Biomedical Microdevices

, 11:1279

A self-contained fully-enclosed microfluidic cartridge for lab on a chip

Authors

    • Institute of Microelectronics, 11 Science Park Road
  • Lih Feng Cheow
    • Institute of Microelectronics, 11 Science Park Road
  • Kum-Cheong Tang
    • Institute of Microelectronics, 11 Science Park Road
  • Shien-Eit Yong
    • Institute of Microelectronics, 11 Science Park Road
  • Eleana Kye-Zheng Ong
    • Institute of Microelectronics, 11 Science Park Road
  • Lionel Wong
    • Institute of Microelectronics, 11 Science Park Road
  • William Cheng-Yong Teo
    • Institute of Microelectronics, 11 Science Park Road
  • Hongmiao Ji
    • Institute of Microelectronics, 11 Science Park Road
  • Siti Rafeah
    • Institute of Microelectronics, 11 Science Park Road
  • Chen Yu
    • Institute of Microelectronics, 11 Science Park Road
Article

DOI: 10.1007/s10544-009-9347-z

Cite this article as:
Yobas, L., Feng Cheow, L., Tang, K. et al. Biomed Microdevices (2009) 11: 1279. doi:10.1007/s10544-009-9347-z

Abstract

We describe a self-contained fully-enclosed cartridge for lab-on-a-chip applications where sample and reagents can be applied sequentially as is performed in a heterogeneous immunoassay, or nucleic acid extraction. Both the self-contained and fully-enclosed features of the cartridge are sought to ensure its safe use in the field by unskilled staff. Simplicity in cartridge design and operation is obtained via adopting a valveless concept whereby reagents are stored and used in the form of liquid plugs isolated by air spacers around a fluidic loop. Functional components integrated in the loop include a microfluidic chip specific to the target application, a novel peristaltic pump to displace the liquid plugs, and a pair of removable tubing segments where one is used to introduce biological sample and while the other is to collect eluant. The novel pump is fabricated through soft-lithography technique and works by pinching a planar channel under stainless-steel ball bearings that have been magnetically loaded. The utility of the cartridge is demonstrated for automated extraction and purification of nucleic acids (DNA) from a cell lysate on a battery-operated portable system. The cartridge shown here can be further extended to sample-in-answer-out diagnostic tests.

Keywords

MicrofluidicsLab on a chipCartridgeNucleic acidPoint of care

1 Introduction

The Lab-on-a-chip (LOC) technology has reached a certain level of maturity where point-of-care diagnostic assays can be realized by simply integrating microfluidic operations such as sampling (Burns et al. 1996), mixing (Sudarsan and Ugaz 2006), filtration (Wilding et al. 1998), extraction (Christel et al. 1999), and detection (Zhiqiang et al. 2007) into a single cartridge (Burns et al. 1998). It is believed that such portable assays would decentralize clinical laboratories and benefit public health tremendously by providing reasonably accurate and rapid test results at a lower cost without resorting to trained personnel and laboratory-bound instruments (Chin et al. 2007).

Ideally, a cartridge for a point-of-care use is preferred to incorporate several key features. The cartridge should be self-contained, inclusive of all the required reagents to simplify its operation in the field. In addition, the cartridge should be fully-enclosed not having any of its content exposed. This is important because some of the reagents, as well as the raw sample itself, might pose a potential threat to the user and environment (e.g. infectious samples, carcinogenic reagents). This feature implies that fluid propulsion must take place inside a closed system without leading to excessive pressure build up. Moreover, the cartridge should be self-sufficient and not rely on a bulky expensive instrument to pump and valve the sample and reagents. Furthermore, the cartridge should be ready to use when it is supplied with the raw sample. The one-time use cartridge should be equipped with all these features at a negligible cost and accomplish its function with almost dip-stick simplicity.

While various microfluidic devices have been successfully demonstrated for point-of-care use (Easley et al. 2006; Lee and Hsing 2006; Liu et al. 2004; Yeung et al. 2006), only few can fulfill the above features. For example, the biochip by Liu et al. for DNA analysis is a self-contained system with integrated microfluidic mixers, valves, and pumps (Liu et al. 2004). The biochip performs sample preparation, polymerase chain reaction (PCR), DNA hybridization, and electrochemical detection in an automated manner. Fluid propulsion is realized via inexpensive pumps using metal electrodes or resistive heaters as opposed to sophisticated moving mechanical parts. These pumps work by gas pressure generated from electrolysis of water or expansion of air pocket at elevated temperatures. Similarly, cost-effective valves are built using paraffin as an actuator material that undergoes solid-liquid phase transition in response to temperature change. Despite the fact that such components greatly simplify device design and manufacturing, complexity in device operation remains since each valve requires supervised control to ensure successful completion of individual assay steps. This intricate operation raises not only the instrument cost but also the risk of assay failure.

To minimize the use of valves and avoid intricate device operation, some researchers suggest employing a single microchamber for more than one function (Yeung et al. 2006). Accordingly, Obeid et al. (2003) and Linder et al. (2005) propose sequential delivery of liquid plugs isolated by air spacers into a microchannel without resorting to valves. The former study alternates sample and wash plugs as they pass through a microchannel reactor in an attempt to investigate high-throughput capability of continuous flow-through PCR. The latter demonstrates self-contained cartridges by arranging reagent plugs in commercial tubes that can be connected to a microfluidic device. The cartridges are capable of performing solid-phase heterogeneous immunoassay in a microchannel for biomarker detection. Nevertheless, the cartridges cannot be considered fully-enclosed because their free end must remain open to atmosphere while driving the reagent plugs into a microfluidic device via suction manually operated by the user. In addition, the user has to deliver the input sample through the same inlet before connecting the cartridge tubing to the device.

In this paper, we further develop this simple valveless concept into a self-contained fully-enclosed cartridge by integrating with a pumping mechanism, magnetic planar peristaltic pump (MP3), recently introduced by our group (Yobas et al. 2008). This integration removes the burden from the user to displace sample and reagents through the cartridge. The cartridge employs a novel concept of removable fluid storage elements through which the biological sample can be easily introduced or the assay product (eluant) can be collected at the completion of the assay. For this demonstration, the cartridge is built around a silicon-based microfluidic chip (as the LOC) by making use of off-the-shelf tubing. With the final goal of attaining a cartridge for sample-to-answer nucleic acid probe assay on complex biological samples, we apply the first prototype here to purification of DNA from an input sample containing lysate of cultured cells. The cartridge described here can also be adopted to facilitate other point-of-care assays which involve sequential injection of sample and reagents.

2 Concept

2.1 The pump

The pump (MP3) will be described here briefly. The pumping mechanism involves a set of stainless-steel ball bearings rolling on a planar substrate in a circular trajectory under the influence of revolving magnets. Aligned with the circular trajectory of the ball bearings is a fluidic channel that is buried inside the planar substrate. While rolling, the ball bearings continuously press on an elastomeric membrane which pinches the fluidic channel underneath and generates a peristaltic wave during each cycle of rotation. This peristaltic wave produces a pumping action which propels the liquid inside the channel in the direction of rotation. The liquid is continuously replenished from the pump inlet to the outlet which can be linked to the LOC.

2.2 The cartridge

Figure 1 diagrammatically describes the self-contained fully-enclosed cartridge introduced here. The cartridge consists of fluidic channel segments arranged in a pattern of concentric serpentine arcs. These channel segments, one of which in fact defines the pump, form a single closed loop when they are fluidically connected through a pair of detachable (non-planar) storage elements. These removable storage elements (e.g. plastic tubes) allow introducing the sample input and collecting the assay product, both of which are only accessible either immediately before or after use of the cartridge but not during manufacturing or loading of the cartridge. The LOC shown at the cartridge centre can be either a monolithic part of the cartridge or a detachable module mounted on the top.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-009-9347-z/MediaObjects/10544_2009_9347_Fig1_HTML.gif
Fig. 1

Planar diagram of the self-contained fully-enclosed cartridge in which the MP3 fulfils the integrated fluid delivery function. A single turn of the serpentine channel forms the peristaltic channel of the planar pump which gets squeezed by the ball bearing(s) rolling under the influence of revolving magnets located on the opposite side of the cartridge (not drawn). The individual channel segments are linked through a pair of removable non-planar fluid storage elements (tubes) to complete a single closed-loop channel and to lead to a fully-enclosed cartridge. The removable storage elements allow introducing the sample and collecting the assay product. Reagent plugs are isolated by air spacers and stored downstream side of the pump. The pump along with the remaining cartridge is filled with a single plug of working liquid (water). The pump propels the working liquid to sequentially displace the sample and reagent plugs through the LOC

The fluidic layout of the cartridge is designed for sequential injection of reagents as in heterogeneous immunoassay or nucleic-acid purification. To avoid use of valves, the design is such that all the reagents are stored and delivered through the LOC in individual fluid segments isolated by air spacers. To prevent the air spacers from entering into the pump and degrading its performance, the reagents and the accompanying air spacers are stored downstream side of the pump. The remaining channels including the pump are filled with an inert working liquid (e.g. water) to sequentially displace the downstream plugs containing the sample and reagents through the LOC.

Ideally, the cartridge contains all the reagents and requires only the input sample to be supplied by the user. This can be simply achieved by replacing a dummy fluid storage element on the cartridge with the one containing the input sample. The cartridge along with the ball bearings is then placed on a portable system which houses the revolving magnets and thereby drives the peristaltic action that displaces the liquid plugs through the LOC. The input sample and the subsequent reagents go through the LOC one at a time and replace the working liquid at the outer segment of the cartridge channel. A single cycle of the cartridge assay is said to be complete when the sample plug reaches the pump at which point the user is expected to stop the pump. Depending on the assay type, the product can be kept in a detachable fluid storage element at the outer segment of the cartridge for the user to collect. This output storage element can be replaced with a dummy element for safe disposal of the cartridge to avoid biohazard (e.g. infectious disease).

3 Experimental

3.1 Pump prototyping

Rapid prototyping of the pump was realized by replica-moulding a planar channel in poly(dimethylsiloxane) (PDMS) with a laser-patterned grinding tape as the template (Duffy et al. 1998; Luo et al. 2007). The grinding tape 130 µm thick (Furukawa Electric, Japan) was first attached to a polished surface on a silicon substrate. The pump layout was then transferred onto the tape by thermal ablation under a CO2 laser at a power 0.3 W (M-300 Laser Platform, Universal Laser Systems Inc., USA). The laser beam cut through the tape without damaging the silicon substrate. Unwanted sections of the tape were then carefully peeled off the wafer. The PDMS mixture (Sylgard 184, Dow Corning Inc., USA) was prepared (base:curing agent at a ratio of 1:10 by weight), degassed, and then poured onto the patterned tape. A poly(methyl methacrylate) (PMMA) ring barrier was placed around the patterned tape to keep the PDMS confined during thermal curing of the polymer. After peeling the cured PDMS layer (>1 mm thick) off the mould, fluidic vias were punched in the PDMS layer and bonded to a flat PDMS layer ∼ 0.5 mm thick by activating their surfaces in oxygen plasma. The pair was annealed at 150°C for 2 h to reinforce their bonding strength.

3.2 Cartridge prototyping

A research prototype of the proof-of-concept cartridge was implemented here by utilizing commercially available plastic tubing for the cartridge channels in a multi-layer structure. We believe that the overall cartridge structure can be greatly simplified in a fully developed version by leveraging from well-established manufacturing techniques.

Figure 2 shows three-dimensional (3D) graphical illustrations which reveal details of the cartridge prototype. The cartridge prototype consists of four main layers. An adhesive tape (Layer 1) is employed to secure the commercial tubing (Layer 2) in grooves behind a rigid planar disk (Layer 3). Each plastic tube is terminated with a pair of pins (stainless-steel tubes) penetrating vias through the disk. The plastic tubes form the concentric channels within the cartridge with the exception of the channel for the pump which is segregated and positioned in an elastomeric sheet (Layer 4a). The elastomeric sheet is directly mounted on the rigid disk such that the inlet/outlet ports mate with the corresponding pins. The rigid disk provides the required mechanical support for the elastomeric sheet as the ball bearings press down and roll on the pump channel. The sheet also contains windows so as to provide access to pins underneath for connecting the LOC and the removable fluid-storage elements (non-planar tubing). Using a double-sided adhesive tape, the LOC is mounted on a block having inlet/outlet holes (Layer 4b). For the sake of clarity, Fig. 2(b) provides an exploded view of the cartridge and Fig. 2(c) shows a detailed view of its fluidic connections.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-009-9347-z/MediaObjects/10544_2009_9347_Fig2_HTML.gif
Fig. 2

3D graphical illustrations of the cartridge prototype. a A fully assembled cartridge depicted with a set of three ball bearings positioned on the channel defining the pump and mounted with a pair of non-planar tubing segments as the removable fluid storage elements for the input sample and the assay product. b Exploded view of the cartridge showing individual structural layers. Layer 1: adhesive tape; Layer 2: commercial tubing segments terminated with L-shaped stainless steel tubes; Layer 3: structured rigid disk with backside grooves and corresponding vias to accommodate the commercial tubing segments; Layer 4a: pump (MP3); Layer 4b: the LOC and the intermediate layers to assist its fluidic interfacing with the remaining cartridge c A partially assembled cartridge showing details of the fluidic connections after leaving out the structured rigid disk and the pump (MP3)

The rigid disk here was made of a poly(methyl methacrylate) (PMMA) block with a diameter of 12 cm and a thickness of 3 mm. The disk was structured with concentric grooves and vias according to the cartridge layout by means of thermal ablation from the CO2 laser. Tygon® tubing segments (inner diameter 0.5 mm, Norton, UK) were cut into appropriate lengths so as to fit into the grooves. Terminals of each tubing segment were connected to pins (inner diameter 0.33 mm, OK International Inc., USA) which were inserted into the vias and bent into an L-shape. The tubing segments were then secured in the grooves with an adhesive grinding tape (Furukawa Electric, Japan). Tygon® tubing segments cut into appropriate lengths were also employed as non-planar removable fluid-storage elements. The elastomeric sheet of the pump was made by casting PDMS as described above. The LOC was selected from our silicon-based microfluidic chips for nucleic acid purification which we reported previously with a slight variation in design (Hui et al. 2007; Yobas et al. 2007; Yobas et al. 2005; Yobas et al. 2006). Fabrication of the LOC involved deep reactive ion etching silicon to a depth of 80 µm and glass anodic bonding. Fluidic access ports were created in silicon by back-side wet etching. Although the LOC was capable of cell separation, cell lysis, and nucleic acid purification all in consecutive steps through an integrated mixer, filter, and reactor combination, it was utilized here for the last step of nucleic acid purification. The LOC was treated as a single-inlet and single-outlet system and directly mounted on a PMMA block via medical-grade double-sided adhesive tape (Adhesive Research Inc., USA). The PMMA block and the adhesive tape were laser-drilled to create a pair of holes as the inlet/outlet ports before having the LOC mounted on. The holes in the block were inserted with short sleeves prepared from Tygon® tubing to attain leak proof seals around the respective pins of the cartridge.

3.3 Nucleic acid purification

To demonstrate the cartridge utility, we used it for solid-phase extraction of nucleic acids from a lysate of cultured cells. Nucleic acid purification by solid-phase extraction was realized in the LOC through selective binding or adsorption of nucleic acids to thermal-oxide silicon surface and borosilicate glass surface. The LOC was conditioned here first with alkaline wash NH4OH (28%) overnight to enrich its surface with negatively charged (OH) groups. Before being mounted on the cartridge, the LOC was equilibrated with 6 M Guanidine Hydrochloride (GuHCl), a chaotropic agent also present in the lysis buffer added to the input sample. GuHCl dehydrates the silica surface as well as the nucleic acid molecules. This allows a positively charged Guanidium ion to form a salt bridge between the negatively charged silica surface and the negatively charged nucleic acid backbone, thereby facilitating the adsorption. A fresh combination of the LOC and cartridge was used for each experiment.

We prepared the input sample by harvesting and resuspending NCI-H460 cells in 2 ml phosphate buffered saline (PBS) solution at a density of 2.5 × 105 cells/ml. Following an incubation period of three hours at room temperature, we further concentrated the sample by pelleting the cells at 5000 g for 10 min and resuspending them in 200 µl PBS. In the subsequent steps, we used reagents from QIAamp DNA Mini Kit (Qiagen, Germany). We added 20 µl QIAGEN Protease and 200 µl AL buffer (lysis) to the sample and mixed thoroughly to produce a homogenous solution. After a 10-min incubation period at 56 °C, we added 200 µl ethanol (96 – 100%) to the sample. A 20 µl aliquot of the sample was then loaded into each cartridge via the non-planar tubing segment. The purification protocol involved passing through the chip the following sequence of sample/reagent plugs: the sample (20 µl); AW1 and AW2 wash buffers (13 µl each); and three AE elution buffers (11 µl each). During control experiments, we replaced the MP3 with a commercial pump (M6, VICI AG, Switzerland).

3.4 PCR amplification

DNA amplification through polymerase chain reaction (PCR) was carried out using Taq PCR Core Kit (Qiagen, Germany). 5 µl of each 11 µl eluant collected at the conclusion of a cartridge run was amplified in a 25 µl reaction mixture containing 10× PCR buffer, 5× Q-Solution, 25 mM of MgCl2, 10 mM of each dNTP, 10 µM of each primer, and 5 units of Taq DNA Polymerase. The coding strand was 5′-ATGGTGGGCATGGGTCAGA-3′ and the non-coding strand 5′-GCCACACGCAGCTCATTG-3′ both designed in house to amplify a 170-bp fragment of the β-actin gene. The thermal cycling protocol was as follows: 95 °C for 5 min (initial PCR activation step), then 35 cycles of 95 °C for 15 s (denaturation), 55 °C for 30 s (annealing), and 72 °C for 15 s, followed by a single final extension for 10 min at 72 °C. DNA amplification was confirmed by observing the PCR products stained with ethidium bromide under UV illumination once they were separated by gel electrophoresis.

3.5 Loading the cartridge

For external loading, all the parts of the cartridge were assembled except the non-planar fluid storage element for the input sample, thereby leaving the input pins exposed. The input pin that leads to the inlet of the LOC was connected to a computer-controlled precision pump (M6, VICI AG, Switzerland) in suction mode at a rate 30 µl/min while the other pin to a tube dipped into a reservoir of loading liquid. The first loading liquid was a plug of ultra-pure water (TKA GenPure, Germany) to function as a working liquid. Subsequent loadings included plugs of other liquids specific to each experiment as described. These liquid plugs were loaded according to their reverse order of sequence passing through the chip. For example, the plug that would pass through the chip last was loaded first whereas the plug to pass through the chip first (e.g. sample) was loaded last (by using a non-planar tubing segment attached to the input pins). Before loading each plug, we drew air into the cartridge and created an air spacer about 10 mm long to prevent the succeeding plugs from mixing with each other. We used the cartridges immediately upon loading of the plugs.

3.6 Determination of the sample/reagent cross-contamination

To quantify the level of cross-contamination within the cartridge as a result of sequential displacement of plugs, we utilized a plug of 20 µM fluorescein in water as the input sample and a sequence of water plugs as the subsequent reagents. We operated the MP3 at 20 rpm displacing all the plugs through the chip once and collected them at the conclusion of a single run cycle for analysis. We measured the collected plugs for traces of fluorescein using a commercial fluorometer (Perkin Elmer, USA) and referenced them to a set of standards spanning a range up to 30 µM. For each water plug, we calculated the percent fluorescein contamination relative to the fluorescein plug.

3.7 Pump characterization

Typical pump characteristics such as flow rate and backpressure were measured on an assembled cartridge for various speeds of the magnetic field revolution. The cartridge was assembled with all the parts after excluding the non-planar fluid storage element corresponding to the assay product, leaving the output pins exposed. The output pins were then connected to two separate water columns through adaptors with negligible fluidic resistance as compared to that of the cartridge. The cartridge, before its connection to the water columns, was filled with water while ensuring no air bubble being trapped. At the beginning of the experiments, both columns were at the same height. The pumping was initiated by continuously revolving the magnets in the clockwise direction at a constant speed. Differential height of the water columns was recorded at specific time intervals to calculate the flow rate and pressure head. The clockwise pumping was maintained until the columns reached a steady level which showed no discernable change. Subsequently, the direction of the pumping was reversed by continuously revolving the magnets counter-clockwise at a constant speed. The measurements were recorded until a steady level of the columns was reached. This back and forth pumping was repeated under various motor speeds to derive the pump characteristics under the load of the cartridge.

3.8 Portable system

Rare-earth magnets, neodymium (Nd2Fe14B), were retrieved from failed hard-disk drives and stripped of their shields. The magnets were positioned off-centered on a 4-inch aluminum disk and clamped in place by counter magnets located on the opposite side of the disk. The disk was then mounted on a DC motor shaft through a gear box (2342S012CR, Faulhaber). The motor was controlled by a microcontroller (MCDC 3006 S, Faulhaber). The cartridge was placed on four identical posts surrounding the disk and ensuring a small separation from the revolving magnets (clearance < 1 mm). Positions of the magnets were adjusted such that the stainless-steel ball bearings Ø 6 mm (AEC, Singapore) disposed on the cartridge aligned with the channel of the pump while rolled under the influence of the revolving magnets.

4 Results and discussion

Figure 3 shows pictorial details of the portable system and the cartridge prototype realized here. The system is depicted also without the cartridge positioned on its top to reveal the rotating aluminum disk and the magnets (Fig. 3(a)). The three identical magnets are symmetrically positioned (120º apart) on the aluminum disk and held in place by counter magnets located on the opposite side of the aluminum disk (not shown). The four identical stands protruding up from the system body provide the necessary support for the cartridge and prevent its direct contact with the revolving magnets. These stands also make sure that the channel of the pump is aligned with the revolving magnets when a cartridge is properly placed on top (Fig. 3(b)). The cartridge is shown entirely filled with dyed water to highlight its fluidic parts including the non-planar tubing segments and the channel within the PDMS layer. The tubing segments in the laser-ablated grooves on the backside of the acrylic disk are not discernable since the grooves are not optically clear (Fig. 3(c)).
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-009-9347-z/MediaObjects/10544_2009_9347_Fig3_HTML.gif
Fig. 3

Pictures of the cartridge prototype (filled entirely with dyed water) and the custom-built portable system for the controlled revolution of the magnets. a The portable system: (left) shown without the cartridge placed on its top (the 100-mm-diameter aluminum disk connected to the motor shaft and the three rare-earth magnets symmetrically positioned ∼ 120° apart on the disk); and (right) shown with the cartridge placed on top. b A close-up view of the cartridge placed on the four stands protruding up from the system body: the three ball bearings shown self-aligned with the channel of the pump as they trace the revolving magnets. c The cartridge shown from above. The bold arrows point at the flow direction in the cartridge when the ball bearings revolve clockwise

Experimentally derived pump characteristics (volumetric water flow rate and backpressure) of a representative cartridge that is entirely filled with water and no air spacer are shown in Fig. 4. Under a steady rotational speed of 40 rpm, the pump can generate backpressure as high as 10 kPa. Maximum backpressure values are observed to scale with the corresponding motor speed by a factor of 1/4. During a steady rotation of the motor in either direction, the pumping rate drops almost linearly with the rising backpressure. The pumping rates in clockwise and counter-clockwise directions may not fully agree largely due to inaccuracy in the alignment of the revolving magnets to the pump channel. The pumping rates are always higher at negative backpressures where the pressure gradient also works for the flow direction. Slope of the trendlines observed at various motor speeds remains more or less same since the slope is defined by the fluidic resistance through the cartridge, particularly the LOC, which is fixed by the geometry. With the fluidic resistance and the range of motor speeds here, the flow rates in the current setup may vary from a few µl/min to > 100 µl/min.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-009-9347-z/MediaObjects/10544_2009_9347_Fig4_HTML.gif
Fig. 4

Measured volumetric water flow rate through a representative cartridge as a function of its backpressure (differential water column). The cartridge was entirely filled with water. Each set of measurements was taken at a constant motor speed (legend). The solid and blank symbols respectively refer to clockwise and counter-clockwise direction of the flow in the pump. For the measurements taken at 0 rpm, the ball bearings were removed and the flow was driven under a negative pressure gradient across the cartridge. At negative pressures, since the gradient also contributes to the flow, the rate of flow increases. The trend lines through each set of measurements are drawn to guide the eye

We can evaluate the pumping efficiency by comparing the experimental flow rates with the expected flow rates. Assuming that the entire volume of the pump channel gets displaced after each revolution of the magnets, the flow rate should then be multiples of the volume of the channel as determined by the rate of revolution of the motor. Since the total volume of the channel here is ∼ 50 µl, the flow rates are expected to be 0.5, 1, and 2 ml/min for the motor speeds 10, 20 and 40 rpm, respectively. In comparison, these estimates far exceed the experimentally observed values here which are 27, 57, and 108 µl/min at minimal backpressure (0 kPa). This indicates that the pump works only at ∼ 5% of its maximum capacity. However, the efficiency can be improved greatly by reducing the ball-magnet separation because a slight reduction in the separation tremendously increases the magnetic force. In our experiments, we have maintained a nominal distance of 9 mm between the magnets’ surface and the centre of the ball bearings (Ø 6 mm). As the clearance between the magnets and the acrylic disk is already less than 1 mm, a more practical way of reducing the ball-magnet distance is to reduce the thickness of the acrylic disk and the PDMS.

We have put the new integrated pump and cartridge into test by demonstrating sequential delivery of sample/reagent plugs to a silicon-based microfluidic chip for solid-phase extraction and purification of nucleic acids. Solid-phase extraction of nucleic acids is based on the principle that nucleic acids readily bind to silica surfaces in the presence of chaotropic salts (Vogelstein and Gillespie 1979; Boom et al. 1990). After washing away impurities in ethanol, nucleic acids can be simply recovered from the surface by eluting them in a low-salt buffer. Conventional procedures using this principle, however, are technically-challenging and time consuming with many steps of pipetting and spinning required. Microfluidic chips offer automation through miniaturization by eliminating the need for pipetting. Previously, our group and others demonstrated silica-based chip surfaces for solid-phase extraction of nucleic acids (both genomic DNA and viral RNA) (Christel et al. 1999), (Hui et al. 2007; Yobas et al. 2005, 2006, 2007), (Cady et al. 2003; Wolfe et al. 2002; Breadmore et al. 2003). However, for such chips to find practical use in the field, they should be supplied in the form of a self-contained fully-enclosed cartridge with an integrated pump.

Prior to using the cartridge for nucleic acid purification, it is important to check whether the integrated pump can perform in the presence of air spacers between the reagent plugs. We tested this capability with cartridges loaded with six water plugs isolated by air spacers as schematically described in Fig. 5(a). Specifically, plug 1, representing the input sample, was loaded in the non-planar tubing segment upstream of the LOC while plugs 2 – 5 as the subsequent reagents were loaded between plug 1 and the pump. Plug 6 contained the working liquid which filled up the remaining volume of the cartridge including the pump and LOC. Plugs 1 and 2 were 100 mm long (∼20 µl) whereas plugs 3, 4, and 5 were about 50 mm long (∼10 µl). To prevent succeeding plugs from mixing with each other as they passed through the LOC, we separated them by air spacers about 10 mm long. To minimize any undesired plug dispersion, we maintained relatively low flow rates with values measured and stated accordingly. By activating the pump, we displaced the plugs and completed a single run cycle. We identified the “single run cycle” as the point where plugs 3 – 5 reached the non-planar tubing segment located at the downstream of the LOC while plugs 1 and 2 were positioned at the upstream of the pump. We ran the cartridges at 20 rpm in a fully-enclosed mode with the non-planar tubing segments mounted on them. The pump managed to displace all the water plugs through the LOC. In Fig. 5(b), photomicrographs taken at subsequent time intervals show an air spacer traveling through the serpentine microchannel (500 µm wide and 80 µm deep) located in the LOC. However, with the presence of air spacers, we noticed an overall decline in the flow rate (average ∼ 35 µl/min) as compared to the case when there was no air spacer (57 µl/min at 0 kPa from Fig. 4). This was because the liquid plugs had to move along the dry surface of the tubing walls. Moving the contact lines where the hydrophobic tubing walls met air and liquid phases required additional energy. Energy was also spent on overcoming the surface tension effects encountered with the air spacers passing through the joints. All these factors contribute to the fluidic resistance and thus should be taken into account when designing for the assays requiring precise flow rates.
https://static-content.springer.com/image/art%3A10.1007%2Fs10544-009-9347-z/MediaObjects/10544_2009_9347_Fig5_HTML.gif
Fig. 5

a Schematic arrangement of six liquid plugs in the cartridge before (above) and after (below) a “single run cycle”. Plug 1 represents the sample while plugs 2 – 5 the subsequent reagents. Plug 6 is an inert working liquid (e.g. water) filling up the remaining of the cartridge including the pump and LOC. b Photomicrographs of the LOC taken at consecutive time intervals (from left to right) showing an air spacer travelling through the serpentine microchannel (from top to bottom) while preventing the two water plugs from merging (at 20 rpm and ∼ 35 µl/min). c Representative pictures of plugs 3, 4, and 5 (dyed water) with their individual lengths compared before (above) and after (below) a single run cycle at 20 rpm. d Picture of a gel separation of PCR products derived from the collected eluants (plugs 3 – 5 in their respective order) to confirm the presence of purified genomic DNA. Lane 1 – 3: Cartridge 1 (the MP3 at 25 rpm, ∼ 10 µl/min for ∼ 10 min); Lane 4 – 6: Cartridge 2 (the MP3 at 40 rpm,  ∼ 24 µl/min for  ∼ 4 min); Lane 7 – 9: Cartridge 3 (the MP3 replaced by a conventional pump, at 7 µl/min for 15 min); Lane 10 – 12: Cartridge 4 (the conventional pump, at 24 µl/min for 5 min). Positive control for PCR used an input amount of 17 ng genomic DNA

At the conclusion of a cartridge operation, we verified the integrity of the plugs. Figure 5(c) shows representative dyed water plugs (plugs 3, 4, and 5) before and after a single run cycle at 20 rpm. The plugs were initially 50 – 51 mm long and separated by ∼ 10 mm air spacers. After a single run cycle, we found that the overall integrity of the plugs was well maintained. However, we also found each plug short by less than 1 mm (<0.2 µl) implying that the plugs left behind residues as they wet the channel walls. We confirmed small residues behind the plugs when we closely inspected air spacers passing through the microchannel (Fig. 5(b)). Some of these residues were collected by the subsequent plugs while the others trapped in the chip’s dead volume. This raised a genuine concern over plug-to-plug cross contamination. To quantify the extent of the contamination, we repeated the above experiments after staining the leading water plug (plug 1) with fluorescein at 20 µM and measured the subsequent four water plugs for traces of fluorescein at the conclusion of a run cycle. We found the highest contamination in plug 2 typically less than 10% while the contamination in subsequent plugs 3 – 5 below 1%, 0.5%, and 0.25%, all relative to the fluorescein plug and based on the order of their sequence.

We used the cartridge with a sample containing cell lysate. The sample enabled us to test the use of a cartridge to purify nucleic acids from contaminants such as cellular proteins and chaotropic agents known to inhibit PCR reactions. We loaded the sample into the cartridge using non-planar tubing segment, replacing plug 1. We also replaced plug 2 with wash buffers in two equal fractions and plugs 3 – 5 with elution buffer in three equal fractions. We tested a pair of cartridges (1 and 2) in a fully assembled configuration and a second pair of cartridges (3 and 4) partially assembled with the integrated pump (MP3) replaced by a conventional pump. We ran cartridges 1 and 2 at 25 and 40 rpm and completed a single run cycle respectively in ∼ 10 and  ∼ 4 min with the average flow rates 10 and 24 µl/min. For cartridges 3 and 4, we set the conventional pump at flow rates 7 and 24 µl/min and completed a single run cycle in < 15 and  < 5 min.

Figure 5(d) shows PCR results after amplifying a 170 bp (β-actin) fragment of genomic DNA present in the individual fractions eluted from each cartridge. For all four cartridges, the amount of DNA eluted in the first fraction (plug 3) is typically the highest and comparable to positive control (17 ng DNA). This amount seems reasonable as it remains below the input amount of DNA  ∼ 100 ng, estimated based on the number of cells (∼16,000) in the 20 µl sample. The second fraction (plug 4) of each cartridge also contains DNA, yet in significantly reduced amounts. The amount of DNA eluted in the third fraction (plug 5) is negligibly small for cartridges 1 and 2 which utilized the MP3 to displace the sequence of plugs. Interestingly, the band intensities produced by both cartridges 1 and 2 are similar despite their difference in flow rates. Moreover, these intensity patterns are indistinguishable from those of cartridge 4 which replaced the MP3 with a conventional pump and displaced the plugs at a rate of 24 µl/min. Cartridge 3, by reducing the pumping rate to 7 µl/min, managed to enhance the DNA yield noticeably both in the second and third fractions (plugs 4 and 5). In contrast, such yield enhancement cannot be observed in the eluants from cartridge 1 despite having subjected to comparable plug transport rate (10 µl/min). This discrepancy might be explained by the pulsating nature of the flow generated by the MP3. The pulsations are an inherent feature of peristaltic pumping and typically alleviated by incorporating a diffuser or a dampener. PCR inhibiting factors can be ruled out since the carry-over contaminants in plugs 4 and 5 are considerably low as indicated by the fluorescence test. About a dozen cartridges subjected to the same test protocol returned results more or less representative of those presented here.

Maximizing the nucleic acid yield would require optimization of the purification protocol by modifying the flow rate and/or the volume of the sample and reagent plugs. However, this may imply an increase in the protocol time. For instance, the flow rate for the sample and water plugs can be reduced to allow nucleic acids more time to reach, bind, and release from the surface. Concurrently, the motor speed can be paused when the plugs pass through the chip to introduce incubation steps. Moreover, the motor speed can be programmed according to each plug since the volume of each plug is known prior to the extraction. Here, for the sake of simplicity, we applied a fixed average flow rate for all the plugs and did not vary the volume of individual plugs. Longer plugs are surely undesired as they increase the total protocol time and overall cartridge footprint.

Use of commercial tubing for the cartridge channels offers several advantages. The tubing provides a round cross section and smooth surface which reduces dispersions behind the displaced plugs and subsequent carry-over contamination. In addition, the tubing can be selected from a variety of commercially available materials such that it is inert and adsorbs less sample and reagents. Moreover, the tubing is capable of storing plugs for a relatively long period of time until their use without the evaporation becoming a major concern. However, the approach is not without its limitation as it requires manual assembly. In order to avoid this problem, the cartridge channels can be fabricated through rapid prototyping methods such as laser-ablation, soft-lithography, and hot-embossing. Nevertheless, these methods return channel profiles likely to increase plug dispersion and carry-over contamination. The methods also limit the set of materials that can be used for constructing the channels. For instance, PDMS, the popular soft-lithography material, is suitable for the pump but not ideal for the remaining cartridge channels, mainly because its gas-permeable structure cannot prevent the evaporation of the plugs during their storage. PDMS may lead to merging of adjacent plugs as the air spacer between them may diffuse through the channel wall under the pressure exerted by the pump. Further, its elastic nature makes the channel walls expand under pressure thereby limiting the pump effectiveness. For those reasons, we opted not to use soft-lithography even if it could have simplified the cartridge prototype by monolithically defining all the channels within the same PDMS layer (as schematically illustrated in Fig. 1). Evaporation of the cartridge content can however still occur through the pump since it is made of PDMS and has to be addressed. Perhaps freezing the cartridges as suggested by Weibel et al. may help to solve this problem and improve their shelf life (Weibel et al. 2007).

The idea of using multiple tubing segments also facilitates modular cartridge design in which a particular tubing segment of the loop can be replaced in order to introduce the input sample or collect the assay product. Instead of having all the tubing segments connected in series, some of the tubing segments can be joined in parallel, much like the components of an electrical circuit. Parallel branches can be useful for storing certain sample and/or reagents that need to be segregated until their use but can automatically merge and mix together once the cartridge is activated. Such a feature may minimize off-the-cartridge sample preparation steps. For instance, parallel branches may involve a plug of lysis buffer and a dummy branch to be replaced with the sample. Without this configuration, the sample and lysis buffer have to be mixed together off the cartridge as performed here.

Last but not least, preserving the self-contained feature of the cartridge while handling the input sample or the assay output would require additional means to automatically make or break a seal when connecting or disconnecting the tubing segments.

5 Conclusion

A self-contained fully-enclosed cartridge is demonstrated by constructing a simple fluidic loop in which several functional components are integrated including (1) a commercial tubing segment containing a sequence of reagent plugs isolated by air spacers, (2) a microfluidic device specific to the target application, and (3) a novel peristaltic pump. The peristaltic pump facilitates fluid propulsion in a closed system by sequentially displacing the reagent plugs around the loop. The pump is manufactured in PDMS via soft-lithography and its peristaltic action is induced by compressing a planar channel under stainless-steel ball bearings that are loaded magnetically. In the loop, removable tubing segments simplify the procedure for sample loading and the eluant collection once either liquid becomes available. These segments also offer the flexibility of handling varying volumes by accordingly sizing the tubing length. The prototype of the cartridge built here involves a silicon-based microfluidic chip as the nucleic acid capturing surface and can successfully purify DNA from a cell lysate on a battery-operated portable system. Current focus in our laboratory is on the incorporation of nucleic acid amplification and detection modules in the cartridge for sample-in-answer-out capability.

Acknowledgement

We would like to acknowledge contributions of the cleanroom staff at the Institute of Microelectronics, Singapore and our co-workers K. Wang for maintaining cell culture, J. Reboud for reviewing our manuscript, and P. Neuzil for his help with art work.

Copyright information

© Springer Science+Business Media, LLC 2009