Visualising single molecules of HIV-1 and miRNA nucleic acids

Background The scarcity of certain nucleic acid species and the small size of target sequences such as miRNA, impose a significant barrier to subcellular visualization and present a major challenge to cell biologists. Here, we offer a generic and highly sensitive visualization approach (oligo fluorescent in situ hybridization, O-FISH) that can be used to detect such nucleic acids using a single-oligonucleotide probe of 19–26 nucleotides in length. Results We used O-FISH to visualize miR146a in human and avian cells. Furthermore, we reveal the sensitivity of O-FISH detection by using a HIV-1 model system to show that as little as 1–2 copies of nucleic acids can be detected in a single cell. We were able to discern newly synthesized viral cDNA and, moreover, observed that certain HIV RNA sequences are only transiently available for O-FISH detection. Conclusions Taken together, these results suggest that the O-FISH method can potentially be used for in situ probing of, as few as, 1–2 copies of nucleic acid and, additionally, to visualize small RNA such as miRNA. We further propose that the O-FISH method could be extended to understand viral function by probing newly transcribed viral intermediates; and discern the localisation of nucleic acids of interest. Additionally, interrogating the conformation and structure of a particular nucleic acid in situ might also be possible, based on the accessibility of a target sequence.


Background
Visualising nucleic acids in situ may provide highly significant biological information at a cellular level. Detecting nucleic acid in a single cell routinely employs fluorescence in situ hybridization (FISH). Traditionally, FISH requires the use of single probes labelled with multiple fluorophores [1][2][3][4][5][6] or multiple probes labelled with a single fluorophore [7][8][9] to allow visualization (for review see [10]). Recent advances in the use of rolling circle amplification from padlock probes [11] and branched DNA probes [12] have significantly improved signal to noise ratios as well as sensitivity during FISH detection. However, the requirement for relatively large target sequences makes these approaches unsuitable for visualizing small size RNAs, such as miRNAs. Alternative approaches include molecular beacons [13], MS2-GFP [14], quantum dots [15] or sub-diffraction microscopy, however, have inherent technical and instrumentation constraints, making them impractical for mainstream use to answer biological questions.
To improve the limitations of nucleic acid detection, we modified a commercially available proximity ligation assay (PLA) to detect individual copies of nucleic acids. PLA was originally designed for detecting co-localization of proteins within a 40 nm distance [16]. The intended detection of co-localized proteins via PLA relies on the use of primary antibodies to the proteins of interest and two species-specific secondary antibodies conjugated to short DNA sequences, which can interact with two short DNA oligonucleotides to form a circularized sequence. This sequence is then ligated, amplified via rolling circle DNA polymerization, and the amplified sequences are hybridized with fluorescent oligonucleotide probes, resulting in an approximate two hundred-fold amplification of the original signal.
Here we have modified the PLA technology to visualise nucleic acids in fixed cells. The method incorporates probing target nucleic acid sequences with a modified FISH protocol combined with detection of probe binding with a commercially available PLA based kit (we have termed this method O-FISH). Initially, target-specific oligonucleotides coupled with biotin are hybridised to the gene of interest. Subsequently an anti-biotin primary antibody is used to bind to the biotin labelled probe, and finally the PLA method detects the conjugated target complex to generate an O-FISH signal ( Figure 1). In this study we have used O-FISH to visualize miR146a in both mammalian and avian cells, demonstrating its capacity to detect miRNAs. In addition, we used a HIV-1 model system to illustrate the sensitivity of O-FISH detection, which may reach as little as 1-2 copies of nucleic acids in a single cell. In this model we were able to detect both HIV-1 genomic RNA and newly synthesized viral cDNA allowing visualisation of nucleic acids at various stages of the viral reverse transcription process. Unexpectedly, we also observed that certain HIV RNA sequences are only transiently available for O-FISH detection, implying O-FISH can potentially be used for in situ probing of temporal nucleic acid structures.

Results and discussion
To test the O-FISH technology we used a HIV-1 virus system since the viral genome has a well-defined copy number within each HIV-1 virion. For retroviruses, such as HIV-1, two near identical copies of the RNA genome are packaged into the virion [17], which makes HIV-1 -infected cells excellent reference samples to assess the sensitivity and specificity of this novel nucleic acid detection system. A biotinylated 26-nucleotide probe targeting the HIV-1 pol region (Pol, see Figure 2 for relative genome position) was used to detect the presence of HIV-1 RNA genomes in HIV-1 infected lymphoid cells. O-FISH signals were detected in the HIV-1 infected cells (Figure 3b), at levels clearly discernible from minimal background signals observed in the mock-infected control ( Figure 3a). The data demonstrates that O-FISH Pol RNA detection is highly specific and is able to distinguish HIV-1 nucleic acids from host cellular RNA. In addition we have shown that fluorescently labelled HIV-1 virions do not co-localise during infection, (Figure 3c-e), therefore the observed O-FISH signals are derived from independent viral RNA genomes ( Figure 3b). These data indicate that the O-FISH procedure may detect as few as 2 copies of nucleic acid.
In order to probe for cDNA viral replicative products in HIV-1 infected cells we used two additional biotinylated Figure 1 Overview of the O-FISH mechanism. Target nucleic acids are initially hybridised with a biotintylated complimentary oligonucleotide probe (step 1). The biotin conjugate is then targeted with an anti-biotin monoclonal antibody (mAb; step 2). The proximal ligation assay (PLA) consisting of a + and -mAb, is then employed to target the specific IgG domain of the biotin-bound mAb (step 3). Oligos conjugated to each of the PLA mAbs are then hybridised to form circularised DNA (step 4) and then rolling circle amplification is used to effectively multiply the target sequence (step 5). Fluorescently labelled oligonucleotide probes are then hybridized with the rolling circle amplified DNA allowing the observation of target with fluorescence microscopy.  HIV-1 replication-cycle. Furthermore, as it is generally accepted that only one copy of viral cDNA is derived from the reverse transcription complex, these data suggest that our O-FISH probes may detect as little as one copy of viral nucleic acid, although, this is somewhat complicated by background levels. To ensure the reproducibility of these observations, a parallel HIV-1 infection was carried out with an alternative lymphoid cell line, Jurkat cells, resulting in similar strong stop cDNA and gag cDNA O-FISH signals (Figure 5c, d). These data further support the reliability and reproducibility of our O-FISH method. To further confirm the reliability of the O-FISH method we designed experiments to test the specificity of the O-FISH protocol. Due to the small size of the target nucleic acid sequences used for detection, the addition of RNase/ DNase enzymatic treatments for background reduction proved impractical due to the difficulty in achieving the complete nucleic acid digestion required to abrogate detection. As an alternative, a reverse transcription inhibitor AZT was used to block synthesis of HIV-1 cDNA and evaluate the specificity of our strong stop cDNA and gag cDNA O-FISH signal. Samples were treated with AZT then analysed at 12 hours post-infection, to allow maximal HIV-1 replication and thus O-FISH signals. As expected, there was a dose-dependent reduction of detected HIV-1  The O-FISH method was also used to examine structural rearrangements of nucleic acid targets. We again used HIV-1 as a model since HIV-1 RNA may undergo structural rearrangements in the early steps of the replication cycle [18][19][20] and thus may limit the O-FISH probe binding to its target due to the initial occupancy of the primer binding site by tRNA and the absence of newly transcribed viral cDNA. Moreover, there is a lack of direct evidence examining the structural rearrangement hypothesis from HIV-1 infected cells [21] thus prompting the use of OFISH to examine structural conformational changes. To examine the HIV-1 structural conformations, a biotinylated 24 nucleotide probe was targeted towards the 5'UTR region of the HIV-1 RNA genome to assess the accessibility of HIV-1 5' U5/primer binding site (PBS) sequence during the early phase of infection (U5/PBS, Figure 2 for the relative genome position). We have found that the 5' R-PBS sequence was transiently available for our 5  RNA genome, the increase in detected 5' U5/PBS sequences between 0 and 2 hours post-infection was an unexpected observation.
As the HIV-1 RNA genome is not replicated in infected cells until later time-points this increase is not due to an increase in the number of RNA genomes in the cell. Instead, we suggest that the increase in detected signals results from either the differential occupancy of primer tRNA on the PBS and/or structural rearrangements of viral RNA genomes during the early steps of HIV infection. RNA rearrangement is likely to be a critical regulator of HIV-1 biology [22] and is therefore an area of much research. The O-FISH method has the potential to provide a means to interrogate the conformational rearrangement of HIV RNA during replication or RNA structural rearrangements in general and thus may provide supporting data for speculated RNA structures.
To validate whether O-FISH can detect small size RNAs that are as small as miRNA, we used a 19-nucleotide O-FISH probe to detect miR146a expression levels in both chicken DF1 and human HeLa cells. MiR146a levels are linked to the control of type 1 interferon [23], and are altered upon stimulation during virus infection [24] or virus mimics such as dsRNA polyinosinic: polycytidylic (pIC) [23]. Upon pIC treatment, significant differential patterns of miR146a distribution were seen in both DF1 (Figure 8c The data shown here not only provide a proof of concept for the O-FISH protocol but also demonstrates specific detection of both the HIV-1 positive sense RNA genome of infecting virions and viral cDNA generated from reverse transcription in the natural target cells of HIV-1. The use of very short oligo probes combined with a novel signal amplification method provide the flexibility for this method to be used to detect very short nucleic acid target sequences in cells. Here we have established that this allows both the detection of specific cDNA products, which are often both short and scarce in nature, during the HIV-1 reverse transcription process and native cellular miRNAs. In addition, we have observed data that suggests this method can also be applied to probing RNA structure and binding events. This method has the potential to be expanded to detection of viral nucleic acids in additional virus models providing an important tool for biologists to unravel complex viral transcriptional processes. Furthermore, the ability of this method to detect short cellular nucleic acid sequences, such as miRNAs, shows that O-FISH can be easily adapted for use outside the field of virology, and may prove to be a useful tool for examining more general processes in cell biology.

Conclusions
In summary, our data show that O-FISH can detect low copy numbers of nucleic acids that are as little as 20nucleotides in length. Additionally, O-FISH provides a new method to identify the subcellular distribution of nucleic acids and miRNAs during biological processes. Furthermore, by taking advantage of certain newly synthesized viral nucleic acids that are unique to infectious viral particles, O-FISH could also be used to discern and

Cell culture
293T cells were maintained in Dulbecco's modified Eagle medium/high modified (with 4500 mg/l dextrose and

Infection of lymphoid cells
Synchronized infections were performed as described previously [26]. MT-2 or Jurkat cells were spinoculated with 300 ng p24 CA, as determined by Vironostika HIV-1 antigen (p24 CA) MicroELISA assay (bioMèrieux, Marcy l'Etoile, France) virus per 1 × 10 6 cells for 2 h at 1,200 × g in either 12 or 24 well plates at a non fusion-permissive temperature (15°C). After spinoculation, cells were washed twice with PBS-to remove unbound virus and incubated with warm media at 37°C, 5% CO 2 to initiate infection.

Removal of extracellular plasma membrane proteins and un-entered virus using pronase
As part of the experiment using the RT inhibitor AZT pronase treatment was used 20 minutes post-infection to remove all proteins from the outside of the cells, thus removing any bound but un-entered virions from the cell surface. Twenty minutes post-spinoculation cells were washed and resuspended in ice-cold Hank's Balanced Salt Solution (HBSS, Invitrogen) containing 2 mg/ml of protease from Streptomyces griseus (pronase E; Sigma-Aldrich) for 10 min on ice. Cells were then washed extensively with HBSS containing 10% FCS then fresh media added and the cells incubated at 37°C.

Reverse-transcription inhibition using Zidovudine (AZT)
The RT inhibitor Zidovudine (AZT) was obtained through the AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH. AZT was added to cell culture supernatants at the point of infection at either 0.1 or 10 μg/ml as separate conditions. AZT was maintained in the cell culture supernatant throughout the infection.

Modulation of miRNA levels via synthetic nucleic acids
Introduction of synthetic nucleic acids to DF1 or HeLa cells were carried out using the transfection reagent Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Briefly, polyinosinic-polycytidylic acid (pIC) was diluted to 10 times the intended final concentration in media, and Lipofectamine was diluted into media at 1 μL / cm 2 of well area. After 5 min the Lipofectamine and pIC were combined 1:1, incubated for 15 min and then added to cell culture as per manufacturer's instructions (Invitrogen).

Quantitative PCR for HIV-1 reverse transcription products
Quantification of HIV-1 reverse transcription products and standardization of cell numbers was performed using quantitative PCR (qPCR). Cells were harvested at various time points post-infection and lysed in PCR lysis buffer containing 10 mM Tris [pH 8.0], 50 mM KCl with 0.5% vol/vol Triton-X100, 0.5% vol/vol NP-40 and 75 mg/ml proteinase K (Roche, Basel, Switzerland). Samples were incubated at 56°C for 2 h before the proteinase K was inactivated by heating to 95°C for 10 min. Samples were then stored at −20°C. Quantitative PCR was

O-FISH detection of HIV-1 nucleic acids and miR146
Cells were fixed with either 4% formaldehyde in PBS-at 4°C overnight (HIV analyses) or in 300 μl methanol for 20 min at 20°C (miR146 analyses) and then washed twice in PBS-. Fixed cells were mounted onto slides using a Cytospin II machine (Shandon, Astmoore, UK) at 67 × g for 5 minutes. The slides were removed and air-dried overnight in the dark. Slides were either processed immediately or stored in sealed bags at −20°C until needed. All reactions were performed as open-droplet reactions with a droplet volume of 15 μl. Cells were rehydrated in PBS-for 5 minutes and permeabilised with 0.2% Triton for 6minutes, then washed twice with PBS-before being treated with 0.005% pepsin (Sigma-Aldrich) in 0.01N HCl for 1 min(pepsin treatment was omitted in the miR146 analyses) and then again washed twice with PBS-. O-FISH probe stocks were then diluted to a final concentration of 500 nM in hybridisation buffer (10 mM TRIS [pH7.4], 600 mM NaCl, 1 mM EDTA, 10 nM DTT, 0.1% SDS, 50% formamide) and added to the cells, which were incubated in a humidity chamber at 37°C for 1 hour. Cells were washed twice with PBS-and blocked in blocking buffer (Olink Bioscience, Uppsala, Sweden) at 37°C for 30minutes. Probe binding was detected via incubation with an anti-biotin primary antibody (Sigma-Aldrich), diluted 1:500 for HIV-1 nucleic acids or 1:1500 for miR146 in antibody diluent (Olink Bioscience) in a humidity chamber at room temperature for 30 minutes. Detection of primary antibody binding was carried out using Duolink II anti-mouse PLA probes and detection kit (Olink bioscience) as per the manufacturer's instructions. After O-FISH detection, the cells were counterstained with Hoechst 33258 (Invitrogen) and then mounted in Fluoromount-G (Electron Microscopy Sciences, Hatfield, PA, USA) or mounted directly using fluromount with DAPI (Olink Bioscience) for miRNA microscopy.

O-FISH Probes used for RNA and cDNA detection
Probes were designed complementary to different parts of the HIV-1 RNA genome, to HIV-1 cDNA synthesised at various stages of the reverse-transcription process, or to miR146. Negative sense probes pol (5' CTG TCA GTT ACA TAT CCT GCT TTT CC 3') and U5/PBS (5' CGG GCG CCA CTG CTA GAG ATT TTC 3') were used to detect positive sense HIV-1 RNA. Positive sense probes gag (5' ATG GGT GCG AGA GCG TCG GTA TTA AG 3') and strong stop (5' TGT GAC TCT GGT AAC TAG AGA TCC CT 3') were used to detect negative sense HIV-1 cDNA. The miRNA probe (5' CCC ATG GAA TTC AGT TCT C) was used to detect miR146a. All probes had a biotin molecule conjugated to their 5' end.

Image acquisition and analysis
For HIV O-FISH analysis, images were acquired using either a DeltaVision-RT (Applied Precision, Issaquah, WA, USA) or Zeiss Axio Observer Z1 (Zeiss) microscope. Images taken on the DeltaVision-RT were acquired in a z-series on a charge-coupled device (CCD) camera (CoolSnap HQ; Photometrics, Tucson, AZ, USA) through either a 60X 1.42 numerical aperture (NA) or a 100X 1.4 NA oil immersion lens. Reference brightfield images were also acquired. Images were deconvolved using softWoRx deconvolution software (Applied Precision) before analysis. All images taken on the Zeiss Axio Observer Z1 were taken in a z-series on a charge-coupled device (CCD) camera (AxioCam MRm Rev. 3, Carl Zeiss, Germany) through a 100X 1.30 NA oil immersion lens. Reference differential interference contrast images (DIC) were also acquired. Images of a minimum of 5 fields were taken per slide for analysis for all experiments. Cell images were analysed by quantification of fluorescent signal using Bitplane Imaris software (Bitplane AG, Zurich, Switzerland). For miRNA analyses, similar procedures were used but the samples were imaged using a Leica (Leica Microsystems, Sydney) SP5 confocal microscope. Fluorescence and DIC images were collected and all images were taken with the same microscope parameters. Quantification of O-FISH signal events was performed using the spots function in Imaris (Bitplane AG, Switzerland). O-FISH signals were designated to be at least 0.5 μm in diameter and intensity was determined on an experimental basis for each set of slides. The total number of O-FISH signals was divided by the number of nuclei for each image. The mean for each sample was then calculated from an average of between 13.4 and 30 cells per time point and condition from at least five randomly acquired images (resulting in data being derived from a total of between 120 and 160 cells per panel in each figure) and is reported as 'signals per cell'.

Statistical analysis
Statistical analysis was performed using GraphPad Prism software (GraphPad Software Inc., San Diego, CA, USA). Paired two-tailed t-tests were used to compare variance between two sets of observations as indicated.