G Model
BIOS-3634; No. of Pages 8
ARTICLE IN PRESS
Biosensors and Bioelectronics xxx (2010) xxx–xxx
Contents lists available at ScienceDirect
Biosensors and Bioelectronics
journal homepage: www.elsevier.com/locate/bios
Polysaccharide microarrays with a CMOS based signal detection unit
Johannes Baader a , Holger Klapproth a , Sonja Bednar a , Thomas Brandstetter a ,
Jürgen Rühe a,∗ , Mirko Lehmann b,1 , Ingo Freund b
a
Laboratory for Chemistry and Physics of Interfaces, Department of Microsystems Engineering (IMTEK), University of Freiburg, Georges-Köhler-Allee 103,
D-79110 Freiburg, Germany
b
Micronas GmbH, Hans-Bunte-Str. 19, D-79108 Freiburg, Germany
a r t i c l e
i n f o
Article history:
Received 13 November 2009
Received in revised form 17 January 2010
Accepted 19 January 2010
Available online xxx
Keywords:
Pneumococcal polysaccharides
Antibody
Microarray
Chemiluminescence detection
Capture molecule density
Photodiodes
a b s t r a c t
Microarray based test assays have become increasingly important tools in diagnostics for fast multiparameter detection especially where sample volumes are limited. We present here a simple procedure
to create polysaccharide microarrays, which can be used to analyze antibodies using an integrated,
complementary metal-oxide-semiconductor (CMOS) based electric signal readout process. To accomplish this chips are used which consist of an array of silicon photodiodes and where different types of
polysaccharides from the bacteria Streptococcus pneumoniae are printed on the (silicon dioxide) chip surface. Typical amounts of polysaccharide deposited in the printing process are around 12 attomol/spot.
In a subsequent reaction step the polysaccharide microarrays were used for the measurement of IgG
antibody concentrations in human blood sera using either chemiluminescence or fluorescence based
detection. To understand the device performance the influence of surface density of the immobilized
polysaccharide molecules and other parameters on the assay performance are investigated. The dynamic
measurement range of the sensor is shown to reach over more than 3 decades of concentration and
covers the whole physiologically relevant range for the analysis of antibodies against a large panel of
pneumococcal polysaccharides.
© 2010 Published by Elsevier B.V.
1. Introduction
A major trend in medicine goes towards personalized therapeutics and consequently the demand rises for reliable, fast
and economic diagnostic tools. In the field of proteomics most
standard clinical diagnostic techniques like turbidimetry only
measure one factor at a time (Salden et al., 1988) and have
limited sensitivity (Roberts et al., 2001). A more sensitive (and
also routinely used) method is the enzyme-linked immunosorbent assay (ELISA) (Engvall and Perlmann, 1971; Wernette et al.,
2003). Excellent sensitivity, variability and a for many applications
sufficient reproducibility have made ELISA processes one of the
standard methods used today in biomedical analysis. Drawbacks
of the standard ELISA process are the rather laborious procedures, which require significant amount of time during sample
preparation and analysis. In addition, the effort required for the
analysis scales directly with the number of parameters to be
tested (Engvall and Perlmann, 1971; Nahm and Goldblatt, 2002).
However, in many cases analysis of a multiplicity of factors is
∗ Corresponding author. Tel.: +49 761 2037160; fax: +49 761 2037162.
E-mail address: [email protected] (J. Rühe).
1
Present address: IST AG, Industriestr. 2, CH-9630 Wattwil, Switzerland.
crucial for a comprehensive diagnosis. Accordingly parallel detection of a series of parameters would reduce time, cost and ease
requirements on sample volume. Point-of-care diagnostics requires
high degrees of automation of sample processing and analysis
to ensure a high test precision in clinical daily routine. Microarrays are promising diagnostic devices to fulfill these specifications:
Due to the high integration density many parameters can be
tested in parallel. Short diffusion path lengths and small geometries provide rapid time to result and require only small sample
volumes (Squires et al., 2008). Additionally, microarrays can be
easily incorporated into lab-on-chip devices equipped with liquid guiding structures to provide simplified and reproducible test
handling.
Although microarrays have widely been used in the fields
of genetics (Bonetta, 2006; Schena et al., 1995; Schenk et al.,
2009) and proteomics (Joos and Bachmann, 2009; MacBeath,
2002; Michaud et al., 2003), they are not yet routinely used to
study carbohydrate–protein interactions. Carbohydrates, or more
precisely polysaccharides (Ps) are a component of bacterial capsules of various pathogens (e.g. Haemophilus influenzae, Neisseria
meningitidis, Salmonella typhi and Streptococcus pneumoniae). Thus
it is an important task in clinical diagnostics to measure antibody titers against polysaccharide structures (Wang, 2007; Wang
et al., 2002). In addition, the analysis of such systems might
0956-5663/$ – see front matter © 2010 Published by Elsevier B.V.
doi:10.1016/j.bios.2010.01.021
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
2
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
help to fully understand host-microorganism interactions. Inspired
by the success of chip-based assays, carbohydrate microarrays
are regarded as a promising tool in the emerging field of glycomics (Kiessling and Cairo, 2002; Wang et al., 2002). They are
used for a broad range of applications, e.g. analysis of cell–sugar
interactions, screening of binding inhibitors or enzyme activities.
Several different procedures to create carbohydrate microarrays have been presented in the literature (Blixt et al., 2008,
2004; Horlacher and Seeberger, 2008; Wang et al., 2002): typically substrates are modified with reactive groups which bind
specifically to functional linkers synthesized to the carbohydrate
structures. Examples for surface activation reagents are NHSesters (Blixt et al., 2004), epoxides (Park et al., 2007) (both
working with amide linkers), maleimides (thiols) (Ratner et
al., 2004) or photoreactive groups (Pei et al., 2007). Obviously,
each of these methods requires supplemental processing steps
and modifies the carbohydrate structure causing changes in the
conformation of the biomolecule and thus can lead to alteration or even destruction of the type-specific epitopes (Biagini
et al., 2003; Pickering et al., 2007). The group of Wang had
successfully adsorbed unmodified polysaccharides onto nitrocellulose coated glass slides (Wang et al., 2002). However, these
coatings also tend to exhibit unspecific adsorption of serum proteins on the surface which might result in high background
signals, which compromises the lower detection limit of the
test (Jones, 1999; Kusnezow and Hoheisel, 2003; Ramsden,
1994). Recently Marchese et al. (2009) reported a multiplex
electrochemiluminescence-based detection assay where different types of pneumococcal polysaccharides were adsorbed on
untreated carbon electrodes.
In our manuscript we describe a simple and versatile pathway
for creating polysaccharide microarrays by spotting unmodified
polysaccharides on a silicon dioxide passivation layer of a semiconductor chip. The chip itself consists of an array of 32 addressable
photodiodes with integrated amplifiers (photodiode-type active
pixel configuration [APS]). With this setup the photodiodes are
highly sensitive (Fossum, 1997) and allow the detection of very low
intensities of light generated in e.g. chemiluminescence reactions.
Compared to conventional detection systems such as fluorescence scanners the concept of placing the detection unit, the
photodiode in very close proximity to the light emitting reaction
provides a strongly increased sensor sensitivity. In such photochip
approaches the detector is only nanometers away from the light
emitting process instead of milli- or centimeters as in conventional
scanners, e.g. microarray fluorescence readers or electrochemiluminescence plate readers (e.g. MSD® sector imager) (Miao, 2008).
In addition the size of the measurement apparatus is considerably reduced by integration of the detection unit. As the light
intensity (i.e. number of photons per photodiode area and time)
drops off quadratically with the distance between light source
and detector, the distance between them becomes very important.
As a model assay we analyzed a panel of IgG antibodies against
different types of polysaccharides in human blood serum. We
used pneumococcal polysaccharides (PnPs) which are contained
in the outer shell of the bacterium S. pneumonia (Wernette et al.,
2003). As some persons show insufficient response to inoculation
against this bacterium (Elkayam et al., 2007; Goldacker et al., 2007),
the determination of a patient’s vaccination status is an important task in clinical diagnostics. In the following the process to
bio-functionalize the chips is presented. The influence of capture
molecule density at the sensor interface with respect to assay performance and the suitability of these biosensors for quantifiable
antibody concentration measurements in human serum are analyzed.
2. Materials and methods
2.1. Pneumococcal polysaccharides and antibodies
All pneumococcal polysaccharides used are part of the commercially available vaccine Pneumovax® 23 (Sweeney et al., 2000),
which is manufactured by Merck & Co., Inc., USA. Labeling of
PnPs type 14 with the amino-modified dye Dy647 from Dyomics
(Jena, Germany) was performed according to standard protocols
(Hermanson, 1996). Unbound dye was removed by centrifugation in a Microcon centrifugal filter device with a 10 kg/mol cutoff
obtained from Millipore (Billerica, MA, USA). Conjugation of secondary antibody anti-human-IgG-F(ab )2 fragment antibody from
Sigma–Aldrich (Munich, Germany) with NHS-ester biotin was done
in our laboratory according to standard protocols (Hermanson,
1996). Streptavidin-Cy5 from GE Healthcare (Munich, Germany)
and streptavidin conjugated horseradish peroxidase (strep-HRP)
were bought from Invitrogen (Eggenstein, Germany).
2.2. CMOS chip production and readout
The photochip was fabricated in a standard mixed signal
0.5 ␮m complementary metal-oxide-semiconductor (CMOS) technology process (Fossum, 1997; Wu et al., 2004) by Micronas
GmbH, Freiburg, Germany. Each pixel of the chip consists of a
silicon photodiode and an integrated active amplifier and can
be addressed individually. Since there is no overlying polysilicon
layer the photodiode-type APS pixels have a high quantum efficiency. The n-well/p-substrate photodiode is operated in reverse
bias and the photocurrent is integrated in the pn-junction’s capacitance (Cpn = 14 fF). The 32 photodiodes on the chip have diameters
of 190 ␮m and are arranged in a 4 × 8-grid with 500 ␮m pitch
center-to-center. The diodes are covered with thermally grown
silicon dioxide with a thickness of about 300 nm. The light generated by the chemiluminescence reaction with luminol ( = 428 nm)
induces a photocurrent of a few femtoampere in the photodiodes. This photocurrent and an additional dark current (thermally
induced) discharge the capacitance of the pn-junction as soon
as the reset transistor is switched off. The resulting potential
drop is amplified and buffered by the pixel’s addressable amplifier (gain externally set to 10) and readout with a conventional
notebook equipped with a measurement card for analog-to-digital
conversion.
2.3. Detection of enhanced chemiluminescence reactions
For enhanced chemiluminescence (ECL) reactions the Supersignal West Pico Kit from Perbio Science (Bonn, Germany) was
used. The two components of the kit (hydrogen peroxide and
buffered luminol) were mixed together directly before the measurement. For signal readout each pixel of the photochip was
serially addressed 15 s after the chips were incubated with the
ECL solution and the course of the voltage was recorded for 0.1 s.
The recorded data was analyzed by calculating the slope of the
capacitance discharge of the photodiode over time. For analysis,
these signals were then corrected by subtracting the dark current
of each photodiode (PBS-T-BSA-buffer was applied) which had been
recorded previous to the chemiluminescence measurement.
2.4. Printing polysaccharide microarrays
The chips were washed in toluene in an ultrasonic bath at
40 ◦ C for 5 min. Additional rinsing steps with toluene were followed by drying of the chips under nitrogen flow. All microarrays
were created by using the piezo-actuated contactless printer
SciFlexarrayerTM S5 from Scienion AG (Berlin, Germany). This print
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
3
Fig. 1. Top: False-color image of the photodiode array spotted with polysaccharide solutions with two photodiodes left blank. Center: Schematic depiction of the ELISA
procedure performed on the surface of a photodiode where chemiluminescence reactions are used to generate light signals. Bottom left: Schematic picture of one photodiode
realized in a CMOS process. Bottom right: Circuit diagram of a photodiode with an integrated signal amplification unit. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of the article.)
robot is equipped with an optical microscope at the print head operating in reflection mode, which scans the print targets and is used
for quality control of the printed chips. Together with a pattern
recognition algorithm the coordinates of the centers of each photodiode of the electric biochip can be identified. Relative air humidity
in the print chamber was kept constant at 85% during the print.
Each spot was created by dispensing 6 droplets of 390 pl printing
solution onto the silicon dioxide surface of the photodiodes. Print
buffer consists of 10 mM phosphate buffered saline sterile filtered
in Millipore stericups 0.22 ␮m (Millipore GmbH, Schwalbach, Germany). Pneumococcal polysaccharide concentrations in the print
buffer were adjusted as described in the text. After imprinting the
chips were kept at room temperature and 85% relative humidity
for 1 h and then stored at 4 ◦ C for at least 3 days. Exceptions are
explained in the text.
2.5. Staining and measurement with PnPs microarrays
Immediately before use, the imprinted silicon chips were
washed three times for 5 min in saline sodium citrate buffer with
0.01% [v/v] dodecyl sulfate sodium and 0.05% [v/v] polysorbate 20
(Tween 20® ) and were rinsed with deionized water. They were
then incubated with human serum diluted in 10 mM phosphate
buffered saline with 1% (w/v) ELISA-grade bovine serum albumin
and 0.1% (v/v) Tween 20® (PBS-T-BSA-buffer). To reduce unspecific binding reactions the diluted serum was preincubated with
5 ␮g/ml C-polysaccharide (Xu et al., 2005) and 10 ␮g/ml PnPs
22F (Henckaerts et al., 2006) according to standard procedures
(Nahm and Goldblatt, 2002). After careful rinsing in wash buffer
and in Millipore water, a biotinylated secondary antibody was
applied to the chips at a concentration of 10 ␮g/ml in PBS-T-BSA
for 30 min. Then the chips were washed again as described above.
Depending on the detection method used, the chips were incubated
with either strep-HRP (1:500 in PBS-T-BSA) for 1 h or strep-Cy5
(1:200 in PBS-T-BSA) for 30 min. Exceptions are explained in the
text. For fluorescence signal detection the chips were washed in
washing buffer, rinsed three times in Millipore water, dried with
compressed air and scanned in a Biodetect 100 biochip scanner (GeneScan Europe AG, Freiburg, Germany). Spot intensities
were analyzed using SignalyseTM Version 2.0 software (Klapproth,
Freiburg, Germany), and the background signal was subtracted
for each spot. If electric readout of the chemiluminescence
reactions was desired the chips were washed in washing buffer
and rinsed with PBS-T-BSA. Finally the electric contact pads of the
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
4
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
Fig. 2. Left: Fluorescence image of a polysaccharide chip after an assay with blood serum of a vaccinated person after 1:50 dilution. Pneumococcal polysaccharides were
printed on the chip in concentrations of 2.5–5 ␮g/ml (see also Table 1 in the S.I.). After the serum was applied on the chip for 1 h, the chip was washed and the remaining
surface-bound immunocomplexes were stained with fluorescently labeled anti-human-IgG (1 h incubation time). The fluorescence image was taken by a Biodetect 100
scanner with an illumination time of 40 s. Right: Layout of the chip (BL = blank spot).
chip housing were wiped to avoid short-circuiting and the chips
were placed in a light-proof box for electric readout.
3. Results and discussion
One of the key concepts of this paper is to perform the optical
readout of the polysaccharide microarray immediately at the surface of the substrate by printing the polysaccharides directly onto
a photodiode array. This simplifies the required instrumentation
for the analysis unit, as the chip directly gives an electrical signal as the result of the bioanalytical reaction without the need for
further cameras and related infrastructure. In addition, as already
discussed above, moving the detection unit orders of magnitude
closer to the site of the light emission bears the chance to very
significantly increase the sensitivity, i.e. the number of photons
available per surface area of the detector.
The silicon photodiodes are arranged in a 4 × 8-array. They are
fabricated in a standard CMOS process in a wafer fabrication facility.
These chips are wire-bonded onto ceramics substrates and packaged in flow cells of polymethylmethacrylamide-like material. For
protection and insulation the photodiode array is coated with a silicon dioxide passivation layer which also serves as a sensor interface
with a broad range of bio-conjugate techniques applicable. These
photodiodes are used as transducers to convert light generated by
chemiluminescence reactions into a voltage signal. To yield photons with an intensity, which correlates to the concentration of
bio-molecules in specimen such as human blood serum, reagents
are applied to the chip following in part a standard enzyme-linked
immunosorbent assay (ELISA) protocol. Here, we use horseradish
peroxidase as the enzyme to catalyze the well known chemiluminescence reaction in luminol. Due to interaction with the enzyme
the luminol molecule cleaves off molecular nitrogen, which leaves
the resulting aminophtalate in an electronically excited state. From
this state it immediately relaxes by emitting a photon with a wavelength of 428 nm. As this light emitting reaction only occurs in the
vicinity and is directly proportional to the amount of horseradish
peroxidase the light intensity detected by the underlying photodiode is a measure of the number of analyte molecules bound at
one spot of the microarray (Fig. 1). An advantage of such a procedure is that like in every ELISA process every binding event of the
analyte molecules induces a catalytic reaction and thus causes the
emission of many photons per binding event. This process leads to
a very significant signal amplification, which partially compensates
that in the chemiluminescence process only one photon per luminol molecule is emitted, in contrast to the behavior of fluorescence
dyes, where many photons per time unit and molecule are emitted
due to the fast nature of the fluorescence process.
sity by the photodiode array, but use conventional fluorescence
readout instead, were performed. Different types of unmodified
pneumococcal polysaccharide (PnPs) were spotted directly onto
the silicon dioxide passivation layer of the electric biosensor using
a print robot and a non-contact printing process. The panel of the
different chosen PnPs serotypes is shown in Fig. 2. As we were
interested to start with well established procedure, the used PnPs
serotypes are part of the commercially available polyvalent vaccine
Pneumovax® 23, comprising the 23 most prevalent pneumococcal capsular polysaccharides of S. pneumoniae which are named
according to Danish nomenclature (AlonsoDeVelasco et al., 1995).
As a proof-of-principle an assay was performed with a serum
where it was known that it had a significant content of antiPnPs immunoglobulin G (IgG). The thus obtained surface-attached
polysaccharide-IgG-complex was exposed to biotin conjugated
anti-IgG, then labeled with streptavidin carrying the fluorescence
dye Cy5. A fluorescence image obtained by scanning the chip in
a conventional fluorescence scanner (Biodetect 100) is shown in
Fig. 2. Different signal intensities for each serotype spot are caused
by variations of the concentrations of the corresponding IgG antibody in the sample. From these simple test experiments we can
conclude that sufficient amounts of PnPs molecules can be immobilized on the silicon substrate by the procedure presented here.
Furthermore, as the signals were rather small in all locations where
the photodiodes were left blank, background signal due to nonspecific surface reactions can be left out of consideration.
3.2. Comparison of fluorescence and chemiluminescence
detection
As we were interested to compare the sensitivity of standard
fluorescence detection and chemiluminescence based readout process, a highly diluted human serum sample (1:27,000) with known
concentrations of anti-PnPs IgG concentrations was applied to the
chips. After the chips had been incubated with biotinylated antihuman-IgG detection antibodies one batch of chips was labeled
with streptavidin-Cy5 and another group with streptavidin-HRP.
The first group was then analyzed with the fluorescence scanner
with maximized exposure times, i.e. times which yielded the optimum signal-to-background ratio. The chips which are labeled with
the peroxidase enzyme were measured by using the chemiluminescence approach with on-chip photodiode readout. The resulting
signals – corrected by signals of spots left blank – are shown in Fig. 3.
While the analysis of the fluorescence image showed no significant signals for any of the different types of polysaccharides except
for type 9V,2 detection of chemiluminescence with the photodiode chip exhibited the presence of IgG antibodies against six of
3.1. Proof-of-principle
To test the assay, first a series of simple measurements, which
do not take advantage of the local measurement of the light inten-
2
Spots of PnPs and blanks exhibit no statistically significant differences in their
signal intensities (determined by a Student’s t-test).
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
5
(more than an hour). Thus the immobilization procedure seems to
yield chips with a sufficient stability of the coating under various
washing procedures.
3.4. Influence of surface capture molecule density on biosensor
performance
Fig. 3. Comparison of microarray analysis using fluorescence readout in a conventional fluorescence scanner versus on-chip CMOS based chemiluminescence
readout. For comparison of detection sensitivity a highly diluted sample of human
serum was applied to the chips. The analyte concentrations ranged from 0.18 ng/ml
(PnPs 4) to 2.51 ng/ml (PnPs 14). Error bars indicate the standard deviation of three
chips printed with triplicates (N = 9).
the seven tested PnPs, even in the used highly diluted sera. The
measured antibody concentrations were in the range of 0.18 ng/ml
(PnPs type 4) to 2.51 ng/ml (PnPs type 14). In contrast to this
reported cutoff values of comparable ELISA measurements lie at
50 ng/ml for all serotypes (Henckaerts et al., 2006). From these
experiments it seems permissible to conclude that by using the
photodiode chip not only the assay complexity was drastically
reduced but also the sensitivity could be increased significantly.
3.3. Investigation of coupling stability
With any layer, where the molecules are only adsorbed to the
surface, the stability of the system against desorption and displacement, especially in complex biological solutions is a concern. To
investigate the stability of the PnPs layer on the silicon dioxide we
washed the chips in a wash buffer with high ionic strength (5 M
NaCl-solution) prior to serum incubation. A reference experiment
was performed with deionized water. Again the signal readout was
performed by using the on-chip photodiodes and chemiluminescence reactions. At all expected locations intense light intensities
caused by the chemiluminescence reaction were recorded. No significant deviations of the signals of the different spots on the
chip between the different washing procedures could be measured
(Supplementary Information, Fig. 1). Only slight decreases in signal
intensity for some types of PnPs were detectable for chips incubated in high ionic strength solutions for a prolonged time span
It is evident, that the surface density of the biomolecules, which
are used as probes on the microarray surfaces can have a profound
influence on the performance of a test (Dyukova et al., 2006). To
investigate this effect we conducted assays with PnPs spotted in
varied concentrations and analyzed them using both fluorescence
and chemiluminescence detection.
Fluorescence signals obtained with dots of varied spotting concentrations of PnPs type 9V exposed to a serum are shown in Fig. 4.
At lower polysaccharide concentrations, i.e. of less than 20 ␮g/ml,
the signal increased almost linearly with spotting concentration,
whereas at higher concentrations signal saturation was reached
indicating that no more accessible capture molecules are anchored
on the surface.
This picture changed completely when chemiluminescence
reactions were used for detection. Here highest signal intensities
are seen at rather low spotting concentrations and after a sharp
maximum the intensity of the luminescence signals drops off again
with increasing PnPs concentration. In Fig. 4 we show exemplarily
the data for two types of polysaccharides. This phenomenon can
be explained with the efficiency of the chemiluminescence process
at higher capture molecule surface densities. It might be that the
enzyme reaction rates in densely packed spots are slightly hindered
compared to reactions in bulk solution due to a reduction in the
enzyme mobility. A comparable phenomenon has been reported
for enzymatic reactions in polymer networks (Bastos et al., 2007).
A list of the optimal spotting concentrations for the chosen PnPs
panel can be found in the Supplemental Information section (S.I.,
Table 1), where also the complete data set of the PnPs described
(S.I., Fig. 2) and further measurements are presented (S.I., Fig. 3).
Additionally, the binding kinetics of both labeling agents can be
affected by the “packing density” in the spot. We studied this more
in detail by analyzing the time dependency of the signal evolution
at spots printed with varying PnPs concentrations. To this assays
were performed in which the incubation times using either strepHRP or strep-Cy5 solutions, were varied. The molar concentration
was chosen equal in both systems.
From the data shown in Fig. 5 it is evident, that the binding rate of the larger strep-HRP molecule (M.W. ≈ 140 kg/mol) is
sensitive towards variations of capture molecule surface density,
while a variation of the concentration of the spotted polysaccharide molecules by even more than a factor of 100 did not notably
alter the kinetics of the assay in the case of fluorescence staining
with strep-Cy5 (M.W. ≈ 60 kg/mol). With strep-HRP labeling satu-
Fig. 4. Chemiluminescence and fluorescence signals as function of the capture molecule density (left: PnPs type 9V; right: PnPs type 19F). Mean values of three chips are
displayed with error bars indicating standard deviation (N = 6).
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
ARTICLE IN PRESS
6
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
Fig. 5. Signal development with increasing incubation time of detection markers (left: horseradish peroxidase; right: fluorescence dye Cy5). The signals are normalized for
three different capture molecule densities of PnPs 1.
ration of spot signal could only be reached at low capture molecule
concentrations (spotting concentration of 4 ␮g/ml, 30 min incubation time). For dots having higher capture molecule concentrations
signal intensity did not reach maximum even after labeling solution was applied on the chips for 120 min. Increased polysaccharide
density turned out to hinder binding processes of proteins with
high molecular weights such as the strep-HRP conjugate—an effect
which is also known from particle motion in surface-bound polymer brushes (Filippidi et al., 2007). A scheme illustrating the effects
of steric hindrance at high density of capture molecules is provided
in the Supplemental Information section (S.I., Fig. 4).
3.5. Immobilization efficiency of pneumococcal polysaccharide on
SiO2
We were also interested in the number of polysaccharide
molecules, which remain on the surface when they were printed
in standard spotting concentration. Therefore exact amounts of
PnPs 14 conjugated with fluorescence labels were printed onto the
chip surface at print concentrations of 5 ␮g/ml (see also Section
2). As a reference experiment unconjugated dye was also printed
on the chip. After storing the chips for 3 days at 4 ◦ C spot signals
were analyzed in a fluorescence scanner for reference. Then chips
were washed following a stringent washing protocol and the signal
intensities were measured again.
After comparing the fluorescence signal intensities of each
spot before and after the washing step we were able to
determine the immobilization efficiency of PnPs 14 on silicon substrates to be 67 ± 4%. With a molecular weight of 1037 kg/mol
for PnPs 14 (Bednar and Hennessey, 1993) this means that
7.8 attomol of polysaccharide molecules are immobilized per
spot leading to a surface density of approx. 0.3 zeptomol/␮m2
(or roughly 180 molecules/␮m2 ). Comparable immobilization efficiencies were determined for both printing of purely labeled
polysaccharides or mixed solutions, where half of the spotted
polysaccharide molecules were kept unlabeled. As no signals could
be observed for dots spotted with pure fluorescence dye, it can be
concluded that no unspecific binding of the dye to the chip surface
occurred. This shows that the binding strength of the molecules
is high enough to perform ELISA assays with repeated, stringent
washing steps.
3.6. Long-term stability of the biosensor
To further study the stability of the chips, we conducted longterm storage experiments. To this end we monitored assay signals
of biochips stored under different conditions: One batch of chips
was stored in a refrigerator at 4 ◦ C, a second batch was kept at 22 ◦ C
and a third batch was put in an incubator at 37 ◦ C for storage times
between 1 and 252 days. After the desired storage time the assay
was performed. In a second series of experiments chips were stored
boxed together with drying agents at the same temperatures to
elucidate the influence of humidity on the assay performance. The
course of the cumulated and normalized signals of repeated assays
for two of the batches of chips stored without addition of drying
agents is shown in the Supplemental Information (S.I., Fig. 5).
As no significantly different behavior was observed between
conventional storage or when the chips were boxed together with
drying agents (data not shown), the humidity during the storage
process seems to be of minor importance. At all three temperatures
storage of the chips for more than 250 days did not impede assay
performance. On the contrary, storage seemed – at least initially
– to even increase assay performance. The increase was strongly
temperature dependent. When chips were kept at 37 ◦ C the signals
increased with storage time until approx. 25 days. After that time
signals decreased slowly again but remain adequate for sensing.
Comparable results were monitored for 22 ◦ C storage temperature.
Interestingly the chips stored at 4 ◦ C showed constantly improving
signal behavior with time. In this case cumulated signals of all eight
types of PnPs were doubled after 252 days of storage. The reason for
the initial improvement of device performance is probably due to a
slow rearrangement process in the polysaccharides, which renders
more attachment loci available.
3.7. Dynamic measurement range
In further experiments the dynamic measurement range of the
sensor against different concentrations of antibodies was determined. A broad measurement range is of crucial interest for
measurements in human sera as the concentration of anti-PnPs
IgGs varies significantly depending on the vaccination status of
patients. It can range from less than 10 ng/ml for patients with a low
immune status to concentrations of more than 300 ␮g/ml in vaccinated persons. To study here additionally the influence of probe
density onto the assays performances polysaccharides from differently concentrated solutions were printed onto the calibration
chips (c = 0.2, 1 and 5 ␮g/ml).
From the data presented in Fig. 6 it can be concluded, that
the sensor signal increases with antibody concentration, if the
concentration range is varied over more than three orders of
magnitude (from less than 1 ng/ml to approx. 1 ␮g/ml). Here, the
lower limit of detection of the sensor was not investigated more
in detail, as we are mainly interested in measurements at analyte concentrations around the clinical decision values, which
lie in the range of 0.9 ␮g/ml (PnPs 18C) and 5.6 ␮g/ml (PnPs
14). This implies that the patient sera can be diluted 50–100
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
7
Fig. 6. Dynamic measurement range of the biosensor functionalized with three different spotting concentrations. Shown are averaged signals obtained for real serum sample
dilution series (analyte concentrations indicated at base 10 logarithmic scale) at spots of serotype 14 (left) and 23F (right). Signals were averaged over three chips and PnPs
were printed in triplicates (N = 9). Fitting is done with a 4-parameter logistic function (r2 > 0.98). Error bars indicate standard deviation.
times and can be still analyzed with the biochip so that very
small sample volumes suffice for diagnostics. The spotting concentrations which had been previously determined as optimal for
measurements at one fixed analyte concentration also show the
highest signal intensities over a broad measurement range. Dots
of lower polysaccharide concentrations show minor signal intensities as well as decreased signal-to-background ratio (background
hereby means signals detected when only dilution buffer was
applied).
4. Conclusions
Highly sensitive microarrays with an on-chip detector capability
were obtained by printing of unmodified pneumococcal polysaccharides directly onto photodiode surfaces. The array can be used
to capture anti-polysaccharide IgG antibodies from an analyte solution such as blood serum. After analyte exposure the array is
incubated with secondary antibodies in an ELISA type of assembly. The thus immobilized enzyme induces a chemiluminescence
reaction, which is detected by the photodiode.
The obtained polysaccharide biochips have a number of interesting characteristics: The adsorbed PnPs capture molecules are
readily accessible by their corresponding IgG antibodies and secondary proteins needed for labeling, if crowding is avoided and the
surface concentration of the capture molecules is chosen in an optimal range. The layers are stable against various rigorous washing
procedures and show stability for a period of at least 250 days. The
sensor can be used to detect antibodies over a dynamic measurement range of more than three decades which renders the biochips
suitable to measure concentrations of anti-PnPs IgG in real serum
samples at clinical decision values. The electric biochips presented
here can therefore be used to monitor immune system response
provoked by vaccination against S. pneumoniae with standard vaccines, e.g. Pneumovax® 23. Additionally, the high sensitivity of the
sensor permits analysis on a volume as small as 40 ␮l of diluted
sample. This means that only 400 nl of blood serum are enough for
the analysis of the 23 different antibodies, which would permit the
use of the biochip also in pediatric diagnostics.
Use of the microarray format allows performing measurements of IgG antibodies against multiple types of pneumococcal
polysaccharides in parallel. The use of chemiluminescence detection reactions in combination with silicon photodiodes for signal
readout permits – due to the close proximity between light source
and detector, which is only on the order of a few nanometers – that
the biochips are highly sensitive, show low background signals and
have a broad measurement range. The obtained results suggest that
CMOS based polysaccharide biochips are a promising pathway for
routine diagnostic applications.
Acknowledgments
We gratefully acknowledge the valuable contributions of H.H. Peter, H. Eibel and M. Schlesier from the University Hospital
Freiburg, Germany, who also provided the pneumococcal polysaccharides and the human blood sera. Micronas GmbH, Freiburg,
Germany is thanked for providing the photodiode arrays and generous financial support.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.bios.2010.01.021.
References
AlonsoDeVelasco, E., Verheul, A.F.M., Verhoef, J., Snippe, H., 1995. Microbiol. Rev. 59
(4), 591–603.
Bastos, E., Ciscato, L., Bartoloni, F., Catalani, L., Baader, W., 2007. Luminescence 22
(2), 113–125.
Bednar, B., Hennessey Jr., J.P., 1993. Carbohydr. Res. 243 (1), 115–130.
Biagini, R.E., Schlottmann, S.A., Sammons, D.L., Smith, J.P., Snawder, J.C., Striley,
C.A.F., MacKenzie, B.A., Weissman, D.N., 2003. Clin. Diagn. Lab. Immunol. 10 (5),
744–750.
Blixt, O., Allin, K., Bohorov, O., Liu, X., Andersson-Sand, H., Hoffmann, J., Razi, N.,
2008. Glycoconj. J. 25 (1), 59–68.
Blixt, O., Head, S., Mondala, T., Scanlan, C., Huflejt, M.E., Alvarez, R., Bryan, M.C., Fazio,
F., Calarese, D., Stevens, J., Razi, N., Stevens, D.J., Skehel, J.J., van Die, I., Burton,
D.R., Wilson, I.A., Cummings, R., Bovin, N., Wong, C.H., Paulson, J.C., 2004. Proc.
Natl. Acad. Sci. U.S.A. 101 (49), 17033–17038.
Bonetta, L., 2006. Nat. Methods 3 (7), 571–578.
Dyukova, V., Shilova, N., Galanina, O., Rubina, A., Bovin, N., 2006. Biochim. Biophys.
Acta 1760 (4), 603–609.
Elkayam, O., Ablin, J., Caspi, D., 2007. Autoimmun. Rev. 6 (5), 312–314.
Engvall, E., Perlmann, P., 1971. Immunochemistry 8 (9), 871–874.
Filippidi, E., Michailidou, V., Loppinet, B., Rühe, J., Fytas, G., 2007. Langmuir 23 (9),
5139–5142.
Fossum, E.R., 1997. IEEE Trans. Electron. Devices 44 (10), 1689–1698.
Goldacker, S., Draeger, R., Warnatz, K., Huzly, D., Salzer, U., Thiel, J., Eibel, H., Schlesier,
M., Peter, H.H., 2007. Clin. Immunol. 124 (3), 294–303.
Henckaerts, I., Goldblatt, D., Ashton, L., Poolman, J., 2006. Clin. Vaccine Immunol. 13
(3), 356–360.
Hermanson, G.T., 1996. first ed. Academic Press, San Diego.
Horlacher, T., Seeberger, P.H., 2008. Chem. Soc. Rev. 37 (7), 1414–1422.
Jones, K.D., 1999. IVD Technol. 5, 32–41.
Joos, T., Bachmann, J., 2009. Front. Biosci. 14, 4376–4385.
Kiessling, L.L., Cairo, C.W., 2002. Nat. Biotechnol. 20 (3), 234–235.
Kusnezow, W., Hoheisel, J.D., 2003. J. Mol. Recognit. 16 (4), 165–176.
MacBeath, G., 2002. Nat. Genet. 32, 526–532.
Marchese, R.D., Puchalski, D., Miller, P., Antonello, J., Hammond, O., Green, T., Rubinstein, L.J., Caulfield, M.J., Sikkema, D., 2009. Clin. Vaccine Immunol. 16 (3),
387–396.
Miao, W., 2008. Chem. Rev. 108, 2506–2553.
Michaud, G.A., Salcius, M., Zhou, F., Bangham, R., Bonin, J., Guo, H., Snyder, M., Predki,
P.F., Schweitzer, B.I., 2003. Science 21, 1509–1512.
Nahm, M., Goldblatt, D., 2002. www.vaccine.uab.edu.
Park, S., Lee, M.R., Shin, I., 2007. Nat. Protoc. 2 (11), 2747–2758.
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021
G Model
BIOS-3634; No. of Pages 8
8
ARTICLE IN PRESS
J. Baader et al. / Biosensors and Bioelectronics xxx (2010) xxx–xxx
Pei, Z.C., Yu, H., Theurer, M., Walden, A., Nilsson, P., Yan, M.D., Ramstrom, O., 2007.
Chembiochem 8 (2), 166–168.
Pickering, J.W., Hoopes, J.D., Groll, M.C., Romero, H.K., Wall, D., Sant, H., Astill, M.E.,
Hill, H.R., 2007. Am. J. Clin. Pathol. 128 (1), 23–31.
Ramsden, J.J., 1994. Q. Rev. Biophys. 27 (1), 41–105.
Ratner, D.M., Adams, E.W., Su, J., O’Keefe, B.R., Mrksich, M., Seeberger, P.H., 2004.
Chembiochem 5 (3), 379–382.
Roberts, W.L., Moulton, L., Law, T.C., Farrow, G., Cooper-Anderson, M., Savory, J., Rifai,
N., 2001. Clin. Chem. 47 (3), 418–425.
Salden, H.J., Bas, B.M., Hermans, I.T., Janson, P.C., 1988. Clin. Chem. 34 (8), 1594–1596.
Schena, M., Shalon, D., Davis, R.W., Brown, P.O., 1995. Science 270 (5235), 467–470.
Schenk, T., Brandstetter, T., zur Hausen, A., Alt-Morbe, J., Huzly, D., Ruhe, J., 2009. J.
Clin. Microbiol. 47 (5), 1428–1435.
Squires, T.M., Messinger, R.J., Manalis, S.R., 2008. Nat. Biotechnol. 26 (4), 417–426.
Sweeney, J.A., Sumner, J.S., Hennessey Jr., J.P., 2000. Dev. Biol. (Basel) 103, 11–26.
Wang, D., 2007. In: Appasani Krishnarao, K., Southern, E.M. (Eds.), Bioarrays—From
Basics to Diagnostics, pp. 201–211.
Wang, D.N., Liu, S.Y., Trummer, B.J., Deng, C., Wang, A.L., 2002. Nat. Biotechnol. 20
(3), 275–281.
Wernette, C.M., Frasch, C.E., Madore, D., Carlone, G., Goldblatt, D., Plikaytis, B., Benjamin, W., Quataert, S.A., Hildreth, S., Sikkema, D.J., Kayhty, H., Jonsdottir, I.,
Nahm, M.H., 2003. Clin. Diagn. Lab. Immunol. 10 (4), 514–519.
Wu, C.Y., Shih, Y.C., Lan, J.F., Hsieh, C.C., Huang, C.C., Lu, J.H., 2004. IEEE Sens. J. 4 (1),
135–144.
Xu, Q., Abeygunawardana, C., Ng, A.S., Sturgess, A.W., Harmon, B.J., Hennessey Jr.,
J.P., 2005. Anal. Biochem. 336 (2), 262–272.
Please cite this article in press as: Baader, J., et al., Biosens. Bioelectron. (2010), doi:10.1016/j.bios.2010.01.021