Salvatore
Petralia
a,
Emanuele Luigi
Sciuto
b and
Sabrina
Conoci
*a
aSTMicroelectronics Stradale Primosole, 50 - 95121 Catania, Italy. E-mail: sabrina.conoci@st.com
bDipartimento di Fisica e Astronomia, Via S. Sofia, 64, 95123 Catania, Italy
First published on 21st November 2016
New miniaturised microfluidic biofilter (BF) devices based on silicon micropillars have been developed and tested regarding their ability to extract HBV (Hepatitis B Virus) bacterial DNA from biological sample solutions. The device is composed of a silicon microchannel in which the pillars are distributed at the bottom surface. The extracted DNA solutions were analysed by real time PCR amplification and the biofilter performance was evaluated. The results obtained show that the DNA binding to the biofilters and the elution efficiency strictly depend on the pillars’ geometrical dimensions and increase proportionally with the surface/volume ratio. The device exhibiting the best extraction efficiency was then tested in combination with a silicon integrated real time PCR amplification chip as a preliminary step towards the development of genetic point-of-care devices.
These miniaturized systems are called “Lab-on-Chip” (LoC) devices, and are sophisticated since they require the integration of several specific technological modules capable of efficiently managing the two main steps needed for molecular analysis: (a) DNA extraction and purification (sample preparation) and (b) DNA detection. The DNA extraction step plays a fundamental role in the outcome of the molecular analysis since the purity and the concentration of the extracted DNA strongly affect the subsequent detection analysis. While for DNA detection there are many examples of LoC able to perform polymerase chain reaction (PCR) or real time PCR, hybridization, capillary electrophoresis etc.,4–13 the sample preparation is still challenging since several starting material samples can be employed (blood, urine, saliva etc.) and, consequently, complex architectures and protocols are needed. Therefore, the literature reports few examples of miniaturized sample preparation devices when compared with those for DNA detection. Many of these are developed using plastic materials and include a complex microfluidic network that manages fluid movement, mixing, splitting etc.14–21 Plastic materials for integrated microfluidics have the advantage of low cost but the miniaturization, integration and the thermal properties required by some of the extraction steps (i.e. lysis) still remain the main limitations. In contrast, silicon is a very appealing material. In fact, it combines good physical aspects for sensing, such as low heat capacity, good thermal conductivity, and possible patterned structures to increase the surface–area ratio, with consolidated production technologies and industrialization processes at high volume. Additionally, it allows the integration of electrodes and microelectronics circuitry that imprint the so-called “intelligence on board”. Recently, we proposed an integrated silicon-based lab-on-chip for sample preparation, PCR amplification and hybridization to detect the betaglobin gene from blood.22 This solution adopts a sample preparation process in one step. It is performed in a silicon chamber using a lysis solution containing a mix of proteases that lyse the human cells and digest most of the proteins present in the blood. In this case, even if the DNA is not purified, this approach works in PCR detection since the concentration of human genomic DNA is quite high in human cells. In contrast, when the DNA to be extracted is viral or bacterial, the above-reported methodology fails since these genomes are present in very low concentration in the starting sample compared to the co-present human DNA. In this case, therefore, the purity of the extracted material must be high and, operatively speaking, after the lysis, two additional steps of purification and concentration are mandatory. Viral and bacterial DNA extractions are usually achieved by Solid Phase Extraction (SPE) based on capturing the DNA present in the lysed solution on solid surfaces (purification step) and then eluted with appropriate buffers (concentration step).23 SiO2 is the most used solid surface since it is well-known to bind DNA in the presence of chaotropic salts and ethanol.24 Most of the current commercial extraction kits use silica in the form of micro-filter mounted in a plastic column (i.e. Qiagen kits https://www.qiagen.com/gb/shop/sample-technologies/dna/qiaprep-spin-miniprep-columns) or layers covering magnetic beads (i.e. Magazorb kit https://ch.promega.com/-/media/files/resources/protocols/technical-bulletins/101/magazorb-dna-mini-prep-kit-protocol.pdf) that are moved by external magnets. Sometimes silica is derivatized with positive amino-groups to bind the negatively charged DNA more effectively than naked silica. However, the technical features (i.e. porosity, surface-to-volume ratio, etc.) of these systems are not explicitly disclosed since they are covered by trademarks. A drawback of the current extraction protocol is that the above-reported chemical agents (ethanol and chaotropic salts) can inhibit the PCR reaction or other molecular techniques.25 Therefore several washing steps are needed to eliminate traces of these compounds. In terms of miniaturization of the device, this implies complex architectures to manage the fluidic steps, with consequent increase of the design complexity and costs.
From the prospect of achieving a chip for DNA extraction based on SPE on the silica surface, one of the most interesting approaches shown in the literature is that involving the micro-fabricated silicon pillars. It merges the advantages to increase the surface within the capture area with the possibility to be monolithically integrated into a miniaturized device. The studies reported so far on analytical samples (pre-purified DNA)17–19,26 on cells18 prove this approach to have great potential towards the development of a genetic point-of-care device.
In this contribution, we present new miniaturised microfluidic biofilter (BF) devices based on silicon micropillars. They are analytically characterised regarding their efficiency to catch and release DNA. We have used HBV (Hepatitis B Virus) clone solutions at a concentration typical of real samples for infectious diseases (105 copies) as analytical samples. The device has been developed in three different geometrical structures, so that the effect of the Surface-to-Volume Ratio (SVR) on the extraction efficiency can be investigated. The total extraction performance was also evaluated and compared with those of commercial kits. Finally, the device exhibiting the best extraction efficiency was tested in combination with a silicon integrated real time PCR amplification chip as a preliminary step towards the development of genetic point-of-care devices.
Filter type | Pillar diameter (mm) | Pillar pitch (mm) | Pillar height (mm) | Total number of pillars | Total SVR (mm−1) |
---|---|---|---|---|---|
BF-1 | 0.012 | 0.008 | 0.1 | 23![]() |
39.9 |
BF-2 | 0.015 | 0.01 | 0.1 | 12![]() |
30.5 |
BF-3 | 0.015 | 0.015 | 0.1 | 7380 | 22.5 |
The microfabrication of these devices was carried out by a standard VLSI (Very Large Scale Integration) technology based on multi-step processes. The first step was the cleaning of the wafer with a solution of H2O2:
H2SO4 (1
:
1) followed by washing with deionized water and drying at 90 °C. After that, the silicon wafer was coated with a positive photoresist layer (KrF (248 nm) purchased from FujiFilm) of 4.7 μm by a spin coating process (1000 rpm, for 1 min). UV light irradiation was performed through a photomask, and the development process with TMAH (tetramethylammonium hydroxide)-based solution (OPD4262 purchased from FujiFilm) was carried out. After that, the microstructures were fabricated by silicon etching using a standard Bosch process (United States Patent 5501893) forming silicon channels (18 mm long × 1.8 mm large), the active pillar areas (2.5 mm long × 1.8 mm large) and the inlet and outlet holes (4 mm diameter), respectively. A next stripping step by oxygen plasma was carried out to remove the residual photoresist. A subsequent final dry oxidation process at 1000 °C with O2, and H2O gas (rate 1 cm s−1) was carried out, leading to a surface silicon oxide layer of 320 nm in thickness. This was measured by Scanning Electron Microscopy (SEM) analysis obtained using a high-performance Schottky field emission, LEO 1550 SEM. The instrument was operated at 5 kV in a secondary electron imaging mode.
The silicon part was then sealed with a glass slide (0.5 mm thickness) by a standard anodic bonding process producing the final microchip (Fig. 2a). The internal volume of the device was 5 μL.
The microstructures of the microfabricated devices were imaged by both optical microscopy (OLYMPUS SZX16) and SEM.
All experiments were performed at room temperature.
The recovered samples (not-bound solution, EB1-elution solution and EB2-elution solution) together with the starting solution were quantified in terms of HBV clone concentration by using real time PCR. The experiments were performed on an AB7500 Thermo-Cycler. 5 μL of the above-mentioned solutions were added to 20 μL of a HBV kit master mix. The PCR thermal cycling was as follows: 10 min at 95 °C, followed by 45 cycles of 95 °C for 15 s and 60 °C for 60 s. The quantitative measurements of HBV clone concentration were performed by a calibration curve calculated with the standard HBV concentrations of 1 × 105, 1 × 104, 1 × 103 and 1 × 102 cps.
For comparison, the same starting analytical sample (solution of the HBV clone containing a total amount of 5 × 105 cps) was extracted using two commercial kits: (a) a Magazorb DNA Mini-Prep Kit (Promega) and (b) a Qiagen QIAamp DNA Mini Kit (Ref. 51306), according their Instructions for Use.
5 μL of both starting and eluted solutions were pipetted into two out of the 6 chambers of the RT-PCR chip, and pre-loaded with 10 μL of the HBV kit master mix. The others 4 chambers were loaded with 1 negative sample (1 μL of water + 14 μL of HBV master mix) and three positive controls (1 μL of 1 × 106, 1 × 104 and 1 × 103 cps μL−1 + 14 μL of HBV master mix) for the calibration curve. The real time PCR process was executed using the thermal protocol reported in section 2.5.
Fig. 2 illustrates a typical image of the fabricated BF microdevice (Fig. 2a) and the structural features of the pillar areas in the three device typologies (Fig. 2b and c). In particular, Fig. 2b clearly highlights the three kinds of packing pillar arrangements differing from each other in pillar diameter (12 μm in the case of BF-1 and 15 μm for both BF-2 and BF-3) and inter-pillar distance (8 μm in the case of BF-1, 10 μm for both BF-2 and 15 μm for BF-3). It can be also noticed that the pillars are arrayed in a close-packed arrangement in order to maximise the packing feature and to increase the surface area as much as possible. In light of this, each microstructure contained a different number of pillars from about 23000 in the case of BF-1, to 12
000 for BF-2 and to 7300 for BF-3. The pillars are cylindrically shaped as shown in the SEM image reported in Fig. 2c.
The above discussed geometrical features yield different surface-to-volume ratios (SVR). As summarised in Table 1, the device exhibiting the highest SVR value is BF-1 (39.90 mm−1) while BF-2 and BF-3 exhibit lower values equal to 30.105 mm−1 (BF-2) and 22.50 mm−1.(BF-3), respectively. It can be noticed that the major contribution to the total S/V comes from the active pillar area. Such an area contributes to the total surface of the BF chip, about 78% in the case of BF-1, 71% in the case of BF-2 and 61% in the case of BF-3. On the basis of this finding, we can reasonably attribute the major impact in the DNA capture to that area.
We firstly evaluated both the binding and elution capability of the above-described microstructures (see section 2.5). After sample loading, the unbound, washed and eluted DNA was measured by real time PCR on the tube. Fig. 3 illustrates the comparison of the obtained data in terms of DNA amount (total cps) for the three typologies of BF devices.
It can be clearly noticed that the DNA binding capacity increases with the SVR of the device. The BF device exhibiting the highest SVR (BF-1) also shows the highest values of captured DNA. This latter is defined as the difference between the total starting DNA (blue bar) and the sum of total unbound (orange bar) and total washed DNA (grey bar). In agreement with this, the calculated binding efficiency (see eqn (1)†) corresponds to 54% for BF1 (SVR 39.9 mm−1) versus 26% for BF-2 (SVR 30.5 mm−1) and 22% for BF-3 (SVR 22.50 mm−1). This is expected on the basis of the increasing number of potential binding surface sites upon SVR increasing. Even if the trials were performed on the uncoated SiO2 surface, a remarkable amount of DNA was captured: about 1.8 × 105 cps for BF1, 1.1 × 105 cps for BF2 and 8.0 × 104 cps for BF3 in the case of EB1 (yellow bars of Fig. 3) and about 1.1 × 105 cps for BF1, 7.5 × 104 cps for BF2 and 5.5 × 104 cps for BF3 in the case of EB2 (blue bars of Fig. 3). This is due to the presence of silanol groups on the surface formed during the chemical treatment with NH4OH:
H2O2
:
H2O. These groups probably induce the formation of hydrogen bonds between the surface and the skeleton of the DNA molecule in the presence of high ionic strength solutions. Further mechanistic investigations are currently under way in our laboratories in this concern.
It should be emphasized that the binding capacity occurs without the employment of chaotropic salts or ethanol that are well-known inhibitors of the PCR or other molecular techniques and make the design of miniaturized systems more complex (vide supra).
In terms of elution capability, we observe that water (EB-1) works better than the low-ionic strength buffer EB-2. In fact, the calculated elution efficiency percentage (see eqn (2)‡) in the case of water was about 50% higher than EB-2 buffer (75% water versus 50% EB-2). This can be tentatively attributed to the better ability of water to make the electrostatic interaction between DNA and surface hydrating DNA molecules thermodynamically unstable.
Finally, we calculated the total extraction efficiency of the three types of BF devices by means of eqn (3)§ and compared them with those related to two commercial state-of-the-art kits, Magazorb and Qiagen based on magnetic beads and silica filter, respectively. Fig. 4 illustrates the obtained results.
It is noteworthy that BF-1 shows the best extraction performance achieving about 36% efficiency using water as elution buffer versus 22% for BF-2, 16% for BF-3 (blue bars) and 20% for both Magazorb and Qiagen (green bars). Using EB-2 as elution buffer (orange bars), the final extraction efficiency of BF-1 is comparable with those of the commercial kits, while those of BF-2 and Bf-3 were slightly lower. The latter finding is in line with the previous results discussed above for elution efficiency. As far as the better result displayed by BF1 is concerned, we believe that a probable synergy between the filter structure and the experimental protocol adopted may play a key role.
The on-chip amplification results on the extracted sample by BF1 are shown in Fig. 5b. The Ct values for both the starting (5 × 105 cps) and the BF1-extracted solutions were 22.4 and 26.3, respectively. The final concentrations of these solutions were calculated by the calibration curves embedded into the chip. Specifically, it was obtained by means of the Ct values exhibited by the three positive controls (Ctrl + 1 × 106, 1 × 104 and 1 × 103 cps) equal to 19.4, 23.9 and 31.0, respectively (the negative (Neg) control gave no results as expected). The calculation led to values of about 1.9 × 105 for the starting solution and 7.8 × 104 for the extracted solution. Based on these results, the extraction efficiencies calculated by eqn (3)§ correspond to about 40%. This is fully in agreement with the results obtained previously in the analytical characterization of BF1, and confirms the best extraction performance with respect to the commercial kits.
Furthermore, our results indicate that technology that uses micropillars in combination with an integrated real time PCR microchip can be very appealing for future developments of a fully-integrated genetic PoC device able to perform molecular genetic analysis close to the patient.
The findings here discussed encourage us in continuing our work towards the full integration of DNA extraction–PCR amplification modules in a monolithic device. This will allow to manage in one step the total molecular analysis process starting from real samples. In this case, the integration of all molecular steps in a monolithic device offers the advantages to perform molecular analyses easily in one step by non-specialised personnel. An additional benefit is the low cost of the genetic analysis since the volumes of reagents involved (few μL) are reduced compared to the current commercial kit (some tens of μL).
This journal is © The Royal Society of Chemistry 2017 |