Serial vs Direct Dilution - Time to apply new thinking to IC50 determination and dose-response analysis?
The serial dilution method is standard practice in the preparation of doseresponse series for IC50 determination. However, it is well recognised that inadequacies in the liquid handling or mixing technique will affect the dilution ratio and hence the compound concentration and any errors will be compounded during each successive serial dilution, mix and transfer.
A recent poll of end users ranked better precision, particularly at lower drug concentrations, and the reduction in compound precipitation as the improvements in dose-response analysis they most desired. In addition, it is now suspected that hydrophobic compounds may be lost from solution during aqueous serial dilutions and absorbed to intermediate plastic surfaces. This in turn adds to concern over the reliability of the results generated and the extent to which they are a true reflection of the potency of the compounds being evaluated.
As part of the general drive to enhance the quality of screening data generated researchers are investigating strategies based on the direct dilution of micro-volumes of compound (ie on a volumetric basis). These investigations have been aided by the availability of low volume dispensing systems with good precision at low nL dispense volumes and a relatively wide dynamic range.
Some groups are now reporting that IC50 values of compounds tend to be lower (more active) when the concentrations are made via direct dilutions. It is increasingly evident that direct dilution has a future role to play in doseresponse analysis and where acoustic droplet ejection is preferentially deployed additional benefits will be derived in terms of reduced waste stream generated, less source material used and no cross-contamination.
The uptake and use of nanolitre dispensing within drug discovery is now widespread, with nearly two-thirds of all groups involved in dispensing activities having access to nanolitre (nL) volumes today and most of the remainder considering or planning to acquire a nL capability in the future (Figure 1) (1).
Although plate replication and compound reformatting still represent the main applications for nanolitre dispensing in drug discovery today, there is increasing use being made of cherry picking low volumes both for, or directly used in the set up of, IC50 or dose-response analysis (57% now using) (Figure 2).
Most of the nL dispenser offerings available today (eg Beckman PicoRAPTR™, CyBi®-NanoJet or Labcyte® Echo®, etc) are standalone and typically have only a small amount of automation built into them. However, in order to fully automate the main applications of nL dispensing it is evident that users will need to integrate these systems with other components (eg input or output plate stacking/hotel; robotic plate handler; bulk reagent dispenser; and controlling software) and possibly other non-essential plate processing peripherals (eg incubator; mixer; lidder/delidder; sealer; reader, etc).
When interest to purchase a standardised fully automated nL dispensing system for the main applications of nL systems was investigated it was apparent that most end users would like to access a system that is flexible enough to be able to perform multiple applications, ie compound reformatting combined with dose-response and IC50 preparation in the same system (Figure 3).
The generation of dose-response curves requires the preparation of solutions of compound at various concentrations over a wide range, often covering six logs of magnitude. These concentrations have more traditionally been made by a serial dilution technique in which a stock solution of the active compound of interest, typically in 100% DMSO, is cherry picked. This may take the form of aspirating an aliquot from a selected well in a source library plate or by punching out a prealiquoted volume stored in mini-tube.
In either case, the aliquot is then diluted with aqueous diluent or buffer at a fixed ratio (eg 1 into 3) and mixed thoroughly, often by repeat aspirate and dispense cycles within the pipette tip. An aliquot of diluted drug is then removed and added to a volume of new aqueous diluent at the same ratio as the first dilution.
Typically dilutions are performed in adjacent wells along the row or column of the plate, with successive serial dilutions made until the dose response range required in the series is achieved. When this intermediate plate is complete, it can be reformatted or replicated to multiple assay plates. The key point with serial dilution being that it has been standard practise to dilute drug stocks that were initially prepared and solubilised with 100% DMS0, with aqueous solutions to minimise the final DMSO concentration in the assay.
In addition, it is usual to prepare relatively large (μL) volumes of a dilution series in a separate intermediate plate or series of tubes, distinct from the ones in which the assay(s) will be undertaken, contributing to the overall greater use of compound than is necessary for the setup of multiple assay dose-response curves. (Figures 4 and Figure 5).
The alternative approach to serial dilution is the direct dilution of micro-volumes of compound (ie on a volumetric basis). In this case the volume actually dispensed is directly proportional to the amount of compound required to give the desired concentration in the chosen final assay volume.
Although the concept of direct dilution has been around and discussed for many years, the recent availability of low volume dispensing systems able to deliver with better precision a minimum volume dispense in the region of 1 to 10nL and a relatively wide dynamic range, has resulted in the approach finally being investigated more widely, with the practicalities of how the technique might be routinely implemented considered and the benefits realised.
As with most of the recent changes adopted in drug discovery processing, many improvements have been driven by the desire to enhance the quality of data generated. In the case of dose-response analysis, the concern has been the reliability of the results generated and the extent to which they are a true reflection of the potency of the compounds being evaluated. The relative ranking of the importance of improvements desired when undertaking dose-response analysis today are presented in Figure 6.
This shows that better precision (%CV) particularly at lower drug concentrations and the reduction in compound precipitation (induced by poor solubility in aqueous solution which could lead to false negatives) were ranked as the improvements respondents most desired. Interestingly, progress towards addressing this entire list of improvements could be expected to be derived from the application of direct dilution approaches.
With this in mind, interest in performing automated IC50 and dose-response analysis by direct dilution was investigated (1). It was found that around 1 in 5 (19%) current users of nL dispensing are already using direct dilution strategies today and a further 45% are considering it (Figure 7), suggesting that the expectation is that tangible benefits will be derived from the direct approach.
In the following vendor updates we will learn about some of the current approaches to serial and direct dilution for IC50 determination and doseresponse analysis.
In particular, we will examine the role that low volume and nanolitre dispensing can make to such activities; whether positive displacement micropipettes or piezo dispensing holds the key to direct dilution; how new automated, including microfluidic, systems could significantly improve such activities; where new tracking methodology is being applied to enhance the quality of dose-response data; where optimisation of dispensing and mixing parameters has yielded faster processing and superior data; and whether the IC50 is affected by the dilution strategy.
Artel (www.artel-usa.com) has enhanced its MVS® (Multichannel Verification System) to provide laboratories with the first standardised technology to verify the accuracy of dilution ratios in serial dilution protocols. This new capability is essential for drug discovery and other laboratories that rely on data generated from dilution-based liquid delivery procedures. To generate a proper serial dilution methodology, during which a sample solution is across a microtiter plate, accurate and repeatable volume transfer and efficient solution mixing are critical.
It is also imperative that the sample concentration in the initial solution be known so that the decreasing concentrations of sample across the plate can be correctly determined. Because sample concentration is directly affected by starting concentration and the amount of transferred volume, an inaccurate liquid handler can cause unseen, compounding errors in serial dilution applications, which can void the integrity of the dilution assay.
Previously, the only option available for laboratories to measure dilution ratios in serial dilution protocols were home-brewed methods requiring internally developed sample solutions. This time-consuming process does not allow for inter-laboratory comparability or traceability. By contrast, the MVS technology and range of standardised solutions allow for rapid verification of dilution ratios and provide traceable results for compliance and reproducibility.
The MVS verifies target volumes via ratiometric photometry, which uses the absorbance values of two proprietary dye solutions to calculate dispensed volume. The enhanced MVS now provides traceable measurements for single- or multiple-point dilutions to a concentration endpoint ratio of 1:2000 of the starting material (Figure 8).
Beckman Coulter’s (www.beckman.com) PicoRAPTR™ (Picolitre Rapid Transfer Robot) offers ultra-low volume aspirate and dispense operations. The system consists of an eight tip piezo head with eight independent fluid paths and dual plate carrier stage which allows for high speed aspiration (1μL to 100μL) and dispense (1nL to 100μL) volumes.
The ultra clean sonication wash, inert fluid pathways, precise spatial positioning and 100% DMSO compatibility enables high-quality dose response curves using direct or serial dilution. With DMSO correction of liquid and the extended pipetting dispense range you are able to achieve a 5 log concentration giving a percentage bias within 10% and precision %CV of less than 10%. 90% of the aspirated solution is available with using an air gap resulting in reagent cost savings and allowing for the filling of an entire microplate from a single aspirate.
The system will accommodate source plates of 96, 384, 1536 well microplates and destination plates of 96, 384, 1536, 3456 well microplates. Beckman Coulter also offers selective tip loading for the Biomek® FXP Laboratory Automation Workstation multichannel pipetting head. This allows for serial dilutions to be performed by configuring the loading of tips for any row(s) or column(s) the user needs to work with (Figure 9).
The Intelligent Sciclone i1000 from Caliper Life Sciences offers for the very first time independent pipetting control over each channel in a 96 channel pipetting head meaning that each channel can pipette its own unique fluid volume. The heart of the system is the real-time feedback, self-diagnosing and adaptive control enabled by the proprietary MEMS micro flow meters built in to each of the 96 independent channels.
One hundred and nine micro-processors monitor performance parameters in real time and facilitate ‘on the fly’ adjustments to compensate for variations like temperature and viscosity. In addition, each channel has the ability to pipette across an extremely wide dynamic range from 5 to 1,000 microlitres. With this capability, researchers can do one-step plate normalisations, IC50 experiments, serial dilutions and other multiparameter experiments at very high throughput.
For example, to perform a single step serial dilution, a volume file in an Excel format of 96 volumes of dilution buffer is given to the head followed by a file of 96 complementary volumes of target compound. Two aspiration steps, first of the buffer then the target compound, are followed by a single dispense step to create a plate with 96 different concentrations of the compound with a dilution factor of 1:200 across the plate. If required, this plate could be further diluted in a second step for a total dilution range of 1:40,000 (Figure 10).
For dilution operations in the low volume range CyBio provides two dedicated instruments. For serial dilutions the CyBi®- DiluSpro can be used, for direct dilutions the CyBi®-NanoJet is the instrument of choice. The CyBi®-DiluSpro is an economically priced 8-channel pipettor which is perfectly designed for the automatic preparation of serial dilutions. The large volume range (250μL head: 0.5-250μL; 25μL head: 0.1-25μL) allows the execution of serial dilution as well as the direct dilution in a special range (1:500) depending on the dispense volume.
The CyBi®-DiluSpro is a modular device. The user has the choice of three different carriage systems (4-, 5- place linear carriage and 10-place circular system), three different tip types for a single device and two different software packages (either optimised programs/ graphical user interface or a Scripting Studio for creating own programs) depending on the requirements of the customer (Figure 11).
The CyBi®-NanoJet is a very flexible, bulk dispensing instrument operating in the low volume range from 50nL to 20uL per shot. The pressure/valve based system can handle up to 16 independent reagent lines and can access every well individually in any SBS plate format. Reagents, cells or beads can be dispensed with a nearly unmatched precision. Direct dilutions can be performed using the instrument stand alone by adding two different reagents into single wells, which allows dilution rates of more than 1,000 (eg 50nL in 60uL). Depending on the operation mode the CyBi®-NanoJet can be used as a flexible assay workstation or as a fast HTS assay processor (Figure 12).
Genapta has been designing and building direct dilution potency systems in collaboration with a major pharma partner for the past five years. Its current product development (the PIC50 system) is for enzymatic assays, and later in 2007 will pick up cell based assays with the instrument being released on to the open market Q3, 2007. Direct dilution, rather than a well by well discrete approach, has mostly become a reality because of the availability of high quality pumping systems.
In particular the advent of nanoscale HPLC systems, has allowed the ability to control liquid flows down to a few nL per minute, which in turn enables detailed external control of the mixing conditions on a microfluidic chip. Thus it is possible to create a dynamic and varying concentration gradient against which to perform a competitive binding assay by controlling flow rates rather than trying to dispense and mix metered volumes of reagents into a well. This approach has a number of advantages that include repeatability and ease of set up. In addition to the flow control, a miniaturised spectrometer is used to assay the reaction.
Together these elements form a miniaturised- screening platform. As the system takes a single slug of compound that is diluted as it is measured, the use of microwell plates as a carrier medium is greatly reduced, and the device needs little or no external automation to generate potency data for the medicinal chemist.
As such it allows this potency to be generated alongside basic toxicity in timescales of hours rather than weeks. Such a reduction in assay time increases the potential for iteration during discovery 10-fold, allowing many more elements in a candidate series to be characterised at the crucial later stages (Figure 13).
Acoustic droplet ejection (ADE) on the Labcyte® Echo® (www.labcyte.com) has been used to investigate direct dilutions of small volumes. Using the protocol outlined in Figure 14, 12-point direct dilutions over six orders of magnitude can be prepared.
Recently researchers at Bristol-Myers Squibb have reported (2) that the IC50 values of compounds tend to be significantly lower (more active) when the concentrations are made via direct dilutions of small volumes than when the concentrations are achieved with serial dilution (Figure 15).
It has been postulated that this was due to compounds adsorbing to the solid surfaces of the dilution plate wells and to the pipette tips used to transfer the fluids. Hydrophobic compounds suffered the most loss during serial dilutions. Recent work using mass spectroscopy to measure the concentrations of solutions has provided independent analytical support to the hypothesis that significant amounts of compound are lost during serial dilution.
The actual concentration of analyte in serially diluted solutions can be far less than anticipated, and may cause false negatives in primary screening tests for potency as well as inaccurate IC50 values. In addition to the loss of high value hits, these false negatives also create difficulties for medicinal chemists and chemo-informatics programmes attempting to draw conclusions about structureactivity relationships.
The recently launched Labcyte® POD™ 810 plate assembler (Figure 16) robotically integrates an Echo® liquid handler for ADE with plate storage, a bulk filler for reagents or diluents, and software designed to optimise direct dilution techniques for dose-response curves.
The POD plate assembler takes source plates and optimises the throughput of ADE to provide destination plates ready for assay or further storage. The need for high quality dose response data continues to drive the demand for serial dilution, especially in the area of low volume dispensing. This is reflected in a continued level of interest for serial dilution functionality in the fully integrated liquid handling systems that RTS Life Science produces.
A system recently developed for one client utilised twin Velocity11 VPrep liquid handlers installed within an automated robotic cell. One of the key challenges that RTS had to overcome was the need to dynamically generate custom plate well information so as to ensure that the correct serial dilution was performed for specific plates being processed through the system. Key to this was the tight integration afforded by RTS’ d-Sprint™ and Sprint™ software platforms.
In addition to the VPreps, the system includes two Thermo Cytomat storage units (one with assisted thawing capability), a plate sealer and piercer from Abgene, RTS’ own-design 384 tube sealer/cutter instrument, as well as numerous lid handlers and barcode readers etc, all of which are controlled via RTS’ Sprint™ scheduling software. Directly fed from an RTS Sample- Store™, itself running d-Sprint™, via the integrated high-speed 384 Multi-Pix™ picking station, this system provides users with highly functional serial dilution capabilities capable of operating 24/7 (Figure 17).
Siliflow designs, manufactures and integrates liquid handling solutions. Its proprietary Nanodrop module is based on a piezo-actuated dispenser which allows ejection of droplets in the nanolitre to picolitre range (50pl to 10nl per drop). So far CVs of less than 3% have been achieved for nanolitre drops. Dispense frequency can be adjusted from 0.1 to 1Khz and exact number of drops (burst) can also be defined.
The drop size is defined by the nozzle diameter while various electronic parameters allow to accurately adjust this drop size. Drop volume can be calculated thanks to a synchronised stroboscopic vision system combined with a proper calculation algorithm. Combining these items Siliflow aims to fully automate the direct dilution process. Siliflow will integrate these functions together with an agitation module, in a 3-axis robot in order to launch a commercial direct dilution system in mid-2007.
TTP LabTech’s mosquito® product family facilitates assay miniaturisation through precise compound dilution on a nL scale in microplates containing up to 1536 wells. At the core of mosquito is a continuous reel of disposable, positive displacement micropipettes which can aspirate, dispense and even mix. The narrow bore pipettes can sample from the bottom of each well, ensuring negligible dead volume.
Since each micropipette is disposable, the instrument guarantees zero cross-contamination of samples and removes the need for unreliable, time-consuming and expensive solvent wash cycles. This means compound dilutions can be produced directly from high-concentration DMSO stock solutions with confidence. Mosquito has two formats applicable to compound dilution: mosquito X1 is a single micropipette ‘hit picking’ version for precision sampling of any individual well; while the mosquito HTS uses 8 or 16 pipettes in a columnar format for more rapid pipetting.
A key application of mosquito is the preparation of assay-ready plates within HTS departments. These plates contain test compounds prepared from library stocks ahead of time at the correct concentrations. DMSO is routinely used for compound dissolution, but its presence at concentrations above 1% can markedly affect assay performance. Thus, accurate nanolitre quantities of DMSO-based stock solution are required to keep concentrations low when preparing assay-ready plates.
On mosquito, serial dilutions are commonly performed in a 500nL total volume using an aspirate, dispense and mix cycle. Where compound carry-over is an issue, tips can be changed for each dilution step. Alternatively, direct dilution can be achieved by pipetting stock solutions throughout the instrument’s volumetric range. Routinely, this involves the repeat pipetting of 3-4 stock solutions at volumes between 50 and 1,200nL (Figure 18). Standard serial dilution methods are tedious, timeconsuming, and often inaccurate at high dilutions due to the propagation of errors.
Velocity11’s Bravo™ Liquid Handling platform is capable of performing serial dilution by row or by column. Coupled with the VWorks™ software package, parameters such as mix height, tip retraction, mixing speed and mixing cycles can cut the time to perform an accurate and precise serial dilution by 75%. The software tools ensure that the serial dilution can be tailored to specific applications.
The ability of one pipette head to perform both serial dilution and full 96- or 384- well dispensing, from low (2μL) high (200μL) volumes means that only one instrument is required to perform backfilling of plates and the serial dilution operation in a single protocol, with no head swap required. In a case study, we performed a standard 1:2 serial dilution by column across a 96 well plate. Using non-optimised parameters, 20 mix cycles were required to achieve satisfactory precision (average CVs under 3%, and accuracy >99%).
The drawback was that this would require 20 minutes to process an individual 96-well plate. After utilising the capabilities of VWorks software platform to optimise parameter values for mix height, speed of mixing, and tip retraction, essentially identical accuracy and precision results with just three mix cycles was achieved; this reduced the time for a serial dilution operation to 5 minutes (Figure 19).
In Table 1 we attempt to draw a comparison between the serial and direct dilution developments reviewed above.
It also serves to highlight how direct dilution approaches can reduce or minimise the perceived inadequacies of current dose response analysis techniques. It is well recognised that even minor problems in the liquid handling or mixing technique will affect the dilution ratio and hence the compound concentration, and these errors will be compounded during the serial dilution process increasing with each successive dilution and transfer.
It is, however, less appreciated that hydrophobic compounds may be readily lost from solution during aqueous serial dilutions with as much as 90-99% of the material absorbed to intermediate surfaces including disposable pipette tips and dilution wells. Direct dilution using ADE eliminates compound absorbance and adherence (sticking) since the sample does not encounter the pipette tips.
It is not known the extent to which hydrophobic compounds might adhere to the inner surfaces of fixed tip pipettors or the nozzles of nanolitre dispensers, some of which may have specialised non-stick surface coatings. If disposable tips are used only one time, there is no chance of cross-contamination, although repeated use would increase the chance of carryover.
As fixed tips pipettors or the nozzles of nanolitre dispensers come in contact with many different compounds on multiple occasions, even with the most stringent washing, there is the potential for cross contamination. To reduce carryover to an acceptable level it is routine to wash fixed tips or nozzles more thoroughly or frequently, and as a result the liquid waste stream generated will be high. If disposable tips are used only once, without washing, then the plastic waste stream also increases.
By adopting a direct dilution strategy significantly less source material and less solvent are required to make sample concentrations than traditional techniques employing serial dilutions. Overall, the case for considering direct dilution would seem to be selfevident, particularly if by using ADE the quality of hits could be improved and their true potency revealed. However, it remains to be seen what value the groups responsible for making dose response comparisons will place on these benefits and whether direct dilution will be adopted further, particularly if this involves the purchase of new ADE liquid handler. DDW
Dr John Comley is Managing Director of HTStec Limited, an independent market research consultancy whose focus is on assisting clients delivering novel enabling platform technologies (liquid handling, laboratory automation, detection instrumentation and assay reagent technologies) to drug discovery. Since it formation in 2003, HTStec has published 23 market reports on drug discovery technologies and Dr Comley has authored 17 review articles in Drug Discovery World. Further information on accessing the market report ‘Automated Low Volume Dispensing Trends 2006’ can be obtained by visiting www.htstec.com or by emailing firstname.lastname@example.org to receive a free copy of the Report’s Executive Summary and Table of Contents.
1 Automated Low Volume Dispensing Trends 2006, 44pp, published by HTStec Limited, Cambridge, UK, June 2006.
2 Spicer, J et al. Pharmacological Evaluation of Different Compound Dilution and Transfer Paradigms on an Enzyme Assay in Low Volume 384-well Format. Drug Discovery Technology Meeting, Boston, August 2005.