on decreasing Kd values of secondary antibodies, in the range of

19

10-6 to 10-10,23 to reduce the concentration detection limits of immunoassay, including LFA tests.

20

Alternatively, the detection limit of a LFA test can be improved by a factor of nearly 10-fold by

21

using different labels and labeling techniques.18,19 However, to improve labeling, usually a high

22

concentration of the labeling reagent is necessary which can increase final cost of the assay.

23

Moreover, conjugating the antibodies with larger labeling molecules can lead to a change or even



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

a complete loss of affinity due to the effect of steric restrictions.6 Recently, there has been a

2

push to enhance the signal in LFAs by developing paper-based microfluidic devices that

3

incorporate sophisticated functions and device designs using capillary action to drive the flow.24–

4

26

5

detection limit improvement reported in these studies is limited to about 3-fold. Furthermore,

6

these methods add more complexity and cost to the final product which may limit their

7

application in POC settings.

In spite their success in integrating multiple functions and assays on a single paper device,

8

Improving reagent affinity, labeling techniques, and complex device features have not alone

9

addressed the challenge of detecting low target concentrations (10−12 to 10−16 M) observed in

10

diseases of interest. At these target concentrations the Kd for a LFA would need to be reduced to

11

a value near that of the strongest affinity of 10-15 reported for streptavidin-biotin interaction27,28

12

to reach a CT/Kd of 10-2. Surface binding of the antibody-antigen in LFAs can be regarded as a

13

second-order reaction where the binding rate depends on the reaction rate constants and

14

concentration of the reactants. At low target concentrations the forward reaction rate is

15

decreased, i.e. reaction is kinetically limited, and as a result it takes impractically long reaction

16

times for the assay to equilibrate and generate a detectable signal. Increasing the surface

17

concentration of the immobilized capturing reagents can improve the forward reaction rate, but is

18

ultimately limited by the protein binding capacity of the membrane. Therefore, to improve the

19

detection limits and slow binding rates in an immunoassay, preconcentration of the target

20

becomes a necessary step.29–31 A number of strategies are currently available to improve

21

detection limits of the surface reaction in biosensors using preconcentration, including solid-

22

phase extraction,32,33 electrokinetic techniques,34,35 as well as chromatographic and membrane

23

preconcentration.36–38 Electrokinetic techniques address the target preconcentration with minimal


Page 4 of 33

Page 5 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

requirements and pretreatment of the sample. Isotachophoresis (ITP) preconcentration has been

2

recently used to accelerate and improve detection limits of surface DNA hybridization assays. In

3

these studies surface-based reaction in microchannels was accelerated by preconcentrating the

4

target analytes and transporting them to the test zone which considerably improved the assay

5

detection limits.39,40

6

ITP is a powerful electrokinetic technique used widely in capillaries and microchannels for the

7

preconcentration, separation, and purification of ionic analytes.41 Under an applied electric field

8

during ITP, target analytes transport electrophoretically and accumulate at the boundary of fast

9

moving leading (LE) and slow moving trailing electrolyte (TE) ions, resulting in substantial

10

increase in analyte concentration in a narrow plug, on the order of 100 μm wide. ITP in

11

microchannels has demonstrated up to a million-fold increase in target concentration during the

12

assay time of less than 30 minutes.42 We previously showed that ITP can increase the width-

13

averaged concentration of target analytes in nitrocellulose membrane strip by up to 900-fold in

14

less than 5 minutes.43,44 We also demonstrated that ITP on nitrocellulose can extract more than

15

80% of the target from initial sample, is compatible with relatively large sample volumes (>100

16

µL) and can be powered by a portable power supply showing its potential to be used in POC

17

devices. It has been shown that ITP can enhance the surface DNA microarray hybridization

18

reaction between a suspended target and an immobilized probe in conventional microfluidic

19

devices. Han et.al demonstrated the acceleration and sensitivity improvement of DNA

20

microarray hybridization by preconcentrating the target molecules and transport them over the

21

immobilized probe site using ITP.40 Their method leverages preconcentration of ITP to

22

overcome the slow reaction kinetics of surface hybridization. Their analytical predictions along



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

with experimental observations show 8-fold improvement in the detection limit at about 30-fold

2

shorter assay time.

3

In this paper, we integrate ITP into a model conventional LFA to show how target

4

preconcentration can significantly improve limit of detection (LoD) of LFAs. ITP increases

5

concentration of the target molecules and transports them in a thin band to the test zone which

6

results in an increase in kinetic reaction rate of the target and the probe and more than two orders

7

of magnitude improvement in the LoD. To achieve this goal, we construct a LFA on

8

nitrocellulose membrane with labeled secondary antibody IgG as the target and compared its

9

detection limits to the same assay enhanced by ITP preconcentration. We do not make specific

10

claims about the absolute reduction in LoD in terms of concentration which may have been

11

achieved with existing LFA tests that use optimized reagents, paper substrate, spotting, labels

12

etc. The goal of this work is not to achieve the lowest LoD concentration reported, but to show

13

how ITP can improve detection limit of LFA through target preconcentration by directly

14

comparing the ITP enhanced-LFA (ITP-LF) to a conventional LFA.

15

We use the analytical model, developed by Han et al.,40 for the ITP-enhanced second

16

order surface reaction and apply it to our system. We perform quantitative fluorescence ITP-

17

enhanced lateral flow (ITP-LF) assay on nitrocellulose membranes and compare its performance

18

to the conventional LFA. We also use antibodies labeled with colloidal gold as our target to

19

investigate compatibility of ITP-LF with colorimetric detection, which is used in conventional

20

LFA tests for visual interpretation of the test results. Finally, we demonstrate an improvement of

21

detection limit in mouse serum showing that ITP can extract and preconcentrate targets present

22

in biological media.


Page 6 of 33

Page 7 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1


2. Theoretical model

2

The basic principle of any immunochemical technique, including LFA immunoassays, is that a

3

specific target, e.g. an antigen, binds with its specific probe, e.g. a secondary antibody, to give an

4

exclusive target-probe complex. The specific binding of antibodies is dependent on hydrogen

5

bonds, hydrophobic interactions, electrostatic forces, and van der Waals forces, therefore it is a

6

reversible process. For a single probe (P) binding site that binds to a single target (T) and

7

forming a complex (C), the surface reaction can be modeled as a second-order reaction with on-

8

and off- rate constants of respectively kon and koff, and the dissociation constant of KD = koff /

9

kon.45

10

By assuming perfect mixing conditions (where the reaction is kinetically limited), and

11

negligible consumption of the target molecules at the test zone, one can develop a relation for

12

this reaction. In this regime, target concentration above the immobilized probe remains

13

unchanged from its initial value, CT0.40,46,47

14

hLFA 





CC C*  * 0 1  exp   C0*  1 koff t C p 0 C0  1



(1)

15

Here hLFA is the fraction of bound probe,

16

bound probe and of the probe at t = 0. C0* is defined as normalized initial target concentration,

17

CT0 /KD where CT0 is the volumetric molar concentration of the suspended target in the immediate

18

vicinity on the immobilized probe in the test zone. We verified the validity of kinetic, reaction

19

limitation assumption using non-dimensional parameters introduced by Pappaert et al.48 and Han

20

et al.40 for a second order surface reaction in capillaries. Damkohler number, Da, which is

21

defined as the ratio of maximal forward reaction rate to maximal normal diffusion rate can be

and

are molar surface concentrations of the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 33

1

used to determine if the system is reaction- or diffusion-limited. Da for our assay can be

2

calculated as ~10-4 using Da  konCP 0 d / D , where D is the effective diffusivity of IgG in

3

nitrocellulose (2×10-10 m2/s),23 and d is the pore radius (5 µm) which is an appropriate

4

characteristic length scale to determine diffusion of the target molecules to the surface.40,48 Very

5

low value of Da effectively means that the time scale for reaction is long compared to that of

6

diffusion. This confirms that the target binding is limited by the reaction kinetics rather than the

7

diffusion of the target molecules to the surface. Da number for LFA is much lower than capillary

8

channels therefore there is no need to calculate other four non-dimensional parameters described

9

in Pappaert et al. work. 48

10

ITP-LF assay can be similarly modeled as a kinetically limited reaction at an elevated target

11

concentration resulted from the ITP preconcentration. Han et al. showed that by modeling the

12

Gaussian concentration profile of the ITP-focused sample as a top-hat pulse with characteristic

13

ITP zone width of δITP = ±2σ and target concentration of pCT0, where p is the effective

14

preconcentration ratio and σ is the standard deviation of the sample Gaussian distribution, an

15

approximate analytical model for ITP enhanced surface binding can be derived as,40

16

hITP  LF





pC0*  1  exp   pC0*  1 koff tITP * pC0  1



(2)

17

where hITP-LF is the fraction of bound probe in ITP-LF. They derived this equation by substituting

18

pCT0 for CT0 and tITP for t in equation (1). The ITP plug migrates at a constant velocity of VITP,

19

therefore each point at the test zone is exposed to the highly focused sample for tITP   ITP VITP .

20

p is calculated such that the area under the top-hat profile, pCT0 δITP, is the same as the area under

21

the Gaussian distribution of the target in the ITP plug,

2 A ITP / 4 , where A is the maximum


Page 9 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

value. We use equations (1-2) to compare the signal strength, LoD and assay time of the ITP-LF

2

and conventional LFA. This theory gives us an insight to better understand the design and

3

optimization of ITP-enhanced LFAs.

4

We similarly use a reaction limited assumption here, which means that the concentration

5

of the target in the ITP zone stays constant above the probe area at all times. The target is being

6

transported to the capture line by electrophoresis in a single tightly focused plug. Once the target

7

is transported to the capture line the target must diffuse across the pore to react at the surface of

8

the membrane. This process continues as the capture reagent consumes the target in the plug over

9

time. To sustain the reaction, the target must diffuse across the pore. We can assume that the

10

ITP-LF is kinetically limited because the typical ITP's advective residence time, i.e. time during

11

which probe area is exposed to the ITP zone, is tITP = 284 sec. This is much longer than the

12

diffusion time scale for the target to diffuse to the probe surface in nitrocellulose membrane,

13

which can be estimated as tdiff = 0.32 sec using tdiff  d 2 D where d is the pore diameter of the

14

membrane (10 µm) and D is the diffusivity of IgG in nitrocellulose membrane.23 ITP is also

15

known to cause secondary flows which help mitigate diffusion-limited regimes.49 Effectively,

16

the longest time scale is the reaction time scale (~103), followed by the advective residence time

17

scale of the ITP plug (~102), and followed by diffusion (~10-1).

18

3. Material and methods

19

3.1. Reagents and the electrolytes system

20

We performed a series of quantitative and qualitative ITP experiments on nitrocellulose

21

membranes demonstrating the improvement of LFA LoD using ITP. We used IgG secondary

22

antibodies as the target and capture reagents since they are extensively used in LFAs as the



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 33

1

capturing and labeling reagents. Moreover, these antibodies have high binding affinity for

2

nitrocellulose membrane therefore higher surface concentration of the immobilized probe can be

3

achieved which improves the capturing. For target analyte we used goat anti-rabbit IgG labeled

4

with Alexa Fluor 488 fluorescence dye (CAS# 2628-2-8, Molecular Probes, Eugene, OR), for

5

quantitative detection, and goat anti-mouse IgG labeled with 40 nm colloidal gold (Arista

6

Biologicals, Allentown, PA) for colorimetric detection. We used bovine anti-goat IgG (Jackson

7

ImmunoResearch Laboratories, West Grove, PA) as the capturing reagent, i.e. probe, since it

8

recognizes and binds with the goat IgG and has minimal cross reactivity for bovine albumin

9

serum (BSA) that we used to block the membrane.

10

To design the ITP electrolyte system we first measured the effective electrophoretic mobility

11

of the target IgG in a wide range of pH (data shown in SI). IgG antibodies are large molecules of

12

about 150 kDa thus they show low electrophoretic mobilities especially around the neutral pH,

13

due to their isoelectric point. Based on the measured effective mobility of IgG as 6.3 nm2/Vs (at

14

pH 7.4) we designed the TE composition as 10 mM Glycine, with the effective mobility of 0.4

15

nm2/Vs, buffered with Bis-Tris to pH 7.4 (σ = 3 μS/cm). At this pH the binding reaction occurs

16

at a biologically relevant condition and the difference between the mobilities of target and TE

17

ions is maximized, which results in more potent sample focusing. We used semi-finite sample

18

injection by mixing the TE with target analyte at different concentrations. The LE contained 40

19

mM HCl buffered with Tris to pH 8.1 to maintain high buffering capacity of the Tris (σ = 4.2

20

mS/cm). The conductivity, σ, ratio of the LE and TE was maximized to 1400 to achieve the

21

highest preconcentration while minimizing the Joule heating. We added 0.5% PVP to the LE to

22

suppress electroosmotic flow. For a detailed discussion on the choice of electrolyte system in

23

ITP, refer to the works of Bagha et al. and Moghadam et al.43,50 All the reagents were purchased


Page 11 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

from Sigma Aldrich (St. Louis, MO) unless mentioned otherwise. All aqueous samples were

2

prepared using water ultrapurified with a Milli-Q Advantage A10 system (Millipore Corp.,

3

Billerica, MA).

4

3.2. Paper device preparation

5

We used backed nitrocellulose membrane (HF-135, Millipore Corp., Billerica, MA) cut with a

6

CO2 laser ablator (Universal Laser Systems, Scottsdale, AZ) to fabricate the immunoassay

7

devices used in both LFA and ITP-LF experiments. Selected membrane has the capillary flow

8

rate of 0.3 mm/s and pore diameter of ~10 µm which provides long reaction time for

9

conventional LFA while offering an optimum internal surface area for maximized amount of

10

immobilized probe on the surface. Paper device dimensions were chosen as 40 mm × 3.5 mm to

11

maximize the target accumulation in the ITP zone while reducing the Joule heating caused by

12

high electrical resistance of the device.43 We immobilized the capture reagent on the membrane

13

by using a custom-made protein spotting system.51 To maintain physiological condition, capture

14

IgG was suspended in 0.25 M NaCl and 0.01 M sodium phosphate, buffered to pH 7.0. This pH

15

is close to the isoelectric point of the IgG, which helps to destabilize the probes in aqueous form

16

therefore they bind more effectively with the membrane surface. The test line was located one-

17

third of the device length away from the LE side where higher sample preconcentration ratio is

18

reached. We spotted the test line with the width of approximately 800 µm and measured probe

19

surface density of CP0 = 2×10-9 moles per internal surface area of the membrane (0.5 fmole of

20

probe per test line).51 Thickness of the test line was chosen to be in the same order of the ITP

21

plug width, δITP, to increase the surface interaction and CP0 was chosen close to the IgG binding

22

capacity of the nitrocellulose reported by Fridley et al.’s measurements.23 After spotting, the

23

membranes were allowed to dry in room temperature for 5 minutes then further dried at 37o C for



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

1 hour. To minimize the non-specific interaction of the target antibody with the membrane, we

2

used 1% BSA as the non-reactive blocking agent.52

3

3.3. ITP-LF and LFA protocol

Page 12 of 33

4

Figure 1A shows our ITP experimental setup where the paper device is placed in an acrylic

5

holder with the test zone closer to the LE reservoir. Before each ITP experiment the membrane

6

was hydrated with the LE and placed on the holder with the left and right reservoirs filled

7

respectively with 100 µL of the TE, mixed with the sample, and LE. To apply electric field

8

across the membrane we placed the positive and negative platinum wire electrodes in the LE and

9

TE reservoirs, respectively. We initiated the ITP-LF by applying 500 µA constant current to the

10

TE reservoir and grounding the LE using a high voltage power supply (HSV488 6000D,

11

LabSmith Inc., Livermore, VT). Higher currents result in a higher preconcentration ratio and a

12

shorter assay time,43,53 but lower the signal strength at the test zone because the ITP zone

13

residence time over the test line is decreased. Therefore, as the ITP zone arrives at the test zone

14

we reduced the current to 50 µA which reduces the ITP velocity thus increases the reaction time.

15

Plug velocity, and therefore the ITP zone residence time, is inversely proportional to the applied

16

current.53 For example, in our system, tITP = 9 sec at 500 µA, and tITP = 284 sec at 50 µA. After

17

the ITP plug crosses the test zone we increased the current back to 500 µA until the plug reaches

18

the LE reservoir. Progression of the ITP plug downstream subsequently acts as an electrokinetic

19

wash step where unbound targets are removed from the probe spots and collected at the

20

migrating ITP plug so no further washing step is required.39,40 For the conditions mentioned

21

above, each ITP experiment takes about 7 minutes to complete.


Page 13 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

For conventional LFAs we used a holder with one reservoir, as shown in Figure 1B. We placed

2

one end of the membrane in the reservoir and the other end on a cellulose absorbent pad (HF-

3

125, Millipore Corp., Billerica, MA) to wick the sample during the assay. To initiate the LFA we

4

filled the reservoir with the target solution mixed with TE which triggers the capillary action and

5

movement of sample across the membrane. The absorbent pad continuously wicks the sample

6

allowing constant flow of the target analytes during the experiment therefore target molecules

7

interact with the immobilized probe throughout the duration of the test. We performed a wash

8

step at the end of each LFA to improve the SNR by removing the unbound targets from the test

9

area. We replaced the target-TE solution with TE and let the wash step to continue for 4 minutes

10

which we found it to be sufficient for effective removal of the unbound targets. We explored

11

longer wash steps and found no significant improvement of the SNR or LoD.

12

3.4. Detection and quantification

13

We performed quantitative fluorescence experiments using a Nikon AZ100 upright microscope

14

equipped with a 0.5x objective (Plan, NA 0.5, Nikon Corporation, Tokyo, Japan), an

15

epifluorescence filter cube (488 nm excitation, 518 nm emission, Omega Optics, Brattleboro,

16

VT), and a 16-bit, cooled CCD camera (Cascade 512B, Photometrics, Tucson, AZ) which

17

capture images at the exposure time of 1 s. Images taken from the test zone after each experiment

18

were processed using an in-house MATAB code to quantify the amount of target captured in the

19

test area, mT, effective stacking ratio, p, and ITP residence time, tITP. We calculated both the

20

mean and standard deviation of the background, determined from a cropped image of the device

21

excluding the test line. The mean-subtracted intensity of each pixel in the test area was then

22

counted as the signal if it was greater than 3 standard deviations of the background, i.e. signal to

23

noise ratio (SNR) of 3. To convert the signal intensity values to the target concentration, we



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 33

1

generated a calibration curve by performing three titration experiments. In each titration

2

experiment we measured fluorescence intensity of the membrane surface fully wetted by a

3

known high concentration of the target solution and generated a calibration curve from a linear

4

regression (example is shown in the SI). The calibration experiments were done after performing

5

each set of experiments, on multiple days, and we found the results to be very consistent and

6

repeatable.

7

4. Results and discussion

8

4.1. Demonstration of ITP enhanced LFA

9

After applying the electric field to the system, a highly focused plug of target molecules

10

electromigrates to the test zone where they react with capture molecules immobilized on the

11

surface. ITP increases the target concentration and interaction kinetics of the target-probe which

12

results in generation of a specific fluorescence signal at the test zone. In Figure 1C, we present

13

five snapshots of a single ITP-LF experiment showing successful capturing of the target using

14

ITP (an example video of this experiment is provided in the SI). In snapshots a-c, an ITP zone

15

containing highly concentrated target forms and migrates across the membrane. As the ITP zone

16

travels downstream it sharpens and its fluorescence intensity increases demonstrating that the

17

target concentration is increasing. As the plug approaches the test zone, we decrease the applied

18

current and the slow moving plug sweeps over the test zone and binds with the immobilized

19

probes. After the ITP zone passed the test zone (snapshots d-e), we observe a fluorescence signal

20

at the test zone indicating that the target molecules have been captured by the immobilized

21

probes. Any unreacted target molecules electromigrate towards the LE reservoir which

22

effectively serves as a membrane wash step.


Page 15 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1


4.2. Experimental determination of the kinetic and focusing parameters

2

Kinetic parameters, KD and koff, as well as the ITP focusing parameters, p and tITP, need to be

3

determined in order to compare the experimental results with the model predictions described in

4

equations 1 and 2. To obtain the kinetic parameters we performed a conventional LFA

5

experiment using 25 mg/l initial target concentration and monitored the signal over time. Figure

6

S-3 of the SI shows the captured amount of target in LFA as a function of time. We fit Equation

7

2 to these data using the least square method using KD and koff as the fitting parameters. The

8

fitting parameters were determined as, koff = 1.75×10-3 s-1 and KD = 1.42×10-6 M which are

9

consistent with the range of values reported for the interaction of secondary antibodies.54,55

10

We obtained the ITP focusing parameters, p and tITP, by conducting independent ITP

11

experiments. We performed ITP experiments and measured the ITP zone velocity, VITP, width of

12

the ITP plug, δITP, and the effective preconcentration ratio, p. For each ITP experiment we

13

obtained images of the ITP plug prior to reaching the test zone, then converted the intensity

14

values to target concentration using our calibration curve. We fit a Gaussian distribution in the

15

form of A exp(( x   )2 2 2 ) to the resulting concentration profile.53 The resulting fitting

16

parameters are the maximum plug concentration A, mean µ, and the plug width δITP = ±2σ (two

17

times the plug standard deviation). We experimentally estimated the effective ITP

18

preconcentration ratio of p = 92 ± 12 for I = 500 µA using a method described earlier. The ITP

19

velocity was measured by dividing the average displacement distance of the Gaussian fit

20

obtained from twenty pairs of images taken with 1 second image-to-image delays. By dividing

21

the value of VITP to δITP we obtained the characteristic ITP reaction time as tITP = 9.4 ± 0.74 s for

22

I = 500 µA, and tITP = 284 ± 47 s for I = 50 µA. The plus/minus values represent 95% confidence



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

interval obtained from 3 measurements. We used the later value for our model since the current

2

is set to 50 µA when the ITP zone is crossing the test zone in ITP-LF.

3

4.3. Capture Enhancement by ITP

Page 16 of 33

4

Figure 2 shows experimental data for the fraction of bound probe h as a function of target

5

concentration for ITP-LF and LFA. Assuming one-to-one binding of target and probe, the

6

fraction of bound probe is defined as the surface concentration of the bound probe, CC,

7

normalized by the initial surface concentration of the probe, CP0. We estimated the fraction of

8

bound probe by normalizing the amount of bound target in the test zone with its maximum value.

9

Each ITP-LF takes about 7 minutes to complete with the ITP zone residence time of 284 s. LFA

10

data were collected at three different assay times of 90 seconds, 5 minutes and 1 hour. h

11

increases with initial target concentration for both LFA and ITP-LF. The amount of captured

12

sample in ITP-LF increases with increasing target concentration, however at concentrations

13

higher than 25 mg/l we did no observe a change in the amount of captured target which suggests

14

that the probe surface was saturated. At each concentration the fraction of bound probe using ITP

15

is higher than the conventional LFA since the ITP preconcentration increases the reaction rate.

16

For example, at 10 mg/l the fraction of bound probe is 6.7×10-1 for ITP-LF and is 8.53×10-3 for

17

LFA after 5 min. This shows improved capture by ITP preconcentration by about 80 fold. It

18

should be noted that according to the Equation 1 hLFA increases with time and reaches its

19

maximum value, C0* (C0*  1) , after long reaction times, i.e. equilibrium conditions. The

20

equilibrium time depends on the initial concentration of the target, for example, our model

21

predicts that LFA reaches near equilibrium condition, 99% of the maximum signal, after 1 hour

22

for CT0 = 10 mg/l.


Page 17 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

Along with the experimental data in Figure 2 we present analytical models (equations 1-2) for

2

ITP-LF and conventional LFA. To obtain these theoretical curves, we used the experimentally

3

determined kinetic and focusing parameters as koff = 1.75×10-3 s-1, KD = 1.42 µM, p = 92 and tITP

4

= 284 s. For LFA, we observe good agreement of predicted trends and our experimental data. For

5

ITP-LF, we notice good agreement at higher concentrations but lower than predicted binding at

6

0.05 mg/l. We hypothesize that this over prediction of h by the model at low concentration may

7

be due to deviation from the kinetically limited assumption as a result of very low number of

8

moles of target present above the reaction zone. The number of moles of target in the ITP zone

9

here is approximately 8 fmoles. At this condition, roughly one-third of the target molecules have

10

been consumed by the surface reaction, therefore the assumption of constant target concentration

11

may not be valid.

12

LoD in our system is defined as the lowest target concentration at which the maximum

13

background-subtracted intensity in the test zone is 3 standard deviations away from the

14

background, i.e. SNR = 3.56 In practice, for example, the normalized standard deviation for ITP-

15

LF is ~ 8×10-4 and the fraction of bound target, h, corresponding to LoD is ~ 5.4×10-3 for ITP-LF

16

and ~ 10-2 for LFA. The bound target concentration of ITP-LF at LoD is marginally lower

17

because the electrokinetic “wash” step during ITP effectively removes the unbound target from

18

the surface and improves the SNR. Table 1 lists the LOD extracted from Figure 2 for our ITP-LF

19

and LFAs for different reaction times. This table shows that the LoD of our LFA is improved by

20

more than two orders of magnitude, i.e. 160, and 400, for assay times of 5 minutes, and 90

21

seconds, respectively. Moreover, the normalized LoD, CT0/KD, achieved for ITP-LF can be

22

calculated as 3.3×10-4 which is close to the LoD range reported for lab-based analytical

23

techniques like ELISA.22 We are not making specific claims about the absolute reduction in



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 33

1

LoD in terms of concentration which may have been achieved with existing LFA tests that use

2

optimized reagents, paper substrate, spotting, labels etc. For example, the work of Fernandez et

3

al. showed an LoD of 10-11 M for a LFA which used customized PSA antibodies as capturing

4

reagents at 1000 times higher concentrations of capture reagens, 1000 times higher gold labeled

5

antibodies, and an additional washing step. Our LFA LoD is 10-8 and we improve it to 10-10 by

6

integrating ITP.

7

significant optimization of the LFA must be completed similar to the efforts by Fernandez and

8

coworkers. These optimized LFA can then be integrated to with ITP to further reduce the LoD

9

as we have demonstrated here.

We recognize that to achieve the lowest possible LoD for a specific target,

10

In addition to improving the detection limits, ITP increases the capturing efficiency of LFAs.

11

In Table 2 we compare the capturing ratio of ITP-LF,  NC NT 0 ITP LF , for 7 minutes assay time

12

with that of 5 minutes LFA time,  NC NT 0 LFA , as a function of initial target concentrations, CT0.

13

Capturing ratio is calculated by normalizing moles of target captured in the test area, NC, by the

14

initial moles of target in the sample reservoir, NT0. Only a fraction, i.e. 18 μl, of the original 100

15

μl initial sample volume is consumed for the case 5 minutes conventional LFA time while this

16

fraction is negligible for ITP-LF because we effectively extract the target from the liquid sample

17

volume. Capture ratio for LFA increases by increasing the initial target concentration and

18

reaches 0.7 % at CT0 = 25 mg/l. Similar trend is observed for ITP-LF with the highest capture

19

ratio of 30 % observed at CT0 = 0.5 mg/l. An apparent decrease in the capturing ratio of ITP-LF

20

is observed at higher concentrations because the test line is saturated due to having excess

21

amount of target molecules in the sample solution, i.e. NT 0 is larger than the initial moles of

22

probe, NP0. Unbound target molecules during ITP-LF travel with the ITP plug toward the LE


Page 19 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

reservoir as shown in Figure 1C-d. This suggests that we may be able to increase the capturing

2

efficiency of ITP-LF by using even smaller volumes of the sample. LFA shows capturing ratios

3

smaller than 1% and we are not able to detect the target at concentrations lower than 8 mg/l

4

which is below the LoD of LFA. At each concentration, capturing ratio of the conventional ITP-

5

LF is one order of magnitude larger than the capturing ratio of LFA. This data suggests that

6

integration of ITP with LFA may offer the capability in using large sample volumes with very

7

dilute concentration of target.

8

4.4. Colorimetric detection in ITP-LF

9

The capability of LFAs to generate an unambiguous and visually read result, i.e. colored band,

10

is critical feature of their success as a POC diagnostic. This is achieved by using detection

11

antibodies labeled with colloidal gold or other colorimetric signal generators such as colored

12

latex particles,57 and colloidal carbon nanoparticles.58 In order to show the compatibility of our

13

ITP-LF with colorimetric detection, we used IgG labeled with 40 nm colloidal gold (IgG-Au) as

14

our target. We measured effective electrophoretic mobility of IgG-Au at pH 7.4 as 8.9 nm2/Vs.

15

Since this value is larger than the effective mobility of TE ions we used in our quantitative

16

analysis, we kept the same TE composition to perform colorimetric ITP-LF experiments.

17

Generation of a pink-colored band at the test zone indicates successful capturing of the target

18

with the color intensity directly proportional to the amount of sample captured. ITP is able to

19

focus IgG molecules labeled with gold nanoparticles and transport them across the membrane

20

toward the test zone (a video of colorimetric ITP-LF is provided in the SI).

21

snapshots of the paper device used for colorimetric detection of the target using both ITP-LF and

22

conventional LFA for the initial target concentration ranging from 0.1 – 15 mg/l. The optical

23

density of the test line increases with initial target concentration as expected.


Figure 3 shows

Qualitatively, a


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 33

1

visually detectable signal was observed using ITP-LF at concentrations as low as 0.1 mg/l while

2

no signal was detected using LFA at concentrations lower than 10 mg/l and 10 minutes assay

3

time. Qualitatively, these data show more than 100-fold improvement in detection limit of LFA

4

using ITP which is consistent with our quantitative fluorescence measurements.

5

4.5. Insights into the design of an ITP-LF

6

We use our analytical model as a guide for the design of an ITP-enhanced LFAs. In Figure 4

 LoDLFA

LoDITP LF  versus the assay time for different effective

7

we present the LoD ratio

8

preconcentration ratios and off-rate constants. Each value on the y-axis shows the improvement

9

in LoD when using ITP at a given LFA time. ITP preconcentration results in a considerable

10

improvement in LoD of the LFA due to the elevated forward reaction rate as a result of higher

11

target concentration in the test zone. LoD ratio decreases at longer times since the binding of

12

target and probe in kinetically limited LFA increases with time. ITP improves the LoD of LFA

13

even at longer times when the LFA reaches its highest signal at near equilibrium levels. ITP

14

increases target concentration over the capture line allowing it to reach higher equilibrium

15

binding than are achieved for long times in the lower concentrations that are experienced in

16

conventional LFAs. Our model suggest that the LoD ratio scales with ptITP koff which is greater

17

than 1 for P > 2. For example, at p = 100, model predicts that ITP improves the LoD of LFA by a

18

factor of about 40 at 1 hour assay times and longer. Figure 4 shows that for shorter assay time

19

much higher LoD ratios can be obtained which means, in addition to the LoD improvement, ITP

20

can speed up the LFA reaction. In the SI we provide a discussion on increasing the reaction rate

21

using ITP as was described in the work of Han et.al for DNA hybridization in microchannels.40

22

Closed symbols in Figure 4 represent three experimental conditions where for ITP-LF CT0 = 0.05


Page 21 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

mg/l, p = 92, tITP = 284 s and for LFA CT0 = 20, 8 and 3 mg/l for 90 second, 5 minute and 1 hour

2

assay times, respectively. Model prediction shows good agreement with the experimental

3

observation. We hypothesize that the under prediction by the model at 1 hour assay time is due to

4

the fact that at very low target concentrations, i.e. 3 mg/l, the LFA experiments may be limited

5

by the diffusion of the target molecules to the probe surface therefore it requires longer than 1

6

hour to reach its equilibrium signal. At this low concentration, the reaction limited assumption

7

may be violated which results in an under under-prediction of LoD ratio. Figure 4 also shows

8

that at higher preconcentration ratios, e.g. p = 500, higher LoD ratios can be achieved, as

9

expected.

10

We also examined effect of kinetic parameters and observed that the koff has marginal

11

improvement effect on the LoD. For example, at 30 min assay time, increasing the koff by one

12

order of magnitude increases the LoD ratio by less than 2-fold. When the koff is increased the kon

13

value also increases, at constant KD, therefore the forward reaction rate is increased. We

14

observed that LoD ratio does not depend on the KD value because ITP improves the binding by

15

increasing the forward reaction rate. The equilibrium dissociation constant is a representation of

16

the affinity of two compounds and does not depend on time. For conventional LFA, the fraction

17

of the bound probe at longer times is directly proportional to the initial target concentration, hLFA

18

≈ CT0/KD for conditions satisfying C0* ≪ 1. Similarly, in ITP-LF, a Taylor series expansion for

19

the limiting case of (pC0* + 1)kofftITP ≪ 1 yields the linear proportionality of hITP ≈ pkofftITP (CT0 /

20

KD) .40 At the LoD concentration, signal strength, i.e. fraction of the bound probe, is the same for

21

LFA and ITP-LF. Therefore, LoDLFA ≈ hLoDKD and LoDITP-LF ≈ hLoD /pkontITP. As a result, the LoD

22

ratio scales as pkofftITP which confirms our model predictions presented in Figure 4 that the LoD



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

ratio is directly proportional to the ITP parameter ptITP and the off-rate constant while it is

2

independent of the KD value.

Page 22 of 33

3

Our scaling analysis and the analytical model show that the amount of captured target for

4

conventional LFA depends on the normalized initial target concentration, i.e. CT0/KD. On the

5

other hand, for ITP-LF the amount of captured target depends on pkofftITP (CT0 / KD) which

6

includes both antibody-specific parameters, koff and KD, and an ITP condition parameter ptITP.

7

This offers flexibility in assay design by controlling the preconcentration ratio and the ITP

8

residence time. For a point of care LFA test, known values can be kinetic parameters of the

9

antibody-antigen pair and the desired test time. From this, one can estimate the ITP

10

preconcentration ratio and residence time required to achieve the desired detection limits.

11

Preconcentration ratio in ITP can be optimized by carefully selecting the electrolyte system,

12

optimizing the applied electric field and the device length.43

13

5. Summary

14

We demonstrated LoD and assay time improvement of LFAs using ITP. In this work we used a

15

model LFA to show how ITP can improve detection limit of LFA by performing target

16

preconcentration. We used two different labeled secondary antibody IgG as the targets which are

17

widely used in currently available LFA tests showing that ITP can be integrated to current LFA

18

tests to enhance their detection limit. Our method leverages high preconcentration power of ITP

19

to overcome the slow reaction kinetics of surface binding in LFA. We focused target antibodies

20

into a narrow ITP zone (order 100 μm) and transported them across the nitrocellulose membrane

21

to the test zone. Our quantitative analysis show that LoD of LFAs can be improved by 400-fold,

22

160-fold and 60-fold at assay time of 90 sec, 5 min and 1 hour, respectively. Our qualitative


Page 23 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1

colorimetric measurements show about 100-fold improvement in LoD using ITP, consistent with

2

our quantitative findings. We used an analytical model describing the ITP-enhanced surface

3

reaction in LFAs which confirms our experimental observations and serves as guide toward the

4

design of LFA enhanced by ITP. By integrating ITP into LFA we can also capture up to 30 %

5

which shows more than one order of magnitude improvement over conventional LFAs. High

6

target extraction and preconcentration of ITP enables the LFAs to use large volume of sample

7

when the target is very dilute, e.g. blood, and waste water. Moreover, we found that detection

8

limit of LFA can be improved by separating and preconcentrating target analytes from 5x diluted

9

mouse serum (data provided in SI) showing the potential of ITP to improve LFA with real-world

10

complex biological samples.

11

Associated Content

12

Videos showing isotachophoresis-enhanced lateral flow assay (ITP-LF) using Alexa Fluor 488

13

(Video S-1) and colloidal gold (Video S-2) labelled IgG, electrophoretic mobility measurements

14

of IgG secondary antibody, an example of calibration curves used for fluorescence quantitative

15

measurements, target capturing in conventional LFA, a discussion on surface reaction rate speed-

16

up by ITP preconcentration and demonstration of ITP-LF in complex samples are available as

17

supporting information. This material is available free of charge via the Internet at

18

http://pubs.acs.org.

19

Author Information

20

* E-mail: [email protected], Phone: (206) 543-9834

21

Notes: The authors declare no competing financial interest.

22

Acknowledgments



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

This work was sponsored by an NSF CAREER Award (J.D.P., grant number CBET-0747917).

2

We thank Carly Holstein and Paul Yager from Bioengineering department of the University of

3

Washington for assistance.

4

References

5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15)

(16) (17) (18) (19) (20) (21) (22)

Gubala, V.; Harris, L. F.; Ricco, A. J.; Tan, M. X.; Williams, D. E. Anal. Chem. 2012, 84, 487–515. Rosenstein, R. W.; Bloomster, T. G. Solid phase assay employing capillary flow. US4855240 A, August 8, 1989. Campbell, R. L.; Wagner, D. B.; O’Connell, J. P. Solid phase assay with visual readout. US4703017 A, October 27, 1987. Yetisen, A. K.; Akram, M. S.; Lowe, C. R. Lab Chip 2013, 13, 2210–2251. Warsinke, A. Anal Bioanal Chem 2009, 393, 1393–1405. Posthuma-Trumpie, G. A.; Korf, J.; Amerongen, A. Analytical and Bioanalytical Chemistry 2008, 393, 569–582. Giljohann, D. A.; Mirkin, C. A. Nature 2009, 462, 461–464. Gordon, J.; Michel, G. Clinical Chemistry 2008, 54, 1250–1251. Nabatiyan, A.; Baumann, M. A.; Parpia, Z.; Kelso, D. Journal of Acquired Immune Deficiency Syndromes January 2010 2010, 53, 55–61. De la Rica, R.; Stevens, M. M. Nat Nano 2012, 7, 821–824. Srinivas, P. R.; Kramer, B. S.; Srivastava, S. The Lancet Oncology 2001, 2, 698–704. Galasko, D. Journal of Alzheimer’s Disease 2005, 8, 339–346. De Jong, D.; Kremer, B. P. H.; Olde Rikkert, M. G. M.; Verbeek, M. M. Clinical Chemical Laboratory Medicine 2007, 45. Barletta, J. M.; Edelman, D. C.; Constantine, N. T. AJCP 2004, 122, 20–27. Rissin, D. M.; Kan, C. W.; Campbell, T. G.; Howes, S. C.; Fournier, D. R.; Song, L.; Piech, T.; Patel, P. P.; Chang, L.; Rivnak, A. J.; Ferrell, E. P.; Randall, J. D.; Provuncher, G. K.; Walt, D. R.; Duffy, D. C. Nat Biotech 2010, 28, 595–599. O’Keeffe, M.; Crabbe, P.; Salden, M.; Wichers, J.; Van Peteghem, C.; Kohen, F.; Pieraccini, G.; Moneti, G. Journal of Immunological Methods 2003, 278, 117–126. Safenkova, I.; Zherdev, A.; Dzantiev, B. Analytical and Bioanalytical Chemistry 2012, 403, 1595–1605. Linares, E. M.; Kubota, L. T.; Michaelis, J.; Thalhammer, S. Journal of Immunological Methods 2012, 375, 264–270. Parolo, C.; de la Escosura-Muñiz, A.; Merkoçi, A. Biosensors and Bioelectronics 2013, 40, 412–416. Chinnasamy, T.; Segerink, L. I.; Nystrand, M.; Gantelius, J.; Svahn, H. A. Analyst 2014, 139, 2348–2354. Fernández-Sánchez, C.; McNeil, C. J.; Rawson, K.; Nilsson, O.; Leung, H. Y.; Gnanapragasam, V. Journal of Immunological Methods 2005, 307, 1–12. Nagasaki, H.; Watanabe, M.; Komatsu, N.; Kaneko, T.; Dubé, J. Y.; Kajita, T.; Saitoh, Y.; Ohta, Y. Clinical Chemistry 1999, 45, 486–496.


Page 24 of 33

Page 25 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46


(23) Fridley, G. E. Methods and models to control and predict behavior of two dimensional paper networks for diagnostics. PhD Dissertation, University of Washington, 2014. (24) Fu, E.; Lutz, B.; Kauffman, P.; Yager, P. Lab on a Chip 2010, 10, 918. (25) Fridley, G. E.; Le, H.; Yager, P. Anal. Chem. 2014, 86, 6447–6453. (26) Govindarajan, A. V.; Ramachandran, S.; Vigil, G. D.; Yager, P.; Böhringer, K. F. Lab Chip 2011, 12, 174–181. (27) Heitzmann, H.; Richards, F. M. Proc Natl Acad Sci U S A 1974, 71, 3537–3541. (28) Becker, J. M.; Wilchek, M. Biochimica et Biophysica Acta (BBA) - General Subjects 1972, 264, 165–170. (29) Cheong, B. H.-P.; Liew, O. W.; Ng, T. W. Analytical Biochemistry 2014, 444, 57–59. (30) Babkina, S. S.; Ulakhovich, N. A.; Moiseeva, E. N.; Filyushina, E. E. Journal of Analytical Chemistry 2003, 58, 651–651. (31) Sheehan, P. E.; Whitman, L. J. Nano Lett. 2005, 5, 803–807. (32) Pohl, P.; Stecka, H.; Jamroz, P. Anal. Methods 2012, 4, 125–131. (33) Gautam, S.; Loh, K.-C. Analytical Biochemistry 2012, 423, 202–209. (34) Quirino, J. P.; Haddad, P. R. Anal. Chem. 2008, 80, 6824–6829. (35) Yang, K.; Wu, J. Biomicrofluidics 2010, 4. (36) Long, Z.; Liu, D.; Ye, N.; Qin, J.; Lin, B. ELECTROPHORESIS 2006, 27, 4927–4934. (37) Gözke, G.; Posten, C. Food Eng. Rev. 2010, 2, 131–146. (38) Timmer, B. H.; van Delft, K. M.; Olthuis, W.; Bergveld, P.; van den Berg, A. Sensors and Actuators B: Chemical 2003, 91, 342–346. (39) Karsenty, M.; Rubin, S.; Bercovici, M. Anal. Chem. 2014, 86, 3028–3036. (40) Han, C. M.; Katilius, E.; Santiago, J. G. Lab Chip 2014. (41) Rogacs, A.; Marshall, L. A.; Santiago, J. G. Journal of Chromatography A 2013. (42) Jung, B.; Bharadwaj, R.; Santiago, J. G. Analytical Chemistry 2006, 78, 2319–2327. (43) Moghadam, B. Y.; Connelly, K. T.; Posner, J. D. Anal. Chem. 2014, 86, 5829–5837. (44) Rosenfeld, T.; Bercovici, M. Lab Chip 2014. (45) Levicky, R.; Horgan, A. Trends in Biotechnology 2005, 23, 143–149. (46) Vijayendran, R. A.; Ligler, F. S.; Leckband, D. E. Anal. Chem. 1999, 71, 5405–5412. (47) Phillips, T. M. Journal of Chromatography B 2010, 878, 113–114. (48) Pappaert, K.; Desmet, G. Journal of Biotechnology 2006, 123, 381–396. (49) Garcia-Schwarz, G.; Bercovici, M.; Marshall, L. A.; Santiago, J. G. Journal of Fluid Mechanics 2011, 679, 455–475. (50) Bahga, S. S.; Bercovici, M.; Santiago, J. G. Electrophoresis 2010, 31, 910–919. (51) Moghadam, B. Isotachophoresis in Paper-Based Microfluidic Devices, University of Washington, 2014. (52) Spinola, S. M.; Cannon, J. G. Journal of immunological methods 1985, 81, 161–165. (53) Khurana, T. K.; Santiago, J. G. Analytical Chemistry 2008, 80, 6300–6307. (54) Welschof, M.; Terness, P.; Kipriyanov, S. M.; Stanescu, D.; Breitling, F.; Dörsam, H.; Dübel, S.; Little, M.; Opelz, G. Proceedings of the National Academy of Sciences 1997, 94, 1902–1907. (55) Kumagai, I.; Tsumoto, K. eLS 2001. (56) Parsons, M. L. J. Chem. Educ. 1969, 46, 290. (57) Lyashchenko, K. P.; Greenwald, R.; Esfandiari, J.; Rhodes, S.; Dean, G.; de la RuaDomenech, R.; Meylan, M.; Vordermeier, Hm.; Zanolari, P. Clin Vaccine Immunol 2011, 18, 2143–2147.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2 3 4 5

(58) Van Amerongen, A.; Wichers, J. H.; Berendsen, L. B. J. M.; Timmermans, A. J. M.; Keizer, G. D.; van Doorn, A. W. J.; Bantjes, A.; van Gelder, W. M. J. Journal of Biotechnology 1993, 30, 185–195.

6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30


Page 26 of 33

Page 27 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1


Figures and tables

2 3

Figure 1. Experimental setup for (A) ITP-enhanced LFA and (B) conventional LFA. For both

4

assays a 40 mm long nitrocellulose membrane is used as the paper device. In ITP-LF targets

5

migrate electrophoretically under the high electric field of 500 µA and accumulate at the narrow

6

interface of the TE and LE. In LFA targets migrate through the device by the capillary action.

7

High wicking properties of the absorption pad guarantees continuous flow of the sample. (C)

8

Experimental snapshots taken at 5 different times showing ITP-focused IgG labeled with AF488

9

is captured at the test zone. (a-c) ITP plug forms and migrates on the membrane, (d) current is



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

reduced to 50 µA while the ITP zone crosses the test zone to increase the reaction time (e) after

2

the plug is passed the test zone we see a specific fluorescence signal is generated. The white

3

arrow indicates instant location of the ITP plug.

4 5 6 7 8 9 10 11 12 13


Page 28 of 33

Page 29 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1 2

Figure 2. Experimental data showing quantified fraction of bound probe, h, for detection of IgG

3

using ITP-LF (closed diamonds) and conventional LFA (open symbols) at 90 second (open

4

triangle), 5 minutes (open diamonds), and 1 hour (open squares) assay time. Along with

5

experimental data, we show results of analytical models for the conventional (dashed lines) LFA

6

and for the ITP-LF (solid line) assays. ITP creates two orders of magnitude improvement in LoD

7

of LFA after 7 minutes assay time. To generate the model curves we used the estimated koff =

8

1.75×10-3 s-1 and KD = 1.42×10-6 M , ITP preconcentration ratio p = 92 and tITP = 284 s. The

9

error bars indicate 95% confidence interval for three measurements.

10 11 12 13 14



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1 2

Figure 3. Qualitative detection limit of ITP-LF (right) compared to LFA (left). Here the target is

3

Goat anti Mouse IgG labeled with 40 nm colloidal gold. LFAs are performed for 5 min to be

4

consistent with ITP assay time.

5 6 7 8 9 10


Page 30 of 33

Page 31 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1 2

Figure 4. Ratio of  LoDLFA LoDITP  LF  , predicted by our analytical model, versus the assay

3

time, t. Each value on y-axis shows the level of improvement in LoD of LFA. Contours are

4

plotted for tITP = 284 s, p = 100 (bottom), p = 500 (top), koff = 1.75×10-3 s-1 (solid lines) and koff =

5

1.75×10-4 s-1 (dashed lines). We note that contours have no dependence on the KD value showing

6

that ITP improves detection limit of LFA by promoting the reaction rate. Solid symbols represent

7

three experimental conditions where for ITP-LF CT0 = 0.05 mg/l, preconcentration ratio is p =

8

92, tITP = 284 s and for LFA tLFA = 90 sec, 5 min and 1 hour corresponding to CT0 = 20, 8 and 3

9

mg/l, respectively.

10 11 12 13 14 15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1

Page 32 of 33

Table 1. Detection limit improvement of conventional LFA using ITP.

Assay time

LoDLFA (mg/l)

LoDITP-LF (mg/l)

Fold improvement of detection limit

90 seconds

20

0.05

400

5 minutes

8

0.05

160

1 hours

3

0.05

60

2 3 4

Table 2. Fraction of captured target, NC NT 0 , by ITP-LF compared to that of LFA. CT 0 is the

5

initial target concentration, NT 0 is the initial number of moles of target, and N P 0 is the number of

6

moles of probe on the surface. The uncertainties represent 95% confidence interval for three

7

measurements.

8

a

 NC

NT 0  ITP  LF

 NT 0 LFA b

 NC

NT 0 LFA c

CT 0 a

N P0 b

 NT 0 ITP LF b

0.05

0.5

0.02

11.62 ± 2.29

0.06

No detection

0.1

0.5

0.04

24.05 ± 5.83

0.13

No detection

0.5

0.5

0.23

30.86 ± 6.91

0.67

No detection

10

0.5

4.6

7.01 ± 0.32

10.6

0.47 ± 0.05

25

0.5

11.6

3.94 ± 0.08

26.6

0.70 ± 0.04

c

in mg/l, b in fmoles, c in %

9 10


Page 33 of 33

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

1


For TOC only

2


Two Orders of Magnitude Improvement in Detection Limit of Lateral

Recommend Documents