Quantitative analysis in nanoliter wells by prefilling of wells using electrospray deposition followed by sample introduction with a coverslip method

Article abstract of DOI:10.1021/ac0400515

In contrast to performing assays on a substrate using immobilization techniques, wet analysis in nanoliter-sized wells allows quantitative monitoring of enzyme-based reactions by measuring luminescence with time. However, a suitable dispensing method is required to accurately deposit stabilized enzyme solutions into nanoliter wells in such a manner that the enzyme activities are preserved prior to and during measurements. Furthermore, an efficient method is required to add sample liquid to these wells in such a manner that evaporation of assay liquid is completely prevented during sample introduction and monitoring. A powerful methodology is presented in this paper allowing quantitative analysis of enzyme-based reactions in identical nanoliter volumes on-chip. In a first step, picoliter amounts of protein solutions are deposited as uniform dry dots into wells using our reported electrospraying technique (Moerman, R.; Frank, J.; Marijnissen, J. C. M.; Schalkhammer, T. G. M.; van Dedem, G. W. K. Anal. Chem. 2001, 73, 2183-2189.). The silicon chips are then stored at temperatures as low as -80 °C. At the time of analysis, a sample solution is slid into the wells using a coverslip. With the edge of the coverslip, sample solution is pushed across the wells at a speed of 1.5-2.5 cm/s to prevent carryover of reagents to neighboring wells. Evaporation of assay liquid from the wells is prevented because the coverslip seals the wells and "bonds" to the chip by adhesion forces. Electrospraying appears to be an excellent method to deposit enzyme solutions containing up to 20% (w/v) of trehalose without being hampered by clogging of the capillary or splashing of droplets. After being sprayed on-chip (silicon nitride), the enzymes pyruvate kinase and lactate dehydrogenase remained stable for a period of 1.5-2 months at a storage temperature of -20 °C. The coverslip method completely prevented evaporation for minutes up to hours allowing monitoring of enzyme-based reactions in arrays of nanoliter wells.

Full text of DOI:10.1021/ac0400515

Anal. Chem. 2005, 77, 225-231
Quantitative Analysis in Nanoliter Wells by
Prefilling of Wells Using Electrospray Deposition
Followed by Sample Introduction with a Coverslip
Method
Robert Moerman,*,† Johan Knoll,† Cristina Apetrei,† Lenard R. van den Doel,‡ and
Gijs W. K. van Dedem†
Kluyver Laboratory of Biotechnology and Department of Pattern Recognition, Delft University of Technology,
Delft, The Netherlands
In contrast to performing assays on a substrate using
immobilization techniques, wet analysis in nanoliter-sized
wells allows quantitative monitoring of enzyme-based
reactions by measuring luminescence with time. However,
a suitable dispensing method is required to accurately
deposit stabilized enzyme solutions into nanoliter wells
in such a manner that the enzyme activities are preserved
prior to and during measurements. Furthermore, an
efficient method is required to add sample liquid to these
wells in such a manner that evaporation of assay liquid is
completely prevented during sample introduction and
monitoring. A powerful methodology is presented in this
paper allowing quantitative analysis of enzyme-based
reactions in identical nanoliter volumes on-chip. In a first
step, picoliter amounts of protein solutions are deposited
as uniform dry dots into wells using our reported elec-
trospraying technique (Moerman, R.; Frank, J.; Marijnis-
sen, J. C. M.; Schalkhammer, T. G. M.; van Dedem, G.
W. K. Anal. Chem. 2001, 73, 2183-2189.). The silicon
chips are then stored at temperatures as low as -80 °C.
At the time of analysis, a sample solution is slid into the
wells using a coverslip. With the edge of the coverslip,
sample solution is pushed across the wells at a speed of
1.5-2.5 cm/s to prevent carryover of reagents to neigh-
boring wells. Evaporation of assay liquid from the wells
is prevented because the coverslip seals the wells and
“bonds” to the chip by adhesion forces. Electrospraying
appears to be an excellent method to deposit enzyme
solutions containing up to 20% (w/v) of trehalose without
being hampered by clogging of the capillary or splashing
of droplets. After being sprayed on-chip (silicon nitride),
the enzymes pyruvate kinase and lactate dehydrogenase
remained stable for a period of 1.5-2 months at a storage
temperature of -20 °C. The coverslip method completely
prevented evaporation for minutes up to hours allowing
monitoring of enzyme-based reactions in arrays of nano-
liter wells.
Downscaling of assay volumes to the nanoliter and picoliter
level reduces the costs associated with expensive and rare (bio)-
chemicals and allows high-throughput screening of high-density
arrays using very small amounts of sample liquid. So, blood
samples, for example, can be screened for more and different
compounds and more frequently than is currently done in clinical
diagnostics. In addition, cell analysis, monitoring cell growth, and
monitoring crystallization of expensive proteins are preferably
performed in arrays of easily accessible and identical nanoliter
or picoliter wells made in silicon1,2 or plastic.3 However, this
requires a platform consisting of effective and cost-saving tech-
niques allowing customers to fully benefit from the advantages
of downscaling. First, a liquid deposition technique should allow
for accurate dotting of picoliter amounts of protein solutions
containing high concentrations of protectants to preserve the
activities of antibodies and enzymes on-chip. As a consequence,
the accuracy of deposition should be independent of liquid
composition, and clogging of capillaries or channels due to high
concentrations of additives should be prevented. Deposition
techniques such as piezodispensing and contact-printing have
restrictions regarding liquid composition, dot volume, and dot size.
Furthermore, piezoheads and print heads are relatively expensive
and vulnerable. In contrast, miniaturized electrospray dispensing4-6
is based on liquid being slowly forced through a capillary toward
the capillary tip where it forms a liquid cone that releases a
continuous spray of nanometer-sized droplets by just applying an
electric field. So, there is no shooting (piezoelectric) or contact-
printing mechanism required, shear stress on biocompounds is
(1) Bratten, C. D. T.; Cobbold, P. H.; Cooper, J. M. Anal. Chem. 1997, 69,
253-258.
(2) Beyer Hietpas, P.; Ewing, A. G. J. Liq. Chromatogr. 1995, 18, 3557 -3576.
(3) Jackman, R. J.; Duffy, D. C.; Ostuni, E.; Willmore, N. D.; Whitesides, G. M.
Anal. Chem. 1998, 70, 2280-2287.
(4) Moerman, R.; Marijnissen, J. C.; Frank, J. Method of the dosed application
of a liquid onto a surface; EP 1140365; Delft University of Technology, 2003.
(5) Moerman, R.; Frank, J.; Marijnissen, J. C. M.; Schalkhammer, T. G. M.;
van Dedem, G. W. K. Anal. Chem. 2001, 73, 2183-2189.
(6) Moerman, R.; van den Doel, L. R.; Picioreanu, S.; Frank, J.; Marijnissen, J.
C. M.; van Dedem, G. W. K.; Hjelt, K. T.; Vellekoop, M. J.; Sarro, P. M.;
Young, I. T. Proc. SPIE, Prog. Biomed. Opt. 1999, 3606, 119-128.
* To whom correspondence should be addressed: (tel) + 31 15 257 46 49;
(fax) + 31 15 278 23 55; (e-mail) R.Moerman@tnw.tudelft.nl.
† Kluyver Laboratory of Biotechnology.
‡ Department of Pattern Recognition.
10.1021/ac0400515 CCC: $30.25 © 2005 American Chemical Society
Published on Web 11/24/2004
Analytical Chemistry, Vol. 77, No. 1, January 1, 2005 225

negligible, and dispensing is not dependent on liquid composition
because the released amount of liquid depends on the liquid feed.
Capillaries are low cost, disposable, and available in various
materials, diameters, and lengths. Furthermore, the electric field
induces convection in the liquid cone, which prevents the
compounds from crystallizing at the capillary tip. Finally, a sprayed
liquid amount of 200 pL (room temperature) evaporates within 1
s upon landing onto substrate resulting in a nearly dry reagent
dot, which is favorable to preserve protein stability.7
the labile phosphofructokinase (PFK) during air-drying13 and
during freeze-drying.14
We report here on novel results concerning preserving activi-
ties of LDH and PK in wells on a silicon nitride surface as a
function of spraying conditions, enzyme and trehalose concentra-
tions, storage time, and storage temperature.
EXPERIMENTAL SECTION
Chemicals. Brij 35, D(+)-trehalose, 2-phosphoglycerate (2-
PG), phosphoenolpyruvate (PEP), and pyruvate were from Sigma
(St. Louis, MO). NADH, adinosine 5′ diphosphate (ADP), enolase,
LDH (from beef heart, 1250 units/mL), and PK (from rabbit
muscle, 2000 units/mL) were from Roche. The enzyme suspen-
sions contained 3.2 M (NH4)2SO4.
Fabrication of Coverslips, Chips, and Storage Boxes.
Coverslips. We used Pyrex coverslips allowing adequate light
transmission for excitation of NADH molecules as reported
previously.15 Coverslips of 2 cm × 1 cm were cut from 1-mm-
thick Pyrex plates. Prior to use and after use, the coverslips were
cleaned by brushing them for 20 s with a 5% w/v solution of Brij
35 followed by rinsing with purified water for 1 min.
Substrates. At the Delft Institute of Microelectronics and
Submicron Technology (DIMES), arrays of 25 (5 × 5) round and
square wells were made in silicon substrates of 2 × 1 cm (500
µm thick) by means of plasma etching. The wells are 400 µm wide,
400 µm apart (edge to edge), and 50 µm deep. The chip and well
surfaces were coated with a 50-nm-thick silicon nitride layer by
means of chemical vapor deposition.
Aluminum Storage Boxes. Prefilled chips were stored in
machined aluminum boxes having a storage volume of 10.5 ×
20.5 × 0.7 mm.
Second, the final analysis of a small amount of sample solution
preferably is performed using a simple and automated setup
instead of using complex dispensing tools such as a piezoelectric
dispenser, which requires precise alignment for each chip
combined with a method to prevent rapid evaporation of the
nanoliters of assay liquid. We have reported on a coverslip
method8,9 (“chimney”) to fill arrays of wells with a sample solution
using a coverslip containing holes to degas wells. The chimney
method is suitable for applications where relatively soft polymer
substrates or fragile coated substrates are used because, after
filling of the wells, the coverslip is just pushed downward onto
the substrate to seal the wells. However, the chimney method
has a few disadvantages: (1) it is difficult to handle manually
because the coverslip has to be moved across the chip forward
and backward in a very precise and fast manner to fill the wells
adequately and at the same time prevent carryover; (2) the
chimney is restricted to well depths of 40 ( 5 µm because deeper
wells are filled with more difficulty while shallow wells may result
in severe carryover of dot compounds; (3) the chimney is
restricted to certain liquids that are pulled between the coverslip
and the chip by capillary forces; (4) the chimney is not suitable
for handling cell suspensions and cell-free extracts because the
micrometer-sized cells and cell debris remain between the
coverslip and the chip after sealing resulting in poor sealing of
the wells; (5) arrays of micrometer-sized holes have to be drilled
or etched into coverslips of quartz, glass, or plastic, which is
relatively expensive.
In this study, we therefore applied an alternative coverslip
method,10 referred to as the “Zamboney” method, allowing one
to perform enzyme-based reactions in nanoliter wells by just
manually sliding a coverslip with test solution across an array of
wells. We show here progress curves (calibration curves) for
enolase, lactate dehydrogenase (LDH), and pyruvate kinase (PK)
by measuring the decrease in NADH fluorescence with time in
wells.
In addition to spraying on flat substrates and in aluminum
microliter wells as reported previously,5 we show here spraying
conditions to prefill 400-µm-wide and 50-µm-deep wells with
relatively complex reagent cocktails containing up to 20% w/v
trehalose. We used trehalose in this study because it is a well-
known preservative that is successfully used in stabilizing LDH
following freeze-drying11 and freeze-thawing12 and in stabilizing
Methodology of Performing and Measuring Enzyme-
Based Reactions in Nanoliter Wells. Enolase and PK based
assays were performed according to reactions 1 + 2 + 3, reactions
2 + 3 to show that the methodology can be used for complex
systems that require spraying of complex cocktails of reagents.
Calibration curves of enolase and PK were obtained by measuring
the decrease in NADH concentration with time using fluorescence
detection.
2 - PG enolase
T
PEP
(1)
PK
PEP + ADP T pyruvate + AMP
(2)
(3)
pyruvate + NADH LTDH lactate + NAD
The complete methodology to perform and measure reactions in
nanoliter wells is shown in Figure 1. We sprayed reagent solutions
into wells of 400 µm wide and 50 µm deep (Figure 1A). A prefilled
(11) Miller, D. P.; Anderson, R. E.; de Pablo, J. J. Pharm. Res. 1998, 15 (8),
1215-1221.
(7) Liu, W. R.; Langer, R.; Klibanov, A. M. Biotechnol. Bioeng. 1991, 37, 177-
184.
(12) Izutsu, K.; Yoshioka, S.; Terao, T. Biotechnol. Bioeng. 1994, 43, 1102-
1107.
(8) Moerman, R.; van Dedem, G. W. K. Reaction plate with slidable cover and
method to use the same. WO 02064252, Delft University of Technology,
2002.
(9) Moerman, R.; van Dedem, G. W. K. Anal. Chem. 2003, 75, 4132-4138.
(10) Moerman, R. Method of filling a well in a substrate. WO 03/041863, Delft
University of Technology, 2003.
(13) Carpenter, J. F.; Crowe, L. M.; Crowe, J. H. Biochim. Biophys. Acta 1987,
923, 109-115.
(14) Arawaka, T.; Prestrelski, S. J.; Kenney, W. C.; Carpenter, J. F. Adv. Drug
Delivery Rev. 1993, 10, 1-28.
(15) Doel, L. van den; Moerman, R.; van Dedem, G. W. K.; Young, I. T.; van
Vliet, L. J. Proc. SPIE, Prog. Biomed. 2002, 4626D.
226 Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

Brij 35 and 1% w/v trehalose in 0.05 M triethanolamine buffer
(TEA). To determine the PK calibration curve, a solution contain-
ing 2 mM NADH and 8 mM PEP was sprayed into five wells of a
single column. In each well, 2 nL of substrate solution was
deposited. Just before measurements, arrays of wells were filled
with solutions containing 0.05, 0.1, and 0.2 unit/mL PK, 12.5 units/
mL LDH, 44 mM KCl, 11 mM MgSO4, and 2 mM ADP in 0.1 M
TEA buffer. So, at the start of the reaction the filled square wells
2
having a volume of 8 nL contained /8(2 mM NADH and 8 mM
PEP) ) 0.5 mM NADH and 2 mM PEP, assuming homogeneous
mixing of the compounds.
To determine the enolase calibration curve, a solution contain-
ing 2 mM NADH and 4 mM 2-PG was sprayed into five wells of
a single column. In each well, 2 nL of substrate solution was
deposited. Prior to measurements, arrays of wells were filled with
solutions containing 0.1 and 0.5 unit/mL enolase, 5 units/mL LDH,
5 units/mL PK, 2 mM ADP, 44 mM KCl, and 11 mM MgSO4 in
0.1 M TEA buffer. So, at the start of the reaction, the filled square
wells (8 nL) contained 0.5 mM NADH and 1 mM 2-PG.
Preservation of Enzyme Activities after Spraying and Storage.
Solutions of 5 units/mL LDH and 10 units/mL PK were sprayed
in a solution of a 0.05 M TEA containing 0.5-2% w/v Brij 35 and
0.5-20% w/v trehalose.
Just before measurements, solutions containing 2 mM pyruvate
and 0.5 mM NADH in 0.1 M TEA buffer were added to arrays of
LDH-containing wells using the coverslip method. To arrays of
PK-containing wells, solutions were added containing 2 mM PEP,
0.5 mM NADH, 2 mM ADP, 44 mM KCl, 11 mM MgSO4, and
12.5 units/mL LDH in 0.1 M TEA buffer.
Control Experiments. The defined spraying and defined sample
solutions that were prepared for on-chip measurements were first
assayed on a photospectrometer in volumes of 200 µL at a
temperature of 25 ( 1 °C. The solutions that gave correct progress
curves (appropriate nominal ratios) were also used on the same
day to prefill wells by means of spraying and to fill wells using
the coverslip method. To differentiate between loss of enzyme
activity caused by spraying or caused by contact with the silicon
nitride or Pyrex surfaces, progress curves were obtained from
nanoliter wells that were filled with premixed assay using the
coverslip method. These assays were made by pipetting 50 µL of
a defined spraying solution and 50 µL of a defined enzyme solution
in a vial followed by mixing on a vortex mixer for 5 s.
Prefilling of Nanoliter Wells by Means of Electrospraying.
A syringe was filled with a fresh enzyme solution that had been
stored for no longer than 1 h at 4 °C. The syringe was refilled
every 15 min. After alignment, electrospraying was started at the
topside of a chip to obtain a stable Taylor cone followed by
spraying into the third column of wells or into the second and
fourth columns of wells. Reagent dots were deposited by spraying
for 1 up to 10 s into each well, depending on the reagent
concentrations required. Rows of dots were sprayed in a computer-
controlled manner by moving the table 150 µm downward
(spraying stops), 800 µm sideways, and 150 µm upward (spraying
resumes) in a total time period of 0.25 s. We sprayed at flow rates
of 160 and 200 pL‚s-1 at spray distances ranging from 250 to 400
µm.
Figure 1. Methodology to perform quantitative reactions in nanoliter
volumes. (A) Prefilling of nanoliter wells: image (180× magnified) of
the spraying in a 400-µm-wide and 50-µm-deep well. (B) Schematic
of the principle of filling of an array of wells with “sample” solution
using the coverslip method. In panel A (topview) of (B), reagent liquid
is pipetted onto a chip and a coverslip is pressed onto the chip and
moved toward the liquid line. In panel B (sideview) of (B), the edge
of the coverslip pushes the liquid across the wells and seals the filled
wells allowing to monitor the well contents for hours up to days.
chip was put in an aluminum box that was stored at low
temperature to preserve the activities of the sprayed reagents.
Before measurement, an aluminum box containing a chip was
allowed to accommodate to room temperature and then the chip
was placed in a holder. Liquid was pipetted onto the chip, and
the array of wells was filled using the coverslip (Figure 1B). This
chip was placed onto a computer-controlled x-y-z table to monitor
the reactions with time in a sequential manner using a specially
adapted microscope-CCD-camera setup. From the initial slopes
of the progress curves, we calculated the initial reaction rates in
picomoles of converted substrate per minute per well.
Compositions of the Spraying and Calibration Solutions. All
substrate solutions that were electrosprayed contained 1% w/v
Coverslip Method. The principle of filling (sub)nanoliter wells
(containing presprayed reagent dots) with a “sample” liquid while
Analytical Chemistry, Vol. 77, No. 1, January 1, 2005 227

avoiding carryover is to fill and seal the wells that rapidly at a
speed of 400 µm/0.03-0.02 s that all dissolved reagent material
is trapped in the wells. As is shown schematically in Figure 1B,
this is achieved by using a cleaned (brushed with Brij 35) and
smooth coverslip allowing a rapid horizontal movement over the
array of wells, thereby pushing the pipetted liquid forward across
the wells, resulting in wetting of the wells and subsequent removal
of the air from the wells with the liquid flow. The filled wells are
immediately sliced off and sealed off by the coverslip, thus
avoiding carryover and evaporation for a long period of time. To
prevent liquid being pulled between coverslip and chip by means
of capillary action, pressure is applied onto the coverslip prior to
and during movement of the coverslip. The ultrathin layer of liquid
that remains between coverslip and chip after sealing allows long-
term “bonding” of chip and cover by powerful adhesive forces.
Prior to pipetting, a prefilled chip was fixed in a laboratory-
made PMMA holder that was placed in a fixed position underneath
a zoom microscope to inspect the wells and the chip surface and
to verify appropriate filling of the wells prior to detection. The
0.5-1.0 µL of aqueous solution was deposited ∼1 mm in front of
the first column of wells using a flexible pipet tip (OD of 0.68
mm) from Bio-Rad (Catalog No. 223-9915). A Pyrex coverslip was
placed on top of the chip, ∼1 mm in front of the liquid line. A
pressure of 1-3 kg/cm2 (measured by placing the PMMA holder
onto a mass balance) was applied by pressing the index finger
(pressure area ∼1 cm2) onto the front half of the coverslip while
the coverslip was pushed with the thumb in a horizontal direction
toward the liquid line and across the wells at a speed of 1-3 cm/
s, which we verified by recording the sliding of the coverslip with
a Sony digital video camera followed by playback of each frame
of 0.04 s. Immediately after filling, the vertical pressure was
increased to 4-7 kg/cm2 while excess liquid was removed using
pressurized air (4-5 bar). After 5-20 s, the pressure was released
resulting in a “bonded” chip and cover.
Carryover. Carryover of sprayed LDH and PK to neighboring
wells that were not prefilled with enzyme solution (“empty”) was
determined by measuring NADH fluorescence with time in both
the prefilled and the wells containing only the test solution. If
carryover of sprayed enzyme to neighboring wells occurs, NADH
is also converted into NAD+ in these wells, resulting in false
positive progress curves. The slope of this progress curve is a
measure for the amount of enzyme that has been carried over.
Detection. Fluorescence Measurements. The PMMA holder
containing a filled chip was fixed at a defined position onto an
x-y-z stage allowing precise and repeatable alignment underneath
the objective of a Zeiss Axioscope microscope that was coupled
to a Princeton Versarray 512B back-illuminated CCD camera with
a Tektronic 512 × 512 DB chip, as reported previously.15,16 Prior
to monitoring of the reactions, the system was calibrated by
measuring NADH solutions in the range of 0-0.5 mM. Enzyme-
based progress curves were obtained by measuring NADH
fluorescence (λexc. ) 340 nm, λem. ) 465 nm) from 25 wells in a
sequential manner (cycle time of 30 s) for a total period of 240-
500 s. With each fluorescence measurement, a fluorescent image
was recorded from a single well to verify the presence of small
air bubbles.
RESULTS
Prefilling of Nanoliter Wells by Means of Electrospraying.
Spraying Conditions. To preserve enzyme activities during spraying
and storage, we sprayed enzyme solutions containing 0.5-20%
w/v trehalose at room temperature at flow rates of 160 and 200
pL‚s-1. During spraying of solutions containing 20% w/v trehalose,
clogging of the capillary, and crystallization of trehalose on the
outside of the capillary were completely prevented because the
electric field induced a severe convection in the liquid cone. As
shown in Figure 1A, we achieved excellent spraying of the reagent
cocktails in the center of the wells without any indications of
splashing. By spraying for 1-3 s into a single well at spraying
distances of 250 and 400 µm, uniform dots were obtained of 250
and 350 µm in diameter, respectively. Spraying for 1 s or less
into each well is preferred because rapid evaporation results in
“dry” dots already within 1 s after electrospraying, which is
favorable because it increases the enzyme stability. Moreover, the
total time required to prefill an array of 25 wells is reduced to 25
s. By applying the correct voltage difference between capillary
tip and wells relative to the spray distance, the liquid cone is
formed instantly resulting in reproducible spraying with a deviation
of 2-3%.
Coverslip Method. Determining quantitative biochemical
reactions requires the following: (i) complete filling of an array
of identical wells to obtain identical liquid volumes of 8 (square
wells) or 6.3 nL (round wells); (ii) prevention of carryover of
sprayed reagents to neighboring wells during filling and after
sealing, especially for applications where wells are prefilled with
different reagents at various concentrations. Therefore, optimal
filling conditions were evaluated using Pyrex coverslips.
Optimal Filling Conditions. Condensation onto the surface and
into the wells of a prefilled chip was prevented in most cases by
storing the prefilled chips in aluminum boxes. In addition, the
chips can be wrapped in Parafilm to eliminate condensation.
Arrays of wells were successfully filled by pipetting an amount
of 0.7-0.8 µL of liquid in the form of a line approximately 0.6 cm
long and 1.3-1.6 mm wide. Pipetting of 0.5-0.6 µL of liquid
resulted in entrapment of small air bubbles whereas pipetting of
1.0 µL resulted in a thicker line, meaning that the time between
liquid entering the well and sealing by the coverslip increased,
which resulted in an increase of carryover by ∼5%.
By applying a pressure of 3 kg/cm2 onto the coverslip, capillary
action was prevented adequately. The vertical pressure should
not exceed 3 kg/cm2 because then the cover and the chip stick
together too much, resulting in an uncontrolled sliding movement
that results in loss of contact between chip and cover. This should
be avoided at all times because from the moment contact is lost,
carryover occurs and most of the wells are filled with air bubbles
because the liquid is merely pulled over the wells (discontinuous
dewetting3).
After filling of an array, a powerful bonding of cover and chip
was achieved by applying a vertical pressure of 7-8 kg/cm2 onto
the cover for 10 s resulting in an array of “microreactors”
containing identical amounts of liquid that remained filled for days
and some for weeks by using cleaned coverslips and cleaned chips.
By applying these optimal filling and sealing conditions, we
achieved complete filling of 25 wells while reducing carryover to
less than 3%.
(16) Doel, L. R. van den, Quantitative Microscope Techniques for Monitoring
Dynamic Processes in Microarrays. Thesis, 2002; Chapter 3, pp 19-33.
228 Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

Enzyme Calibration Curves. When measuring enolase and PK
activities from a sample solution in nanoliter wells, the reaction
rates of enolase and PK depend on the concentrations of active
enzymes, cofactor (NADH), and substrates (PEP and 2-PG) in
nanoliter wells. The activities of these reagents may be influenced
by the spraying procedure, coverslip method, and silicon nitride
and Pyrex materials. Therefore, calibration curves of enolase and
PK were made. An array of 25 wells, of which only the wells of
the middle column were prefilled with reagents, was filled with a
defined solution containing the enzymes followed by 20 s of
pressing and 15-20-s transfer time before the PK- and enolase-
based reactions were monitored by measuring NADH fluores-
cence eight times in each well (8 cycles) over a total period of
240 s. Figure 2A represents 25 fluorescence images recorded from
25 wells of an array at the start of the reactions during the first
cycle. Only in the wells of the third column are fluorescence
signals obtained because only in these wells do the enzyme
reactions occur. In the middle column of Figure 2B, three progress
curves are shown obtained from wells of the third column of a
chip representing the conversion of 0.5 mM NADH into NAD+
with time that corresponds to 0.5 mM 2-PG converted by 0.5 unit/
mL enolase, according to reaction schemes 1-3. The graphs
presented in columns 1 and 3 of Figure 2B (columns 2 and 4 of
a chip) represent background signals (3000 ( 200 AU) caused
by silicon nitride and Pyrex (1100 AU) and are caused by the
defined sample solution (1900 AU). In contrast to quenching of
NADH occurring in wells on 96-well microtiter plates, we obtained
straight initial slopes from 50-µm-deep wells indicating that there
was no quenching of NADH molecules in these shallow wells.
Moreover, this indicated that the enzyme molecules were very
well mixed with the dissolved reagents, which is another advan-
tage of shallow wells.
In Table 1, the corresponding initial reaction rates (picomole
of converted substrate per minute per well) for known amounts
of PK and enolase (µunits/well) are listed. Each reaction rate is
an average obtained from four to five progress curves measured
from five wells (middle column) on a single chip. Standard
deviations are ∼10% as a result of carryover (0-3%), pipetting
errors, and spraying errors (2-3% deviation). In Table 1, the initial
reaction rates (y), given in picomoles of converted substrate per
minute per well, of PK and enolase are listed with the correspond-
ing enzyme concentrations (x) given in microunits per well. In
addition, nominal ratios are listed for on-chip measurements as
well as for microtiter plate measurements. The nominal ratios for
PK and enolase obtained from on-chip measurements are ap-
proximately 50 and 25% lower than the nominal ratios obtained
from microtiter plate measurements. Premixing experiments on-
chip (controls) also resulted in 50 and 25% lower nominal ratios.
So, the lower reaction rates were caused by a reduced activity of
the enzymes PK and enolase (adsorption of PK and enolase on
the Pyrex glass or silicon nitride) instead of being caused by a
reduced activity of the NADH, PEP, or 2-PG as a result of the
spraying or freezing procedure. The 25% activity loss of enolase
on-chip was probably also affected by the activity loss of PK on-
chip since PK serves as an auxiliary enzyme (reaction 2). Nominal
ratios of LDH obtained from premixing and prefilling experiments
on-chip were identical to those obtained from microtiter plate
Figure 2. Enolase-based reactions performed in nanoliter wells on-
chip. (A) The five wells of the middle column were prefilled with
enolase-based reagents, and a defined solution containing NADH,
LDH, PK, and enolase was added to all 25 wells using the Zamboney
method. So, only in these wells isNADH fluorescence visible while in
the other wells background fluorescence is negligible. These 25
fluorescence images were taken at the initial sequence of measure-
ments, ∼20 s after filling. The order of numbering corresponds to
the sequence of measurement. (B) Three characteristic enolase (0.5
unit/mL) progress curves obtained from the wells (prefilled with
reagents) from the middle column by measuring the decrease in
NADH fluorescence with time in a sequential manner. These three
progress curves represent the conversion of 0.5 mM NADH into NAD+
with time, which corresponds to 0.5 mM 2-PG converted by 0.5 unit/
mL enolase, according to reaction schemes 1-3. The graphs from
the left and right columns show background fluorescence signals
(3000 ( 200 AU units) obtained from the wells that were not prefilled
with reagents but only contain the defined enzyme solutions. The nine
graphs shown here consist of eight fluorescence measurements each
with 30 s of interval between two measurements.
control experiments (y ) 0.65x, R2 ) 0.99), so, LDH activity was
not inhibited by the Pyrex or silicon nitride.
Long-Term Storage of Sprayed Enzymes. A future picoliter dosing
platform should provide a way to prefill many chips with various
enzyme and reagent solutions that remain stable on-chip during
Analytical Chemistry, Vol. 77, No. 1, January 1, 2005 229

after spraying of LDH solutions containing 1 wt % trehalose onto
silicon nitride followed by additional freezing for 2 days, LDH had
lost its activity completely, probably due to freeze denaturation.
Surprisingly, addition of 10 or 20% w/v trehalose to 5 units of
LDH/mL resulted in >90% preservation of the activity of LDH
after spraying at room temperature followed by storage at -80
°C for 2 months. This means that trehalose is a very powerful
preservative that stabilizes LDH on silicon nitride during air-
drying, freezing, and defrosting.
Table 1. Reaction Rates and Nominal Ratios Obtained
from Progress Curves of Enolase and Pyruvate Kinase
in Nanoliter Wellsa
PK
reaction rate
(pmol min-1
per well)
reaction rate
(pmol min-1
per well)
(µunit/
well)
enolase
(µunit/well)
0.40
0.40
0.80
1.60
0.11 ( 0.01
0.13 ( 0.02
0.25 ( 0.02
0.53 ( 0.07
0.80
0.80
3.15b
4.00
0.17 ( 0.02
0.19 ( 0.02
0.94 ( 0.08
0.92 ( 0.10
Comparing the stability results obtained from PK as listed in
Table 2, it appears that the activity of PK reduces drastically at
relatively high electric fields, which is in agreement with our
previous results.5 So, spraying of labile enzymes should occur at
normal electric fields to impose not too much stress onto these
enzymes. In addition, a higher concentration of Brij 35, e.g., 2%
w/v, may reduce this stress. To fully stabilize PK at -20 °C, 20%
w/v trehalose should be added to a 10 units/mL solution of PK.
Moreover, PK preferably is stored at +4 or -20 °C instead of at
-80 °C, which resulted in an extra 10% loss of activity, most likely
caused by freeze denaturation. These results suggest that by
storing sprayed LDH at +4 or -20 °C instead of at -80 °C, the
LDH activity is most likely 100% preserved.
y ) 0.30x (R2 ) 0.99)c
y ) 0.233x (R2 ) 0.99)c
PK control (microtiter)
enolase control (microtiter)
y ) 0.60x (R2 ) 0.99)c
y ) 0.30x (R2 ) 0.99)c
a Reaction rates (picomoles of converted substrate per minute per
well) and corresponding enzyme concentrations (microunits per well).
b Experiment performed in round wells, which contain 0.79 × 4.00
µunits ) 3.15 µunits. c Nominal rates: y represents the reaction rate
in picomoles of converted substrate per miute per well, and x represents
the corresponding enzyme concentration in microunits per well.
Table 2. Relative Activities of LDH and PK as a
Function of Trehalose Concentration, Spraying
Conditions, and Storage Temperaturea
spray
distance,
µm
enzyme
activity,
storage
trehalos,
% w/v
voltage
kV
temp, °C
% relativeb
DISCUSSION
LDH Stored for 2 Days
Our complete methodology allows quantitative analysis of
biochemical reactions in identical nanoliter wells without the need
to immobilize enzymes or other reagents. Already upon filling of
the prefilled wells with a test solution, the sprayed reagents
dissolve and mix rapidly with the test compounds by means of
diffusion, resulting in homogeneous reaction mixtures that are
similar to reaction mixtures obtained on microtiter plates.
In contrast to other techniques such as piezoelectric dispensing
that suffer from clogging, our spraying technique allows prefilling
of nanoliter wells with solutions containing up to 20% w/v
trehalose, which is required to stabilize relatively labile enzymes
such as LDH and PK on-chip for months. As a result, the chips
do not need to be used for quantitative measurements directly
but can be shipped to other analytical laboratories where they
can be stored for a few months and still be used for quantitative
biochemical analysis in a reproducible manner. For high-
throughput applications, dotting preferably is performed by placing
capillaries in rows or arrays at distances of 600-800 µm.17 In
addition, the speed of spraying can be increased by using high-
speed actuators and by decreasing the spraytime to 0.5 s or less.
It is most likely that enzymes other than LDH and PK might
be successfully stored on-chip in solid state at a temperature of
-20 °C, since stability studies have shown that PFK,14 glucose-
-80
-80
-80
-80
-80
0.5c
1
400
400
400
400
400
1.27
1.27
1.27
1.26
1.26
0
0
4d
20
95
95
10
20
LDH Stored for 2 Months
-80
-80
10
20
400
400
1.26
1.26
90e
95e
PK stored for 4 days
+4
-80
-20
10
10
20
300
300
250
1.20
1.19
0.97
60e
50e
85
PK Stored for 1.5 Months
-20
-20
-20
-20
20
20
20
20
300
250
250
250
1.00
100
40e
0
1.04
1.12
1.14
0
a The LDH and PK solutions contained 1% Brij 35 and 5 units of
LDH/mL and 10 units of PK/mL, respectively. b Percentage of enzyme
activity relative to the nominal ratios obtained from premixing experi-
ments. c The solution contained 0.5% w/v Brij 35. d The solution
contained 2% w/v Brij 35. e Duplicate measurements (2 chips).
long-term storage. Therefore, we assessed optimal solution
compositions and optimal spraying and storage conditions to
optimally preserve the activities of the model enzymes LDH and
PK. Solutions containing 5 units/mL LDH and 10 units/mL PK
were sprayed into wells of columns 2 and 4 resulting in enzyme
concentrations of 0.25 (column 2) and 1 unit/mL (column 4) prior
to monitoring. The spraying and storage conditions and related
relative enzyme activities, given in percentages of the nominal
ratios obtained from premixing experiments, are listed in Table
2. In our previous study,5 we showed that the activities of LDH
and PK were not reduced by the spraying procedure alone when
enzyme solutions containing 0.5 and 1 wt % trehalose were sprayed
into aluminum microliter wells. However, as is shown in Table 2,
6-phosphate dehydrogenase,18
L
-asparaginase,19 and catalase20
remained stable during freeze-drying by using additives such as
trehalose, sucrose, maltose, lactose in combination with borate,
Zn2+, Co2+, Cu2+, or Ni2+. In addition, effective protectants such
as hydroxypropyl-â-cyclodextrin, 3-[(3-cholamidopropyl)dimethyl-
(17) DeFrancesco, L. Anal. Chem. 2001, 73, 307A.
(18) Sun, W. Q.; Davidson, P.; Chan, H. S. O. Biochim. Biophys. Acta, 1998,
1425, 245-254.
(19) Hellman, K.; Miller, D. S.; Cammack, K. A. Biochim. Biophys. Acta 1983,
749, 133-142.
(20) Hanafusa, N. In Freezing and Drying of Microorganisms; Nei, T., Ed.;
University Park Press: Baltimore, 1969; pp 117-129.
230 Analytical Chemistry, Vol. 77, No. 1, January 1, 2005

ammonio]-1-propanesulfonate (CHAPS), and sucrose fatty acid
monoester proved to stabilize LDH upon freeze-drying and freeze-
thawing.12
CONCLUSIONS
A straightforward platform is presented here for the perfor-
mance of quantitative analyses of (bio)chemical reactions in
nanoliter wells based on two independent methods: (i) reproduc-
ible spraying of (sub)nanoliter amounts of (bio)chemicals into
arrays of 50-µm-deep nanoliter wells on-chip followed by storage;
(ii) rapid filling of arrays of wells using a coverslip followed by
powerful well sealing to completely prevent evaporation. Promising
stability results were obtained after spraying of solutions contain-
ing LDH and PK, nonionic surfactant in 10-20% w/v trehalose
creating the possibility to preserve activities of a wide variety of
proteins in nanoliter wells for at least 1.5-2 months.
Arrays of prefilled wells can be very easily filled by using our
coverslip method with a wide variety of solutions, and the wells
can be sealed for minutes up to days which opens up interesting
possibilities in the fields of combinatorial chemistry, small-scale
PCR, and metabolomics (single cell) among others.
Future automation of the Zamboney method just requires a
powerful actuator to move the coverslip in horizontal direction
across a chip while a second actuator presses the coverslip onto
the chip. In contrast to the chimney method, the Zamboney
method is also suitable to fill arrays of wells with cell suspensions
and cell-free extracts because the cells are pushed forward across
the wells and chip surface instead of being trapped between the
coverslip and chip surface. In addition, the Zamboney method is
suitable for filling wells with viscous liquids such as glycerol or
volatile liquids such as ethanol. A coverslip preferably is thicker
than 0.2 mm and can be a few centimeters wide to fill several
arrays at a time by pipetting several lines of one or more sample
liquids. In addition to glass coverslips, PMMA coverslips can be
used as long as the surface is reasonably hydrophilic, rigid enough
to withstand the vertical pressure, and as smooth as possible for
long-term sealing. The silicon chips, which are coated with silicon
nitride, can be reused up to 30-40 times before the silicon nitride
layer gets damaged. In addition to silicon nitride-coated chips,
polymer-based chip material can be used as long as it is
hydrophilic and rigid enough to achieve successful filling of wells
followed by adequate bonding of coverslip and chip. A properly
filled and covered chip can be monitored with commercially
available microscopes or array scanners. In addition, (fast) kinetics
can be monitored by integrating a photodiode array into a
coverslip or chip.
ACKNOWLEDGMENT
This work was supported by the Delft Interfaculty Research
Center (DIOC) Intelligent Molecular Diagnostic systems. We thank
Venzteslav Jordanov for designing the plasma-etched chips. We
thank Heidi Dietrich for their contribution to the enzymatic assay
experiments. We thank Rob Suijkerbuijk, Ronald Bode, Arnold
van den Berg, Jacques Jansen, and Joop de Lange for their skilled
technical support.
Preliminary results have shown that yeast cells grow and
multiply in these silicon nitride wells after the cell suspensions
are “pushed” into the wells with a Pyrex coverslip. In addition,
preliminary results have shown that peroxide-based reactions
(using an Amplex Red kit6) can be successfully monitored in these
wells.
Received for review March 16, 2004. Accepted August 5,
2004.
AC0400515
Analytical Chemistry, Vol. 77, No. 1, January 1, 2005 231