Geoffrey Grove, Susan Ryan and Ed Alderman
Applied Science and Technology Department, Caliper Life Sciences, Inc., Hopkinton, MA 01748
The High Volume Head is one of four multi-channel liquid handling options available for the Sciclone Advanced Liquid Handler (ALH). The architecture of the device includes an array of 96 pistons, which are moved simultaneously by a single drive mechanism. This device is designed for use with up to 96 disposable tips for the transfer of up to 200 μL per channel. In this note the plate-to-plate precision of the device across its entire working range is evaluated, including results for a variety of typical liquids and solutions. Plate-to-plate precision results demonstrate that CV's of <5% for volumes in the 1-5 μL range are easily achieved; however, CV's of <2-3% are possible in wet-transfers, when backlash is eliminated. We conclude that the characteristics of the selected solutions inherently carry more variability than the instrument itself. Therefore, experimental design for low-volume techniques requiring high precision may be improved by choosing solutions with better pipetting properties.
The effort to miniaturize and automate assays for drug discovery consistently drives researchers to lower volume liquid handling. In evaluating the feasibility of a given experiment, the specifications for liquid-handling systems serve as a guideline; however, these specifications are conditional statements based on a variety of variables. One of the principle variables is the sample itself. Differences in surface tension, evaporation rate, and wetting properties have a direct effect on the precision and accuracy of sample transfers.
The term "liquid class" is typically applied in the liquid handling industry to refer not only to the liquid or solution being handled, but also to the instrument settings that are required as a result of the physical properties of the sample. There is no current industry standard to encompass the parameters included in the discussion of this term; however, the volume and speed of aspiration and dispensing are two of the principle components of a liquid class. Additional components often included are calibration curves and air gaps.
Air gaps, in particular pre-sample air gaps, are one of the ways to address backlash. Backlash is the amount of distance, usually expressed in units of volume, which it takes a gear to change direction before a force is applied in the opposite direction. The result of this is that if a gear driven piston is used to aspirate a sample, when it comes time to dispense that sample, if the same number of gear turns (or motor steps) are used in the opposite direction, a small amount will be left in the tip. Similarly, if a dispense action is taken, and then an aspiration is done, at the beginning of the aspiration, no liquid will be drawn. A pre-sample air gap can be used to compensate for the backlash by setting the volume of the air gap to be larger than the backlash of the device. It is also possible to eliminate backlash entirely by aspirating more sample than is required, dispensing an aliquot of this sample back into the source, which is larger than the backlash, and then proceeding with dispensing the sample to the destination. This work employs the latter method and maintains the same aspirating and dispensing speeds for all of the samples tested.
Another variable to consider with automated systems is the travel speed. Fast vertical and horizontal movements may have an effect on precision and accuracy of sample delivery. It is particularly important to retract slowly from the sample when the tips are below the surface, to avoid leaving droplets on the outside of the pipetting device. Typically a setting of 10% of the maximum travel speed (Fig. 1) wa s used. For all movements other than these retractions, the Sciclone ALH was run at 100% travel speed.
Finally, the issue of evaporation must be addressed. Although this is an accuracy, rather than a precision issue, variations in accounting for evaporation rates interfere with the precision results at the 1-2 μL data points. Therefore, all data are corrected for evaporation. Evaporation varies depending on environmental conditions; however, the magnitude of evaporation is reported to give the researcher a general guideline as to what to expect, and perhaps when it may be necessary to take evaporation into account.
Materials and Methods
All experiments were run using a Sciclone ALH 3000 workstation with software version 3.1.14, firmware version 1.14 and a High Volume Head (S/N SS0309N4928).
Liquids evaluated in this study include:
Purified Water [Culligan System]
BupH Modified Dulbecco's Phosphate Buffered Saline Packs [Pierce No. 28374]
Ethanol [J.T. Baker A-478-09]
Dimethyl Sulfoxide (DMSO) [J.T. Baker 9224-01]
1 mg/mL Bovine Serum Albumin (BSA) [Sigma A-7906] solution in purified water
10% Glycerol [ICN Biomedical, Inc. No. 806688] in purified water
1% Triton TX-100 [Sigma A-7906] in purified water.
CaliperLS Automation Certified TempoTM-100 and TempoTM-200 tips were used for all experiments. Fresh tips were used for each sample. Tempo-200 tips were only used for the 180 μL dispenses while the Tempo-100 tips were used for all other data points.
Plates: A variety of untreated 96-well, flat-bottomed, polystyrene plates were used. No evidence for differences in sample delivery were o bserved due to differences in plate brand.
A HyperTask Grooved Reservoir (P/N RES03384) was used for all sample types.
Density and evaporation rates were determined gravimetrically using a Sartorius BP211 D analytical balance. Evaporation rates were measured daily, by dispensing 100 μL of the sample into a plate, and recording the weight change in time. Day to day variation in the evaporation rates were found to be as much as 1.5% for a 1 μL water sample.
The liquid class used for all of this work ran at 20 μL/sec rate, with no air-gap and a linear curve assigned for the full range. The backlash for this HVH device (<5 μL) was determined visually, by aspirating 1 μL increments until liquid entered the tips.
All machine movements were at 100% travel speed except during the first 2 dispense steps, where a 5% retraction travel speed was used. At each sampling point (1, 2, 5, 10, 100, and 180 μL) excess sample (>5 μL) was aspirated from the reservoir. Some sample was then dispensed back to the reservoir (>5 μL) to eliminate the backlash.
To standardize the evaporation rate for each data point, data acquisition times were measured with a VWR laboratory timer. For the 180 μL sampling point, a dry plate was weighed and placed on the Sciclone deck. The timer was started when the sample began dispensing into a dry plate, and the second weight was taken 90 seconds later. For all other sampling points, the plate started wet, with at least 60 μL of sample in it. The timer was started when the first weight was taken, the plate was placed on the Sciclone deck, the protocol was run, and the plate was removed and placed on the balance. The second weight was recorded 90 seconds after the first.< /p>
For the gravimetrically determined dispensing of the Sciclone HVH, weights dispensed were corrected for evaporation. The magnitude of evaporation over the 90 second experimental time frame can be seen for all sample types in Figure 2: A and B.
Since this is a wet transfer and there is sample in the plate at the start of the experiment, an evaporation rate of >100% at 1 μL is possible. During the course of our experiments, with water taken as the "standard" sample type, we see that most samples, with the exception of Ethanol and DMSO, did not have evaporation rates that varied by more than 2-3% from water (Fig. 3).
In addition to the %CV data generated, it was also noted that the behavior of BSA solution was not ideal under the experimental conditions used here. At 20 μL/sec as an aspiration rate, small bubbles formed. Although this did not affect the precision greatly, the surface tension properties of bubbles are known to damage proteins1.
At pipetting volumes >5 μL, plate-to-plate %CV's attained for the Sciclone 3000 ALH were <1%. In the 1-2 μL range, a CV of <2.5% is possible with the proper pipetting technique. The addition of a surfactant (Triton X-100) dramatically improved the precision, possibly due to better wetting properties2 . This is particularly clear at 1 μL, where the CV is close to 1%. Since different liquids accountant for more than a 4% change in CV (at 1 μL), this demonstrates that variability in sample properties introduces more error into pipetting protocols than is inherent in the precision of the Sciclone ALH High Volume Head device (<3% CV). These results indicate that slight modifications to sample solutions can yield better pipetting results. Although the method for compensating for or eliminating backlash will not have an e ffect on the precision performance of a given system, it is likely that the calibration curves used in the liquid class will differ depending on which method is used. It has been found in this and other work that exiting the sample slowly after a wet transfer pro - duces a cleaner breakoff at the tip or cannula, and can have a significant impact on precision performance.
Since the physical properties of the sample carries more variability than the HVH device at low volumes, the choice of liquid or solution for critical low volume work will have a significant impact. To obtain optimal results when working with samples that are viscous, or have poor wetting properties, the geometry and materials of the tip should also be considered3.
The authors would like to thank Jim O'Keefe for his initial efforts and advice in beginning this work.
1 Niven, R.W. et al. 1996. Protein Nebulization. 2. Int. J. Pharm. 127 iss. 2 pp. 191-201.
2 Massignon, D. 1987. Science des Interfaces, Mouillabilit, Adhsion et Frottement. Synthse de la Runion et Prospective. Vers une Approche des Problmes Industriels. J. Chim. Phys. PCB. 84 no.2 pp.135-140.
3 Barton, A. 1982. The Application of Cohesion Parameters to Wetting and Adhesion- A Review. J. Adhesion. 14 pp.33-62.