by Jennifer Campbell, Paul Mirsky, Anirban Chatterjee, Andre Sharon and Alexis F. Sauer-Budge
Nucleic acid sample preparation is lengthy, making it a leading candidate for laboratory automation. A parallel processing device that performs three liquid-handling functions – dispensing, pipetting and pressurising – has been developed. The device comprises an array of nozzles connected by fluidic channels that precisely distribute air and fluids to an array.
The manual preparation of nucleic acid samples is an ideal candidate for laboratory automation. It is a routine, yet time-consuming task that often creates a bottleneck that slows down drug discovery and diagnostics research processes. The fact that sample preparation protocols are fairly standardised and widely-used underscores the broad utility of such automation. Most microscale nucleic acid purification columns require the same fundamental liquid-handling functions: pipetting, dispensing and pressurising. That is, samples need to be transferred between certain wells; reagents and buffers need to be distributed from a common source into a number of wells, and solutions need to be pressurised through the columns.
Usually automated pipetting is accomplished with multichannel pipettor heads that use air-filled plungers. To improve the accuracy of smaller volumes, syringe-based pipettors utilise a liquid-filled syringe instead. These methods work, but are mechanically complex and require moving parts. With the advent of ultra-high throughput screening, these devices are reaching their practical density limit. Other devices, including the SciClone from Caliper Life Sciences and the Quadra Nano from TomTec, apply a vacuum across a calibrated flow resistance to draw liquid into a tip. Multi-head dispensers typically require stepping with an x/y motion system to fill a microtitre plate from a common source. Automated dispensers can incorporate syringe pumps to deliver precise volumes of liquid . Alternatively, pressurised liquid can be dispensed for a specified amount of time. Many recent innovations in pipetting and dispensing have improved both accuracy and throughput; however, much of this advancement is focused on handling very small (<1 µL) volumes, which tends to have limited application.
Forcing liquid through a column with air pressure is a common method used to increase throughput. Pressure domes or vacuum manifolds are often used for this purpose. Alternatively, centrifugation can be used to force a liquid through a column, but moving columns in and out of centrifuges is often tedious and inefficient.
TriPette integrates pipetting, dispensing, and pressurising tasks into one compact device [Figure 1], . The dispensing function supplies an array of small nozzles with pressurised liquid in order to produce a controlled flow, while the pressurising and pipetting functions are implemented by controlling the flow of air through the nozzle array. To date, the performance of the device has been validated through DNA and RNA extraction using columns of both Q-sepharose resin and porous polymer monolith (PPM) .
TriPette has been integrated into an instrument for nucleic acid isolation [Figure 1]. The instrument was built to accommodate a standard microtitre plate that sits on a platform. The manifold – a polycarbonate block with a grid pattern of small nozzles – is positioned above the platform. The manifold has four channels and 24 nozzles; however, it is possible to scale up the design to 16 channels and 384 nozzles. Located at the heads of the channels, narrow sections act as distribution restrictors to distribute the liquid evenly across all four channels. Tubing connects the manifold to the controller, a fluidic circuit of valves, sensors, supply bottles, regulators and other components. A host computer provides a user interface. Removable frames holding arrays of pipette tips or microscale columns (shown) can be attached to the manifold. All non-liquid handling steps (plate manipulation, column array attachment, etc.) are performed manually, making this system a low-cost, semi-automated option for life scientists.
Before dispensing, the manifold must first be primed with buffer. To do so, vacuum draws liquid from a source jar into the buffer tank. Next, air pressure forces buffer into the plenum and the drain. Valve switching then pushes buffer into the manifold channels and outlet tubing. Pressure applied to the channels forces buffer to jet out through the nozzles at a precisely-calibrated rate. The dispensed volume is governed by valve timing.
Following dispensing, the manifold is drained of buffer. The buffer tank drains very efficiently, leaving little or no residue, as it is cone-shaped and made of a fluorinated polymer. Buffer is then drained from the plenum and drain tubing. Finally, valve switching drains the buffer from the channels and sends it to waste. The controller performs all valve switching automatically.
The manifold interfaces with an array of pipette tips when operating in pipetting mode. For future versions of the design, we envision an array of pipette tips moulded as a single consumable part with an integral sealing gasket. However, in the present version, commercially-available pipette tips are attached to a grid of seats in a polycarbonate plate. On the opposite face of the plate, a grid of O-rings rests against the manifold and seals each tip around a nozzle.
During pipetting, the buffer tank and manifold remain empty. To pipette in, the tips are dipped into the wells of a microtitre plate. Vacuum then draws air through the nozzles, pulling liquid into the pipette tips. To expel the liquid, air flows out through the nozzles and into the pipette tips.
The manifold can also provide air pressure to a framed array of microscale columns topped by reservoirs of liquid. The column housings are made of extruded plastic and are held in place by a grid of modified one-touch pneumatic fittings mounted in a polycarbonate frame. Short pieces of tubing, fitting snugly to the upper tip of the columns, form the reservoirs. O-rings lie in glands around the reservoirs and seal the frame to the face of the manifold. In the future, we envision the column housings could be moulded as a single consumable part with integral reservoirs and an integral gasket for sealing.
Metrics important for assessing fluidic performance include dispensing accuracy, uniformity, and minimum dispense volume.
Uniformity and accuracy
In discussing uniformity, it is necessary to distinguish between shot-to-shot variation and nozzle-to-nozzle variation. Data sheets for commercial instruments quote a single CV; typical values range from 1.5–2.0% for dispensers, and 2.0–3.0% for pipettors. TriPette has variations within these ranges.
For dispensing, the shot-to-shot variation was on average 0.32%; for pipetting, it was 0.55%. For dispensing, the nozzle-to-nozzle variation had a mean of 0.76%; for pipetting, it was 1.2%. For dispensing, overall RMS error was 0.8% and the maximum error was 1.9%. This is comparable to the quoted accuracies for typical commercial instruments of 1.0–3.0%. For pipetting, overall RMS error was 0.36% and the maximum error was 4.0%. This is also comparable to the quoted accuracies for typical commercial instruments of 2.5–5.0%.
Minimum dispense volume
Another important figure of merit for a dispenser is the minimum dispense volume. Figure 2 displays TriPette’s actual dispensed volume vs. its commanded volume. These data show that actual dispensed volumes below 5 µL are less than ideal. The device is comparable to other commercially-available, automated dispensers, many of which have minimum volumes in the range of 5–10 µL.
Pipetting 200 µL takes approximately 25 seconds, while dispensing 200 µL of water takes approximately 192 seconds. At present, TriPette takes several times longer than other automated dispensers, even though these other devices have fewer channels and must step-and-repeat. Yet these times are not fundamental or even practical limits. Opportunities for TriPette improvement include shortening tubes, reducing flow resistances, and using higher-resistance nozzles to achieve faster filling and emptying of the manifold. Pipetting and dispensing times could be cut significantly with these enhancements, perhaps by 10-fold, making our device both more inexpensive and faster than other commercially available automated liquid handlers.
Application in nucleic acid purification
The industry standard for nucleic acid isolation uses manual processes with solid phase extraction columns from manufacturers like Qiagen, Invitrogen, Roche, Promega, etc. Although the manual processes are simple and provide high extraction efficiencies, they suffer from variability in extraction efficiency and nucleic acid quality. The largest contributor to this variability is human error in liquid handling, leading to different reagent volumes and varied timing of reagent delivery. TriPette’s fluidic performance has been demonstrated with a bacterial DNA isolation protocol using an anion exchange matrix (Q-Sepharose) and a porous polymer monolith (PPM) solid phase extraction matrix. The same extraction protocols were used for both automated (TriPette) and manual (HT-SNAP)  processes. The automated extraction method was found to be both more efficient and less variable than the manual process for both the PPM and the Q-sepharose matrices [Table 1]. Because both procedures used the same reagents and materials, TriPette’s superior performance can be attributed to more accurate liquid handling.
Conclusions and future perspectives
A novel design principle for a liquid handling system has been implemented and tested in biomolecular sample preparation. By most metrics, TriPette’s performance equals or surpasses that of typical commercial instruments. One key benefit is that it reduces lab set-up costs by combining three common liquid-handling devices into a single instrument. It is also less mechanically complex than other devices, allowing for denser arrays of nozzles. One potential disadvantage is that the nozzles are not controlled independently, meaning that the same volumes and/or reagents are pipetted and/or dispensed to all samples. However, because liquid is added manually to the reservoirs of the microcolumns, different reagents can be tested in parallel in pressurising mode.
TriPette’s accuracy and volume range are both comparable to other automated liquid handlers. While the operation time is currently long, reductions in the duration can easily be achieved with some simple design enhancements. Compared to a manual process, it has been shown that extraction efficiency can be increased and variability can be decreased using our automated process. Because the nucleic acid isolation application is one where simplicity, reliability and low cost are important, TriPette is an attractive option. We envision that it will be adapted to other liquid handling applications in the future.
1. Kong FW, Zheng YF and. Chen WD. Automatic Liquid Handling for Life Sciences - A Critical Review of the Current State-of-the-Art 2008. IEEE International Conference on Robotics and Biomimetics 2009; Vols 1-4: 480-486.
2. Mirsky P et al. An automated, parallel processing approach to biomolecular sample preparation. J Lab Autom 2012; 17(2): 116-24.
3. Chatterjee A et al. RNA Isolation from Mammalian Cells Using Porous Polymer Monoliths: An Approach for High-Throughput Automation. Anal Chem 2010; 82(11): 4344-4356.
Jennifer Campbell, Paul Mirsky,
Anirban Chatterjee, Andre Sharon and Alexis F. Sauer-Budge*
Fraunhofer Center for Manufacturing Innovation
15 Saint Mary’s Street
Brookline, MA 02446, USA