A Strategy for Probing the Function of Noncoding RNAs Finds a Repressor of NFAT

See allHide authors and affiliations

Science  02 Sep 2005:
Vol. 309, Issue 5740, pp. 1570-1573
DOI: 10.1126/science.1115901


Noncoding RNA molecules (ncRNAs) have been implicated in numerous biological processes including transcriptional regulation and the modulation of protein function. Yet, in spite of the apparent abundance of ncRNA, little is known about the biological role of the projected thousands of ncRNA genes present in the human genome. To facilitate functional analysis of these RNAs, we have created an arrayed library of short hairpin RNAs (shRNAs) directed against 512 evolutionarily conserved putative ncRNAs and, via cell-based assays, we have begun to determine their roles in cellular pathways. Using this system, we have identified an ncRNA repressor of the nuclear factor of activated T cells (NFAT), which interacts with multiple proteins including members of the importin-beta superfamily and likely functions as a specific regulator of NFAT nuclear trafficking.

Noncoding RNAs (ncRNAs) are surprisingly prevalent. A systematic analysis of transcription observed ∼10 times more transcriptional activity than can be accounted for by predicted protein-coding genes (1). Much of this activity was subsequently shown to be regulated (2). Moreover, large-scale cDNA analysis and genome annotations predict thousands of ncRNAs (3-6), and computational analysis suggests that over 20% of human genes are regulated by ncRNAs known as microRNAs (7). Whereas several strategies have been used to identify probable ncRNAs (8, 9), a systematic approach to explore their biological functions is lacking. Traditional genetic studies with chemical mutagens are unsatisfactory because ncRNAs are likely resistant to the nonsense and frameshift mutations typically generated in such screens. Furthermore, many of the biochemical methods used to characterize protein complexes are not useful for identifying RNA components. Consequently, we have developed a genomics-based strategy to computationally identify ncRNAs conserved between mouse and human, and we subsequently characterized their biological function by knockdown of the ncRNA transcripts with RNA interference (RNAi) in a series of cell-based pathway screens (10).

In a large-scale analysis of full-length mouse cDNAs, the Functional Annotation of the Mouse (FANTOM) Consortium identified 454 mouse ncRNAs with significant human genome homology (3, 11). We used a de novo analysis and expanded searches to more recent versions of both the Celera and the public assemblies of the human genome (12). Ultimately, 512 ncRNAs with significant human homology were identified; 88 were over-lapping with the FANTOM data set and four were independently identified in a genome-wide analysis of ultraconserved elements (13). These ncRNAs are significantly larger (averaging ∼2 kb) than many characterized small ncRNAs and, given their initial isolation in the FANTOM cDNA cloning project, are likely transcribed by RNA polymerase II. Because these longer ncRNAs do not have readily identifiable functional domains (unlike tRNA or microRNA), a priori classification of the mouse-human conserved ncRNA data set is not practical. Instead, their cellular function was explored using an RNAi-based approach. Two DNA vector-encoded short hairpin RNAs (shRNAs) were designed for each of the mouse-human conserved ncRNAs, for a total of 1024 shRNAs (12); human ncRNAs were selectively targeted because of the many available cell-based assays that use human cell lines. This shRNA collection was arrayed in 384-well tissue culture plates and screened to identify genes that modulate the activity of nuclear factor of activated T cells (NFAT) by using an NFAT-responsive luciferase (luc) reporter (12).

NFAT, a remarkably sensitive transcription factor responsive to local changes in calcium signals, is essential for T cell receptor-mediated immune response and plays a critical role in the development of the heart and vasculature, musculature, and nervous tissue (14). Upon stimulation, the calcium-regulated phosphatase calcineurin dephosphorylates cytoplasmic subunits of NFAT complexes, thus promoting accumulation of NFAT in the nucleus, where it becomes transcriptionally active. The calcium ionophore ionomycin increases intracellular calcium levels, which promotes NFAT translocation to the nucleus, while low levels of the phorbol ester PMA (phorbol 12-myristate 13-acetate) lead to moderate activation of the activating protein 1 (AP1) transcription factor, which binds cooperatively with NFAT. Because shRNAs were used to knock down putative ncRNAs, “activators” in this screen represent ncRNAs whose actual function is repressive in nature. One ncRNA was found, which, when targeted with shRNAs, resulted in a dramatic activation of NFAT activity. This noncoding repressor of NFAT (NRON) was then further characterized.

When human embryonic kidney (HEK) 293 cells were stimulated with ionomycin and PMA, shRNA knockdown of NRON resulted in significantly increased NFAT activity (Fig. 1A and fig. S1), which was blocked by the addition of the calcineurin inhibitor cyclosporine A. A different set of shRNAs directed against mouse NRON also showed activity in mouse 3T3 cells (fig. S1). Furthermore, shRNA knockdown of NRON in the T cell-derived Jurkat cell line elevated NFAT activity in cells stimulated by both chemical (Fig. 1B) and T cell receptor activation. To address potential shRNA and interferon related artifacts, small interfering RNAs (siRNAs) corresponding to the shRNA sequences were also tested and shown to be active (fig. S1). In addition, little activity was seen in an interferon-responsive reporter with the NRON shRNAs. Quantitative polymerase chain reaction (PCR) was used to confirm siRNA-mediated knock-down of NRON, and a ∼40% reduction was observed (Fig. 1C). Transfection efficiency of siRNAs was shown to be over 90% for these experiments; lentiviral delivery of an shRNA (with nearly 100% infection rate) also reduced NRON transcript a comparable amount. Because NRON is likely a rare transcript (see below), moderate changes in transcript levels may have pronounced cellular effects.

Fig. 1.

Noncoding repressor of NFAT (NRON). (A) Five different shRNAs (shRNA1 and shRNA5 share 17 bp of overlap) targeting NRON were tested with and without ionomycin (Iono) and PMA stimulation. Three shRNAs showed significant activation of an NFAT-responsive luciferase reporter (mean ± SD). Samples were normalized to a nonspecific shRNA (NS) in unstimulated cells. Similar results were observed with human U2OS and mouse 3T3 cell lines (fig. S1). (B) These shRNAs were also tested in T cell-derived Jurkat cells, with two shRNAs showing ∼2-fold or greater activation of the NFAT-luc reporter. Cells were either chemically stimulated with Iono/PMA or activated via a T cell receptor with antibodies targeting CD3 (data not shown). (C) Quantitative real-time PCR (qPCR) measure of NRON transcript knockdown by siRNA in HEK293 cells. Transfection efficiency for siRNAs was greater than 90%. NRON siRNAs caused a reduction in NRON RNA, compared with a nonspecific control siRNA: siRNA1, 42.4% reduction; siRNA2, 13.2%; and siRNA5, 36.7% (mean ± SD). Transcript knockdown was also measured by lentiviral delivery of shRNA1 (with nearly 100% transfection transduction efficiency), which reduced NRON RNA 36.0%, compared with a nonspecific shRNA.

To gain further insight into the function of NRON, we characterized its gene structure and expression. Based on rapid amplification of cDNA ends (RACE) and cDNA sequence data, we found that the NRON gene is composed of three exons, which can be alternatively spliced to yield transcripts ranging in size from 0.8 to 3.7kb(Fig.2A). The NRON gene also has two large 300-to-400-base pair (bp) regions of near-perfect sequence conservation between rodents and primates (fig. S2). An analysis of the coding potential of NRON transcripts and the surrounding genomic interval further supports the FANTOM Consortium's classification of NRON as an ncRNA (12). The expression of NRON was analyzed by reverse transcription (RT)-PCR, and the high number of amplification cycles required to detect NRON suggests that it is a relatively rare transcript (Fig. 2 and fig. S2). A survey of total RNA from a variety of human and mouse tissues showed that NRON is enriched in placenta, muscle, and lymphoid tissues such as the thymus, spleen, and lymph node (Fig. 2, B and C). Furthermore, all three mouse FANTOM cDNA clones were originally isolated from thymus libraries. A Northern blot of mouse RNA showed significant NRON expression in the embryo and the thymus (Fig. 2D), which is consistent with the enrichment in the placenta and the thymus seen by RT-PCR. Finally, NRON transcripts have a distinct tissue-specific distribution of splice forms, which likely serve a currently unknown biological function. NRON's tissue-specific expression, particularly its enrichment in lymphoid tissues, is consistent with its role as a modulator of NFAT signaling.

Fig. 2.

NRON expression data. (A) Gene structure for mouse NRON based on 5′ RACE analysis and RIKEN cDNA clones (12). Gray bars in exon 3 indicate regions highly conserved between human, chimp, mouse, and rat (fig. S2): region A (298 bp, 90% identity) and region B (400 bp, 89% identity). Most of the siRNA or shRNA sequences used in this study target areas near region B. (B) RT-PCR of NRON from human tissues showed relative enrichment in the placenta, thymus, and spleen, with detectable expression in the testes, kidney, brain, and adrenal glands. (C) RT-PCR of NRON from mouse tissues showed elevated expression in skeletal muscle and the thymus, with expression also seen in the spleen, lymph node, and lung. (D) A Northern blot of mouse poly-A+ mRNA probed with the long splice form of NRON exon 3 showed significant NRON expression in the embryo and thymus, with lower expression in other tissues. The size differences in transcripts, ranging from 2 to 4 kb, are consistent with probable splice variants.

In order to define the molecular mechanism by which NRON represses NFAT activity, a biochemical approach was used to identify possible RNA-protein interactions. The 3′ terminus of exon 3 of NRON was tagged with an RNA hairpin, which itself was bound tightly by the MS2 phage protein (15). An MS2/maltose binding-protein fusion was then used to purify NRON-interacting proteins from a whole-cell protein extract, and their identities were determined by mass spectrometry (12). After comparison to a nonspecific RNA control, 11 proteins were found to bind NRON specifically (Table 1), including three members of the importin-beta superfamily, factors which directly mediate the nucleocytoplasmic transport of cargoes such as NFAT (16). siRNAs directed against four of these 11 putative interactors—a calmodulin-binding protein (IQGAP1), a nuclear transport factor (KPNB1), the structural subunit of a phosphatase (PPP2R1A), and a component of the proteasome (PSMD11)—all activated NFAT activity (a two- to sixfold increase in NFAT activity was observed in a sensitized setting where NRON was also knocked down). cDNA overexpression of these four interactors had the opposite effect and repressed NFAT activity. Therefore, these four proteins, together with NRON, have a repressive effect on NFAT signaling (Fig. 3A and fig. S3).

Fig. 3.

NRON interacts with nuclear transport factors. (A) For each NRON-interacting protein, three unique siRNAs were assayed for their activity on the NFAT-luc reporter. The siRNAs by themselves had no effect (fig. S3); however, in the presence of an siRNA targeting NRON (siRNA5), significant stimulation-dependent synergy was observed (mean ± SD). These same proteins were also identified as repressors of NFAT when overexpressed in the presence of a small amount of NFAT cDNA to stimulate basal levels of activity. See fig. S3 for further controls and qPCR data. (B) In vitro RNA-binding experiments show that NRON interacts with KPNB1. The precipitation of epitope-tagged proteins (because of its size, IQGAP1 was split into two fragments) from cell extracts incubated with radiolabeled NRON (2.7 kb of exon 3) demonstrated that NRON specifically coprecipitates with KPNB1, when compared with a nonspecific RNA (nsRNA). In an RNase T1 protection experiment, protein extracts containing overexpressed KPNB1 protected radiolabeled NRON from digestion. (C) Two NRON shRNAs show no effects on reporters for NFκB, p53, and AP1 transcription. NFAT data are from stimulated cells. (D) U2OS cells were selected for microscopy analysis because of their morphology and adherence. NFATc1 fused to GFP is an effective reporter (19), localizing from cytoplasm to nucleus upon stimulation. (E) Cotransfection with NRON shRNA1 results in significantly increased NFAT nuclear localization (NFATc1-GFP), compared with a control shRNA (mean ± SD). The forkhead transcription factor FOXO1, which also shuttles from cytoplasm to nucleus (24), was used as a control and was shown to be unaffected by NRON knockdown.

Table 1.

NRON-interacting proteins identified by an RNA-protein affinity purification strategy using the long splice form (2.7 kb) of exon 3 (12). Recovered proteins were identified by mass spectrometry [see (12) for peptide sequences and quality scores]. trans., transport; CAS, cellular apoptosis susceptibility protein; JNK, Jun N-terminal kinase; ATP, adenosine triphosphate; CaM, calmodulin.

Unigene Function Description Genbank
CSE1L Nucleocytoplasmic trans. Importin-alpha export (CAS) AAC35008
KPNB1 Nucleocytoplasmic trans. Importin-beta 1 (karyopherin) NP_002256
TNPO1 Nucleocytoplasmic trans. Importin-beta (transportin 1) AAB68 948
EIF3S6 Protein biosynthesis Translation initiation factor AA H17887
CUL4B Proteolysis Cullin-based E3 ligase complex AAK16812
PSMD11 Proteolysis Proteasome 26S non-ATPase NP_002806
UREB1 Proteolysis E3 ubiquitin protein ligase BAC06833
DDX3X RNA helicase DEAD-box protein O00571
IQGAP1 Signal transduction CaM-binding scaffolding protein NP_003861
PPP2R1A Signal transduction Protein phosphatase 2 subunit A P30153
SPAG9 Signal transduction JNK-assoc. leucine-zipper protein AAN61565

In vitro RNA protein-binding assays (12) were used to further characterize the interaction between NRON and the four proteins that showed NFAT modulatory effects. Protein open reading frames were cloned into mammalian expression vectors containing an N-terminal FLAG tag. Then, protein extracts from cells expressing these plasmids were incubated with radiolabeled NRON, and tagged proteins were precipitated. Cell extracts containing overexpressed importin-beta 1 (KPNB1) bound significantly more NRON than did a nonspecific RNA control, suggesting that importin-beta 1 and NRON directly associate (Fig. 3B). This interaction is supported by ribonuclease (RNase) protection experiments, which show that NRON is protected from digestion by extracts containing high levels of KPNB1 (Fig. 3B). Furthermore, interactions between PPP2R1A and KPNB1 (17), as well as between PPP2R1A and IQGAP1 (18), have been previously reported. Altogether, this data argues that NRON functions as an RNA component of a protein complex that acts to repress NFAT activity.

The initial isolation of three importin-beta family members and the subsequent demonstration that importin-beta 1 can bind NRON and alter NFAT activity suggests that NRON may act as a modulator of NFAT nuclear trafficking. Furthermore, NRON acts on multiple NFAT family members; cotransfection of NFAT cDNAs and NRON shRNA showed that NRON modulates the calcium-regulated transcriptional activity of NFATc1, NFATc2, NFATc3, and NFATc4 (fig. S1). shRNAs targeting NRON were also assayed with p53, nuclear factor κB (NFκB), AP1, and the fork-head FOXO1 reporters, which are transcription factors that translocate from cytoplasm to nucleus (Fig. 3, C and E). No NRON-dependent phenotypes were seen, suggesting that NRON is specific for NFAT translocation. Finally, using an NFATc1-green fluorescent protein (GFP) fusion as a visible marker for NFAT subcellular localization (19), it was shown that an NRON shRNA results in a significant increase in nuclear levels of NFAT protein (Fig. 3E and fig. S4). NRON knockdown elevated nuclear NFAT even in the absence of Ca2+ stimulation, which is probably a direct result of elevated cellular levels of NFAT protein from the introduction of the NFATc1-GFP fusion. This is consistent with the observation that overexpression of NFAT cDNAs obviates the need for chemical stimulation for NRON shRNA activity (fig. S1) (20, 21). Therefore, rather than directly modulating the transcriptional activity of NFAT itself, NRON likely regulates NFAT's subcellular localization.

The interaction of NRON with nuclear import factors suggests that NRON exists in a complex with importin-beta and specifically regulates the nuclear trafficking of NFAT, a hypothesis supported by studies of NFATc1-GFP translocation. In light of the complicated networks of nuclear-cytoplasmic transport and the seemingly limited number of available importin-beta family members, specific ncRNAs may play a role in regulating the complexity of intracellular trafficking. The interplay between importins, the nuclear localization signal of NFAT, and NRON remains unknown, as does whether the native function of NRON is to repress nuclear import or to promote the nuclear export of NFAT. In addition, given the effects of both calcium and phosphorylation on NFAT activity, the interactions of NRON with the structural subunit of the phosphatase PP2A and the calmodulin binding IQGAP1 may also be significant. Further characterization of all 11 putative NRON interactors may identify more complex interactions and regulatory features for this presumed RNA-protein macromolecular complex.

NRON is but one of the potentially thousands of RNA regulators, which, through RNA-RNA, RNA-DNA, or RNA-protein interactions, may effectively amplify the complexity of a human genome with a limited number of protein-coding genes (22, 23). The application of this library of ncRNA-specific shRNAs to additional cellular pathway and phenotypic screens is likely to reveal additional functional roles for these transcribed RNAs. Preliminary experiments have identified seven other functional ncRNA genes: six essential for cell viability and one repressor of Hedgehog signaling.

Supporting Online Material

Materials and Methods

Figs. S1 to S4

Tables S1 to S5


References and Notes

View Abstract

Navigate This Article