RNA-binding protein immunoprecipitation as a tool to investigate plant miRNA processing interference by regulatory proteins of diverse origin

Background Due to the nature of viral RNA genomes, RNA viruses depend on many RNA-binding proteins (RBP) of viral and host origin for replication, dissemination and evasion of host RNA degradation pathways. Some viruses interfere with the microRNA (miRNA) pathway to generate better fitness. The development of an adjusted, reliable and sensitive ribonucleoprotein immunoprecipitation (RIP) assay is needed to study the interaction between RBP of different origin (including viral origin) and miRNA precursors. The method could be further applied to transiently expressed heterologous proteins in different plant species. Results Here we describe a modified RIP assay applied to nuclear epitope-tagged proteins of heterologous origin and transiently expressed in Nicotiana benthamiana. The assay includes a combination of optimized steps as well as the careful selection of control samples and rigorous data analysis. It has proven efficient to detect and quantify miRNA processing intermediates associated with regulatory proteins. Conclusions The RIP method described here provides a reliable tool to study the interaction of RBPs, such as transiently expressed regulatory proteins with lowly represented host RNA, as is the case of miRNA precursors. This modified method was efficiently adjusted to recover nuclear proteins and reduce unspecific background. The purification scheme optimized here for GFP-tagged proteins can be applied to a wide array of RBPs. The subsequent application of next-generation sequencing technologies will permit to sequence and characterize all RNA species bound in vivo by a given RBP. Electronic supplementary material The online version of this article (10.1186/s13007-018-0276-9) contains supplementary material, which is available to authorized users.


Background
MicroRNAs (miRNAs) represent a large family of small RNAs that function as regulators of plant and animal gene expression [1,2]. Plant primary miRNAs transcripts (pri-miRNAs) are synthesized by RNA pol II and have a 5′cap and 3′poly-A tail [3,4]. Pri-miRNAs form hairpinlike structures and are sequentially processed by RNAse III-like proteins, namely DICER-like 1 (DCL1) in Arabidopsis thaliana, to generate miRNA precursors (pre-miRNAs) and, ultimately, the mature miRNA/miRNA* duplex [5,6]. Plant and animal viruses can interfere with miRNA-mediated regulation in the host at transcriptional or post-transcriptional level [7][8][9][10][11][12][13][14]. Post-transcriptional alterations may include miRNA processing, accumulation and activity. Recent studies revealed that some viral proteins (VP) interfere with pre-miRNA nuclear export and processing by DICER, as is the case of adenovirus [15] and the Ophiovirus Citrus psorosis virus (CPsV) [16]. CPsV 24K protein interacts with pre-miR156a and pre-miR171a in Nicotiana benthamiana plants, causing a higher accumulation of unprocessed precursor species and a concomitant downregulation of mature miRNA species. Consequently, target transcript
RNA immunoprecipitation (RIP) is a powerful technique used to detect the association of individual proteins with specific RNA molecules in vivo. This assay has been successfully employed to purify ribonucleoprotein (RNP) complexes from plant tissue extracts, mostly from Arabidopsis [17][18][19][20]. RNA-binding proteins (RBP)-RNA associations can be identified in RNP complexes involved in well-conserved core RNA processes, but also in specialized processes regardless of whether the interaction occurs in a specific subcellular location like the nucleus. RIP relies on a simple bead-or resin-based affinity purification step that takes advantage of a specific interaction between an antibody and its antigen. The RIP strategies currently employed fundamentally differ in capturing the native RBP either from the wild-type or the epitopetagged RBPs from transgenic plants. However, little is known about RIP assays using transiently expressed RBP of heterologous origin such as VP [21]. In the case of miRNA precursors, it is well known that they interact with nuclear RBPs like DCL1 in order to be processed, and that the structural features of these precursors are specifically recognized by processing proteins [22]. In plants, miRNA biogenesis is fast and intermediates are in many cases hardly detected [23]. Therefore, methodologies aimed at detecting specific interactions between RBP of heterologous origin and lowly represented miRNAprocessing intermediates need to be optimized.
One of the first steps to be adjusted in a RIP assay is tissue preparation, including plant growth, chemical fixation and extraction conditions. Many RBPs are partitioned within the cell. In VP affecting miRNA processing, RBPs are supposed to be located in the nucleus, normally included in the specialized D-bodies [24][25][26]. Thus, efficient extraction of nuclear and membrane-associated proteins through specific buffer composition should be performed. The selection of a high-affinity antibody for purification of the targeted RNP is another relevant condition to take into consideration. RIP has been successfully implemented in plants using epitope-tagged proteins expressed in transgenic plants and the corresponding commercially available monoclonal antibodies [17,19,27]. In transiently expressed heterologous epitope-tagged proteins, RBP level of expression should be high and must be checked. RIP experiments greatly depend on the differentiation between in vivo specific interactions and irrelevant interactions. For instance, an important source of background arises from unspecific interactions of RNPs with the antibody or affinity matrix and therefore RNA binding to co-precipitated proteins leading to false positives. Such background can be reduced by previous clarification of the lysate with a mock preparation or pre-incubating the matrix with an unspecific protein (such as BSA), and also by using stringent washing conditions. Therefore, suitable negative controls are essential. Negative controls for tagging approaches must include tissue which does not express the tagged-RNP under study and/or a tagged-unspecific protein fused to the amino acid sequence required for the interaction with the matrix. Finally, the last step of adjustment in a RIP assay is monitoring the associated candidate target RNAs with the immunoprecipitated RBP. The most sensitive approach is the use of real-time PCR (RT-qPCR) with specific primers. Careful controls are also required in this step since abundant RNAs will inevitably contaminate the affinity preparation. Quantitative data analysis should also be correctly performed to assess fold enrichment of the target RNAs between immunoprecipitated (IP) and Input fractions.
Here we describe a sensitive RIP method modified from Köster and Staiger [19] and applied to nuclear epitope-tagged proteins of heterologous origin transiently expressed in N. benthamiana. Moderate levels of transiently expressed VP have shown to accumulate in N. benthamiana leaves, as compared with the high accumulation normally detected in natural infection. The present method provides a feasible and promising tool to be applied in experiments with miRNA processing intermediates.

Modified RIP assay
CPsV 24K and 54K proteins are the viral suppressors of RNA silencing (VSR) and have been shown to present affinity for long synthetic double-stranded RNA molecules [28]. These findings, together with the nuclear localization of both proteins, allow us to propose a role for these proteins in the regulation of pre-miRNA processing and miRNA activity [16]. GFP-fused versions of these two viral proteins (VP) were transiently expressed in N. benthamiana plants and the RIP method was adjusted to study the differential accumulation of unprocessed miRNA precursors in tissue expressing VP. Nonfused GFP and RFP were included as controls (Fig. 1).
The RIP method presented here includes a combination of optimized steps (Fig. 2), such as adjusted cross-linking conditions, adequate extraction buffer composition for membrane-bond and partially insoluble proteins, use of a high-quality antibody together with modified immunoprecipitation and elution steps, and inclusion of a highly sensitive RT-qPCR for miRNA precursor detection. A careful selection of negative controls also contributed to the success of the method (Fig. 1).

Cross-linking
The first step to increase the sensitivity and reliability of RIP results was the optimization of cross-linking conditions. In vivo cross-linking stabilizes transient and weak RNA-protein complexes, thus allowing the application of more stringent washing conditions. These washing steps reduce contaminants and eliminate unspecific RBP-RNA interactions that may form after cell lysis [29,30]. In Arabidopsis, chemical cross-linking through formaldehyde fixation has led to an 800-fold enrichment of small spliceosomal U2 snRNA precipitated by spliceosomal U2B protein compared with unfixed seedlings [17]. We first checked the expression of VP-GFP (54K and 24K) and control proteins (GFP and RFP) by fluorescence visualization under the microscope (Additional file 1: Fig.  S1), and then we performed formaldehyde cross-linking of N. benthamiana leaves. To ensure that the cross-linking agent penetrated the plant tissue efficiently and displaced gas at the intercellular space, vacuum was applied and released for short periods. Pressure and time intervals were also adjusted.

Extract preparation
The use of rapidly frozen tissue with liquid nitrogen for the preparation of whole cell plant extracts was advantageous for the recovery of intact, uncontaminated RNPs [31]. For instance, RNPs extracted from pulverized frozen leaf tissue were used for the immunoprecipitation of This outline represents the RIP-qPCR method as described in the text maize chloroplast polysomes translating specific proteins [32] and for the affinity purification of cytosolic mRNAs in Arabidopsis associated with ribosomes [27]. We then quickly harvested and froze N. benthamiana leaves transiently expressing VP (54K or 24K) and controls in liquid nitrogen.
The interpretation of RIP data can be helped by an understanding of the spatial distribution of the target RBP. For the case of the CPsV 54K and 24K proteins studied here, nuclear localization has already been described [16,33]. The chosen buffer needed to stabilize different RNPs, particularly nuclear-bond and membraneincluded proteins for immunoprecipitation, is likely to differ. Since 24K protein is difficult to solubilize, special care must be taken in buffer composition to effectively recover it. This step was optimized by the addition of mild detergents to the lysis buffer.

Antibody description and immunoprecipitation conditions
RBP capturing can be performed by a specific antibody directed against its native form or through the expression of an epitope-tagged version of the RBP. The latter allows the recovery of RBP using highly specific and high-affinity commercially available antibodies for the tag. We used the GFP-TRAP ® system (Chromotek, Germany) that consists of agarose beads containing a covalently linked GFP-binding protein, which is the GFP-recognizing domain of a heavy-chain antibody raised in Camelids [34,35]. The system is extremely stable (up to 70 °C, functional in 2.0 M NaCl or 0.5% SDS) and has a high binding affinity (dissociation constant in the sub-nanomolar range). Additionally, this experimental approach allows the application of the GFP purification scheme optimized here to a wide array of RBPs.
Interactions of RNPs with the antibody or affinity matrix lead to serious background problems. Besides, differentiation of direct from indirect RNA-protein interactions is hindered by RNA binding to co-precipitated proteins. Unspecific binding can be reduced by pre-blocking the beads with a non-cognate protein. For this reason, we included a pre-incubation step of TRAP ® beads with BSA. Assays performed without pre-incubation with BSA, showed unspecific interaction of GFP to different RNAs (data not shown). Another consideration is that detergents in the lysis buffer should not exceed 0.2% to avoid unspecific binding to the matrix. Therefore, we diluted the extracted samples in order to keep a low detergent concentration. Washing conditions after immunoprecipitation should also be established carefully to reduce unspecific binding and to avoid the dissociation of specific RNA-protein interactions and RBP-antibody interaction. We then adjusted stringency by including urea (0.5-3 M) in the washing buffer.

Cross-linking reversal and RNA isolation
One of the main advantages of formaldehyde crosslinking is its reversibility, since it allows further characterization of immunoprecipitates. Commonly, samples consisting of the washed beads bound to the RBP-RNA complexes are incubated at 70 °C for 45 min to reverse cross-linking [26,36]. We optimized this step by lowering the temperature and time of incubation. The new conditions were enough to reverse cross-linking and prevent RNA from prolonged incubation and hydrolysis. RNA was then extracted from these samples using Tri Reagent ® according to the manufacturer's protocol.

Controls
The use of suitable negative controls is essential for the RIP assay [18]. In the present approach, an aliquot of the cellular lysate was taken before immunoprecipitation (Input fraction). This Input sample represented the total RNA employed for RIP and served as positive control for the presence of the transcripts under study. Negative controls for immunoprecipitation experiments typically use a unspecific antibody or start with tissues lacking the bait RBP. For tagging approaches, tissue not expressing the tagged isoform is used [19]. We expressed the non-fused GFP driven by identical regulatory elements as in VP-GFP proteins, and both were immunoprecipitated using GFP-TRAP ® . A comparative analysis of VP-GFP samples or non-fused GFP was then performed by RT-qPCR and VP-GFP was relativized to non-fused samples (see data analysis below). We included RFP immunoprecipitation as an additional negative control of a protein without binding capacity to anti-GFP beads.

Pre-miRNA detection and quantitation
Pre-miRNAs are biogenesis intermediates very relevant for studying variations in miRNA processing, such as those reported in CPsV-infected citrus plants [16]. Northern blot is a gold-standard approach that can detect all sizes, ranging from the long pri-miRNA to the mature form, but limited by its low throughput and low sensitivity. qPCR-based approaches are straightforward for measuring the primary transcript, and they can be adjusted to detect low levels of highly structured miRNA precursors [37,38]. We set conditions to quantify two conserved miRNA precursors (pre-miR156a and premiR171a) from N. benthamiana samples by RT-qPCR and included the well-characterized ubiquitin transcript as internal control [39]. qPCRs were carried out using SYBR-GREEN Master mix (Bio-Rad) and the production of a single PCR product was verified for each primer pair (pre-miR156a, pre-miR171a and ubiquitin) by detecting a single peak in the melting curve.

Data analysis
We calculated fold enrichment of each RIP reaction from qPCR data (see formulas below). We processed four samples: VP 24K and 54K and two controls, nonfused GFP and RFP. We first normalized the Ct value of the four IP RNA to the Input RNA fractions to eliminate possible differences in RNA sample preparation (ΔCt normalized RIP). Presenting RIP signals as fold enrichment of the studied proteins over signals in the nonfused GFP control (unspecific background) would be the most convenient way, accounting also for the specificity of the procedure. Thus, to calculate fold enrichment, the normalized RIP fraction value (ΔCt normalized RIP of 24K, 54K or RFP) was first normalized to unspecific background (ΔCt normalized of non-fused GFP sample), obtaining the ΔΔCt. Finally, the linear conversion of this ΔΔCt rendered fold enrichment, which was plotted for 24K, 54K and RFP (Fig. 3). Fold enrichment of the qPCR internal control, ubiquitin, was also calculated in all samples, and no significant enrichment was observed (Fig. 3).

Protein analysis
Immunoprecipitated proteins were recovered from TriReagent ® organic phase. The expression, size and integrity of the viral and control proteins were confirmed by Western blot in the Input and IP fractions (Additional file 2: Fig. S2). Silver-stained polyacrylamide gels were also run to check immunoprecipitation progress (Additional file 3: Fig. S3).

Materials and reagents Plant growth and infiltration
Nicotiana benthamiana plants were cultivated in pots, under long days conditions (16 h light and 8 h darkness) at 24 °C in growing chambers. They were used to agroinfiltrate when they reached the 4 leaves. Agrobacterium tumefaciens strain GV3101 carrying VP-GFP, non-fused GFP or RFP constructs were used at OD 600 between 0.2 and 0.4. Infiltration was performed by pressing a syringe (no needle) containing the cultures on the underside of the leaf. Samples were collected 4 days after infiltration.

RT-qPCR analysis
For RT-qPCR analysis of pre-miR156a and pre-miR171a, we synthesized first-strand cDNA using MMLV from total DNase-treated RNA. Specific primers [pre156aF 5′-TGACAGAAGAGAGTGAGCAC-3′ pre156aR and 5′-GCTGACAGAAAGAGCAGTGA-3′ for pre-miR156a, and NB.pre171aF 5′-TATTGGTGCGG TTCAATGAGA-3′ and NB.pre171aR 5′-GGCACGGCTC AATCAAAAAG-3′ for pre-miR171a] were used. cDNA was then used as a template for qPCRs. qPCR was performed with an iCycler iQ (Bio-Rad) and SYBR-GREEN Master mix (Bio-Rad) in a reaction held at 95 °C for 10 min, then 44 cycles of 20 s at 95 °C, 30 s at 48 °C and 20 s at 72 °C, followed by melting curve. The presence of a unique product of the expected size was verified on ethidium bromide-stained agarose gels. The absence of contaminant genomic DNA was confirmed in reactions with DNase-treated RNA as the template. N. benthamiana ubiquitin amplification was used to normalize the amount of template cDNA. For this, the primers used were NB.UBQF 5′-ATCCACCCGACCAGCAGAG-3′ and NB.UBQR 5′-TAGAAACCACCACGGAGACG-3′.
Three independent experiments and three biological replicates per experiment were performed. The reproducibility of the assay was monitored by running technical triplicates. Data analysis was performed as described above.

Protein analysis
VP-GFP, GFP and RFP in Input and IP fractions were detected by Western blot with anti-GFP JL-8 monoclonal antibody (BD Biosciences Clontech, USA) and anti-mRFP 3F5 monoclonal antibody, respectively (Chromotek, Germany). Horseradish Peroxidase conjugated anti-mouse (Bio-Rad, USA) was used as secondary antibody. Silver-stained polyacrylamide gels (10%) were run of Input and IP fractions as previously described [40]. Protein sizes were estimated using the ECL ™ Rainbow ™ Marker Full-range protein ladder 12-225 kDa (Amersham ™ ). Different exposure times were taken according to the protein accumulation.