Custom-synthesized peptides, HTLV-1 Rex, HIV-1 Rev and bacteriophage λN, were purchased from GeneScript, (Piscataway, NJ) (Table 1). RNA aptamers as well as all DNA primers and templates were from IDT (Coralville, IA) (Table 1s of the Appendix). To assure correct folding of RNA aptamers, RNAs were first denatured by heating at 95 °C in buffer containing 50 mM Tris-HCl (pH 8.0), 50 mM KCl for 3 min and then slowly cooled to room temperature. Re-folded RNAs were mixed with increasing concentrations of the peptides in the same buffer and incubated at 30 °C for 15 min. Peptide-RNA aptamer complexes were analyzed by gel-shift assay using 15% polyacrylamige gel electrophoresis in Tris-Borate-EDTA buffer and then stained with ethidium bromide.

The EGFP gene was split between amino acid residues 158 and 159 into two fragments, termed α-EGFP (N-terminal fragment) and β-EGFP (C-terminal fragment), respectively. A multi-step PCR was performed to create a DNA fragment that encodes two fusion proteins, HTLV-1 Rex peptide-α-EGFP and bacteriophage λN peptide-β-EGFP, plus T7 promoter in between (the resulting plasmid and the detailed PCR scheme are shown in Figures 1s and 2s of the Appendix). HTLV-1 Rex peptide was fused to the C-terminus of the α-EGFP via a flexible 10-mer oligopeptide linker (Gly-Ser-Ser-Gly-Ser-Ser-Gly-Ser-Gly-Ser); bacteriophage λN peptide was fused to the N-terminus of the β-EGFP via the same 10-mer linker. The DNA construct was inserted into the pACYCDuet-1 vector (Novagen) between the restriction sites NcoI and AvrII, which placed the insert after the first T7 promoter region and before the T7 terminator region (Figure 1s). All constructs were verified by sequencing. Thus, two fusion proteins were expressed from the two-cystronic pACYCDuet-1 plasmid to ensure similar amounts of both proteins. Expression of both fusion proteins was verified by protein gel electrophoresis.

To clone aptamers, a DNA sequence was designed which encodes two RNA aptamers, one of which binds the λN peptide and the other the HTLV-1 Rex peptide. The aptamer sequences were separated by a dT₁₀ linker, and the restriction sites for XbaI and AvrII were added at the ends. Custom synthesized DNA template was PCR-amplified, inserted between the XbaI and AvrII restriction sites in the pETDuet-1 vector (Novagen) under the control of the T7/LacO promoter, and cloned in E. coli strain BL21 (DE3). As a result, RNA was transcribed from this plasmid as a short untranslated transcript without a ribosome binding site and containing two adjacent aptamers separated by 10Ts. As a negative control, DNA sequence encoding two identical RNA aptamer sequences binding HTLV-1 Rex peptides was synthesized and similarly cloned into pETDuet-1 vector. DNA templates and primers are listed in Tables 1s and 2s of the Appendix.

Two plasmids encoding protein and RNA components of the complementation complexes were co-expressed in E. coli BL21 (DE) 3 (B F⁻ dcm ompT hsdS (r_B⁻ m_B⁻) gal λ (DE3)) (Stratagene). As a negative control, the cells were transformed with a plasmid expressing two identical aptamers. Single colonies of transformed cells were grown first at 37 °C in LB medium supplemented with antibiotics for 3 h. Then, cultures were diluted 300-fold into fresh LB medium containing the inducer isopropyl-β-D-thiogalactopyranoside (IPTG) and grown overnight at room temperature. Optical density of the cultures (OD₆₀₀) was between 0.4 and 0.6 at the time of examination.

Fluorescence measurements were obtained with a Becton-Dickinson FACSCalibur flow cytometer with a 488-nm argon excitation laser and a 515–545 nm emission filter (FL1). 500 μL of the cells induced overnight (OD₆₀₀ = 0.4 to 0.6) were washed once with PBS buffer prior to assaying. In total 100,000 cells have been analyzed.

Induced bacterial cells in culture were immobilized between a cover slip and a thin slab of 0.8% agarose in 1X PBS. Time-lapse fluorescence microscopy was performed at room temperature with a Nikon Eclipse 80i inverted microscope equipped with an epifluorescence system X-Cite 120. Images were taken with exposure times of 150–300 ms using a digital black and white camera (12 bit; 20 mHz) with 100x magnification objective controlled by IPLab v.3.7 software (Scanalytics, Inc). An ND4 filter was used to reduce cell photo-damage. Pseudo-green color was added according to the fluorescence level in the florescent images. Image processing was performed using ImageJ 1.36 B software (Wayne Rasband, NIH).

To detect RNA in live bacterial cells we tagged it with two different RNA aptamer sequences which bind two different viral peptides with high affinity. Each peptide was expressed in the cell as a fusion with one of the two fragments of a split enhanced green fluorescent protein (EGFP) (Figure 1). Interaction of each peptide with its cognate aptamer should bring together the two fragments of split EGFP. High local concentrations of the split fragments of EGFP will result in their re-association and development of the fluorescent signal. This approach is a further elaboration of a PC method used by Rackam and Brown to study RNA/protein interactions [14]. In this study, an MS2 binding motif was artificially introduced into RNA, while the second RNA/protein contact was probed by PC. To do so the authors expressed MS2 protein and RNA-binding proteins (FMRP or IMP1) as fusions with the split fluorescent proteins. In other words in this study one aptamer was introduced into RNA, while the other site was an endogenous RNA site which interaction with the RNA-binding protein was in question [14].

To develop an efficient PC method based on two aptamers/two peptides interactions, several issues should be considered. The affinity of each interacting peptide/aptamer pair should be high enough (in the nanomolar range) and ideally it should be comparable for both pairs. There should be no cross-reactivity between the two peptide/aptamer pairs. Also, there should be no interaction between the RNA-binding peptides in absence of RNA, which could otherwise bring the fragments of a detector protein together and thus increase the non-specific background. The RNA-binding peptides should be of comparable length to avoid distorting the assembly of the protein complementation complex. Finally, the placement of the two RNA aptamers on the target RNA should allow accessibility of the corresponding RNA-binding peptides to the cognate RNA aptamers. This implies that a flexible linker of sufficient length should be placed between the RNA aptamer tags.

It is known that the high-affinity and high-specificity interactions of many RNA-binding proteins with the corresponding RNAs are determined by the peptide sequences which contain arginine-rich motifs (ARMs) [15]. Recent studies aimed to understand the mechanism of ARM peptides interaction with the corresponding RNAs concluded that specific binding is determined by a particular pattern of arginines in the peptide and flexibility of the peptide backbone [16]. However, many ARM peptides display a promiscuous behavior by binding several different RNA targets, although with lesser affinity [17]. Keeping all this in mind, we chose three RNA-binding peptides from viral ARM peptides [18-20], and first tested their cross-reactivity with the corresponding RNAs before employing them in our protein complementation system.

In these in vitro experiments the increasing concentrations of the ARM peptides were combined with RNA aptamers at fixed concentrations. The complexes were then qualitatively analyzed by a non-radioactive gel-shift assay (see Figure 2 for the exemplary gel picture). The results showed that the bacteriophage λN peptide and HTLV-1 Rex peptide bind their corresponding RNA aptamers in the nanomolar concentration range and did not display cross-reactivity with the non-cognate aptamers. At the same time, HIV-1 Rev peptide did show some cross-reactivity with the two non-matched aptamers (Figure 2 and Table 2). Based on these results, we concluded that the bacteriophage λN and HTLV-1 Rex peptides along with their corresponding RNA aptamers can be used in the PC-based RNA detection method.

The C-terminus of the EGFP fragment (1–158 aa) was fused to the N-terminus of the HTLV-1 Rex peptide (16 aa-long) via a flexible linker consisting of serine and glycine residues [21,22]. We used the same GS-rich linkers that we used earlier in our eIF4A-based complementation system [1-3]. Similarly, the N-terminus of the second EGFP fragment (159–238 aa) was fused to the C-terminus of λN peptide (22 aa) via the same polypeptide linker. The vector pETDuet-1 (Novagen) was used for the expression of an untranslated T7-transcript containing two aptamer sequences linked by the T₁₀ sequence. Preliminary experiments with T₅ and T₁₀ linkers did not show substantial difference. Therefore, we used the constructs with T₁₀ linkers through out this study.

E. coli cells expressing the entire complementation complex and appropriate controls were grown overnight at room temperature in the presence of the inducer, isopropyl-ß-D-thiogalactopyranoside (IPTG) for co-expression of proteins and RNA. Fluorescence of these cells was compared with fluorescence of the cells expressing two protein fusions in RNA absence and cells expressing two protein fusions plus an incorrect combination of the two aptamers (two HTLV-1 Rex peptide-binding aptamers linked with the T₁₀-linker).

Our experiments with the complementing fusion proteins containing short viral peptides, as well as the fragments of the split eIF4A protein, revealed that background fluorescence caused by spurious self-assembly of the protein fragments can be modulated by cell growth conditions and IPTG concentration (Figure 3). These two parameters, the temperature and IPTG concentration, have an effect on the split protein concentrations by two different mechanisms. Low IPTG concentrations decrease the overall concentration of split proteins, and thus effectively lower the incidence of their spurious re-assembly. The temperature of overnight culture incubation also has a large effect on cell fluorescence by affecting the concentration of properly folded split proteins. At lower temperature, the amount of properly folded and functional proteins synthesized is higher than at higher temperature. Therefore, at 20 °C a high incidence of spurious re-assembly results in a higher background signal than at 30 °C (∼100–200 a. u. versus 20–40 a. u.), in the system based on the eIF4A protein (Figure 3A). It should be emphasized that this background is two orders of magnitude lower than in the cells expressing native EGFP (Figure 3). The cells expressing fusions of the split EGFP with the viral peptides usually displayed higher background than the cells expressing split eIF4A fusion proteins (compare Figures 3A and 3B). This can probably be explained by the positive charges of these arginine-rich peptides that non-specifically bind to the negatively charged molecules or surfaces. To overcome this non-specific background in the cells expressing fusion proteins with the viral peptides we changed the second parameter affecting signal-to-background ratio: we varied concentrations of the inducer, IPTG.

An example of signal-to-background optimization experiments with the protein complementation system using viral peptides is shown in Figure 4. At 25 °C E. coli cells expressing two fusion proteins induced with 1 mM IPTG displayed high fluorescence in the absence of RNA target, and there was also no difference in fluorescence distribution in the cells expressing correct or incorrect aptamer sequences (Figure 4). Decreasing the concentration of IPTG 10-fold resulted in separation of the fluorescence distributions for the cells expressing only fusion proteins and those expressing fusion proteins and RNA aptamers (Figure 4). By decreasing the concentration of IPTG to 0.01 mM it was possible to resolve the cognate aptamer-dependent fluorescence from that of the targets with incorrect aptamer sequences (Figure 4). Under optimized conditions, the average fluorescence of the cells expressing the entire complementation complex exceeded background fluorescence (no RNA component) 10–15 fold, and cells with correct RNA-tagging aptamer sequences displayed 4–5 times higher fluorescence than cells with the non-matched RNA tags (Figures 4 and 5).

Bacterial cells expressing a short untranslated RNA tagged with the two aptamers at optimized conditions were analyzed using fluorescent microscopy. In most cells bright fluorescent spots were seen at the cell poles (Figure 5F) similar to the results obtained in the experiments when an untranslated transcript was labeled by PC triggered by eIF4A-aptamer interactions [1] (Figure 5E).