SDE5, a putative RNA export protein, participates in plant innate immunity through a flagellin-dependent signaling pathway in Arabidopsis

In eukaryotes, RNA silencing, mediated by small interfering RNAs, is an evolutionarily widespread and versatile silencing mechanism that plays an important role in various biological processes. Increasing evidences suggest that various components of RNA silencing pathway are involved in plant defense machinery against microbial pathogens in Arabidopsis thaliana. Here, we show genetic and molecular evidence that Arabidopsis SDE5 is required to generate an effective resistance against the biotrophic bacteria Pseudomonas syringae pv. tomato DC3000 and for susceptibility to the necrotrophic bacteria Erwinia caratovora pv. caratovora. SDE5, encodes a putative mRNA export factor that is indispensable for transgene silencing and the production of trans-acting siRNAs. SDE5 expression is rapidly induced by exogenous application of phytohormone salicylic acid (SA), methyl jasmonate (MeJA), phytopathogenic bacteria, and flagellin. We further report that SDE5 is involved in basal plant defense and mRNA export. Our genetic data suggests that SDE5 and Nonexpressor of PR Gene1 (NPR1) may contribute to the same SA-signaling pathway. However, SDE5 over-expressing transgenic plant exhibits reduced defense responsive phenotype after flagellin treatment. Taken together, these results support the conclusion that SDE5 contributes to plant innate immunity in Arabidopsis.


Results
SDE5 gene expression is upregulated upon SA, MeJA and flg22 application and Pto DC3000 inoculation. To determine the expression of SDE5 during plant basal defense, wild-type (WT) seedlings were infiltrated with the virulent Pto DC3000 at 2 × 10 6 colony-forming units per mL (cfu mL −1 ) and SDE5 transcript levels were then analyzed using quantitative reverse transcription-polymerase chain reaction (qRT-PCR) at different time points. As shown in Fig. 1A, SDE5 expression was significantly up-regulated at 3 hours post inoculation (hpi). This upregulation was transient as gene expression returned to resting levels by 12 hpi.
To analyze the signaling pathway that leads to SDE5 expression, we examined pathogen-induced changes of SDE5 transcript levels in WT, as well as in mutants altered in the production of SA (sid2), JA signaling (jar1) and ethylene perception (ein2) following pathogen Pto DC3000 challenge. We also considered mutation in NPR1, one of the central downstream regulators of SA signaling. As shown in Fig. 1B, the Pto DC3000-induced expression of SDE5 is unchanged in ein2 suggesting that the induction of SDE5 by the virulent bacterial pathogen is independent on EIN2. Interestingly, the level of SDE5 expression was significantly reduced in sid2, jar1 and npr1, indicating dependency on SA-and JA-signaling (Fig. 1B). The contribution of two PAMP receptors (FLS2 and EFR), involved in perception of flg22 and the elongation factor EF-Tu, respectively, to SDE5 expression were evaluated using their respective mutants. The Pto DC3000-induced expression of SDE5 was also reduced in the PAMP receptor-defective mutants (fls2 and efr) indicating a contribution of PTI (Fig. 1B). To further validate the involvement of SA and JA in the induction of SDE5, we tested WT plants 3 h after spraying with SA and MeJA and observed that SDE5 expression is 2-2.5-fold up-regulated (Fig. 1C).
The results obtained in Fig. 1B and previous reports 27, 28 , suggest a critical involvement of the FLS2-dependent signaling pathway in control of SDE5 expression. To confirm the induction of SDE5 during PTI, we next monitored the mRNA levels of SDE5 over a 6 h time course experiments. We found a significant induction of SDE5 expression at 3 h after flg22 peptide treatment (Fig. 1D). As expected, expression was not induced by flg22 in the fls2 mutant indicating that an important role of FLS2 in flg22 mediated SDE5 induction. Taken together, SDE5 expression data indicate a potential role for SDE5 in SA signaling, JA signaling and PTI in Arabidopsis.
Disruption of SDE5 decreases plant basal defense. In order to further elucidate the possible role of SDE5 in plant-pathogen interactions, a reverse genetic approach using mutant alleles of the SDE5 gene containing a T-DNA insertion has been performed. The T-DNA insertion is in the sixth exon of the SDE5 gene and resulted in the loss of detectable SDE5 transcript, indicating that sde5-3 26 is a loss-of function mutant. Another mutant line used in this work was sde5-2 which is known to be a partial loss-of-function mutant in mRNA export 25 . We also generated SDE5 overexpressing transgenic plants in the sde5-3 background using SDE5 coding sequence under the control of the Cauliflower mosaic virus-derived 35S promoter (Fig. S1). Two transgenic lines showing high expression of SDE5 were selected for further study and named OE-5 and OE-6. Then, we inoculated WT, sde5-2, sde5-3 and two overexpressing lines, OE-5 and OE-6, with a biotrophic pathogen, Pto DC3000, and monitored both bacterial growth (at 2 × 10 5 cfu mL −1 ) and disease symptom development (at 2 × 10 6 cfu mL −1 ). As shown in Fig. 2A, sde5-3 permitted nearly 10-fold more bacterial growth than the WT plants. The sde5-3 plants also developed significantly more severe SCIeNtIfIC RepoRts | 7: 9859 | DOI:10.1038/s41598-017-07918-x disease symptoms than WT plants at 4 dpi (Fig. 2B). However, bacterial growth and disease development in partial loss-of-function mutant sde5-2 were not significant compared with WT. As expected, this disease phenotype has been overcome in OE plants indicating that SDE5 may act as a positive regulator in plant innate immunity.
SA is a major plant defense hormone, central to the activation of a range of defenses including the induction of PR (pathogen-related) genes, systemic acquired resistance, and hypersensitive response 29 . Therefore, expression analysis of the well-known SA-dependent effector gene, PR1, was also performed in leaves of WT, sde5-3 mutant and OE-5 plants before and at various time points after inoculation with Pto DC3000. The result indicated that PR1 transcript accumulation was significantly reduced in sde5-3 compared with WT plants (Fig. 3A). Consistent with these findings, analysis of both loss-of-function SDE5 mutants and gain-of-function SDE5-overexpressing plants indicates that SDE5 contributes to SA mediated plant basal defense by modulating responses to bacterial strain Pto DC3000. The biosynthesis of SA is strongly induced upon pathogen infection. This pathogen-induced SA biosynthesis is believed to be controlled by several key components including PAD4, EDS5 and SID2 [30][31][32] . To further clarify the roles of SDE5 in SA-mediated basal defense, we examined the expression of PAD4 and SID2, in WT, sde5-3 mutant and OE-5 plants following Pto DC3000 inoculation. No significant difference was found between WT plants and other genotypes tested ( Fig. 3B and C), supporting the hypothesis that the effect of SDE5 on PR-1 expression is independent of expression of genes involved with SA production and signaling. To study the epistatic relationships between SDE5 and SA, we also treated SA before Pto DC3000 inoculation in Col-0, sde5-2, sde5-3, OE-5 and npr1, and monitored symptoms. Disease symptom of sde5-3 was not rescued by SA treatment suggesting that SDE5 is acting downstream of SA signaling rather than affecting SA biosynthesis (Fig. S2).
The SA-mediated signaling pathway regulated by NPR1 is one of the most important pathways in plant defense 33,34 . The observed down-regulation of SDE5 gene expression in npr1 mutant plants indicates that functional NPR1 protein is required for pathogen responsiveness of SDE5 (Fig. 1B). This result prompted us to investigate whether the accumulation of NPR1 transcript levels are altered in WT and sde5-3 mutant upon Pto DC3000 inoculation. However, no significant difference in the NPR1 expression level was observed between the plants tested, indicating that SDE5 may function downstream of NPR1 in SA-signaling (Fig. 3D). Moreover, when challenged with virulent pathogens, the sde5-3/npr1 double mutant was not more susceptible to Pto DC3000 than npr1 (Fig. 3E), indicating that SDE5 and NPR1 might contribute to the same SA-signaling pathway in Arabidopsis. The sde5-3 mutant confers elevated disease resistance to ECC. Since, SDE5 expression was induced by exogeneous MeJA treatment, we evaluated the contribution of SDE5 to plant responses to the necrotrophic pathogen, Erwinia caratovora pv. caratovora (ECC). We inoculated WT, sde5-2, sde5-3, OE-5 and OE-6 plants with ECC and observed bacterial accumulation at 4 dpi. As shown in Fig. 4A, sde5-3 mutant plants restricted the growth of ECC relative to WT plants, indicating that SDE5 promotes the growth of ECC. ECC susceptibility was restored to WT levels in the OE-5 and OE-6 plants, further confirming that expression of the SDE5 gene promotes infection by this necrotrophic bacterial (Fig. 4A). In an effort to further examine the role of SDE5 in JA-mediated defenses, we determined the expression levels of plant defensin gene PDF1.2, a characteristic molecular response of plants to necrotrophic pathogen attack in different time points after pathogen inoculation. The induction of PDF1.2 was significantly increased in sde5-3 plants following the inoculation with ECC (Fig. 4B). This observation is congruent with the observation that sde5-3 plants showed enhanced disease resistance to this pathogen. Collectively, these results indicate that SDE5 plays positive and negative roles in SA-and JA-mediated pathogen defense, respectively, possibly by participating in the cross-talk between these signaling pathways.

SDE5 is involved in PAMP triggered immunity in Arabidopsis.
SA is required for the full activation of PTI 35,36 . To better understand the contribution of SDE5 in PTI, we utilized the type-three secretion system (TTSS)-defective mutant Pto DC3000 hrcC − (Pto hrcC − ) strain, which can elicit, but not suppress, PTI responses due to its inability to inject effector proteins within host cells 37 . The behavior of WT, sde5-2, sde5-3, OE-5 and OE-6 plants was analyzed following foliar inoculation with the Pto hrcC − strain, and, as expected, limited bacterial growth was observed in WT plants compared to the fully virulent bacteria following inoculation at 2 × 10 5 cfu mL −1 . No significant differences were detected between WT and the sde5 mutant lines (Fig. 5A). In contrast, OE-5 and OE-6 plants were more susceptible to Pto hrcC − relative to the WT (Fig. 5A). We next examined the expression pattern of PR1 in response to Pto hrcC − inoculation in various genotype plants. As shown in Fig. 5B, PR1 gene expression was highly induced in sde5-3 plants with significant up-regulation at 24 hpi in comparison to other plants. These results reveal an opposite regulation of PR1 gene expression by SDE5 following infection with virulent Pto DC3000 versus the TTSS-deficient Pto hrcC − strain. These data further support the hypothesis that the responses observed with Pto DC3000 are TTSS-dependent and therefore may involve the activity of bacterial effectors in plant cells. Thus, SDE5 restricts susceptibility to virulent Pto DC3000, but has a less discernible effect on resistance to Pto hrcC − that cannot deliver effectors.
The observed results indicate that SDE5 may negatively regulate PTI. We further analyzed the formation of cell wall depositions of callose, a PTI response that plays a critical role in the establishment of basal immunity [38][39][40] in different genotype upon Pto hrcC − inoculation. Callose accumulated significantly less in OE-5 and OE-6 plants compared with WT and sde5-3 mutant plants, reinforcing a negative role for SDE5 in PTI response (Fig. 6).
Next, a model bean (Phaseolus vulgaris) pathogen P. syringae pv. phaseolicola 1448a 41 , Pph, that is unable to efficiently suppress defense reactions in Arabidopsis was further introduced 42, 43 . Forsyth et al. 44 established that a major determinant of non-host resistance to Pph in Arabidopsis is FLS2. Therefore, the pathosystem Arabidopsis-Pph is considered a classical model to study FLS2-mediated defenses and PTI. We examined the behavior of sde5-3 genotypes upon inoculation with the Pph strain. Significantly enhanced bacterial growth was observed in the transgenic OE-5 plant compared to WT (Fig. S3) at 7 dpi. This is consistent with the susceptibility of OE-5 and OE-6 plants to Pto hrcC − strain and confirms that SDE5 restricts PTI-dependent defense system. Transgenic overexpressing plant exhibits altered molecular and cellular responses to flagellin application. The up-regulation of SDE5 transcript by exogenous flg22 treatment prompted us to examine the function of SDE5 protein in the FLS2 signaling pathway. Because callose formation is induced in response to PAMPs 45 , phenotypic assays for flg22-induced callose deposition were performed on sde5-3 genotypes 46, 47 . Callose accumulation is undetectable in fls2. And, as was observed with Pto hrcC − inoculation, flg22 induced callose accumulation is significantly reduced in transgenic OE-5 and OE-6 plants compared to WT and sde-5 plants (Fig. 7). These results suggest that SDE5 has a negative role in the production and/or deposition of callose probably by regulating FLS2-mediated signaling pathway.
The perception of flagellin by plant cells also leads to important changes in gene expression 27,28 . Therefore, we next analyzed the transcript level of two PTI marker genes including Flg22 RECEPTOR KINASE 1 (FRK1) and WRKY29, using mRNA from seedlings of WT, sde5-3, OE-5 and OE-6 lines prepared at 3 h before and after 1 μM flg22 treatment. Both FRK1 and WRKY29 transcripts increased in WT seedlings, but not in fls2 mutant, after flg22 elicitation, confirming that these genes are PAMP-responsive. Transcript accumulation of FRK1 and WRKY29 was similar among WT, sde5-3, OE-5 and OE-6 lines after flg22 treatment (from 0 to 3 h) (Fig. S4). These results indicate that SDE5 involves in the FLS2-signaling pathway at downstream or independently of the induction of these PAMP-responsive genes. SCIeNtIfIC RepoRts | 7: 9859 | DOI:10.1038/s41598-017-07918-x SDE5 is required for mRNA export. A previous report suggested that SDE5 has some similarity (12% identity and 58% similarity) to the C-terminal domain (PF03943) of human mRNA export factor TAP (or NXF1) 48 . The TAP C-terminal domain is particularly important for the function of TAP as an mRNA export mediator because it binds to nucleoporin complexes 48,49 . Therefore, we speculated that mRNA export may also be affected in the sde5-3 mutant. To test this hypothesis, we performed an in situ hybridization assay 50 to localize poly(A) signals in WT and sde5-3 mutant plants. The poly(A) signals were examined by confocal microscopy, using grp7 as a positive control 51 . As shown in Fig. 8, the fluorescein poly(A) signals were stronger in the nuclei of sde5-3 and grp7 than WT, OE-5 and OE-6, indicating that mRNA export is diminished in sde5-3 plants, resulting in mRNA accumulation in the nucleus. The signal was not observed in not probed sde5-3 mutant. This result indicates that Arabidopsis SDE5 is likely a contributing factor in the mRNA export pathway.

Discussion
Our results convey that SDE5, a putative RNA export protein and an essential component of the trans-acting small interference RNA (tasiRNA) pathway, may have dual roles in plant defense mechanism. For instance, SDE5 acts as a positive regulator of plant defense system upon fully virulent Pto DC3000 strain (Figs 2 and 3). In contrast, less callose deposition and PR1 expression in SDE5-overexpressing lines upon flg22 application or Pto hrcC − inoculation indicate that SDE5 may act as a negative regulator of the flagellin signaling pathway (Figs 6 and 7).
Our results show that the SDE5 expression is rapidly induced in response to phytopathogenic bacteria and PAMP treatment (flagellin). Using a reverse genetics approach, we clearly demonstrated that SDE5 acts as a positive regulator of plant defense in response to biotrophic pathogen Pto DC3000. The knockout mutant exhibits higher levels of susceptibility than WT to Pto inoculations, and, conversely, SDE5 over-expressing transgenic lines exhibit fewer discernible symptoms and a significantly lower level of bacteria development compared to WT and knockout plants (Fig. 2). In addition, SA-regulated defense marker gene, PR1 expression was significantly downregulated in sde5-3 relative to WT and OE-5 plants (Fig. 3A).
Signaling cross-talk between plant hormones, such as SA, ET and JA, fine tunes the plant defense response 52 . In general, it is believed that SA signaling plays an important role in resistance to biotrophic pathogens and ET/ JA signaling plays a crucial role in resistance to necrotrophic pathogens 53 . And both synergistic and antagonistic interactions between SA and ET/JA signaling pathways have been reported 54 . Up-regulation of JA-mediated  responses and inhibition of SA-inducible defenses result in enhanced resistance to necotrophs but increased susceptibility to biotrophs 55 . Data reported in this study also indicates that SDE5 regulates disease susceptibility to the necrotrophic pathogen, ECC and the expression of PDF1.2 in an opposite way to the biotrophic pathogen (Fig. 4). Depending on the type of invader, a particular subset of defense responses might be activated, such as SA-and JA-mediated signaling pathways, to specifically fend off specific classes of pathogens.
To examine the role of SDE5 in PTI-mediated restriction of bacterial growth, we challenged Arabidopsis with two pathogens, Pto hrcC − and Pph. The Pto hrcC − bacteria trigger PTI and lack effectors to suppress it. FLS2-dependent PTI makes a critical contribution to the resistance of Arabidopsis to Pph, which lacks effectors required to efficiently block PTI 43,44 . When either Pto hrcC − or Pph were introduced into WT, sde5-3 and over-expressing lines, the OE-5 and OE-6 plants showed increased susceptibility compared to WT control and sde5-3 plants (Figs 5 and S3). These results align with the observation of reduced cell wall depositions of callose and lower PR1 gene expression in the OE-5 and OE-6 plants (Figs 5 and 6). On the other hand, sde5-3 showed an opposite tendency, as indicated by the increased PR1 expression. Thus, SDE5 may have a negative role in the PAMP-signaling pathway resulting in reduced plant defense responses. Although SDE5 negatively regulates callose deposition, it did not affect the expression of early induced PTI marker genes such as FRK1 and WRKY29, indicating that SDE5 might regulate a later stage of plant defense signaling (Fig. S4). In addition, the results presented here convey the involvement of SDE5 in PTI through a flagellin-dependent signaling pathway. Further research would be required in order to explore the molecular mechanism controlled by SDE5 in plant defense signaling.
Previous studies revealed that siRNA-related components, such as AGO4, DRB4 and HPR1, involve in plant defense system. A mutation in the AGO4, that is associated with siRNAs showed enhanced susceptibility to the bacterial pathogen Pto DC3000 19 . An important observation presented by the Lopez et al. 56 showed that the co-existence of an enhanced disease resistance to a biotrophic bacteria, like Pto DC3000, with an enhanced susceptibility to necrotrophic fungi in RNA Polymerase V (Pol V)-defective mutant. Pol V is crucial for the RNA-directed DNA methylation (RdDM) pathway that is an epigenetic control mechanism driven by siR-NAs. A mutation in double-stranded RNA binding protein 4 (DRB4) had a more severe effect on RPS2-and RPM1-mediated resistance to Pto DC3000 expressing avrRpt2 or avrRpm1, respectively 57 . Biochemical studies also suggest that DRB4 is required for the stability of RPS2 and RPM1 proteins and thereby resistance mediated by these R proteins 57 . Similarly, an mRNA export factor in Arabidopsis, HPR1, which is also involved in the production of endogenous and exogenous siRNA, acts as a positive regulator in plant defense signaling 23 . Based on these findings, we hypothesize that SDE5 functions in plant defense system in the same pathway as other siRNA components above upon biotrophic pathogen infection. Many studies also have shown that the mutants impaired in mRNA export exhibit enhanced susceptibility to pathogens 58 . Such as, in Arabidopsis, mutations in MOS3 (modifier of snc1) and MOS11 (suppressor of npr1-1, constitutive 1) lead to defects in mRNA export 58,59 . Similar to sde5-3 plants, both mos3 and mos11 single mutants are more susceptible to the virulent strain than WT 58,59 . Pan et al. 23 reported that HPR1, another mRNA trafficking protein, contributes to the basal defense against virulent pathogens. And accordingly, similar to hpr1, mos3 and mos11, mRNA export was affected in the sde5-3 mutant (Fig. 8). Consistent with these observations, we further conclude that SDE5, HPR1, MOS3 and MOS11 probably belong to the same pathway. Interestingly, prl1 mutant, a loss-of-function mutant of a second SDE5 homologue in Arabidopsis (at AT58720) 60 , exhibited enhanced susceptibility to various kinds of virulent and avirulent pathogens 61 . Thus, it would be speculated that SDE5 and PRL1 may function together to regulate innate immunity in Arabidopsis.
In this study, we found that SDE5, a homologue of a human mRNA export protein, is upregulated upon pathogen inoculation, exogenous PAMP and hormonal application and sde5 mutant plants are defective in mRNA export from the nucleus to cytoplasm. Loss of SDE5 function leads to increased susceptibility to the biotrophic pathogen Pseudomonas syringae and enhanced resistance toward necrotrophic Erwinia caratovora pv caratovora (ECC). In addition, knockout mutants and over-expressing transgenic plants also exhibit delayed defense responses after flg22 treatment. Taken together, our results establish that SDE5 contributes to plant innate immunity.
Although, we showed that SDE5 is involved in plant disease resistance, knowledge regarding the role of the mRNA export in plant defense responses is just emerging. Therefore, further functional analysis will be required to determine the detail molecular basis of mRNA export and defense responses in plants.

Plant lines, growth conditions and chemical treatments. Arabidopsis plants used in all experiments
were derived from ecotype Columbia-0 (Col-0). Mutant lines were sde5-2 25 , sde5-3 26 , sid2 32 , ein2 62 , jar1 63 and npr1 8 , and fls2 obtained from the Arabidopsis Biological Resource Center. Genotyping of the Transfer DNA (T-DNA) insertion lines was performed by PCR, using allele-specific primers. The double mutant was produced by crossing the above mutants and genotyped using the primer sets listed in Supplemental Table S1. Plants were grown either on soil or on plates containing Murashige and Skoog (MS) medium (Sigma-Aldrich) with 1% sucrose and 0.6% agar (Sigma-Aldrich) in a growth chamber (16 h of dark and 8 h of light) unless otherwise indicated. SA (250 μM), MeJA (200 μM), flg22 (10 μM) and sterile water (for mock) treatment was carried out in fifteen-day-old seedlings grown in MS medium with 1% sucrose. Seedlings were collected as indicated time after treatment, immediately frozen in liquid nitrogen, and stored at −80 °C until RNA purification.
Bacterial strains, growth conditions and inoculations. The P. syringae strains and ECC strain SCC1 64 used in this study were grown at 28 °C on King's B (KB) medium supplemented with the appropriate antibiotics: 50 μg/ml rifampicin and 50 μg/ml kanamycin (for Pto DC3000) or 50 μg/ml rifampicin (for Pto hrcC − , Pph and ECC). Inoculation was performed as described 65 . In brief, five-week-old leaves were infiltrated with a needless syringe on the abaxial side at the indicated densities. Mock-treated plants were infiltrated with 10 mM MgCl 2 alone. Disease symptoms and quantification of bacterial growth was performed at the indicated times. These experiments were repeated at least three times with similar results.

Plasmid constructs and plant transformation.
Constructs overexpressing SDE5 were generated using the Gateway cloning system (Invitrogen). The SDE5 coding sequence was amplified by PCR using cDNA synthesized from total RNAs of Arabidopsis WT Col-0 seedlings as the template. The amplified fragment was cloned into the pENTR/D-TOPO vector and inserts were confirmed by sequencing. The entry clones were subsequently transformed into the destination vector pSITE-4CA (to overexpress the protein fused to RFP). These constructs were transformed into WT plants through Agrobacterium tumefaciens (GV3101 strain)-mediated floral dip method 66 . Homozygous transgenic lines were selected and transgene expression was analyzed by qRT-PCR and by confocal microscopy to the 35 S::SDE5-RFP transgenic plants.
Callose staining. Callose detection was performed as described 65 . In brief, four-week-old leaves were syringe-infiltrated with 2 × 10 6 cfu/ml of Pto hrcC − , 10 μM flg22, 10 mM MgCl 2 (Mock for bacterium) and sterile water (Mock for flg22) and collected after 16 h. Whole leaves were collected, stained with 0.1% methyl blue, mounted in 50% glycerol, and examined with OPTIKA fluorescence microscope. Four leaves were prepared for each treatment. Three independent biological assays were performed. Representative views of these pictures were randomized, and the number of callose deposits was counted blind.
Gene expression analyses. Total cDNA was synthesized from total RNA of fifteen-day-old seedlings either pathogen inoculation or chemical treatment as indicated using the SuperScript III first strand synthesis system (Invitrogen), according to the manufacturer's instructions. qRT-PCR was performed using a Bio-Rad CFX96 Real-Time System. Amplification curves and gene expression were normalized using ACTIN as an internal standard. The primers used for qRT-PCR were listed in Supplemental Table S1. Triplicate biological and technical replications were performed. Data were analyzed using BioRad CFX Manager 2.0 Software.
Poly(A) mRNA in situ localization assay. Poly(A) mRNA in situ hybridization was conducted essentially as described previously 50,51 . Briefly, leaf samples of 2-week-old seedlings were fixed in a fixation buffer (3 mM NaH 2 PO 4 , 7 mM Na 2 HPO 4 , 120 mM NaCl, 2.7 mM KCl, 80 mM EGTA, 0.1% Tween 20, 5% formaldehyde, 10% DMSO, and 50% heptane), were subsequently incubated in 1:1 ethanol:xylene, and were washed with ethanol, methanol and finally with 1:1 methanol:fixation buffer. The samples were post-fixed in the fixation buffer for 30 min at room temperature, and were rinsed with Hyb Plus hybridization buffer (Sigma-Aldrich). After prehybridization in hybridization buffer for 1 h at 50 °C, 10 pmol of 45-mer oligo(dT) labeled with fluoresceine at the 5′-end was added and hybridized at 50 °C in darkness. The samples were then washed in 2 × SSC (1 × SSC is 0.15 M NaCl and 0.015 M sodium citrate), 0.1% SDS at 50 °C and in 0.2 × SSC, 0.1% SDS at 50 °C in darkness. The samples were immediately observed under an OLYMPUS 1 × 71 FV500 confocal laser-scanning microscope (Olympus America Inc.) with a 488-nm excitation laser and a 522/DF35 emission filter at identical laser strength. Each experiment was repeated at least three times, and similar results were obtained.