Acoustic Droplet Ejection Technology and Its Application in High-Throughput RNA Interference Screening

Abstract

The development of acoustic droplet ejection (ADE) technology has resulted in many positive changes associated with the operations in a high-throughput screening (HTS) laboratory. Originally, this liquid transfer technology was used to simply transfer DMSO solutions of primarily compounds. With the introduction of Labcyte’s Echo 555, which has aqueous dispense capability, the application of this technology has been expanded beyond its original use. This includes the transfer of many biological reagents solubilized in aqueous buffers, including siRNAs. The Echo 555 is ideal for siRNA dispensing because it is accurate at low volumes and a step-down dilution is not necessary. The potential for liquid carryover and cross-contamination is eliminated, as no tips are needed. Herein, we describe the siRNA screening platform at Southern Research’s HTS Center using the ADE technology. With this technology, an siRNA library can be dispensed weeks or even months in advance of the assay itself. The protocol has been optimized to achieve assay parameters comparable to small-molecule screening parameters, and exceeding the norm reported for genomewide siRNA screens.

Introduction

In recent years, the idea of “moving liquids with sound” has been used to develop methods for acoustic dispensing. Acoustic liquid handling makes use of the fact that sound waves or acoustic energy can travel through solids and fluids. This principle and the physics of this technology have been extensively covered by Ellson and coworkers.^1–4 To summarize, during acoustic droplet ejection (ADE) transfer, the destination plate is upside down and the technique transfers liquid from the source well to the destination well without making contact with either well. After the source plate and destination plate are placed in the designated plate holders inside the liquid handler, a droplet is ejected from the source plate into the destination plate, which is suspended upside down above it. Surface tension ensures that the droplets stay in the wells. This eliminates the potential for contamination, unlike conventional transfer technology such as disposable tips or pin tools that contact the liquid and have the potential to contaminate the library.

RNA interference (RNAi), a mechanism for RNA-guided regulation of gene expression, is a widely used tool for target identification and reverse genetics. In the last 10–15 years, targeted knockdown of the expression of individual genes using small interfering RNAs (siRNAs) introduced into cells by the process of transfection has become a staple technique with which to investigate the function of specific genes.^5–7 A number of large-scale siRNA libraries are currently available to target complete genomes for multiple organisms, making it possible to carry out genomewide high-throughput screens. However, unlike small-molecule screens, which are generally robust and reproducible, the metrics for siRNA screens are usually not as good. The reasons behind this are both biological and technical. Downregulation of target gene expression by siRNA results in a certain level of knockdown of protein expression (as opposed to complete loss of expression), which may produce only partial effects, such as 50% knockdown versus 90+% kill in a virus assay. Technically, the transfection step in siRNA screening introduces additional variability, and since siRNAs may take several days to achieve maximum effect, siRNA screening assays are usually longer in duration than small-molecule screening assays. These limitations may also negatively impact the reproducibility of siRNA assays. One way to reduce this variability is through equipment optimization. We showed in a previous publication⁸ that certain factors should be monitored in order to achieve good assay metrics when designing an siRNA screen. Here, we focus more on one of these factors, which is the use of ADE technology for handling reagents that are not dissolved in DMSO, such as siRNA libraries. This is now feasible since ADE instruments include aqueous dispensing capabilities.

Materials and Methods

Cell Culture

Human embryonic kidney 293 (HEK293) cells (item CRL-11268; ATCC, Manassas, VA) were cultured in high-glucose Dulbecco’s modified Eagle medium (DMEM) (Gibco/Life Technologies, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS) (Life Technologies). Cells were grown in a humidified incubator at 37 °C, 5% CO₂. For screening experiments, a sufficient quantity of the cells for the entire screen was frozen at the same passage in Recovery Cell Freeze media (Life Technologies) and stored in liquid nitrogen prior to use.

siRNA Library

siRNA screening was performed with a subset of Silencer Select Human Genome siRNA library V4 (Life Technologies). Silencer Select Negative Control #1 siRNA and Silencer Select siRNA targeting Wee1 (s21; Life Technologies) were used as negative and positive siRNA controls, respectively. Library plates were resuspended in DNase/RNase-free water to the final concentration of 5 μM. An aliquot of the library was further diluted to 1 μM and transferred to Echo-qualified 384-well polypropylene source plates (Labcyte, Sunnyvale, CA) to use for dispensing into assay plates.

siRNA Transfection

siRNA transfection was performed by reverse transfection, with 0.2% RNAiMAX transfection reagent (Life Technologies) and 5 nM siRNA, unless otherwise indicated. Screens were performed in single siRNA/well format. siRNAs (150 nL of a 1 μM stock; 0.15 pmol/well) were dispensed into 384-well assay plates (Corning 3712; Corning, NY) using an Echo 555 Series Liquid Handler (Labcyte) or, where indicated, a Biomek FX (Beckman Coulter, Fullerton, CA). Assay plates prepared using the Echo had the siRNA droplets air-dried and were then kept at −20 °C until use. On the day of the screening, assay plates that were preplated with siRNAs were thawed and equilibrated to room temperature. RNAiMAX transfection reagent (0.06 μL/well) diluted in OptiMEM media (Life Technologies) was added to the assay plates (lipofectamine RNAiMax was diluted 1:167 in OptiMEM and 10 μL added to plates) using a Multidrop Combi dispenser (Thermo Fisher Scientific, Waltham, MA), with a standard tube dispensing cassette (Thermo Fisher Scientific item 24072675). After 1 h incubation at room temperature, cells (20 μL of 80,000 cells/mL; 1600 cells/well) were added to the assay plates with the Multidrop Combi dispenser and standard tube dispensing cassette. For the tip dispense, a working stock siRNA plate was made on the day of the assay (35 nM siRNA in OptiMEM) and 5 μL of this 35 nM stock (5 nM final concentration) was dispensed to assay plates using a Biomek FX liquid handler. The rest of the transfection protocol (addition of the transfection reagent and cells) was identical to that for the plates dispensed with the Echo. Assay plates were incubated at 37 °C, 5% CO₂ in a controlled-humidity incubator (Forma 3307; Thermo Fisher Scientific), and cell viability was assessed after 6 days using either CellTiter-Glo (CTG) (Promega, Madison, WI) or alamarBlue (TREK Diagnostic Systems, Oakwood Village, OH), as per the manufacturer’s instructions. Plates were read on an EnVision multilabel plate reader (PerkinElmer, Waltham, MA), collecting luminescence (CTG) or fluorescence (alamarBlue; excitation 535 nm, emission 590 nm).

Data Analysis

Data were imported into the ActivityBase (IDBS, Bridgewater, NJ) data management system for analyses. Data were normalized and reported as percent viability, which was calculated using the formula 100*(Data Value/Scrambled siRNA Control Value), where Data Value is the reading from each sample siRNA tested.

Results and Discussion

Compared to small-molecule screens, siRNA screens are less robust and show much more variability.⁹ The transfection step in siRNA screening is one of the sources of this variability. Part of this variability stems from the liquid handling irregularities during the transfer of siRNAs into the assay plate. The metrics for siRNA screening can be improved by minimizing the variables that affect these screens, and ADE technology has proven to be a great tool in reducing this assay variability.

Some Advantages of Acoustic Dispensers

There are several reasons why acoustic dispense is our preferred choice for siRNA transfer. First, there is a significantly reduced risk of contamination of the library source plate due to its noncontact nature. It is also very versatile, as any volume that is a multiple of 2.5 nL can be transferred from any well in a source plate to a completely different position on the destination plate. Having complete automated control of volumes transferred from each source well to the destination wells is a very useful feature, particularly during validation experiments when it is necessary to try multiple conditions. Transfer between different plate formats (384 to 1536, 1536 to 384, etc.) is also relatively easy. Cost is always a factor in HTS screening campaigns. ADE may reduce overall costs.^10–13 We reduced our disposable tip costs by $100,000 per year when we started using the Echo, and in just a few years, the instrument paid for itself. With the use of the Echo, we have also been able to move more assays into the 1536-well format, which generally requires less reagent than screening in the 384-well format.

Since only nanoliter volumes will be dispensed, the siRNA library can be kept at a higher concentration and therefore remain more stable (Thermo Fisher Scientific R&D, unpublished data, and Olivieri et al.¹⁴). No predilution in media is necessary, reducing the up-front preparation that is usually a part of the process when using disposable tips. Once a dispense protocol has been established during assay validation, no additional effort is needed for subsequent batches, as the same protocol will be used each time. Another attractive feature is that the nanoliter droplets can be dispensed to destination plates weeks or even months in advance, the siRNAs dried-down for storage, and the assay performed on a much later date without loss of efficiency. In the dry state, the siRNAs are very stable; in addition, using predispensed, dried siRNAs streamlines the HTS process by eliminating or reducing the preparation needed on the day of the assay. In contrast, a dispenser that uses tips generally requires that the siRNAs be prediluted in media before transfer to assay plates. This not only necessitates some preparatory work every day a screen is run but also is a source of increased variability.

Acoustic Liquid Handling versus Disposable Tips

We evaluated the delivery of siRNAs for transfection, using both acoustic dispense and disposable tip protocols. The ADE instrument chosen for this project was Labcyte’s Echo 555 Series Liquid Handler. The Echo 555 has both DMSO and aqueous dispense capability, which was important since the siRNA library is dissolved in water. The Beckman Coulter Biomek FX was used as an example of a conventional liquid handler that uses disposable tips.

We first examined the quality of data from each instrument and compared them to data from manual pipetting. As seen in Table 1, the precision and accuracy of siRNA transfer was high for both the Echo and the FX, and both were comparable to data obtained by hand pipetting. All coefficient of variation (CV) values were below 10% and Z′ factor values were comparable at 0.7. In another experiment, the correlation between the Echo and FX was evaluated. When two plates were dispensed with the Echo, the correlation between them was very high with a Pearson’s correlation value of 0.95 (Fig. 1A). Similarly, two plates dispensed with the FX showed high correlation with a Pearson’s correlation value of 0.96 (Fig. 1B). The mean values from the two plates (Echo 1 and Echo 2; FX 1 and FX2) were used to compare the plates dispensed with the Echo to the ones dispensed with the FX (Fig. 1C). The correlation between the Echo and FX was also very good, with a Pearson’s correlation value of 0.89 (Fig. 1C). We noticed that in the Echo versus FX comparison, most of the points with viability less than 75% fall below the regression line, suggesting a trend of lower viabilities for the FX data than the Echo data. However, we believe that this is probably due to variability between experiments, as the two experiments were performed on separate days. Some day-to-day variability is possible due to subtle differences from reagents or equipment when the same experiment is performed on two different days.

Table 1.

siRNA Transfection by ADE Technology and the Use of Disposable Tips.

Assay Parameter	Echo 555	Biomek FX	Manual Pipette
CV of positive control	9.9%	8.7%	8.3%
CV of negative control	6.2%	6.3%	5.4%
Z′ factor	0.70	0.70	0.71
Signal-to-background ratio	5.3	5.1	4.3

Data from positive and negative control siRNAs where the positive control is siRNA targeting Wee1 and the negative control is scrambled, nontargeting siRNA are shown. siRNAs were dispensed either with a noncontact dispenser or with disposable tips. For the noncontact dispense, 1 μL of a 125 nM stock of each siRNA was transferred into 384-well assay plates using an Echo 555 Series Liquid Handler (Labcyte), for a final concentration of 5 nM (25 μL final volume per well). For the tip dispense, a 35 nM working stock was made in OptiMEM and 5 μL of this 35 nM stock dispensed to assay plates using a Biomek FX liquid handler (Beckman Coulter). Five microliters of this 35 nM stock was also dispensed by hand using a multichannel pipetter for a final concentration of 5 nM (35 μL final volume per well). The rest of the transfection protocol was as stated in the Materials and Methods section. All plates included a negative control (scrambled, nontargeting) siRNA and a positive control siRNA (transfection efficiency control) at 5 nM. The number of wells for each control on each plate was 32. n = 2 for the Echo, n = 1 for the FX, n = 2 for manual pipetting. The average value is provided for the signal-to-background ratio, CV, and Z′ factor.

Figure 1.

Pearson’s correlation plots of percent viability values comparing siRNAs dispensed by acoustic transfer and by disposable tips. For comparison, the same set of siRNAs was dispensed with the Echo and the FX at a concentration of 5 nM and viability was assessed 7 days posttransfection using CellTiter-Glo. Two replicate plates are shown for each dispenser, named Echo 1, Echo 2, FX 1, or FX 2. The correlation between the two Echo plates is 0.95 (A), and that between the two FX plates is 0.96 (B). The mean of the two Echo plates is compared to the mean of the two FX plates (C), with a Pearson’s correlation of 0.89.

Dried-Down siRNAs Can Be Used for Transfections Several Years Later

Drying-down siRNAs is both convenient and an excellent way to preserve the siRNAs to prevent degradation. The convenience of predispensing siRNAs by acoustic transfer is elaborated further in Figure 2, which shows that the library can be dispensed even years in advance of the assay. In Figure 2A, the transfection reaction was performed immediately after the siRNAs were freshly transferred to assay plates (referred to here as “Fresh”). At the same time, siRNAs were transferred to replicate plates, air-dried, and stored at −20 °C for 3 days before transfection (referred to here as “Dried”). Another set of plates was air-dried and stored at −20 °C for 5 years (the siRNAs were transferred in October 2010 and the transfection done in September 2015) (referred to here as “Dried Stored”) (Fig. 2B). The heat map in Figure 2A is similar to that in Figure 2B, suggesting that the siRNAs show a similar effect even 5 years after the transfer was done. The data from all three conditions were very reproducible, with a Pearson’s correlation value of 0.97 between Fresh and Dried (Fig. 2C), a value of 0.92 between Fresh and Dried Stored (Fig. 2D), and a value of 0.90 between Dried and Dried Stored (Fig. 2E). The R² value is a little lower (more scatter) when the Dried Stored data are compared to the Fresh or Dried data (Fig. 2D,E). The Dried Stored transfection experiment was done at a different facility, with cells at a different passage, using a different plate reader, and all these can account for some variability when compared to the Dried or Fresh. However, overall data are very reproducible, even between the experiments that were performed 5 years apart and at two different institutions using different equipment and cells at a different passage. The good correlation between Fresh, which is considered the benchmark, and Dried Stored, shows that one of the advantages of using the Echo is to preprint several copies of the siRNA library and store them for years until needed.

Figure 2.

Heat maps and Pearson’s correlation plots of percent viability values comparing siRNAs dispensed by acoustic transfer on the same day, with transfections done 5 years apart. (A) Heat map of raw values from a 384-well plate in which siRNAs were transferred and transfection performed immediately after (Fresh siRNAs). (B) Heat map of raw values from a 384-well plate in which siRNAs were dried-down and stored at −20 °C for 5 years before transfection (Dried Stored siRNAs). The correlation between freshly dispensed siRNAs and siRNAs dried-down and frozen for 3 days is shown (C), while D is the correlation between freshly dispensed siRNAs and siRNAs dried-down and stored at −20 °C for 5 years. (E) Comparison of siRNAs dispensed, dried, and stored at −20 °C on the same day, but transfection reactions performed 5 years apart. There is no loss of efficiency, even after 5 years. All data shown are for replicate siRNA plates and are representative of each condition tested. The mean percent viability value from replicate plates was used for C, D, and E (n = 3 for C; n = 2 for D and E). The plate consisted of 352 sample siRNA wells and 32 control siRNA wells. All controls were in columns 23 and 24. There were three positive controls on each plate: Wee1 siRNA, which strongly affects viability of the cells, and two other siRNAs with different levels of inhibition (four wells for each positive control). The negative control (20 wells) was a scrambled siRNA (Silencer Select Negative Control #1) with no effect on cell viability. Viability was assessed 6 days posttransfection using either CellTiter-Glo or alamarBlue. The Dried Stored assay was performed at the high-throughput siRNA/miRNA screening facility at the University of Alabama at Birmingham, and these plates were read on a CLARIOstar plate reader (BMG Labtech, Cary, NC). Data values are normalized to the scrambled siRNA control wells in columns 23 and 24.

Advantages of Predispense and Nanoliter Volumes by Acoustic Transfer

As stated in the Materials and Methods section, after the siRNA library dispense process was complete, the plates were placed in a sterile biosafety cabinet to air-dry. They were then sealed with sterile aluminum foil seals and stored at −20 °C until needed for the assay.⁸ When prepared this way, the library can be dispensed and frozen weeks or even years in advance of the assay itself, making the assay protocol more amenable to HTS and more convenient. Additionally, with ADE technology, low-nanoliter volumes of each siRNA are dispensed into the wells of an assay plate. Acoustic liquid dispensers are self-calibrating and can determine how much liquid is present in each well. The low-volume dispensing ability ensures that the siRNA library can be maintained at a relatively high concentration, siRNAs being stable at concentrations of ≥1 μM and not affected by multiple freeze thaws (Thermo Fisher Scientific R&D, unpublished data). Another significant advantage to having a concentrated sample is increased dispense speed. ADE ejects droplets in 2.5 nL increments; therefore, the bigger the volume to be transferred to a well, the longer it will take to complete the transfer, which is why more concentrated samples need less dispense time. For instance, it takes approximately 45 s to transfer 15 nL/well across a 384-well plate, 1.5 min to transfer 150 nL, and 8.75 min to transfer 1500 nL (transfer time only, not including opening and closing the Echo door or getting the plates in and out of the instrument).

Conclusion

We have shown that by using ADE technology, data from siRNA screens can be as robust and reproducible as those using disposable tips. With this technology, the problem of carryover and cross-contamination from disposable pipette tips is eliminated due to the noncontact nature of the transfer process, which occurs with very high accuracy and precision. Because the transfer is noncontact and because the Echo liquid handler is housed within a BIOProtect II biological safety cabinet, the potential for contamination with RNases has been minimized. In addition, since ADE technology dispenses nanoliter volumes, the siRNA library plates can be kept at higher concentrations, which increases the shelf life of the library. Most importantly, several copies of the siRNA library can be made and stored for years or until needed. The siRNA data presented here are comparable to those from small-molecule screens.