Automated, simple, and efficient influenza RNA extraction from clinical respiratory swabs using TruTip and epMotion

Abstract

Background

Rapid, simple and efficient influenza RNA purification from clinical samples is essential for sensitive molecular detection of influenza infection. Automation of the TruTip extraction method can increase sample throughput while maintaining performance.

Objectives

To automate TruTip influenza RNA extraction using an Eppendorf epMotion robotic liquid handler, and to compare its performance to the bioMerieux easyMAG and Qiagen QIAcube instruments.

Study Design

Extraction efficacy and reproducibility of the automated TruTip/epMotion protocol was assessed from influenza-negative respiratory samples spiked with influenza A and B viruses. Clinical extraction performance from 170 influenza A and B-positive respiratory swabs was also evaluated and compared using influenza A and B real-time RT-PCR assays.

Results

TruTip/epMotion extraction efficacy was 100% in influenza virus-spiked samples with at least 745 influenza A and 370 influenza B input gene copies per extraction, and exhibited high reproducibility over four log₁₀ concentrations of virus (<1% CV). RNA yields between the three automated methods differed by less than 0.5 log₁₀ gene copies. 99% of clinical specimens that were PCR-positive after easyMAG or QIAcube extraction were also positive following TruTip extraction. Overall C_t value differences obtained between TruTip/epMotion and easyMAG/QIAcube clinical extracts ranged from 1.24 to 1.91. Pairwise comparisons of C_t values showed a high correlation of the TruTip/epMotion protocol to the other methods (R² >0.90).

Conclusion

The automated TruTip/epMotion protocol is a simple and rapid extraction method that reproducibly purifies influenza RNA from respiratory swabs, with comparable efficacy and efficiency to both the easyMAG and QIAcube instruments.

1. BACKGROUND

Influenza viruses continue to be a global public health threat, causing epidemics resulting in millions of clinical cases and up to 500,000 deaths each year.^1,2 Although influenza seasons vary in severity, approximately five to 20% of people are infected with influenza viruses annually in the United States,³ with an estimated 200,000 hospitalizations and up to 49,000 deaths each season^4,5. Following the 2009 H1N1 pandemic, influenza testing demands significantly increased, reinforcing the need for simple, rapid, sensitive, and increased throughput influenza testing capabilities by diagnostic laboratories⁶.

Due to the high sensitivity and specificity of molecular detection techniques, increasing numbers of diagnostic laboratories have implemented them to detect and characterize influenza viruses. High quality viral RNA is required for such techniques, and isolation of viral RNA from respiratory samples can be one of the most time-consuming steps, creating workflow bottlenecks in the diagnostic process. Several commercially-available, automated RNA sample preparation systems are available but can be expensive, involve tedious or confusing instrument setup, require large quantities of consumables, have large footprints, or are inefficient at RNA extraction.^{7, 8, 9, 10, 11} We recently described a simple, rapid, manual extraction method for purifying influenza RNA from respiratory samples.¹² Known as TruTip extraction, the process involves the use of a nucleic acid extraction matrix embedded inside a pipette tip.¹³ Here we developed and evaluated an automated TruTip extraction method using a robotic liquid handler, which improves sample throughput without sacrificing influenza RNA extraction and detection efficacy.

2. OBJECTIVES

To evaluate an automated TruTip RNA extraction procedure developed for respiratory samples using an epMotion 5070 liquid handling system (Eppendorf, Hauppauge, NY), and to evaluate the performance of the system relative to other automated extraction instruments commonly used in diagnostic laboratories (easyMAG and QIAcube).

3. STUDY DESIGN

3.1 Viral culture

Cultured viruses used for spike-recovery experiments included influenza A/New York/1669/2009(H1N1) and an influenza B virus isolated in 2008. Both viruses were isolated from respiratory samples submitted to the reference and surveillance programs at the Wadsworth Center Laboratory of Viral Diseases (WC-LVD), New York State Department of Health (Albany, New York). Viruses were propagated in primary rhesus monkey kidney (pRhMK) cells (Diagnostic Hybrids Inc.; Athens, OH) using classical virus culture procedures.

To determine stock viral RNA concentrations, viral RNA was extracted from 60 µL of culture harvests and eluted into 60uL using the QIAamp Viral RNA kit and QIAcube instrument (Qiagen, Valencia, CA) as per the manufacturer’s instructions. Using the real-time RT-PCR assays described below, influenza RNA was quantified against standard curves created by serial dilutions of gene-specific influenza A and B RNA transcripts, developed in-house and previously quantified by UV spectrophotometry.

3.2 Virus-spiked samples

Influenza-negative nasopharyngeal swab specimens (NPS) in viral transport medium were pooled and used as sample matrix for epMotion protocol development, and for comparative evaluation of the easyMAG (bioMerieux, Durham, NC) and QIAcube extraction instruments. Quantified influenza A and B viruses were serially (log₁₀) diluted in sample matrix from 2.98 × 10⁷ to 298 RNA gene copies/mL (influenza A) and 1.48 × 10⁷ to 148 gene copies/mL (influenza B). Spiked samples were dispensed into single-use aliquots and frozen at −70°C. Quantitative real-time RT-PCR was performed on RNA extracts in duplicate.

3.3 Clinical specimens

In total, 170 influenza A or B positive respiratory swabs (nasopharyngeal, oropharyngeal, nasal, and throat) were randomly chosen from the 2008–2011 respiratory specimen archives of the WC-LVD. Samples were originally extracted using an easyMAG instrument, confirmed positive with the CDC Human Influenza Virus Real-time RT-PCR Diagnostic Panel,¹⁴ and stored at −70°C. Influenza A subtypes included both A(H3N2) and A(H1N1)pdm09 viruses.

3.4 Real-time RT-PCR

Testing for influenza A and B RNA was performed using laboratory-developed influenza A and B real-time RT-PCR assays as previously described¹⁵, using either Quanta BioSciences’s qScript™ One-Step qRT-PCR (Gaithersburg, MD) kit with a Stratagene Mx3005P thermal cycler (Agilent Technologies, Santa Clara, CA), or an ABI TaqMan® Fast Virus 1-step Master Mix kit (Applied Biosystems, Foster City, CA) with an ABI 7500 Fast Dx real-time PCR instrument.

3.5 TruTip/epMotion RNA purification from virus-spiked samples

Automated TruTip RNA extraction was performed on an Eppendorf epMotion 5070 using a modified baseline protocol as recently described.¹⁶ Briefly, an 8-channel pipetting tool was used with 1mL epMotion (EPM) TruTips containing a small pore (SPT), 2 mm thick TruTip binding matrix (Figure 1A). Off-board lysis was performed on virus-spiked NPS (250uL sample:375uL TruTip lysis buffer) followed by a 10 min. incubation at room temperature. Lysed samples were manually transferred to a 2 mL 96-well deep-well sample plate, and the plate positioned on the epMotion worktable. After manual addition of reagents to bulk reservoirs the automated script was started. The epMotion first added reagents to the appropriate columns on the 96-well extraction plate (Figure 1B). Next, the machine added 375uL of 95% ethanol to the lysed samples and mixed each thoroughly by repeated pipetting. Extraction by the epMotion continued by aspirating and dispensing the lysed samples and wash buffers through the TruTip binding matrix, 8 samples at a time. Purified RNA was then eluted into the extraction plate by cycling 100µL RNase-free Tris-HCl elution buffer through the TruTip binding matrix.

Figure 1.

(A) EPM TruTips on an epMotion 8-channel tool. (B) 96-well plate layout for extraction of 24 samples with the automated epMotion protocol. Lysed clinical samples are manually added to columns 1, 5, and 9. TruTip solutions are added automatically to the remaining wells by the epMotion script.

Eluates were manually transferred to microfuge tubes and analyzed, in duplicate, by quantitative real-time RT-PCR. A total of six to eight replicate Ct values and RNA quantifications were obtained for each concentration. Reproducibility was assessed by calculating standard deviations (SD) and coefficients of variation (CV) from replicate crossing threshold (C_t) values.

Replicate aliquots of each concentration were also processed on easyMAG and QIAcube instruments according to manufacturers’ instructions. 250 µL sample was extracted on the easyMAG; RNA was eluted in 100 µL. On the QIAcube, 250 µL exceeded the maximum sample input; therefore 125 µL corresponding sample was extracted and eluted in 50 µL. The TruTip/epMotion system processed up to 24 samples in one run. Instrument capacities for the corresponding automated easyMAG and QIAcube protocols were 24 and 12 samples, respectively. Extraction methods were also timed for comparison.

3.6 TruTip/epMotion RNA purification from blinded clinical specimens

TruTip/epMotion extraction of original clinical specimens proceeded as above, with additional modifications: 125 µL of specimen was lysed in 500 µL lysis buffer, 2 mm SPT EPM TruTips were replaced with large pore (LPT), 4 mm thick EPM TruTips, the TruTip binding matrix was pre-wet with lysis buffer or 95% ethanol, and RNA was eluted in 50 µL of elution buffer. Extraction using the easyMAG was also modified to 125 uL sample and 50 uL elution. Thus, all three automated extraction methods used identical input (125 uL) and elution (50 uL) volumes. To assess the RNA recovery of the three automated sample preparation systems (TruTip/epMotion, easyMAG, and QIAcube), aliquots from each nasopharyngeal specimen were extracted on all three platforms on the same day, analyzed together on the same PCR instrument with the same master mix preparation, and compared by assessing differences in C_t values.

4. RESULTS

4.1 Initial evaluation and comparison using influenza-spiked nasopharyngeal samples

The total RNA extraction time for 24 samples was approximately 30 minutes for the TruTip/epMotion, 40 minutes for the easyMAG, and 120 minutes with the QIAcube, the latter requiring two 60-minute runs of 12 samples each. The average C_t values obtained from real-time RT-PCR testing of the influenza A and B dilution panel extracts, from all three automated systems, are shown in Table 1. The TruTip/epMotion performance was comparable to the easyMAG and QIAcube, with influenza RNA detected by real-time RT-PCR at all but the lowest dilutions in all three systems. Pearson correlation coefficients between average influenza A and B C_t values were greater than 0.99, over all detected dilutions, for TruTip compared to easyMAG or to QIAcube.

Table 1.

Comparison of efficacy and reproducibility of influenza A and B extraction on three automated systems.

	TruTip/epMotiona			easyMAGa			QIAcubeb
	Input (gc)	Average Ct ± StDev	% CV	Input (gc)	Average Ct ± StDev	% CV	Input (gc)	Average Ct ± StDev	% CV
Influenza A	7.45 × 106	24.77 ± 0.08	0.33	7.45 × 106	24.01 ± 0.08	0.33	3.73 × 106	24.09 ± 0.26	1.08
	7.45 × 105	28.30 ± 0.49	0.49	7.45 × 105	27.22 ± 0.12	0.44	3.73 × 105	27.41 ± 0.30	1.09
	7.45 × 104	31.92 ± 0.38	0.38	7.45 × 104	30.72 ± 0.19	0.62	3.73 × 104	30.66 ± 0.22	0.72
	7.45 × 103	35.12 ± 0.34	0.97	7.45 × 103	34.74 ± 0.44	1.27	3.73 × 103	34.04 ± 0.23	0.68
	7.45 × 102	38.32 ± 1.15c	3.00	7.45 × 102	36.58 ± 1.07	2.93	3.73 × 102	37.68 ± 0.35	0.93
	7.45 × 101	1/8 detected	NC	7.45 × 101	0/8 detected	NC	3.73 × 101	3/8 detected	NC
Influenza B	3.70 × 106	25.01 ± 0.16	0.64	3.70 × 106	23.44 ± 0.06	0.26	1.85 × 106	23.54 ± 0.39	1.66
	3.70 × 105	27.82 ± 0.28	1.01	3.70 × 105	26.79 ± 0.08	0.30	1.85 × 105	26.98 ± 0.29	1.07
	3.70 × 104	31.21 ± 0.15	0.49	3.70 × 104	30.14 ± 0.17	0.17	1.85 × 104	30.04 ± 0.35	1.17
	3.70 × 103	34.68 ± 0.18	0.52	3.70 × 103	33.52 ± 0.10	0.30	1.85 × 103	33.25 ± 0.46	1.38
	3.70 × 102	38.54± 0.78d	2.02	3.70 × 102	36.71 ± 0.59	1.61	1.85 × 102	36.31 ± 0.30	0.83
	3.70 × 101	1/8 detected	NC	3.70 × 101	3/8 detected	NC	1.85 × 101	4/8 detected	NC

^aValues are the average from 3 or 4 extractions for each dilution (input volume = 250 µL, elution volume = 100 µL) analyzed in duplicate by RT-PCR (n = 6 or 8).

^bValues are the average of 4 extractions per dilution (input volume = 125 µL, elution volume = 50 µL) analyzed in duplicate by RT-PCR (n = 8).Thus the calculated amounts used in equivalent PCR assays are equal in all three methods.

^c6 of 8 RT-PCR replicates detected

^d7 of 8 RT-PCR replicates detected

NC=Not calculated

Although the easyMAG and QIAcube extractions resulted in higher overall RNA recovery for both influenza A and B (Figure 2), the absolute difference in recovered RNA between any of the systems was less than 0.5 log₁₀ RNA gene copies, with the single exception of influenza B at 3.17 log₁₀ gene copies, which was 0.60 and 0.72 log₁₀ gene copies less with TruTip compared to easyMAG and QIAcube, respectively.

Figure 2.

Comparative recovery of influenza A and B RNA from spiked NPS samples. Error bars represent standard deviations from 3 or 4 replicate extractions at each sample concentration. Ratios of input/elution volume were the same for all three methods (TruTip and easyMAG=250/100uL, QIAcube=125/50uL). Recovery represents log₁₀ RNA copies/50uL elution volume.

4.2 Blinded respiratory samples

Preliminary experiments with true clinical samples (not shown) indicated that RNA recovery could be enhanced by using a 4 mm LPT EPM TruTip (larger pore size, thicker binding matrix) and by pre-wetting the binding matrix with either lysis buffer or 95% ethanol before the binding step. These modifications were therefore introduced into the automated TruTip/epMotion protocol for comparative performance evaluations influenza-positive, original clinical samples. The net effect on total extraction time was an increase of approximately 3 to 4 minutes, to a total 33 minutes for 24 samples.

Of the 170 clinical samples extracted using TruTip/epMotion and easyMAG, 168 (99%) were positive when nucleic acid from both extraction methods was subsequently tested for influenza A or B using the Quanta qRT-PCR kit. Seventy of the samples were also analyzed using the ABI Fast Virus PCR master mix, and 69 (99%) were positive with both extraction methods. When clinical sample results were grouped according to C_t range obtained from the easyMAG extractions during the comparisons, differences in average C_t values (ΔC_t) ranged from +0.97 to +2.40 cycles (average = 1.89) with the Quanta kit (Table 2). Given that a log₁₀ change in RNA concentration results in a C_t difference of about 3.3 cycles, a ΔC_t of 1.89 represents an approximate 0.5 log₁₀ variation in starting concentration. Similarly, samples analyzed using the ABI master mix ranged from −0.14 to +2.04 ΔCt (average = 1.29), again less than a 0.5 log₁₀ difference in RNA recovery.

Table 2.

TruTip/epMotion and easyMAG extraction comparison from clinical specimens with two commercially-available real-time one-step RT-PCR master mixes.

qScript™ One-Step qRT-PCR Kit					ABI TaqMan® Fast Virus 1step RT-PCR Master Mix
Number of Samples	easyMAG Ct range	Mean ΔCta	StDev	95% confidence intervalb (CI)	Number of Samples	easyMAG Ct range	Mean ΔCta	StDev	95% confidence intervalb (CI)
4	17.00 – 20.30	1.42	0.92	1.42 ± 0.90	Only 1 sample with Ct < 19.00
30	20.31 – 23.60	1.82	1.02	1.82 ± 0.37	7	19.00 – 23.60	2.04	0.90	2.04 ± 0.67
74	23.61 – 26.90	2.14	0.98	2.14 ± 0.22	19	23.61 – 26.90	1.39	1.04	1.39 ± 0.47
40	26.91 – 30.20	2.40	0.65	2.40 ± 0.20	27	26.91 – 30.20	1.40	1.00	1.40 ± 0.38
12	30.21 – 33.50	1.60	0.86	1.60 ± 0.49	9	30.21 – 33.50	1.04	0.47	1.04 ± 0.31
5	33.51 – 36.80	1.57	1.23	1.57 ± 1.08	3	33.51 – 36.80	0.47	0.58	0.47 ± 0.66
3	36.81 – 40.10	0.97	1.62	0.97 ± 1.83	4	36.81 – 40.10	−0.14	2.05	−0.14 ± 2.01
N = 168	Average =	1.89	1.03	1.89 ± 0.16	N = 69	Average =	1.29	1.09	1.29 ± 0.26

^aMean ΔC_t calculated as the average TruTip/epMotion C_t value minus the average easyMAG C_t value.

^b95% CI values represent MeanΔC_t ± CI

A similar analysis was performed on 124 samples extracted using TruTip/epMotion and the QIAcube; 123 (99%) were positive with extracts from both methods when tested with the Quanta kit. Of these samples, 32 were also analyzed with the ABI Fast Virus kit, and 31 (97%) were positive. TruTip/epMotion and QIAcube comparisons (Table 3) using the Quanta kit show less than 1 log₁₀ difference in estimated viral titer between samples extracted with the two systems (averageΔCt = 1.91), and less than 0.5 log₁₀ difference after analysis with the ABI chemistry (averageΔCt = 1.24). There were only three discrepant detection results, all in clinical specimens with very low viral titers (C_t values ranging from 38.9 to 42.6).

Table 3.

TruTip/epMotion and QIAcube extraction comparison from clinical specimens with two commercially-available real-time one-step RT-PCR master mixes.

qScript™ One-Step qRT-PCR Kit					ABI TaqMan® Fast Virus 1step RT-PCR Master Mix
Number of Samples	QIAcube Ct range	Mean ΔCta	StDev	95% confidence interval (CI)b	Number of Samples	QIAcube Ct range	Mean ΔCta	StDev	95% confidence interval (CI)b
3	17.00 – 20.30	0.98	0.18	0.98 ± 0.20	No samples with Ct < 20.30
29	20.31 – 23.60	1.84	1.03	1.84 ± 0.37	3	20.31 – 23.60	2.42	1.34	2.42 ± 1.52
54	23.61 – 26.90	1.84	1.02	1.84 ± 0.27	12	23.61 – 26.90	1.19	1.27	1.19 ± 0.72
25	26.91 – 30.20	2.00	0.92	2.00 ± 0.36	8	26.91 – 30.20	0.99	0.92	0.99 ± 0.63
6	30.21 – 33.50	2.30	0.61	2.30 ± 0.48	5	30.21 – 33.50	1.24	0.53	1.24 ± 0.46
3	33.51 – 36.80	2.05	1.07	2.05 ± 1.21	3	33.51 – 37.84	0.95	0.15	0.95 ± 0.17
3	36.81 – 40.10	1.28	0.75	1.28 ± 0.85	No samples with Ct > 37.84
N = 123	Average =	1.91	0.96	1.82 ± 0.17	N = 31	Average =	1.24	1.06	1.24 ± 0.37

^aMean ΔC_t calculated as the average TruTip/epMotion C_t value minus the average QIAcube C_t value.

^b95% CI values represent MeanΔC_t ± CI

Linear regression analysis of average C_t data for all positive specimens and pair-wise comparisons between the TruTip/epMotion and the other extraction systems revealed similar R² values (0.92 or 0.93), regardless of the RT-PCR detection chemistry used (Figure 3).

Figure 3.

Linear regression analysis for C_t values obtained from influenza A and B positive clinical NPS samples.

5. DISCUSSION

The efficacy of manual TruTip extraction for recovery of influenza RNA from respiratory specimens was reported previously.¹² Results from this study demonstrate that an automated TruTip/epMotion system can reproducibly purify influenza RNA from respiratory swabs with an efficacy and efficiency comparable to both the easyMAG and QIAcube instruments. Overall quantitative differences in RNA recovery between the TruTip/epMotion and the other instruments was less than 0.5 log₁₀ gene copies. When using qualitative molecular methods with detection capability across several logs of viral concentrations, this disparity would not correlate to lower influenza RNA detection rates in clinical specimens, except for those containing extremely low viral titers.

C_t variations between TruTip/epMotion and easyMAG or QIAcube instruments may be attributable to inhibition from residual drying solvent or sample impurities in the TruTip eluants since the ABI Fast Virus master mix, which is less sensitive to PCR inhibitors¹⁷ resulted in lower ΔC_t values than the Quanta master mix. In previous experiments, the ABI Fast Virus master mix was found to increase RNA amplification in samples known to contain inhibitors (such as stool) in some instances by more than 1 log₁₀, as compared to the Quanta kit (data not shown). TruTip/epMotion extraction efficiency or RNA purity cannot be excluded as a cause for the observed differences in C_t values among the three extraction systems. Adding wash steps or drying cycles to the automated epMotion program to remove impurities might help decrease the ΔC_t values, but would add to the run time and cost.

It should be noted that influenza RNA extraction performance was only assessed for upper respiratory swabs in this study, and other sample types, such as nasopharyngeal aspirates, bronchoalveolar lavage and sputum, may require a different protocol to obtain comparable efficiency with those of the QIAcube and easyMAG instruments. If so, to achieve desired results, users may be limited to one sample type per run. The epMotion/TruTip system can successfully purify influenza RNA from nasopharyngeal aspirates¹⁶; however its performance using an automated protocol as described here has yet to be evaluated and compared to those of other automated extraction systems.

The bi-directional flow over the binding matrix and the simplicity of the TruTip fluidics enable the method to be performed without high-precision pipettors and leads to fewer instrument requirements, such as linear actuators, syringe pumps, valves and ultimately, lower capital equipment costs. However, since the actual solution aspiration volumes (by the robotics) and solution hold-up volumes (in the binding matrix) will vary and are unknown, especially during the elution step, nucleic acid yield and efficiency calculations may be affected. It is therefore possible that some of the measured differences in ΔC_t values between TruTip/epMotion and easyMAG or QIAcube are also related to the liquid handling system and elution fluidics, rather than differences in extraction efficiency or RNA purity.

Nevertheless, the TruTip/epMotion protocol successfully extracted influenza RNA in 99% of the previously positive nasopharyngeal samples and shows comparative clinical efficacy between the automated TruTip/epMotion procedure and the easyMAG and QIAcube instruments. Of the three discrepant results, only one sample was negative for influenza RNA after extraction with TruTip. However, this sample was also negative with easyMAG, and positive with the QIAcube (C_t = 39.3). TruTip/epMotion purified RNA is also amenable to multiple PCR master mix chemistries, which may perform at different efficiencies or be variably sensitive to co-extracted PCR inhibitors (i.e., see average ΔC_t values in Table 2).

The TruTip/epMotion system extracted 24 clinical samples in 33 minutes, which is significantly faster than either the easyMAG (40 min) or QIAcube (120 min) systems. Setup and sample preparation times for all three methods were similar, as all used off-board lysis and manual transfer of lysed samples to the instruments before extraction.

Based on the data presented here, similar TruTip/epMotion performance would be expected for the extraction of other respiratory RNA viruses from nasopharyngeal swabs, and this experimental framework could be used for further evaluation of TruTip efficacy from other respiratory sample types. This study demonstrates that the TruTip extraction technology can be successfully formatted for automated, influenza RNA extraction from clinical specimens, while maintaining sensitive and reproducible RT-PCR detection results.

Abbreviations

WC-LVD

Wadsworth Center Laboratory of Viral Diseases

pRhMK

primary rheusus monkey kidney

NPS

nasopharyngeal swab

RT-PCR

reverse transcriptase-polymerase chain reaction

CDC

Centers for Disease Control and Prevention

ABI

Applied Biosystems

EPM

epMotion

SPT/LPT

small/large pore tips

SD

standard deviation

CV

coefficients of variation

C_t

crossing threshold