The long noncoding RNA NEAT1_1 is seemingly dispensable for normal tissue homeostasis and cancer cell growth

Abstract

NEAT1 is one of the most studied lncRNAs, in part because its silencing in mice causes defects in mammary gland development and corpus luteum formation and protects them from skin cancer development. Moreover, depleting NEAT1 in established cancer cell lines reduces growth and sensitizes cells to DNA damaging agents. However, NEAT1 produces two isoforms and because the short isoform, NEAT1_1, completely overlaps the 5' part of the long NEAT1_2 isoform; the respective contributions of each of the isoforms to these phenotypes has remained unclear. Whereas NEAT1_1 is highly expressed in most tissues, NEAT1_2 is the central architectural component of paraspeckles, which are nuclear bodies that assemble in specific tissues and cells exposed to various forms of stress. Using dual RNA-FISH to detect both NEAT1_1 outside of the paraspeckles and NEAT1_2/NEAT1 inside this nuclear body, we report herein that NEAT1_1 levels are dynamically regulated during the cell cycle and targeted for degradation by the nuclear RNA exosome. Unexpectedly, however, cancer cells engineered to lack NEAT1_1, but not NEAT1_2, do not exhibit cell cycle defects. Moreover, Neat1_1-specific knockout mice do not exhibit the phenotypes observed in Neat1-deficient mice. We propose that NEAT1 functions are mainly, if not exclusively, attributable to NEAT1_2 and, by extension, to paraspeckles.

Keywords: NEAT1 isoforms; RNA exosome; cell cycle; lncRNA; mouse genetics; paraspeckles.

FIGURE 1.

NEAT1 isoforms are differentially regulated by distinct p53 activating agents. (A) RefSeq representation of human NEAT1 isoforms in the UCSC genome browser (hg38), nucleotide-level conservation among 30 mammals and 100 vertebrates, and the location of repeats in the genome sequence. Red and blue dotted lines represent RNA-FISH probes targeting both and the long NEAT1_2 isoform specifically, respectively. Note that when targeting the long isoform, blue and red probes will overlap and thus show PSs in pink. Small arrows represent approximate locations of RT-qPCR primers used in this study. The poly(A) tail on the short NEAT1_1 isoform is not drawn to scale. (B) Representative confocal images of RNA-FISH targeting NEAT1 isoforms in U2OS cells in nontreated (NT), 10 µM Nutlin-3a (24 h) and 1 mM hydroxyurea (HU, 48 h) conditions. Scale bar, 15 µm. (C) Quantification of the percentage of cells in which the short isoform can be observed outside of PSs. Each dot represents an independent experiment (N = 13, 8, and 6, respectively). (D–F) Relative levels of NEAT1 (both isoforms; D), NEAT1_2 (E), and the canonical p53 target CDKN1A (F) after 0, 3, and 48 h of HU treatment (1 mM) in RT-qPCR.

FIGURE 2.

NEAT1 isoforms are differentially regulated during the cell cycle. (A) Representative DNA content (DAPI) distribution of U2OS cells in G₀ (3 d starvation) and G₁/S (G₀ cells released in 20% serum + 5 µg/mL aphidicolin for 24 h). Gray background plots are control nonsynchronized (NS) cells in culture. (B) Representative images of NEAT1/NEAT1_2 RNA FISH of the cells in A. Scale bar, 10 µm. (C) Quantification of the percentage of cells in which NEAT1_1 was detected independently of NEAT1_2 PSs (red boxplots, left y-axis). Percentage of cells in which only NEAT1_2 PSs were detected (right y-axis, purple boxplots). Tukey plots of individual data points (one point per picture). Significance was calculated using an unpaired, two-sided t-test on independent biological replicates (N = at least three). (***) P < 0.001. (D) Like in A, but from U2OS cells synchronized by double thymidine block and released for 18 (G₁), 2 (early S), 4 (mid/late S), and 8 h (G2M). (E) Representative images of NEAT1/NEAT1_2 RNA FISH of the cells in D. (F) Same as C for the cells in D. Significance was calculated using a two-way ANOVA with Dunnett's correction for multiple comparisons on independent replicates (N = at least three). (***) P < 0.001 for G₁ vs. early S, mid/late S, and G₂. N1_2 = NEAT1_2. (G) Percentage of cells in S and G₂ phases at different time points relative to time = 0 h in HeLa cells upon release into 20% serum after 3 d starvation. (H) Box plots (Min to Max) of individual data points (N = 5 and 3 pictures per replicate, respectively) quantifying the percentage of cells in E in which NEAT1_1 was detected using RNA-FISH against both and the long isoform specifically. (I) Relative increase of NEAT1_2 by RT-qPCR of the cells in H. Error bars are standard deviation of N = 2 experiments in G and H. (J) Representative pictures of cells in G–I. Scale bar, 20 µm.

FIGURE 3.

NEAT1_1 is selectively degraded by the RNA exosome. (A) Representative NEAT1/NEAT1_2 RNA-FISH images of U2OS cells in which RRP40, a subunit of the RNA exosome complex, was knocked down. (B) Tukey plots of quantified RNA-FISH signal shown as NEAT1_2 nuclear intensity over total NEAT1 nuclear intensity in N = 5 independent experiments. Significance was calculated by an unpaired Mann–Whitney U nonparametric test with each cell as a data point combining data from five independent experiments. (***) Two-tailed P < 0.001. (C) Mean and standard deviation of relative RNA levels by RT-qPCR of the same experiments as in B (N = 5) in eGFP and RRP40 KD cells suggesting specific up-regulation of the NEAT1_1 isoform. Dots are individual data points. (D,E) Line plots of the intensity in eGFP KD (D) and RRP40KD (E) conditions showing NEAT1 localization in poly(A)⁺ accumulations upon exosome inhibition. (F) NEAT1 (blue)/poly(A) (red) RNA-FISH and IF in HeLa cells showing NEAT1/poly(A)⁺ accumulations in RRP40 KD conditions (arrows) are distinct from PSs (NEAT1 + NONO, asterisk). (G–I) Relative RNA levels of both NEAT1 isoforms (G) and the long NEAT1_2 isoform specifically (H) upon addition of 2 mg/mL of the transcription inhibitor actinomycin D 48 h after eGFP or RRP40 KD in HeLa cells. (I) Bar graph showing relative RNA levels of RRP40 for the experiments in G–H at time = 0 h. Statistical significance was tested using a two-way ANOVA with Dunnett's correction for multiple testing for N = 3 independent experiments. Bars and error are mean and standard deviation. Individual data points are independent experiments.

FIGURE 4.

Loss of NEAT1_1 does not impact on cell growth. (A) Scheme of the CRISPR strategy used to knock out NEAT1_1 by deletion of the regulatory sequences and PAS at the 3′-end of the NEAT1_1 short isoform genomic sequence and the resulting RNAs in wild-type (WT) and NEAT1_1 knockout (KO) cells. Arrows and dotted lines in the RNA represent RT-PCR primers and RNA-FISH probes, respectively. Arrows on the DNA sequence represent approximate locations of genotyping primers. (B) Representative inverted gel image of the PCR product from U2OS gDNA used for genotyping individually isolated single cell clones after CRISPR with the primers depicted in A. (L) DNA ladder. (C) Relative RNA levels of NEAT1 (total) and NEAT1_2 specifically showing up-regulation of NEAT1_2, whereas the total levels of NEAT1 remain the same in WT versus KO clones. The WT without a number is the mother population from which the WT and KO clones were derived. Significance was calculated using a two-way ANOVA comparing RNA levels in WT and KO clones with Dunnett's correction for multiple testing. (***) P < 0.001 for N = at least three independent experiments. (n.s.) Not significant. Dots represent data points from each independent experiment. (D) Representative NEAT1/NEAT1_2 RNA-FISH image of WT and NEAT1_1 KO cells showing the complete loss of NEAT1_1 upon poly(A) site knockout. Scale bar, 10 µm. (E) Quantification of percentage of area covered (left) and representative images (right) of colony assays 14 d after seeding 2000 cells per well in N = 3 independent replicates, three wells per replicate each. Statistical testing was done using a one-way ANOVA. (n.s.) Not significant. Dots represent the average of three wells of the independent experiments. Bars are mean + standard deviation. (F) Short-term growth measured by WST-1 relative to day 0 (1 d after seeding) in nontreated (NT), 10 µM Nutlin-3a and a low dose of doxorubicin (150 ng/mL) in NEAT1_1 WT and KO clones. All data is the average of N = 3 independent experiments. Standard error is depicted as dotted lines above and below the data points. In Nutlin-3a and doxorubicin conditions, values in nontreated conditions are shown as light gray lines in the back of the graph. All data is not significant as tested by two-way ANOVA with Dunnett's correction for multiple testing in the different time points. (G) Quantification of percentage of EdU-positive cells (S phase) in flow cytometry upon CTRL (siC) and NEAT1_2 (siLg) knockdown in WT and KO clones. N = 3 independent experiments. Significance was determined using two-way ANOVA with Dunnett's correction for multiple comparisons. (***) P < 0.001. (n.s.) Not significant. (H) NEAT1_2 knockdown efficiency as assessed by RT-qPCR 48 h after transfection for siCTRL (dark purple bars) and siNEAT1_2 (light purple bars) for the NEAT1_1 KO cells shown in the pictures on the right (panel I). (I) Representative image of siCTRL and siNEAT1_2 KO cells showing decreased cell density 48 h after transfection. (J) Representative flow cytometry graphs of EdU/DAPI staining in eGFP and RRP40 KD conditions. (K) Percentage of cells in G₁, S, and G₂M phases of the cell cycle upon eGFP and RRP40 KD as in Figure 3A–C. N = 5 independent experiments. Nonsignificance was determined using a two-way ANOVA comparing eGFP and RRP40 conditions in each of the phases with Dunnett's correction for multiple testing.

FIGURE 5.

NEAT1_1 KO only causes limited changes in gene expression. (A) Representative NEAT1/NEAT1_2 FISH in G₀ and G₁S conditions as in Figure 2A–C in WT and NEAT1_1 KO cells used for the Smartseq2 RNA sequencing experiment. (B) Number of genes detected in G₀ and G₁S conditions. Red/pink dots are WT conditions; blue dots represent KO conditions. Significance was tested using a two-sided unpaired t-test comparing the number of genes detected in WT and KO conditions. (C,D) Volcano plot of gene expression changes (−log₂) in G₀ (C) and G₁S (D) plotted against their P-value (−log₁₀). Dots are color-coded red if the adjusted P < 0.05, orange if log₂ fold change >1, and green if both. (E) Hierarchical clustering of G₀ samples based on 156 unique differentially expressed genes (FC > 1.5, P-adj < 0.05). (F) Hierarchical clustering of G₁/S samples based on 23 unique differentially expressed genes (FC > 1.5, P-adj. < 0.05).

FIGURE 6.

Comparison of Neat1 versus Neat1_1 KO phenotypes. (A) CRISPR strategy to knock out Neat1_1 in mouse embryonic stem cells resulting in a 39-base pair deletion spanning the Neat1_1 PAS and strategy to knock out both isoforms as described in Nakagawa et al. (2011). (B) Genotyping of Neat1 full KO mice. (C) Genotyping of Neat1_1 (PAS) KO mice. The asterisk indicates an unspecific band in the heterozygous sample. (D) Genotype distribution of pups born from heterozygous parents in full Neat1KO (middle bar) and PAS KO nests (lower bar) as compared to the expected Mendelian ratios (upper bar). N = 53 and 15 litters for Neat1 and PAS KO, respectively. P-values were calculated using the χ² test. (***) P < 0.001. (n.s.) Not significant. (E,F) Pup weight of offspring from females with the respective genotypes at 3 (E) and 6 (F) weeks of age. Tukey plots of pups from N = between four and 19 females per genotype. Statistical significance was calculated using two-way ANOVA with Dunnett's correction for multiple testing. (***) P < 0.001. (n.s.) Not significant. (G) Strategy and timeline for short-term DMBA/TPA carcinogenesis protocol. (H) Representative H&E staining of back skin sections from mice treated as in G. Scale bar, 50 µm. (I) Neat1 RNA-FISH and Keratin 5 immunofluorescence on back skin section of mice treated with DMBA and TPA as in G. (J,K). Quantification (J) and representative images (K) of sections from DMBA/TPA treated back skin stained with the DNA damage marker γ-H2A.X of N = at least three mice per genotype according to the method described in Adriaens et al. (2016). The Tukey plot in J graphs individually quantified pictures. Statistical testing was done on biological replicates (averages for individual mice) using a one-way ANOVA with Sidak's correction for multiple testing. (**) P < 0.01. (n.s.) Not significant. White arrows in K indicate γ-H2A.X-positive cells, whereas dotted lines separate the dermis from the epidermis. Scale bar, 20 µm.