Genomic SELEX to Identify RNA Targets of Plant RNA Binding Proteins

1

Genomic SELEX to identify RNA targets of plant RNA binding proteins

Olga Bannikova and Andrea Barta

Max F. Perutz Laboratories, Medical University of Vienna, Dr. Bohrgasse 9/3, A-1030 Wien, Austria

Address correspondence to: Andrea Barta or Olga Bannikova, Max F. Perutz Laboratories, Medical University of Vienna, Dr. Bohrgasse 9/3, A-1030 Wien, Austria ; email: ; ;

1. Abstract

Systematic evolution of ligands by exponential enrichment (SELEX) is an elegant technique and allows the isolation of RNA and DNA sequences which directly interact with a protein of interest. Genomic SELEX is an expression level independent selection method which is useful when multiple RNA targets are expected. These RNAs might be expressed in different conditions, or are differentially localized, or have diverse expression levels. Therefore, this method allows the determination of RNA sequences within a particular genome which are potential binding partners of a particular protein. At first a DNA library is constructed by random priming of sheared Arabidopsis DNA with a direct and reverse primer and selection of fragments with a desired length of 200-300 nucleotides. The RNA library is constructed by transcribing the DNA fragments with T7 polymerase. Several rounds of selection with a protein of choice yield a highly specific pool of potential RNA targets. This pool is best sequenced by a deep sequencing method such as 454 sequencing technology and sequences of the selected library are analyzed by bioinformatics methods. The result in a two month time experiment is a collection of RNAs which are binding targets for the protein used for selection. These RNAs usually allow the determination of binding motifs.

Keywords: RNA selection, RNA binding proteins,

2. Introduction

Systematic evolution of ligands by exponential enrichment (SELEX) is a combination of a combinatorial chemistry approach and experimental molecular biology techniques allowing the isolation of high affinity binding partners to a given molecular target [1]. The first SELEX experiments used randomized artificial RNA-aptamer libraries from which strong binders for proteins or small molecules were selected. Typically, the initial aptamer library contains around 1015 to 1016 DNA oligonucleotides with a randomized central part and fixed flanking regions [2]. Such a pool can be easily converted to RNA by in-vitro transcription and then, after the selection step, back to DNA via RT-PCR. The selection is based on incubation of the library with the target molecule followed by separation of the unbound fraction from the formed complex which is usually performed on nitrocellulose filter [3, 4]. The nucleic acids of the selected complex are isolated and then amplified. Such cycles are repeated several times to select for high affinity binders for the target molecule. The aim of this selection is the isolation of RNA or DNA oligonucleotides which have the strongest binding affinity to the target of interest. These tight binding partners are often used in diagnostic and therapeutic applications [5].

Several problems can emerge throughout the selection procedure. One of the possible troubles is losses of the bound oligonucleotides or enrichment of unspecific binding targets due specificity selection procedure. To overcome that problem few other methods for the separation step were described [for review see 6]. As the described SELEX method is optimized to select the winners for best binding to a particular molecule, they have their benefits if such artificial aptamers are used in clinical studies. Applications range from inhibition of a particular protein with this aptamers to using them as substitutions for antibodies. However, these selected aptamer sequences might not be present in the genome of the organism as such strong binders might be detrimental for the function of the protein used in the selection. Biological procedures necessitate a certain dynamic equilibrium as proteins have to bind specifically in one situation but often have to be released again in order to continue the biological process. Therefore, using a randomized aptamer library for the SELEX procedure is not optimal for finding natural binding sites for your favorite RNA binding protein, as the best binder might not corresponded to real binders occurring in vivo in the cell.

To overcome some of the problems in the field of RNA –protein interactions, another type of SELEX was developed termed Genomic SELEX. In contrast to a randomized library, only sequences occurring in the genome of a specific organism are used for library construction allowing the search for real-existing DNAs or RNAs from the same organism of interest.

The first DNA library for Genomic SELEX was developed in 1997 by B.S. Singer [7] for E.coli, S.cerevisiae and human genomes. Subsequently, several proteins have been used to find in vivo RNA targets via Genomic SELEX experiments [8, 9]. These experiments showed the ability of Genomic SELEX not only to find known targets for these proteins, but detected many new targets which then were proven in in vivo studies.

The selection procedure does not vary significantly from a randomized aptamer SELEX. The only difference lies in the library construction where short genomic sequences are prepared with a Klenow dependent addition of adapters to the DNA. These adaptors are chosen to permit the construction of an RNA library which now contains most genomic DNA sequences of an organism in form of RNA sequences. This allows now to select for protein binding regions independent of their level of transcription. Unsurprisingly, many problems which have to be dealt with are similar in Genomic SELEX as in randomized aptamer-based selection.

A critical point for this genomic approach is the loss of weak but biologically significant binders during selection. Consequently the stringency of selection conditions especially in the first rounds of SELEX should be decreased which boosts the diversity of RNA targets in comparison to selecting only a few winners. Other interference in such experiments is the possibility of forming secondary structures between the middle part and the adapter sequences in a library which could lead to unspecific selection of targets. An approach to handle this issue was suggested by Wen and Grey [10] which developed a primer-free Genomic SELEX. Finally, genomic SELEX as any other SELEX procedure remains an in vitro technique and results of the selection procedure have to be proven by other in vitro methods like a gel shift assay and by experiments to show binding and activity in vivo (CLIP or ChIP-chip technology) [1].

Here we present our variant of Genomic SELEX which was developed for an Arabidopsis thaliana (as well as an S. pombe) RNA-library where we tried to avoid possible problems inherently residing in SELEX techniques. This protocol can be performed in two months and results in a highly-specific RNA-pool saturated with targets sequences for the protein of interest. These RNAs are suitable for further sequencing and bioinformatic analysis.

3. Protocol

Protocol 1: Genomic library construction

3.1 DNA preparation

3.1.1 Genomic DNA isolation

Genomic DNA was isolated from 2-weeks old seedling of wild-type Arabidopsis thaliana (Columbia) using the DNAeasy plant mini kit (Qiagen) by following the manufactures instructions. The optimal amount for 30 µg of pure genomic DNA is about 2 gram plant tissue grounded in liquid nitrogen. DNA should be in TE (10 mM Tris/HCl pH 8.0, 1 mM EDTA) buffer, its concentration quantified by measuring OD260nm and DNA purity should be checked on a 1% agarose gel. If DNA is still contaminated with RNA an additional RNase treatment is required following the standard protocol.

3.1.2. DNA fragmentation

30 µg of pure genomic DNA was placed in a 13 ml round-bottom falcon tube and fragmented by ultrasound treatment using a Bandelin Sonoplus UW2070 device with a MS73 microtip. DNA was sonicated 8 times with 10 pulses for 10 seconds at 70% power. Sonication should produce fragments from 100 bp to 4 kb in length, which should be checked by agarose gel electrophoresis and compared with unshared control DNA (Figure 1A).Then sheared DNA was precipitated overnight in the presence of 1/10 volume 3 M NaOAc pH 5.4, and 3 volumes of absolute EtOH. The pellet was resuspended in 100 μl of TE buffer.

3.2 Library development

3.2.1 Primers design

Two pairs of primers were used in the protocol. The first pair was needed for the introduction of adaptor sequences to the sheared DNA. The second pair introduced the T7 promoter sequence for creating the RNA library. The adapters should meet several requirements: 1. they should be able to prime at any sequence in genome, using randomized 8 nucleotides. 2. Have fixed sequence for further manipulation. 3. Sequences of the primers should not be present in the genome.

Based on the features above, forward and reversed primers for the Klenow reaction were:

Fran: AGGGGAATTCGGAGCGGGGCAGCNNNNNNNNN

Rran: CGGGATCCTCGGGGCTGGGATGNNNNNNNNN

The second pair of primers should be complementary to the fixed part of the first primer pair. In addition, the forward primer must have a T7 promoter sequence for the in vitro transcription.

Fclcf: CCAAGTAATACGACTCACTATAGGGGAATTCGGAGCGGG

Rclcr: CGGGATCCTCGGGGCTG

3.2.2 Primer labeling

In order to visualize the incorporation of the randomized primer-adaptors to the genome, a part of the adaptors was radioactively labeled at their 5’ end with [γ32P] ATP and T4 polynucleotide kinase using standard protocols. Unincorporated nucleotides were separated by a G-50 column from GEHealthcare following the manufacturer’s instructions.

3.2.3 Introducing adaptor sequences to the genomic DNA by Klenow reaction

The starting material was about 25 µg of sheared and purified DNA (3.1.2.) at a concentration of 1mM. The concentration of the primers should be enough to allow annealing once every 40 nucleotides as described in [8]. To monitor the reaction process, part of the forward and the reverse primer (containing the randomized sequence) were kinased. Then the reaction mixture was split into two tubes to control the introduction of the forward and reverse primer separately. The DNA solution was mixed with 255 μM primer Rran to a final concentration of 12 μM and one tube was supplemented with 2 µM radioactive Rran. Both tubes were incubated for 3 min at 93oC, and then placed on ice and further treated in parallel. After addition of 10x Klenow buffer and deoxyribonucleotides to a final concentration of 1 mM, the reaction was started with 67 U of Klenow exo-minus enzyme (Fermentas) and incubated for 5 min on ice. Incubation was for 25 min at room temperature and then 5 min at 50oC. The reaction was inactivated by adding EDTA (final conc. 15 mM) and heating for 10 min at 75oC. The reaction mixture was cleaned up from low molecular weight substances with YM-30 Millipore columns. At this point the efficiency of incorporation could be monitored in the radioactive sample by a denaturing 8% polyacrylamide gel with 7 M urea (Figure 1B). The same protocol was applied for the forward primer reaction and now the radioactive Fran primer was added to the non-radioactive sample. Note, that the specific activity of the primers was low, so no high incorporation of radioactivity was expected. But, typically a smear of labeled DNA above the radioactive primer should be visible (Figure 1B).

3.2.4. Gel-purification and size selection of DNA fragments

At this point the DNA has to be separated from unincorporated primers. The two samples from the previous steps were combined and fractionated for about 2h at 100V on a preparative 8% denaturing polyacrylamide (7 M urea) gel with labeled size markers.

Then DNA of 100 to 700 bp was extracted from the gel which was divided into small pieces; frozen to -80oC for 15 min, and then eluted with buffer (10 mM Tris-HCl pH=8.0, 2 mM EDTA pH=8.0, 0.3 M NaOAc pH=5.4). The mixture was heated for 5 min at 95oC and left overnight at 25oC shaking (900 rpm). Alternatively shaking could be at 65oC for 3 hours.

Next, the gel mixture was filtrated though a 0.22 µm nitrocellulose filter (Millipore) and precipitated with 2 volume of EtOH for at least 3 hours.

3.2.5. Introduction of the T7 promoter

For the introduction of the T7 promoter sequence the second pair of primers (Fclcf and Rclcr) were used for the PCR reaction. The number of PCR cycles shouldn’t be more then 10 to avoid artificial byproducts. One of the available proof-reading DNA polymerases (we usually use Phusion polymerase from Finnzyme) should be used to prevent mutations. Typically, PCR was performed: 40 s denaturation at 95oC, 40 s annealing at 55oC (-3oC below the Tm of primer), 20 s elongation at 72oC. The PCR reaction was cleaned via phenol/chloroform extraction followed by PCR-clean up kit which has a low cut off to leave small DNA fragments in the library (e.g. Nucleospin extract II (Macherey-Nagel).

3.2.6. Library cleaning and verification

At his point the DNA library was almost ready, but still can contain some undesirable features, such as fragments containing the same primer (forward or reverse) at both ends. To get rid of such products, the library was subjected to an in vitro transcription reaction followed by a reverse transcription and PCR reaction (RT-PCR).

In order to produce a lot of RNA (usually around 50 µg) from a limited amount of DNA, a High Yield Transcription kit (e.g. from Fermentas) is strongly recommended. Transcribed RNA was extracted with phenol/chloroform and precipitated with 2 volumes of EtOH overnight. For the precise quantification of RNA, it was crucial to purify the RNA with any clean-up kit (e.g. MegaClear RNA clean-up kit from Ambion). After that the quality of the RNA was checked by agarose gel electrophoreses.

For reverse transcription of RNA and a one-step RT-PCR kit from Qiagen was used, because it contained two types of reverse transcriptases which allowed reverse transcription of low and high abundant transcripts from the mixture. In case of using this kit, the number of PCR cycles should be kept to 7-9 to decrease formation of unspecific products. After checking the concentration of the DNA library, it could be further amplified using one of the proof-reading polymerases. Now, the DNA library is ready for a selection procedure and can be store at -20oC for at least 6 months or for a longer period at -80oC.

It is important to check the quality and comprehensiveness of the library by choosing 5 - 15 single copy gene primers for amplification of an appropriate size product from the library by standard PCR [7].

Protocol 2: Affinity selection of RNA targets

3.3. In vitro transcription

In each cycle of the SELEX procedure, the same amount of starting DNA from the previous cycle was used. Optimally, about 1 µg of DNA should be utilized as starting material for in vitro transcription. Transcription should be performed as described in section 3.2.6.

3.4. Selection of RNAs bound to the protein of choice

The protein of interest was usually a recombinant protein purified from a prokaryotic or eukaryotic protein expression system. In our case, a recombinant GST-tagged protein (the RRM and Zn-knuckle domain from atCyp59 [11]) expressed in E.coli was used for selection. The protein should be dialyzed into binding buffer with a 1:1000 excess.

There were several important points to take care off in this step. First, at each cycle 10µg clean RNA (without remaining DNA and nucleotides) was used. Second, each RNA-protein complex behaved differently; meaning that in each case information about binding conditions (such as buffer, working pH and binding temperature) should be empirically assembled. If nothing is known about the protein of interest or of a similar protein, PBS buffer (0.135 M NaCl, 27 mM KCl, 8mM Na2HPO4, 2 mM NaH2P04) with 10 mM Mg2+ could be used. Third, for binding and washing siliconized eppendorf tubes were recommended to prevent loss of glutathione beads.

Before performing the binding reaction of the protein to the GST- beads, an aliquot of 40 µl 50% 4B glutathione sepharose (which is sufficient to bind up to 8 µg protein) was blocked with 0.5 ml tRNA 200 µg/ml (Sigma) and incubated 30 min at 4oC by slow rotation and then washed three times with 10 volumes of binding buffer (1xPBS with 10mM MgCl2). In parallel, the protein-RNA binding reaction was performed in solution (100 µl reaction mixture contained 10 µg of RNA in binding buffer, supplemented with protease inhibitor cocktail (Roche) and RNAse inhibitor (Invitrogen)). This mixture was heated for 5 min at 70oC and then left for 10 min at 25oC (refolding process). Then the dialyzed protein was added in a 3:1 molar excess of RNA over protein in the first 3 cycles. Then the stringency was increased and a ratio of 10:1 was used in later cycles. The binding reaction was incubated for 30 min at 4oC (or another temperature appropriate for the protein of interest). Then, the blocked and washed GST-beads were added to the binding reaction and incubated for another 30 min. Then the beads were washed 3 times with 10 volumes of binding buffer and then eluted twice with 100 µl of reduced glutathione elution buffer according to the manufactures instructions (GEHealthcare). When performing the control – anti-GST selection, purified GST-tag protein at the 10:1 molar ratio was used instead of the protein of interest. To release RNA from the RNA-protein complex, 400 µl FES buffer (20 mM citric buffer pH 5.0, 7 M urea, 1mM EDTA pH 8.0) and 400 µl of Phenol pH 6.0 was added and vigorously shaken for 10 min at 900 rpm. Then 200 µl of H2O was added and the mixture extracted with an equal volume of phenol/chloroform/isoamylalcohol : 25/24/1. The RNA was precipitated overnight by adding 40 μg of glycogen, 1/10 volume of 3 M NaOAc pH 5.4, and 2 volumes of EtOH. The precipitate was dissolved in 20 µl water, cleaned from residuals of phenol with Megaclear RNA clean up kit and then the yield of the selected RNA was measured.