Design and characterization of immobilized enzymes in microfluidic systems

Article abstract of DOI:10.1021/ac010822u

Herein we report the fabrication, characterization, and use of total analytical microsystems containing surface-immobilized enzymes. Streptavidin-conjugated alkaline phosphatase was linked to biotinylated phospholipid bilayers coated inside poly(dimethylsiloxane) microchannels and borosilicate microcapillary tubes. Rapid determination of enzyme kinetics at many different substrate concentrations was made possible by carrying out laminar flowcontrolled dilution on-chip. This allowed Lineweaver-Burk analysis to be performed from a single experiment with all the data collected simultaneously. The results revealed an enzyme turnover number of 51.1 ± 3.2 s^-1 for this heterogeneous system. Furthermore, the same enzyme immobilization strategy was extended to demonstrate that multiple chemical reactions could be performed in sequence by immobilizing various enzymes in series. Specifically, the presence of glucose was detected by two coupled steps employing immobilized avidinD-conjugated glucose oxidase and streptavidin-conjugated horseradish peroxidase.

Full text of DOI:10.1021/ac010822u

Anal. Chem. 2002, 74, 379-385
De s ig n a n d Ch a ra c t e riza t io n o f Im m o b ilize d
En zym e s in Mic ro flu id ic S ys t e m s
Ha nbin Ma o, Tinglu Ya ng, a nd Pa ul S. Cre m e r*
Department of Chemistry, Texas A&M University, P.O. Box 30012, College Station, Texas 77842-3012
Herein we report the fabrication, characterization, and use
of total analytical microsystems containing surface-im-
mobilized enzymes. Streptavidin-conjugated alkaline phos-
phatase was linked to biotinylated phospholipid bilayers
coated inside poly(dimethylsiloxane) microchannels and
borosilicate microcapillary tubes. Rapid determination of
enzyme kinetics at many different substrate concentra-
tions was made possible by carrying out laminar flow-
controlled dilution on-chip. This allowed Lineweaver-
Burk analysis to be performed from a single experiment
with all the data collected simultaneously. The results
Several investigators have also begun to explore the role that
enzymes could play in microfluidics for developing lab-on-a-chip
technologies. For example, Duffy and co-workers developed a
centrifugal microfluidic system capable of measuring dozens of
enzymatic assays simultaneously in a homogeneous format.13 They
exploited this apparatus to measure a substrate inhibition constant
for binding to an enzyme in a single experiment. In another
homogeneous assay, Hadd and co-workers employed computer-
controlled electrokinetic transport for analysis and electrophoretic
4
separation of enzyme inhibitors. Extending such work to hetero-
geneous assays with immobilized enzymes would provide distinct
advantages over homogeneous formats for applications such as
chemical synthesis and assay reuse because reactants can flow
in and products flow out without disrupting the catalysts. Indeed,
a few groups have already begun exploring procedures for
immobilizing enzymes in microfluidic devices for use in hetero-
geneous assays. Examples include work by Kim and co-workers,
which showed that enzymes could be imbedded inside sol-gel
structures.14 Another example is the work of Eteshola and
Leckband that undertook the design of an ELISA assay inside
PDMS microfluidic channels.15
-
1
revealed an enzyme turnover number of 5 1 .1 ( 3 .2 s
for this heterogeneous system. Furthermore, the same
enzyme immobilization strategy was extended to demon-
strate that multiple chemical reactions could be per-
formed in sequence by immobilizing various enzymes in
series. Specifically, the presence of glucose was detected
by two coupled steps employing immobilized avidinD-
conjugated glucose oxidase and streptavidin-conjugated
horseradish peroxidase.
In the studies presented here, we developed a simple method
for immobilizing biocatalysts on the walls of PDMS microfluidic
channels for three specific goals. First, we wanted to find a simple
strategy for protein immobilization within lab-on-a-chip devices
for use in rapid prototyping experiments. In particular, we wanted
to do this for PDMS molds bonded to glass as this system is in
wide use. The challenge for such work, however, lay in the fact
that immobilization methods needed to be developed such that
both the PDMS and glass surfaces accommodated protein attach-
ment while still preserving reasonable enzymatic activity. Second,
we wanted to exploit such methods for obtaining enzyme turnover
rates and related kinetic data in a massively parallel fashion with
good signal-to-noise properties. Finally, we wanted to test the
feasibility of employing such methods for the immobilization of
several different catalysts in a series for use in multistep or coupled
chemical reactions.
Microfluidic devices have gained a great deal of attention over
the last several years due to their potential for creating inexpensive
analytical tools with minute volumes and high throughput as well
1
as for their potential for mass replication. A host of diagnostic
applications can be coupled to these systems for immunoassay
development, biosensor design,4,5 genetics,6-8 and cell screen-
2,3
9
ing. Furthermore, microfluidics can be exploited for patterning
surfaces with biological materials at micrometer-scale resolution.
There have been several reports dealing with patterning proteins,10
c
e
l
l
s
,
1
1
a
n
d
p
l
a
n
a
r
l
i
p
i
d
b
i
l
a
y
e
r
s
.
1
2
*
To whom correspondence should be addressed: (tel) (979) 862-1200; (fax)
(
979) 845-7561, (e-mail) cremer@mail.chem.tamu.edu.
(
(
(
(
1) Figeys, D.; Pinto, D. Anal. Chem. 2 0 0 0 , 72, 330A-335A.
2) Chiem, N.; Harrison, D. J. Anal. Chem. 1 9 9 7 , 69, 373-378.
3) Chiem, N. H.; Harrison, D. J. Electrophoresis 1 9 9 8 , 19, 3040-3044.
4) Hadd, A. G.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1 9 9 9 , 71, 5206-
5
212.
(
(
(
5) Santini, J. T.; C., R. A.; Scheidt, R.; Cima, M. J.; Langer, R. Angew. Chem.,
Int. Ed. 2 0 0 0 , 39, 2397-2407.
6) Khandurina, J.; McKnight, T. E.; Jacobson, S. C.; Waters, L. C.; Foote, R.
(10) Delamarche, E.; Bernard, A.; Schmid, H.; Michel, B.; Biebuyck, H. Science
1 9 9 7 , 276, 779-781.
(11) Chiu, D. T.; Jeon, N. L.; Huang, S.; Kane, R. S.; Wargo, C. J.; Choi, I. S.;
Ingber, D. E.; Whitesides, G. M. Proc. Natl. Acad. Sci. U.S.A. 2 0 0 0 , 97,
2408-2413.
(12) Yang, T.; Jung, S. Y.; Mao, H.; Cremer, P. S. Anal. Chem. 2 0 0 1 , 73, 165-
169.
(13) Duffy, D. C.; Gillis, H. L.; Lin, J.; Sheppard, N. F. J.; Kellogg, G. J. Anal.
S.; Ramsey, J. M. Anal. Chem. 2 0 0 0 , 72, 2995-3000.
7) Burns, M. A.; Johnson, B. N.; Brahmasandra, S. N.; Handique, K.; Webster,
J. R.; Krishnan, M.; Sammarco, T. S.; Man, P. M.; Jones, D.; Heldsinger,
D.; Mastrangelo, C. H.; Burke, D. T. Science 1 9 9 8 , 282, 484-487.
8) Lagally, E. T.; Simpson, P. C.; A., M. R. Sens. Actuators, B 2 0 0 0 , 63, 138-
(
(
1
46.
Chem. 1 9 9 9 , 71, 4669-4678.
(14) Kim, Y.; Park, C. B.; Clark, D. S. Biotechnol. Bioeng. 2 0 0 1 , 73, 331-337.
(15) Eteshola, E.; Leckband, D. Sens. Actuators, B 2 0 0 1 , 72, 129-133.
9) Sohn, L. L.; Saleh, O. A.; Facer, G. R.; Beavis, A. J.; Allan, R. S.; Notterman,
D. A. Proc. Natl. Acad. Sci. U.S.A. 2 0 0 0 , 97, 10687-10690.
1
0.1021/ac010822u CCC: $22.00 © 2002 American Chemical Society
Analytical Chemistry, Vol. 74, No. 2, January 15, 2002 379
Published on Web 12/12/2001

the same for both systems within experimental error. Further-
more, the use of the PDMS system made data collection more
than 1 order of magnitude more rapid than in the microcapillary
platform, while possessing an improved signal-to-noise ratio. In
the final section of this work, we immobilized glucose oxidase
and horseradish peroxidase in series to create a simple glucose
sensor that works in two steps.
EXPERIMENTAL SECTION
Materials. Phospholipid vesicles were formed from 1,2-
dilauroyl-sn-glycero-3-phosphocholine (DLPC), 1,2-dipalmitoyl-sn-
glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-
yl) (NBD-Cap-PE), and N-Biotinyl-Cap-PE (16:0) (Avanti Polar
Lipids, Inc, Alabster, AL). Texas Red-labeled streptavidin, strepta-
vidin-conjugated alkaline phosphatase (1:1 stoichiometry), strepta-
vidin-conjugated horseradish peroxidase (1:1 stoichiometry),
Amplex Red, and 4-methylumbelliferyl phosphate were purchased
from Molecular Probes (Eugene, OR). This particular phosphate
derivative was chosen because dephosphorylation generates the
highly fluorescent soluble product, 7-hydroxy-4-methylcoumarin.
AvidinD-conjugated glucose oxidase (1:1 stoichiometry, 0.5 mg/
mL) was purchased from Vector Laboratories (Burlingame, CA).
Cleaning solution (7×) came from ICN Biomedicals, Inc, while
Fig u re 1 . Schematic diagram of streptavidin-conjugated alkaline
phosphatase bound to a supported lipid bilayer containing biotinylated
lipids. The declivities in the enzyme represent its two active sites,
while the four declivities in the streptavidin represent its four biotin
binding sites.
bovine serum albumin (BSA, fraction V, 99%) and R-
D
-glucose were
-glucose
purchased from Sigma. R- -Glucose equilibrates with â-
D
D
in water. Borosilicate capillary tubes (5 µL, 388-µm i.d.) were
obtained from VWR. Photoresist Microposit S 1813 and Microposit
Developer were purchased from Shipley Co. (Marlborough, MA).
Syringe needles came from Becton Dickinson & Co. (Franklin
Lakes, NJ).
The strategy adopted for this work was to employ biotin-
streptavidin chemistry in conjunction with supported phospholipid
membranes for surface immobilization of the enzymes1
6-18
(Figure
P reparation of Small Unilamellar Vesicles. SUVs formed
from phospholipids, fluorescently labeled phospholipids, and
biotinylated phospholipid mixtures were prepared using the
Barenholz method.19 Briefly, lipids dissolved in chloroform were
mixed together to the desired mole ratio. The solvent was
evaporated in a stream of dry nitrogen followed by vacuum
evaporation for an additional 4 h. The dried lipids were reconsti-
tuted in phosphate buffer (pH 7.4, ionic strength 150 mM) and
sonicated with a titanium tip to form SUV solutions. The samples
were centrifuged at 38 000 rpm (94500g) for 30 min, and then the
supernatant was centrifuged again at 52 000 rpm (176900g) for 3
h using a Beckman ultracentrifuge L7 with a Ti-75 rotor.
Forming P hospholipid Bilayers inside Capillary Tubes.
Borosilicate capillary tubes were cleaned in boiling surfactant (4:1
mixture of water and 7× cleaning solution) for 30 min. After
extensive rinsing with water from a NANOpure Ultrapure water
system (Barnstead, Dubuque, IA), they were baked in a kiln for
1
). We have previously shown that lipid bilayers containing specific
recognition elements can be completely coated on the inside of
PDMS microfluidic channels bound to glass supports.12 The
bilayer coating is formed by the fusion of small unilamellar vesicles
(
SUVs) to the surfaces of both the glass and PDMS, if the latter
is rendered hydrophilic. When a small percentage of the lipid
headgroups is biotinylated, an enzyme of interest can be bound
to the surface, if it is in turn linked to streptavidin. In this case,
the lipid bilayer serves as a buffer to prevent the protein from
denaturing on the underlying surface and could, under proper
circumstances, provide a method for obtaining uniform orientation.
Indeed, this method has been shown to be a viable means of
maintaining a high degree of protein structure and orientation in
corresponding planar systems.17,18 Furthermore, because the
bilayer is two-dimensionally fluid, enzymes can reorganize at the
interface to achieve high packing densities. It may also be possible
to modify bilayer-based assays to accommodate membrane
proteins.
5
h at 400 °C and cut into 1-cm-length pieces using a glass cutter.
DLPC SUVs, which contained 3 mol % NBD-PE as fluorescent
tags, were made to flow through the capillary tubes where they
fused to the surface to form single planar supported membranes.
Excess vesicles were washed out with fresh PBS buffer (pH 7.4).
The fluidity of the supported phospholipid bilayers inside the
microcapillary tubes was confirmed by fluorescence recovery after
photobleaching (FRAP)20,21 using an upright epifluorescence
This paper is laid out in the following manner. In the first
section, we immobilized phosphatase enzymes inside glass mi-
crocapillary tubes containing no PDMS. The turnover rate of the
catalysts were measured and compared with results in the second
section for enzymes inside an array of PDMS microchannels. The
results indicated that the turnover rate for dephosphorylation was
(
(
(
16) Darst, S. A.; Ahlers, M.; Meller, P. H.; Kubalek, E. W.; Blankenburg, R.;
Ribi, H. O.; Ringsdorf, H.; Kornberg, R. D. Biophys. J. 1 9 9 1 , 59, 387-396.
17) Herron, J. N.; Muller, W.; Paudler, M.; Riegler, H.; Ringsdorf, H.; Suci, P.
A. Langmuir 1 9 9 2 , 8, 1413-1416.
(19) Barenholz, Y.; Gibbes, D.; Litman, B.; Goll, J.; Thompson, T.; Carlson, F.
Biochemistry 1 9 7 7 , 16, 2806-2810.
(20) Axelrod, D.; Koppel, D. E.; Schlessinger, J.; Elson, E.; Webb, W. W. Biophys.
18) Edmiston, P. L.; Saavedra, S. S. J. Am. Chem. Soc. 1 9 9 8 , 120, 1665-1671.
J. 1 9 7 6 , 16, 1055-1069.
380 Analytical Chemistry, Vol. 74, No. 2, January 15, 2002

microscope (E800, Nikon) equipped with a 1024 Micromax CCD
camera. Images were captured with Metamorph software from
Universal Imaging Corp.
Forming P hospholipid Bilayers inside P oly(dimethylsi-
loxane) Microchannels. The formation of phospholipid bilayer-
coated microchannels has been described elsewhere.12 Briefly,
channels were formed by placing lithographically patterned poly-
(
dimethylsiloxane) (PDMS) molds (Dow Corning Sylgard Silicone
Elastomer-184, Krayden, Inc.) into conformal contact with planar
borosilicate coverslips. The molds were made by curing the low
molecular weight polymer with a cross-linker on a photolitho-
graphically patterned master prepared by exposing and developing
a photoresist patterned surface. The elastomeric mold, which bore
the negative pattern of the master, was carefully peeled off and
washed repeatedly with acetone and ethanol. Before contact was
made with the glass substrate, the PDMS surface was rendered
hydrophilic by oxygen plasma treatment for 15 s (PDC-32G plasma
cleaner, Harrick Scientific, Ossining, NY). Glass coverslips, which
served as floors for the microchannels, were cleaned in hot
surfactant solution (ICN ×7 detergent, Costa Mesa, CA), rinsed
at least 20 times in purified water from the NANOpure Ultrapure
Water System and then baked in a kiln at 400 °C for 4 h before
use. A Harvard PHD 2000 syringe pump (Harvard Apparatus,
Holliston, MA) was employed to infuse all solutions into the PDMS
microchannels. Supported bilayer formation and quality inside the
channels was confirmed by FRAP and quantitative fluorescence.
Fig u re 2 . Fluorescence from the production of 7-methoxy-4-
methylcoumarin versus reaction time from phosphatase enzymes
immobilized on lipid bilayers containing various amounts of biotin.
Each reaction was performed with an initial concentration of 0.136
mM 4-methylumbelliferyl phosphate.
RESULTS AND DISCUSSION
Immobilization and Characterization of Phosphatase Cata-
lyst in Microcapillary Tubes. Vesicles containing 0.1 mol %
biotinylated lipids were fused to the microcapillary walls as
described above in the Experimental Section. A solution of 5 mg/
mL BSA in PBS buffer (pH 3.8) was injected and incubated for at
least 1 h. This ensured that defect sites in the bilayer would be
passivated before enzyme introduction. The BSA solution was
Fig u re 3 . Fluorescence intensity from Texas Red-labeled strepta-
vidin bound to DLPC supported bilayers as a function of biotin
concentration in the bilayers. The initial concentration of streptavidin
in the bulk solution was 0.01 mg/mL.
2 3 3
washed out using a pH 9.8 buffer (Na CO / NaHCO , ionic
strength 150 mM). The tube was incubated in the buffer overnight,
as waiting several hours allowed BSA to fully reorganize at the
interface. At this point, a solution of streptavidin-conjugated
alkaline phosphatase (0.02 mg/ mL) was made to flow into the
tubes, incubated for 15 min, and washed out with the same pH
T
It is necessary to estimate the total enzyme concentration, [E] ,
inside the microcapillary tubes in order to find the absolute
turnover rate of the catalyst. An approximate measure can be
obtained by comparing the relative footprint sizes of streptavidin-
conjugated phosphatase to streptavidin alone, since the size of
the latter is well known. To achieve this, binding experiments of
fluorescently labeled streptavidin were performed as a function
of the mole percent of biotin in the membrane. The results are
plotted in Figure 3. As can be seen from the data, the maximum
amount of streptavidin that could be bound to the surface
continued to rise with increasing biotin concentration until ∼4
mol % and then saturated. This indicated that phosphatase-linked
streptavidin occupied ∼4 times greater surface area per molecule
than streptavidin alone. Such a result is in line with expectations,
as streptavidin has a molecular weight of 60 000, while the
phosphatase (alone) is ∼140 000 and the conjugate is 200 000.
Two-dimensionally crystalline streptavidin has a 5.8 nm × 5.8 nm
unit cell.22 The footprint of crystalline streptavidin (33.6 nm )
therefore provides a lower bound for streptavidin-linked phos-
9.8 buffer. This relatively high protein concentration was employed
to quickly saturate the surface biotin binding sites, and the pH of
the washing buffer was chosen for maximum catalytic activity. In
the next step, 0.136 mM 4-methylumbelliferyl phosphate in pH
9.8 buffer was added to an enzyme-coated capillary tube. The tube
was placed under a fluorescence microscope, and the rate of
product formation (7-hydroxy-4-methylcoumarin) was monitored
by following the change in blue fluorescence intensity as a function
of time (Figure 2, O). The same reaction then was repeated with
several different concentrations of biotinylated lipids in the
membrane (Figure 2). As can be clearly seen from the data, the
observed reaction rate increased as the biotin concentration was
raised from 0.1 to 1 mol % and leveled off at this value for higher
concentrations. It should be noted that even the 0.0 mol % bilayer
gave a small residual response, which was probably due to a tiny
amount of enzyme nonspecifically adsorbed to the underlying
substrate.
2
2
phatase of ∼4 times this value, 134 nm .
(
21) Soumpasis, D. M. Biophys. J. 1 9 8 3 , 41, 95-97.
(22) Kim, E. E.; Wyckoff, H. W. J. Mol. Biol. 1 9 9 1 , 218, 449-464.
Analytical Chemistry, Vol. 74, No. 2, January 15, 2002 381

enzymes. The first stems from the obstruction of active sites at
the interface. Steric effects have often been observed for im-
mobilized proteins when their binding pockets face toward the
surface.24 As shown in Figure 1, the alkaline phosphatase used
here had two active sites for dephosphorylation. Depending upon
the packing density and protein orientation, some of these may
not have been available for reaction. Furthermore, since the
streptavidin-conjugated alkaline phosphatase was prepared by
generic chemical cross-linking, the active sites on even neighbor-
ing alkaline phosphatases may have had different orientations
relative to the underlying streptavidin. If these steric effects
strongly blocked the substrate from getting to the most buried
locations, this could have caused a significant reduction in turnover
rate.
Fig u re 4 . Lineweaver-Burk plots of the reciprocal initial reaction
rates, 1/v0, versus the reciprocal of substrate concentration, 1/[S],
for soluble, O, and immobilized, b, streptavidin-conjugated alkaline
phosphatase.
Another reason for the apparent difference in turnover between
the immobilized enzymes and those in solution probably stemmed
from an overestimation of enzyme number density at the surface,
leading to an underestimation of kcat in the immobilized case. Two
important factors could have contributed to this. First, the footprint
size of the streptavidin-conjugated phosphatase was almost
certainly underestimated because of the use of crystalline strepta-
vidin dimensions. Furthermore, our calculations assumed 100%
enzyme coverage on the inside of the microcapillary tubes. This
was not completely accurate as BSA rather than the supported
lipid bilayers covered some portions of the surface. It therefore
should be stated that the factor of 6 difference in turnover rate
represented an upper bound to the difference. Finally, any portion
of the bound enzyme that was denatured by this interface would,
of course, have led to a decrease in the measured turnover rate.
Immobilization and Characterization of Phosphatase Cata-
lyst in Microchannels: One-Shot Lineweaver-Burk Studies.
The next set of dephosphorylation reactions was carried out inside
a parallel array of bilayer-coated PDMS microchannels (Figure
To obtain quantitative information on the turnover rate of the
bilayer-supported enzyme, the initial reaction rate, v
sured as a function of substrate concentration, [S]. v
0
, was mea-
was
0
determined by finding the slope (extrapolated to the origin) during
the initial stage (first 300 s) for the change in fluorescence
intensity versus time at each concentration tested. These data were
plotted using the Lineweaver-Burk equation:23
1
/ v ) (K / Vmax)/ [S] + 1/ Vmax
0 M
M
where K is the Michaelis constant and Vmax is the velocity of
the reaction when the active sites of the enzyme are saturated
with substrate. A series of substrate concentrations ranging from
0.273 to 3.41 mM were made to flow into capillary tubes containing
streptavidin-linked phosphatase on a 3 mol % biotinylated mem-
brane (Figure 4, b). The results revealed that Vmax ) 7.7 ( 1.2
5
). The apparatus was designed to simultaneously measure the
µM/ s and K
cat, for this reaction, it was necessary to obtain a reasonable
estimate of [E] , as the turnover number is calculated from kcat
M
) 12.5 ( 2.9 mM. To calculate the turnover number,
reaction rates of 12 independent concentrations of substrate
created by series dilution from the introduction of a single aliquot.
k
T
)
Several methods for designing concentration diluters have already
V
max/ [E]
T
. The surface area inside the capillary tube was 1.22 ×
been shown for microfluidics.2
5-27
In our apparatus,28 two streams
2
-13
1
013 nm , so ∼1.5 × 10 mol of the enzyme could adsorb inside,
of liquid were combined at a Y-junction and allowed to diffuse
into each other as they flowed downstream side by side.29 Such
diffusional mixing was possible because the Reynolds number
inside the channel was sufficiently low to prevent turbulence. After
a fixed distance (2 cm), the main channel was partitioned into a
series of smaller channels that emanated from the main flow
stream. Substrate could be infused in one inlet port while pure
buffer was injected into the other. This method created a dilution
series that ranged over a factor of 33 in concentration from the
first to the last microchannel. At the opposite end of the device,
the channels were recombined into a common outlet. Inspired
using the footprint size obtained above. Because the volume of
-
6
each capillary tube was 1.2 × 10 L, the effective total enzyme
-
7
concentration was 1.3 × 10 M, leading to a turnover number,
cat ) 60.6 ( 9.4 s-1
The turnover number for the immobilized enzymes needs to
k
.
be compared to enzymes in solution to determine what fraction
of activity was maintained on the surface. The bulk solution value
was measured using the same Lineweaver-Burk analysis with a
-
7
streptavidin-phosphatase concentration of 0.02 mg/ mL (1 × 10
M) and substrate concentrations ranging from 0.136 to 2.73 mM.
The data are plotted as open circles (O) in Figure 4. In this case,
V
max ) 36.5 ( 6.9 µM/ s and K
M
) 14.0 ( 2.7 mM. The turnover
(
24) Spinke, J.; Liley, M.; Guder, H. J.; Angermaier, L.; Knoll, W. Langmuir 1993,
, 1821-1825.
(25) Jacobson, S. C.; McKnight, T. E.; Ramsey, J. M. Anal. Chem. 1 9 9 9 , 71,
455-4459.
-
1
number in bulk solution was kcat ) 365 ( 69 s
.
9
From the calculations above, the turnover rate in solution was
roughly a factor of 6 greater than the surface-bound enzyme.
There are probably several factors contributing to the apparent
difference in reaction rate between the immobilized and soluble
4
(
26) Jeon, N. L.; Dertinger, K. W.; Chiu, D. T.; Choi, I. S.; Stroock, A. D.;
Whitesides, G. M. Langmuir 2 0 0 0 , 16, A.-F.
(27) Dertinger, S. K.; Chiu, D. T.; Jeon, N. L.; Whitesides, G. M. Anal. Chem.
0 0 1 , 73, 1240-1246.
2
(
23) Voet, D.; Voet, J. G. Biochemistry, 2nd ed.; John Wiley and Sons: New York,
995.
(28) Holden, M. A.; Jung, S. Y.; Cremer, P. S., to be published.
(29) Weigl, B. H.; Yager, P. Science 1 9 9 9 , 283, 346-347.
1
382 Analytical Chemistry, Vol. 74, No. 2, January 15, 2002

Fig u re 5 . Schematic diagram of the apparatus used for a one-shot
measurement of phosphatase activity. Twelve channels (100 µm wide
×
10 µm high) spaced 50 µm apart are connected to Y-shaped inlets
through a 2 cm × 600 µm (length × width) mixing channel.
by the work of Unger and co-workers,30 simple mechanically
operated switches were designed by making elliptical holes in a
plane ∼2 mm above the 12-channel array and inserting elliptical
metal needles whose shapes matched those of the holes. Turning
the needles such that the wide axis was parallel to the array
allowed liquid to flow inside the channels. Alternatively, when the
needles were turned such that the wide axis was normal to the
array, the top of the channels pressed against the floor causing
complete shutoff of liquid flow.
Fig u re 6 . (a) Fluorescence intensity of 7-methoxy-4-methylcoumarin
produced in the first 300 s of reaction by the phosphatase immobilized
inside 12 channels: b, 1; O, 2; 1, 3; 3, 4; 9, 5; 0, 6; [, 7; ], 8; 2,
9; 4, 10; `, 11; ", 12. The inset shows the same data set for the
entire length of the experiment (∼3000 s) (b) Lineweaver-Burk plot
of the reciprocal initial reaction rates obtained from (a) versus the
reciprocal of substrate concentrations for the immobilized streptavidin-
conjugated phosphatase inside the arrayed microfluidic channels. The
concentrations of 7-methoxy-4-methylcoumarin ranged from 0.103 to
The device was operated in the following manner. Small
unilamellar DLPC vesicles containing 3 mol % biotinylated lipids
flowed through the two inlet ports and fused to the walls and floor
of the device to create a uniform supported bilayer coating. Excess
vesicles were then washed out by flowing pure buffer. The
channels were passivated with BSA in the same fashion as
described above for the microcapillary tubes and incubated
overnight. At this point, switch 1 (Figure 5) was closed and
streptavidin-conjugated phosphatase (0.2 mg/ mL, pH 9.8) was
added through the common outlet channel. Mixing occurred by
continuously rotating switch 2 between the open and closed state
for a period of 10 min. The switching process created a series of
squeeze/ release events, which efficiently mixed the protein
solution into the channels. This procedure allowed the phos-
phatase to coat the microchannels between switch 1 and the
common outlet, while leaving the front of the device protein free.
After protein binding was completed, both switches were rotated
to the open position and pure buffer was used to flow out excess
enzyme. Following this, a solution of 3.41 mM 4-methylumbel-
liferyl phosphate in pH 9.8 buffer was pumped through one inlet
port while pure buffer was pumped through the other. After
3
.41mM.
allowing sufficient time to completely fill the channel array with
substrate solution (several minutes), both switches were closed,
the pump was turned off, and the fluorescence intensity of the
product in all channels was followed simultaneously by epifluo-
rescence microscopy. The rate of product formation as a function
of time is plotted in Figure 6a.
0
The initial reaction rate (v ) was obtained by measuring the
slope of the change in fluorescence intensity versus time for each
channel in Figure 6a for the first 300 s of the experiment as before.
To determine the initial concentration of substrate in each channel,
the total fluorescence intensity upon completion of the reaction
(
inset of Figure 6a) was compared to that of known product
concentrations and it was assumed that the reaction went to
completion in each case. Control experiments revealed that the
fluorescence intensity was linearly related to the initial substrate
concentration under these conditions. The data obtained from
these measurements were then utilized to make a Lineweaver-
Burk plot (Figure 6b). The results revealed that kcat ) 51.1 ( 3.2
s-1 and K
M
) 16.2 ( 1.3 mM, which were the same as the values
obtained inside the microcapillary tubes within experimental error
(Figure 4). This demonstrated that the combined PDMS-glass
(
30) Unger, M. A.; Chou, H. P.; Thorsen, T.; Scherer, A.; Quake, S. R. Science
0 0 0 , 288, 113-116.
2
Analytical Chemistry, Vol. 74, No. 2, January 15, 2002 383

Fig u re 8 . Line scan of resorufin fluorescence intensity across
channels A2 and B2.
A schematic diagram of the microfluidic device used for the
two-step reaction is shown in Figure 7b. It was capable of running
two experiments simultaneously. On one side (A) the experiment
was run in the presence of glucose and on the other (B) the
chemistry was identical, but no glucose was used. Each micro-
channel was divided into two parts (A1 and A2, B1 and B2) and
connected by a reversibly attachable U-shaped plastic tube (700
µm in diameter). With the tubes unattached, SUVs containing 3
mol % biotinylated lipids were fused to the channel surfaces
throughout the device, excess SUVs were removed, and the whole
device was passivated with BSA. After washing out the excess
BSA with pH 7.4 PBS buffer, avidinD-conjugated glucose oxidase
(0.5 mg/ mL) was made to flow into A1 and B1, while streptavidin-
conjugated horseradish peroxidase (0.2 mg/ mL) was infused into
A2 and B2. Both were incubated in their respective channels for
20 min in pH 7.4 PBS buffer solution and washed out for 40 min
in pure buffer. At this point, the U-shaped tubes were attached.
Fig u re 7 . (a) Reaction scheme for the sequential reactions
catalyzed by glucose oxidase and horseradish peroxidase. (b)
Schematic representation of the microfluidic device used for the two-
step glucose detection reaction.
surface was approximately as amenable for preserving enzyme
activity upon immobilization as were the simpler microcapillary
tubes. It should be further noted that the signal-to-noise ratio from
the microchannel experiments was superior to that of the
microcapillary tube data (Figures 4 and 6). Indeed, the error bars
for this experiment were approximately the size of the filled circles
used to plot the data. This was only about one-third the size of
those from the microcapillary tube measurements. One important
reason for the difference is the fact that all the data were collected
simultaneously. Errors, therefore, associated with flicker in the
Hg arc lamp in producing the fluorescent signal would be greatly
reduced. Furthermore, taking all the data simultaneously took ∼1
order of magnitude less time than preparing and measuring each
stable concentration separately. The assay was stable enough that
the product could be flowed out, new substrate introduced, and
the experiment repeated. This was done up to 10 times over the
course of 2 days without noticeable changes in the data obtained.
Exploiting Immobilized Enzymes inside Microchannels
To P erform a Coupled Set of Chemical Reactions. The
methods presented above for enzyme immobilization inside
microfluidic channels affords the opportunity to run multistep
chemical reactions within lab-on-a-chip devices. Substrates can be
introduced over a first immobilized enzyme and the products may
then continue downstream to additional sets of enzymes for further
reactions. Indeed, it could be easily envisioned that complex
multistep syntheses could be performed in this manner. To
demonstrate this principle with the chemistry developed above,
we chose a simple two-step process often exploited for the
detection of glucose (Figure 7a). The process operates by
To commence the reactions, freshly prepared â-D -glucose (10
mg/ mL in PBS 7.4 buffer, pregassed with oxygen) was infused
into the inlet of A1 while pure buffer was infused into B1. Amplex
Red (1 mM in DMSO/ H O ) 1/ 10) flowed in via branched
2
sidearms of A2 and B2 where it met up with any products of A1
and B1 and flowed together into A2 and B2. The red fluorescent
signal of the resorufin product was monitored by fluorescence
microscopy. The line scan of the fluorescence across the channels
is shown in Figure 8. The results indicated that the fluorescence
intensity from the control channel (B) was ∼2% of that from
experimental channel. This residual intensity in the control
channel was due to the fact that Amplex Red itself is weakly
fluorescent. The entire experiment was repeated by introducing
both the Amplex Red and glucose through A1 and Amplex Red
through B1. The results were identical. This procedure clearly
demonstrated the utility of employing immobilized enzymes in
microfluidic channels for carrying out multistep chemical reac-
tions.
CONCLUSION
In the experiments presented here, enzymes were immobilized
in phospholipid bilayer-coated microcapillary tubes and micro-
fluidic channels. The turnover numbers for phosphatase im-
mobilized inside microcapillary tubes and microfluidic channels
were similar and represented a reasonable fraction of the corre-
sponding bulk value. A one-shot Lineweaver-Burk plot using
employing glucose oxidase in a first step to oxidize â- -glucose
D
to δ- -gluconolactone and hydrogen peroxide. The H O is then
D
2 2
used in a second step to convert Amplex Red, a molecule with
little fluorescence, to the highly fluorescent product, resorufin.
384 Analytical Chemistry, Vol. 74, No. 2, January 15, 2002

arrayed microfluidic channels demonstrated the feasibility of using
microchannels to obtain kinetic data rapidly with a high S/ N ratio.
Serial enzyme reactions inside a microfluidic device were estab-
lished to show the potential application of this general methodol-
ogy to multistep chemical synthesis.
nology Program (Grant 010366-0181-1999), startup monies from
Texas A&M University, a Nontenured Faculty Award from 3M
Corp., and a Research Innovation Award from the Research
Corporation of America (RI0437).
Received for review July 20, 2001. Accepted October 10,
ACKNOWLEDGMENT
2001.
This work was supported by ARO (DAAD19-01-1-0346), an
ONR-YIP Award (NOOO14-00-1-0664), the Texas Advanced Tech-
AC010822U
Analytical Chemistry, Vol. 74, No. 2, January 15, 2002 385