Low Nanoliter Acoustic Transfer of Aqueous Fluids with High Precision and Accuracy of Volume Transfer and Positional Placement

Introduction 

Acoustic droplet ejection (ADE) was introduced less than 5 years ago as a commercial technique using sound energy to transfer nanoliter volumes of dimethyl sulfoxide (DMSO).1, 2, 3 ADE has been widely applied to DMSO-based solutions and has been demonstrated to improve screening results in drug discovery and reduce the use of plastic consumables.4, 5

To expand the utility of ADE as a basic laboratory technique, we tested the transfer of a variety of aqueous salt and buffer solutions. These tested fluids encompass solutions of biological interest including such common solutions as phosphate buffered saline (PBS), a solution that closely mimics polymerase chain reaction (PCR) master mix, Dulbecco's modified Eagle's medium (DMEM), and buffers such as 4-morpholinepropanesulfonic acid (MOPS), 4-(2-hydroxyethyl)pepierazine-1-ethanesulfonic acid (HEPES), acetate, borate, and tris(hydroxymethyl)aminomethane/ethylenediamine tetraacetic acid (TRIS-EDTA). Further, we tested ADE for the transfer of a viscous glycerol-containing solution. Such viscous solutions can be difficult to transfer with pipettes, nozzles, and pin tools.

Materials and Methods 

We prepared various water-based solutions according to procedures in the literature or based on manufacturers' recommended procedures. These solutions included DMEM (Sigma D5796, St. Louis, MO) and PBS (Sigma D1408), both prepared according to manufacturers' suggestions. Luria-Bertani broth with 7.5% glycerol (LB broth, Teknova L8070, Hollister, CA) was prepared according to the manufacturer's protocol with the addition of glycerol. Solutions of 50mM HEPES (Calbiochem 391340, EMD Biosciences, San Diego, CA), pH 7.4; 50mM MOPS (Sigma M3183), pH 7.0; 50mM acetate (Sigma S7899), pH 7.2; and 50mM borate (Sigma B7660), pH 8.0 were all prepared from the free acid forms and pH adjusted with sodium hydroxide. A 50mM potassium phosphate solution, pH 7.3 was prepared by adding 1M dibasic potassium phosphate (Sigma P8584) solution to 1M monobasic potassium phosphate (Sigma 8709) solution until the pH measured 7.25. This stock solution was diluted with deionized water to 50mM, at which time the pH was measured at 7.3. A solution of TRIS-EDTA was prepared by mixing 0.5mL of 1M TRIS hydrochloride (Gibco 15567-027, Invitrogen, Carlsbad, CA), and 10μL of 0.5 M disodium EDTA solution (Gibco 15575-038) and then diluting to 50mL with deionized water. The pH of this solution measured 7.5. A solution of 10mM TRIS hydrochloride, 50mM KCl (Sigma P9333), 2mM MgCl2 (Sigma M1028), pH adjusted to 7.5 with sodium hydroxide (10N, JT Baker #5674-02, Phillipsburg, NJ) was used as a nucleotide-free substitute for PCR master mix. All solutions prepared from solids were filtered through a 0.2-μm filter (Nalgene 450-0020, Rochester, NY) before use. Solutions of dimethyl sulfoxide (DMSO, Sigma D5879) were prepared by mixing anhydrous DMSO with deionized water and allowing the solution to cool overnight in a sealed container. Fluorescein (Sigma F6377) was added to each solution so that the final concentration of fluorescein was 0.15mM. In the array-making section, glycerol (Sigma G5516) solutions were doped with Dylight 649 NHS ester (catalog # 46415, PerkinElmer, Boston, MA) and the arrays made on gamma-aminopropylsilane (GAPS)-coated slides (catalog # 40004, Corning Life Sciences, Corning, NY). In the experiments showing the spot-on-spot placement of two solutions, DMSO solutions of Cy3 and Cy5 mono-reactive dyes (product codes 25-8009-86 and 25-8009-87, respectively from GE Healthcare, Chalfront St. Giles, Great Britain) were spotted onto untreated glass microscope slides (catalog number MS-1000, Lab Storage Systems, St. Peters, MO). The source plate used in all acoustic transfers was a Labcyte Echo qualified, polypropylene 384-well plate (catalog number LP-05525) with the exception of the low-density arrays made on GAPS-coated slides which used an Echo qualified, low dead volume, COC, 384-well plate (384 LDV, catalog number LP-0200).

We used both the Echo 550 and the Echo 555 liquid handlers (Labcyte, Inc., Sunnyvale, CA) to transfer nanoliter volumes of solutions acoustically as indicated in the text. Adjustments to the acoustic power and focus were made to accommodate the different acoustic and fluidic properties of the ejected fluids and to achieve a nominal drop volume of 2.5nL. The information for these adjustments was stored in the Echo, enabling it to dispense the standard DMSO range and other fluids. These adjustments for fluids can be obtained commercially as options for the instrument. We measured precision and accuracy of transfer via fluorescein fluorescence by the procedure of Harris and Mutz6 in which they compensated for signal drift of the analytical plate reader both before and after samples were analyzed. In the case of bulk filling with 10mM NaOH, we used a Labcyte miniGene LD-01 liquid handler. Fluorescence was measured on a Wallac 2100 EnVision reader (PerkinElmer, Boston, MA). Arrays were read with a GenePix 4000B reader (MDS Analytical Technologies, Sunnyvale, CA).

Results 

Precision and Accuracy of the ADE Transfer Volume of Different Fluids 

We transferred 50nL of each solution from the wells of a 384-well source plate to each well of a 384-well destination plate. The volumes of the fluids transferred with an Echo 550 system were measured by fluorescence, as described by Harris and Mutz.6 Transfer volume statistics over the range of fluids are shown in Table 1, Table 2. As seen in the tables, the deviation from expected volume (accuracy) was less than 5% in all cases. The coefficient of variation (CV) was <5% in all cases. The precision of transfer but not the accuracy was also measured for 50-nL transfers of 50-mM buffers of sodium borate (pH 8.0), sodium acetate (pH 7.2), potassium phosphate (pH 7.3), HEPES (pH 7.4), and MOPS (pH 7.0). In each case, the CV was <2.0%. The highest CV over all tested fluids was 4.7% for the 50nL transfer of Luria-Bertani broth with 7.5% added glycerol. A limited number of experiments were performed with this fluid and no measurements to determine the accuracy of transfer were made.

Table 1. Transfer volume precision and accuracy over a range of fluids

	Precision and accuracy of ADE transfer volumes over a range of biochemical solutions
	Sample	Average transfer volume (nL)	σ (nL)	Volume deviation from expected (%)	Drop volume (CV%)
	DMEM	51.4	1.2	2.8	2.3
	10mM TRIS, 0.1mM EDTA, pH 7.5	47.7	0.7	−4.7	1.5
	10mM TRIS, 50mM KCl, 2mM MgCl2, pH 7.5	49.9	0.8	−0.2	1.6
	PBS	49.3	0.7	−1.5	1.5
	10mM TRIS, pH 7.5	49.8	0.8	−0.4	1.7
	50% Glycerol	51.0	0.8	2.0	1.5
	70% DMSO	50.5	1.0	0.9	2.0
	95% DMSO	47.7	0.7	−4.7	1.5

ADE, acoustic droplet ejection; CV, coefficient of variation; TRIS, Tris(hydroxymethyl)aminomethane; PBS, phosphate buffered saline; DMSO, dimethyl sulfoxide; DMEM, Dulbecco's modified Eagle's medium. DMEM and PBS were prepared as directed by the manufacturers. All transfers were nominally of 50nL. Measured volumes transferred ranged from 47.7 to 51.4nL. All results were within 5% of the expected 50nL volume. The CV for a 50-nL transfer ranged from 1.5% to 2.3%. A total of 384 measurements were used to determine the average and the CV% of transfers for each fluid. All transfers performed with an Echo 550 system.

Table 2. Transfer volume precision for some biochemical buffers

	Precision of ADE transfer volumes for five common buffers
	Sample, all buffers at 50mM	Drop volume precision (CV%)
	HEPES (pH 7.4)	1.63
	MOPS (pH 7.0)	1.90
	Potassium phosphate (pH 7.3)	1.51
	Sodium acetate (pH 7.2)	1.54
	Sodium borate (pH 8.0)	1.55

ADE, acoustic droplet ejection; CV, coefficient of variation. All transfers of 50nL. All buffers tested at 50mM concentration showed similar precision for transfer volumes. Transfers performed with an Echo 550 system.

Linearity, Accuracy, and Precision of Nanoliter Transfer Volume—10mM TRIS Hydrochloride Buffer 

TRIS is a commonly used aqueous buffer and we used it as a test solution for ADE. We measured the precision of 384 transfers of four different volumes (5, 10, 25, and 50nL) of 10mM TRIS with an Echo 550 system. Measured volume varied linearly with the expected volume, with a coefficient of determination, R2 of 0.9999 (Fig. 1). Error bars (three standard deviations) are indicated on the graph. The precision (CV) of the transfer of 10mM TRIS was <2.5% at all measured transfer volumes (n=384 for each volume). Deviation from expected volume was <2% in all cases (Table 3).

View Large Image Download to PowerPoint

Figure 1 A graph of the correlation between expected volume, that is, the volume set on the instrument, and the measured volume as determined by the method of Harris and Mutz. Data values shown in Table 3. The fluid in this experiment was 10mM TRIS hydrochloride, pH 7.3. Error bars show three standard deviations from measured values. Each average volume is based on 384 individual transfers.

Table 3. Transfer volume statistics for 10-mM TRIS buffer

	Precision and accuracy of ADE transfer volumes for 10-mM TRIS hydrochloride buffer
	Expected volume (nL)	Average measured volume (nL)	σ (nL)	CV (precision) (%)	Volume deviation from expected (accuracy) (%)
	5	5.0	0.11	2.20	0.81
	10	10.2	0.20	1.98	1.56
	25	25.3	0.41	1.61	1.29
	50	49.8	0.84	1.69	−0.39

ADE, acoustic droplet ejection; CV, coefficient of variation; TRIS, Tris(hydroxymethyl)aminomethane. Both precision and accuracy of transfer are independent of the volume transferred over the range tested. Each average volume and each CV measurement are based on 384 individual transfers.

Accuracy and Precision of Transfer are Independent of Buffer Concentration 

We measured the accuracy of the volume transfer with an Echo 550 system over a range of different concentrations (10, 50, 100, and 200mM) of TRIS hydrochloride buffer (Fig. 2 and Table 4). The extremely low coefficient of determination (R2=0.0063) in the curve of Fig. 2 indicates a lack of dependence of the transfer volume on the concentration of buffer. Likewise, the precision, as measured by CV (Table 4) is independent of buffer concentration.

View Large Image Download to PowerPoint

Figure 2 Measured transfer volume as a function of TRIS buffer concentration. Each average volume is based on 384 individual transfers. The error bars show three standard deviations. Data presented in Table 4.

Table 4. Precision and accuracy of volume transfer for different concentrations of TRIS buffer

	Precision and accuracy of ADE transfer volumes are independent of buffer concentration
	Concentration	Expected volume (nL)	Average measured volume (nL)	σ nL	CV (%)	Volume deviation from expected (%)
	10mM TRIS	50	49.8	0.84	1.69	−0.39
	50mM TRIS	50	48.6	0.77	1.58	−2.73
	100mM TRIS	50	49.3	0.75	1.53	−1.44
	200mM TRIS	50	49.5	0.69	1.40	−0.96

ADE, acoustic droplet ejection; CV, coefficient of variation; TRIS, Tris(hydroxymethyl)aminomethane. The CV values ranged from 1.40% to 1.69%, whereas deviation from the expected volume ranged from −0.39% to −2.73%.

Precision and Accuracy for Transfer of 50% Glycerol 

Viscous solutions are often difficult to transfer by traditional methods. When transferring viscous solutions, both the precision and accuracy of the transfer may degrade. Glycerol is often added to biological solutions to modify viscosity, osmotic pressure, and as a carbon source. We measured the precision of acoustic transfer of a 50% glycerol solution, for both 5-nL (three 384-well plates for 1152 measurements) and 50-nL (one 384-well plate) volume transfers. The actual volume transferred with an Echo 550 system was measured via fluorescence for each well, and the results are shown in Table 5. The set of 1152 five-nanoliter transfers had a CV of less than 2.5%; the 50-nL transfers showed a CV of 1.54%.

Table 5. Precision and accuracy of transfer volumes of 50% aqueous glycerol

	Precision and accuracy of ADE transfer volumes of 50% aqueous glycerol
	Transfer volume (nL)	σ (nL)	CV (%)	Measured volume	Volume deviation from expected (%)	N (number of wells)
	5	0.11	2.09	5.10	1.80	384
	5	0.12	2.32	5.10	2.40	384
	5	0.10	1.99	5.10	1.60	384
	50	0.78	1.54	51.0	1.98	384

ADE, acoustic droplet ejection; CV, coefficient of variation. At both 5- and 50-nL transfer volumes, acoustic droplet ejection is extremely precise with CV less than 2.5% in all cases.

Array Spot Size and Placement on GAPS-Coated Slides 

An array was produced using an Echo 555 liquid handler. The solution of 50% aqueous glycerol containing Dylight 649 NHS ester was transferred to a GAPS-coated slide. The spot spacing was 800μm, with each spot resulting from one ejected 2.5-nL droplet. The resultant spots showed little obvious distortion in any direction.

The array was optically scanned using a GenePix 4000B scanner, and the GenePix software was used to extract the size and location of the spots. The average spot diameter was measured to be 235μm, with a CV for spot size of 1.8% based on n=384. The spot locations determined by the scanner were fit to a square grid of 800-μm spacing. The fitting was done by globally adjusting the translation and rotation of the ideal 800-μm grid so as to minimize the squared error between the ideal grid and measured spot locations, summed over all spots (see Fig. 3). For the 50% glycerol array, the average magnitude of the spot displacement in the x-direction from the best-fit grid locations was found to be 29μm; the average magnitude of the spot displacement in the y-direction was found to be 21μm.

View Large Image Download to PowerPoint

Figure 3 An array was created by acoustically ejecting a 2.5nL drop of fluorescein-doped 50% glycerol from each well in a 384LDV plate. Image scanned on a GenePix 4000B scanner.

Spot Placement Properties of Other Fluid Arrays 

Spot arrays were produced in the same manner, using other source fluids. Results of the scanned and fitted spot array data for such fluids are shown in Table 6. The spot placement statistics are found to be relatively insensitive to the composition of the fluid or the concentration of buffer.

Table 6. Spot size and placement accuracy for arrays of different fluids

	Spot size and average displacement with respect to fluid
		Average spot diameter (μm)	Average spot displacement in x-direction (μm)	Average spot displacement in y-direction (μm)
	10mM TRIS, 0.1mM EDTA, pH 7.5	289	50	34
	50mM TRIS, 0.1mM EDTA, pH 7.5	294	45	34
	100mM TRIS, 0.1mM EDTA, pH 7.5	284	45	36
	200mM TRIS, 0.1mM EDTA, pH 7.5	251	43	32
	PBS	293	38	41
	10mM TRIS, 50mM KCl, 2mM MgCl2, pH 7.5	295	43	39
	DMEM	237	41	33

TRIS, Tris(hydroxymethyl)aminomethane; PBS, phosphate buffered saline; DMEM, Dulbecco's modified Eagle's medium. The spot displacement is that from a grid of 800×800-μm spacing, translated to produce the best overall fit to the measured spot locations.

The average radial displacements of the spots from the best-fit grid were between 30 and 50μm. The average spot size for each liquid varied somewhat, with more viscous solutions (DMEM and 200mM TRIS) producing smaller spots as was also seen with 50% glycerol.

Spot-on-Spot Grid Formation 

To further demonstrate the arraying capability of the Echo 555 system, Cy3 and Cy5 reactive dyes were diluted 1:1000 in neat DMSO (Cy3-DMSO and Cy5-DMSO), and then spotted onto clean untreated glass microscope slides using the Echo 555 liquid handler. In Figure 4, an array of 4×24 spots is shown, as imaged by the GenePix 4000B scanner. The array is composed of two subarrays. The first subarray consists of 4×16 spots of fluid with Cy5, deposited along the 16 leftmost columns in the image. The second subarray consists of 4×16 spots of fluid containing Cy3, deposited along the 16 rightmost columns of the image. The spots containing only Cy5 appear red; those containing only Cy3 appear green. The spots in which both Cy3 and Cy5 drops have been deposited appear yellow. Each single spot results from one 2.5-nL droplet, and the overlapping spots correspond to a total spot volume of 5nL.

View Large Image Download to PowerPoint

Figure 4 Spot-on-spot array made with an Echo 555 liquid handler. The spots, 400μm in diameter, are 1000μm center to center. Spots of Cy5 in DMSO are red. Spots of Cy3 are green. Where the spots of Cy5 and Cy3 are coincident, the resulting spot is yellow. Image scanned on a GenePix 4000B scanner (MDS Analytical Technologies, Sunnyvale, CA).

Arrays in Multiwell Plates 

We used Echo liquid handlers to create microarrays in the bottom of 96-well plates. In Figure 5, each well of the 96-well plate contains a 4×4 array at 1000-μm center-to-center spacing. Each array spot was created by ejecting a single 2.5-nL droplet of 70% DMSO.

View Large Image Download to PowerPoint

Figure 5 4×4 Arrays of 70% DMSO were created in each well of a 96-well plate with an Echo 555 liquid handler. Close-up view shows nine wells. Image scanned with a 4990 Photo scanner (Epson, Long Beach, CA).

Conclusion 

ADE is compatible with many aqueous solutions such as those used in the transfer of oligonucleotide solutions (e.g., PCR, RT-PCR, sequencing), protein solutions (e.g., protein–protein interactions, protein binding), cell-based analyses, and the high-throughput screening of water-soluble analytes. All solutions tested performed well. Transfer volume precision, as CV, was determined to be <5%, and accuracy, as measured by deviation from expected volume, was determined to be <5%. Precision and accuracy are relatively independent of volume transferred, concentration of solutes, and the types of solutes.

ADE can be used to manufacture low-density grids or spot-on-spot arrays without contamination inherent in pin tool transfers. Transfer of fluids from a source to a grid target showed an average displacement of droplets from a best-fit grid to be less than 50μm with little or no distortion in circular shape.

Because ADE has high precision and accuracy when transferring low nanoliter volumes, it offers the possibility of miniaturizing assays. Because ADE is not limited to DMSO solutions, the applicability of the technique is broad. ADE also offers the potential for low-density arrays of fluids as an alternative to other array making procedures.