Dynamic Article LinksC<Analyst

Cite this: Analyst, 2012, 137, 2381

www.rsc.org/analyst PAPER

Publ

ishe

d on

11

Apr

il 20

12. D

ownl

oade

d by

Uni

vers

ity o

f C

alga

ry o

n 10

/09/

2013

14:

32:0

0.

View Article Online / Journal Homepage / Table of Contents for this issue

Fast measurement of binding kinetics with dual slope SPR microchips

Tridib Ghosha and Carlos H. Mastrangeloab

Received 10th January 2012, Accepted 7th March 2012

DOI: 10.1039/c2an35045a

We demonstrate a new dual slope SPR technique that is ten-fold faster than the conventional step-

response method. The new scheme utilizes rapid slope-based measurements followed by rapid reset, and

it separates association and dissociation half reaction measurements at two separate sites inside a dual-

chamber PDMSmicrofluidic chip. For a model CAII-ABS test system, the association and dissociation

slopes were measured in 30 seconds compared to 5 minutes for step-response. The values of ka and kdcalculated from the slope method are 3.66� 0.19� 103 M�1 s�1 and 4.83� 0.17� 10�2 s�1, respectively,

matching well with step-response values while facilitating �10 to 15 fold faster detection and

quantification.

Introduction

Surface Plasmon Resonance (SPR) has formed the basis for

real-time label-free detection of bio-interactions for the last

twenty years. In the conventional SPR step response method

a functionalized surface is subject to a single association and

a single dissociation step producing a sensorgram.1–4 The

observed experimental response is then fitted to a model expo-

nential response in order to obtain the characteristic association

and dissociation kinetic constants.5,6 Single-pulse sensorgrams

have two characteristics that make the determination of the

kinetic constants intrinsically slow: (1) the sensorgram must

achieve steady state for both association and dissociation steps

and (2) the two reactions are sequential. The development of

faster detection schemes is an important area of research as SPR

is presently being considered as a candidate label-free technology

for high-throughput drug screening.

In this paper we demonstrate a new dual slope SPR technique

that provides a faster means for the measurement of the reaction

kinetics. The new scheme utilizes rapid slope-based measure-

ments followed by rapid reset, and it separates association and

dissociation half reaction measurements at two separate sites

within a PDMS microfluidic chip.

The dual-slope method

Bio-interactions involving well behaved macromolecules follow

a simple reversible bimolecular Langmuirian equation7 given by

Aþ B �����!kon

���koff

AB (1)

aDepartment of Bioengineering, University of Utah, Salt Lake City, UT,USA. E-mail: t.ghosh@utah.edubDepartment of Electrical and Computer Engineering, University of Utah,Salt Lake City, UT, USA. E-mail: carlos.mastrangelo@utah.edu

This journal is ª The Royal Society of Chemistry 2012

where A, B and AB are analyte, ligand and formed complex,

respectively. The rate of formation of the complex AB as

a function of time t subject to a step change of analyte concen-

tration [A] can be written as

d½AB�dt¼ kon½A�½B� � koff ½AB� (2)

Analytical integration of this rate equation yields a piecewise

exponential solution.6 The SPR sensorgram specified by the

imager intensity I(t) is proportional to the bound complex

concentration by a factor GSPR. In the association cycle then

IAðtÞ ¼ GSPR

�kon½A�½Bo�

kon½A� þ koff

��1� e�ðkon ½A�þkoffÞt

�(3)

where [A] is the analyte concentration and [B0] is the maximum

surface concentration of active ligand. Similarly, for the disso-

ciation phase subject to buffer flow, the sensorgram intensity is

IDðtÞ ¼ GSPR½ABd�e�koff t (4)

where [ABd] is the concentration of bound complex AB imme-

diately before dissociation. In the conventional step response

method, eqn (3) and (4) are fitted to the experimental sensorgram

to determine kon and koff. This sometimes requires reaching

equilibrium in both association and dissociation cycles and

a correspondingly long measurement cycle.

The kon and koff information however can be obtained in

a much shorter time from the association and dissociation initial

slopes as follows. Differentiating eqn (3) with respect to time

gives

SA ¼ limt/0

dIAðtÞdt

¼ GSPRkon½A�½Bo� ¼ KSPRkon½A� (5)

where SA is the associative slope. At a given flow rate, the

associative slope is proportional to the analyte concentration.

Similarly from eqn (4), subsequent differentiation yields

Analyst, 2012, 137, 2381–2385 | 2381

Fig. 1 Basic principle used in the new fast dual-slope SPR technique.

Two identical functionalized sites are exposed to different sequences of

analyte, buffer, pre-dissociation rapid equilibrium and regeneration

solutions. The association and dissociation constants are obtained from

the two decoupled slopes SA and SD, and the equilibrium sensorgram

intensity level IRE.

Publ

ishe

d on

11

Apr

il 20

12. D

ownl

oade

d by

Uni

vers

ity o

f C

alga

ry o

n 10

/09/

2013

14:

32:0

0.

View Article Online

SD ¼ limt/0

dIDðtÞdt

¼ �GSPRkoff ½ABd� (6)

with SD being the dissociative slope. In order to extract kon and

koff from (5) and (6), we need to eliminate the constant terms.

This can be done by deliberately reacting the species to equilib-

rium prior to the dissociation phase. Since a high analyte

concentration ensures rapid surface saturation, we can use a high

analyte concentration [ARE] to achieve a rapid complex equi-

librium level [ABRE] with corresponding SPR intensity IRE:

IRE ¼ GSPR½ABRE� ¼ KSPRkon½ARE�kon½ARE� þ koff

(7)

Rearranging the terms in eqn (7), (8) and (9) we can eliminate

GSPR and KSPR. Since [A] and [ARE] are both known, and IRE is

extracted from the sensorgram, kon and koff can then be written as

koff ¼ � SD

IRE

(8)

and

kon ¼ SA½ARE� þ SD½A�IRE½ARE�½A� (9)

Therefore the kinetic constants can be determined from

a measurement of the association slope with analyte concentra-

tion [A] and a second measurement of the dissociation slope with

pre-dissociation equilibrium achieved with analyte concentration

[ARE]. It should be noted that similar to a single cycle of

association and dissociation in a conventional step response

cycle, the dual slope technique estimates a single set of kon and

koff for one set of [A] and [ARE].

In this work, the association and dissociation slopes are

determined from measurements at two identical but separate

functionalized surface sites, thus producing two distinct sensor-

grams as shown in Fig. 1. The association site in the function-

alized surface is exposed to a brief positive association slope (SA)

measurement step by exposing the surface to analyte of

concentration [A], followed by a brief regeneration step and

subsequent short buffer wash. At the dissociation site, the surface

is exposed to a solution of high analyte concentration [ARE]

quickly producing pre-dissociation equilibrium level IRE.

For the analyte solution at the dissociation site, [ARE] [ [A]

in order to reach rapid equilibrium. The rapid equilibrium step is

followed by a brief negative dissociation slope (SD) measurement

in pure buffer. Both chemical excitation sequences are repeated

multiple times to produce measurements of progressively

improved quality.

The time required for the slope measurement is short, in the

order of few seconds; hence the process of detection and sensing

is fast. The minimum measurement time is limited only by the

amount of noise present in the system. Generally speaking

the slope measurement time is much less than that required by

the full step response. In this paper we demonstrate a measure-

ment realized ten-fold faster than the step response method.

Fig. 2 Schematic of a dual-slope SPR chip consisting of four SPR spots

and five microvalves.

Dual-slope SPR chip

In order to utilize the dual-slope technique we need to introduce

short flow cycles with steep chemical gradients. Such flow

2382 | Analyst, 2012, 137, 2381–2385

switching profiles can only be achieved in microfluidic environ-

ments15 which exhibit low dispersive mixing. Fig. 2 shows

a schematic of a dual-slope SPR chip. The experiments and data

acquisition are carried out in two separate dedicated chambers.

The association chamber performs the association half reaction

using two gold sensor spots (sensor + control) and three micro-

valves, each for the flow control of buffer, analyte and regener-

ation solutions. The dissociative chamber utilizes two gold spots

and two microvalves each of which control the flow of analyte

and regeneration solutions.

The dual-slope SPR scheme was implemented using

a conventional two-level PDMS microchip.4,8 Fig. 3 shows an

image of the fabricated PDMS chip. It consists of two separate

microfluidic chambers for associative and dissociative half

reactions, respectively, each of which is connected to three

This journal is ª The Royal Society of Chemistry 2012

Fig. 3 Photograph of dual-slope SPR microchip.

Publ

ishe

d on

11

Apr

il 20

12. D

ownl

oade

d by

Uni

vers

ity o

f C

alga

ry o

n 10

/09/

2013

14:

32:0

0.

View Article Online

pressure driven input sources controlled by separate microvalves

and two output channels. The two output channels in each

chamber are lined by gold sensor spots which record the sensor

and control sensorgrams, respectively.

Materials and methods

The slope technique was demonstrated using a model system of

Carbonic Anhydrase-II (CAII) analyte and immobilized 4-(2-

aminoethyl) benzenesulfonamide (ABS) ligand4,9 on amine

reactive PEG modified gold surface.

Reagents

Polyethylene glycol (PEG) compounds were purchased

from Laysan Bio. These include 5 kDa carboxymethyl-PEG-

thiol (cm-PEG) and 2 kDa methoxy-PEG-thiol (m-PEG).

N-Ethyl-N0-(3-dimethylaminopropyl) carbodiimide (EDC),

sulfo-N-hydroxysuccinimide (S-NHS) and phosphate buffered

saline (PBS) tablets were purchased from Thermo Scientific.

Carbonic anhydrase II (CA-II, Mw z 29 kDa), 10% sodium

dodecyl sulfate (SDS) and 4-(2-aminoethyl) benzenesulfonamide

(ABS) were purchased from Sigma-Aldrich. Sylgard 184 Poly-

dimethylsiloxane (PDMS) kit was purchased from Dow Chem-

icals. Deionized 18 MU water (DI water) was provided by the

University of Utah nanofab facility. SF10 Schott glass substrates

(2 � 2 sq. inch) were custom ordered from Schott.

Surface plasmon resonance

The sensorgram data are acquired using a manufacturer modi-

fied GWC Technologies SPRimager2 that accommodates our

chip. Additional modifications are reported elsewhere.4

Gold/glass substrate

SF10 glass substrates were cleaned with piranha (3 : 1,

H2SO4 : H2O2) solution for 10 min, rinsed in DI water for 10 min

and blow dried with N2 gas followed by baking at 80 �C in an

oven for 10 min. They are then transferred to a TM Vacuum

Sputtering machine and a 3 nm adhesion layer of Ti/W is

This journal is ª The Royal Society of Chemistry 2012

deposited followed by a 40 nm layer of gold. The metal is then

patterned using photolithography in the University of Utah

nanofab facility. The array of Au spots patterned using an AZ

1813 photoresist has dimensions of 200 � 200 mm2 and a thick-

ness of �43 � 2 nm. The metals are then etched away, glass

slides are cleaned with acetone and DI water to remove excess

photoresist, blow dried and finally stored in a vacuum desiccator

until being used.

Microfluidics

The dual-slope SPR microchip of Fig. 3 is fabricated stepwise

using two-level PDMS stamp process on patterned glass

substrate. The steps of fabrication are reported elsewhere.8

Briefly, Si molds for two separate layers are generated photo-

lithographically using photoresists (SU-8 and AZ 9260 for valve

and flow layers, respectively). PDMS is then poured over these

defined molds and cured overnight for 15 hours at 100 �C. Thepatterned PDMS chunks are then cleaned with acetone, baked at

100 �C on a hot plate for 10 min, and exposed to oxygen plasma

for 20 s at 400 mTorr pressure in a March Plasmod system.

The layers are aligned and bonded together using an in-house

alignment system. The same process is repeated for binding the

two-level PDMS chip to glass substrate with patterned gold

spots. Prior to binding of PDMS with glass substrate, it is

cleaned using a protocol in our previous reporting.4 All flow

channels have dimensions of 250 � 20 mm2. The microchip has

the final dimensions of 12 � 13 mm.

Gold spot functionalization

The bare gold surface is first modified with PEGs similar to our

previous paper and elsewhere.4,10 The surface modification using

PEGs is quite effective in reducing non-specific adsorption.

Briefly, gold spots inside the microchannels are first rinsed with

0.01 N HCl solution followed by PEGylation in PBS buffer

(50 mM phosphate, 1 M NaCl, pH 7.4) using cm-PEG. A short

underbrushed layer of m-PEG further minimizes the non-specific

adsorption. Excess PEG is removed by a short rinse of 50 mM

NaOH. This is followed by immobilization of ABS ligand using

standard amine-coupling procedure to form a two-dimensional

sensing surface. A solution of sulfo-NHS (0.1 M) and EDC

(0.4 M) is used to facilitate crosslinking of activated cm-PEG to

ABS. The control surface is blocked with ethanolamine (50 mM)

in PBS buffer (25 mM phosphate, pH 8.4 with 0.01% SDS) after

cm-PEG activation.

Experimental

The chip-on-substrate is placed onto a SPR mounting cell and

coupled to a prism using an index matching fluid (n ¼ 1.72). The

input sources are then connected to pressure driven fluid reser-

voirs using plastic tubing (0.01 inch diameter, Tygon�). Due to

gas permeable properties of PDMS, the valve control lines are

filled with water to displace air and are connected directly to

pressure sources. The input reservoirs contain PBS (10 mM

phosphate, pH 7.4) as a flow buffer, CA II as an analyte and

0.01% SDS as a regeneration solution, respectively. The fluid

flow and valve actuating pressures are 10 psi and 25 psi,

respectively. The reservoir pressure corresponds to a flow

Analyst, 2012, 137, 2381–2385 | 2383

Publ

ishe

d on

11

Apr

il 20

12. D

ownl

oade

d by

Uni

vers

ity o

f C

alga

ry o

n 10

/09/

2013

14:

32:0

0.

View Article Online

velocity of �8 cm s�1 in output channels which is sufficient to

overcome mass transport effects.4,9 The analyte samples are

prepared by dissolving dry CA II powder in PBS followed

by concentration measurement using a spectrophotometer

(GENSYS5, Thermo Scientific) at a wavelength of 280 nm.

Reported values of extinction coefficient (a280) for bovine CA II

range from 55300 to 57000 L mol�1 cm�1.11,12 The protein

concentration (C) is then given by the relation C (mol L�1) ¼A280/(a280 � b), where ‘A280’ is the value of absorbance at 280 nm

and ‘b’ being the optical path length (1 cm for the given

photometer). The high concentration analyte ([ARE]) solution is

first prepared followed by appropriate dilution using PBS to

make low concentration solution [A]. Two different concentra-

tion sets ([A] and [ARE]) were prepared and used in experiments

to calculate two sets of kon and koff. The analyte concentrations

used to record the conventional step-response sensorgrams were

used for associative half reactions on the same sensor surface.

Values of analyte concentration sets {[A],[ARE]} were

{1.63,54.4} and {3.4,56.67} mM, respectively. The lowest allow-

able concentration of [A] is dictated by the background noise of

the sensorgram, and the maximum concentration [ARE] was

selected as the saturation limit of protein concentration for our

spectrophotometer (GENSYS5, Thermo Scientific). The

sampling rate was �7.143 Hz for all data recordings. A Wasabi

camera control software integrated with EMCCD camera is used

for data acquisition and analysis. The sensorgram intensity I(t)

for sensing and reference spots (1160 pixels) for all data frames

were first retrieved using the in-built software followed by further

analysis using MATLAB. The SPR sensorgrams from the sensor

spots were subtracted from control spots to account for the RI

change due to the bulk effect and extract the signal changes from

the bio-interaction on the sensor surface. Prior to this subtrac-

tion, the baseline signal level at the point of solution injection in

all datasets is offset to zero in order to account for the fixed

imager intensity.

Results and discussion

In order to serve a basis for comparison we first ran a conven-

tional step-response curve of a functionalized spot within

a PDMS chip. Fig. 4 shows the step-response sensorgram for an

analyte concentration of 3.4 mM requiring 5 minutes of recording

time.

Fig. 4 Conventional SPR step-response. The sensorgram consists of one

association and one dissociation step (5 min for CAII-ABS).

2384 | Analyst, 2012, 137, 2381–2385

Next, using a dual-slope SPR chip, we tested the feasibility of

the new technique of parameter estimation using two different

sets of associative and dissociative analyte concentrations.

Fig. 5(a) shows a sensorgram of ten half reaction cycles at analyte

concentration of [A] ¼3.4 mM recorded from Au spots in asso-

ciation chamber indicating associative slope SA. The time periods

for analyte, buffer and regeneration flows are 10, 7 and 13

seconds, respectively.

The flow cycle periods were adjusted to be sufficiently long as

to achieve signal to noise ratios much greater than one; yet the

cycle times were much shorter than expected exponential asso-

ciation and dissociation time constants.

Fig. 5(b) shows a sensorgram of ten dissociative half reaction

cycles indicating dissociative slope SD for analyte concentration

of [ARE] ¼ 56.67 mM. While pre-dissociation saturation is

allowed for 17 s, the dissociation is carried out for 13 s using

buffer flow. The kinetic constants for conventional step response

were calculated first by exponential curve fitting of sensorgrams

using the method of least squares. The results are summarised in

Table 1. The rates reported were averaged over 10 cycles. The

rates are in good agreement with previously reported values.4

Fig. 5 Rapid multi-step measurement of association and dissociation

slopes for CAII-ABS. (a) For the association slope we use a three-step

buffer, analyte and regeneration solution cycle. The top trace is unre-

ferenced SPR trace and the bottom is referenced to a passivated spot. (b)

For the dissociation slope we use a two-step short rapid equilibrium and

dissociation cycle. The analyte concentrations [A] and [ARE] are 3.4 and

56.67 mM, respectively.

This journal is ª The Royal Society of Chemistry 2012

Table 1 Comparison of rate constants

Method

Analyteconcentration(mM)

Kineticon-ratekon (M

�1 s�1)

Kineticoff-ratekoff (s

�1)

Conventionalstep-response

1.63 4.75 � 0.42 � 103 4.81 � 0.15 � 10�2

3.4 8.68 � 0.35 � 103 4.76 � 0.20 � 10�2

Dual-slopemethod

[A] ¼ 1.63,[ARE] ¼ 54.4

3.66 � 0.19 � 103 4.83 � 0.17 � 10�2

[A] ¼ 3.4,[ARE] ¼ 56.67

3.60 � 0.22 � 103 4.74 � 0.21 � 10�2

Publ

ishe

d on

11

Apr

il 20

12. D

ownl

oade

d by

Uni

vers

ity o

f C

alga

ry o

n 10

/09/

2013

14:

32:0

0.

View Article Online

Although the off-rate estimation using the method of least

squares is more accurate and reproducible, the on-rate tends to

vary with analyte concentration and is hence less reproducible.6

Our studies confirm that as well. However, as indicated by our

estimate using the dual-slope method, both the on- and off-rates

appear to be more reproducible for different sets of analyte

concentrations. One possible explanation for the apparent

reproducibility may be the result of the short measurement of the

slope technique which intrinsically has a much lower sensitivity

to drift induced errors when compared to the longer conven-

tional step response measurement.

Signal-to-noise ratio

The experimentally observed signal-to-noise ratio or SNR for the

dual slope SPR technique is about 18 dB compared to 28 dB of

the step-response of Fig. 4. The SNR was calculated as the ratio

of the root-mean-square intensity of the fitted response over that

of the recorded signal in the absence of any excitation. The SNR

is improved when the measurement is performed over a longer

period of time or averaged over many cycles because the

standard deviation of random noise does not grow as much as

the strength of the signal does. Therefore for a shorter

measurement period, the slope scheme yields a lower SNR. A

figure of merit more useful for comparison is

FOM ¼ SNRffiffiffiffiffiffiffiffiffiffiffiTmeas

p (10)

where Tmeas is the total measurement time. The ratio of FOMs

for the slope versus the step response measurement is approxi-

mately 1.4 when using measurement periods of 30 s and 5 min,

respectively. This indicates that the FOM for the dual slope

methodology is slightly better than that of the conventional

step response. This technique should therefore be utilized in

situations where random noise is not a limiting factor for the

measurement when one can trade speed with SNR or when the

noise is not random.

This journal is ª The Royal Society of Chemistry 2012

Discussion

While a complete on–off cycle in conventional method takes

about 250–300 s,9,13,14 the new dual slope technique requires only

30 s for one cycle. While the associative cycle takes only about

10 s, the saturation level prior to dissociation (17 s in our case)

can be achieved even faster using a higher concentration, hence

reducing the overall time to complete the two half reactions.

Besides, rate estimation using equilibrium analysis6,14 requires

association phase going to equilibrium, hence consuming time

and larger volumes of bio-samples. The dual-slop technique

circumvents such drawbacks. It must however be noted that the

rapid equilibration time also depends on the value of rate

constants for a particular bimolecular binding system. In our

case, the detection process can be achieved �10 times faster

without compromising the quality or accuracy of rate estimation.

Conclusions

A new methodology has been developed for the rapid determi-

nation of kinetic constants of bio-interactions via SPR. The

technique is based on the measurement of association and

dissociation slopes in two half reactions. The dual-slope tech-

nique was implemented in a microfluidic chip and demonstrated

to be ten-fold faster than the conventional step response.

References

1 U. J€onsson, L. F€agerstam, B. Ivarsson, B. Johnsson, R. Karlsson,K. Lundh, S. L€of�as, B. Persson, H. Roos, I. R€onnberg,S. Sj€olander, E. Stenber, R. St�ahlberg, C. Urbaniczky, H. €Ostlinand M. Malmqvist, BioTechniques, 1991, 11, 620–627.

2 E. Stenberg, B. Persson, H. Roos and C. Urbaniczky, J. ColloidInterface Sci., 1991, 143, 513–526.

3 S. L€ofas and B. Johnsson, J. Chem. Soc., Chem. Commun., 1990,1526–1528.

4 L. Williams, T. Ghosh and C. Mastrangelo, Anal. Chem., 2010, 82,6025–6031.

5 L. G. Fagerstam, A. Frostell-Karlsson, R. Karlsson, B. Persson andI. Ronnberg, J. Chromatogr., 1992, 597, 397–410.

6 D. J. O’Shannessy, M. Bringham-Burke, K. K. Soneson, P. Hensleyand I. Brooks, Anal. Biochem., 1993, 212, 457–468.

7 R. Karlsson, A. Michaelsson and L. Mattson, J. Immunol. Methods,1991, 145, 229–240.

8 F. Azizi and C. Mastrangelo, Lab Chip, 2008, 8, 907–912.9 J. Lahiri, L. Isaacs, B. Brzybowski, J. Carbeck and G. Whitesides,Langmuir, 1999, 15, 7186–7198.

10 K. Uchida, H. Otsuka, M. Kaneko, K. Kataoka and K. Nagasaki,Anal. Chem., 2005, 77, 1075–1080.

11 M. Krishnamurthy, G. Kaufman, A. Urbach, I. Gitlin, K. Gudiksen,D. Weibel and G. Whitesides, Chem. Rev., 2008, 108, 946–1051.

12 P. Nyman and S. Lindskog, Biochim. Biophys. Acta, 1964, 85, 141.13 D. G. Myszka, Anal. Biochem., 2004, 329, 316–323.14 M. Corr, A. E. Slanetz, L. F. Boyd, M. T. Jelonak, S. Khilko,

B. K. Al-Ramadi, Y. S. Kim, S. E. Maher, A. L. M. Bothwell andD. H. Marguiles, Science, 1994, 265, 946–949.

15 R. F. Probstein, Physicochemical Hydrodynamics: An Introduction,Wiley, NY, 2nd edn, 2003.

Analyst, 2012, 137, 2381–2385 | 2385