In a previous report [15], we investigated the localization and interaction of fluorescently labeled HuR and p37AUF1 proteins. Herein, we expand this examination to include KSRP and TIA-1, both exogenous and endogenous, thereby increasing substantially the number of permutations of interacting proteins to be examined. As was shown previously, and now in Figures 1 and 2, under basal conditions, HuR and AUF1 are predominantly nuclear, a pattern consistent for both endogenous and exogenously produced proteins. Under basal conditions, KSRP expression is also predominantly nuclear, a result consistent for both endogenous and exogenous tagged proteins [13,36], Figures 1 and 2. As documented previously [37-39], and now herein, in unstressed cells, TIA-1 is expressed uniformly throughout the nucleus and cytoplasm, Figures 1 and 2. Interestingly, all four ARE-BPs appear to be excluded from the nucleoli. Localization and interaction of all RNA-BPs in response to various stimuli is summarized in Table 1.

Cytoplasmic SGs have been described as sites of triage for RNA transcripts when cells are undergoing stress [40-43]. It is generally accepted that SGs contain proteins associated with mRNA translation initiation, translation control, and to some degree, with mRNA decay [44]. Upon initiation of cellular stress, these proteins, as well as a vaguely described subset of mRNA molecules, rapidly relocalize to SGs. SGs have also been shown to be dynamic structures, originating as smaller punctate bodies that can ultimately fuse to form larger, irregularly shaped structures [45]. They have also been shown to be in dynamic contact, exchanging components with PBs [43].

As demonstrated by the relocalization of TIA-1, induction of oxidative stress with arsenite induces the rapid formation of cytoplasmic SGs [15]. Similarly, herein, TIA-1 has been shown to rapidly relocate to SGs in DDT1-MF2 cells within 30 minutes post arsenite treatment, Figures 1 and 2, Panels C, D and E, Row 2 (of each panel). HuR also rapidly relocalizes to SGs [46], Figures 1 and 2, Panels A, B and C, Row 2 (of each panel). Unlike TIA-1, HuR is likely relocalizing from the nucleus to the cytoplasm and then to SGs whereas TIA-1 is already in the cytoplasm prior to the initiation of oxidative stress. Further, we demonstrate that KSRP and AUF1 relocalize from the nucleus to cytoplasmic SGs with rapid time frame, similar to that of HuR [14], Figure 1, Panels B, D and F, Row 2 for KSRP and Figure 1, Panels A, E and F, Row 2 for AUF1. In images of arsenite treated cells, it should be readily apparent that although TIA-1, HuR, AUF1 and KSRP all populate SGs, the proportion of each protein that relocalizes to SGs varies considerably with the relocalization of TIA-1 being the most robust. Based on relative fluorescent intensity, the concentration of endogenous TIA-1 present in SGs, post arsenite, is noticeably greater, in general, than in either the nucleus or cytoplasm. Although still marked, HuR shuttles to a lesser degree than TIA-1 with more HuR protein remaining in the nucleus than is present in the SGs. For KSRP and AUF1, only a small portion of total cellular protein relocalizes to SGs.

Figure 1 also documents the interaction of ARE-BPs present within SGs. Panel A, Row 2 shows the colocalization and interaction of HuR and AUF1 in DDT1-MF2 cells after treatment with arsenite. The merged view shows that HuR is likely to be at a higher concentration in SGs than is AUF1, whereas the two proteins are more or less equally abundant in the nucleus, as demonstrated previously by western blotting [15]. The FRETc image shows that HuR and AUF1 are in stable contact in the nucleus and in SGs. As the peak intensity of the FRETc signal is proportional to the proximity and amount of fluorophors present, a lower intensity FRETc signal from the presumably lesser quantity of AUF1 present in SGs is to be expected (i.e., the concentration of AUF1 is limiting).

Figure 1, Panel B, Row 2 shows the colocalization and interaction of KSRP and HuR. As stated previously, both HuR and KSRP are present dominantly in the nucleus, to a small extent in the cytoplasm, and under conditions of oxidative stress, in SGs. In the merged view, it is easy to see that KSRP and HuR colocalize in the nucleus and to a smaller extent the cytoplasm. Colocalization within SGs is much less readily detectable. This, again, is presumably due to the much lower concentration of KSRP in SGs compared to that of HuR. The FRETc view demonstrates that there is close and stable interaction between HuR and KSRP in the nucleus and to a small extent in cytoplasm and SGs. That being said, the SG FRETc signal is quite distinct if image intensity is adjusted to permit visualization.

Figure 1, Panel C, Row 2 shows the colocalization and interaction of TIA-1 and HuR in arsenite treated cells. Both TIA-1 and HuR are present in the nucleus and in SGs. Under these conditions, TIA-1 is present in SGs and cytoplasm to a much greater extent than is HuR. In the merged view, it is readily apparent that HuR is dominant in the nucleus and TIA-1 dominant in SGs. A FRETc signal is present in the nucleus but is much stronger in the SGs and is indicative of close and stable contact of proteins in high relative abundance.

Figure 1, Panel D, Row 2 shows the localization and interaction of KSRP and TIA-1. Under stress conditions, both KSRP and TIA-1 relocalize to SGs. The merged image shows that, relatively speaking, more KSRP is in the nucleus and more TIA-1 is in SGs. The FRETc image shows a close and stable contact between KSRP and TIA-1 in SGs and nucleus.

Figure 1, Panel E, Row 2 shows the colocalization and interaction of AUF1 and TIA-1. Both TIA-1 and AUF1 are present in nucleus and SGs. The merged view shows that the TIA-1 signal intensity is far greater in SGs than is AUF1 but that they are both readily detectable within the nucleus. A FRETc signal is present in the nucleus and in SGs and shows the close and stable interaction of TIA-1 and AUF1 in both of these locations.

Figure 1, Panel F, Row 2 shows the localization and interaction of KSRP and AUF1. Both KSRP and AUF1 are present in the nucleus and only a small portion of either of these proteins moves to SGs. The merged view shows the colocalization of KSRP and AUF1 to nucleus and SGs. A FRETc signal is present in both the nucleus and SGs indicating that these proteins are in a close and stable interaction.

In previous experiments, as well as in the current Figure 2 representing exogenous proteins, we were unable to detect AUF1 in SGs [15]. Supporting our previous conclusion was an absence of literature describing the localization of AUF1 to SGs. On average, the AUF1 signal intensity present in SGs is about 1/10th of the maximum fluorescence intensity present in the cell, with the most intense signal always being within the nucleus. Increasing the overall signal intensity allowed the detection of a weak fluorescence signal for AUF1 when present in SGs (Figure 1). It should be noted that our previously reported results were based solely on expression of the p37 splice form of AUF1. In contrast, under the currently used experimental conditions, immunostaining of endogenous AUF1 protein detects all 4 splice-forms (37, 40, 42 and 45), the results therefore being representative of the totality of intracellular AUF1 protein.

Pertaining to KSRP, the reason(s) underlying the variability of exogenous protein expression in SG is currently unknown. Two potential explanations might be level of KSRP protein expression or epitope tagging of the protein. The non-uniformity of response between exogenous and endogenous protein demonstrates the utility of immuno-FRET over transient transfection in studying cellular protein localization.

PBs are generally recognized as sites associated with RNA degradation. They are distributed throughout the cytoplasm and contain numerous proteins involved in RNA decay [32,47,48]. Interestingly, PBs appear to be constitutively present in mammalian cells but need to be induced in yeast [43]. In the current studies, we were interested in determining whether or not ARE-BPs colocalize to PBs and if so, could interactions of proteins generally assumed to be in these intracellular compartments, be detected. Figure 3, Panel A, Row 1 details the intracellular localization of HuR and Hedls. As before, in unstimulated cells, HuR is localized predominantly to the nucleus whereas Hedls is localized diffusely in the cytoplasm as well as in obvious PB structures. The merged view shows that there is no colocalization of HuR and Hedls, a finding concordant with a lack of detectable FRET. A similar result is found for the (lack of) interaction between KSRP and Hedls, Figure 3, Panel B, Row 1.

Figure 3, Panel C, Row 1 details the results for interactions between TIA-1 and Hedls. TIA-1 demonstrates a nuclear and cytoplasmic distribution that includes punctuate bodies in the cytoplasm whereas Hedls is present in both PBs and generally in the cytoplasm. Interestingly and importantly, the FRETc view demonstrates a unique interaction between TIA-1 and Hedls delimited to PBs. This finding is concordant with previous descriptions delineating the dynamic nature of PBs and their potential for interaction (protein exchange) with SGs [49].

To investigate the potential for interaction of proteins compartmentalized to SGs and PBs, SGs were first induced by oxidative stress. Figure 3 indicates that the morphology of PBs, as identified by the cellular localization of Hedls, is unaffected by oxidative stress or by the formation of cytoplasmic SGs. This result has been recapitulated using Dcp1a as another marker of PBs [50], data not shown. Further, oxidative stress induced relocalization of ARE-BPs to SGs but did not alter the interaction of each specific protein with Hedls when present in PBs. Interaction between SGs and PB proteins can be seen in the merged view of Figure 3, Panel C, Row 2. To examine this in greater detail, the interaction of SGs and PBs is seen via 3D reconstruction collected along the Z-axis, (Additional file 1). The 3D reconstruction of KSRP/Hedls interaction demonstrates that SGs and PBs are not cytoplasmic spheres but rather, elongated oval structures with substantial depth throughout the cytoplasm. Of note is the observation that more than one PB can be in contact with a single SG. This image also shows that an observed protein/protein colocalization does not necessarily lead to a FRET detectable interaction. Figure 3 Panel C Row 2 demonstrates a FRET signal is generated for the interaction of TIA-1 and Hedls. This result strongly supports the conclusion that the TIA-1/Hedls interaction is specific and not secondary to a packing artifact.

TIA-1 was detected in PBs during both unstimulated and MAPK stimulated conditions. In unstimulated cells, a portion of the cytoplasmic TIA-1 appears to be present in PBs leading to the generation of a FRET signal between TIA-1 and Hedls. In stressed cells, TIA-1 is present in both SGs and PBs. In this case, the strongest FRET signal was generated at the interface of TIA-1 containing SGs and Hedls containing PBs. Why TIA-1, along with TTP [43,51] and BRF1, [43] are present in PBs, but HuR, AUF1 and KSRP are not, is currently unknown. The potential for interaction between AUF1 and Hedls (or Dcp1a) was unable to be investigated, as the only antibodies available to detect these proteins were rabbit polyclonal.

Stimulation of MAP kinase pathways has the biochemical effect of stabilizing ARE containing RNAs [52-54]. MAPK stimulation also has the effect of intracellular redistribution of several ARE binding proteins. Upon activation, both HuR and AUF1 are relocalized from the nucleus to a combined nuclear/diffuse cytoplasmic distribution, Figures 4 and 5. However, as shown previously [15], the time course of redistribution for the two proteins is markedly different. When stimulated with anisomycin, HuR shuttles from nucleus to cytoplasm in < 90 minutes whereas movement of AUF1 requires on the order of 24 hours (both visually and by Western Blot). This disparate time frame may underlie the biological plausibility of the opposing nature of these ARE-BPs, that is, the rapid stabilizing influence of HuR on ARE containing RNAs.

In the current study, we extended this exploration to examine the effects of MAPK stimulation on KSPR and TIA-1. As TIA-1 possesses a general cytoplasmic distribution in unstimulated cells, we speculated that MAPK stimulation might not have a discernable effect on its distribution and indeed, it does not, Figures 4 and 5, Panels C, D and E. Figure 4 summarizes the results of pair-wise endogenous protein/protein interactions following MAPK activation by anisomycin treatment of DDT1-MF2 cells. In each case, the upper panel is representative of control, vehicle treated cells, whereas the bottom panel is representative of anisomycin treated cells (24 hr exposure to 75 nM). As can be seen, in Figure 4 Panels B, D, and F, MAPK activation induces nucleocytoplasmic shuttling of KSRP. In both the merged and FRET image panels, relocalization of endogenously expressed protein pairs to the cytoplasm demonstrates FRET between KSRP/TIA-1, KSRP/HuR, and KSRP/AUF1. Comparing Figure 4 to Figure 5, variability exists in the response of endogenous versus exogenous KSRP to MAPK activation. No shuttling of exogenous KSRP is seen due to MAPK stimulation, compared to readily detectable shuttling of the endogenous proteins.

Using FRETc, we are able to demonstrate that all of the ARE-BPs studied interact in all the pair-wise combinations tested when localized to the nucleus and in the cytoplasm when relocalization is due to MAPK stimulation. These results are supportive of the notion that ARE-BPs are bound to multiple AREs as part of multiprotein complexes. One implication of this conclusion is that fine-tuning of the half-life of an RNA molecule is almost certainly tightly regulated by a large number of factors and conditions that remain to be elucidated.

Figure 6 examines the effect of MAPK stimulation on PBs and the localization of the ARE-BPs currently being investigated. As MAPK stimulation and inhibition has been shown to reciprocally stabilize and destabilize a number of ARE containing RNAs [55], and degradation of mRNA has been associated with protein components of SGs and PBs [32,43,48,56], we hypothesized that certain ARE-BPs may differentially localize to PBs. As was shown in Figures 4 and 5, upon MAPK stimulation, HuR, AUF1 and KSRP all move from the nucleus to the cytoplasm with the extant cytoplasmic localization of TIA-1 being unaffected. Figure 6, Panels A, B and C, demonstrate that the morphology and location of PBs, as defined by Hedls, appears entirely unaffected by MAPK stimulation. We have also shown this to be true for Dcp1a (data not shown).

In Figure 6, Panel A, HuR and Hedls did not colocalize nor was FRET in evidence. Similarly, Figure 6, Panel B demonstrates that MAPK activation did not cause KSRP and Hedls to interact. However, Figure 6, Panel C shows that the Hedls/TIA-1 do interact, a result consistent with that shown in Figure 3, Panel C. As the cellular localization of both the TIA-1 and Hedls are apparently unaffected by MAPK stimulation, it follows that their interaction may also unaltered by this stimulation. Thus, TIA-1 and Hedls show a FRETc detected interaction in both unstimulated and stimulated conditions.

Direct investigation of ARE-containing RNA molecules allows us to begin to interconnect data for ARE-BP localization with specific RNAs. Currently, overcoming the obstacle of visualizing RNA molecules with short half-lives can be approached in one of two ways: plasmid based expression of RNAs encoding stem loops recognized by the MS2-coat protein [57] or in vitro transcription and transfection of fluorescently labeled RNAs with or without 2'F (dUTP) chemical stabilization. For this set of experiments, we chose the latter approach. To visualize target RNA, full length, 2'F-dUTP modified, 5' capped, polyadenylated, hamster β₂ adrenergic receptor (β-AR) RNA was in vitro transcribed, utilizing an appropriate stoichiometry of Cy3-labeled UTP. Importantly, by nondenaturing gel shift assay, the fully modified RNA was able to bind ARE-BPs with normal (low nM) affinity (data not shown). As is evident in Figure 7, the transfected Cy3-labeled, ribonuclease resistant RNA accumulated in punctate cytoplasmic structures. We hypothesized that since 2'F-modified RNA is generally not translationally competent, it may undergo sequestration into translationally silent compartments, i.e., SGs or PBs. Nonetheless, we wished to determine whether or not the ARE was accessible and would bind ARE-BPs under resting and stimulated conditions. Figure 7, Panel A, Row 1 shows a DDT1-MF2 cell containing the Cy3-labeled RNA, which was immunostained for Hedls and HuR. The merged view demonstrates that Hedls colocalizes with the Cy3-labeled RNA as well as being present in small, punctate PB-like structures. In this instance, no FRET data was able to be collected due to a disparity in fluorescent signal intensities.

Figure 7, Panel A, Row 2 shows the effect of oxidative stress on the Cy3-labeled RNA in conjunction with Hedls and HuR. Image 1 shows the mRNA by itself. Image 2 confirms that oxidative stress does not affect Hedls localization. Image 3 shows that HuR readily relocalizes to SGs in the presence of the introduced mRNA. Image 4 shows the merged view with colocalization of Hedls and the Cy3-labeled mRNA. Image 5 shows the FRET detectable interaction between Hedls and the mRNA.

We speculate that what underlies this result is the ribonuclease resistant RNA is attracting or sequestering Hedls protein. Figure 7, Panel B shows the same experiment but this time with the PB component, Dcp1a, which was detected by immunostaining (in addition to HuR). Similar results were obtained as outlined in Figure 7, Panel A; Dcp1a is present in the cytoplasm in PBs but also colocalizes with Cy3-labeled RNA. Image 4 of Row 2, Panel B shows the merged view. HuR is present in SGs but Dcp1a and the RNA are not. The RNA and Dcp1a are in close and stable contact as indicated by the FRETc image. This result, consistent with that found with Hedls, shows that Dcp1a recognized and interacted with the stabilized mRNA molecule in a location associated with RNA degradation.

Construction of the HuR-eYFP, HuR-eCFP, p37AUF1-eCFP and p37AUF1-eYFP vectors was as described previously [15]. An Xho1/Kpn1 fragment of KSRP was subcloned into pECFP.C1 (Clonetech) yielding the KSRP-eCFP vector construct. Similarly a Bgl II/EcoRI fragment of the variant 1 splice form of TIA-1 [86] was subcloned into pEYFP.C1 (Clonetech) yielding TIA-1-eYFP, (plasmid was courtesy of Dr. P. Anderson). In all constructs the fluorophor is N-terminal to the protein of interest.

DDT1-MF2 Hamster smooth muscle cells were grown in 100 mm dishes at 37°C, 5% CO₂ in Dubelcco's Modified Eagle Medium (Invitrogen) supplemented with 5% Fetal Bovine Serum (Hyclone) and 0.1% penicillin/streptomycin solution (Invitrogen). For transient transfection DDT1-MF2 cells were grown on 25 mm glass cover slips to approximately 80% confluence. For immunocytochemistry DDT1-MF2 cells were grown on 18 mm glass cover slips to approximately 60% confluence. Where indicated, cells were treated with anisomycin (75 nM) or sodium arsenite (0.5 mM).

Vectors were transfected into DDT1-MF2 using Fugene 6 (Roche) following the manufacturers protocol. The amount of plasmid DNA transfected was adjusted empirically to yield the desired 1:1 molar ratio of the CFP to YFP protein.

DDT1-MF2 cells were fixed with 4% paraformaldehyde (Electron Microscopy Sciences), permeabilized with 0.5% Triton X-100 in PHEM buffer (60 mM PIPES, 25 mM Hepes, 10 mM EGTA, 2 mM MgCl₂, pH 6.9), treated with 0.1% sodium borohydride (Sigma) and blocked with 5% horse serum. Cells were incubated at 4°C, overnight with 1°Ab and in 2°Ab for 1 hour at room temperature in the dark. Cover slips were mounted on glass slides using a 12.5% solution of Mowiol (Calbiochem) in 50% glycerol. Detection of HuR was via a 1:400 dilution of monoclonal 1°Ab (Santa Cruz Biotechnology) or a 1:200 dilution of a rabbit polyclonal 1°Ab; detection of AUF1 was via a 1: 500 dilution of a polyclonal 1°Ab (Upstate) which recognizes all 4 isoforms of AUF1; detection of TIA-1 was via a 1:500 dilution of goat polyclonal 1°Ab (Santa Cruz Biotechnology); detection of KSRP was via a 1:1000 dilution of a mouse monoclonal 1°Ab (courtesy of Dr. D.L. Black) and detection of Hedls was via a 1:8000 dilution of a rabbit polyclonal 1°Ab (courtesy of Dr. J. Lykke-Anderson). Secondary antibodies used were either a 1:200 dilution of goat anti-rabbit 2°Ab labeled with Cy5; a 1:400 dilution of a goat anti-rabbit 2°Ab labeled with Cy3; a 1:800 dilution of goat anti-mouse 2°Ab labeled with Cy5; a 1:1000 dilution of goat anti-mouse 2°Ab labeled with Cy3; a 1:2000 dilution of goat anti-mouse 2° Ab labeled with Alexa Fluor 488; a 1:200 dilution of a donkey anti-goat Cy3 or a 1:400 dilution of a donkey anti-goat 2°Ab labeled with Alexa Fluor 647. With the exception of TIA-1, all 1° and 2° Abs were applied together. To prevent cross-reaction between 2°Ab, the TIA-1 donkey anti-goat 2°Ab was added first and separately. All Cy3 and Cy5 labeled 2° antibodies were F(ab')₂ IgG fragments from Jackson Immuno Research; Alexa Fluor dye labeled antibodies were F(ab')₂ IgG fragments from Molecular Probes.

For detection of a FRET signal via labeled 2° Ab, DDT1-MF2 cells were processed for immunocytochemistry as indicated above. The fluorophors Cy3 and Cy5 or Alexa Fluor 647 (Invitrogen) conjugated to 2° Ab were used. In this instance Cy3 is the donor fluorophor and Cy5 or Alexa Fluor 647 is the acceptor fluorophor. In DDT1-MF2 cells, the signal intensity ratio of Cy3 to Cy5 required for energy transfer to occur and generate a FRET signal was determined to be 1:1 or greater (data not shown). The requirements for FRET; a FRET fluorophor pair, specific distance between fluorophors and fluorophor alignment, are not changed due to placing the fluorophor pair on 2° Ab.

Images of live and fixed cells were captured at 100× magnification under oil immersion, (lens numerical aperture 1.40), using 2 × 2 or 4 × 4 binning mode on an inverted Nikon Eclipse TE3000 fluorescence microscope using a Cooke SensiCam QE CCD camera running Slidebook software (3I, version 4.0.1.43). All images were collected at room temperature using an integration time greater than 25 ms. Additionally, all images were background corrected by subtracting the intensity of a cell "devoid" area from each individual image. For live cell images, imaging medium was DMEM and integration time was between 25 and 250 ms. For fixed cell and immunostained images, integration time was between 25 and 1000 ms. Live cells containing CFP and YFP labeled proteins were viewed via the CFP channel to prevent photobleaching of YFP. All images presented had a signal intensity greater than 100 after background correction. FRETc images were produced using Slidebook software (3I, version 4.0.1.43) using spectral bleed through numbers that had been experimentally determined as detailed below. Images were edited using Adobe Illustrator.

All experimental data was collected from a minimum of two separate experiments. For each immunocytochemical experiment, a minimum of 400 cells was examined. For each transfection experiment, a minimum of 200 cells was examined. In general, images were captured at a level that would provide best resolution of the nucleus.

The CFP and YFP spectral bleed through numbers were calculated as indicated previously [15]. Additionally, the spectral bleed through number of TIA-1-eYFP vector was determined. As a further proof of the bleed through numbers, and to show that the mere cellular presence of FRET pair fluorophors does not generate a FRET signal, the transient transfection of TIA-1-eYFP and the eCFP vector was carried out to add to the controls previously conducted for empirical correction of the spectral bleed through number [15]. The Cy3 and Cy5 spectral bleed through numbers were determined by imaging KSRP, AUF1, Dcp1a, Hedls, HuR and TIA-1 with Cy3-labelled 2°Ab; KSRP, AUF1, Dcp1a, Hedls and HuR with Cy5-labelled 2°Ab and TIA-1 with Alexa Fluor 647-labeled 2°Ab. The spectral bleed through number was also determined for the Cy3-labeled β-AR mRNA. As a further proof of the bleed through numbers, and a control for the FRET signal generation immunostaining of non interacting proteins with the FRET pair, Cy3-Cy5, was used for empirical correction of the spectral bleed through number; Cy5 labeled 2°Ab detection of KSRP, AUF1, Hedls with Cy3 labeled anti β-tubulin 1°Ab (Sigma) and Alexa Fluor 647-labelled 2°Ab detection of TIA-1 with Cy3 labeled anti-β-tubulin 1°Ab (Sigma). A similar control was conducted for the Cy3-labeled mRNA with transiently transfected TTP-GFP, AUF1-GFP and the GFP vector. FRET positive control were conducted using Cy3 and Cy5 2°Ab staining to a single 1°Ab recognizing KSRP, AUF1, Dcp1a, Hedls, HuR and TIA-1.

To calculate FRETc, the following equation was used:

FRETc = raw FRET − ( D * X ) − ( A * Y )

FRETc is the corrected FRET value, raw FRET is the uncorrected FRET intensity, D is the donor fluorophor intensity, X is the donor fluorophor spectral bleed through number, A is the acceptor fluorophor intensity, and Y is the acceptor fluorophor spectral bleed through number.

To create ribonuclease resistant RNA labeled with a fluorescent Cy3 moiety, we incorporated Cy3-UTP (Amersham) and 2'F substituted CTP and UTP (Epicenter Biotechnologies) into β1 Adrenergic Receptor mRNA molecules via T7 RNA polymerase (Epicenter Technologies). We altered previously published methods [87], to incorporate a 5' m⁷G cap analogue (Promega) and 3' polyadenylation to the stabilized, fluorescent RNA molecule. A DNA template (1 μg) with an incorporated T7 promoter sequence was transcribed using Durascribe T7 enzyme (2 μl) in a 20 μl reaction containing 5 mM ATP, 1 mM GTP, 5 mM 2'F-dCTP, 5 mM 2'F-dUTP, 0.75 mM Cy3-UTP, 10 mM DTT and 4 mM m⁷G cap analogue for 2 hours at 37°C. Polyadenylation was carried out subsequent to the in vitro transcription. After removal of a small aliquot (1 μl) for gel electrophoresis, 3.125 mM MnCl₂, 1.25 mM ATP, 2 ul E-PAP enzyme, 10 ul 5× E-PAP buffer and 1 μl SUPERase, (Ambion) were added directly to the in vitro transcription reaction tube, total volume 40 μl, incubate 2 hours at 37°C. This reaction creates a poly A tail of ~100-200 nucleotides. Based on a calculation of the base:dye ratio (i.e., UTP to Cy3-UTP ratio), the capped and polyadenylated Cy3-RNA products contain ~2-4 Cy3 molecules per transcript. The optimal concentration for observation of diffuse Cy3 signal in cells has been determined empirically to be ~200 pM of RNA when DDT1-MF2 cells were transfected using Lipofectamine 2000, as per the manufacturer's guidelines (Invitrogen).

Cy3-labeled RNAs were transfected into DDT1-MF2 cells using Lipofectamine 2000 (Invitrogen). Briefly, DDT1-MF2 cells were resuspended and plated at 80% confluence onto 100 mm plates containing 25 mm cover slips and allowed to adhere, about 2 hours. The RNA was melted at 90°C for 1 min and allowed to cool slowly to room temperature prior to transfection. The Lipofectamine 2000 and Cy3-labeled RNA (2 μg) were diluted separately into Optimem 1 media (4 fold and 140 fold respectively) (Invitrogen) and incubated at room temperature for 30 minutes. DDT1-MF2 cells were washed with Optimem 1 and changed into this media. The Lipofectamine 2000 and RNA solutions were mixed together and added to the media bathing the cells. The cells were incubated overnight at 37°C, 5% CO₂ prior to the media being changed back to DMEM (37°C). Cells were left to recover for 4 hours prior to any subsequent treatment.