In-gel imaging of RNA processing using broccoli reveals optimal aptamer expression strategies

Abstract

RNA aptamers can be expressed in cells to influence and image cellular processes. Aptamer folding is maintained by inserting the aptamers into highly structured RNA scaffolds. Here, we show that commonly used RNA scaffolds exhibit unexpected instability and cleavage in bacterial and mammalian cells. Using an in-gel staining approach for rapid and simple detection of Spinach- or Broccoli-tagged RNAs in cells, we monitored the processing of RNAs tagged with scaffolded aptamers, revealing endonucleolytic cleavage, RNA instability, and poor expression. We reengineered a natural three-way junction structure to generate an alternative scaffold that enables stable aptamer expression in cells. This scaffold was used to create cassettes containing up to four Broccoli units, markedly enhancing the brightness of mammalian cells expressing cassette-tagged RNAs. These experiments describe methods for screening RNA cleavage events in cells and identify cell-compatible scaffolds that enable efficient tagging of RNAs with aptamers for cellular expression.

Figure 1. DFHBI-1T-binding aptamers can be detected in gel alone or in a context of other RNAs

(A) Schematic representation of the in-gel detection of vegetable aptamer-tagged RNAs and their cleavage products. In this approach, Spinach, Spinach2, or Broccoli-tagged RNAs and their cleavage products can be simultaneously detected in complex RNA mixtures. For example, a cellular lysate can be queried for different sized fragments containing a vegetable tag by running the sample on a gel. Use of a denaturing gel can be particularly valuable for accurately determining the size of the RNA fragments. RNAs that contain a vegetable tag can be visualized based on specific binding of DFHBI-1T to the aptamer tag (star and hairpin). Imaging of all the RNA can be subsequently performed by using a nonselective nucleic acid stain. (B) Broccoli and Spinach2 can be detected using either native or denaturing PAGE. In vitro transcribed Broccoli and Spinach2 were subjected to native or denaturing PAGE. Gels were stained first with DFHBI-1T and subsequently with SYBR Gold. (C) Flanking sequences do not prevent Broccoli from refolding in the gel after denaturing PAGE (7M urea). Four RNAs were in vitro transcribed: Broccoli in a context of the aptamer-stabilizing tRNA scaffold (tBroccoli), tBroccoli linked to the 3' terminus of 5S, tBroccoli linked to the 5' terminus of 7SK, and Broccoli between fragments of the β-actin gene. Each of these RNAs is readily detectable after denaturing PAGE using the protocol described in (B).

Figure 2. Sensitivity and linearity of Broccoli-based in-gel RNA detection

(A) To determine DFHBI-1T staining sensitivity we prepared a serial dilution of in vitro transcribed 5S-tBroccoli or 5S-tSpinach2. The RNA was resolved by urea-PAGE followed by staining with DFHBI-1T and SYBR Gold. Both types of staining showed similar sensitivity with the ability to detect as little as 100–200 pg (~1 fmole) of RNA. (B) The gel from (A) was used to plot the calibration curve. Both DFHBI-1T and SYBR Gold show excellent linearity down to 300 pg (R² for both dyes and both aptamers is ~0.998). The bands from the gel on panel were quantified and the resulting values were plotted on a log scale.

Figure 3. The tRNA scaffold triggers RNA cleavage in E. coli

(A) Aptamers with the tRNA scaffold are processed into two forms in E. coli and the ratio of these forms depends on the bacterial strain. We induced the expression of tBroccoli from pBAD E tBroccoli or pET28c-tBroccoli in LMG194, Rosetta and BL21 Star (DE3) strains of E. coli. tBroccoli in all strains shows a major fluorescent form of ~110 nt (filled arrow), which is 50 nt shorter than the expected full-length RNA expressed from the plasmid. However, some strains also revealed the presence of a longer RNA (~160 nt, open arrow) that was consistent with the expected size. (B) tRNA scaffold processing explains presence of two bands in E. coli. tBroccoli is recognized as bacterial tRNA, resulting in the 3' portion of the transcript distal to the tRNA being cleaved off as a part of tRNA maturation process. Thus, the tBroccoli transcript is highly processed, with only levels of the full-length product detectable. (C) Broccoli without the tRNA scaffold does not get cleaved. In this experiment, we expressed either Broccoli or tRNA-Broccoli from the pET28c plasmid in the BL21 Star (DE3) strain. The full-length Broccoli transcript is 98 nt (filled arrowhead). Unlike the tRNA-Broccoli transcript which is highly cleaved, the Broccoli transcript is uncleaved, as evidenced by just one band on the gel. No other smaller products are detected. Broccoli constructs were expressed in BL21 Star (DE3).

Figure 4. The tRNA scaffold induces RNA cleavage and instability in mammalian cells

(A) Processing of tRNA-scaffolded Broccoli in mammalian cells. We expressed tBroccoli and Broccoli linked to 5S rRNA from the 5S promoter, and tBroccoli and Broccoli from the U6 promoter and linked to the U6+27 leader sequence. tBroccoli from both promoters is cleaved, as evidenced by multiple bands on the DFHBI-1T stained gel. Open arrows indicate full-length product and filled arrows indicate cleaved products. (B) Analysis of the band pattern on the gel on the panel (A) suggests a possible series of processing events for both 5S-tBroccoli and U6+27-tBroccoli. Numbers correspond to specific cleavage sites and also to specific bands seen in panel (A). (C) tRNA scaffold promotes faster degradation. To determine if tRNA recognition and processing affects the half-life of expressed RNA we treated HEK293T cells expressing 5S-Broccoli or 5S-tBroccoli with actinomycin D (5 μg/ml). Total RNA from these cells was collected at the indicated time points. Broccoli without tRNA is noticeably more stable. (D) Quantification of the RNA half-life from panel (C). A monoexponential decay curve was fitted onto the data points. The SYBR Gold-revealed 5S rRNA band was used for loading normalization.

Figure 5. 5S-tBroccoli processing in vitro in mammalian nuclear extracts

(A) To further confirm that the processing of tBroccoli is due to cleavage events, we wanted to establish a precursor-product relationship. Thus, we added full-length transcript and observed whether it undergoes cleavage in vitro. In vitro transcribed 5S-tBroccoli was incubated with HEK293 nuclear extracts (Active Motif) for the indicated times. As can be seen, 5S-tBroccoli is cleaved in a time-dependent manner which can be attributed to both specific processing (bands, labeled with 1, 2, 3) and non-specific degradation. Open arrows indicate full-length product and filled arrows indicate cleaved products. (B) Schematic representation of the possible order of events in vitro. In vitro processing, unlike in cells, apparently starts with 3' terminus cleavage with RNase Z, then proceeds via 5' terminus cleavage with RNase P and finally generates a short fragment which can attributed to Broccoli excised as a result of the tRNA splicing process. Steps 1, 2, and 3 correspond to the bands on the gel on panel (A) labeled with the same numbers. RNase P seems to be less active in vitro compared to in cells as the band 2, presumably generated by the cleavage with this enzyme, is very faint.

Figure 6. Stability and cleavage of V5- and F29-scaffolded Broccoli RNAs in cells

(A) Test of V5 and F29 scaffold cleavage in bacteria. V5-Broccoli expression in E. coli resulted in two bands: one corresponding to the full-length transcript and one which reflects cleavage downstream of the scaffold. F29-Broccoli also showed two bands, however the shorter, presumably cleaved fragment, was very faint and the full-length product was the major form of the expressed construct. Open arrows indicate full-length products and filled arrows indicate cleaved products. (B) Expression of V5-Broccoli and F29-Broccoli in HEK293T cells. All constructs were transcribed off the pAVU6+27 plasmid and had the U6+27 leader sequence at the 5' end and the transcription terminator on the 3' end. V5-Broccoli showed one band with a size corresponding to the full-length product. No cleavage products were observed. F29-Broccoli showed two bands. The upper band corresponded to the full-length product while the lower one is potentially the result of a cleavage event. Open arrows indicate full-length products and filled arrows indicate cleaved products. (C) Half-lives of RNAs containing V5 and F29 scaffolds. HEK293T cells expressing aptamers as described in (B) were treated with actinomycin D. Cells were harvested at the indicated time points and total RNA was analyzed. Broccoli fluorescence was revealed with DFHBI-1T staining (Figure S14) and the band intensity of the full-length transcript was quantified. The intensity of the 5S bands stained with SYBR Gold were used for signal normalization. The numbers were plotted and fitted to a monoexponential decay curve to calculate RNA half-life. An unscaffolded Broccoli was used as a control. The scaffolded RNAs were more stable relative to the unscaffolded RNA.

Figure 7. Engineering and validation of the F30 scaffold

(A) Schematic representation of U6+27-F29-Broccoli (in Arm 1) with the terminator at the 3' end. We proposed that the lower molecular weight product observed on the gel is a result of an early transcription termination on the UUUGUU sequence (indicated with red arrow). (B) In-gel staining to identify F29 mutants that do not form the lower molecular weight bands. F29mut1-Broccoli and F29mut2-Broccoli expressed in HEK293T cells showed suppressed formation of either one or the other lower molecular weight band. This allowed us to engineer the final F30 scaffold, which contains mutations found in both F29 variants. F30-Broccoli shows only one major product expressed in cell. Though the total amount of the expressed RNA remained similar in all the lanes, suppressed formation of the unwanted bands lead to higher amount of the desired, full-length transcript. Open arrows indicate full-length products and filled arrows indicate cleaved products. (C) F30 enhances Broccoli fluorescence in cells and allows engineering of brighter fluorescent tags. Flow cytometry analysis of DFHBI-1T-treated HEK293T cells expressing unscaffolded Broccoli, F30-Broccoli, F30-Broccoli(Arm 2), F30-2xBroccoli, F30-2xdBroccoli or 2x(F30-2xBroccoli). mCherry expressed from another plasmid was used for assessing transfection efficiency. Transfected cells were analyzed in the green and red fluorescence channels. F30 substantially improves Broccoli fluorescence signal when compared to unscaffolded Broccoli. Insertion of Broccoli into Arm 2 of F30 also generates bright fluorescent signal, which lead to engineering of F30 with both arms bearing either two Broccoli or two dimeric Broccolis. All the constructs, containing two or more Broccoli units demonstrated significantly increased cellular fluorescent signal.