© 2001 Oxford University Press Nucleic Acids Research, 2001, Vol. 29, No. 17 e89

Linear 2′ O-Methyl RNA probes for the visualization ofRNA in living cellsC. Molenaar*, S. A. Marras1, J. C. M. Slats, J.-C. Truffert2, M. Lemaître2, A. K. Raap,R. W. Dirks and H. J. Tanke

Department of Molecular Cell Biology, Wassenaarseweg 72, 2333 AL Leiden, The Netherlands, 1Department ofMolecular Genetics, Public Health Research Institute, 455 First Avenue, New York, NY 10016, USA and2Eurogentec s.a. Parc Scientifique du Sart Tilman, 4102 Seraing, Belgium

Received May 16, 2001; Revised and Accepted July 16, 2001

ABSTRACT

U1snRNA, U3snRNA, 28 S ribosomal RNA, poly(A)RNA and a specific messenger RNA were visualizedin living cells with microinjected fluorochrome-labeled 2′ O-Methyl oligoribonucleotides (2′ OMe RNA).Antisense 2′ OMe RNA probes showed fast hybridiza-tion kinetics, whereas conventional oligodeoxyribo-nucleotide (DNA) probes did not. The nucleardistributions of the signals in living cells were similarto those found in fixed cells, indicating specifichybridization. Cytoplasmic ribosomal RNA, poly(A)RNA and mRNA could hardly be visualized, mainlydue to a rapid entrapment of the injected probes inthe nucleus. The performance of linear probes wascompared with that of molecular beacons, which dueto their structure should theoretically fluoresce onlyupon hybridization. No improvements were achievedhowever with the molecular beacons used in thisstudy, suggesting opening of the beacons by mecha-nisms other than hybridization. The results show thatlinear 2′ OMe RNA probes are well suited for RNAdetection in living cells, and that these probes can beapplied for dynamic studies of highly abundantnuclear RNA. Furthermore, it proved feasible tocombine RNA detection with that of green fluores-cent protein-labeled proteins in living cells. This wasapplied to show co-localization of RNA with proteinsand should enable RNA–protein interaction studies.

INTRODUCTION

The eukaryotic cell nucleus is highly organized to tightlycontrol and coordinate gene expression. This picture hasemerged from fluorescence in situ hybridization and immuno-cytochemical studies, showing that chromosomes occupydistinct territories within the nucleus and that many of thenuclear factors that play a role in transcription and RNAprocessing are compartmentalized in nuclear bodies, asreviewed by Lamond and Earnshaw (1). The chromosometerritories contain regions in which the degree of chromatincondensation varies. Transcriptionally active genes have been

shown to be located at the borders of the condensed chromatinregions to have access to the transcription, RNA processingand transport machinery (2,3). Nuclear bodies, like coiledbodies, PML bodies and speckles may then serve as pre-assembly sites of the components of this machinery or asstorage sites from which factors are recruited to active genes(4,5). Alternatively, or additionally, transcription and RNAprocessing complexes may be functional at these sites (6).

To relate the spatial organization of nuclear components tofunctional activities the dynamic behaviors of DNA, RNA andproteins need to be studied in living cells. Various strategiesfor in vivo labeling of DNA, RNA and proteins have beenproposed. DNA can be labeled by the transient incorporationof fluorescent nucleotides during replication, which results in apartial labeling of chromatin in successive cell generations (7).Proteins can be tagged with the spectral variants of the greenfluorescent protein (GFP) and their dynamic behavior can beanalyzed by time-lapse experiments and by fluorescence redis-tribution after photo bleaching (FRAP) (8,9). More recently,Tsukamoto et al. (6) described a system to visualize a gene andits protein product in living cells. To detect the various RNAclasses in living cells, several approaches have been developed(reviewed in 10–13). Ainger et al. (14) injected in vitro tran-scribed fluorescent RNAs, which act as duplicates of endo-genous RNAs, and studied their localization upon delivery intocells. As an alternative, mRNA was visualized using GFP as afluorescent tag on the RNA-loop binding protein MS2 (15–17).In addition to these two methods, in vivo hybridizationmethods were used that are based on the annealing of shortantisense fluorescent probes to specific cellular RNAs. In moststudies, short linear DNA oligonucleotides (20–40 bases) wereused (18–21). As an alternative, Carmo-Fonseca et al. (22)described the use of 2′ O-Methyl (2′ OMe) RNA and 2′ O-allylRNA probes for the detection of small nuclear RNAs. 2′ OMeRNA probes were considered to perform better then DNAoligonucleotides because they are not only nuclease resistant,but also possess a higher affinity, increased specificity, fasterhybridization kinetics and a superior ability to bind to struc-tured targets compared to DNA oligonucleotides (23,24).Recently, so-called molecular beacons were introduced forRNA detection in living cells (19,25). Molecular beacons arehairpin-shaped oligonucleotide probes that fluoresce onlyupon hybridization (26,27). The rationale for using molecularbeacons to detect RNAs in living cells was to improve

*To whom correspondence should be addressed. Tel: +31 71 527 6278; Fax: +31 71 527 6180; Email: c.molenaar@lumc.nl

e89 Nucleic Acids Research, 2001, Vol. 29, No. 17 PAGE 2 OF 9

signal-to-noise ratios by eliminating fluorescence signalsderived from non-hybridized probe sequences. In order to cometo a generally applicable technique for the detection of RNAs inliving cells, we systematically evaluated the performance oflinear probes and molecular beacons synthesized both fromDNA and 2′ OMe RNA monomers. The RNA targets detectedwere two small nuclear RNAs, U1 snRNA and U3 snRNA,ribosomal RNA of the 28S ribosomal subunit, poly(A) RNAand a specific mRNA, the mRNA encoding the human cytome-galovirus immediate early antigen (HCMV IE) in transformedrat fibroblasts. We demonstrate that of the four probe typestested, linear 2′ OMe RNA probes are the most suitable forspecific detection of these RNAs, representing differentclasses of RNA, in the nuclei of living cells. Under the condi-tions used, molecular beacons did not result in images withimproved signal to noise ratios, thereby leading to better detec-tion sensitivity.

MATERIALS AND METHODS

Probes

For the various RNA targets described, probes were selectedon the basis of sequences in the NCBI genome database.

Single-stranded regions or partly single-stranded regions in thesecondary RNA structure for U1 snRNA and U3 snRNA (28)were targeted. They were varied in length, structure andnucleotide modification. Sequences and GenBank IDs arelisted in Table 1. DNA oligonucleotides were synthesizedusing standard phosphoramidite chemistry and purified byHPLC. 2′ OMe RNA probes were synthesized using standard2′ OMe phosphoramidite monomers. For linear probes fluores-cent dyes were covalently linked to the 5′-end via a succinimidylester derivative (Molecular Probes, Eugene, OR). Molecularbeacons were synthesized automatically with the 3′-4-dimethylaminoazobenzene-4′-sulfonyl (dabcyl)-quencherlinked to the solid phase support (DABCYL-CPG, GlenResearch, Sterling, VA). Oligonucleotide chain elongation wasperformed via standard cyanoethyl-phosphoramidite chem-istry and fluorophores were covalently linked to the 5′-end viaiodoacetamide (Molecular Probes) or phosphoramidite deriva-tives (Glen Research) following release from the solid support.All 2′ OMe RNA probes were purified twice by reverse phaseHPLC with a Waters 600E instrument, equipped with a Waters996 Photodiode Array Detector for simultaneous detection atthree different wavelengths. Ion-molecular weights of purifiedprobes were determined by mass spectrometry using a

Table 1. Probe sequences

aAdditional bases in longer probes for the same target are indicated within brackets.bLonger linear probes.All sequences were synthesized as DNA and as 2′ OMe RNA oligonucleotides, except for the longer linear probes that were only used asDNA probes.All probes were labeled with TAMRA; molecular beacons were labeled additionally with DABCYL as quencher.

RNA target Design Sequence GenBank ID Nucleotidepositions

U1 snRNA linear probea (c)cugccagguaaguau(g) 340085 1–17

MB probe gcgac cugccagguaaguau gucgc

U3 snRNA Linear probea (c)ggcuucacgcucagg(a) 2736114 205–221

MB probe (c/g)cgacggcuucacgcucagggucg(c/g)

28S rRNA Linear probe 17 bp uaccaccaagaucugca 337381 2321–2337

Linear probe 35 bpb gaatatttgctactaccaccaagatctgcacctgc 2316–2350

MB probe gcgacuaccaccaagaucugcagucgc

Poly(U) Linear probes (u)18(u)22(t)40b

MB probes gucgac(u)18gucgac;cgucgac(u)22gucgacg

Poly(A) Linear probe (a)18

MB probe gucgac(a)18gucgac

CMV1 Linear probe aaacauccucccauca 330617 119–134

MB probe cgcacaaacauccucccaucagucag (exon 4)

CMV2 Linear probe caucuccucgaaaggcu 169–186

MB probe cgcacaucuccucgaaaggcugugcg

CMV3 Linear probe acaucaugcagcuccuu 271–287

MB probe ccagacaucaugcagcuccuucugg

CMV4 Linear probe gagcacugaggcaaguuc 465–482

MB probe ccagggagcacugaggcaaguucccugg

CMV5 Linear probe ugugaucaaugugcguga 551–568

MB probe ccagugugaucaaugugcgugcugg

PAGE 3 OF 9 Nucleic Acids Research, 2001, Vol. 29, No. 17 e89

Dynamo time-of-flight instrument. The initial molecularbeacon probes were designed following the guidelinesdescribed by Tyagi and Kramer (26) (http:\\www.molecular-beacons.org). The intramolecular configuration of molecularbeacons and their target sequences was modeled using theDNA mfold program (www.ibc.wustl.edu/~zuker/dna/form1.shtml), which uses thermodynamic parameters estab-lished by SantaLucia (26). The molecular beacons did notcontain self-complementary sequences in the loop. Beaconswere equipped with stem sequences of varying length.

Cell culture and microinjection

For microinjection experiments and life cell observation,U2OS (human osteosarcoma) cells and R9G cells werecultured on coverslips in 3.5 cm Petri dishes (Mattek, Ashland,MA) in RPMI 1640, without phenol red supplemented with 5%fetal calf serum and buffered with 25 mM HEPES buffer to pH 7.2(Life Technologies, Breda, The Netherlands). R9G is a ratfibroblast cell line harboring a tandem repeat of 50 copies ofthe HCMV IE gene in one of the rat chromosomes (29–31). Toinduce the S-phase dependent HCMV IE gene expression,50 µg/ml cycloheximide (Sigma, St Louis, MO) was added tothe culture medium, 3–4 h prior to injection. The temperatureof the culture medium was kept at 37°C using a heated ringsurrounding the culture chamber and a microscope objectiveheater (Bioptechs, Butler, PA). For microinjection, a NarishigeIM 300 microinjector (Narishige, Tokyo, Japan) was used.Needles with a tip inner diameter of ∼0.5 µm were pulled fromglass capillaries (no. MTW100F-3 WPI, Aston, UK) on aSutter micropipette puller (p97, Sutter instrument company,Novato, CA). Probes were dissolved in a buffer containing 80 mMKCl, 10 mM K2PO4, 4 mM NaCl pH 7.2 to a final concentra-tion of 5 µM and microinjected in the cytoplasm of cells. Theoptimal concentration and the amount of probe moleculesinjected was determined on the basis of a series of initialexperiments taking the intensity of the hybridization signalsand cell viability into account. Typically, for each probe about50 cells were injected and monitored over time. Representativecells were selected and analyzed by digital fluorescence micro-scopy.

RNA-FISH on fixed cells

To detect the various RNAs in fixed cells, in situ hybridizationwas performed essentially as described previously (32). Inbrief, U2OS cells were fixed in 3.7% formaldehyde, 5% aceticacid in PBS for 15 min at room temperature. The cells werepretreated for 1 min with 0.1% pepsin pH 2.0 at 37°C beforehybridization. The 2′ OMe RNA probes were dissolved in 40%deionized formamide/2× SSC pH 7.4 and the DNA probeswere dissolved in 10% deionized formamide/2× SSC. Hybridi-zation mixture (10 µl) containing 50 nM probe was applied oncells grown on glass slides and covered with an 18 × 18 mmcoverslip. Slides were put on an 80°C hot plate for 2 min todenature nucleic acids and then incubated for 2 h at 37°C forhybridization. To remove excess of probe, slides were rinsed in2× SSC for 5 min, 2× SSC/10% formamide for 10 min, 2× SSCfor 5 min. Cells were then mounted in VectaShield (VectorLaboratories, Burlingame, CA).

Construction and expression of GFP-fusion proteins

The cDNAs encoding ASF/SF2 and fibrillarin were generatedby RT–PCR and cloned into the pEGFP-C1 vector (Clontech,Palo Alto, CA) as described previously (9). The resultingplasmid constructs were microinjected into cells (50 ng/µlH2O), and cells showing GFP-expression were injected 24 hafter transfection with antisense probes.

Life cell imaging

Microinjection and life cell imaging was performed on a ZeissAxiovert 135 TV microscope (Zeiss, Jena, Germany),equipped with a 100 W mercury arc lamp. The objectives usedwere a 40×/NA 1.30 Plan Neofluar phase and a 100×/NA 1.30Plan Neofluar phase. Tetramethylrhodamine (TAMRA)labeled probes were detected using a filter set consisting of a546/10 nm BP filter, a 580 nm DM and a 590 LP barrier filter.Images were acquired using a cooled charged coupled devicecamera (type PXL, Photometrix, Tucson, AZ). At various timepoints after microinjection, images were acquired and fluores-cence distribution patterns were analyzed. Typical exposuretimes were 1–2 s.

RESULTS

U1 snRNA and U3 snRNA, 28 S rRNA and poly(A) arepresent in high amounts in a cell, and specific nuclear compart-ments, such as nucleoli, speckles and Cajal bodies are enrichedfor some of these RNA species. U1 snRNA is a component of‘spliceosomes’, structures that are responsible for mRNAsplicing (33); U3 snRNA has a function in rRNA processing,and is mainly localized in nucleoli (34). Poly(A) RNA,representing mRNA or pre-mRNA, is present throughout thenucleoplasm and concentrated in speckles and also in thecytoplasm (35,36), and 28S rRNA is localized in nucleoli andcytoplasm. The high expression from the HCMV IE gene integ-ration sites generates a local nuclear track consisting of tran-script (31). For each of these targets four different types ofprobes were synthesized: linear 2′ OMe RNA probes, molecularbeacon 2′ OMe RNA probes, linear DNA probes and molecularbeacon DNA probes, and used for hybridization in fixed andliving cells. Probe sequences and the fluorescent labels thatwere used are listed in Table 1.

Linear 2′ OMe RNA probes

Probes were injected in the cytoplasm of U2OS cells, causingless damage to the cells compared to nuclear injections. Within60 s of injection, all probes accumulated in the nucleus, anobservation that is in agreement with earlier studies (37,38).Microinjected fluorescent linear 2′ OMe RNA probes toU1 snRNA were visible in speckles, Cajal bodies andthroughout the nucleoplasm (Fig. 1C), excluding nucleoli, inagreement with earlier studies (22,39,40). The U3 snRNA2′ OMe RNA probe was found to localize in nucleoli and in thenucleoplasm (Fig. 1G). Nucleoli were identified in phasecontrast images (Fig. 1B, D, F and H). Similar staining patternswere observed when these probes were hybridized on fixedcells (Fig. 1A and E), indicating that they specifically hybrid-ized to their respective RNA targets in living cells. In generalthe signal intensity of hybridized probe was higher in fixedcells compared to that in living cells. The pepsine-treatment

e89 Nucleic Acids Research, 2001, Vol. 29, No. 17 PAGE 4 OF 9

and the denaturation step that are applied in FISH on fixedcells make the RNA more accessible for the probe, causing amuch higher hybridization efficiency compared to that inliving cells. Post-hybridization washing steps removed theunbound probe from fixed cells. The homogeneous nucleo-plasmic staining observed with the U3 snRNA probe in livingcells, that obscured the detection of U3 snRNA in Cajal bodies,was not observed in fixed cells (Fig. 1E). The microinjectedcells were followed for several days and underwent mitosis.Similar localization patterns were observed in daughter cells(not shown). This indicated that the probes did not interferewith cell vitality. When non-fluorescent probes for U1 snRNAand U3 snRNA were injected in a 5-fold higher concentration

prior to injection of fluorescent probes, the specific hybridiza-tion pattern was not obtained (data not shown), indicatingblocking of the hybridization sites.

Following microinjection in U2OS cells, fluorescent linear2′ OMe RNA probes designed to hybridize to 28S rRNA orpoly(A) RNA localized at the expected nuclear sites. RibosomalRNA was detected in nucleoli (Fig. 2C) and poly(A) RNA wasdetected in speckles and throughout the nucleoplasm (Fig. 2G).Similar nuclear hybridization patterns were obtained whenthese probes were hybridized to fixed cells (Fig. 2A).However, in fixed cells, the nucleoplasmic staining was lesspronounced and the contrast between speckles and nucleo-plasm was more apparent (Fig. 2E). The cytoplasmic stainingof 28S rRNA and poly(A) RNA observed in fixed cells(Fig. 2A and E) was not observed in living cells (Fig. 2C andG), due to a rapid accumulation of the probes in the nucleus.Only a very low concentration of all injected probes could beobserved in the cytoplasm.

To further prove the specificity of hybridization in live cells,cells were microinjected with control probes without a specifictarget sequence. Microinjection of the poly(A)18 2′ OMe RNAprobe resulted in a diffuse staining throughout the nucleoplasmwithout revealing any specific hybridization pattern (Fig. 2K).A similar diffuse staining was observed when non-induced rat9G cells or U2OS cells were microinjected with a 2′ OMeRNA probe specific for HCMV IE RNA (not shown).

Detection of the transcription site of CMV RNA in livingR9G cells

As 2′ OMe RNA oligonucleotides were shown to hybridize tohigh abundant RNAs in living cells these probes weresubsequently used to detect CMV IE mRNA in a transformedrat fibroblast cell line. Induced R9G cells were microinjectedwith up to five TAMRA-labeled 2′ OMe RNA oligo-nucleotides specific for IE mRNA. Injection of single probesdid not result in specific hybridization signals. Nevertheless,when a mixture of three 2′ OMe RNA oligonucleotides (I, IIand IV in Table 1) was injected, a fluorescent dot representingIE RNA at the integrated IE gene repeat was visible (Fig. 3A).This nuclear signal was similar to the one observed in fixedcells after FISH with these probes (Fig. 3B), and as observed inprevious studies (31,41). In addition to the fluorescent dot, livecells revealed a homogeneous fluorescent staining throughoutthe nucleus including nucleoli. This can be ascribed to non-hybridized probes and non-specifically hybridized probes.Cytoplasmic IE mRNA could not be detected in living cells.

2′ OMe RNA molecular beacons

To test their performance in living cells, 2′ OMe RNA molecularbeacon probes specific for U1 and U3 snRNA, for 28S rRNAand for poly(A) RNA were designed. As control probes, apoly(A) beacon and an HCMV IE mRNA specific 2′ OMeRNA beacon were used (Table 1). Since not much is knownabout the hybridization kinetics of these types of probes inliving cells, all molecular beacons were tested in solution forspecificity for their target sequences and their ability to fluoresceupon hybridization (not shown). 2′ OMe RNA molecularbeacons specific for U1 snRNA, U3 snRNA, 28S rRNA andpoly(A) RNA microinjected in U2OS cells showed initially avery low signal intensity because they are in a closed and thusquenched form. However, after 5–10 min, the signal intensity

Figure 1. Similar localization patterns of 2′ OMe RNA oligonucleotideshybridized to U1 snRNA and U3 snRNA in fixed and living U2OS cells(A, C, E and G). The distinct localization patterns observed for U1 snRNA(C) and U3 snRNA (G) detected in living cells after microinjection of the anti-sense oligonucleotides are similar to those observed in fixed cells (A and E),indicating a specific hybridization to RNA in living cells. Corresponding phasecontrast images (B, D, F and H). The intense spots in (C) and (E), indicated byarrows, are Cajal bodies, the faint structures in the nucleoplasm in (C) arespeckles.

PAGE 5 OF 9 Nucleic Acids Research, 2001, Vol. 29, No. 17 e89

increased steadily throughout the nucleus following injectionand similar localization patterns were observed as obtainedwith linear 2′ OMe RNA probes (Fig. 4). Overall, the maximalsignal intensity achieved with the molecular beacons was 2–3-foldlower compared to the linear probes. This difference in intensitywas also observed when the comparison was performed on

fixed cells (data not shown). In addition to nuclear compart-ments a diffuse background staining of the nucleoplasm wasobserved following microinjection of the beacon probes inliving cells. This background staining was independent fromspecific hybridization to an RNA target, since opening of themolecular beacon also occurred when 2′ OMe RNA beaconswere injected with no RNA target in U2OS cells. This wasobserved after injection of a beacon with a loop containingonly 18 adenines or a beacon specific for HCMV IE RNA (notshown).

To find the optimal beacon probe structure for hybridizationin living cells, probes for U3 snRNA and poly(A) RNA withdifferent stem–loop lengths were tested (Table 1). The alteredprobe structures did not result in improved signal-to-noiseratios. For example, a poly(U) molecular beacon with a lengthof 28 bases (loop, 18 bases; stem, 5 bp) and a molecular beaconwith a length of 36 bases (loop, 22 bases; stem, 7 bp) were usedto detect poly(A) tails in live cells. Microinjection of the 28 baseprobe resulted in the specific ‘speckled’ pattern that wasobtained with the linear probe, although the intensity of thesignals was 2–3-fold lower (not shown). Injection of the 36 baseprobe resulted in a diffuse nuclear staining (Fig. 4E), similar tothat obtained with the non-specific poly(A) beacon (Fig. 4G).

The performance of linear and molecular beacon DNAprobes

To test the theoretical advantage of 2′ OMe RNA probescompared to unmodified oligodeoxynucleotides for detectingRNA species in vivo, we directly compared the two types ofprobes. Linear fluorescent DNA probes with the samesequences as the 2′ OMe RNA probes (Table 1) were synthesized.Upon microinjection, all DNA probes rapidly accumulated inthe cell nucleus which resulted in a diffuse staining, withoutrevealing any specific hybridization pattern (Fig. 5). Hybrid-ization on fixed cells indicated that the probes were specific fortheir RNA target (data not shown). These results suggest that inliving cells, linear DNA probes have an insufficient affinity forthe RNA target sequences to allow specific and stable hybrid-ization. The inability to show specific hybridization was notcaused by degradation of the probes, since the observationswere done within minutes of injection of the probes, and thehalf-life of DNA oligonucleotides injected in living cells wasshown to be 15 min (42). It is feasible that a low percentage ofthe injected DNA oligonucleotides did in fact hybridize to thetarget RNA, but that the relatively dim specific signal was verymuch disturbed by the high amount of the non-specificstaining. Even 35 and 40 base long DNA probes for 28S rRNA(Fig. 5A) and poly(A) RNA (Fig. 5B), possessing higheraffinities compared to the 15–22 base probes, did not result indistinct localization patterns in living cells. Also molecularbeacon DNA probes for the same targets did not reveal aspecific hybridization pattern in living cells upon injection.

Combined RNA and protein detection in living cells

Having established that linear 2′ OMe RNA probes are suitedfor detection of RNA species in living cells, we examined thepossibility to simultaneously detect RNA and protein mol-ecules in living cells. U2OS cells were transiently transfectedwith GFP expression vectors containing the coding sequencesfor alternative splicing factor (ASF) and fibrillarin andexpressing cells were microinjected with a 2′ OMe RNA probe.

Figure 2. Hybridization of linear 2′ OMe RNA probes to abundant RNAtargets in living cells. Probes for 28S rRNA (A and C) and poly(A) RNA(E and G) hybridized to fixed and living cells show a similar nuclear hybrid-ization pattern but a different cytoplasmic localization pattern. Probes to 28SrRNA and poly(A) RNA accumulate in the nucleus within minutes of injectionand the level of hybridization to the RNA present in the cytoplasm is negligible.The (A)18 probe shows no specific hybridization pattern (K). Correspondingphase contrast images (B, D, F, H and L).

e89 Nucleic Acids Research, 2001, Vol. 29, No. 17 PAGE 6 OF 9

Figure 6A–D shows the simultaneous detection of ASF–GFPtogether with poly(A) RNA in a living cell. As expected fromprevious localization studies on fixed cells, ASF andpoly(A) RNA were shown to co-localize in a speckledpattern. Fibrillarin–GFP, which localizes to nucleoli and Cajalbodies (9), was shown to co-localize with U1 snRNA in CajalBodies (Fig. 6E–H). These results indicate that proteins andRNA species can be visualized simultaneously in living cellswithout disturbing their integrity. The simultaneous labeling ofRNA and proteins in living cells not only allows microscopicco-localization studies as was shown here, but could alsoprovide the opportunity to study their interaction, by applyingRNA–Protein Fluorescence Resonance Energy Transfer(FRET).

DISCUSSION

In this study we evaluated the performance of different probetypes, for the detection of representatives of the major RNAclasses, rRNA, snRNA, poly(A) RNA and a specific mRNA,in living cells. Our results show that 2′ OMe RNA probes canbe designed such that they hybridize with high specificity tothe corresponding target sequences in living cells withoutaffecting cell vitality, and that they are superior to DNA oligo-nucleotides. Cells microinjected with the 2′ OMe RNA probescould be monitored for several days and underwent mitosis,showing that processes like transcription and proliferationwere not irreversibly disturbed. Long-term observationhowever was only found possible when low light level illumi-nation was applied, in order to prevent fluorescence fading andphototoxic effects. As was observed before (37,38,43), oligo-nucleotide probes rapidly entered the cell nucleus after micro-injection in the cytoplasm, and specific hybridization patternswere observed within 1–2 min irrespective of the target RNA.This suggests that oligonucleotide probes can move freelythroughout the cell nucleus.

RNA targets in the cytoplasm could not be labeled presumablybecause rapid entrapment of the probes in the nucleus causeddepletion of the cytoplasm and prevented successful hybridiza-tion to cytoplasmic RNA targets. One possibility to overcomethis nuclear entry of probes when injected in the cytoplasm, isto couple a high molecular weight tag to the oligonucleotide.

Tsuji et al. (18) used streptavidin to tag oligonucleotide probesto detect mRNA in the cytoplasm. Alternatively, oligo-peptideconjugates, where the peptide sequence is a nuclear exportsignal, could be tested (44). Interestingly, 24 h after injectionof a probe to poly(A) tails, an increase in cytoplasmic signalwas observed, suggestive for the export of mRNA from thenucleus to the cytoplasm. This signal, however, could alsohave been derived from degradation products.

Using linear 2′ OMe RNA probes, U1 and U3 snRNA, 28SrRNA and poly(A) RNA were detected with high signal-to-noiseratios, indicating that the majority of the injected probewas hybridized specifically to the RNA of interest. This wasparticularly true when the injected amount of probe wasestimated to be lower or equal to the number of target RNAmolecules. The probes for U1 and U3 snRNA and poly(A)RNA that localized to specific nuclear compartments asspeckles, Cajal bodies and nucleoli, were also found presentthroughout the nucleoplasm. Since this diffuse staining wasmuch weaker after FISH on fixed cells, it is possible that partof this signal is derived from non-hybridized fractions ofprobes.

HCMV IE mRNA was detected as a single confined spot inthe nucleus. This shows that the high concentration of HCMVIE mRNA that is present at the transcribing HCMV genecluster can be detected in living cells, and suggests that thismethod provides a useful additional tool for the visualizationof gene activity in vivo. The IE mRNAs emanating from thetranscription site towards the cytoplasm, as revealed by FISHon fixed cells (31), could however not be detected. This couldbe attributed to a low signal-to-noise ratio. Interestingly,Sei-Iida et al. (45) observed that DNA probes hybridizeefficiently to RNA molecules in vitro during or just aftersynthesis and less efficiently after folding. Thus, also in vivo,the nascent RNA transcripts are probably better accessible forprobes compared to the folded mRNA–protein complexes intransport through the nucleus. The efficiency of hybridizationof antisense probes to mRNA varies along the target sequence,depending on secondary RNA structures and proteinsoccupying the probe binding sites (18). Testing a series ofprobes along the sequence may identify the most optimalprobes for a particular mRNA. In this study 2′ OMe RNAoligonucleotides were shown to be well suited for detection of

Figure 3. Detection of the transcription site of CMV RNA, transcribed from the integrated HCMV IE 1 gene in R9G cells. (A) Transcription sites and cytoplasmicmRNA are visible. Detection of transcription sites in living cells. The small bright spot shows transcription from the integration sites. A high background wasobserved in the nucleus and nucleoli. The signal in the cytoplasm was very weak. (B) Representative example of an in situ hybridization with three linear 2′ OMeRNA oligonucleotides on fixed R9G cells.

PAGE 7 OF 9 Nucleic Acids Research, 2001, Vol. 29, No. 17 e89

highly abundant nuclear RNA targets in living cells. The factthat 2′ OMe RNA probes and RNA form a hybrid that is notdegraded by RNase H, whereas DNA/RNA and RNA/RNAhybrids do form substrates for this enzyme, is an additionalargument to use the 2′ OMe modification. DNA analogs thatwere shown to have a higher affinity or specificity for RNA,such as 2′ allyl RNA (46), 2′ methoxy-ethoxy (2′ Moe) RNA(47,48), peptide nucleic acids (PNAs) (49) and several others,could possibly be used to further optimize hybridization inliving cells.

Reducing or eliminating signals from unbound or non-specifically bound probe

The key question is whether the signals obtained by in vivohybridization are actually derived from probes specificallyhybridized to RNA. To reduce or circumvent the contributionof unwanted signals from unbound or non-specifically boundprobes, a method may be applied that enables discriminationbetween signals derived from probes bound specifically to theRNA target and signals from unbound or non-specificallybound probes.

Therefore, we tested a series of molecular beacons for RNAdetection in living cells. In vitro experiments demonstrated theimproved specificity of 2′ OMe RNA molecular beaconscompared to 2′ OMe RNA linear probes in vitro (data notshown). We could not confirm the beneficial properties ofmolecular beacons for detecting RNA targets in living cells.Although 2′ OMe RNA molecular beacons showed specifichybridization in our study and were applied before in live cellsby Matsuo et al. (25) and Sokol et al. (19), for all RNAs in thisstudy the linear probes showed better signal-to-noise ratiosthan molecular beacons. In general, the molecular beaconsused by us showed lower affinity for RNA targets in fixed andliving cells. Specific as well as control beacons were shown toopen up shortly after entering the cell nucleus, indicating thattheir non-specific interaction with nuclear proteins or nucleicacids imposes a conformational change leading to fluores-cence. Recently, it has been shown that a beacon type structurecan be recognized by lactate dehydrogenase (50), and sincemany nuclear proteins can interact with structural motivespresent in RNA molecules, it is well possible that such interac-tions occur with molecular beacons, resulting in fluorescenceregardless of specific hybridization.

Sixou et al. (51) proposed an alternative method to discriminatebetween specific and non-specific hybridization based on

Figure 4. Hybridization of molecular beacons in living cells. U1 snRNA(A) and U3 snRNA (C) molecular beacons show similar hybridization patternsas the linear probes (Fig. 1), but have a lower signal-to-noise ratio. A molecularbeacon targeted to poly(A) RNA consisting of a 22 base loop and a 7 bp stemshowed only a very vague specific staining (E), compared to the linear probe(Fig. 2G). The overall intensity was lower for molecular beacons. For none ofthe RNAs that were targeted, a molecular beacon showed hybridization with animproved contrast. The control poly(A) molecular beacon probe does not showa specific hybridization pattern (G). Corresponding phase contrast images(B, D, F and H).

Figure 5. No specific hybridization visible after microinjection of DNA oligo-nucleotides. Two examples of microinjected cells where RNA was targetedwith DNA oligonucleotides. Thirty-five and 40 base antisense DNA probes for28S ribosomal RNA (A) and the poly(A) tail of mRNA (C) were microinjectedinto living U2OS cells. The probes accumulated in the nucleus with time aswas expected but did not show any further specific localization as was observedwith the 2′ OMe RNA probes (as in Figs 1 and 2). (B and D) Correspondingphase contrast images.

e89 Nucleic Acids Research, 2001, Vol. 29, No. 17 PAGE 8 OF 9

FRET. Recently Tsuji et al. (18) applied this method by usingadjacently hybridizing probes, labeled 3′ and 5′ with donor andacceptor molecules, respectively, thereby facilitating FRET.This approach was used to detect cFos mRNA in transfectedCOS cells using Bodipy 493/503 and Cy5 labeled DNA oligo-nucleotides. For the detection of FRET it is required that bothprobes are simultaneously hybridized on the target molecule,demanding very efficient probe hybridization. For this reason,high affinity probes should improve FRET efficiency.

In conclusion, linear 2′ OMe RNA antisense probes promiseto be a general applicable probe type allowing the detection of

various endogenous nuclear RNAs in living cells, withoutinterfering with cell vitality. In addition, we demonstrate thatsimultaneous detection of RNA species with GFP-taggedproteins is very feasible, opening a way to study the molecularinteractions involved in RNA transcription, processing andtransport in living cells.

ACKNOWLEDGEMENTS

We would like to thank S. Snaar, S. Khazen, H. Albus andE. Manders for their excellent technical assistance, J. Bonnetand H. Vrolijk for their expertise in digital imaging micro-scopy and C. Jost for her helpful comments on the manuscript.We would like to thank Michel Dechamps and Dario Largana(Eurogentec) for technical assistance in the synthesis of theprobes used in this study. C. Molenaar is supported by theDutch Scientific Organization NWO program ‘4D imaging ofliving cells and tissues’, grant no. 901-34-144.

REFERENCES

1. Lamond,A.I. and Earnshaw,W.C. (1998) Structure and function in thenucleus. Science, 280, 547–553.

2. Verschure,P.J., van Der Kraan,I., Manders,E.M. and van Driel,R. (1999)Spatial relationship between transcription sites and chromosometerritories. J. Cell Biol., 147, 13–24.

3. Leonhardt,H., Rahn,H.P., Weinzierl,P., Sporbert,A., Cremer,T., Zink,D.and Cardoso,M.C. (2000) Dynamics of DNA replication factories inliving cells. J. Cell Biol., 149, 271–280.

4. Dirks,R.W., Hattinger,C.M., Molenaar,C. and Snaar,S.P. (1999)Synthesis, processing, and transport of RNA within the three-dimensionalcontext of the cell nucleus. Crit. Rev. Eukaryot. Gene Expr., 9, 191–201.

5. Misteli,T. (2000) Cell biology of transcription and pre-mRNA splicing:nuclear architecture meets nuclear function. J. Cell Sci., 113, 1841–1849.

6. Tsukamoto,T., Hashiguchi,N., Janicki,S.M., Tumbar,T., Belmont,A.S.and Spector,D.L. (2000) Visualization of gene activity in living cells.Nature Cell Biol., 2, 871–878.

7. Zink,D., Cremer,T., Saffrich,R., Fischer,R., Trendelenburg,M.F.,Ansorge,W. and Stelzer,E.H. (1998) Structure and dynamics of humaninterphase chromosome territories in vivo. Hum. Genet., 102, 241–251.

8. Phair,R.D. and Misteli,T. (2000) High mobility of proteins in themammalian cell nucleus. Nature, 404, 604–609.

9. Snaar,S., Wiesmeijer,K., Jochemsen,A.G., Tanke,H.J. and Dirks,R.W.(2000) Mutational analysis of fibrillarin and its mobility in living humancells. J. Cell Biol., 151, 653–662.

10. Pederson,T. (1999) Movement and localization of RNA in the cellnucleus. FASEB J., 13 (supp. 2), S238–S242.

11. Pederson,T. (2001) Fluorescent RNA cytochemistry: tracking genetranscripts in living cells. Nucleic Acids Res., 29, 1013–1016.

12. Bassell,G.J., Oleynikov,Y. and Singer,R.H. (1999) The travels of mRNAsthrough all cells large and small. FASEB J., 13, 447–454.

13. Dirks,R.W., Molenaar,C. and Tanke,H.J. (2001) Methods for visualizingRNA processing and transport pathways in living cells. Histochem. CellBiol., 115, 3–11.

14. Ainger,K., Avossa,D., Diana,A.S., Barry,C., Barbarese,E. andCarson,J.H. (1997) Transport and localization elements in myelin basicprotein mRNA. J. Cell Biol., 138, 1077–1087.

15. Rook,M.S., Lu,M. and Kosik,K.S. (2000) CaMKIIalpha 3′ untranslatedregion-directed mRNA translocation in living neurons: visualization byGFP linkage. J. Neurosci., 20, 6385–6393.

16. Beach,D.L., Salmon,E.D. and Bloom,K. (1999) Localization andanchoring of mRNA in budding yeast. Curr. Biol., 9, 569–578.

17. Bertrand,E., Chartrand,P., Schaefer,M., Shenoy,S.M., Singer,R.H. andLong,R.M. (1998) Localization of ASH1 mRNA particles in living yeast.Mol. Cell, 2, 437–445.

18. Tsuji,A., Koshimoto,H., Sato,Y., Hirano,M., Sei-Iida,Y., Kondo,S. andIshibashi,K. (2000) Direct observation of specific messenger RNA in asingle living cell under a fluorescence microscope. Biophys. J., 78,3260–3274.

Figure 6. Combined detection of RNA and expression of GFP in living cells.(A–D) Co-localization of poly(A) RNA and ASF–GFP in living U2OS cellsin speckles and nucleoplasm. (E–H) Co-localization of U1 snRNA andfibrillarin–GFP in coiled bodies. Arrows in (F) indicate Cajal Bodies contain-ing fibrillarin–GFP. Cells expressing ASF–GFP (B) were microinjected witha probe for poly(A) RNA (A) and cells expressing fibrillarin–GFP (F) with aprobe for U1 snRNA (E).

PAGE 9 OF 9 Nucleic Acids Research, 2001, Vol. 29, No. 17 e89

19. Sokol,D.L., Zhang,X., Lu,P. and Gewirtz,A.M. (1998) Real timedetection of DNA.RNA hybridization in living cells. Proc. Natl Acad.Sci. USA, 95, 11538–11543.

20. Politz,J.C., Tuft,R.A., Pederson,T. and Singer,R.H. (1999) Movement ofnuclear poly(A) RNA throughout the interchromatin space in living cells.Curr. Biol., 9, 285–291.

21. Paillasson,S., Robert-Nicoud,M. and Ronot,X. (1996) Specific detectionof RNA molecules by fluorescent in situ hybridization in living cells.Cell Biol. Toxicol., 12, 359–361.

22. Carmo-Fonseca,M., Pepperkok,R., Sproat,B.S., Ansorge,W.,Swanson,M.S. and Lamond,A.I. (1991) In vivo detection of snRNP-richorganelles in the nuclei of mammalian cells. EMBO J., 10, 1863–1873.

23. Pitts,A.E. and Corey,D.R. (1998) Inhibition of human telomerase by2′-O-methyl-RNA. Proc. Natl Acad. Sci. USA, 95, 11549–11554.

24. Majlessi,M., Nelson,N.C. and Becker,M.M. (1998) Advantages of2′-O-methyl oligoribonucleotide probes for detecting RNA targets.Nucleic Acids Res., 26, 2224–2229.

25. Matsuo,T. (1998) In situ visualization of messenger RNA for basicfibroblast growth factor in living cells. Biochim. Biophys. Acta, 1379,178–184.

26. Tyagi,S. and Kramer,F.R. (1996) Molecular beacons: probes thatfluoresce upon hybridization. Nat. Biotechnol., 14, 303–308.

27. Tyagi,S., Bratu,D.P. and Kramer,F.R. (1998) Multicolor molecularbeacons for allele discrimination. Nat. Biotechnol., 16, 49–53.

28. Luhrmann,R., Kastner,B. and Bach,M. (1990) Structure of spliceosomalsnRNPs and their role in pre-mRNA splicing. Biochim. Biophys. Acta,1087, 265–292.

29. Boom,R., Geelen,J.L., Sol,C.J., Raap,A.K., Minnaar,R.P., Klaver,B.P.and van der Noordaa,J. (1986) Establishment of a rat cell line inducible forthe expression of human cytomegalovirus immediate-early gene productsby protein synthesis inhibition. J. Virol., 58, 851–859.

30. van de Corput,M.P., Dirks,R.W., van Gijlswijk,R.P., van de Rijke,F.M.and Raap,A.K. (1998) Fluorescence in situ hybridization usinghorseradish peroxidase-labeled oligodeoxynucleotides and tyramidesignal amplification for sensitive DNA and mRNA detection.Histochem. Cell Biol., 110, 431–437.

31. Dirks,R.W., de Pauw,E.S. and Raap,A.K. (1997) Splicing factorsassociate with nuclear HCMV-IE transcripts after transcriptionalactivation of the gene, but dissociate upon transcription inhibition:evidence for a dynamic organization of splicing factors. J. Cell Sci., 110,515–522.

32. Dirks,R.W., van de Rijke,F.M., Fujishita,S., van der,P.M. and Raap,A.K.(1993) Methodologies for specific intron and exon RNA localization incultured cells by haptenized and fluorochromized probes. J. Cell Sci., 104,1187–1197.

33. Sleeman,J.E. and Lamond,A.I. (1999) Nuclear organization of pre-mRNAsplicing factors. Curr. Opin. Cell Biol., 11, 372–377.

34. Tollervey,D. and Kiss,T. (1997) Function and synthesis of small nucleolarRNAs. Curr. Opin. Cell Biol., 9, 337–342.

35. Huang,S., Deerinck,T.J., Ellisman,M.H. and Spector,D.L. (1994) In vivoanalysis of the stability and transport of nuclear poly(A)+ RNA. J. CellBiol., 126, 877–899.

36. Fay,F.S., Taneja,K.L., Shenoy,S., Lifshitz,L. and Singer,R.H. (1997)Quantitative digital analysis of diffuse and concentrated nucleardistributions of nascent transcripts, SC35 and poly(A). Exp. Cell Res.,231, 27–37.

37. Chin,D.J., Green,G.A., Zon,G., Szoka,F.C.,Jr and Straubinger,R.M.(1990) Rapid nuclear accumulation of injectedoligodeoxyribonucleotides. New Biol., 2, 1091–1100.

38. Leonetti,J.P., Mechti,N., Degols,G., Gagnor,C. and Lebleu,B. (1991)Intracellular distribution of microinjected antisense oligonucleotides.Proc. Natl Acad. Sci.USA, 88, 2702–2706.

39. Lamond,A.I. and Carmo-Fonseca,M. (1993) Localisation of splicingsnRNPs in mammalian cells. Mol. Biol. Rep., 18, 127–133.

40. Carmo-Fonseca,M., Pepperkok,R., Carvalho,M.T. and Lamond,A.I.(1992) Transcription-dependent colocalization of the U1, U2, U4/U6, andU5 snRNPs in coiled bodies. J. Cell Biol., 117, 1–14.

41. Lawrence,J.B., Singer,R.H. and Marselle,L.M. (1989) Highly localizedtracks of specific transcripts within interphase nuclei visualized by in situhybridization. Cell, 57, 493–502.

42. Sixou,S., Szoka,F.C.,Jr, Green,G.A., Giusti,B., Zon,G. and Chin,D.J.(1994) Intracellular oligonucleotide hybridization detected byfluorescence resonance energy transfer (FRET). Nucleic Acids Res., 22,662–668.

43. Fisher,T.L., Terhorst,T., Cao,X. and Wagner,R.W. (1993) Intracellulardisposition and metabolism of fluorescently-labeled unmodified andmodified oligonucleotides microinjected into mammalian cells.Nucleic Acids Res., 21, 3857–3865.

44. Meunier,L., Mayer,R., Monsigny,M. and Roche,A.C. (1999) The nuclearexport signal-dependent localization of oligonucleopeptides enhances theinhibition of the protein expression from a gene transcribed in cytosol.Nucleic Acids Res., 27, 2730–2736.

45. Sei-Iida,Y., Koshimoto,H., Kondo,S. and Tsuji,A. (2000) Real-timemonitoring of in vitro transcriptional RNA synthesis using fluorescenceresonance energy transfer. Nucleic Acids Res., 28, e59.

46. Iribarren,A.M., Sproat,B.S., Neuner,P., Sulston,I., Ryder,U. andLamond,A.I. (1990) 2′-O-alkyl oligoribonucleotides as antisense probes.Proc. Natl Acad. Sci. USA, 87, 7747–7751.

47. McKay,R.A., Miraglia,L.J., Cummins,L.L., Owens,S.R., Sasmor,H. andDean,N.M. (1999) Characterization of a potent and specific class ofantisense oligonucleotide inhibitor of human protein kinase C-alphaexpression. J. Biol. Chem., 274, 1715–1722.

48. Elayadi,A.N., Demieville,A., Wancewicz,E.V., Monia,B.P. andCorey,D.R. (2001) Inhibition of telomerase by 2′-O-(2-methoxyethyl)RNA oligomers: effect of length, phosphorothioate substitution and timeinside cells. Nucleic Acids Res., 29, 1683–1689.

49. Nielsen,P.E., Egholm,M. and Buchardt,O. (1994) Peptide nucleic acid(PNA). A DNA mimic with a peptide backbone. Bioconjug. Chem., 5, 3–7.

50. Fang,X., Li,J.J. and Tan,W. (2000) Using molecular beacons to probemolecular interactions between lactate dehydrogenase and single-strandedDNA. Anal. Chem., 72, 3280–3285.

51. Sixou,S., Szoka,F.C.J., Green,G.A., Giusti,B., Zon,G. and Chin,D.J.(1994) Intracellular oligonucleotide hybridization detected byfluorescence resonance energy transfer (FRET). Nucleic Acids Res., 22,662–668.