Self-powered enzyme micropumps

Article abstract of DOI:10.1038/nchem.1895

Non-mechanical nano- and microscale pumps that function without the aid of an external power source and provide precise control over the flow rate in response to specific signals are needed for the development of new autonomous nano- and microscale systems. Here we show that surface-immobilized enzymes that are independent of adenosine triphosphate function as self-powered micropumps in the presence of their respective substrates. In the four cases studied (catalase, lipase, urease and glucose oxidase), the flow is driven by a gradient in fluid density generated by the enzymatic reaction. The pumping velocity increases with increasing substrate concentration and reaction rate. These rechargeable pumps can be triggered by the presence of specific analytes, which enables the design of enzyme-based devices that act both as sensor and pump. Finally, we show proof-of-concept enzyme-powered devices that autonomously deliver small molecules and proteins in response to specific chemical stimuli, including the release of insulin in response to glucose.

Full text of DOI:10.1038/nchem.1895

ARTICLES
PUBLISHED ONLINE: 30 MARCH 2014 | DOI: 10.1038/NCHEM.1895
Self-powered enzyme micropumps
Samudra Sengupta , Debabrata Patra , Isamar Ortiz-Rivera , Arjun Agrawal , Sergey Shklyaev ,
1
†
1†
1†
1
2
1
3
1
1
Krishna K. Dey , Ubaldo C o´ rdova-Figueroa , Thomas E. Mallouk and Ayusman Sen
*
*
Non-mechanical nano- and microscale pumps that function without the aid of an external power source and provide
precise control over the flow rate in response to specific signals are needed for the development of new autonomous nano-
and microscale systems. Here we show that surface-immobilized enzymes that are independent of adenosine triphosphate
function as self-powered micropumps in the presence of their respective substrates. In the four cases studied (catalase,
lipase, urease and glucose oxidase), the flow is driven by a gradient in fluid density generated by the enzymatic reaction.
The pumping velocity increases with increasing substrate concentration and reaction rate. These rechargeable pumps can
be triggered by the presence of specific analytes, which enables the design of enzyme-based devices that act both as
sensor and pump. Finally, we show proof-of-concept enzyme-powered devices that autonomously deliver small molecules
and proteins in response to specific chemical stimuli, including the release of insulin in response to glucose.
elf-powered nano- and micropumps that precisely control flow molecules and proteins (for example, insulin) in response to a
rate in response to external stimuli are critical to the design of specific chemical trigger (for example, glucose). This opens up the
S
the next generation of smart devices. Ideally, the pump should possibility of designing novel stimuli-responsive autonomous
enable fluid flow to be controlled both by the presence and concen- cargo and drug-delivery systems .
tration of the specific analyte, such as a substrate, promoter (cofac-
3
2
tor) or a related biomarker. This coupling between sensing and Results and discussion
transport should allow new applications such as the bottom-up We designed triggered fluid pumps using four different classes of
assembly of dynamic structures, cargo delivery at specific locations enzymes. Gold (Au) was patterned on a polyethylene glycol
1,2
(
for example, drug delivery) and related functions .
Recently, autonomous motion that arises from the catalytic har- alized with a quaternary ammonium thiol , which formed a self-
(PEG)-coated glass surface. Next, the patterned surface was function-
3
3
nessing of chemical free energy from the surrounding environment assembled monolayer (SAM) on the Au surface. On incubating
has been demonstrated at the nano- and microscale1 . Tethering the SAM-modified Au surface with enzyme, the negatively
these catalytic systems to surfaces enables the transfer of the mech- charged groups on the enzyme bound selectively to the modified
anical force to the surrounding fluid. Utilizing this, a variety of Au surface via electrostatic self-assembly, which resulted in an
pumps have been designed and fabricated that operate on the micro- enzyme pattern on the glass surface (Fig. 1a, Supplementary
scale and function as delivery vehicles for fluids, small molecules Fig. 1). To demonstrate the pumping ability of immobilized
and colloids1 . A major drawback of these artificial micropumps enzymes, a spacer (20 mm diameter, 1.3 mm height) was placed
is their non-biocompatibility with respect to the catalyst, fuel or on top of the enzyme-patterned surface to seal the pump chamber
even the ionic-strength regime in which they operate. Biocatalysts, and create a closed system. A buffered solution of substrate with sus-
–10
1–28
such as enzymes, provide an obvious solution to this problem.
pended tracer particles was injected into the chamber and the fluid
The diffusive mobility of single enzyme molecules has been flow was monitored with an optical microscope.
shown to increase in the presence of a substrate in a concentration-
dependent manner and, when exposed to a substrate concentration Pumping in the presence of the substrate. We examined catalase34
2
9–31
gradient, ensembles of enzyme molecules exhibit chemotaxis
.
as our first example of an ATP-independent, enzyme-powered
Enzymes that move by generating a continuous surface force in a micropump. The enzyme was immobilized selectively on the Au
fluid should, when fixed in place, function as micropumps that pattern (6 mm diameter) as described above and sulfate-
move fluid and colloidal particles in a directed manner. Here we functionalized polystyrene microspheres, 2 mm in size, were used as
describe a novel enzyme-based platform that combines sensing tracer particles to analyse the fluid flow. In the presence of
and microfluidic pumping into a single self-powered microdevice. substrate (hydrogen peroxide) the tracer particles moved towards
These micropumps are substrate specific and their pumping speed the Au surface, which indicates that the surrounding fluid is
can be tuned by altering the substrate concentration. In addition, pumped inwards (Supplementary Movie 1). The fluid flow was
the pumps can be triggered by different analytes that, in situ, observed in a closed system, so by fluid continuity the fluid
generate the substrate for the respective enzymes. This approach flow showed an outwards motion when viewed above the
demonstrates that even adenosine triphosphate (ATP)-independent enzyme-patterned surface. The fluid-pumping velocity showed a
enzymes can generate movement and can be used as micropumps in substrate concentration and a reaction-rate-dependent increase
2
1
the presence of the appropriate fuel (substrate). Finally, we show from 0.37 mm s
in 0.001 M hydrogen peroxide (enzymatic
21 21
that an enzyme-immobilized hydrogel can be used as a scaffold reaction rate (n), 12.60 mM s ) to 4.51 mm s in 0.1 M hydrogen
for micropumps that actively pump out gel-entrapped small peroxide (n, 613.5 mM
21
s ) (Fig. 2a). The Supplementary
1
2
Department of Chemistry, The Pennsylvania State University, University Park, Pennsylvania 16802, USA, Institute of Continuous Media Mechanics, Ural
3
Branch of the Russian Academy of Sciences, Perm 614013, Russia, Department of Chemical Engineering, University of Puerto Rico-Mayag u¨ ez, Mayag u¨ ez,
PR 00681, Puerto Rico; These authors contributed equally to this work. *e-mail: asen@psu.edu; tem5@psu.edu
†
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
1
©
2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
a
Au pattern
6
mm
Au
Enzyme
SAM formation
Stability
Tracer
particles
Substrate
for enzyme
Anchoring group
HS
Biocompatibility Interacting group
+
N
O
O
Spacer
O
O
Catalase, urease, lipase, glucose oxidase
Convective flow
b
Glucose oxidase
Glucose
H O
2
Glucose
Catalase
2
Spacer
Spacer
Tracer particles
Immobilized catalase
on surface
Triggered pumping
Figure 1 | Schematic that shows the enzyme pattern on a surface and triggered fluid pumping by enzymatic micropumps. a, Au was patterned onto a
PEG-coated glass surface using an e-beam evaporator. The patterned surface was functionalized with a quaternary ammonium thiol, which forms a self-
assembled monolayer (SAM) on the Au surface. The negatively charged groups on the enzyme bound selectively to the SAM-functionalized Au patterned
surface via electrostatic assembly, which resulted in an enzyme pattern on the surface. b, Catalase enzyme immobilized on the Au pattern causes fluid
pumping triggered by the presence of both GOx and glucose, which generates hydrogen peroxide in situ.
Information gives detailed calculations of the reaction rates: the kcat Temporal and spatial variations in pumping. The temporal
(turnover number) and KM (substrate concentration at which the velocity profile was investigated for all four enzyme-powered
reaction rate is half of the maximum rate for the system) values pumps over both short and long time intervals. In the case of
used are for enzymes in solution; these values will be different for catalase, fluid pumping was monitored for a duration of ten
immobilized enzymes that are restricted dimensionally. No fluid minutes, at a distance 50–100 mm away from the enzyme
pumping was observed in the absence of substrate (Supplementary pattern and for time intervals of one minute. No significant
3
5
Movie 2). Similar pumping behaviour was also observed for lipase
change in the velocity of tracer particles was observed at each
and glucose oxidase (GOx)36 in the presence of their respective of the three different concentrations of hydrogen peroxide
substrates, 4-nitrophenyl butyrate and glucose, with inwards fluid (10 mM, 50 mM and 100 mM) within the time frame of ten
flow near the Au surface and an outwards flow when viewed above minutes (Supplementary Figs 2 and 3). Similar time-dependent
the surface. As with catalase, the pumping velocity increased with studies of pumping speed with urease in 0.75 M urea, GOx in
increasing substrate concentration and, in turn, enzymatic reaction 1 M glucose and lipase in 0.5 M 4-nitrophenyl butyrate showed
rates in general (Fig. 2c,d).
no appreciable change in pumping velocity at short time
The opposite fluid flow was observed for urease anchored to the intervals (Supplementary Fig. 2). As expected, over longer time
Au surface. Close to the glass surface, the tracer particles moved scales, the pumping velocity decreased. As the substrate was
away from the Au pattern, which indicates that the surrounding consumed, the reaction rate decreased and thereby slowed the
fluid was pumped outwards (Supplementary Movie 3). When fluid-pumping speed. This was demonstrated with catalase in
viewed upwards in the solution (away from the glass surface), by the presence of 0.050 M hydrogen peroxide at monitored at
fluid continuity an inwards fluid flow was observed. As expected, regular time intervals of 30 minutes for a duration of four
the pumping velocity increased on increasing substrate concen- hours (Fig. 3a). Similar behaviour was observed with urease-
2
1
21
tration from 0.24 mm s in 0.001 M (n, 44.83 mM s ) urea to powered pumps in the presence of 1 M urea (Supplementary
0
21
21
.80 mm s in 0.75 M urea (n, 102.9 mM s ) (Fig. 2b). No fluid Fig. 4). Significantly, these pumps can be recharged by
pumping was observed in the absence of urea.
introducing fresh substrate solution after the initial substrate
The newly designed enzyme-powered micropumps clearly have solution is exhausted and fluid pumping stops (Supplementary
the ability to sense substrate in the surrounding media and initiate Fig. 5). For both catalase and urease, fluid pumping resumed
fluid pumping in response. Using glucose and GOx, fluid pumping with a velocity similar to that observed previously at that
in the catalase pump was triggered by in situ generation of hydrogen specific substrate concentration.
peroxide (Fig. 1b, Supplementary Movie 4). In the presence of 50 mM
The spatial velocity profile was also examined for each of these
glucose and 0.1 mM GOx, with catalase immobilized on the Au enzyme pumps. The fluid-pumping velocity was examined at set
21
pattern, the fluid was pumped inwards at a speed of 1.2 mm s
.
distances moving away from the enzyme-functionalized Au
Pumping was not observed in the absence of either glucose or pattern. At shorter distances (50–400 mm), the pumping velocities
GOx, or both. Thus, in principle, the enzyme pumps can be did not show significant variations for catalase-, urease-, GOx-
triggered by a variety of analyte molecules, which opens up the and lipase-powered pumps (Supplementary Fig. 6). As expected,
possibility of designing enzyme-based devices that act both as the pumping velocity decreases at longer distances, as observed
sensor and as pump.
for catalase- (Fig. 3b) and urease-powered (Supplementary Fig. 4)
2
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
©
2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
a
b
Catalase
Urease
+
–
H O
H O + O
(NH ) CO
NH4 + HCO3
2
2
2
2
2 2
[Hydrogen peroxide] (M)
[Urea] (M)
0
.001
0.010 0.050
0.100
0.001
0.010
0.100
0.250
0.500
0.750
1.0
0.8
0.6
0.4
0.2
0.0
5
4
3
2
1
0
12.60
115.0
414.0
613.5
44.83
91.25
101.8
102.6
102.8
102.9
Reaction rate (μM s–1
)
Reaction rate (μM s–1
)
c O
O
O
OH HO
+
d
Glucose
oxidase
Lipase
NO2
NO2
β-D-glucose + O2
Gluconic acid + H O
2 2
[
4-nitrophenyl butyrate] (M)
[Glucose] (M)
0.010 0.100
0.001
0.010
0.100
0.500
0.001
1.000
0
0
0
0
0
0
0
0
.7
.6
.5
.4
.3
.2
.1
.0
1.0
0.8
0.6
0.4
0.2
0.0
0.01706
0.02311
0.02400
0.02404
0.02937
0.2276
0.7004
0.8841
Reaction rate (μM s–1
)
Reaction rate (μM s–1
)
Figure 2 | Fluid pumping velocity in an enzyme-powered micropump as a function of substrate concentration and reaction rate. a, Pumping velocity in a
catalase-powered micropump increases in the presence of its substrate in a reaction-rate-dependent fashion, at substrate concentrations that ranged from
0
.001 M to 0.1 M hydrogen peroxide. b, Pumping velocity in a urease-powered micropump increases on increasing the substrate concentration from 0.001 M
to 0.75 M urea. c, Pumping velocity in a lipase-powered micropump shows a concentration-dependent increase at substrate concentrations from 0.001 M to
.5 M 4-nitrophenyl butyrate. d, Pumping velocity in a GOx-powered micropump increases in a substrate concentration- and reaction-rate-dependent manner
0
from 0.001 M to 1 M glucose. The reaction-rate calculations are based on kcat and KM values for enzymes in solution. Error bars represent standard
deviations. The means and standard deviations are calculated for 30 tracer particles. The pumping velocities at different substrate concentrations are
statistically different (P , 0.01) (see the Supplementary Information).
4
1
pumps monitored at distance intervals of 1,000 mm, for an overall self-electrophoresis , has been demonstrated in the past for
1
8–24,27
distance of 5,000 mm.
surface-anchored catalytic particles
. Symmetry breaking by
anchoring catalysts to solid surfaces can lead to chemical
Pumping mechanism. Figure 2 suggests that pumping velocity is gradients because of the asymmetric production or depletion of
generally proportional to the reaction rate, which in turn is solute molecules (charged or uncharged) 8,42–44. Directional
controlled by both substrate concentration and inherent catalytic movement of tracers in the catalase-powered pump can arise from
activity. A detailed understanding of the mechanism will allow us a non-electrolyte diffusiophoretic mechanism because of a
to a priori predict the limits of reactive sensing and detection for gradient caused by the conversion of hydrogen peroxide (two
specific analyte–pump combinations. It is possible to rule out reactant molecules) into water and oxygen (three product
3
4
5,46
several alternative mechanisms. Pumping that arises from phoretic molecules) . However, such a mechanism can be ruled out
mechanisms3 , such as diffusiophoresis , osmophoresis and from our observations with an inverted pump set-up. When the
7,38
39
40
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
2014 Macmillan Publishers Limited. All rights reserved.
3
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
a
Catalase
b
Catalase
H O
H O + O
H O
H O + O
2 2
2
2
2
2
2
2
2
1
1
0
0
.0
.5
.0
.5
.0
2.0
1.5
1.0
0.5
0.0
5
30 60 90 120 150 180 210 240
Time (min)
100
1,000
2,000
3,000
4,000
5,000
Distance (µm)
Figure 3 | Temporal and spatial changes in fluid-pumping velocity for catalase-powered micropumps. a,b, The fluid-pumping velocity in catalase-powered
micropumps in the presence of 50 mM hydrogen peroxide was monitored 50–100 m away from the enzyme pattern as a function of time at intervals of
0 minutes for a total duration of four hours (a), and as a function of distance away from the Au pattern every 1,000 m for a total distance of 5,000
b). As shown, the pumping velocity decreased over time and distance. Error bars represent standard deviations. The means and standard deviations are
calculated for 30 tracer particles. The pumping velocities at different time intervals are statistically different from the pumping velocity at time t ¼ 5 minutes
P , 0.01). The pumping velocities at different distance intervals are statistically different from the pumping velocity at distance d ¼ 100 m (P , 0.01) (see
the Supplementary Information).
m
3
m
mm
(
(
m
experimental set-up for the catalase-driven device was turned upside temperature increase at the pump surface should give rise to
down such that the Au disk was on top, the direction of fluid flow thermal convection because of the local decrease in fluid density.
relative to the glass surface reversed (Supplementary Movie 5). Thus, in an upright device the flow should be directed upwards
Fluid flowed outwards from the Au pattern at the glass surface from the pump. Fluid continuity means that near the glass surface
and, by fluid continuity, moved inwards when viewed away from the flow should be directed towards the Au pattern. For the inverted
the surface. If, indeed,
a
non-electrolyte diffusiophoretic set-up, the flow direction should be reversed because the lighter
mechanism was in operation, the direction of fluid flow should fluid tries to occupy the upper layers and spreads along the glass
remain the same irrespective of whether the pump device was surface away from the Au pattern.
upright or inverted.
To validate our hypothesis, fluid flow was monitored in the
Transport of fluid in urease-, lipase- and GOx-powered pumps inverted device to determine the pumping velocity. For all four
may be the result of an electrolyte diffusiophoretic mechanism enzyme-powered pumps, the pumping velocities in the inverted
4
5,47
because of the generation of charged reaction products . Similar set-up were similar to those in the upright one, which strongly
to its non-electrolyte counterpart, electrolyte diffusiophoresis can be suggests a density-driven mechanism as the governing factor
ruled out from our observations with inverted pumps. In the case (Fig. 4a). Further, the intensity of thermal convective flow within
of urease, the direction of fluid flow reversed when the experimental a horizontal layer of liquid in the presence of a temperature gradient
4
8
set-up was turned upside down (Au disk on the top). Closer to the is governed by the Rayleigh number (Ra) , such that
surface, the fluid flow was inwards (Supplementary Movie 6), with
4
tracers moving outwards when monitored away from the surface.
Further, for both lipase and GOx a similar effect was observed, that
is the direction of fluid flow was reversed relative to the pump
surface in the inverted set-up. The zeta potential (surface charge) of
the tracer particles has a profound effect on the direction of electrolyte
diffusiophoretic transport; tracers with opposite charges move in
opposite directions. The negatively charged sulfate-functionalized
polystyrene tracers moved towards the enzyme-tethered Au pattern
for lipase and GOx systems, and moved outwards for urease. If a dif-
fusiophoretic mechanism was in operation, reversing the charge on
tracer particles should reverse the direction of their movement.
However, when positively charged amine-functionalized polystyrene
tracers were used, the direction of their movement remained exactly
the same as that of the negative tracers. Moreover, the speed of
fluid pumping, monitored with positively charged tracers, was
similar to that of their negative counterparts for all the enzyme
pumps, which thereby conclusively rules out the possibility of a diffu-
siophoretic mechanism (Supplementary Fig. 7).
gbh dT
vx dx
Ra =
(1)
where g, h, b, n and x represent the gravitational acceleration, thick-
ness of the liquid layer, volumetric coefficient of thermal expansion,
kinematic viscosity and heat diffusivity of the liquid, respectively.
The magnitude of the vertical component of the temperature gradi-
ent dT/dx can be estimated by calculating the heat flux Q
2
J cm
2 21
(
s
) as dT/dx ¼ Q/k, where, k is the thermal conductivity
of the liquid. The heat flux depends on the rate r and enthalpy DH of
the chemical reaction, as follows:
rDH
Q = pR
(2)
2
where R is the radius of the pump surface. Assuming the flow to be
steady and small in magnitude, the speed can be scaled as:
As described above, the direction of fluid flow generated by all
four enzyme pumps reversed as the device cavity was inverted.
The simplest explanation for this observation is a density-driven
x
V ≈ h Raf (a)
(3)
mechanism. The enzymatic reactions are exothermic and the where the function f(a) depends on the aspect ratio of the
4
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
2014 Macmillan Publishers Limited. All rights reserved.
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
a
b
6
5
4
3
2
1
0
1
00 mM
10 mM
hydrogen peroxide
hydrogen peroxide
Single spacer (h)
Double spacer (2h)
Upright pump set-up
Inverted pump set-up
4
3
2
1
0
1
00 mM PNB
10 mM urea
5
00 mM urea
10 mM
100 mM PNB
β-D-glucose
Catalase
Lipase
Urease
GOx
Catalase
Lipase
Urease
Figure 4 | Fluid pumping in enzyme micropumps generated by density-driven flows. a, The fluid-pumping velocity monitored in the upright and inverted
pump set-ups showed no significant difference for any of the four enzyme micropumps. Error bars represent standard deviations. The means and standard
deviations are calculated for 30 tracer particles. The pumping velocities monitored in the upright and inverted pump set-ups are not statistically different
(
P . 0.01). b, The fluid-pumping velocity monitored in the double-spacer (two × height of chamber, h) set-up showed an approximately seven-fold increase
as compared with the single-spacer (h) set-up for three enzyme micropumps. Error bars represent standard deviations. The means and standard deviations
are calculated for 30 tracer particles. The pumping velocities monitored in the single- and double-spacer set-ups are statistically different (P , 0.01) (see the
Supplementary Information).
micropump, a ¼ R/h. The flow, therefore, can be characterized by a drives the fluid flow inwards near the glass surface. Therefore, in
speed given by:
this case the fluid density can be written as:
3
gbh rDH
r = r (1 − b(T − T ) + b (C − C ))
(5)
0
0
C
0
V ꢀ
f (a)
(4)
nkpR2
where r is the final fluid density, T is the absolute temperature, C is
At small Ra the function f(a) can be found by solving two the concentration of reaction products, r , T and C are the reference
0
0
0
uncoupled boundary-value problems: first, to derive the tempera- values of these three characteristics, and b is the solute’s coefficient
C
ture of the fluid by solving the Laplace equation with the pre- of expansion. For b . 0 the fluid density grows as the concentration
C
scribed heat flux at the reactive patch and constant temperature of products increases. Therefore, for urease the situation is more com-
at the upper plate; second, the fluid velocity can be found via plicated and double-diffusive convection sets against the competing
the linearized Navier–Stokes–Boussinesq equation. To reiterate, impacts of the reaction on the flow density and hence on the flow.
within this linear model, f(a) only changes its sign when the To verify our hypothesis, the fluid flow was examined in two
gravity is inverted.
different systems.
Numerical calculations within this approach showed that
The movement of tracer particles was monitored for the urease
2
3
f(a) grows from zero up to a value of 10 , with saturating pump in a vertical device set-up (Supplementary Fig. 8). The fluid
beyond R . 3h. Therefore, for R . 3h, any increase in the layer flowed downwards when viewed both below and above the
thickness h should result in an increase in the flow speed pro- enzyme-patterned surface, which indicates an overall downwards
3
portional to h . For smaller a, the prefactor f(a) slightly diminishes flow at the enzyme-patterned surface in the vertical set-up
this effect. For example, for the experimental set-up with a ¼ 2.3 (Supplementary Movie 7). Again, by fluid continuity the fluid
(
R ¼ 3 mm, h ¼ 1.3 mm), the velocity grows by a factor of 6.6 flowed upwards away from the surface. The reaction-generated pro-
when the layer thickness is doubled. For three different enzymes, ducts, being denser than the reactants, settled to the bottom layers of
the speed increased approximately by a factor of 6.8, which confirms the device and thereby drove the fluid-flow downwards. The mech-
our theoretical proposition (Fig. 4b). Further, assuming that anism proposed for the urease-powered micropump can also be
27
21
the values of the reaction rate (r ¼ 10 mol s ), enthalpy established by monitoring the fluid flow using a sink-reservoir
2
1
(
DH ¼ 100 kJ mol ) and height of the cavity (h ¼ 1 mm) are of model (see Supplementary Information).
the same order of magnitude for all the enzymes, the speed was
Similar experiments that used vertical set-ups were performed with
21
determined as ꢀ1 mm s , in good agreement with the experimen- catalase- and lipase-powered micropumps (Supplementary Fig. 8).
tal results. Interestingly, even though Ra for our system is fairly high In both of these cases, the fluid flow was upwards when viewed
(
ꢀ10), the smaller magnitude of f(a) helps in keeping the flow speed below the Au pattern close to the surface, against gravity
linear in Ra.
(Supplementary Movie 8). This upwards fluid movement also sup-
In the case of urease, the observed effect is opposite to the expected ports the mechanism proposed previously: convective flows result
one. Although the enzyme catalytic reaction is exothermic, the fluid is from a thermal gradient. The increase in temperature at the pump
pumped outwards in the upright device. We hypothesized that as surface caused by these enzyme catalytic reactions decreases the
þ
2
the products of urea hydrolysis are all ionic (NH4 and HCO3 ), local fluid density for catalase and lipase, and thereby drives the
these solvated ions can increase the density of the fluid near the fluid flows upwards. The effect of temperature on solute–particle inter-
4
9
enzyme pattern. This local increase in density causes the fluid to action may also play a role in the observed pumping .
spread along the glass surface, which results in a density-driven con-
vective flow that directs the fluid away from the pattern. In the Substrate-triggered release of molecules. The ability of enzyme-
inverted set-up, the denser fluid generated on the top of the device powered micropumps to respond to an external stimulus (for
settles down to lower layers in the cavity, and by fluid continuity example, substrate) and produce a change in the surrounding
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
2014 Macmillan Publishers Limited. All rights reserved.
5
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
a
saline (PBS). This ensured that the enzymatic activity was
retained and that absorbance analyses were not subject to changes
in solution pH, because fluorescein is known to show a
pH-dependent change in absorbance.
Enzyme
Substrate
Cargo
In another proof-of-concept demonstration, the release of insulin
from GOx-immobilized hydrogels was shown at different concen-
trations of glucose in sodium acetate trihydrate (SAT) buffer (pH
Positively charged
gel
Electrostatic assembly
of enzymes on gel
Powered release
of cargo from gel
5
.23). Increasing levels of insulin release from the hydrogel were
b
10
8
achieved with increases in glucose concentration in the surrounding
solution (Fig. 6, Supplementary Tables 5–8, Supplementary Figs 12
and 13, and see Supplementary Information for the detailed calcu-
lations). The release profile for insulin is somewhat different to that
of the dye, presumably because of differences in interaction with
the host hydrogel. Our results suggest the design for a rechargeable
enzyme pump that can release insulin actively at a rate proportional
to the ambient glucose concentration (one of the glucose concen-
trations employed (0.005 M) is in the physiologically relevant
range). The autonomous delivery device described above contrasts
with other recently described systems, which involve the passive
release of insulin through scaffold decomposition . Currently, we
are exploring our approach with different enzymes and hydrogel
systems. We assume that the structural variation of hydrogel is a
key parameter for the sustained release of molecules from the
cross-linked gel network.
Buffer
0
0
0
.005 M urea
.05 M urea
.5 M urea
6
4
2
3
2
0
0
100
200
300
400
500
600
700
800
900
Time (min)
Figure 5 | Urease-powered stimuli-responsive autonomous release of dye.
a, A general schematic that shows the functionalization of enzyme
molecules on a positively charged (quaternary-ammonium-terminated)
hydrogel, followed by the triggered release of cargo in the presence of the
enzyme substrate. b, The concentration of dye (fluorescein) molecules
Conclusion
In conclusion, we have demonstrated successfully the first examples
of ATP-independent, non-mechanical, self-powered enzyme micro-
pumps. These novel devices were fabricated by simple patterning
and electrostatic assembly of enzymes on surfaces. The fluid-pump-
ing speed shows a substrate concentration- and reaction-rate-depen-
dent increase. Catalysis-induced density-driven convective flow is
the driving mechanism for the directional fluid pumping. Further,
these micropumps can be utilized to attain both spatial and tem-
poral control over fluid transport, as well as delivery of colloids
and small molecules. We have also shown that the pumps can be
(
units of mM) released from urease-anchored hydrogel as a function of time
in the presence of different concentrations of urea, monitored using a UV-vis
spectrophotometer. The profile shows an increase in the amount of dye
released from the hydrogel with increasing urea concentration. The
concentration of fluorescein dye molecules released was calculated from the
absorbance values by using a calibration curve measured for the dye
(
Supplementary Fig. 9). The initial absorbance measurement was recorded
0 minutes after substrate (urea) addition.
3
8
0
5
.0
.5
4.0
.5
.0
.5
2.0
.5
.0
0.5
.0
environment by generating fluid flows makes them suitable
candidates for applications such as drug delivery, in which a
controlled response to an external stimulus is required to attain a
specific goal (for example, the triggered administration of a drug).
We fabricated a proof-of-concept design to demonstrate the
potential ability of the enzyme pumps as autonomous stimuli-
responsive drug-delivery devices. Positively charged hydrogels
were used as scaffolds for immobilizing enzymes, as well as
reservoirs for small molecules. We anticipated the active release of
small molecules and proteins from the hydrogel in the presence of
the enzyme substrate via a self-pumping mechanism. Hydrogels
with quaternary ammonium functionality were synthesized and
used as the template for enzyme immobilization via electrostatic
self-assembly, similar to the previous pump set-up. As the
hydrogel also serves as a reservoir for small molecules (cargo),
simultaneous incubation of the hydrogels with enzyme and cargo
4
70
60
3
50
40
3
2
30
1
20
Sodium acetate trihydrate buffer
0.005 M glucose
1
10
0
0
.050 M glucose
.500 M glucose
0
0
0
200
400
Time (min)
600
800
molecules (to be released) led to their absorption in the gel Figure 6 | GOx-powered stimuli-responsive release of insulin. The solution
network. The release of fluorescein dye molecules (used as a concentration and percentage of insulin molecules released from a
model cargo) as a function of time was monitored from the GOx-immobilized hydrogel as a function of time in the presence of different
urease-immobilized hydrogel in the presence of varying urea concentrations of glucose monitored using a UV-vis spectrophotometer.
concentrations (Fig. 5) using an ultraviolet–visible (UV-vis) The profile shows an increase in the amount of insulin released from the
spectrophotometer. Although there was some leaching of dye hydrogel with increases in glucose concentration in the surrounding solution.
molecules through passive diffusion in the absence of any The observed behaviour is a direct consequence of the enzymatic reaction-
substrate, the dye release rate from the hydrogel increased regulated fluid pumping. The concentration of insulin released was
21
21
with increasing substrate concentration (Fig. 5b, Supplementary calculated using the molar extinction coefficient, 1276 ¼ 6,100 M cm
Figs 9–11). This is a direct consequence of fluid pumping (Supplementary Information and Supplementary Tables 5– 8)50. The initial
regulated by enzymatic reaction. To maintain a stable solution absorbance measurement was recorded ten minutes after substrate
pH, all the measurements were performed in phosphate-buffered (glucose) addition.
6
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
2014 Macmillan Publishers Limited. All rights reserved.
©

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
triggered by the presence of specific analytes. In principle, the or SAT) to remove any excess enzyme and guest molecules. The gel pieces were
transferred to a UV-vis cell and buffer or substrate solution (urea or glucose) of
varying concentrations was added to promote the release of the cargo. A Beckman
DU-800 Spectrophotometer with a six-cell sampler was used for the measurements
analyte can be a toxic substance that will be drawn towards
the pump and be consumed as substrate, and thereby reduce its
ambient concentration (for example, a phosphate-based nerve
in kinetics and time mode, with the analytical wavelength set to 488 nm for
agent as a substrate for a phosphatase pump). Currently, work is fluorescein dye (maximum absorbance wavelength of fluorescein in PBS) and
5
0
2
76 nm for insulin (maximum absorbance wavelength of insulin in SAT). Both
in progress to pattern pumps that involve multienzyme cascades
to enable regulation and microfluidic logic. Finally, we describe
a proof-of-concept device that acts as a self-regulated, stimuli-
responsive, active-delivery vehicle.
experiments were carried out at 25 8C using 250 nm as the background wavelength.
Measurements were set to be taken at periods of 30 or 60 minutes for a total
duration of ꢀ15 hours. Initial measurements were taken after 15–30 minutes for the
dye-release experiment and after 10–15 minutes for the insulin release one. Dye
The ability of immobilized enzymes to pump fluids and particles release from the urease-immobilized hydrogels was analysed in 0 M, 0.005 M, 0.05 M
and 0.5 M urea solutions in PBS. Insulin release from GOx-immobilized hydrogels
autonomously with velocities that are dictated by the concentration
of a specific analyte, either a second enzyme or a chemical, may
enable the development of smart micro- and nanoscale devices
was analysed in 0 M, 0.05 M, 0.25 M and 0.5 M glucose solutions in O -saturated
2
SAT. Ultraviolet absorbance was plotted as a function of time at each of these
concentrations, which gives a direct correlation between substrate concentration
that contain sophisticated levels of control over the direction and and the amount of dye or insulin molecules released from the gel (Supplementary
velocity of fluid and particle transport. From a technological stand- Tables 5–8, and Supplementary Fig. 9).
point, this work provides a novel intrinsic energy source for fluid
Syntheses. Detailed information on the synthesis of quaternary ammonium thiol
movement that will enhance microfluidic device design and over-
linker, the selective attachment of enzyme molecules to the SAM-modified Au pattern
and the synthesis of hydrogel are provided in the Supplementary Information.
come a critical barrier in the field in which pressure-driven
pumps are used to move fluids. Further, these self-powered
pumps can, in principle, remain viable and be capable of ‘turning Received 8 August 2013; accepted 11 February 2014;
on’ even after prolonged storage.
published online 30 March 2014
Methods
References
Micropump design, enzyme immobilization and particle tracking. An electronic-
beam (e-beam) evaporator was used to produce the Au pattern on a PEG-coated
glass surface (MicroSurfaces). The e-beam evaporated a thickness of 90 nm of Au on
the PEG-functionalized surface, with a 10 nm adhesion layer of Cr. The radius of the
Au pattern was 3 mm. The surface was cleaned thoroughly with isopropanol
followed by acetone and dried by flowing nitrogen. Previously synthesized
quaternary ammonium thiol was used for SAM formation on the Au surface. The
ligand was dissolved in methanol and the surface was incubated in this overnight
at room temperature under an inert atmosphere. Later, the surface was washed
several times with methanol followed by PBS buffer, and dried under an inert
atmosphere. The SAM-modified surface was incubated in an enzyme solution for
1. Sengupta, S., Ibele, M. E. & Sen, A. Fantastic voyage: designing self-powered
nanorobots. Angew. Chem. Int. Ed. 51, 8434–8445 (2012).
2. Patra, D. et al. Intelligent, self-powered, drug delivery systems. Nanoscale
5, 1273–1283 (2013).
3. Wang, J. Can man-made nanomachines compete with nature biomotors?
ACS Nano 3, 4–9 (2009).
4. Sanchez, S. & Pumera, M. Nanorobots: the ultimate wireless self-propelled
sensing and actuating devices. Chem. Asian J. 4, 1402–1410 (2009).
5. Hong, Y., Velegol, D., Chaturvedi, N. & Sen, A. Biomimetic behavior of synthetic
particles: from microscopic randomness to macroscopic control. Phys. Chem.
Chem. Phys. 12, 1423–1425 (2010).
4
–5 hours. The negatively charged enzymes bound selectively to the thiol-
6. Mei, Y., Solovev, A. A., Sanchez, S. & Schmidt, O. G. Rolled-up nanotech
on polymers: from basic perception to self-propelled catalytic microengines.
Chem. Soc. Rev. 40, 2109–2119 (2011).
7. Mirkovic, T., Zacharia, N. S., Scholes, G. D. & Ozin, G. A. Nanolocomotion—
catalytic nanomotors and nanorotors. Small 6, 159–167 (2010).
8. Paxton, W. F., Sen, A. & Mallouk, T. E. Motility of catalytic nanoparticles
through self-generated forces. Chem Eur. J. 11, 6462–6470 (2005).
9. Pumera, M. Nanomaterials meet microfluidics. Chem. Commun. 47,
5671–5680 (2011).
10. Mallouk, T. E. & Sen, A. Powering nanorobots. Sci. Am. 300, 72–77 (2009).
11. Laser, D. J. & Santiago, J. G. A review of micropumps. J. Micromech. Microeng.
14, R35–R64 (2004).
12. Nisar, A., Afzulpurkar, N., Mahaisavariya, B. & Tuantranont, A. MEMS-based
micropumps in drug delivery and biomedical applications. Sens. Actuat. B Chem.
130, 917–942 (2008).
13. Dash, A. K. & Cudworth II, G. C. Therapeutic applications of implantable
drug delivery systems. J. Pharmacol. Toxicol. Methods 40, 1–12 (1998).
14. Jakeway, S. C., de Mello, A. J. & Russell, E. L. Miniaturized total analysis systems
for biological analysis. Fresenius J. Anal. Chem. 366, 525–539 (2000).
15. van der Schoot, B., Jeanneret, S., van den Berg, A. & de Rooij, F. A. A silicon
integrated miniature chemical analysis system. Sens. Actuat. B Chem.
6, 57–60 (1992).
16. Khandurina, J. et al. Integrated system for rapid PCR-based DNA analysis
in microfluidic devices. Anal. Chem. 72, 2995–3000 (2000).
17. Woolley, A. T. et al. Functional integration of PCR amplification and
capillary electrophoresis in a microfabricated DNA analysis device. Anal. Chem.
functionalized Au-patterned surface via electrostatic assembly. The enzyme-
functionalized surface was thoroughly washed with PBS to remove any unbound
enzyme molecules from the PEG-coated glass surface. The enzyme-patterned
surface was covered with a secure-seal hybridization chamber (Electron Microscopy
Sciences) with dimensions of 20 mm diameter and 1.3 mm height.
To monitor the fluid flow, in all our experiments functionalized polystyrene
microspheres (Polysciences Inc.), 2 mm in size, were introduced as tracers suspended
in a buffered solution of substrate. Videos were captured using two different
optical set-ups. One set-up comprised an upright optical microscope (Olympus
BX60M) with a halogen lamp (12 V maximum, 100 W). Excitation light was focused
into the sample through a 50× objective (LMPlanFLN 50× /0.5 BD 1/0/FN26.5,
Olympus). Emission light was collected by the objective, passed through interference
filters and finally detected by a high-sensitivity CCD (charge-coupled device)
camera at 30 frames per second. Videos were recorded using this CCD camera
attached to the optical microscope. The other system also included an optical set-up,
which comprised an inverted microscope (Zeiss Axiovert 200 MAT) with a halogen
lamp (12 V maximum, 100 W). Excitation light was focused into the sample
through a 20× objective (EC Epiplan-NEOFLUAR 50× /0.55 HD DIC 1/0, Zeiss)
or 50× objective (LD EC Epiplan-NEOFLUAR 20× /0.5 HD DIC 422472-9960,
Zeiss). Emission light was collected by the objective, passed through interference
filters and finally detected by a high-sensitivity Flea 3 CCD camera (FL3-U3-32S2C-
CS, Point Grey), which has a resolution of 2,080 × 1,552 pixels at 30 frames per
second. This camera was attached to the optical microscope and videos were
recorded. To measure fluid-pumping velocity in each experiment, 30 tracer particles
were tracked for a 25 second time interval using PhysVis software (Kenyon College).
68, 4081–4086 (1996).
Preparation of enzyme-functionalized hydrogels. For the preparation of enzyme-
18. Zhang, H. et al. Self-powered microscale pumps based on analyte-initiated
depolymerization reactions. Angew. Chem. Int. Ed. 51, 2400–2404 (2012).
19. Kline, T. R. et al. Catalytic micropumps: microscopic convective fluid flow
and pattern formation. J. Am. Chem. Soc. 127, 17150–17151 (2005).
20. Ibele, M. E., Wang, Y., Kline, T. R., Mallouk, T. E. & Sen, A. Hydrazine fuels
for bimetallic catalytic microfluidic pumping. J. Am. Chem. Soc. 129,
7762–7763 (2007).
3
functionalized, dye-loaded hydrogels, small pieces of hydrogel (2 mm ) were cut
2
1
and soaked overnight in a mixture of urease (2 mg ml , 3 ml) and fluorescein
dye (1 ml saturated) solution in PBS. The solution was kept in the refrigerator
overnight and allowed to come to room temperature before the experiments
were performed.
For the preparation of enzyme-functionalized, insulin-loaded hydrogels, small
3
pieces of hydrogel (2 mm ) were cut and soaked overnight in a mixture of GOx
21. Jun, I. K. & Hess, H. A. A biomimetic, self-pumping membrane. Adv. Mater.
22, 4823–4825 (2010).
21
21
(
2 mg ml , 3 ml) and insulin (2 mg ml , 3 ml) solution in SAT buffer (pH 5.23,
adjusted with 0.1 M HCl). The solution was kept at 4 8C overnight, and brought
to room temperature before the experiments were performed.
22. Hong, Y., Diaz, M., C o´ rdova-Figueroa, U. M. & Sen, A. Light-driven titanium-
dioxide-based reversible microfireworks and micromotor/micropump
systems. Adv. Funct. Mater. 20, 1568–1576 (2010).
Release of fluorescein dye from enzyme-functionalized hydrogels monitored with
UV-vis spectroscopy. Hydrogel pieces were washed 3–4 times with fresh buffer (PBS
23. Paxton, W. F. et al. Catalytically induced electrokinetics for motors and
micropumps. J. Am. Chem. Soc. 128, 14881–14888 (2006).
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
7
©
2014 Macmillan Publishers Limited. All rights reserved.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1895
ARTICLES
2
4. Solovev, A. A., Sanchez, S., Mei, Y. & Schmidt, O. G. Tunable catalytic
tubular micro-pumps operating at low concentrations of hydrogen peroxide.
Phys. Chem. Chem. Phys. 13, 10131–10135 (2011).
5. Zhang, L. et al. Measurements and modeling of two-phase flow in
microchannels with nearly constant heat flux boundary conditions.
J. Microelectromech. Syst. 11, 12–19 (2002).
6. Hogg, T. & Freitas, R. A. Chemical power for microscopic robots in
capillaries. Nanomedicine: Nanotech. Biol. Med. 6, 298–317 (2010).
7. Yadav, V., Zhang, H., Pavlick, R. A. & Sen, A. Triggered ‘on/off’ micropumps
and colloidal photodiode. J. Am. Chem. Soc. 134, 15688–15691 (2012).
8. Andersson, H., van der Wijngaart, W., Nilsson, P., Enoksson, P. & Stemme, G.
A valve-less diffuser micropump for microfluidic analytical systems. Sens.
Actuat. B Chem. 72, 259–265 (2001).
43. C o´ rdova-Figueroa, U. M. & Brady, J. F. Osmotic propulsion: the osmotic
motor. Phys. Rev. Lett. 100, 158303 (2008).
44. Ke, H., Ye, S., Carroll, R. L. & Showalter, K. Motion analysis of self-propelled
Pt2silica particles in hydrogen peroxide solutions. J. Phys. Chem. A 114,
5462–5467 (2010).
45. Anderson, J. L. Colloid transport by interfacial forces. Ann. Rev. Fluid Mech.
21, 61–99 (1989).
46. Anderson, J. L., Lowell, M. E. & Prieve, D. C. Motion of a particle generated
by chemical gradients. Part 1. Non-electrolytes. J. Fluid Mech. 117,
107–121 (1982).
2
2
2
2
47. Prieve, D. C., Anderson, J. L., Ebel, J. L. & Lowell, M. E. Motion of a particle
generated by chemical gradients. Part 2. Electrolytes. J. Fluid Mech. 148,
247–269 (1984).
2
9. Sengupta, S. et al. Enzyme molecules as nanomotors. J. Am. Chem. Soc.
48. Strutt, J. W. On convection currents in a horizontal layer of fluid, when the
higher temperature is on the underside. Phil. Mag. 32, 529–546 (1916).
49. Zhao, G. & Pumera, M. Macroscopic self-propelled objects. Chem. Asian J.
7, 1994–2002 (2012).
135, 1406–1414 (2013).
3
0. Muddana, H. S., Sengupta, S., Mallouk, T. E., Sen, A. & Butler, P. J.
Substrate catalysis enhances single-enzyme diffusion. J. Am. Chem. Soc.
1
32, 2110–2111 (2010).
50. Gammeltoft, S. Receptor binding of biosynthetic human insulin on isolated
pig hepatocytes. Diabetes Care 4, 2, 235–237 (1981).
3
1. Yu, H., Jo, K., Kounovsky, K. L., de Pablo, J. J. & Schwartz, D. C. Molecular
propulsion: chemical sensing and chemotaxis of DNA driven by RNA
polymerase. J. Am. Chem. Soc. 131, 5722–5723 (2009).
2. Gu, Z. et al. Glucose-responsive microgels integrated with enzyme nanocapsules
for closed-loop insulin delivery. ACS Nano 7, 8, 6758–6766 (2013).
3. Miranda, O. R. et al. Enzyme-amplified array sensing of proteins in solution
and in biofluids. J. Am. Chem. Soc. 132, 5285–5289 (2010).
4. Gaetani, G. et al. Predominant role of catalase in the disposal of hydrogen
peroxide within human erythrocytes. Blood 87, 1595–1599 (1996).
5. Svendsen, A. Lipase protein engineering. Biochim. Biophys. Acta 1543,
Acknowledgements
3
3
3
3
3
3
3
We gratefully acknowledge financial support by the Penn State Materials Research Science
and Engineering Centers under National Sciences Foundation (NSF) grant DMR-0820404
and, in part, by the Defense Threat Reduction Agency (HDTRA1-13-1-0039). This
publication was supported by the Pennsylvania State University Materials Research
Institute Nanofabrication Lab and the SNF Cooperative Agreement No. ECS-0335765. This
publication is also based on work supported by Award No. RUP1-7078-PE-12 of the US
Civilian Research & Development Foundation (CRDF Global) and by the NSF under
Cooperative Agreement No. OISE-9531011 (joint grant with the Ural Branch of the
Russian Academy of Sciences). I.O-R. acknowledges a NSF Fellowship (DGE-1255832).
2
23–228 (2000).
6. Raba, J. & Mottola, H. A. Glucose oxidase as an analytical reagent. Crit. Rev.
Anal. Chem. 25, 1–42 (1995).
7. Golestanian, R., Liverpool, T. B. & Ajdari, A. Designing phoretic micro-
and nano-swimmers. New J. Phys. 9, 126 (2007).
8. Golestanian, R., Liverpool, T. B. & Ajdari, A. Propulsion of a molecular
machine by asymmetric distribution of reaction products. Phys. Rev. Lett.
Author contributions
S.S., D.P., T.E.M. and A.S. designed the research. S.S., D.P., I.O-R. and A.A. performed the
research. S.S., D.P. and I.O-R. contributed new reagents and analytical tools. S.S., I.O-R.,
K.K.D., S.Sh., U.C-F., T.E.M. and A.S. analysed the data. S.S., D.P., I.O-R and A.S. wrote
the manuscript.
94, 220801 (2005).
3
9. Anderson, J. L. & Prieve, D. C. Diffusiophoresis: migration of colloidal
particles in gradients of solute concentration. Separ. Purif. Method 13,
Additional information
6
7–103 (1984).
Supplementary information is available in the online version of the paper. Reprints and
permissions information is available online at www.nature.com/reprints. Correspondence and
requests for materials should be addressed to T.E.M. and A.S.
4
4
4
0. Pavlick, R. A., Sengupta, S., McFadden, T., Zhang, H. & Sen, A. A
polymerization-powered motor. Angew. Chem. Int. Ed. 50, 9374–9377 (2011).
1. Moran, J. L. & Posner, J. D. Electrokinetic locomotion by reaction induced
charge auto-electrophoresis. J. Fluid Mech. 680, 31–66 (2011).
2. Howse, J. R. et al. Self-motile colloidal particles: from directed propulsion
to random walk. Phys. Rev. Lett. 99, 048102 (2007).
Competing financial interests
The authors declare no competing financial interests.
8
NATURE CHEMISTRY | ADVANCE ONLINE PUBLICATION | www.nature.com/naturechemistry
©
2014 Macmillan Publishers Limited. All rights reserved.