What is RNA?
The abbreviation RNA stands for ribonucleic acid. These molecules are indispensable for virtually all processes of life, since they mediate all steps of gene-expression: messenger RNA (mRNA) is transcribed from genes, carrying information out of the nucleus. During transcription in eukaryotes, mRNAs mature, which involves removal of intervening sequences (introns). This process – termed splicing – is mostly mediated by the sn/U RNA containing spliceosomes. Upon export from the nucleus, most mRNAs get translated by the rRNA containing ribosomes – the code is interpreted through transfer RNAs carrying activated amino acids for protein synthesis. Translation and lifetime of particular mRNAs is partly controlled by short RNAs – microRNAs, short interfering RNAs and piRNAs. Not only the aforementioned RNA species, but actually a large fraction of the genome is transcribed – the significance of which is currently actively researched.
What does microscopy have to do with RNA?
Nearly three decades ago, β-actin mRNA molecules were found to be localized nonhomogenously within the cytoplasm of chicken myoblasts and fibroblasts. Soon, many other examples in other model systems– such as in budding yeast, Drosophila and in Xenopus oocytes and in mammalian fibroblasts and neurons – followed. Recently, genome wide studies of RNA localization in Drosophila embryos and oocytes showed that large fraction (up to 80 %) of the expressed transcripts is distributed in a distinctive, non-homogenous pattern within the cytoplasm. Also, despite being tiny and without well-defined organelles, bacteria was demonstrated to express localized mRNA.
Nearly a decade ago, by the introduction of single molecule sensitivity microscopic techniques, it became possible to measure absolute RNA quantities – by counting RNA molecules – with unprecedented cellular or subcellular resolution. These techniques are currently on the verge of allowing "deep transcriptome sequencing" of individual cells on the microscope stage.
Microscopy, of course, not only provided snapshots – positions and numbers – of RNA molecules, but with appropriate imaging methods, profound mechanisms of transcription, RNA localization, translation and decay were uncovered.
How can you visualize RNA?
RNA – similarly to DNA and polypeptides – is a polymeric macromolecule. Its identity is determined by the primary structure or sequence of building blocks, the ribonucleotides (A, C, G and U). Most RNA molecules are single stranded - although they fold into complex secondary and tertiary structures often involving formation of intramolecular duplexes – thus their identity can be probed with a matching complementary sequence. These probes are mostly single stranded natural nucleic acids – ssDNA or RNA – that carry some detectable labels ranging from radioactive isotopes, small, immuno-detectable molecules – haptenes – or fluorescent dyes. During a process termed in situ hybridization (ISH), the probe finds and binds its target RNA. A mismatch-free hybrid is usually more stable energetically than other, aspecific complexes formed by the probe, thus by including appropriate wash steps in the ISH protocol a specific signal is developed. However, due to these obligatory differentiation steps – beside the harsh treatment some ISH protocols require (use of high temperature, organic solvents, high salt etc.) – none of these ISH techniques is compatible with life cell imaging.
Then how do you follow RNA live?
RNA never exists "naked" within cells – it is always in complex with a dynamic set of protein molecules forming ribonucleoprotein particles (RNPs). Many of these protein molecules bind the RNA directly (RBPs). Fusing these RBPs to fluorescent proteins (FPs) is a widely used method to visualize RNA live. Most of these RBPs found in the repertoire of the eukaryotic cell – e.g. Staufen, eIF4AIII, Hrp48A, PABP, etc., have limited target specificity and only part time residents of any given RNP. Some RBPs however, such as the Pumillio Homology Domain (Pum-HD) or CRISPR/Cas9, can be engineered to match a given RNA target. However, to efficiently visualize mRNPs – usually containing a single mRNA molecule – several copies of RBP-FP fusions per particle is necessary. To minimize the number of transgenic constructs per target molecule, the most commonly applied in vivo RNA visualization technique utilizes RBPs orthologous to the host. The first of their kind was the MS2 system, consisting of a tandem array of 6–24 copies of an RNA stem-loop structure specific to the MS2 bacteriophage and the matching MS2 Coat Protein fused to an FP. By introducing the MS2 loop array into the target Ash1 mRNA and sequestering the unbound MCP-FP to the nuclei, the Singer lab could image dynamic Ash1 mRNPs in living budding yeast.
A common drawback of these phage-loop/loop binding coat protein systems is that they rely on transgenic modification of the RNA target – mostly introducing an extra copy in addition to the endogenously present two alleles – and that the array of loops loaded with the coat protein add about 1.5–2 MDa to the size of the RNP of interest. Such changes might influence the dynamics (transcription, transport, translation and decay) of target RBPs and thus, ideally, the obtained results should be validated with other independent, complementary approaches.
Then there must be other methods to follow RNPs live. How do they differ from the previous ones?
In 1996, the Tyagi lab introduced molecular beacons (MBs), pioneering the use of fluorogenic probes. MBs are short oligonucleotide probes with a fluorophore and a quencher pair at their two opposing termini. These termini (usually 6–7 nucleotides) are complementary to each other forming a hairpin structure thus positioning the fluorophore next to the quencher (dark state). Upon hybridization of the target specific middle portion (loop) of the MB, the stem of the hairpin is melted, separating the fluorophore and the quencher, resulting in activation of fluorescence (bright state). The use of MBs – of fluorogenic probes in general – requires no differentiating wash steps and no transgenic modification of the target.
Unfortunately, neither RNA/DNA hybrid nor double stranded RNA are stable in vivo. To solve this issue, the backbone of these probes is chemically modified to hide the forming hybrid from intracellular nucleases. As a positive side effect, these modifications – mostly 2’O methylation of the phosphoribose backbone and/or incorporation of locked nucleic acids – increase the stability of the hybrid allowing for shorter probes.
What does FIT stand for?
Forced Intercalation of Thiazole orange (FIT) is another way to obtain fluorogenic probes. These probes, developed by the Seitz lab, include an intercalating DNA dye, Thiazole orange (TO) as a base surrogate – that is replacing a non-terminal base in an oligonucleotide sequence. This dye is non-fluorescent when in a low viscosity environment, such as in solution (dark state). Upon hybridization of the probe, the hydrogen bonds of base pairs on both sides of TO greatly limit torsional motion of the dye, thus forcing it into an intercalation like state, activating fluorescence (bright state). As a consequence, FIT probes – unlike MBs – are very sensitive to mismatches – especially to those that are in the immediate vicinity of TO.
What do these FIT probes "look like"?
One of the most important properties of a fluorescent reporter is brightness. The emission of TO (quantum yield) greatly depends on its microenvironment; not only on the hybridization state of the probe, but also on the position and the neighbours of the TO nucleoside. In the bright state, typical brightness of a FIT probe hybridized with a complementary ssDNA molecule is around 7–14 ml ⋅ mol–1 ⋅ cm–1, around 20–40 % of that of EGFP and 10–20 % of AlexaFluor 488. However, another key parameter of imaging is contrast, i.e. the ratio of signal to its immediate background. While conventional fluorophores – like EGFP or AlexaFluor 488 – are equally bright independently from being bound to target or in solution, the dark state of TO labelled FIT probes is 8–20-fold dimmer than their bright state (responsiveness). This property makes FIT probes suitable for e.g. perform rapid wash-free fluorescent in situ hybridization (FISH) even within otherwise challenging tissues, such as the Drosophila ovary.
Can you make brighter FIT probes by incorporating more than one TO dye per probe?
Not really – since these probes are short, typically 15–30 nucleotide long (5–10 nm), two or more fluorophores are positioned well within the range of Förster resonance energy transfer (FRET) and thus quench each other, resulting in only 20–40 % increase in brightness. However, we managed to double the brightness (to around 24–28 ml ⋅ mol–1 ⋅ cm–1) by introducing a slightly red shifted and brighter fluorophore, oxazolopyridine (JO) into a TO containing FIT probe. JO, while being an intercalating dye, is fairly bright even in solution; that is it has low responsiveness. In single stranded probes, however, TO efficiently quenches JO by collisions. Hybridisation leads to the bright state. Due to the rigidity of the formed duplex, TO no longer quenches JO. Rather, excitation energy is transferred by FRET from TO to JO, restoring maximal brightness of JO.
Another option to increase brightness of a single FIT probe is to increase the local constraint experienced by TO. We demonstrated that introduction of a locked nucleic acid (LNA) 3’ of TO doubles the maximal brightness of a given probe while it leaves responsiveness effectively unchanged. This effect is due to LNA-induced reduction of stacking distances between adjacent bases in a duplex increasing local viscosity.
Can you use such a brightness improved FIT probe for in vivo imaging of RNPs?
Almost – a single LNA modification within the probe backbone does not confer sufficient nuclease resistance to the forming duplex however, with additional methylation of the 2’ OH of ribose (mixmer design), the probes remain stable for over an hour after being injected into Drosophila oocytes. Also, a single probe is rarely bright enough to reveal individual RNPs. By combining three to five different mixmer probes against oskar mRNA, we managed to obtain brightness and contrast similar to 10x MS2 tagged transgenic oskar RNPs labelled with MCP-EGFP.
Fig. 4: RNA imaging in vivo: A mixture of three mixmer FIT probes targeting oskar mRNA was injected into living, wild-type oocytes (G, marked by arrow). oskar RNP motility (H, arrowheads) could be followed even 50 minutes after the injection. Scale bars represent 50 micron and 5 micron in G and H, respectively. Source: Hövelmann and Gaspar et al., 2014.
How do you know that the injected probes haven’t had any effects on the target?
We used the aforementioned oskarMS2(10x)/MCP-EGFP system, well documented by us and others, as reference. Even when injecting all five mixmer probes simultaneously, we observed no differences in oskar RNP motility compared to oskarMS2. When such an independent assay is unavailable, a good practice can be to test different combinations of several different FIT probes.
Of course, not all probes are inert. By targeting a known double stranded secondary structure important for oskar mRNA localization (a so-called localization element) with a mixmer probe complementary to one of the strands (osk6), we induced the same defects of oskar RNP motility as was reported previously for a series of mutated oskar transgenes. The FIT probe based manipulation was not only much less time consuming than generating and analysing transgenic copies of RNA but also had effect only on a particular step of the RNP biogenesis – the transport, in this instance. Mutations, on the other hand, may affect fate of the mRNA from transcription onward, and thus, the observed phenotype may be due to the sum of a series of defects.
In summary, careful choice of FIT probes is not only a tool of RNA visualization but also of RNP function discovery.
Can you combine these probes with other live cell markers, such as FP tagged proteins?
Yes, you can. Since most model organisms have by now vast libraries of (E)GFP tagged fusion proteins, it was vital for us to find conditions under which these two fluorophores – EGFP and TO – can be co-visualized. Absorption and emission spectrum of TO is sufficiently red-shifted compared to EGFP that with carefully choosing the excitation light – 470 nm for EGFP and 525 nm for TO – there is virtually no cross-talk between the two molecules in a sequential scan. These excitation wavelengths may seem a little exotic – indeed, for such dual-colour imaging, we use a state-of-the-art Leica confocal microscope equipped with a super-continuum light source. Naturally, we first used the oskarMS2(10x)/MCP-EGFP system in combination with FIT probes to check for co-labelling and co-migration of the two signals.
More interesting application, however, is to test co-localization with proteins that may have functions during RNP biogenesis. Classically, such discoveries are done in en mass biochemical assays. Microscopy, however, has clear advantages over such analyses: it can define the place, the time and the dynamics of the interactions in question. Unfortunately, RNPs are usually smaller than the diffraction limit and contain only a handful – maybe only one copy – of a given RBP. Thus, as most immuno-stainings inevitably result in dot like aspecific labelling – even with specific antibodies – real RNPs are indistinguishable from aspecific background. Autofluorescence of EGFP fusion proteins, on the other hand, has an almost hundred percent specificity and as few as a single GFP molecule can be detected with today’s most sensitive detectors.
If RNPs are so small, have you considered any super-resolution imaging?
Indeed – mainly not because of the RNP size but because of the crowdedness they can achieve at certain locations. Since the Drosophila egg-chamber is a quite thick specimen (~100 microns along the axial axis) and the interesting parts are separated from the glass surface by a 10–15 micron thick layer of cells, this limited the choice of methods. We found that STED microscopy worked under these conditions. LNA modified FIT probes have been proven to be excellent STED labels, since upon hybridization not only their brightness but also the fluorescence lifetime increases considerably, and thus gated STED further improves target specificity of the detected signal. Also, because of the overlap in the emission spectra of EGFP and TO, a single 592 nm depletion laser is sufficient to achieve dual-colour STED imaging. Although the light scattering of the specimen limits maximal lateral resolution to about 80–120 nm, both individual RNPs and RNP substructures are resolved to much greater detail than with conventional confocal microscopy. Such imaging allows us to perform precise object based co-localization analysis.
Can you similarly co-visualize two different RNPs or different segments of an mRNA within an RNP?
TO belongs to the thiazole orange family of visco-sensitive dyes. In this family, there are numerous hybridization sensitive dyes with different absorption and emission spectra. The cyan emitting pyridium benzothiazole organge (BO) has been already used successfully in combination with TO to label two different mRNA of the Influenza A virus. Unfortunately, BO has almost no emission at 592 nm and thus it is not available for commercial STED solutions.
We recently found another dye, quinoline blue (QB) a cyanine family member. Its maximal emission is at 605 nm and it is a relatively bright dye (brightness is 56.4) but more importantly, this dye is also sensitive to the viscosity of its environment. QB containing FIT probes were found to be as bright as TO containing ones, while their responsiveness was an order of magnitude higher (~100-fold difference between the dark and bright states). Using a QB containing LNA modified oligo-dT probe together with the TO labelled oskar probes we could co-visualize the whole poly-A(+) transcriptome and oskar mRNA in the aforementioned wash-free setup. The emission spectrum of QB allowed us to use the 775 nm pulsed depletion laser; in a sequential STED scan, together with the 592 nm depletion line, we obtained greatly detailed super-resolution images of Drosophila egg-chambers.
Different RNA imaging methods have been recently reviewed in
- Gaspar I, and Ephrussi A: Strength in numbers: quantitative single-molecule RNA detection assays. Wiley Interdiscip. Rev. Dev. Biol. 4 (2): 135–50 (2015); doi: 10.1002/wdev.170. Epub 2015 Jan 21.
- Köhler O, Jarikote DV, and Seitz O: Forced intercalation probes (FIT Probes): thiazole orange as a fluorescent base in peptide nucleic acids for homogeneous single-nucleotide-polymorphism detection. Chembiochem. 6 (1): 69–77 (2005).
- Hövelmann F, Gaspar I, Ephrussi A, and Seitz O: Brightness enhanced DNA FIT-probes for wash-free RNA imaging in tissue. J. Am. Chem. Soc. 18; 135 (50): 19025–32 (2013); doi: 10.1021/ja410674h. Epub 2013 Dec 10.
- Hövelmann F, Gaspar I, Loibl S, Ermilov EA, Röder B, Wengel J, Ephrussi A, and Seitz O: Brightness through local constraint--LNA-enhanced FIT hybridization probes for in vivo ribonucleotide particle tracking. Angew. Chem. Int. Ed. Engl. 13; 53 (42): 11370–5 (2014); doi: 10.1002/anie.201406022. Epub 2014 Aug 28.
- Kummer S, Knoll A, Socher E, Bethge L, Herrmann A, and Seitz O: PNA FIT-probes for the dual color imaging of two viral mRNA targets in influenza H1N1 infected live cells. Bioconjug. Chem. 17; 23(10): 2051–60 (2012); doi: 10.1021/bc300249f. Epub 2012 Sep 14.
- Hövelmann F, Gaspar I, Kasper M, Chamiolo J, Steffen J, Ephrussi A, and Seitz O: LNA-Enhanced DNA FIT-Probes for Multicolour RNA Imaging. Chem. Sci. doi: 10.1039/C5SC03053. Epub 2015 Nov 02.