O.
Mosley
a,
L.
Melling
a,
M. D.
Tarn
b,
C.
Kemp
b,
M. M. N.
Esfahani
c,
N.
Pamme
*b and
K. J.
Shaw
*a
aFaculty of Science and Engineering, Manchester Metropolitan University, Chester Street, Manchester, M1 5GD, UK. E-mail: k.shaw@mmu.ac.uk
bDepartment of Chemistry, University of Hull, Cottingham Road, Hull, HU6 7RX, UK. E-mail: n.pamme@hull.ac.uk
cSchool of Engineering, University of Hull, Cottingham Road, Hull, HU6 7RX, UK
First published on 10th May 2016
Despite recent advances in microfluidic-based integrated diagnostic systems, the sample introduction interface, especially with regards to large volume samples, has often been neglected. We present a sample introduction interface that allows direct on-chip processing of crude stool samples for the detection of Helicobacter pylori (H. pylori). The principle of IFAST (immiscible filtration assisted by surface tension) was adapted to include a large volume sample chamber with a septum-based interface for stool sample introduction. Solid chaotropic salt and dry superparamagnetic particles (PMPs) could be stored on-chip and reconstituted upon sample addition, simplifying the process of release of DNA from H. pylori cells and its binding to the PMPs. Finally, the PMPs were pulled via a magnet through a washing chamber containing an immiscible oil solution and into an elution chamber where the DNA was released into aqueous media for subsequent analysis. The entire process required only 7 min while enabling a 40-fold reduction in working volume from crude biological samples. The combination of a real-world interface and rapid DNA extraction offers the potential for the methodology to be used in point-of-care (POC) devices.
Rapid and efficient diagnosis is thus important in eradicating the infection and reducing the risk of gastric cancer development. To this end, microfluidic and lab-on-a-chip (LOC) devices offer considerable advantages for use in point-of-care (POC) diagnostics6,7 due to increased analysis speed and sensitivity, reduced reagent usage and the possibility for full automation. Despite this great potential, the development of real-world sample introduction interfaces remains challenging. Currently, the majority of published integrated devices either use simulated samples or require excessive off-chip or on-chip sample pre-treatment to achieve the desired specimen volume reduction and target concentrations. Simulated samples include the use of a few microlitres of highly concentrated bacterial cell cultures8 or high virus titre matrices9 that rarely represent target concentrations and purities found in clinical samples. Low target concentrations, such as those present in urine and stool samples, therefore require the use of larger sample volumes to assure sensitivity of the assay. In particular, the analysis of stool samples results in the need for considerable off-chip sample pre-treatment, such as centrifugation and filtration steps10 and chemical lysis11 prior to addition to a microfluidic device, all of which can be somewhat time-consuming. Furthermore, research in this area has focussed on the detection of infectious agents, such as Clostridium difficile, which cause diarrhoea resulting in liquid stool samples that are easier to introduce into microfluidic systems for analysis.10,12,13 Thus, the development of a real-world interface for the direct manipulation of crude biological samples has largely been ignored, and represents a barrier between the research and clinical environments.
Superparamagnetic particles (PMPs) have become very popular as solid supports for nucleic acid purification in the detection of infectious diseases.14 With a suitable surface functionality, such as silica or chitosan, the particles will capture DNA in a sample and their magnetic properties allow them to be held in place by an external magnet while the sample is removed and washing steps are applied. However, these methods typically require a great deal of time and manual handling. As a consequence, magnetic particle-based procedures have been incorporated into microfluidic devices with great success,15–17 thanks to the reduction in diffusion distances, procedural times, and the ease of particle manipulation. Magnetic particle-based procedures integrated with microfluidic devices have proven particularly effective for the purification of nucleic acids prior to their amplification and analysis,18 but many techniques involve complex chip setups19,20 and laborious multi-step procedures involving the trapping of magnetic particles while solutions are pumped over them.20
A simple method of achieving DNA or RNA extraction involves the introduction of PMPs into a contained sample volume, before moving the particles via a magnet through multiple washing solutions, leaving behind any unwanted and unbound material. Early examples of this mechanism employed the use of droplets on open, superhydrophobic microfluidic platforms, in which magnetic particles would be moved between stationary sample and washing droplets separated via an immiscible phase such as oil21–24 or air.25–27 However, these techniques often require either mechanical25,26 or electromagnetic21,23,24 actuation, adding complexity to the system in terms of both fabrication and operation.
A recent development from the group of Beebe is that of immiscible filtration assisted by surface tension (IFAST), in which rather than having solutions contained in droplets they are instead added to interconnected microwells separated by small “gated” regions.28 The chambers are filled with alternating aqueous and oil phases to form “virtual walls” between each chamber, controlled by the surface tension, but allowing magnetic particles to be pulled through these walls and thus through each chamber in one smooth yet fast motion via a handheld magnet. This allows simple and rapid DNA extraction to be performed with minimal setup and materials, and by the “unskilled” end-user. The standard IFAST design features three chambers consisting of (i) aqueous sample solution to which PMPs are added, (ii) an immiscible oil phase for washing of particles, and (iii) elution buffer that can be collected for off-chip nucleic acid amplification and analysis. So far this method has been applied to the purification of RNA28–31 and DNA,31,32 as well as for cell isolation33–35 and immunoassays.36 Further developments have included automation of the devices,37,38 their fabrication from wax,30,38 and variants such as vertical IFAST (VerIFAST),34,35 and SNARE (selective nucleic acid removal via exclusion).31 Similar techniques have also been developed by other research groups, in which different immiscible phases have been employed including liquid wax,39 paraffin wax,40 and air.41–44 Furthermore, miscible phases have recently been employed for particle washing by using phaseguides to pattern interfaces between adjacent aqueous solutions.45,46
Here, we have exploited and considerably modified the IFAST principle to develop a sample introduction interface that enables direct processing of stool samples. Stool samples are particularly challenging for diagnostic analysis through molecular biology techniques as they exhibit high variability in terms of consistency of samples, the presence of PCR inhibitors and low target analyte concentrations, hence the requirement of the sample pre-treatment and pre-concentration steps described earlier. IFAST is usually conducted in chambers of 10 μL volumes, while in our high volume IFAST system the issue of low biomarker concentration is negated by the use of a large sample chamber, while the IFAST process itself allows rapid DNA purification, concentration, and elution, in a single device (Fig. 1). Initial experiments were performed using E. coli as a model Gram-negative pathogenic target before moving onto analysis of H. pylori (also Gram-negative) from clinical stool samples.
Fig. 1 Schematic of the DNA extraction process, showing (a) sample loading and cell lysis, (b) mixing of PMPs with the sample for DNA binding, (c) transfer of PMPs through the immiscible phase for washing, and (d) elution of DNA from the PMPs followed by collection for off-chip analysis. The design was later amended to include two extra downstream chambers for additional washing (see Fig. 3 for further details). Schematics are not to scale. |
The novelty of the reported approach lies in (1) the design of the sample chamber which enables a 40-fold reduction in working volume, (2) the choice of detergent-free solid cell lysis and DNA binding agent that is reconstituted by the sample itself, (3) a unique PDMS/optical adhesive bonding approach which facilitated the formation of a stable but immiscible barrier, and (4) the real-world interface created using a septum-based sample introduction design.
Fig. 3 (a) Schematic of the 5-chamber DNA purification device. (b) Photograph of the PDMS chip filled with inks and oil. |
A real-world interface was constructed for sample introduction via the holes in the optical film lid above the sample chamber, consisting of a septum for sample introduction and an air vent (Fig. 2c and d). The septum was prepared by cutting a standard capillary gel electrophoresis septum to size and seating it in the top of a cut-to-size pipette tip (100 μL) that was attached to the optical film lid via double-sided tape. The vent was fabricated from a filter pipette tip (10 μL) that was also attached to the lid via double-sided tape. The vent allowed air to be expelled when the sample chamber was filled. The assembly and interfacing of the IFAST the device was the same for both the 3-chamber and 5-chamber chip designs.
As described earlier, a number of immiscible phases have been used as the washing solution in IFAST, including liquid wax,39 paraffin wax,40 olive oil,28 and air.41–44 Here, mineral oil was chosen as the immiscible phase due to its purity and compatibility with downstream biochemical applications. Upon addition of the sample to the chamber, a handheld magnet was used to mix the PMPs with the sample for 5 min, reconstituting the GuHCl to a concentration of 5 M and allowing binding of the DNA to the particles. Finally, the PMPs were quickly transferred across the immiscible barrier by the handheld magnet and into purified water, where the DNA was allowed to elute from the particles for 2 min (Fig. 1). Unwanted components of the stool sample matrix were left behind in the sample chamber. The use of oil phases in both the 3- and 5-chamber chip designs acted to remove potential PCR inhibitors, such as complex polysaccharides, from the stool samples. In addition, the 5-chamber chip contained an additional wash step (5 M GuHCl) to ensure the DNA remained bound to the PMPs and could be separated from any remaining inhibitors, as previously described with other types of biological samples.28
Several parameters were tested using the described DNA extraction process, including: (i) lysis efficiency of Gram-negative bacterial cells (E. coli) using powdered GuHCl stored in the sample chamber (‘Evaluation of stored reagents parameters’ section), (ii) DNA extraction efficiency from cultured bacterial cells (E. coli) (‘DNA extraction efficiency’ section), (iii) evaluation of purity of DNA extracted from real clinical stool samples (‘Evaluation of clinical stool samples’ section), and (iv) amplification of H. pylori targets following IFAST-based extraction (‘Evaluation of clinical stool samples’ section).
DNA amplification was achieved using a 25 μL polymerase chain reaction (PCR) mixture prepared from the following: 5 μL of purified template DNA solution (taken directly from the IFAST device), 0.4 μM each primer, 1× Biomix™ containing 0.2 mM each dNTPs, reaction buffer (100 mM Tris-HCl [pH 8.3], 500 mM KCl, 2.5 mM MgCl2, and 0.01% (w/v) gelatin), and 2.5 U of Taq DNA polymerase. Samples were run on a Q-cycler 96 thermal cycler under the following conditions: an initial denaturation at 94 °C for 10 minutes followed by 40 cycles of 94 °C for 2 minutes, 55 °C for 2 minutes and 72 °C for 2 minutes, with a final extension of 72 °C for 10 minutes.
Following amplification, PCR products and a DNA size ladder (Hyperladder V) were run on a 2% (w/v) agarose gel until adequate separation had been achieved and visualised using a UV transilluminator.
Fig. 4 Lysis efficiency of the stored 5 M GuHCl reagent both on- and off-chip (n = 6) for 2.54 × 106 (black) and 2.54 × 104 cells (grey). |
The amount of particles that can be used in the IFAST device was restricted by the geometry of the microfluidic conduits. Previous studies demonstrated that 0.24 mg of MagneSil PMPs could be transported across the immiscible phase without particle loss or blocking of the device, and therefore this amount was chosen to achieve maximum DNA binding and transport.50
Fig. 5 DNA extraction efficiency showing a strong linear correlation (Pearson's R, R2 = 0.98259) between the amount of DNA added to the system (ng) and the amount of DNA recovered from the system (ng). |
Negative controls were also included, in which samples containing no DNA were added to the IFAST device and underwent the DNA extraction process. No DNA was detected in the eluent of these samples (n = 3).
Sample | DNA concentration (ng μl−1) | Purity (260 nm/280 nm) |
---|---|---|
1 | 31.8 | 1.3 |
2 | 63.5 | 1.1 |
3 | 19.1 | 1.3 |
4 | 295.0 | 1.1 |
5 | 93.8 | 1.2 |
6 | 154.0 | 1.3 |
Average | 109.5 | 1.2 |
Following PCR, weak or no PCR products were observed, indicating that the samples were not sufficiently pure and free of inhibitors for successful amplification to be achieved. Therefore the chip design was modified to include an additional wash step, yielding the 5-chamber design shown in Fig. 3. An improvement was seen in the purity of the extracted samples using the modified chip design, with purity values up to 2.0 obtained (Table 2). In order to account for the wide variety in composition of stools, all subsequent samples analysed were assigned a value based on the Bristol Stool Chart which classifies samples on a 7-point scale from separate hard lumps (type 1) to entirely liquid (type 7), with an ideal stool being smooth and sausage-like (type 4).52 Comparison of the purity of the samples to their original appearance (based on values assigned from the Bristol Stool Chart) showed a strong correlation (R = 0.96 and P < 0.001 Pearson's R), with more liquid samples (e.g. types 6 and 7) producing higher extracted DNA purities. However, there was no correlation between the appearance of the stool and the amount of DNA that was obtained. In addition, successful amplification of the UreC target gene (PCR product size = 274 bp) was achieved on those samples which were extracted using the IFAST device (Fig. 6). As expected, no PCR products were observed for those stool samples which had previously tested negative for H. pylori. None of the clinical samples tested proved positive for CagA (expected PCR product size = 400 bp).
Fig. 6 Gel electrophoresis image showing: (L) DNA size ladder; (1–8) amplified faecal samples extracted using the IFAST device; (N) negative control. |
IFAST has been previously demonstrated for rapid nucleic acid and cell purification purposes, and here we have significantly adapted the procedure to enable analysis of stool samples via a real-world interface. Firstly, our design includes a large sample reservoir that accommodates 400 μL of crude sample without the need for sample pre-treatment. For clinical samples, the target analyte concentration may be very low and therefore the larger the sample volume that can be accommodated the more likely the chances of successful extraction of the target of interest. Secondly, on-chip cell lysis and DNA binding to the solid phase supports (PMPs) was achieved on-chip by preloading the chamber with the solid chaotrope49 and dried PMPs. Furthermore, sample loading was facilitated by an incorporated septum (Fig. 2b and c), keeping the sample sealed within the chamber. In a recent publication, cell lysis and DNA binding was performed on an IFAST device and proven to be as efficient as off-chip cell lysis prior to IFAST extraction,32 but required the addition of lysis buffer to the sample and incubation of 30 minutes in an oven. By comparison, the method described here requires only the addition of the crude sample to reconstitute the solid GuHCl and PMPs. This allows easy storage on the microfluidic device; increasing analysis speed and user friendliness. Detergent-free lysis is not a requirement for IFAST-based analysis as such systems have been shown to be compatible with common lysis and elution buffers containing detergents such as 1% Triton X-100, 1% LiDS or 2% SDS.28 The ability to reconstitute the reagent to a known concentration in the sample itself also makes it easier for the operator to use and reduces the number of manual steps required. Thirdly, instead of bonding the PDMS microfluidic layer to a glass substrate, we opted for an optical adhesive film as the bottom substrate of the chip. This very simple and rapid bonding approach has the added benefit that its hydrophobic surface properties allow the transfer of magnetic particles through the immiscible barrier without the addition of detergents to lower the interfacial energy. It also has advantages over plasma bonding in terms of ease of use and accessibility to equipment. Not every lab has access to a plasma oven for bonding but the optical adhesive is readily available from a number of suppliers and is more cost effective and easier to use. It is also specifically designed for PCR-based applications and has good optical properties which would be beneficial for future integration of real-time isothermal amplification to create a complete point-of-care system.
This miniaturised approach offers advantages over current commercially available stool DNA extraction kits, such as the QIAamp Stool DNA Mini Kit, as it offers a 7 fold reduction in the time taken for analysis, enables further pre-concentration of target DNA by eluting in a volume of 10 μL compared to 200 μL and is easy to use (e.g. multiple heating and centrifugation steps are not required, no proprietary chemicals are used).53
The simplicity and ease of use of the presented real world interface is perfectly suited for the requirements of a POC diagnostic device as results can be obtained whilst the patient is waiting, ensuring rapid identification of pathogenic strains of H. pylori from stool samples. Future work would look at evaluating the IFAST system with additional stool samples, allowing replicates of all possible sample types (based on the Bristol Stool Chart) to be performed, particularly with respect to the purity of the eluted DNA. This would also allow a more in depth study to be carried out on the number of H. pylori positive samples which express CagA. In addition, future work aims to integrate this work with real-time isothermal amplification of the pathogenic target to create a complete point-of-care system. This could lead to more immediate therapy and potentially a reduction in adverse conditions as a result of infection, such as gastric carcinogenesis. The microfluidic system could also be readily adapted to accommodate other pathogenic targets present in stool samples, such as Clostridium difficile or rotavirus.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6lc00228e |
This journal is © The Royal Society of Chemistry 2016 |