Abstract
One of the first steps in drug discovery involves identification of novel compounds that interfere with therapeutically relevant biological processes. Identification of ‘lead’ compounds in all therapeutic areas included in a drug discovery program requires labor-intensive evaluation of numerous samples in a battery of therapy targeted biological assays. To accelerate the identification of ‘lead’ compounds, Janssen Research Foundation (JRF) has developed in the past an automated high throughput screening (HTS) based on the unattended operation of a custom Zymark tracked robot system. Automation of enzymatic and cellular assays was realized with this system adapted to the handling of microtiter plates. The microtiter plate technology is the basis of our screening. All compounds within our chemical library are stored and distributed in micronic tube racks or microtiter plates for screening. An efficient in-house developed mainframe based laboratory information management system supported all screening activities. Our experience at JRF has shown that the preparation of test compounds and serial dilutions has been a rate-limiting step in the overall screening process. In order to increase compound throughput, it was necessary both to optimize the robotized assays and to automate the compound supply processes. In HTS applications, one of the primary requirements is highly accurate and precise pipetting of microliter volumes of samples into microplates. The SciClone(tm) is an automated liquid handling workstation capable of both 96- and 384-channel high precision pipetting. For high throughput applications, the SciClone™ instrumentation is able to pipette a variety of liquid solutions with a high degree of accuracy and precision between microplates (inter-plate variability) and tip-to-tip (intra-plate variability) within a single plate. The focus of this presentation is to review the liquid handling performance of the SciClone™ system as a multipurpose instrument for pipetting aqueous or organic solutions, and virus suspensions into 96- and 384-well microplates. The capabilities of the system and the resulting benefits for our screening activities will be described.
INTRODUCTION
We have developed automated microplate-related assays for new drug discovery in the field of oncology 1 as well as virology. 2 We started using a Zymark robotic system in 1985 making us one of the first users of such applications at the time. New approaches and rather unexplored domains of automation were tried out. The use of robotics for performing cell-based assays is rather complicated. It differs in several ways from more standard assays such as ELISAs. For example, the need to maintain a sterile environment in cell-based assays requires the use of a laminar flow hood which has a limited area for robotics peripherals if the robot arm is fixed. In addition there is a need to give the robot access to the incubators. Therefore our first approach, introduced almost 15 years ago, was a custom configuration with the robot arm mounted on a track. It proved to be essential for optimal performance of the system. This idea is now used in most robot configurations. 1 Many custom peripherals were designed, like a 96-channel aspiration block mounted on a robot hand and a device to give the robot automatic access to the incubators. In 1991 we moved to a brand-new lab and this gave us the opportunity to redesign the whole system. The device was installed in a temperature- and humidity-controlled room functioning as one big incubator with a capacity of almost 600 microplates. In a continuous effort to further optimize our antiviral drug discovery and to meet the ever changing assay requirements, the system was redesigned again in 1998 with the integration of new hardware and software technologies. 3 A Reagent Addition Module, a plate washer, a fluorescence reader and a custom tipbox turntable were integrated. The introduction of the Zymark Zypettor™, an 8 or 12 channel pipettor, working on a new custom microplate turntable, was a crucial factor in achieving our throughput goals. Wide-oriented or narrow-oriented microplates can now be processed with high precision in a rapid and flexible way. The redesign of the system resulted in a flexible configuration able to run cell-based and enzymatic assays.
Over the past 15 years the different robotics systems which we have presented in detail at previous ISLARs enabled us to increase sample throughput significantly year after year.2,3.
Graph 1 illustrates the number of datapoints obtained over the past 14 years in the Antiviral Screening Department at JRF. It clearly illustrates the gradual but significant increase in datapoints that is a direct consequence of the implementation of automated screening systems. The asterisks indicate the years in which new systems were implemented. Before 1991, the number of manually generated results remained almost constant. Our first system increased the output between 1991 and 1996 almost six-fold from 260,000 to 1,400,000. The replacement by a new system in 1997, did not result in an increase in the number of datapoints. This could be ascribed to the lack of a proper laboratory information system. Last year an in-house developed mainframe based LIS was introduced and now sufficiently supports all screening activities and this boosted throughput significantly.
This graph proves beyond any doubt the success of the automated antiviral screening in our lab. It is important to note that this graph reflects an accomplished output, and not a projection of the theoretical maximum capacity of the system. The rate-limiting steps in realizing another throughput increase shifted to the lack of automated compound supply processes, together with the relative slowness of making serial dilutions.
ZYMARK'S STACCATO™ SYSTEM
In 2000, Zymarks Staccato™ system was introduced. Staccato™ application-focused workstations provide a uniform architecture using reliable, standard building blocks. At the heart of the workstation are the SciClone™ Liquid Handling and Presto™ Autostacking technologies. Controlled by Clara™ 2000 software this automation tool provides unmatched flexibility, throughput and capacity. This Staccato™ approach allows to set-up and run multiple applications, such as:
high speed plate replications or reformatting from: plate to plate mother to daughter
serial dilutions: between different plates Column-to-column or row-to-row within a single plate
versatile assay formats using 96 or 384-well plates, deep wells.
In May 2001, a new Staccato™ system was installed at our lab (Figure 1). The Reagent Addition Module was repositioned to increase deck layout flexibility. It was also critical to check the alignment of the pipettor head to the deck in order to avoid inaccurate tip loads.
Staccato at JRF
The high density pipettor SciClone™ is available in either a 9- or 15-position deck, it offers a large work surface and flexible deck configurations. It is able to pipette a variety of liquid solutions with a high degree of accuracy and precision between microplates (inter-plate variability) and tip-to-tip (intra-plate variability) within a single plate. The system is a multipurpose instrument for pipetting aqueous, organic, and protein solutions into 96- and 384-well microplates or custom racks. The automated reagent dispenser system (Figure 2a) and the reservoir options provide additional liquid handling capabilities to serve a large variety of protocols.
Reagent Dispenser
The Presto™ Autostacker provides high speed access to the local storage of plates, tips and custom racks. It has 70 independent shelves for flexible storage, up to 630 standard microplates. It contains a pneumatic shuttle mechanism with barcode option to transfer the stacker shelves between the storage unit and the SciClone™ pipettor. The optional De-Lidding system provides vacuum activated lid removal and placement for each microplate or tipbox lid within a shelf (Figure 2).
De-Lidding system
EXPERIMENTAL
One of the most common and time-consuming activities in the screening operation is the plate replication process, transferring microliter volumes from a source rack to a destination plate or rack for further storage or analysis. It is often desirable to dispense a sample or reagent onto a dry plate and a challenging task to prepare assay plates with compound samples at the appropriate reaction concentration. Therefore, tips must be kept clean of contaminants or samples may “wick” up the side of the tip instead of being deposited onto the dry plate. Uniformity of the tip length is also very important. Though the tolerance for tip length is very tight, there may be slight variations across the tip array and the Tipbox Load position must be defined so that the tips are firmly mounted on the tip adapters otherwise it always resulted in worse percentage CV's. To investigate the limits of precision dispensing with the SciClone™, experimental data were obtained using the 96-channel dispense head with 100 μl disposable tips.
Protocol for Wet & Dry Dispensing
Tartrazine solutions were prepared at a concentration of 0.6 mg/ml for each test criteria in water for the aqueous tests, in 100% DMSO (dimethylsulfoxide) for the organic solvent tests and in 5% FCS (fetal calf serum) for the protein solution tests. Volumes of 100 μL down to 5 μL were dispensed into both “wet” and “dry” Nunc 96-well microplates. For the wet plates, a buffer volume of 100 μL was pre-dispensed to all plates and the stated volume was dispensed onto the liquid surface. The dispense height was positioned at the meniscus of the liquid, allowing capillary action to aid in drawing the fluid from the tips. For dry dispense tests, the Tartrazine solution was added directly onto the plate surface followed by buffer addition as described above for wet dispensing. A new set of tips was used for each run of four plates. After mixing for two minutes with an orbital shaker the plates were read with the Titertek Multiscan™ plate reader at 450 and 620 nm. A four plate replicate set was performed for each volume with each solution to determine the precision (percentage CV) of each dispense. This was calculated from the standard deviation of each plate divided by the mean and expressed as a percentage.
RESULTS
Table 1 lists the percentage CV obtained for wet dispensing. In the last column, the Zymark Rapidplate™ specifications are listed as reference because the SciClone™ uses the same piston driver technology. It is important to emphasize that these tests are done without any fine-tuning of the protocol. The results obtained in DMSO are somewhat better then those obtained in water. Our results were as expected in general agreement with the specifications reported by Zymark.
Wet dispensing (percentage CV)
Performance Fine-tuning
In this protocol 11 μl Tartrazine, dissolved in DMSO, is dispensed onto dry plates. A four plate replicate set was performed to determine the precision (percentage CV) of each incremental dispense. For all speed codes tested from 2 to 100 μl/sec it was found (but not shown here) that a slower aspirating/dispensing speed resulted in a better reproducibility or performance.
Table 2 shows the calculated percentage CV's for each of the two runs of four microplates which make up test A to D. The following parameters were investigated:
variation of the dispense heights and vertical tip dips (A,B);
horizontal instead of vertical tip touches (C);
the effect of using a pre-dispense volume which is larger then the final dispense volume (D).
Incremental Dry dispensing
The most critical factor in dry pipetting is to just touch the surface of the microplate. This should ensure uniform contact across the plate surface and allowing the liquid to be pulled from the tips resulting in accurate and precise dispensing. Nevertheless, not all of the droplets remain in the middle of the well to allow capillary action to pull the liquid from the tips. While dispensing, some of them directly spread to the out-side of the well. This unequal spreading phenomena clearly affects the results. Therefore, a tip touch moving left or right to a width specified just after dispensing, equals the spreading to the outside of the wells and allows to just touch the meniscus of the droplet. It always resulted in better CV's.
In addition, the use of both a pre-dispense and a carry volume significantly enhances the pipetting performance in any protocol. The pre-dispense function starts the piston movement in the proper direction with liquid before the actual dispense. The overfill volume keeps liquid following the dispense volume to keep the volume accurate. Of all conditions tested, the combination of a large pre-dispense volume (20 μl) with a critical dispensing height and a horizontal tip touch resulted in the lowest percentage CV. The data also indicate that the SciClone™ is capable of dispensing volumes of 11 μl with a CV of less than 2% using 100 microliter disposable tips. A SciClone™ configured with a replaceable cannula array with 96 or 384 fixed tips could even do better. We ourselves have chosen the 96 disposable tip configuration to be able to work with sterile tips in cell-based assays.
CUSTOM NON-TRADITIONAL APPROACHES
The following examples of new approaches illustrate how one could use a SciClone™ in a non-traditional way.
Example 1
As mentioned earlier, our standard system uses 9-position trays. But in one of the protocols we do need 10 positions (three tipboxes, three stock racks and four daughter racks). Three 80-format stock racks (96-well plate or custom rack) must be reformatted into a 60-format, i.e., three mother plates to four daughter plates. In a traditional routine every tray is loaded in the same way and the same routine is executed on every tray. In this case it was necessary to access two trays within the same routine. Therefore, two trays are loaded with the 10 disposables needed but the trays are defined as two different materials: Tray1 and Tray2 (Figure 3).
SciClone™ Deck Multiple Tray Layout-Plate reformatting 80 to 60
The Clara™ Execution Manager software allows the pick up of different materials in the same run and presents them to the SciClone™. This overcomes the restriction that you only can run protocols based on a maximum of a 9-position deck layout.
Example 2
Another example of creative thinking is the incorporation of a custom 96-channel aspiration tool incorporated in the SciClone™ environment (Figure 4–5). This allows aspiration on the fly, 96- or 384 well plates on the SciClone™ deck. Tip adapters, mounted on top of the aspiration block, can be picked up or ejected just like a tipbox being handled by the SciClone™ head. Using this strategy a plate washer and a robot in combination with the SciClone™ are not needed to do ELISA-like assays as all attributes are located on the SciClone™ deck itself.
Parked Custom 96-channel Aspiration Head
FUTURE PERSPECTIVES
A few applications have already been transferred to the Staccato™ environment, but a lot more can probably be done on our system in the future. It is our intention to change our current screening approach of making serial dilution within a plate. Dose response curves (DRCs) will be made in a set of plates from one plate to another. On our older robot system, with a six channel dispenser, it takes almost five minutes to make a DRC of five concentrations for 12 compounds. With the SciClone™ we can make the same DRCs for 60 compounds in the same period of time. This means we can increase our throughput by five times, at least theoretically.
Another example of an increased throughput was the migration of one of our antiviral enzymatic screens to our Staccato™ platform. In about one week, the previously manually performed anti-neuraminidase assay was transferred and validated on the Staccato™ system. From compound distribution into assay plates to assaying itself, the whole procedure was automated. In the past we were able to test 1600 compounds in a five hour run in a 96-well format. From now on, even 384-well formats can be performed, that was manually impossible. Again a dramatic increase in throughput is achieved: 6400 compounds can be processed in a three and a half hour run. Within an eight hour work day, two runs can be processed on the system. On a weekly base the theoretical throughput has been increased 10-fold, from 6000 to 64000 compounds.
Graph 2. Weekly Throughput - Migration from manual (96-well format) to automated (96 - or 384 well format).
CONCLUSION
In conclusion, we have successfully validated our Staccato™ system. It allows us to perform complex applications with a minimum of manual interventions. Staccato's ™ user-friendly programming environment allowed our operators to perform sophisticated programming after only one day of training. It gives the ability to create new applications in a shortened time frame. Every day we discover new opportunities that increase even more the benefit of the Staccato™ system. The new system gives us the ability to handle large numbers of microplates (96- and 384-well) and to automate our sample distribution work previously done manually. In addition the implementation of the Staccato system increased the throughput of cell-based and enzymatic assays five and ten times respectively.
The 100 μl disposable tips tested performed well when dispensing volumes down to 5 μL of aqueous, protein based, and DMSO solutions into wet and dry microplates. When dispensing onto dry plates, somewhat higher CV's are seen, but results are still very good, percentage CV below two for an 11 μl volume. After fine-tuning the liquid-handling protocols the accuracy increased even further below the expected values. Using the proper dispense protocol, more accurate and precise 96-channel dispensing can be achieved at any throughput level.
Acknowledgements
The author would like to thank Danny Sas of the JRF Laboratory Automation Department for his technical assistance.
TRADEMARKS
Zypettor, Rapidplate, Staccato, SciClone, Clara, Presto are registered trademarks of Zymark Corporation. Titertek Multiscan is a registered trademark of ICN Biomedicals, Inc.