High-throughput mouse genotyping using robotics automation
MATERIALS AND METHODS
Kaari L. Linask and Cecilia W. Lo
Two-millimeter mouse-tail biopsies are collected into a 96-well plate with 100 µL lysis buffer [50 mM KCl, 1.5 mM MgCl2, 10 mM Tris, pH 8.0, 0.001% gelatin, 0.045% Nonidet™ P40 (NP40), 0.045% Tween® 20, and 500 µg/mL proteinase K; modified from Reference 2]. The tails in the lysis buffer are incubated for 4 h at 55°C using an Eppendorf® Thermomixer R (Brinkmann Instruments, Westbury, NY USA) with agitation at 1000 rpm, followed by incubation at 95°C for 10 min to inactivate the proteinase K. The lysates are then cleared by a 5 min centrifugation using an Eppendorf Multipurpose Centrifuge (Model 5804) set at 2250× g and used directly for PCR amplification.
National Institutes of Health, Bethesda, MD, USA BioTechniques 38:219-223 (February 2005)
The use of mouse models is rapidly expanding in biomedical research. This has dictated the need for the rapid genotyping of mutant mouse colonies for more efficient utilization of animal holding space. We have established a high-throughput protocol for mouse genotyping using two robotics workstations: a liquid-handling robot to assemble PCR and a microfluidics electrophoresis robot for PCR product analysis. This dual-robotics setup incurs lower start-up costs than a fully automated system while still minimizing human intervention. Essential to this automation scheme is the construction of a database containing customized scripts for programming the robotics workstations. Using these scripts and the robotics systems, multiple combinations of genotyping reactions can be assembled simultaneously, allowing even complex genotyping data to be generated rapidly with consistency and accuracy. A detailed protocol, database, scripts, and additional background information are available at http://dir.nhlbi.nih.gov/labs/ldb-chd/autogene/.
Primer Design
INTRODUCTION Transgenic and mutant mouse models have become increasingly important in biomedical research, but often a constraining factor in the full utilization of new mouse models has been the limited availability and prohibitive cost of animal holding space. Given this, the rapid genotyping of transgenic and mutant mouse colonies has become imperative for efficient utilization of animal holding space. Mouse genotyping is generally straightforward but time-consuming when there are large numbers of samples with a myriad of different genotypes to be processed and analyzed. Typically, biopsied tail samples are subject to overnight lysis, followed by purification of the DNA to remove salts and detergents. The purified DNA is then PCR-amplified using the appropriate primers and analyzed by gel electrophoresis (1). Although genotyping a single mouse litter comprising 6–12 samples can be completed in 2 days, the turnaround time for 10 or more litters is often substantially greater. In a large mouse colony, the complexity of the analysis is often compounded by the need for PCR analysis using different primer pair combinations, depending on the genotyping analysis required. Vol. 38, No. 2 (2005)
Tissue Lysis
The tedium and complexity involved in working with such large sample sizes provide ample opportunity for human error. In short, mouse genotyping can quickly become a bottleneck in the efficient management of a breeding mouse colony. To address this critical need for accurate and high-throughput mouse genotyping, we sought to automate mouse-tail genotyping using commercially available robotics systems. To reign in the instrumentation cost and bench space required to accommodate such a robotics system, we compromised in developing a system that is just short of full automation. A low-end liquid-handling robot is used for the assembly of the PCR, together with a separate microfluidics robotic system for PCR product analysis. Human intervention is required for moving the 96-well sample plates from the liquid-handling robot to the thermal cycler and from the thermal cycler to the microfluidics robot. This roboticsassisted protocol yields accurate genotyping data and represents a rational compromise between cost, level of automation, and the throughput suitable for the average transgenic/ mutant mouse facility.
The PCR primers either worked without modification or, in some cases, were redesigned to allow all reactions to be carried out in the same plate using a 58°C annealing temperature (3). We used MacVector™ 7.2.2 (Accelrys, San Diego, CA, USA) or LightCycler® Probe Design software 2.0 (Roche Applied Science, Indianapolis, IN, USA) for determining the optimal annealing temperatures and for primer design. Software Scripts All scripts required for data handling and robotics control are available at http://dir.nhlbi.nih.gov/labs/ldb-chd/ autogene/. Information for each sample is entered into a FileMaker Pro® 6.0v4 (FileMaker, Santa Clara, CA, USA) database that tracks the following: plate identification (ID), column ID, row ID, sample ID, and the genotyping required (AutomationScripts.zip on the web site). From this information, the database’s scripts automatically enter the PCR required for each genotype and calculate the well ID, based on the column and row ID information. Next, another script is run that transforms and sorts the sample list to include values required for assembly of the PCR. These data are then exported as .csv BioTechniques 219
SHORT TECHNICAL REPORTS Table 1. Automation Outline Including Time Estimates Procedure Tissue Lysis Add lysis buffer and seal plate Incubate at 55°C with shaking Transfer to thermal cycler Proteinase K inactivation Database Import sample records Run scripts and print recipes PCR Prepare master mixtures MultiPROBE setup MultiPROBE run Cap and transfer to thermal cycler Thermal cycling
Method
Min
M A M A
20 240 1 10
M/A M/A
2 2
M M A M A
30 20 50 5 190
thermal cycling program consists of a 5-min preincubation at 95°C, followed by 40 cycles at 95°C for 45 s, 58°C for 30 s, 0.4°C/s ramp to 72°C, 72°C for 30 s, and one 10-min extension step at 72°C, after which the plate is held at 4°C until electrophoresis. Electrophoresis
The PCR products are analyzed on the AMS 90 SE LabChip electrophoresis system with Electrophoresis the HT DNA 5000 SE30 LabChip preparation and setup M 5 LabChip kit according to LabChip run A 45 protocols provided by the Time values are based on extrapolated averages per 96-well plate measured from various counts of samples. M, manual incompany. Briefly, the chip tervention; A, automated program. is manually primed with a ® gel-dye matrix mixture, and files for use in the MultiPROBE II HT marker is added. The chip and 96-well robotics workstation (Perkin Elmer, plate containing the PCR products and Torrance, CA, USA) and the AMS 90 a molecular weight standard are loaded SE LabChip® electrophoresis system into the instrument. The robotics (Caliper Life Sciences, Hopkinton, system then samples each reaction. MA, USA). The database also calcuSample information is imported from lates and prints recipes that include the the .csv file that had been generated total volumes required for making the earlier. The genotypes are interpreted PCR master mixtures. from the PCR products displayed by the electropherograms or virtual gels Polymerase Chain Reaction generated by the LabChip HT software (Caliper Life Sciences). The PCR master mixture is prepared according to the general protocols for AmpliTaq Gold® polymerase (Applied RESULTS AND DISCUSSION Biosystems, Foster City, CA, USA), a hot-start polymerase that eliminates An outline of the genotyping the priming of nonspecific products strategy, including manual and before activation (4). The 1× PCR automated steps, is shown in Table 1. buffer contains 15 mM Tris-HCl, pH In preparation for robotics automation, 8.0, 50 mM KCl, 0.8 mM dNTP, 2–2.5 tail snips are harvested in 96-well plates mM MgCl2, 0.5 µM each primer, and 1 using a modified lysis buffer containing U polymerase in a 40-µL volume. The a reduced concentration of nonionic reactions are assembled by the Multidetergents. Using this procedure, the PROBE II HT robotics workstation tail lysate can be used directly, without according to a WinPrep ® script further purification, for PCR ampli(MousePCR.MPT on the web site) by fication followed by PCR product combining the DNA lysates and the analysis using the microfluidics chip. required master mixture in a 96-well The reduction in detergent concenplate according to values specified in a tration in the lysis buffer is essential comma-delimited (.csv) file generated for compatibility with the LabChip by the FileMaker software scripts. electrophoresis system. Assembly of The 96-well plate containing the the PCR is carried out in a 96-well plate PCR assembled by the MultiPROBE using the MultiPROBE II HT robotics II HT is then incubated with a heated ® liquid-handling system. This robotics lid in a DYAD thermal cycler (MJ system has a relatively small footprint Research, Waltham, MA, USA). The 220 BioTechniques
and eight probes that are individually programmable, providing the flexibility needed for assembling different combinations of PCR amplifications in a single 96-well plate. Once the PCRs are assembled, the 96well plate is manually retrieved from the MultiPROBE and placed onto a thermal cycler for amplification. We use an annealing temperature of 58°C because this has been found to be suitable for amplification with a wide range of primer sets, making it possible for different combinations of primers to be used on the same plate. Short PCR products tend to work more reliably, with a product range of 150–900 bp reproducibly generated. In general, our experience shows that primer melting temperature (Tm) is not as critical as PCR product length because primers with a wide Tm range of 43.3°–71.6°C have been used successfully. Primers with low Tm tend to yield lesser amounts of product, but sufficient products are usually generated with 40 PCR cycles. Primers with high Tm may yield some nonspecific products, but generally the main product bands still predominate. After completion of the thermal cycling, the 96-well plate is moved to the microfluidics electrophoresis system, the AMS 90 SE LabChip electrophoresis system. This instrument has a robotic arm that holds a microplate and a microfluidics LabChip with a capillary that samples each reaction individually for PCR product analysis. The product separation profile generated by the chip can be viewed in either a gel display or an electropherogram format. For convenience, we typically divide this three-part protocol for completion over 1.5 working days, with the lysis and assembly of the PCR on the first day, followed by thermal cycling for PCR amplification overnight. The next morning, the PCR products are then analyzed on the electrophoresis system. Here we briefly describe the scripts used for automation and some comparisons on throughput and cost using this robotics-assisted protocol versus manual processing of mouse genotyping protocols. We note that both robotics systems can work with a 384-well plate format with minor protocol and script modifications, which should provide considerable savings in reagent costs. The script for PCR assembly using Vol. 38, No. 2 (2005)
samples assembled by the MultiPROBE II HT (data not shown). Nevertheless, we routinely included negative Method Costs Consumables Labor Total Hands-On Hands-Off controls for each primer set and Automation $160,400.00 $1.06 $0.72 $1.78 1.4 8.9 also incorporated a tip-washing step Manual $19,200.00 $0.67 $4.13 $4.81 7.9 12.8 between each set of eight DNA samples Outsource $28,000.00 N.A. N.A. $27.00 N.A. N.A. (5). Positive controls are usually not Labor costs are based on a $50.00 hourly wage including benefits. Outsourcing start up is included, but these can be added if based on 40 PCR primer sets requiring setup and validation for genotyping. Automation there is concern with false-negative start up includes all equipment and software used in the Materials and Methods section. results. Figure 2 shows an example Manual start up includes the thermal cycler, two thermomixers, and a centrifuge. Time valof data generated from a row of 12 ues are based on extrapolated averages measured for various counts of samples. N.A., not samples in a 96-well plate. Amplificaapplicable. tions using three different PCR primer pairs (Figure 2, Cx43, DHFR, MHC43) the MultiPROBE II HT was designed to such as incorrect template and master were used to identify the genotypes maximize the number of 96-well plates mixture combinations or confusion of the transgenic mice. Lanes 2 and 6 that can be handled simultaneously of well numbers during pipeting. In are negative controls and, as expected, without manual interaction (Figure addition, optional scripts can be used show no products. Samples exhibiting 1). Sample location values are taken that allow for restriction digestion the expected product band correspond directly from a .csv file, obviating the following PCR to assay for single to transgenic mice, while samples need to map the location for each new nucleotide polymorphisms in the PCR showing no products are nontransgenic sample plate. Because the .csv file products. littermates. is generated by software-automated It should be noted that we have To examine the efficacy of the sorting of records, there is no chance tested for and found no evidence of automation, we calculated the approxof errors from manual manipulations, cross-contamination between PCR imate time required for the automation protocol, as compared to manual processing of the same numbers of samples (Table 2). As expected, there is substantial savings in hands-on time, from 1.4 h for the automation protocol versus 7.9 h for manual processing of a full 96-well plate. In addition, the automation protocol provides a timesavings of 4 h in the handsoff time required. We note that the amount of time required for this automation protocol does not scale-up directly with increased numbers of 96-well plates. The time required depends on numerous factors, including the number of different master mixtures required, the order of samples in the 96-well plates, the layout of the plates on the robotics deck, and the number of PCR product plates being processed in one run. The plate orientation on the deck of the MultiPROBE II HT can significantly impact the time required to assemble the PCR. For instance, if the PCR plate orientation is 90° clockwise relative to the DNA template source plate on the robotics deck, Figure 1. Protocol for automation using the MultiPROBE II HT robotics workstation. (A) Flowchart of the the MultiPROBE II HT robotic basic WinPrep robotics protocol used to assemble PCR. Bold letters refer to the comma-separated columns of the .csv file. (B) Layout of the .csv file and possible values used in the robotics protocol. (C) Deck layout of locations arm can pick up and dispense more template samples at the same time. on the MultiPROBE II HT liquid handling system. Table 2. Cost and Time Analysis of Various Genotyping Methods Cost Per Sample Hours Per 96 Samples Start-Up
Vol. 38, No. 2 (2005)
BioTechniques 221
SHORT TECHNICAL REPORTS Such an alternate layout requires half the time to run, but the rotated plates take up more deck space and therefore limit the maximum number of PCR plates that can be processed at one time without extra manual interaction. Processing multiple plates through the Caliper microfluidics system at one time is also advantageous. It saves on chip reagents since four 96well plates can be processed with a single priming of the chip. Overall, our experience shows that with the robotics automation, typically there is a 50% reduction in total time required for processing a typical 96-well plate, and perhaps more significantly, the total hands-on time is reduced by more then 80% (Table 2). Not factored into these calculations is the important advantage that the analysis is virtually error free because human manipulation is minimized. We carried out a comparison of the cost required for genotyping 96 tail samples either manually, using the robotics-assisted automation, or outsourcing to a commercial service
vendor (Table 2). This comparison shows that manual genotyping would cost almost three times as much as the robotics-assisted protocol, all arising from higher total labor cost. Not factored into this calculation is the difference in start-up costs for the robotics automation. We estimate this could be recovered within a year, assuming a genotyping load of 3000 mice per month, and even faster if multiple genotyping reactions are needed for mice carrying two or more different transgenes. In addition, further savings could come from reduced animal housing costs made possible from the more rapid turnaround of mouse genotype assignments. In summary, we have successfully automated mouse genotyping using two robotics systems. This protocol is high-throughput and provides consistency and accuracy by eliminating human sampling or pipeting errors. Our database scripts and protocols are flexible and can be adapted for use with other database software and liquid-
handling systems. They also can be easily modified for other genotyping applications. Using our robotics automation scheme, we estimate that approximately 2000 genotypes may be assayed in 24 h of manual processing time (corresponding to a total time of as little as 3–4 days, assuming thermal cyclers and incubator availability are not rate-limiting), thereby maximizing precious animal holding space. A detailed protocol, database, scripts, and additional background information are available at http://dir.nhlbi.nih.gov/ labs/ldb-chd/autogene/. ACKNOWLEDGMENTS
This work is supported by funding from grant no. NHLBI Z01-HL005701 to C.W.L. We would like to thank Ms. Heidi Dudik for assistance with running the robotics protocols. The mention of commercial products in this article does not imply recommendation or endorsement by the National Institutes of Health and is by way of example only. COMPETING INTERESTS STATEMENT
The authors declare no competing interests. REFERENCES
Figure 2. PCR fragment analysis on the AMS 90 SE electrophoresis system. Shown is a gel view of PCR fragments following electrophoresis on the AMS 90 SE using an HT DNA 5000 SE30 LabChip. These products were generated from a single row from a 96-well plate of PCR samples. Lane labels are imported values that include the mouse Tag identification (ID) and the specific PCR amplifications carried out. Negative controls are indicated by a minus sign (-). The background bands below 50 bp are nonspecific primer products. Ladder D contains molecular weight standards with fragment lengths indicated in base pairs. In addition, two markers are added to each sample at 15 and 7000 bp to allow for alignment of each sample with the ladder. 222 BioTechniques
1.Miller, S.A., D.D. Dykes, and H.F. Polesky. 1988. A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res. 16:1215. 2.Nagy, A., M. Gertsenstein, K. Vintersten, and R. Behringer. 2003. Detection and analysis of mouse genome alterations and specific sequences, p. 507-539. Manipulating the Mouse Embryo: A Laboratory Manual. CSH Laboratory Press, Cold Spring Harbor, NY. 3.Rychlik, W., W.J. Spencer, and R.E. Rhoads. 1990. Optimization of the annealing temperature for DNA amplification in vitro. Nucleic Acids Res. 18:6409-6412. 4.Grunenwald, H. 2003. Optimization of polymerase chain reactions, p. 89-99. In J. Bartlett and D. Stirling (Eds.), PCR Protocols. Humana Press, Totowa, NJ. Vol. 38, No. 2 (2005)
5.Kwok, S. and Higuchi. 1989. Avoiding false positives with PCR. Nature 339:237-238.
Received 27 May 2004; accepted 17 September 2004. Address correspondence to C.W. Lo, Laboratory of Developmental Biology, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD 208168019, USA. e-mail:
[email protected]
