method validation of in vitro rna transcript analysis on the agilent 2100 bioanalyzer

Bruce I. SodowichIrini FadlCraig Burns

Roche Molecular Systems,Branchburg, NJ, USA

Received October 16, 2006Revised December 22, 2006Accepted December 28, 2006

Research Article

Method validation of in vitro RNA transcriptanalysis on the Agilent 2100 Bioanalyzer

The Agilent 2100 Bioanalyzer can characterize in vitro RNA transcripts for their integrity,purity, concentration, and size. The results are comparable to denatured agarose electro-phoresis with ethidium bromide staining and UV spectrophotometry combined. In thisreport, we describe our strategy for validating this method following the InternationalConference on Harmonization guidelines. The assay has a linear range of quantitation be-tween 500 and 25 ng/mL. Quantitation accuracy is within 620% of measurements pro-duced from spectrophotometry and sizing accuracy is within 67% based on theoreticalsizes. Concentration and sizing measurements within a single assay produce RSDs that are,10 and ,2%, respectively, indicating good precision. The method also maintains a toler-able precision when altering operator, day, and reagent kit lot. The RSD for quantitation is�25 and ,2% for sizing. The LOQ and LOD are 15.4 and 5.4 ng/mL based on experi-mentation, producing values similar to those specified by the manufacturer. The Bioanaly-zer can differentiate between the RNA transcript and DNA contamination, and proteincontamination quenches the RNA transcript signal. The effect of the ionic strength of thebuffer on RNA analysis is also examined. Limitations of this method and future applica-tions are discussed.

Keywords:

In vitro RNA transcription / Microchip electrophoresis / Method validationDOI 10.1002/elps.200600673

2368 Electrophoresis 2007, 28, 2368–2378

1 Introduction

In vitro RNA transcripts derived from DNA templates have awide variety of molecular applications including RNAprobes, microinjection studies, in vitro translation, micro-array studies, RNA interference studies, reference standards,and as controls for PCR-based diagnostics [1–13] to namejust a few. In vitro RNA transcripts are characterized usingdenatured agarose gels stained with ethidium bromide toassess integrity [14], and UV spectrophotometery to assesspurity and concentration [15]. However, both methods havelow sensitivity and require a significant amount of RNAmaterial (,0.2–4 mg) for reliable results.

Another way in vitro RNA transcripts may also be ana-lyzed is on the Agilent 2100 Bioanalyzer. This instrument is amicrochip-based CE system which automates and digitizesseparation analysis of nucleic acids or proteins. Using theRNA 6000 Nano assay reagent dye kit for RNA analysis, up to12 RNA samples can be analyzed during a single assay.

Samples migrate through individual microchannels withinthe chip through a gel–dye matrix. The dye intercalates intothe nucleic acid and fluoresces as it passes a detector. Sam-ples run along with an internal marker for normalization ofmigration times. The RNA 6000 Nano ladder is run in aseparate channel as both a sizing and quantitation standard.Results from each sample are displayed as an electro-pherogram, plotting fluorescence against migration time, a“gel-like” format, and a tabular format. The advantage of theBioanalyzer is that it can assess integrity, purity, and quanti-tation on a single platform. In addition, its increased sensi-tivity allows for smaller sample requirements to obtain relia-ble results.

The RNA assays designed for the Bioanalyzer are forassessing total or messenger RNA [16, 17] and not specifi-cally for in vitro RNA transcripts. However, Bioanalyzeranalysis of a library of in vitro RNA transcripts produced fromcDNA intermediates has been reported [18]. Herein, wedemonstrate that the Bioanalyzer can both quantitate andsize uniform in vitro RNA transcripts derived from DNAplasmid templates.

Validation strategies for DNA assays on the Bioanalyzerto both quantitate and size DNA fragments have previouslybeen reported [19–21]. Using similar parameters, we havevalidated our method for in vitro RNA transcripts on theBioanalyzer. Our laboratories operate in a regulated environ-

Correspondence: Irini Fadl, Roche Molecular Systems, 1080Route 202 South, Branchburg, NJ 08876-3733, USAE-mail: [email protected]: 11-908-253-7609

Abbreviations: RE, relative error; nt, nucleotides

© 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com

Electrophoresis 2007, 28, 2368–2378 Nucleic Acids 2369

ment, and require that all methods employed be appro-priately validated. Therefore, we have stringently validatedthis method following the guidelines from the InternationalConference on Harmonization “Q2B Validation of AnalyticalProcedures: Methodology” (Guidance for Industry: Q2BValidation of Analytical Procedures: Methodology, 1996.http://www.fda.gov/cber/gdlns/ichq2bmeth.pdf). In thisreport, we compare the Bioanalyzer method for the analysisof in vitro RNA transcripts to traditional methods. In addi-tion, we report the results of our validation study using thefollowing parameters: Linearity and range, accuracy, preci-sion (repeatability and intermediate precision), LOQ andLOD, specificity, and robustness.

2 Materials and methods

2.1 DNA template purification

Plasmid DNA templates were purified as described byRosenstraus et al. [13]. Escherichia coli cells derived frommaster banks containing plasmid with sequences regulatedby an SP6 or T7 promoter were grown overnight andextracted by the alkaline lysis method. Plasmid was recov-ered with isopropanol and purified through CsCl2 densitygradient centrifugation. Plasmid DNA was linearized withan appropriate restriction endonuclease, purified using phe-nol–chloroform extraction, and precipitated by sodium ace-tate and ethanol.

2.2 In vitro RNA transcript production and

purification

Using the linear plasmid as templates, RNA transcripts weregenerated using an appropriate MegaScript High Yield Tran-scription Kit (Ambion, Houston TX, USA). RNA transcriptsare named after the plasmid from which they are derived. AllRNA transcripts except pTRI-Xef were extracted with phenol–chloroform followed by purification using an Oligo d(T) solidphase; all transcripts are engineered with a poly(A) tail. ThepTRI-Xef transcript was purified using the MegaClear kit(Ambion, Houston, TX, USA). RNA transcripts were thenprecipitated with 3.0 M sodium acetate and ethanol, andresuspended in nuclease-free, non-DEPC-treated water. TheRNA transcripts used are pAW109 (1026 nt), pNAS2 (255 nt),pSDL149 (255 nt), pSDL150 (255 nt), and pSYC52 (351 nt).

In addition to our in-house transcripts, we also preparedand analyzed the pTRI-Xef transcript (1920 nt) from thecontrol template which accompanies all MegaScript in vitroRNA transcript kits (Ambion).

2.3 RNA markers

Millennium and Century RNA markers were purchasedfrom Ambion, Houston, TX, USA. The RNA Century markeris composed of five bands (100, 200, 300 400, and 500 nt).

The Millennium RNA marker is composed of ten bands(500, 1000, 1500, 2000, 2500, 3000, 4000, 5000, 6000, and9000 nt).

2.4 Analysis of RNA transcripts

RNA transcripts and sizing markers were analyzed usingdenaturing agarose electrophoresis with ethidium bromidestaining [14] and quantitated using UV spectrophotometry[15]. RNA transcript analysis on the Bioanalyzer was per-formed following the RNA 6000 Nano assay reagent kit guide(Agilent, Palo Alto, CA, P/N G2938-90030) using the Prokar-yote Total RNA Nano assay. All transcripts were diluted innuclease-free, non-DEPC-treated water except where noted.

3 Results

3.1 Comparison of RNA transcript analysis between

agarose electrophoresis and the Agilent 2100

Bioanalyzer

We examined the performance of the Bioanalyzer method todetect and resolve in vitro RNA transcripts by running five ofour prepared transcripts both on the Bioanalyzer and on adenatured agarose gel stained with ethidium bromide.

Figures 1A and B are a side-by-side comparison of theresults from these two methods of separation. Flanking thetranscripts are the Millennium and Century RNA markers(Ambion, Houston, TX, USA). Both the agarose gel and theBioanalyzer resolve the markers in similar fashion from 100to 2000 nucleotides (nt), and each in vitro transcript migratesalong side these markers in an equivalent fashion.

The electropherograms for each transcript produce asingle peak representing the transcript (Fig. 1C). In addition,the Bioanalyzer provides information about what is mostlikely secondary structure of the larger pAW109 and pTRI-Xef transcripts. A small additional peak running behind theprimary peak in the graph of transcript pAW109 is detectedon the Bioanalyzer but not on the agarose gel. Also, thesmear and lower band appearing on the agarose gel underthe primary band of the pTRI-Xef transcript lane is moredefined on the Bioanalyzer, and is detected as a sloping tracerising to a major peak on the electropherogram (Fig. 1D).The additional patterns produced by these transcripts mostlikely represent persistent secondary structure because theyare reproducible and are distinct from the electropherogramsproduced by the corresponding DNA templates (Fig. 3).

The Bioanalyzer method also provides direct sizing dataon RNA transcripts. Although, the RNA 6000 Nano assaydoes not show the values for the size fragments as Bioanaly-zer DNA assays do [19–21], sizing data can still be acquiredby positioning the cursor on-screen at the apex of a transcriptpeak in an electropherogram (see Fig. 1D). In order to getsimilar data from the agarose gel, a standard curve of themigration distances for each transcript in the markers is


2370 B. I. Sodowich et al. Electrophoresis 2007, 28, 2368–2378

Figure 1. Comparison of RNA transcript analysis between denaturing agarose gel electrophoresis and the Agilent 2100 Bioanalyzer. Sizingstandards and RNA transcripts run on (A) agarose electrophoresis, (B) Agilent 2100 Bioanalyzer. (C) Electropherograms for each sizingstandard and transcript. (D) Screen shot of the pTRI-Xef electropherogram for size determination.



required. Taken together, these results demonstrate that theBioanalyzer can separate RNA transcripts in an equivalentmanner as denaturing agarose electrophoresis, and providesuperior data on integrity, structure, and transcript size.

3.2 Linearity and range

Agilent has published in the RNA 6000 Nano assay reagentkit guide that the quantitative range for RNA concentrationusing the total RNA assay is 25–500 ng/mL. We evaluated thisrange using the pTRI-Xef transcript by serially diluting it to500, 375, 250, 100, 50, and 25 ng/mL. Each concentration wasrun on the Bioanalyzer on a single microchip in duplicate.The assay was performed in triplicate. Figure 2, which plotsthe curve of the measured concentration of each dilutionagainst the expected concentration, produces a slope of1.0191 and an R2 value of 0.999. These results confirm thelinearity of the quantitative range for RNA transcript quanti-tation.

The RNA 6000 Nano ladder, along with the internalmarker, contains seven transcripts with sizes of 25, 200, 500,1000, 2000, 4000, and 6000 nucleotides (nt) in length. Weevaluated the Bioanalyzer method to resolve and size RNAtranscripts within this range by challenging the assay withtwo RNA ladders, the RNA Century (100–500 nt) and Mil-lennium (500–9000 nt) markers, which between the twocontain transcript sizes ranging from 100 to 9000 nt. Eachladder was run in all 12 sample wells of a microchip. The sizeof each transcript peak was determined from each electro-pherogram. The transcripts in the Century marker were allwell resolved (Fig. 1C) and the average size of each transcripthad a relative error (RE) no greater than 14.0% (Table 1).These measurements also showed high precision; no tran-script had a RSD greater than 2.04%.

The results from the Millennium marker were similarfor the transcripts between 500 and 2000 nt in size. At2500 nt and above, however, the Bioanalyzer was unable to

Figure 2. Linear range of the Agilent 2100 Bioanalyzer for RNAtranscript concentration measurements.

Table 1. Transcript marker sizing range of the RNA 6000 Nanoassay on the Agilent 2100 Bioanalyzer

Transcripttarget size(nt)

Observedvalue (nt)n = 12

RE (%) SD(nt)

RSD(%)

RNACenturymarker

100 114 14.0 1 0.87200 198 21.0 2 0.98300 309 3.0 6 2.04400 400 0.0 5 1.12500 469 26.2 5 1.07

RNAMillenniummarker

500 489 22.2 7 1.381000 962 23.8 14 1.461500 1511 0.8 21 1.392000 1943 22.8 32 1.662500 2765 10.6 76 2.753000 3796 26.5 60 1.584000 4387 9.7 75 1.715000 4911 21.8 62 1.266000 5302 211.6 81 1.539000 7017 222.0 45 0.65

resolve the transcript peaks efficiently. The RE for the2500 nt peak increased to 10.6%, and the peaks 3000 nt andabove broadened and merged (Fig. 1C). The RE for the3000 nt was particularly large at 26.5%, but the apex of any ofthese larger transcripts were difficult to differentiate makingtheir sizing difficult. Although outside the range, 9000 ntband was detected and resolved from the others, but had anRE of 222.0%. These results demonstrate that for transcriptsat 3000 nt and above, the Bioanalyzer method is less efficientin resolving several RNA transcripts separated by 1000 ntsteps than agarose electrophoresis.

3.3 Accuracy and repeatability

Next, the accuracy of the Bioanalyzer method to measure theconcentration and size of the RNA transcripts that are pro-duced in-house was verified. Our specification for quantita-tion accuracy is 620% of the concentration as determined byspectrophotometry. According to ICH guidelines, accuracyshould be assessed using a minimum of three replicates atthree concentrations covering the specified range. We seri-ally diluted each transcript to three concentrations within thelinear range and measured each of these concentrationsthree times on the Bioanalyzer on the first nine wells of asingle chip. The total concentration of each measured sam-ple was determined by multiplying the measured concentra-tion with the dilution factor. The sizing accuracy for eachmeasured sample was also assessed using the results fromthese assays.

Table 2 presents the results from each transcript. Each ofthe six transcripts measured produced an average con-centration that was within 620% of its concentration as



Table 2. Accuracy and repeatability of RNA transcript measurements on the RNA 6000 Nano assay, data given for(A) concentration and (B) sizing

(A)

Transcriptname

Target conc.(ng/mL)

Acceptablerange 6 20%(ng/mL)

Average(ng/mL) n = 9

SD (ng/mL) RE(%)

Confidenceinterval ofthe mean at95% (ng/mL)

RSD(%)

pAW109 404 323–485 336 16.18 216.74 324–345 4.81pNAS2 323 259–388 326 29.79 0.84 303–349 9.14pSDL149 417 334–501 371 14.56 210.91 361–383 3.92pSDL150 418 320–502 417 15.16 20.15 405–429 3.63pSYC52 244 195–293 249 23.86 1.89 230–267 9.60pTRI-Xef 500 400–600 495 20.68 20.93 479–511 4.17

(B)

Transcriptname

Target size(nt)

Acceptablerange 615%(nt)

Average(nt) n = 9

SD (nt) RE(%)

Confidenceinterval ofthe mean at95% (nt)

RSD(%)

pAW109 1026 872–1180 1057 15 3.00 1046–1068 1.39pNAS2 255 217–293 233 5 28.63 229–237 1.97pSDL149 255 217–293 240 3 25.80 238–243 1.33pSDL150 255 217–293 237 4 26.93 234–241 1.87pSYC52 351 298–404 358 5 1.90 354–361 1.38pTRI-Xef 1920 1632–2208 1884 18 21.85 1871–1898 0.94

determined by spectrophotometry. The confidence interval ofthe mean was calculated at 95% confidence for each dataset.The interval for each transcript fell within the 620% range ofthe concentration target as determined by spectro-photometry. Since there is no published range for sizingaccuracy for the RNA 6000 Nano assay, we evaluated tran-script sizing measurements using a specification of 615%,the accuracy range published by Agilent for the sizing ofDNA fragments using the Bioanalyzer DNA 7500 and 12000assays (Agilent, Palo Alto, CA, P/N G2941-90104) The size ofeach RNA transcript fell well within this range. These resultsindicate that the Bioanalyzer produces acceptable accuracyfor the concentration and sizing of in vitro RNA transcripts.

We also examined the repeatability of the assay, that is,the precision of the assay between measurements madeduring a single assay. Agilent has published an RSD of 10%between concentration measurements. We evaluated thisclaim using the same datasets, and determined that eachmeasured transcript produced an RSD below 10% (Table 2).Two RNA transcripts, pNAS2 and pSYC52, had precisionsthat approached the 10% limit. Sizing measurements wereparticularly precise; no transcript measurement producedan RSD that exceeded 2%. The results indicate that theBioanalyzer produces measurements with acceptable preci-sion for both the concentration and size of in vitro RNAtranscripts.

3.4 Intermediate precision

The next parameter examined was intermediate precision,also known as ruggedness, of the Bioanalyzer method.Intermediate precision is the effect random events have onthe scatter of the measurements. We reasoned that if a singleassay had an inherent RSD of 10%, then altering three vari-ables would produce an RSD of 30%. Therefore, the methodwas challenged by measuring four transcripts while varyingthe operator, day, and reagent kit. These variables were ran-domized using the experimental matrix shown in Table 3A.Intermediate precision was examined on the concentrationand sizing measurements of the pAW109, pNAS2, pSDL149,and pTRI-Xef RNA transcripts. Each transcript was run intriplicate during a single assay altering the three variables inthe matrix for a total of 27 measurements per transcript.

The results for concentration are shown in Table 3B.Varying the operator, day, and reagent kit lot had a clear effecton the precision of the concentrations measurements asindicated by the larger RSD values. The pSDL149 and pTRI-Xef RNA transcripts, which had RSDs of less than 12%, alsohad the greatest accuracy; the RE remained within 620%specification set in the accuracy experiments. As the RSDincreased, accuracy decreased; transcript pAW109 had anRSD of 13.77% and a RE of 221.12%, just outside the 620%target range. Transcript pNAS2 displayed the largest RSD of



Table 3. Intermediate precision of RNA transcript measurements on the RNA 6000 Nano assay(a)

Day 1 Day 2 Day 3

Operator AReagent Kit L/N GF22RK01 Reagent Kit L/N GH20RK01 Reagent Kit L/N GK12RK01Operator BReagent Kit L/N GH20RK01 Reagent Kit L/N GK12RK01 Reagent Kit L/N GF22RK01Operator CReagent Kit L/N GK12RK01 Reagent Kit L/N GF22RK01 Reagent Kit L/N GH20RK01

(b)

Transcript Number ofmeasurements

Targetconc. (ng/mL)

Average(ng/mL)

RE (%) SD ofconcentration (ng/mL)

RSD (%)

pAW109 27 404 319 221.12 43.89 13.77pSDL149 27 417 464 11.28 14.56 11.72pNAS2 27 323 428 32.30 107.22 25.06pTRI-Xef 27 500 447 210.58 53.48 11.96

(c)

Transcript Number ofmeasurements

Target size (nt) Average (nt) RE (%) SD of size (nt) RSD (%)

pAW109 27 1026 1056 2.88 12 1.17pSDL149 27 255 244 24.37 3 1.20pNAS2 27 255 235 27.70 3 1.45pTRI-Xef 27 1920 1887 21.70 20 1.03

(a) The experimental matrix used for randomizing variables. Data given for (b) concentration and (c) sizing.

25.06%, while its RE was greater than 30%. However, eachtranscript met the RSD requirement of less than 30%, indi-cating that an acceptable precision was achieved. We believethe variability observed in these experiments was primarilydue to the operator technique and less to variability betweenreagent lots, since each reagent kit comes with a declarationof conformity as supplied by the manufacturer.

However, sizing measurements did not appear to beaffected by randomizing operator, day, or reagent kit. Table 3Cindicates that the RSD generated by the sizing measurementsare�2%, regardless of the transcript measured. The RSD andRE for sizing measurements appear similar to those gener-ated from the accuracy and repeatability assays (Table 2).These results suggest that the sizing of RNA transcripts is in-dependent of the technique used to prepare a microchip priorto running the assay and that the Bioanalyzer performsseparation and migration in a highly reproducible fashion.

3.5 LOQ and LOD

The LOQ and LOD are the minimum concentrations of ana-lyte that can be accurately measured or reliably detectedrespectively. We defined these limits for the Bioanalyzermethod using the following equations as described in the

ICH guidelines Q2B (Guidance for Industry: Q2B Validationof Analytical Procedures: Methodology, 1996, http://www.fda.gov/cber/gdlns/ichq2bmeth.pdf):

LOQ = 106(SD/S) (1)

LOD = 3.36(SD/S) (2)

where SD is the standard deviation of the y-intercepts of sev-eral regression lines derived from calibration curves aroundthe expected limits, and S is the average slope of theseregression lines. Agilent has published that the lower end ofthe quantitative range of the RNA 6000 Nano assay is 25 ng/mL. To confirm this LOQ, three assays were performed usingthe pTRI-Xef transcript serially diluted to 50, 25, 10, and5 ng/mL, respectively. Each concentration was run on thechip in triplicate. Table 4 shows the data derived from curvesfrom each assay. The LOQ was determined to be 15.440 ng/mL, which suggests that the limit for accurate concentrationmeasurements of RNA transcripts may be below the pub-lished limit. This result is in close agreement with thequantitation limit published by Ricicová and Palková [17].

The LOD was determined by establishing three calibra-tion curves around the published LOD of 5 ng/mL. The pTRI-Xef transcript was diluted to 25, 5, 1, and 0.2 ng/mL. Each



Table 4. LOQ and LOD of RNA transcripts on the RNA 6000 Nano assay

LOQ 106(SD/S) (ng/mL) LOD 36(SD/S) (ng/mL)

Average of y-intercepts from the calibration curves 20.842 0.045SD of the y-intercepts (SD) 0.230 0.117Average slope of the calibration curves (S) 0.149 0.720Calculation of limit 15.440 5.362

dilution was run on an assay in triplicate. Table 4 shows thedata derived from these curves. The LOD was determined tobe 5.362 ng/mL, which is in agreement with the publishedspecification. These experiments were performed using theoriginal Biosizing software version A.02.12 and the intensityof peak signals was determined visually from printed elec-tropherograms using a straightedge. The later Bioanalyzerexpert software version B.01.02 calculates the signal of peaksautomatically. Using the newer software, it is likely that moreaccurate limits may be calculated.

3.6 Specificity

We examined the specificity of the Bioanalyzer method byassessing the ability of the assay to distinguish the RNA

transcript from other materials that may be present in asample. There are two primary contaminants derived froman in vitro RNA transcript reaction, the DNA template, andthe enzyme mix used in the transcription reaction.

We first examined how a DNA template might affect theelectropherograms and measurements of RNA transcripts.The in vitro RNA transcripts and the respective templates ofpAW109, pSYC52, and pTRI-Xef were run individually andas RNA/DNA mixes with the DNA template at a concentra-tion of either 100 or 200 ng/mL, 100 ng/mL representing theconcentration of DNA template typically used in an in vitroRNA transcript reaction. Figure 3 shows that a DNA templatealone produces an electropherogram that contains twopeaks, one peak running between 39 and 44 s and secondpeak running at 54 s or beyond which appears to split into

Figure 3. Electropherograms of RNA transcripts, DNA templates, and RNA/DNA mixes. The concentration outside the parentheses ismeasured RNA concentration.



two or three sub-peaks. This pattern occurs regardless oftemplate length, the pAW109 and pSYC52 templates areabout 3300 and 3500 bp respectively, and the pTRI-Xeftemplate is about 6000 bp.

When the RNA transcript was spiked with DNA templateit was expected that the electropherogram would appear as acomposite of the RNA and DNA electropherograms, al-though this was not always observed. The pAW109 and pTRi-Xef RNA/DNA mixes appear to display a composite electro-pherogram when the DNA template was spiked at 100 ng/mL. At 200 ng/mL, however, the DNA peaks initially detectedto the far right of the graphs are no longer seen and thetemplate produces a single peak running just after the RNApeak. The pSYC52 RNA/DNA mix produced a single peakrunning after RNA transcript peak at both 100 and 200 ng/mL. In all three analyses, however, the RNA/DNA mix pro-duced an electropherogram that is distinct from the RNAtranscript electropherogram alone, and the observed RNAconcentration is elevated, which allows DNA contaminationin an RNA transcript to be identified.

The reason for the differences in the electrophoretic pat-terns from the different RNA/DNA mixes is unclear, al-though we expect the DNA material to behave differently inan RNA assay. The chemistry of the RNA 6000 Nanoreagents is distinct from that of the various DNA assays. Theelectrophoretic parameters applied to the samples by thesoftware are also different between RNA and DNA assays.RNA samples are heated at 707C for 2 min to reduce sec-ondary structure prior to chip loading, a processing step notused to analyze DNA. Finally, Agilent technical support havesuggested that loading high amounts of DNA into samplewells can lead to clogging of the microchannels leading topoorer separation (personal communication). Whatever theprecise reasons, we have observed that this phenomenon isreproducible.

Next, we examined the effect of contamination from en-zyme mix on the analysis of an RNA transcript. A 250 ng/mLsample of the pTRI-Xef transcript was spiked with transcrip-tion enzyme mix to a final concentration v/v of 10, 0.2, or0.1% of the total sample volume. A 10% spike is the approx-imate concentration of enzyme mix used during an in vitrotranscription reaction, while the lesser concentrations repre-sent a concentration of mix that may remain in an RNAtranscript even after purification. These spiked samplesalong with a pure RNA sample were each run on an assay intriplicate. Figure 4 shows that the trace of the electro-pherogram from each sample appears identical with respectto migration despite different amounts of spiked enzymemix. However, there is an apparent difference between theconcentration measurements of the sample spiked with 10%enzyme mix to that of the pure sample and those spiked with0.2 or 0.1% of the enzyme mix. At 95% confidence (a = 0.05)a two-tailed t-test comparing the pure sample to that of the10% spiked sample demonstrates that the difference be-tween the means was statistically significant (n = 3,p = 0.0071). On the other hand, t-tests comparing the mean

Figure 4. Electropherograms of transcript pTRI-Xef spiked withdifferent concentrations of protein from the MegaScript enzymemix.

of the pure sample with those of the 0.2 and 0.1% spikedsamples were not n = 3, p = 0.3429 and p = 0.6239, respec-tively.

We measured our spiked transcript samples on a spec-trophotometer to determine its A260/A280 ratio to deter-mine if the interference was due to the protein found in theenzyme mix. Pure RNA is demonstrated when this ratio isequal to or greater than 2.0 [15]. Our observations indicatethat even with a 10% enzyme mix spike, an in vitro RNAtranscript still generates a ratio above this limit. Therefore,we cannot directly correlate the presence of protein to thequenching of the RNA signal on the Bioanalyzer. Other



components of the mix such as glycerol or the salt con-centration may also have an effect on the RNA signal. Theexact formulation of the enzyme mix is proprietary. Theseresults indicate that while contamination from the enzymemix cannot be detected by the Bioanalyzer, when the con-tamination is great enough it can interfere or quench thesignal generated by the RNA transcripts.

3.7 Robustness

Robustness is the capacity for an assay to remain unaffectedby small but deliberate variations in the assay parameters.Agilent has published that the maximum buffer strength forreliable results is 10 mM Tris, 0.1 mM EDTA. Therefore therobustness of the method was investigated by varying theamount of Tris and EDTA in which an RNA transcript isdiluted. The pTRI-Xef transcript was diluted to 100 ng/mL, inseveral different concentrations of Tris-EDTA buffers asindicated by the matrix in Table 5A. All buffers were for-mulated to a pH of 8.0 6 0.1. Each sample was run in fourwells over three microchips for a total of 36 measurements.Each sample set was compared against the set measured inwater alone using a two-tailed, t-test at 95% confidence.

Table 5B shows that Tris alone at a concentration of 10 or20 mM quenches the fluorescence signal produced by theRNA transcript. But when in combination with 0.1 mMEDTA, the signal is restored. While EDTA alone at a con-centration of 0.1 or 1 mM does not affect the fluorescencesignal, 1 mM EDTA in combination with either concentrationof Tris does appear to quench the signal. These results suggestthat the buffer strength can affect the RNA signal, and con-firm the specification published by the manufacturer.

The buffer strength had no effect on sizing based onthese experiments. However, we have also run the 0.16–1.77and 0.1–2 kb RNA ladders from Invitrogen on the Bioanaly-zer (Carlsbad, CA, USA) which are formulated in 10 mMHEPES, 2 mM EDTA, pH 7.2. Unlike the ladders fromAmbion, the Bioanalyzer had difficulty resolving the tran-scripts in these ladders and individual peaks were more dif-ficult to discern (data not shown).

4 Discussion

In this report, we have demonstrated that the Agilent 2100Bioanalyzer can characterize in vitro RNA transcripts by pro-ducing data equivalent to traditional agarose electrophoresisand spectrophotometric measurement on a single platform.In addition, we have communicated our results from thesubsequent validation study. The Bioanalyzer method canreliably resolve RNA transcripts between 100 and 2000 nt inlength, and has an acceptable degree of linearity in the rangeof 25–500 ng/mL. The Bioanalyzer produces concentrationmeasurements that are within 620% from the measurementgenerated by UV spectrophotometery, and produces sizingmeasurements that are within 615% from a theoretical size.

It maintains an acceptable degree of precision for con-centration measurements, and an excellent degree of preci-sion for sizing measurements. Even when varying operator,day, and reagent kit lot, precision for both concentration andsizing remains acceptable. The Bioanalyzer method has beenshown to have LOQ and LOD consistent with manu-facturer’s published specifications.

The only instance where traditional agarose electropho-resis appears to have an advantage over the Bioanalyzer is itsability to resolve RNA transcripts that are 3000 nt or larger in1000 nt steps. Larger transcripts are clearly resolved on theagarose gel but are poorly resolved on the Bioanalyzer, whichlead to difficulty in transcript sizing. This issue may not be ofconcern when characterizing a single, large RNA transcript.The electropherogram of the RNA 6000 Nano marker alsohas broad peaks for transcripts above 2000 nt, but are moreeasily resolved due to the larger difference in size (Fig. 1C).When run in sample wells, the 4000 and 6000 nt transcriptsof this ladder produced measurements that had an accuracyand precision similar to that of the RNA Century marker(data not shown).

Our results have demonstrated that the Bioanalyzer issensitive enough to detect small amounts of secondarystructure that would otherwise go undetected on an agarosegel. We have also demonstrated that the Bioanalyzer can dif-ferentiate an RNA transcript from its DNA template. DNase Itreatment of the RNA transcript reaction should always beperformed and is usually sufficient to eliminate the tem-plate. However, if the persistence of secondary structure andthe presence of DNA template are both of concern andrequires differentiation, it may be necessary to run a pre-vious produced in vitro RNA transcript of known purity dur-ing the same assay as a reference.

We have shown that while the presence of contaminationfrom the transcript enzyme mix cannot be detected on theBioanalyzer, it has the ability to quench the fluorescent signalwhen it represents at least 10% of the total sample volume.However, purification of RNA transcripts with either a com-bination of phenol extraction and solid-phase purification[22–24], or using the Ambion MegaClear transcription clean-up kit [25, 26] as we have done, provides in vitro RNA tran-scripts with a high degree of purity. We have not examinedthe effects of phenol contamination on RNA measurementon the Bioanalyzer, although it has been published elsewherethat the Bioanalyzer is entirely insensitive to phenol (Light-foot, S. 2002. http://www.chem.agilent.com/temp/radF7488/00045106.pdf). If phenol contamination is a con-cern, measurement via UV spectrophotometry using theA260/A270 ratio will absolutely be required [27].

Finally, we verified the maximum published limit bufferstrength of 10 mM Tris, 0.1 mM EDTA for reliable resultsusing the Bioanalyzer method. Even the RNA 6000 Nanoladder is formulated in 0.1 mM EDTA. Therefore, for reliableresults, we advise using sample buffer strength weaker thanthose published by Agilent when performing RNA analysison the Bioanalyzer.



Table 5. Effect of sample buffer strength on RNA transcript con-centration

(A) pTRI-Xef transcript diluted to 100 ng/mL in nine different con-centrations of Tris-EDTA buffers, pH 8.0

1 Nuclease-freewater

2 10 mM Tris0 mM EDTA


4 0 mM Tris0.1 mM EDTA



7 0 mM Tris,1 mM EDTA



(B) Data of the nine different concentrations of Tris-EDTA buffers, pH 8.0

Bufferstrength

N Mean(ng/mL)

SD(ng/mL)

Mean difference(ng/mL)

95% Confidenceinterval of the meandifference (ng/mL)

p-value Significantdifference


4 90 3.97 N/A N/A N/A N/A


4 57 3.38 32.33 26–39 ,0.0001 Yes


4 69 7.44 20.61 10–31 0.0028 Yes


4 89 7.00 0.92 29–11 0.8276 No


4 85 6.94 5.08 25–15 0.2505 No


4 85 3.80 4.76 22–11 0.1335 No


4 84 5.40 5.63 23–14 0.1443 No


4 72 7.71 17.38 7–28 0.0071 Yes


4 70 3.80 19.60 13–26 0.0004 Yes

We required an in vitro RNA transcript reference stand-ard with known characteristics for establishing system suit-ability of our method. System suitability is the checking of asystem before or during the measurement of unknowns toensure the system performance. We required a referencestandard which is produced in a similar fashion to our usualmanufactured RNA transcripts, yet remain distinguishedfrom them. We decided on the pTRI-Xef RNA transcript dueto the convenient availability of its DNA template. The pTRI-Xef transcript we produced was made using the SP6 Mega-Script kit, characterized by traditional denaturing agaroseelectrophoresis and UV spectrophotometry and diluted to aconcentration of 100 ng/mL. This transcript is run in severalwells on a chip along with an uncharacterized transcript. Thewells measuring the pTRI-Xef transcript must produceacceptable electropherograms and have an average con-centration measurement between 80 and 120 ng/mL in orderfor the results from the unknown transcript to be consideredvalid. The results from the reference standard are also usedto monitor the performance of the method over time. The

results from the reference standard are control charted soany changes that may occur to the system over time may beidentified and corrected if necessary.

In this manuscript, we have demonstrated that the anal-ysis of in vitro RNA transcripts on the Agilent 2100 Bioana-lyzer produces results that are both reliable and reproduci-ble. In addition, we have demonstrated this method can beeasily validated. We manufacture RNA molecules as controlsfor molecular diagnostics, but RNA molecules are alsorapidly emerging as potential therapeutic candidates. Forexample, RNA interference is becoming a pharmacologicallyrelevant technology for variety of human diseases such asmacular degeneration, spinal cord injury, Parkinson’s dis-ease, cystic fibrosis, and variety of viral pathogens includingHIV-1 [28, 29]. Silencing MicroRNAs with small oligonu-cleotides may also have therapeutic uses [30]. As these tech-nologies emerge into realized therapies, RNA moleculesmust be manufactured in a regulated environment and vali-dated methods for characterizing them will be required forin-process testing, quality control testing, or process analyti-



cal technologies (PATs). The use of the Agilent 2100 Bioana-lyzer for the characterization of in vitro RNA transcripts andthe validation strategy we have described could be used to fillsuch requirements.

We wish to thank Michael Geier PhD for his assistance withour validation strategy as well as Ganapathy Muthukumar PhDand Robert Kosecki for their review of our manuscript.

5 References

[1] Lindberg, R. L., De Groot, C. J., Certa, U., Ravid., R. et al., J.Neuroimmunol. 2004, 151, 154–167.

[2] Volokhov, D., Chizhikov, V., Chumakov., K., Rasooly, A., J.Clin. Microbiol. 2003, 41, 4071–4080.

[3] Bittcher, R. T., Pellet., N., Delius, H., Niehrs, C., Nat. Cell Biol.2004, 6, 38–44.

[4] Severance, E., Zhang, H., Cruz, Y., Pakhlevaniants, S. et al.,Mol. Pharmacol. 2004, 66, 420–429.

[5] Sampath, P., Mazumder, B., Seshadri, V., Fox, P. L., Mol. Cell.Biol. 2003, 160, 77–87.

[6] Smith, L. L., Coller, H. A., Roberts, J. M., Nat. Cell Biol. 2003,5, 474–479.

[7] Saghizadeh., M., Brown, D. J., Tajbakhsh, J., Chen, Z. et al.,Biomol. Eng. 2003, 20, 97–106.

[8] Liu, C. L., Schreiber, S. L., Bernstein, B. E., BMC Genomics2003, 19.

[9] Kiger, A., Baum, B., Jones, S., Jones., M. et al., J. Biol. 2003,2, 27.

[10] Goto, A., Blandin, S., Royet, J., Reichhart, J. M. et al., NucleicAcids Res. 2003, 31, 6619–6623.

[11] Zhang, H., Emmons, S. W., Genetics 2002, 160, 799–803.

[12] Golubovskaya, V., Kaur, A., Cance, W., Biochim. Biophys.Acta 2004, 1678, 111–125.

[13] Rosenstraus, M., Wang, Z., Chang, S-Y., DeBonville, D. et al.,J. Clin. Microbiol. 1998, 36, 191–197.

[14] Sambrook, J., Russel, D. W., Molecular Cloning: A Labora-tory Manual, 3rd Edn. Cold Spring Harbor Laboratory Press,Cold Spring Harbor, NY 2001, pp. 7.31–7.34.

[15] Wilfinger, W. W., Mackey, K., Chomczynski, P., BioTechni-ques 1997, 22, 474–481.

[16] Randall, R. L., Damron, T. A., Coffin, C. M., Baster, J. D. et al.,Clin. Orthop. Relat. Res 2005, 438, 149–157.

[17] Ricicová, M., Palková, Z., FEMS Yeast Res. 2003, 4, 119–122.

[18] Che, S., Ginsberg, S. D., Lab. Invest. 2004, 84, 131–137.

[19] Nachamkin, I., Panaro, N. J., Li, M., Ung, H. et al., J. Clin.Microbiol. 2001, 39, 754–757.

[20] Panaro, N. J., Ki, Yuen, P. K., Sakzume, T., Fortina, P. et al.,Clin. Chem. 2000, 46, 1851–1853.

[21] Jabasini, M., Zhang, L., Dang, F., Xu., F. et al., Electrophore-sis 2002, 23, 1537–1542.

[22] Wang, E., Miller, L. D., Ohnmacht, G. A., Liu, E. T. et al., Nat.Biotech. 2000, 18, 457–459.

[23] Hakak, Y., Walker, J. R., Li, C., Wong, W. H. et al., PNAS 2000,98, 4746–4751.

[24] Jost, J.-P., Oakeley, E. J., Zhu, B., Benjamin, D. et al., NucleicAcids Res. 2001, 29, 4452–4461.

[25] Daum, L. T., Chambers, J. P., Santiago, J, Hcikman, J. R. etal., Mol. Cell. Probes 2004, 18, 207–209.

[26] Kabos, P., Matundan, H., Zandian, M., Bertolotto, C. et al.,Biochem. Biophys. Res. Commun. 2004, 318, 955–963.

[27] Stulnig, T. M., Amberger, A., BioTechniques 1994. 16, 403–404.

[28] Mahanthappa, N., Pharmocogenomics 2005. 6, 879–883.

[29] Huelsmann, P. M., Rauch, P., Allers, K., John, M. J. et al.,Antiviral Res. 2006. 64, 1–9.

[30] Krutzfeldt, J., Rajewsky, N., Braich, R., Rajeev, K. et al., Na-ture 2005, 438, 685–689.


method validation of in vitro rna transcript analysis on the agilent 2100 bioanalyzer

Documents