Progress towards a DNA-Field-Effect
Transistor
| Oct
1, 2005 |
| By:
Patrizia
Di Pietro, Enrico
Alessi, Floriana
San Biagio, Luigi
La Magna, Gaetano
Panvini, Gianfilippo
Scicolone, Salvatore
Oliveri, Salvo
Coffa |
| Pharmaceutical
Discovery |
|
Microarray analysis allows scientists to detect thousands of genes in
a small sample simultaneously and to analyze the expression of those
genes. It promises to enable biotechnology and pharmaceutical companies
to more efficiently identify and develop drug targets. Because it also
can help identify clinical trial subjects with common biological
patterns, microarray analysis can assist drug companies in choosing the
most appropriate subjects for participating in clinical trials of new
drugs. In the future, this technology has the potential to help medical
professionals select the most effective drugs, or those with the fewest
side effects, for individual patients.
Identifying drug targets is the main market for the microarray. By
comparing the ways in which genes are expressed in a normal individual
with a diseased individual, for example, scientists might be able to
identify the genes—and hence the associated proteins—that are part
of the disease process. Researchers then could use that information to
synthesize drugs that interact with these proteins, thus reducing the
disease's effect on the body.
By connecting drug response to individual biology, DNA chips will
facilitate the integration of diagnosis and therapeutics, as well as the
introduction of personalized medicines. (1).
The enormous potential of this technology and its wide applicability
are urging the search of new devices to detect DNA hybridization to
improve the reliability, ease of use, sensitivity, etc., of the DNA
microarray. In fact, conventional microarrays, fabricated on a glass
slide and analyzed by confocal scanning fluorescence, generally reflect
an organism's entire genome and generate tremendous volumes of data. In
many cases, only specialized bioinformatics centers are equipped to
properly analyze this information. So it would be particularly
attractive to make use of the information processing capability of
silicon-integrated circuits to acquire, store and process the genetic
information achievable by detecting the hybridization signal of probes
and targets on a DNA microarray. The appropriate DNA immobilization on
silicon-based surfaces needs to be developed to optimize hybridization,
as well as the proper design of silicon-based electronic circuits.
The efficiency of an array mainly depends upon the quality of the
surface, the type of the surface treatment and the arraying and covalent
immobilization of DNA probes. In particular, the aim is to obtain flat
surfaces that are coated uniformly, without patches, because poor
coating can lead to poor retention of DNA and, consequently, low quality
analytical results (2).
Our proposed solution is based upon substrate 820 Ĺ thermal
oxide-coated silicon and a vapor-phase method for coating slides.
Vapor phase and solvent-based coating methods were compared with
aminopropiltrimethoxysilane (APTMS) as a coating agent (8). The coated
surface's topography and morphology were characterized by atomic force
microscopy (AFM) analysis and contact angle measurements. Microarray
experiments on glass- and silicon oxide-based slides were performed in
order to compare their performances in terms of sensitivity. Probes and
targets at different concentrations were tested in order to measure the
detection limit and to discover the best couple of concentrations.
Finally, the quality of data was measured by computing reproducibility,
sensitivity, accuracy, dynamic linear range and local background.
Although label-dependent methods achieve the highest sensitivities,
eliminating the labeling steps simplifies the readout and the speed and
ease of nucleic acid assays. Thus, we developed a label-free approach to
detect DNA hybridization using a silicon field-effect sensor. The
process is based upon monitoring the changes in surface potential when
DNA hybridizes on the sensor surface. This sensor is similar in
structure and behavior to a metal/oxide/semiconductor (MOS) device,
which essentially is a capacitor with variable capacitance, depending on
the potential applied. Our field-effect device, however, lacks the MOS
device's metallic layer, employing instead an an electrolytic solution,
that is conveniently polarized by a reference electrode. The overall
structure is then formed by the solution in contact with the oxide (gate
dielectric) on top of the silicon substrate, and is called EIS
(electrolyte/insulator/semiconductor).
Results of the preliminary test to validate the detection
system—using a double-stranded DNA instead of the above mentioned
oligonucleotides—are reported here (see Results and Discussion). In
one experiment, the cDNA was immobilized on the EIS surface, and changes
in capacitance-voltage characteristics during the immobilization and
denaturation processes (double-helix vs. single-strand form) were
observed. The silicon dioxide used as insulator was functionalized both
with APTMS and glycidoxypropyltrimethoxysilane (GOPS). However, after
the measuring and washing steps, we observed, by variation of contact
angle and shift of capacitance-voltage characteristics, that GOPS
yielded the most reproducibility and the greatest stability of the
silane layer, in agreement with previous reports(6, 7). The results
reported here therefore refer to GOPS-functionalized surfaces.
This approach demonstrates the most
direct and simple translation of genetic information to
microelectronics.
Experiment
Materials. Isopropylic
alcohol 100%, acetone 100%, sulphuric acid 96% and hydrogen peroxide 30%
(all electronic chemicals) were purchased from Ashland Italia s.p.a.
(San Giuliano-Mi, Italy). 3- APTMS 97% and GOPS (98%) were obtained from
Aldrich (Milano, Italy)
Tris-Cl 1M pH 7.5 (Sigma), Na2HPO4
(Sigma) NaCl powder (Sigma), Tween 20 (Sigma), 20X SSC (0.3 M sodium
citrate/3M sodium chloride; Euroclone Ltd., (Paignton-Devon TQ4 5DN-
UK), sodium dodecyl sulphate (SDS; Sigma) and NaOH (Sigma) were used.
Oligonucleotides were purchased
from MWG Biotech AG (Ebersberg, Germany). As substrates, 820 Ĺ thermal
oxide-coated silicon wafers (manufactured by ST) were used for the
grafting experiments. The EIS structure was composed by n+ (1020
cm-3)/ n- (1015 cm–3)/ SiO2
(350 Ĺ, thermal) and a back-side contact with a three-layer metal. The
Ag/AgCl reference electrode used to ensure a stable electrical contact
with the solution was by Metrohm (Herisau, Switzerland). All references
to water indicated ultrapure water (resistivity >18 Ω/cm).
Substrate preparation. The
substrate was washed in the cleaning solution (MeOH/HClconc.
1÷1 ratio) under stirring and at room temperature, for 30 min. The
cleaned surface was activated by immersion in NaOH 0.1 M solution for 30
min under stirring at room temperature. Subsequently it was washed in
water and dried under a nitrogen flow. Finally, the substrate was cured
in O2 plasma (20 W, 20 sccm of O2, for 45 sec.).
The substrate surface was silanized with 3-APTMS in a CVD (chemical
vapor deposition) reactor at 75 °C. (This process is called vapor-phase
silanization, or VPS.) Then it was washed in acetone in an ultrasound
bath and dried under nitrogen flow.
Another substrate was prepared by
liquid-phase silanization (LPS, in which the silanizing agent is mixed
with a liquid organic solvent), which after the activation step has been
immersed for 30 min in 1% APTMS solution in acetone. Afterwards, the
substrate was washed in acetone and dried for 45 min at 110 °C.
Substrate characterization. A
silanized surface was characterized by AFM and contact angle measures.
AFM was carried out in air by using a multimode/nanoscope IIIa module
(Digital Instruments, Santa Barbara, CA) equipped with an extender
electronics module (Digital Instruments) in tapping mode. Commercially
available tapping etched silicon probes (Digital Instruments) with a
pyramidal shape tip having a nominal curvature of 10 nm and a nominal
internal angle of 35° were used. Height and phase images were recorded
contemporaneously by collecting 512 x 512 points for each scan and
maintaining the scan rate below 1 Hz.
The contact angle of water on the
SAMs (self-assembled monolayers) was measured at room temperature using
the Cam 200 contact angle meter (KSV Instruments Ltd., Helsinki,
Finland) The resolution of the CCD (charged-coupled device) was 512 x
512. All the reported results are to be intended as obtained from
measurements carried out on manifold samples, varying the contact area
on the surface and taking the average of the left and right angles.
Immobilization of a DNA
oligonucleotide. We used a 50-mer oligonucleotide as a probe for
microarray experiments. Moreover, two distinct types of controls were
used—the negative control and a performance control—which aimed to
measure the microarray's accuracy. The probe was spotted at different
concentrations—10, 5, 1, 0.1, 0.01 µM—while the negative control
and performance control were spotted at 1 µM. Two types of targets were
used, one complementary to probe and labeled at its 5' end with Cy3; the
second complementary to the performance control and labeled with Cy5.
They were hybridized at different concentrations—10, 1, 0.1, and 0.01
nM. A grid of seven rows by six columns was designed; each row had a
distinct concentration of the same probe, the grid was replicated five
times. Oligonucleotides were diluted at different concentrations in
sodium phosphate buffer 0.1 M pH 8 and spotted onto activated slides by
the Piezoarray (PerkinElmer, Boston, Massachusetts, USA) non-contact
piezo-driven microdispensing system. After spotting, the slides were
kept in a humidified chamber and incubated in water bath at 60 °C for 2
h.
After incubation time, the slides
were dipped (three times) into washing buffer at high ionic force (10 mM
TRIS-Cl; 1M NaCl; 0.05% Tween 20) to remove non-covalently bound nucleic
acids; dipped into water (three times) and then dried in a stream of N2
before proceeding to the hybridization step.
Hybridization. The printed
slides were pre-hybridized for 45 min at 65 °C in a buffer solution
containing 6X SSC and 0.5% SDS. Then the slides were washed in water at
room temperature and dried in a stream of N2.
5'-Cy5-labeled and 5'-Cy3-labeled
oligonucleotides were diluted in pre-hybridization buffer solution and
applied to the slides; hybridization was performed under cover slips in
a humidifier chamber at 65 °C for a minimum of 1.5 h.
At the end of hybridization time,
un-hybridized oligonucleotides were removed by washing with 0.1X SSC and
0.1% SDS for 10 min. After washing, the slides were dried in a stream of
N2 and scanned with a confocal laser scanner.
Image acquisition and data
analysis. The fluorescent images were acquired by the ScanArray
Express (Perkin Elmer) confocal laser scanner at different settings of
laser power and photomultiplier (pmt): 60%, 70% and 80%. Microarray
image analysis was performed by the Microarray Analysis Tool of the ST
Gene Platform™ (ST Microelectronics).
Figure 1. Standard deviation of
spot intensity values average over grids at different
concentrations.
|
Capacitance–voltage measurements.
The slices forming the EIS structure were mounted in a Teflon holder
and sealed from the solution with a rubber O-ring, defining an active
surface area of 0.6 cm2 (Figure 1). An Ag/AgCl reference
electrode was used to ensure a stable electrical contact with the
solution when it was biased with respect to the back-side electrode. The
PTFE cell is capable of a 4-mL solution volume, into which the reference
electrode was bathed. The field-effect readout is most sensitive when
using a solution with low ionic strength, which is when counter-ion
screening of the charged molecules is minimized; we used a phosphate
buffer 10 mM NaCl, 2 mM NaH2PO4. Adding NaOH, the
pH of the solution was set to 7 and continuously monitored to assure its
constancy.
The capacitance-voltage (C-V)
output characteristics were recorded using a potentiostatic/galvanostatic
system (AUTOLAB PGSTAT30 by EcoChemie, Utrecht, The Netherlands)
interfaced with a personal computer through a GPES/USB (general purpose
electrochemical system) card. Capacitance was measured at different dc
biases, from –1.5 V to 1.2 V, and at various ac frequencies in the
range of 10 Hz to 10 KHz.
We used a polymerase chain reaction
(PCR) product, amino-terminated of cDNA (1500 bp) 90 ng/µL in
concentration. Due to the high contact angle value of the GOPS silanized
surface, as well as to the large solution volume being used, a great
deal of DNA is required. To overcome this issue, the sample was bathed
in NaOH solution at pH = 11 for 30 min. and then rinsed. This way we
were able to use a volume of only 20 µL. After the immobilization and
the relative measures, denaturation was performed by placing the sample
in 80 °C water, followed by rinsing and drying.
Results and Discussion
Figure 2. An AFM image shows a
silicon oxide surface obtained by thermal growth.
|
In order to investigate the surface
changes during the silanization process, the slides treated with both
type of silanization methods were examined by AFM and compared to the
pre-treated measurement results. Figure 2 shows AFM analysis of a
typical silicon oxide surface obtained by thermal growth. The beginning
surface was very flat, with a mean roughness of 1.6 Ĺ. After
silanization, the roughness values increased, though the CVD-associated
roughness was much lower than that for LPS. Figure 3 shows an AFM
inspection of a surface silanized by VPS. The surface is homogenous and
its mean roughness was 2.6 Ĺ. We also conducted an AFM inspection on an
LPS-silanized sample. In this sample (Figure 4), the roughness value
increased to 4.4 Ĺ. Moreover, the surface appeared irregular, with
irregular organic structures, presumed to be polymeric aggregates.
Figure 3. An AFM image shows the
silanized surface prepared by VPS.
|
The preparation of the silane surface
is a crucial step in microarray technology because often it runs into
some problems, like an increased roughness due to polymerization
reactions in aqueous solution. So, as shown above, the surface results
are not very homogeneous or regular for this type of application because
the increased roughness introduces failures as gaps or bridges in the
final silane film. The absence of water in our protocols suppressed the
unwanted polymerization and produced a low roughness value in a
homogeneous surface suitable for further processing.
Figure 4. An AFM image shows the
silanized surface prepared by LPS.
|
On the basis of these studies, we
think that 820 Ĺ thermal oxide-coated silicon is a good substrate for
microarray applications, and our coating method with APTMS in vapor
phase produces a surface coating particularly suitable for microarray
studies.
The contact angle measurements of
silanized surface performed with water have yielded a value of 62±3
degrees, according to the literature (3). This value is typical of a
rather hydrophobic aminated surface.
Figure 5. Silicon oxide hybridized
with a target at 10 nM. The scanner setting was 70% laser/70%
PMT.
|
In order to verify the reproducibility
and the sensibility of the ST microarray system, we conducted a series
of experiments on thermal silicon oxide and glass slides. In figure 5 is
an images obtained by the scanner. The image has been acquired to 70% of
the laser power and PMT and refers to one slide hybridized with a target
at the concentration of 10 nM. The image shows how the controls have
worked well, the negative control (fourth row) not showing any
hybridization and the performance control (sixth row) showing specific
hybridization (no cross hybridization). Moreover, at these settings and
concentration of the target, it also is possible to see also the lower
concentration probe (0.01 µM, fifth row).
Figure 6 shows compares the
fluorescent signals elicited from silicon and glass slides. Both samples
were hybridized with the same concentration of target, but the glass
slide shows a much weaker signal, even at higher settings of laser power
and pmt.
Figure 7. Reproducibility over
experiments and the linear dynamic range in the interval of
10–0.01 uM for silicon slides. Negative and positive controls
work properly. Power laser = 70% and PMT gain=70%
|
The data from all images have been
processed by statistical analysis to estimate some significant
parameters for microarray system quality, such as reproducibility and
background. Figure 7 shows the reproducibility among three experiments
of the same type and the linear dynamic range. Concentration of spotted
probe is plotted on the x-axis, with fluorescence intensity on the
y-axis. The three curves agree well, almost perfectly overlapped.
Moreover, the microarray hybridization data are linear, since at the
diminishing of the concentration of the spotted DNA we have a
proportional decrease of fluorescence intensity.
Figure 8. Local background for
each spot computed as the median of the region around the spot.
Local background = average of all background medians.
|
Figure 9 shows a graph similar to that
in Figure 6, comparing the performance of silicon and glass. Although
the glass shows good reproducibility, comparable to the silicon, its
signal intensity was twenty times lower.
Figure 1 analyzes the
reproducibility of the same experiment. The x-axis plots different
concentrations of spotted DNA, while in the y-axis shows the the
standard deviation of spot intensity values averaged across each row.
For example, the first bar represents the standard deviation of spot
intensity values averaged in the first row of spots (concentration 10 µM)
of the first grid; the second bar represents the same parameter for the
row of equal concentration of the second grid, and so on. As the chart
demonstrates, the standard deviation values are very low, reflecting
good reproducibility in the tested system.
Figure 9. A comparison between
glass and silicon. At the same concentration the signal captured
by SiO2 slides is higher than that of glass slides
|
Local background, computed as the
median of noise in the region around the spot, is another critical
parameter. Figure 8 shows local background values for a set of
experiments. A set of hybridization experiments, carried out with the
four concentrations of target, ranges along the x axis. The y axis plots
the median local background. It is worth noting that all values fall in
intervals with very low background values.
The limit of detection—defined as
the lowest measurable concentration of probe at a specific target
concentration—was computed. At the lowest target concentration (0.01
nM) and at 70% of laser scanner power and pmt, 1 µM of spotted probe
was detected as the limit.
The results shown above clearly
demonstrate that the use of silicon, rather then glass, provides for a
more sensible and reliable microarray system. The method assures
sensitivity, reproducibility and accuracy—critical parameters for a
technology whose main application are in medical and pharmaceutical
fields.
Figure 10. Capacitance–voltage
characteristics of the successively EIS functionalized
structure.
|
Figure 10 shows the series of C-V
characteristics for the sequence of modifications that the surface
underwent. Silanization with GOPS—the first step—is on the left;
after the immobilization—the second step—the curve translates to the
furthermost right. After denaturation—the last treatment—the graph
falls between teh other two.
Shifts of the curves along the
voltage axis are indicative of charge density variations at the
functionalized SiO2 surface in contact with the solution (4).
In fact, changes in the potential at the electrolyte-insulator interface
produce variations at the silicon-insulator interface, namely relative
charge carrier density variations in the space-charge layer of the
semiconductor: this brings the structure to changing its threshold
voltage (5). The measured shifts are easily explained on the basis of
the negatively charged DNA molecule: the augmented negative charge on
the surface causes a less negative necessary potential to obtain
substrate inversion.
While the trends of the shifts were
always in the same directions, we did not obtain a constant change
between the curves. The data do show a strong correlation between the
observed shifts and the alteration of the surface, so this label-free
method can be used to quickly and readily probe the presence of DNA on a
functionalized substrate. Not only can the DNA immobilization phase be
monitored, but its chemical state modifications can be monitored as
well. Silicon is the basis for most microelectronics devices, so it
follows logically that it should be possible to fabricate field-effect
devices to directly detect and transduce hybridization of
oligonucleotides, as our results indicate.
Conclusions
The development of
label-independent methods that can monitor hybridization and are simple
and scalable still is in its infancy. To approach these goals, we
anticipate further research toward improvement of concentration
sensitivity and of specificity through optimization of experimental
conditions, improved surface functionalization and integration in
small-volume fluid handling systems. Particularly, sensitivity and
specificity are of critical importance for applications in the medical
and pharmaceutical fields.
Patrizia Di Pietro*, Enrico
Alessi, Floriana San Biagio, Luigi La Magna, Gaetano Panvini,
Gianfilippo Scicolone and Salvatore Oliveri are on the
Lab-on-Chip R&D Team at STMicroelectronics. Salvo Coffa, is
director RIG-MLD group, STMicroelectronics. Patrizia Di Pietro can be
reached at STMicroelectronics, Stradale Primosole, 50, 95121 Catania
Italy. E-mail patrizia.dipietro@st.com
References
1) K. Jain. "Applications of
biochip and microarray systems in pharmacogenomics, Pharmacogenomics."
2000 Aug;1 (3):289-307.
2) J. Worley, K. Bechtol, S. Penn,
D. Roach, D. Hanzel, M. Trounstine and D. Barker. "A system
approach to fabricating and analyzing DNA microarrays," Microarray
biochip technology, Edited by Mark Schena. 2000 BioTechniques Books,
Natick, MA. pp 73-76
3) V. Balladour, A. Theretz and B.
Mandrand, J. Colloid and Interface Sc. 194, 408–418
(1997).
4) D.E. Yates, S. Levine and T.W.
Healy, J. Chem. Soc. Faraday I , vol. 70, pp. 1807–1818,
1974.
5) S.M. Sze Physics of
Semiconductor Devices 2nd edition, Wiley & sons, New York, 1981.
pp. 589-594
6) J.P Cloarec, J.R. Martin, C.
Polychronakos, I. Lawrence, M.F. Lawrence, E. Souteyrand, "Functionalization
of Si/SiO2 substrates with homooligonucleotides for a DNA
biosensor," Sensors and Actuators B 58, 394 – 398
(1999)
7) J.P Cloarec, N. Deligianis, J.R.
Martin, I. Lawrence, E. Souteyrand, C. Polychronakos M.F. Lawrence,
"Immobilization of homooligonucleotide probe layers onto Si/SiO2
substrates: characterization by electrochemical impedance measurements
and radiolabelling," Biosensors and Bioelectronics 17,
405–412 (2002)
8) N. Zamatteo, L. Jeanmart, S.
Hamels, S. Courtois, P. Louette, L. Hevesi and J. Remacle.
"Comparison between different strategies of covalent attachment of
DNA to glass surface to build DNA microarray." Analytical
Biochemistry 280, 143-150 (2000).
|
