Automated High-Throughput mRNA Selection from Eukaryotic Total RNA

Abstract

We describe a new technology (patent pending) for high-throughput selection of poly(A)⁺ RNA from total RNA. A novel binding solution is used to ensure the efficient and specific binding of mRNA to oligo(dT) magnetic beads with high stringency, virtually eliminating the non-specific binding of ribosomal RNA (rRNA) either to oligo(dT) beads or to the poly(A)⁺ RNA bound to the beads. As quantified by real-time RT-PCR, more than 99% of the rRNA is removed in a single round selection and mRNAs are fully recovered for both highly-expressed (GAPDH) and poorly-expressed (DDPK) genes from a few μg total RNA. The protocol is adaptable to any generic robotic workstation and takes ∼30 minutes to process 96 samples.

Introduction

The presence of a poly(A) tail is an ubiquitous feature of eukaryotic messenger RNA (mRNA), which makes up 1-5% of total cellular RNA. Many techniques have been developed to purify mRNA away from the ribosomal, transfer, and other RNA species which can create problems such as obscuring rare messages in direct and indirect detection assays and high backgrounds in microarray experiments. ¹ This is especially pronounced when random priming is used to create cDNA probes or intermediates. Most of these techniques take advantage of a hybrid selection with the poly(A) tail, using immobilization matrices such as oligo(dT)-cellulose, ² oligo(dT) attached to latex or other plastic beads,^3,4 or biotinylated oligo(dT) and streptavidin beads. These procedures usually utilize the physical separation of the matrix on filter supports or by centrifugation. However, when paramagnetic beads are used, the separation is simply achieved by applying a magnetic field, ⁴ which eliminates the need for filtration or centrifugation, and therefore is very convenient for automation integration.

High quality mRNA selection requires both efficient removal of rRNA and other non-mRNA species, and quantitative recovery of poly(A) RNA. The problems with published procedures, ⁵ no matter what matrix was used, is that they require several separation procedures or ‘rounds’ of isolation. After one round, these procedures routinely have a pattern on standard ethidium bromide-stained agarose gels that is predominantly the 18S and 28S rRNA bands. The multiple selection rounds required to achieve a high level of mRNA enrichment are time consuming and often lead to low recovery of mRNA. On examining this phenomenon, we have found that this rRNA contamination that plagues mRNA enrichment is not exclusively caused by non-specific adsorption to the oligo(dT) matrix, but also by binding to, and co-purifying with, messenger RNA (mRNA). We have developed hybridization reagents and a protocol to minimize these unwanted interactions while still promoting efficient oligo(dT) hybridization and mRNA selection. This process, which uses novel, proprietary components in the hybridization and wash solutions, was initially developed as a larger-scale manual procedure, but it performs extremely well in an automated mini-format, requiring less than 16 μg of total RNA as input per well. With this procedure a single round of selection removes more than 99% of the rRNA and both abundant (e.g. GAPDH) and rare (DDPK) mRNAs are fully recovered even when the total RNA input is only a few micrograms. Although the protocol was developed on the MultiPROBE®II HT EX (Perkin Elmer) and Biomek®2000 (Beckman Coulter) robotic systems, it is adaptable to any generic robotic workstation, and takes only about a half hour to process 96 samples with an 8-channel pipetting tool.

Material and Methods

Poly(A)Purist-MAG-96 automated kit (including plates, oligo(dT) magnetic beads, and reagents) was used and the standard protocols (downloadable from Ambion automation Web site: http://www.ambion.com/techlib/automation) were strictly followed. The Ambion Magnetic Stand-96 was used for bead separation. The Poly(A) Purist technology uses proprietary binding and wash solutions provided with the kit. The input of 0.5-16 μg total RNA in 25μl water was dispensed to each well, followed by addition of 15 μg oligo(dT) magnetic beads in 25 μl 2X Binding Solution. The mixture was heated at 65°C for five minutes, then the plate is moved to a shaking platform at ambient temperature (22-25°C) and oscillated for four minutes. During this stage, mRNA is selectively bound. After pelleting magnetic beads with a magnetic stand, supernatants are removed. The beads with mRNA bound are washed with two separate wash solutions in sequence, pelleting the beads and removing supernatants after each wash. After the second wash, 25 μl hot (-75°C) water is applied to each well, and the mRNA is eluted during a three-minute oscillation step. It takes about 30 minutes for processing 96 samples with an 8-channel liquid handler.

Figure 1.

Schemes showing the principle of Ambion Poly(A)Purist-MAG technology (Panel A) and deck layout of automated template on a MultiPROBE® II HT EX (Perkin Elmer) (Panel B).

Ambion mouse liver total RNA (Cat# 7810) was used for all experiments. RNA amounts were quantitated using the Ribogreen assay ⁵ (Molecular Probes, Eugene, OR), which was performed on a fluorescence microplate reader (SpectraMax, Molecular Devices). RNA samples were separated on glyoxal-denatured 1% agarose gels, and Northern blotted using Ambion's NorthernMax Kit (Cat#1940). The blots were probed for GAPDH and 18S rRNA, using RNA probes. Probes were made using Ambion's MAXIscript Kit (Cat#1312), following manufacturer's recommendations, with templates purchased commercially (Ambion, Inc., cat# 7330 and 7339). Each transcription mixture contained 1 μg of each template, 80μCi of α-³²P UTP (specific activity 800 mCi/mmol), and 500 μM of unlabeled ATP, CTP, and GTP in a 20 μl reaction volume. The 18S reaction also included 500 μM of unlabeled UTP as well, so that the 18S and GAPDH signals were comparable in total RNA (lane 1, Figure 2A). One-step real-time RT-PCR was performed on ABI7700. Primers and Taqman probe sets were custom synthesized at Synthegen, LLC. Parameters for the thermal profile were: 42°C for 15 minutes, 95°C for one minute, then 40 cycles of 15s at 95°C and 45 s at 60°C.

Figure 2.

Efficient removal of ribosomal RNA with Poly(A)Purist-MAG-9 Automated Kit. A. Samples before and after mRNA selection were fractionated on 1% agarose gel using glyoxal as a denaturant and blotted onto a positively-charged nylon membrane. 8 μl of each 25 μl elaute (32%) from each isolation was applied to the gel. This was probed with radioactively labeled 18S and GAPDH RNA probes (top and bottom bands). The 18S bands after selection were almost invisible. Lane 1 contains total RNA before polyA isolation (1.6 μg), lane 2 a non-hybridizing size marker. B. Real-time quantitative RT-PCR of 18S rRNA. The same fractional amounts of sample (20%) before and after mRNA selection were used. A C_t difference of 8.9 cycles corresponds to a (2^8.9 =) 478-fold decrease in substrate between the two samples, indicating 99.8% of the rRNA was removed from the sample.

Results

EFFICIENT REMOVAL OF RRNA

The majority of total RNA is ribosomal RNA (rRNA, <75%), with mRNA making up only 1-5% of total RNA. Any residual rRNA will greatly misrepresent the apparent mRNA yield when quantified by absorbance at 260 nm or assayed fluorimetrically using Ribogreen. ⁶ Northern blotting is a common way of visualizing rRNAs and mRNAs. Nineteen identical isolations were performed on 5 μg apiece of the mouse total RNA and 32% of the eluate was used for Northern Blotting, as well as 1.6 μg (32% of the input amount) of the total RNA run in the first lane of the gel. The blot was probed for both GAPDH and 18S rRNA (Figure 2A). Whereas the GAPDH band was buried in the smear of total RNA before mRNA selection (Figure 2A, lane 1), clear GAPDH bands showed in all samples after mRNA selection and the 18S band was visible only on overexposure of the Northern. Real-time RT-PCR was carried out to further quantify the efficiency of rRNA removal (Figure 2B), again performing the assay on equivalent fractions (20%) of the input and output samples. After mRNA selection, the C_t of rRNA (exemplified by 18S) was lowered by 8.9 cycles, equivalent to 99.8% removal, indicating extremely high efficiency of rRNA removal in one single round selection.

FULL RECOVERY OF mRNA

In many selection assays, high selectivity is achieved through sacrificing the yield. Published mRNA selection protocols routinely require two to three rounds of selections to remove 99% rRNA, and a significant amount of mRNA is lost by the end of the procedure. Real-time RT-PCR was employed to quantify mRNA recovery after the selection, again using equal percentages of the input and output (here, 20% of each). As shown in Figure 33 (top panel), GAPDH's C_t after selection is almost identical to that of the original sample, indicating negligible loss of mRNA. Benefiting from the novel high stringency of the Poly(A)Purist-MAG kit binding and elution conditions, we have achieved not only a high level of rRNA removal in a single round selection, but recovery of a very high percentage of the mRNA, as judged by GAPDH mRNA.

Figure 3.

Non-biased recovery of abundant and rare mRNA with Poly(A)Purist-MAG-96 Automated Kit. 5 μg total RNA is used for the selection, and 1/5 of of both the input total RNA (‘before”) and output poly(A)⁺ (“after”) samples were used for real-time quantitative RT-PCR. The coincidence of the “before” and “after” curves for both GAPDH and DDPK indicate they each were fully recovered.

CONSERVATION OF mRNA PROFILE

There is always a concern that any manipulation of the sample may lead to distortion of the original mRNA profile. This is especially the case when a method used doesn't fully recover mRNAs, therefore most likely has some bias in recovery of mRNAs of low and high abundance. Since we have shown that our procedure obtains full recovery of mRNA, it less likely leads to distortion of the mRNA profile. Again, real-time RT-PCR was used to quantify the recovery of high abundant mRNA (exemplified by GAPDH, Figure 3, top panel) and very low abundant mRNA (exemplified by DDPK, Figure 3, bottom panel). DDPK in the input total RNA sample has a C_t of - nine cycles lower than GADPH (- 500-fold lower, assuming both mRNAs have the same RT and PCR efficiencies). This C_t difference remained the same after the selection process, showing that capture efficiency was identical for both mRNA species.

LINEAR RECOVERY DOWN TO AS LITTLE AS 0.5 μg TOTAL RNA

The efficiency of mRNA selection and recovery using previously published mRNA selection protocols often drops significantly when total RNA input is at very low concentration. To test this factor for our system, we applied a range of total RNA inputs from 0.5 to 16 μg, with multiple replicates at each level. The output from each was quantified using the Ribogreen fluorescence assay. ⁶ As shown in Figure 4A and B, the yield of mRNA is proportional in a linear manner to the total RNA input for the entire range tested, with no drop-off in the low end. The output levels were also verified qualitatively using a Northern blot probed for GAPDH, to ensure no loss of integrity. The mRNA yield is equivalent to 1.3% of total RNA input. Since the quantitative RT-PCR and Northern blotting data indicated virtually complete recovery of mRNA, this provides the precise percentage mRNA present in this total RNA sample.

Figure 4.

Recovery of mRNA from mouse liver total RNA with input varying from 0.5 μg to 16 μg, each performed in 12 replicates. Panel A shows the quantity of mRNA recovered from all samples as determined by the Ribogreen fluorescence assay. The data was averaged for each input level, plotted, and fitted to a line in Panel B.

LOW WELL-TO-WELL CROSS-CONTAMINATION

Well-to-well cross-contamination is a major concern of open-well high-throughput processing. We designed an experiment to test the cross-contamination when the standard protocol is used. As shown in Figure 5A, 24 RNA samples were surrounded by empty wells. After the protocol was completed, samples from all wells (with or without RNA input) were subjected to Ribogreen assay ⁶ (Figure 5B) to look for the presence of RNA in the blank wells. Well-to-well cross contamination is not observable by this assay. For a much more sensitive test to detect cross contamination, real-time RT-PCR was employed to determine the presence of GAPDH, a very abundant mRNA in liver. As shown in Figures 5C, the calculated cross contamination levels are less than 0.1%, probably reflecting the error inherent in the assay used.

Figure 5.

Well-to-well cross contamination of Poly(A)Purist-MAG-96 Automated Kit. 5 μg total RNA was spiked into 24 individual wells as indicated in the 96-well diagram in panel A. The remaining wells were filled with 25 μl water, and processing according to the protocol was performed on all wells. After processing, 5 μl of sample was used for a Ribogreen Assay ⁵ to measure total mass of RNA (Panel B), and 5 μl of sample from all wells were used for real-time quantitative RT-PCR, assaying the levels of abundant GAPDH mRNA (Panel C). The 11-C_t difference indicates (2¹¹ =) 2,048 times less RNA in the empty wells.

Conclusions

The system presented here (sold as the Ambion Poly(A)Purist-MAG-96 Automated Kit) provides a short, robust protocol for the quantitive retrieval of mRNA from total RNA preparations, removing greater than 99% of the rRNA initially present. The robotic protocol is very open and can be adapted to any generic liquid handling system. On most platforms, the entire procedure takes only 30 minutes. It fully recovers mRNA of both abundant and rare mRNAs therefore conserves the original mRNA profile. This performance level is maintained even with as little as 0.5 μg total RNA input, and is applicable to a total RNA input of up to 16 μg, with no drop-off in recovery efficiency at any point in this range. For those researchers who wish to quantify their yields after this procedure, it should be noted that the absolute mass of RNA obtained will be lower than with other procedures, due to the more thorough removal of rRNA.

Acknowledgments

We would like to thank Patricia Powers, Vince Pallotta, and Quoc Hoang, for technical assistance.