The hnRNP RALY regulates transcription and cell proliferation by modulating the expression of specific factors including the proliferation marker E2F1

The heterogeneous nuclear ribonucleoproteins (hnRNP) form a large family of RNA-binding proteins that exert numerous functions in RNA metabolism. RALY is a member of the hnRNP family that binds poly-U–rich elements within several RNAs and regulates the expression of specific transcripts. RALY is up-regulated in different types of cancer, and its down-regulation impairs cell cycle progression. However, the RALY's role in regulating RNA levels remains elusive. Here, we show that numerous genes coding for factors involved in transcription and cell cycle regulation exhibit an altered expression in RALY–down-regulated HeLa cells, consequently causing impairments in transcription, cell proliferation, and cell cycle progression. Interestingly, by comparing the list of RALY targets with the list of genes affected by RALY down-regulation, we found an enrichment of RALY mRNA targets in the down-regulated genes upon RALY silencing. The affected genes include the E2F transcription factor family. Given its role as proliferation-promoting transcription factor, we focused on E2F1. We demonstrate that E2F1 mRNA stability and E2F1 protein levels are reduced in cells lacking RALY expression. Finally, we also show that RALY interacts with transcriptionally active chromatin in both an RNA-dependent and -independent manner and that this association is abolished in the absence of active transcription. Taken together, our results highlight the importance of RALY as an indirect regulator of transcription and cell cycle progression through the regulation of specific mRNA targets, thus strengthening the possibility of a direct gene expression regulation exerted by RALY.

The transcription of DNA into RNA mediated by RNA polymerase II (RNAPII) 3 is a complex process that involves numer-ous factors, which co-operate with the transcriptional machinery and interact with newly synthesized RNA to ensure its correct processing (1)(2)(3)(4). In this context, RNA-binding proteins (RBPs) exert a broad range of functions from the early steps of RNA synthesis until its maturation. For example, they can participate in gene expression regulation by binding chromatin components and interact with nascent RNAs, promoting post-transcriptional modifications, including splicing, nuclearcytoplasmic shuttling, and RNA transport and stability (5)(6)(7)(8)(9)(10)(11)(12)(13). By affecting one or more of these processes, a dysregulation of RBPs might alter the correct expression of fundamental proteins, leading to pathological conditions (14 -18). Thus, the tight regulation of transcription and post-transcriptional processing of RNA exerted by RBPs is required for the upstream maintenance of cellular physiology and for the regulation of different processes, including transcription and cell proliferation, which are closely associated. Rapidly proliferating cells show high levels of transcription to sustain the intense demand of gene products necessary to maintain quick growth rates. To this aim, these cells tend to overexpress versatile transcription factors, such as members of the MYC and E2F family, which supply the necessary gene products to promote their growth and division (19).
Among the RBPs involved in the transcriptional regulation and post-transcriptional modification of RNAs, the heterogeneous nuclear ribonucleoproteins (hnRNPs) form a large family of RBPs possessing a broad range of functions in the metabolism of RNA (20 -23). Several members of the hnRNP family have been described as interacting with chromatin, DNA, and RNA to influence transcription and RNA processing (24). For example, the hnRNPs C1/C2 interact with newly synthesized transcripts through their RNA-binding domain, regulating RNA stability and RNA nuclear export (12,25,26). hnRNP-U, which binds chromatin and is involved in pre-mRNA processing, also regulates RNAPII transcription by binding to its C-terminal domain or to the preinitiation complex (27,28). Moreover, hnRNP-K exerts a transcriptional control activity by interacting with DNA on regulatory elements upstream of the promoters but also by binding specific RNAs to recruit the termination factor XRN2 (24,29).
RALY is a member of the hnRNP family that was initially identified as a cross-reacting autoantigen with the Epstein-Barr nuclear antigen 1 (EBNA1), a viral protein encoded by the Epstein-Barr virus (30). RALY contains an RNA-recognition motif (RRM) at the N terminus, two predicted nuclear localization signals, and a glycine-rich region at the C terminus (25,31). We recently identified RALY RNA targets by performing RIPseq analysis in MCF7 cells and studied the gene expression profile of RALY-silenced MCF7 cells. The two analyses identified 2929 RNAs bound by RALY and 217 differently expressed genes consequent to its down-regulation (32). Crossing these results, we selected 23 genes that are differently expressed after RALY silencing and for which the transcript is bound by RALY. Among these targets, we focused on ANXA1 and H1FX, which were found, respectively, to be up-and down-regulated both at the RNA and the protein level upon RALY silencing. We also characterized the RNA-binding property of RALY by showing that it binds poly-U stretches, preferentially positioned in the 3Ј-UTR of protein-coding transcripts (32). Through proteomic analysis, RALY was found to interact with the C-terminal domain of RNAPII as well as with proteins involved in RNA metabolism and translational control, including polysomes (31)(32)(33).
In a recent study, RALY was found to interact with the longnoncoding RNA LeXis and with promoters of genes involved in the cholesterol biosynthesis pathway in mouse liver (34). An increased expression of LeXis determined a decrease in the cholesterol content of mice blood serum. Interestingly, when LeXis was up-regulated, Sallam et al. (34) detected a lesser presence of RALY on the promoters of genes involved in cholesterol biosynthesis, with a consequent reduced expression of their mRNA. The same phenotype was observed when RALY was down-regulated with LeXis in a wild-type condition, and no effects of LeXis expression were observed in RALY-defective mice. These results, which highlight a transcriptional regulatory activity of RALY, although not in detail, induced Sallam et al. (34) to propose RALY as a transcriptional co-factor. Altogether, these data suggest that RALY plays differing roles in the processing and post-transcriptional modification of RNA.
Here, we studied the gene expression profile of HeLa cells in response to RALY silencing to better characterize the impairment in cell cycle progression observed previously by Rossi et al. (32). We found that mRNAs coding for several factors involved in the regulation of cell proliferation and transcription were down-regulated upon RALY silencing. In particular, we focused on the interaction between RALY and E2F1 mRNA, a well-characterized marker of cell proliferation and regulator of the cell cycle (35). We demonstrated that RALY regulates the expression and the stability of E2F1 mRNA and therefore the amount of E2F1 protein inside the cells. In agreement with our findings, the deletion of RALY in HeLa cells caused a reduction of cell proliferation and a stall in the G 1 phase of the cell cycle. Moreover, RALY silencing caused a global reduction of RNAPIIdependent transcription. To understand in more detail the role of RALY in gene expression regulation, we then studied the interaction of RALY with chromatin. We show that RALY can interact with transcriptionally active chromatin through two different regions in both an RNA-dependent and -independent manner. Taken together, our results demonstrate that RALY regulates cell proliferation and transcription by modulating the expression of several key factors of the two processes, and add evidence of the direct involvement of RALY to gene expression regulation by characterizing its binding to transcriptionally active chromatin.

The down-regulation of RALY impairs the expression of cell cycle-and transcription-related genes
We recently observed that the silencing of RALY in MCF7 cells caused the down-regulation of different genes related to cell cycle progression and that RALY-down-regulated cells showed a reduction in cell growth rate compared with control cells (32). Furthermore, transcripts coding for cell cyclerelated proteins were enriched among the RNA bound by RALY (32). To characterize the role of RALY in cell proliferation, we studied the gene expression profile of HeLa cells after RALY down-regulation. We used a HTA2.0 microarray (Affymetrix Human Transcriptome Array) to analyze three different biological replicas of HeLa cells transfected with either siRNA against RALY (si-RALY) or control siRNA (si-CTRL) for 72 h. The levels of RALY mRNA measured by quantitative real-time PCR (qRT-PCR) were detected at levels below 10% in si-RALYtransfected cells compared with control cells (Fig. 1A).
After RALY silencing, we found 1971 differentially expressed genes, of which 919 were up-regulated and 1052 down-regulated ( Fig. 1B and supplemental Table S1). The majority of the variations was observed in protein coding transcripts (93.5%), with a small percentage of noncoding RNAs, generally higher among up-regulated genes (3% of the long noncoding RNA) (Fig. 1C). Functional annotation enrichment analysis by gene ontology and pathways databases identified extracellular matrix organization, regulation of cell motility, and cell adhesion as the biological processes that were significantly associated with the up-regulated genes. On the other hand, cell cycle control and RNA splicing were the two broad themes strongly associated with the down-regulated genes upon RALY silencing ( Fig. 1D and supplemental Table S2). Interestingly, several genes coding for factors involved in transcription regulation were present in the list of down-regulated genes (supplemental Table S1).
To have a more comprehensive understanding of the general function of RALY on its target mRNAs, we intersected the list of differentially expressed genes detected in HeLa cells with the list of RALY targets identified previously by RIP-seq in MCF7 cells (32). We found that 193 up-regulated genes and 359 downregulated genes were present in both lists (Fig. 1E). Although this intersection could be underestimated due to cell line differences, the increased frequency of RALY RNA targets among genes down-regulated upon RALY silencing (34.1 versus 21%, p value, 1.3e-10) suggests that the loss of a direct RALY-mRNA interaction might explain the observed down-regulation (Fig. 1E). the release of RNAPII from promoter-proximal pausing and for the activation of RNAPII productive elongation; GTF2A1 and GTF2E2, subunits of general transcription factors involved in transcription initiation; and ELL2, a member of the transcription-promoting super elongation complex ( Fig. 2A) (39 -42). In addition, the two subunits of the facilitating transcription (FACT) complex, SUPT16H and SSRP1, were found to be down-regulated at the mRNA level ( Fig. 2A). The FACT complex has been described as associating with active RNAPII and promoting nucleosome disassembly to facilitate transcription (43)(44)(45). All of the target mRNAs were found to be down-regulated in si-RALY-transfected HeLa cells, which confirms the microarray data ( Fig. 2A). In addition, we measured the protein levels of CCNE1 and CCNT1 by Western blot analysis and found both down-regulated in si-RALY cells (Fig. 2, B and C).

RALY binds E2F1 mRNA and regulates its stability
To improve the functional characterization of the deregulated transcripts upon RALY silencing, we performed gene set enrichment analysis (GSEA) over the collection of "hallmark" annotated gene sets provided by the Molecular Signatures Database (MSigDB) (46). This analysis identified the targets of the E2F transcription factors family as the class of genes most significantly enriched among the genes down-regulated upon RALY silencing (Fig. 3A). This gene set, according to the MSigDB definition, includes "genes encoding cell cycle related targets of E2F transcription factors." The E2F family encodes for eight transcription factors that play key roles in the regulation of cell cycle progression. Its members can be divided into activators (E2F1-3) and repressors (E2F4 -5, E2F6, and E2F7-8) of cell proliferation (47). In particular, the E2F1-3 group, which regulates the transactivation of genes necessary for the G 1 /S-phase transition of the cell cycle, was found to be down-regulated in the microarray data upon RALY silencing (supplemental Table S1). On the contrary, only the repressors E2F7 and E2F8 showed an altered expression in si-RALY-treated cells, being up-and down-regulated, respectively, whereas no variations were found for repressor group E2F4 -6 (supplemental Table S1). Interestingly, E2F1 and E2F2 mRNAs were identified previously as specific targets of RALY in the RIP-seq experiment performed in MCF7 cells, whereas all the proliferation repressors of the E2F family were not detected as RALY RNA interactors (Fig. 3B) (32).
Among the misregulated activators of the E2F family, we focused on E2F1, a key regulator of the cell cycle and apoptosis, also involved in carcinogenesis when misexpressed (19, 48 -51).
First, we confirmed its down-regulation in RALY silenced cells both at the mRNA and protein level by qRT-PCR and Western blot analyses, respectively (Fig. 3, C and D). The down-regulation of E2F1 was confirmed also in stable RALY knock-out HeLa cells (RALY KO) generated by CRISPR/Cas9 technology compared with control vector-transfected HeLa cells (PX330) (supplemental Fig. S1 and Fig. 3E) (32).
Next, we analyzed whether E2F1 mRNA was enriched in RALY-containing ribonucleoparticles (RNPs) by performing an RNA immunoprecipitation (RIP) assay after UV cross-linking in HeLa cells followed by qRT-PCR analysis. As expected, E2F1 transcript was significantly enriched in RALY immunoprecipitates compared with both rabbit IgG and the amount of GAPDH mRNA, which was used as a negative control (Fig. 4A). RALY mRNA was used as a positive control (32).
We recently demonstrated that RALY preferentially interacts with poly-U stretches with four or more uridines within the 3Ј-UTR of transcripts (32). To confirm the direct interaction and to identify the region bound by RALY on E2F1 mRNA, we selected a suitable poly-U stretch in the 3Ј-UTR of E2F1 mRNA and performed RNA pulldown experiments using either a wildtype or a mutated RNA probe (Fig. 4B). As a positive control, we used a probe containing the poly-U stretch of the H1FX 3Ј-UTR (32). RALY was pulled down by the wild-type but not by the mutated probe of E2F1 3Ј-UTR poly-U stretch (Fig. 4B). In contrast, neither tubulin nor actinin was detected upon incubation with all the probes, excluding the possibility of unspecific bindings (Fig. 4B). As expected, the H1FX RNA probe pulled down RALY from the cell lysate (Fig. 4B). Taken together, these results show that RALY interacts with E2F1 mRNA.
Because the 3Ј-UTRs are often targeted by trans-acting factors to determine the fate of the mRNAs, we decided to investigate whether RALY could regulate E2F1 mRNA at the post-transcriptional level. In the range of possible post-transcriptional modifications, we analyzed transcript stability, as other members of the hnRNP family, in particular hnRNP-C, which shares a high similarity with RALY in the RRM, are described as regulating the stability of specific transcripts (12,21,(52)(53)(54). HeLa cells were transfected with either si-CTRL or si-RALY for 72 h and then treated with actinomycin D (ActD, 5 g/ml) for different periods of time to prevent new RNA synthesis. The total RNA was then extracted and analyzed by qRT-PCR. As shown in Fig. 5A, the stability of E2F1 in RALY-silenced HeLa cells was measurably lower over time compared with control cells, with a half-life of 7.64 Ϯ 1.32 h. In contrast, E2F1 was stable over the 9-h analyzed time-frame in cells trans- Three independent microarray experiments were performed using RNA preparations from three independent biological replicates of HeLa cells transfected with either si-RALY or si-CTRL for 72 h. A, to assess RALY silencing, total RNA and protein extracts were analyzed through qRT-PCR (normalized on GAPDH) and subjected to SDS-PAGE and Western blot analysis, respectively, with the indicated antibodies. The cells analyzed by microarray showed down-regulation of RALY at both the mRNA (upper graph) and protein (lower graph) levels. The graphs show the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test (***, p Ͻ 0.001). B, MA plot of RALY-silencing transcriptome profiling. For each gene, the average log10 signal against the RALY-silencing log2 fold change (si-RALY versus control) is plotted. Genes significantly up-regulated (blue) or down-regulated (red) upon RALY silencing are highlighted. C, classification of DEGs upon RALY silencing according to RNA classes. D, functional annotation enrichment analysis of RALY-up-regulated and -down-regulated genes. The heat map, colored-coded according to enrichment p value, displays enriched classes from gene ontology terms and the KEGG or REACTOME pathways. The number of DEGs falling in each category is displayed inside each tile. E, intersection between the lists of si-RALY up-regulated (blue) or down-regulated (red) genes in HeLa cells and the list of RALY RIP-seq targets identified previously in MCF7 cells (32). The percentage of overlap with respect to the number of DEGs is displayed beside the corresponding bar. Although the intersection could be affected by cell line differences, the increased frequency of RALY RNA targets among genes down-regulated upon RALY silencing suggests that the loss of a direct RALY-mRNA interaction is associated with the down-regulation of the target.

RALY regulates E2F1 stability
fected with si-CTRL. As a control, we used the mRNA coding for GAPDH, a non-target of RALY (Fig. 4A) considered a suitable control for assessing general mRNA stability (55). GAPDH mRNA was found to be stable over 9 h in both si-CTRL and si-RALY-transfected cells (Fig. 5B).
Taken together, these results show that the silencing of RALY induces a broad down-regulation of cell cycle-and transcription-related genes. Moreover, we show that RALY binds E2F1 mRNA and that in its absence, the stability of E2F1 mRNA is reduced. Consequently, both E2F1 mRNA and protein are down-regulated in RALY-silenced cells. Interestingly, cells expressing high levels of endogenous RALY, such as the cancerous cell lines Panc-1 and MDA-MB-231, also showed a higher expression of E2F1, compared with the less aggressive HPNE and MCF7 cell lines (Fig. 5C).

The silencing of RALY affects cell cycle progression
To understand whether these down-regulated genes could induce specific effects in HeLa cells, we measured cell proliferation, cell cycle distribution, and global RNA transcription of RALY KO and RALY-silenced HeLa cells. Cell proliferation was measured through the real-time cell analysis platform, xCELLigence (56,57). The proliferation rate of RALY KO cells was reduced compared with control cells (PX330) (Fig. 6A). To analyze the cell cycle distribution, we sorted the cells into the different phases of the cell cycle by measuring their DNA content through the incorporation of the thymidine analogue 5-ethynyl-deoxiuridine into newly synthesized DNA, successive staining with 5-FAM, and analysis with the high-content imaging system Operetta TM (58). In accordance with our previous results, RALY KO cells showed a significant enrichment in the G 1 phase and a consequent lower number of cells in S phase compared with control cells (Fig. 6B). For both experiments, similar results were obtained for RNAi-RALY-down-regulated cells (supplemental Fig. S2). These results show that the downregulation of RALY affects the normal progression of the cell cycle and impairs cell proliferation.

The silencing of RALY affects RNA transcription
To evaluate any general effect of RALY silencing on RNA transcription, we transfected HeLa cells with either si-RALY or si-CTRL for 72 h and measured the nuclear levels of newly synthesized RNA over different periods of time through the incorporation of 5-ethynyl uridine (5-EU) into new transcripts and successive staining with the fluorescent molecule 5-FAM using a click reaction (59). The down-regulation of RALY was verified and quantified by immunostaining using the high-content imaging system (Fig. 7A). Notably, for the quantification of newly synthesized RNA we selected only the successfully RALY-down-regulated cells by setting a threshold on the RALY fluorescence signal. As expected, the levels of newly synthesized RNA showed a slower increase over time in cells lacking RALY compared with control cells (Fig. 7B).
Because ribosomal RNAs are the most abundant and actively transcribed RNA species inside cells, RNA staining yielded an intense fluorescent signal in the nucleoli, making the differences in the nuclear RNA levels difficult to measure accurately. To overcome this problem, and given the fact that RALY is not present inside the nucleoli, we treated the cells with a low dose of ActD (125 ng/ml) to specifically block the activity of RNA polymerase I (59). Following this treatment, the nucleolar RNA signal was reduced (Fig. 7C). The down-regulation of RALY was verified by immunofluorescence (Fig. 7D). Under these conditions, si-RALY-transfected cells showed a slower increase over time of nuclear newly synthesized RNA compared with si-CTRL transfected cells (Fig. 7E). As in the microarray analysis, all of the analyzed genes were found to be less expressed in si-RALY cells compared with si-CTRL cells. B and C, total protein extracts were produced from HeLa cells transfected with either si-RALY or si-CTRL for 72 h. Total lysates were subjected to SDS-PAGE, and Western blot analysis with the indicated antibodies (left panels). The levels of CCNE1 (B) and CCNT1 (C) were analyzed by band densitometry analysis (right panels) and found to be down-regulated in si-RALY-compared with si-CTRL-transfected cells. All of the graphs show the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001) between the signal detected in si-RALY-and si-CTRL-transfected cells.

RALY regulates E2F1 stability
To confirm that the effect on general transcription was specifically caused by RALY down-regulation, we also measured RNA synthesis in RALY KO cells. As for the RNAi experiments, RALY KO cells showed a decreased synthesis of RNA compared with PX330 cells (supplemental Fig. S3A). The decrease in RNA synthesis was also observed in RALY KO cells upon RNA poly- Enrichment results from GSEA were mapped as a network of gene sets (nodes) related by mutual overlap (edges), where black identifies the up-regulated and light gray the downregulated gene sets. The size of each node is proportional to the size of the gene set. The size of each edge is proportional to the mutual overlap between two nodes. The gene set E2F_TARGETS, which contains "genes encoding cell cycle-related targets of E2F transcription factor," is the most enriched in downregulated genes upon RALY silencing. B, the mRNAs coding for E2F1 and E2F2 were found to be enriched in RALY-containing immunoprecipitated particles in the RIP-seq analysis performed in MCF7 cells. On the contrary, no enrichment was detected for the other members of the E2F family. The graphs represent the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test between the respective samples (***, p Ͻ 0.001). C and D, HeLa cells were transfected with either si-RALY or si-CTRL for 72 h, and both total RNA and total protein extracts were collected. RNA was analyzed by qRT-PCR and the signals normalized on GAPDH, whereas proteins were subjected to SDS-PAGE and Western blot analysis with the indicated antibodies followed by band densitometry analysis. Both the E2F1 mRNA (C) and E2F1 protein (D) are down-regulated in cells lacking RALY expression. E, PX330 and RALY KO lysates were subjected to SDS-PAGE and Western blot analysis using the indicated antibodies followed by band densitometry analysis. E2F1 expression is reduced in RALY KO cells. The graphs show the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test (**, p Ͻ 0.01).

RALY regulates E2F1 stability
merase I block, confirming that the effect observed with RNAi was specifically related to RALY down-regulation (supplemental Fig. S3B). These results show that in absence of RALY, the synthesis of new RNA by RNAPII is significantly decreased.

The absence of RALY does not impair RNAPII elongation
To determine whether the reduction in RNA synthesis could be the result of a slower RNAPII-RNA polymerization, we measured the rate of RNAPII elongation (60). We blocked RNAPIIdependent transcription by treating the cells with DRB (100 M) for 3 h. During this time, any pre-mRNAs could be processed but no new RNA was transcribed. The cells were washed extensively with PBS and incubated with new media to recover transcription, and total RNA was extracted at 5-min intervals until 120 min of reactivation was reached. qRT-PCR was performed with different exon-intron-spanning primers to measure the RNAPII elongation rate (60).
We measured the elongation speed of RNAPII on three genes: ITPR1, where the level did not change upon RALY downregulation; and OPA1 and CTNNBL1, which were, respectively, up-regulated and down-regulated in cells in which RALY was silenced (Fig. 8A). These genes were selected based on length, so that RNAPII transcription could be appreciably measured at 5-min intervals by qRT-PCR. For all of the analyzed genes, the elongation speed in cells lacking RALY was comparable with that of the control cells (Fig. 8, B-E). However, the amount of synthesized pre-mRNA was lower in cells where RALY was down-regulated, suggesting that a lower number of transcriptional events had taken place compared with control cells (Fig.  8, B-D). Taken together, these results show that the downregulation of RALY affected RNAPII-dependent transcription by lowering the total amount of newly synthesized RNA without affecting RNAPII dynamics.

RALY associates with chromatin in both an RNA-dependent and -independent manner
Our results showed the involvement of RALY in the regulation of cell proliferation and transcription. We demonstrated that RALY controlled the expression of the proliferation marker E2F1 at the post-transcriptional level, thus regulating its stability. However, RNA-binding proteins may have a plethora of roles both in the biogenesis and post-transcriptional modifications of RNA. Regarding this possibility, RALY was recently described as a transcriptional co-factor (34). For these reasons and to better define a possible direct role of RALY in the regulation of gene expression, we decided to study the interaction of RALY with chromatin more in details.
As reported previously, RALY interacts with proteins that have the ability to bind chromatin to exert different functions related to gene expression and transcriptional control (13, 24, . RALY interacts with E2F1 mRNA A, qRT-PCR was used to compare the indicated mRNAs isolated upon UV-cross-linked RALY immunoprecipitation. RNA was recovered after immunoprecipitation with anti-RALY and control anti-IgG antibodies. The relative abundance was compared with 10% of input. E2F1 mRNA is enriched in RALY-containing RNPs. Bars represent means Ϯ S.D. of three independent experiments. The p value was calculated by comparing the amount of each mRNA with the amount of GAPDH using an unpaired two-tailed t test (*, p Ͻ 0.05). B, sequences of the biotinylated wild-type and mutant E2F1 3Ј-UTR probes (top). Total protein extract from HeLa cells was incubated with either E2F1 wild-type (WT) or mutant (MUT) biotinylated RNA probes (50 pmol) and captured by streptavidin-Dynabeads. H1X 3Ј-UTR biotinylated probe was used as a positive control (CTRL) as described by Rossi et al. (32). Western blot analysis of probe-bound proteins showed the positive in vitro interaction of RALY with the wild-type poly-U sequence but not with the mutant probe. Immunoblotting with anti-actinin and anti-tubulin served as negative controls. The graph shows the mean values of the densitometry analysis of three independent experiments. Bars represent mean Ϯ S.D. The p value was calculated by an unpaired two-tailed t test (***, p Ͻ 0.001). 31). We fractionated HeLa cells into four different fractions: cytosolic, nuclear-soluble, and low-salt-and high-salt-soluble chromatin fractions (13,61). The cytosolic fraction contains all of the extranuclear proteins, whereas the nuclear-soluble fraction comprises the nuclear proteins not attached to chromatin. The low-salt-soluble chromatin fraction contains proteins associated with transcriptionally active chromatin and with transcription start sites. Finally, the high-salt-soluble chromatin fraction contains proteins associated with transcriptionally inactive chromatin and bulky complexes strongly associated with chromatin (13,61). The enrichment of RALY in these different fractions was analyzed using Western blot analysis.

RALY regulates E2F1 stability
To verify the success of the fractionation, we used ␤-tubulin, linker histone H1X, and histone H3 as the cytosolic, inactive chromatin, and whole chromatin markers, respectively. As expected, H1X was found in the high-salt-soluble fraction but not in the low-salt-soluble fraction of chromatin (Fig. 9A, lanes LS and HS). In contrast, H3 was present in both chromatin fractions (Fig. 9A). ␤-Tubulin instead was enriched exclusively in the cytosolic fraction (Fig. 9A, lane C). When we analyzed the distribution of RALY, we found that an equal amount was present in the fractions containing transcriptionally active and inactive chromatin-bound proteins (Fig. 9A). Moreover, RALY was also present in the nuclear-soluble fraction. Finally, we compared RALY distribution with FUS/TLS, a well-characterized member of the hnRNPs (62)(63)(64)(65)(66). RALY and FUS showed a similar association with the nuclear-soluble and transcriptionally active chromatin fractions but had a different interaction with the high-salt-soluble chromatin (Fig. 9A). The association of FUS with chromatin has been described as strictly RNA-depen-dent (13), whereas a portion of RALY persists on chromatin even in the absence of RNA, suggesting different molecular functions of the two RBPs. However, considering that FUS has been identified as an interactor with RALY through mass spectrometry and that RALY is characterized as a transcriptional co-factor (31,34), this shared association opens the interesting possibility of a co-operative function for the two hnRNPs in transcriptional control and RNA maturation.
To assess the involvement of RNA in the RALY-chromatin interaction, we treated cell pellets with RNase A prior to the fractionation process (Fig. 9B). Upon RNase A treatment, a statistically significant amount of RALY shifted from the low-saltsoluble chromatin fraction to the cytosolic fraction ( Fig. 9, B,  lanes 1 and 2 and 5 and 6, and C). Interestingly, the fraction of RALY associated with transcriptionally inactive chromatin did not shift upon RNA degradation (Fig. 9B, lanes 7 and 8). Taken together, these results demonstrate for the first time that RALY binds both transcriptionally active and inactive chromatin. Moreover, our data show that the interaction of RALY with active chromatin is only partially dependent on RNA.
To identify the regions of RALY involved in the interaction with chromatin, we produced constructs coding for different domains of RALY predicted previously by computational analysis (31). The constructs were tagged with a c-Myc tag at the C terminus and encompassed the N-terminal region (amino acids 1-225), the C-terminal region (amino acids 143-306), and the unstructured glycine-rich region (amino acids 225-306) (Fig.  10A). A c-Myc-tagged full-length RALY (FL, amino acids 1-306) (Fig. 10A) and the empty expression vector were used as positive and negative controls, respectively. The expression of

RALY regulates E2F1 stability
each construct in HeLa cells was analyzed by Western blot analysis (Fig. 10B).
We analyzed the subcellular localization of the recombinant proteins by fluorescence microscopy. All RALY fragments localized into the nucleus, with the exception of region 143-306, which also showed a signal in the cytoplasm (Fig. 10C). Then, we assessed the distribution of each fragment upon cell fractionation. Although all constructs were enriched in the nuclear fractions, their distribution changed upon RNase A treatment (Fig. 10D). In the absence of RNA, the N-terminal part of RALY containing the RRM (amino acids 1-225) disappeared from the nuclear-soluble and low-salt-soluble chromatin fractions (Fig. 10D, NS and LS). This behavior was not observed for the C-terminal part (amino acids 143-306) (Fig.  10D, NS and LS). As expected, full-length RALY decreased in the low-salt-soluble chromatin fraction after RNase A treatment but not in the high-salt-soluble chromatin fraction, confirming that RALY-Myc behaved the same as endogenous RALY (Fig. 10D, LS and HS).
These results underline the importance of RNA in the interaction of RALY with transcriptionally active chromatin and show that RALY can interact with chromatin using either the N-or C-terminal region. However, only the binding with tran-scriptionally active chromatin, which involves the N-terminal RRM domain, is dependent on the presence of RNA.

The localization of RALY is dependent on active transcription
Because the activity and localization of RBPs is directly related to RNA, we focused on the RNA-dependence quality of RALY interaction with transcriptionally active chromatin. We asked whether RALY localization and its association with chromatin could be affected by the inhibition of transcription. Therefore, we blocked transcription with ActD (5 g/ml) for different periods of time and analyzed the intracellular localization of RALY with a high-content imaging system (Operetta™). The inhibition of transcription was verified by measuring the newly synthesized RNA through the incorporation of 5-EU into new transcripts and successive staining with 5-FAM using Click-iT reaction (59). As expected, no incorporation of 5-EU was detected in HeLa cells treated with ActD (Fig. 11, A and B).
Interestingly, RALY nuclear staining decreased over time, suggesting that the ActD treatment activated either RALY degradation or translocation from the nucleus to the cytoplasm (Fig. 11, A and C). Because the increase in the RALY signal in the cytoplasm was close to the sensitivity limit of the highthroughput microscope for reliable quantification, we analyzed the localization of RALY by cell fraction after treating the cells with either ActD or DMSO (as control). After treating the cells with ActD, the level of RALY in the low-salt-soluble chromatin fraction significantly decreased over time (Fig. 11, D and E). In parallel, we observed a progressive increase in RALY in the cytosolic fraction. The same result was observed when we incubated the cells with both ActD and the proteasome inhibitor MG132 (supplemental Fig. S4A), indicating that RALY had shifted from the nucleus to the cytoplasm and was not degraded by the proteasome and de novo synthesized to remain outside the nucleus. The inhibition of proteasome activity by MG132 was confirmed by the accumulation of ubiquitin in the cytosolic fraction (supplemental Fig. S4B). These results show that RALY localizes on transcriptionally active chromatin in a transcriptiondependent manner and that the inhibition of transcription leads to a shuttling of RALY from the nuclear compartment to the cytoplasm. Taken together, these data demonstrating that RALY interacts with chromatin better characterize the direct role of RALY in gene expression regulation. Further experiments will be necessary to study this novel feature of RALY from the mechanistic point of view.

Discussion
The RNA-binding protein RALY has been described as interacting with numerous mRNAs to regulate the expression of specific transcripts. Moreover, the down-regulation of RALY is associated with reduced cell proliferation (32). Here we studied the association of RALY with cell cycle progression in greater detail. Our microarray analysis detected an altered expression of several transcription-, cell proliferation-, and cell cyclerelated genes and revealed enrichment of members of the E2F transcription factors family and of cell cycle-related targets of the E2F family in the down-regulated genes upon RALY silencing (supplemental Table S1 and Fig. 3A). The family of the E2F transcription factors comprises eight members that regulate

RALY regulates E2F1 stability
cell physiology by either promoting or repressing cell cycle progression (47). Interestingly, all of the cell cycle-promoting E2Fs (E2F1-3) were found to be down-regulated upon RALY silencing in our microarray experiments (supplemental Table S1). We focused on E2F1, a well-known marker of cell proliferation and promoter of the S phase, which can also induce apoptosis through different p53-dependent and -independent mechanisms (35,45,67,68). The overexpression of E2F1 has been reported in numerous types of cancer, and its ectopic expression was described to drive S-phase entry in quiescent cells and hyperplasia (48,69). We show that RALY interacts with and stabilizes E2F1 mRNA, consequently regulating the amount of E2F1 protein within the cell (Figs. 3, D and E, 4, and 5A). E2F1, together with its activating partner transcription factor DP1 (TFDP1), is normally kept in an inactivated state by the retinoblastoma protein (Rb) and is released at the G 1 /S-phase transition after the CDK-mediated phosphorylation of Rb. The E2F1-TFDP1 complex will induce the expression of S phase-promoting factors, such as cyclin E1 and E2 (CCNE). In a positive feedback loop, CCNE will induce the further phosphorylation of Rb, consequently increasing the levels of free E2F1-TFDP1 complexes (70 -72). A decreased expression of E2F1 therefore depotentiates the G 1 /S-phase transition, lowering the expression of proliferation-promoting genes such as CCNE (70).
Consequently, we observed a lower expression of direct targets of E2F1 transcriptional control activity in RALY-downregulated cells such as CCNE1, CCNE2, CDC25A, CDK1, and CCNB1 (Fig. 2, A and B) (37,38,70,(72)(73)(74). Except for CDC25A, the other mRNA targets were also detected in the RIP-seq analysis of MCF7 cells by Rossi et al. (32). Although we did not validate these direct interactions in HeLa cells, we did not observe any adverse effect on their stability upon RALY down-regulation (data not shown). Thus, we cannot exclude the possibility that, together with E2F1 down-regulation, the absence of a direct interaction with RALY might also affect their In both the ActD-treated and untreated conditions, the levels of newly synthesized RNA show a slower increase over time in si-RALY cells compared with si-CTRL cells. The absence of RALY affects RNA polymerase II-dependent transcription. All of the graphs represent the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test (*, p Ͻ 0.05; **, p Ͻ 0.01) between the respective samples of si-RALY-and si-CTRL-transfected cells. C, HeLa cells were incubated with 5-EU and treated with ActD (125 ng/ml) or DMSO for 1 h. Newly synthesized RNA was stained with 5-FAM and RALY through immunofluorescence with anti-RALY antibody. The mild ActD treatment only inhibited RNA polymerase I, abrogating the synthesis of RNA in the nucleoli. Scale bar ϭ 10 m. and CTNNBL1 mRNAs was quantified by qRT-PCR and normalized on GAPDH. The level of ITPR1 mRNA did not change significantly, whereas OPA1 and CTNNBL1 levels, respectively, increased and decreased in the absence of RALY compared with control cells. The graph shows the mean of three independent experiments Ϯ S.D. The p value was calculated using an unpaired two-tailed t test between si-RALY-and si-CTRL-transfected cells (**, p Ͻ 0.01). B-D, HeLa cells transfected either with si-RALY or si-CTRL siRNA for 72 h were treated with 100 M DRB for 3 h to block RNAPII activity. After washing with PBS, the cells were incubated in DMEM to recover transcription. Successively, total RNA was collected at 5-min intervals. qRT-PCR analysis with different exon-intron primer pairs for ITPR1 (B), OPA1 (C), and CTNNBL1 (D) pre-mRNAs was used to measure RNAPII elongation rate. The pre-mRNA expression values are plotted relative to the expression level of the untreated controls, which was set to 1 in all experiments. RALY did not impair the RNAPII elongation rate. E, the table depicts the calculated elongation rate of RNAPII along the analyzed genes for both si-RALY-and si-CTRL-transfected cells. The down-regulation of RALY did not impair RNAPII elongation rate.

RALY regulates E2F1 stability
expression. In fact, even though preliminarily, we observed that the loss of a direct interaction between RALY and mRNA could be associated with the down-regulation of the target mRNA (Fig. 1E).
Our observation of the down-regulation of E2F1, together with the lower expression of numerous transcription-and cell proliferation-promoting factors ( Fig. 2A and supplemental  Table S1), explains the decreased cell proliferation and stall in the G 1 phase of cells lacking RALY expression (Fig. 6, A and B, and supplemental Fig. S2) and the diminished RNAPII-dependent transcription in RALY-silenced cells (Fig. 7, B and E, and  supplemental Fig. S3). In this view, the two processes are directly impaired by the down-regulation of the respective activating factors in response to RALY silencing, and they also mutually affect each other. RALY therefore can be classified as a novel indirect regulator of transcription and cell cycle progression. Further experiments will be necessary to better   Figure 11. The localization of RALY is dependent on active transcription. A, HeLa cells were treated with 5-EU and either ActD (5 g/ml) or vehicle (DMSO) for 2 h. RALY was stained with anti-RALY antibody and RNA with 5-FAM. After ActD treatment, there is a higher presence of RALY in the cytoplasm compared with the DMSO-treated cells (arrowheads). Scale bar ϭ 10 m. B and C, HeLa cells were treated with 5-EU plus either ActD (5 g/ml) or vehicle (DMSO) for 2, 4, or 6 h, and the amount of nuclear newly synthesized RNA (B) and nuclear RALY (C), respectively, were assessed after staining with 5-FAM and after immunofluorescence with anti-RALY antibody using a high-throughput fluorescence microscope. The graphs represent the mean of three independent experiments and highlight the absence of new RNA synthesis (B) and the decrease of nuclear RALY (C) in HeLa cells treated with ActD. In B, the remaining RNA signal was considered background noise. The p value was calculated using an unpaired two-tailed t test comparing the 5-FAM signal (B) and the nuclear level of RALY (C) of the treated cells (ActD or DMSO) with the respective untreated sample (t ϭ 0) (***, p Ͻ 0.001). D, HeLa cells were treated with either ActD (5 g/ml) or DMSO for 0, 2, 4, or 6 h. At each time point, the cells were fractionated into cytosolic (C), nuclear-soluble (NS), low-salt-soluble chromatin (LS), and high-salt-soluble chromatin (HS) fractions. Then, each fraction was subjected to SDS-PAGE and Western blot analysis with the indicated antibodies. Along with the ActD treatment, the amount of RALY decreases in the low-salt-soluble chromatin fraction and in parallel increases in the cytosolic fraction. E, the levels of RALY in the low-salt-soluble chromatin fraction were quantified by band densitometry analysis. The graphs represent the mean of three independent experiments and highlight the significant decrease in RALY in the low-salt-soluble chromatin fraction during ActD treatment (right graph) compared with DMSO treatment (left graph). The p value was calculated comparing the level of RALY in the low-salt-soluble chromatin fraction of the treated cells with the respective untreated samples (t ϭ 0) using an unpaired two-tailed t test (*, p Ͻ 0.05; ***, p Ͻ 0.001). n.s., not statistically significant.

RALY regulates E2F1 stability
understand the whole spectra of molecular functions of RALY. For example, the analysis of nuclear and cytoplasmic RNAs could clarify whether the RALY-mediated stabilization of E2F1 mRNA, and in general of other transcripts, occurs at the level of nascent or rather mature RNA. Nevertheless, an involvement of RALY in the 3Ј-end processing and termination can be hypothesized. In fact, Fasken et al. (75) observed functional homologies between human RALY and the yeast protein Nab3, an RNA-binding protein involved in the exosome-mediated processing, termination, and degradation of noncoding RNAs (ncRNA). Nab3 and RALY share 31% sequence homology in the RRM and a certain degree of similarity also in the C-terminal region, and the expression of RALY in Nab3-negative cells rescued the thermosensitive yeast phenotype. Interestingly, to be efficient, both proteins need a functional RRM (75). RALY could be involved in the 3Ј-end processing of coding as well as noncoding RNAs, ultimately affecting RNA stability, but further experiments are needed to better elucidate this possibility. Recently, RALY was described as having a direct role in regulating the splicing of the protein arginine methyltransferase 1 (PRMT1) mRNA, coding for a factor involved in the post-translational methylation of proteins and found to be aberrantly regulated in different cancers (76). A spliced isoform of PRMT1 that includes the alternative second exon (PRMT1v2) was found to be up-regulated in breast cancer and was correlated with high aggressiveness (77). Interestingly, PRMT1 mRNA is a direct target of RALY and RALY knockdown determines a significant decrease in PRMT1v2 mRNA and protein inside the cell, consequently determining a reduction of invasiveness. The overexpression of PRMT1v2 in RALY-down-regulated cells rescues the aggressive phenotype of the breast cancer cells (76). These results demonstrate the direct involvement of RALY in the splicing of a specific mRNA with consequent effects on cell phenotype. It is therefore possible that RALY caused part of the effects observed in our study by modulating the splicing of specific factors. Further experiments will be performed to study the involvement of RALY in splicing regulation.
Our results characterized the effect of RALY on cellular physiology by the regulation of its target mRNAs. RALY was recently characterized as binding gene promoters to act as a transcriptional co-factor together with the ncRNA Lexis in the cholesterol biogenesis pathway in mouse liver (34). This prompted us to characterize in detail the association of RALY with chromatin. Here, we have demonstrated that RALY interacts with transcriptionally active chromatin and that the association is partially abrogated upon RNA degradation (Fig. 9, B and C). Based on our results, we propose that the association of RALY with transcriptionally active chromatin is mediated by two types of interaction: one, an RNA-dependent interaction through the RRM domain to bind and process RNAs possibly in synergy with other factors; and another, an RNA-independent interaction mediated by the C-terminal domain (Fig. 10D). Interestingly, the fragment of RALY (amino acids 143-306) involved in the RNA-independent interaction with chromatin contains a basic leucine zipper-like motif (bZLM) typical of DNA-binding proteins, more precisely between the amino acids 146 -214 (25,34). Because the construct containing only the GRR (glycine-rich region) domain was not found on tran-scriptionally active chromatin, we concluded that the bZLM is necessary for the RNA-independent association of RALY with active chromatin. This interaction may allow RALY to function as a transcriptional co-factor (34), by binding DNA, but could also serve to anchor RALY on chromatin to allow its N-terminal RRM domain to contact RNAs. Further experiments will better characterize the function of the two diverse bindings of RALY to transcriptionally active chromatin.
To gain more information about the cellular distribution of RALY, we analyzed the localization of the RBP FUS/TLS, also classified as hnRNP-P2, finding that the two RBPs share the localization in the nuclear-soluble and the transcriptionally active chromatin fractions (Fig. 9A) (13,64). FUS is a well-studied RBP because of its association with multiple diseases, particularly those affecting the nervous system, and has different biological functions, including transcription regulation, RNA biogenesis and post-transcriptional modification (65,66,78). Further experiments will explore the interaction and possible shared functions between RALY and FUS.
RALY showed a partial RNA dependence in its interaction with chromatin when RNA was degraded in vitro. In vivo, the presence of RNA on chromatin is mainly due to ongoing transcription, and the inhibition of RNA synthesis may lead to changes in the association of RBPs with chromatin. treatment with ActD induced a progressive delocalization of RALY from transcriptionally active chromatin to the cytoplasm, suggesting that the interaction of RALY with this fraction of chromatin requires the active synthesis of new transcripts (Fig. 11, D and  E). The transcription block might abrogate the dynamic recruitment of RALY on chromatin, disrupting the existing interactions and preventing the formation of new associations, but also make the RALY pool intended to interact with RNA unnecessary. As a consequence of the transcription block, RALY migrates from the nucleus to the cytoplasm. Whether the RNA-dependent and -independent interactions are equally affected by the block of transcription remains to be elucidated. Taken together, our results characterize RALY as an indirect regulator of cell proliferation and transcription and better define the direct interaction of RALY with chromatin, making the transcriptional control activity of RALY an interesting topic for study.
Plasmid transfection was performed using the TransIT transfection reagent (Mirus Bio LLC) according to the manu-

RALY regulates E2F1 stability
facturer's protocols. The transfections of ON-TARGETplus SMARTpool siRNAs and control siRNAs (Thermo Fisher Scientific) were performed with the INTERFERin transfection reagent (Polyplus-transfection) according to the manufacturer's protocols.

qRT-PCR
Total RNA was purified from cells using TRIzol reagent (Thermo Fisher Scientific) and retrotranscribed using the RevertAid first-strand cDNA synthesis kit (Thermo Fisher Scientific). The KAPA SYBR FAST qPCR kit (KAPA Biosystems) was used for the qRT-PCR. All primers were purchased from Integrated DNA Technologies and are listed in supplemental  Table S3. The samples were incubated in a Bio-Rad CFX96 Thermo Cycler for 40 cycles, and the results were analyzed with Bio-Rad CFX Manager, version 2.1. The relative expression was calculated according to the 2 Ϫ⌬⌬ C t method.

Constructs
Reverse transcription was performed on total RNA isolated from HeLa cells using the TRIzol reagent. Human RALY cDNA was amplified with Phusion high-fidelity DNA polymerase (New England Biolabs) and then cloned in-frame with the c-Myc/DDK tag (pCMV6-Entry Vector, Origene). The sequences of the primers are listed in supplemental Table S3.

Micrococcal nuclease cell fractionation
Cell fractionation was performed as described by Yang et al. (13). Briefly, HeLa cell pellets were incubated for 10 min on ice in nucleus separation buffer (NSB) (10 mM KCl, 1.5 mM MgCl 2, 0.34 M sucrose, 10% glycerol, 1 mM DTT, and 0.1% Triton X-100) supplemented with protease inhibitors (Roche Applied Science) and RNase inhibitors (New England Biolabs) depending on the experiment. For RNase treatment, RNase A (100 g/ml) was added to NSB, and after 10 min on ice, both the RNase A-treated and untreated extracts were incubated at 37°C for 10 min. The samples were then centrifuged at 1400 ϫ g at 4°C for 10 min to pellet the nuclei. The supernatant (cyto-solic fraction) was collected, and the pellet containing the nuclei was resuspended in NSB supplemented with 1 mM CaCl 2 and 2000 units/ml micrococcal nuclease (MNase, New England Biolabs) and incubated at 37°C for 10 min. EGTA was then added to arrive at a 2 mM concentration to stop the MNase reaction. The samples were centrifuged at 1400 ϫ g at 4°C for 10 min to pellet chromatin. The supernatant (nuclear soluble fraction) was collected, and chromatin was resuspended in NSB supplemented with 150 mM NaCl. The tubes were left in rotation at 4°C for 2 h. The samples were then centrifuged at 1400 ϫ g at 4°C for 10 min to pellet chromatin again. The supernatant was collected (low-salt-soluble fraction, transcriptionally active chromatin), and the remaining pellet was dissolved in NSB supplemented with 600 mM NaCl. The samples were left in rotation at 4°C overnight. The tubes were finally centrifuged at 1400 ϫ g at 4°C for 10 min to pellet the insoluble fraction of chromatin, and the last supernatant was collect (high-salt-soluble fraction, transcriptionally inactive chromatin). After the supernatant collections, every fraction was clarified with a centrifugation at 16,400 ϫ g at 4°C for 2 min.

RNA synthesis assay
HeLa cells were grown for 24 h to obtain 70 -80% confluence. The cells were incubated in complete DMEM supplemented with 0.5 mM 5-EU (Jena Bioscience) for the desired amount of time. When required, actinomycin D was added at the concentration of 125 ng/ml at the same moment. The cells were then fixed for 30 min at room temperature in fixing solution (125 mM Pipes, pH 6.8, 10 mM EGTA, 1 mM MgCl 2 , 0.2% Triton X-100, and 3.7% formaldehyde). Click reaction with 5-FAM-azide (5-FAM, Jena Bioscience) was performed in staining solution (100 mM Tris, pH 8.5, 1 mM CuSO 4 , 10 M 5-FAM, and 100 mM ascorbic acid (in water)). Finally, the cells were washed three times in TBS supplemented with 0.5% Triton X-100. Staining of the proteins was achieved by incubating the cells in blocking solution according to the immunocytochemistry protocol described below.

RNA pulldown
RNA pulldown was performed as described previously by Rossi et al. (32). Briefly, wild-type or mutant probes (50 pmol) were incubated with 35 l of streptavidin-coupled Dynabeads (Thermo Fisher Scientific) for 20 min at room temperature in RNA capture buffer (20 mM Tris, pH 7.5, 1 M NaCl, 1 mM EDTA). HeLa cells were washed with PBS and then lysed with the lysis buffer (20 mM Tris, pH 7.5, 50 mM NaCl, 2 mM MgCl 2 , and 0.1% Tween 20). The lysate (200 g) was incubated with biotinylated RNA probes coupled to streptavidin-Dynabeads for 1 h at 4°C under rotation. Dynabeads were then washed three times with washing solution (20 mM Tris, pH 7.5, 10 mM NaCl, and 0.1% Tween 20), solubilized in Laemmli reducing buffer, and boiled for Western blot analysis.

Immunocytochemistry and fluorescence microscopy
Cells were washed in prewarmed PBS and then fixed in 4% paraformaldehyde for 15 min at room temperature. Immunocytochemistry was carried out as described previously (32). The

RALY regulates E2F1 stability
following primary antibodies were used: rabbit polyclonal anti-RALY (A302-070A, Bethyl Laboratories), mouse monoclonal anti-c-Myc (M4439, Sigma Aldrich). Alexa 594 -and Alexa 488 -coupled goat anti-mouse and anti-rabbit IgG were used as secondary antibodies (Santa Cruz Biotechnology). Microscopy analysis was performed using the Zeiss Observer Z.1 microscope implemented with the Zeiss ApoTome module. Pictures were acquired using Zen Blue imaging software package (Zeiss) and assembled with Adobe Photoshop CS6. Images were not modified other than by adjustments to the level, brightness, and magnification.

RNA immunoprecipitation
RIP was performed as reported by Keene et al. (80) with some modifications. In brief, 4 ϫ 10 7 cells were grown in 15-cm culture dishes and irradiated once with 150 mJ/cm 2 at 254 nm using a UVLink UV cross-linker (Uvitec Cambridge). The cells were lysed in lysis buffer (10 mM HEPES, pH 7.4, 100 mM KCl, 5 mM MgCl 2 , 0.5% Nonidet P-40, 1 mM DTT, 100 units/ml RNase Out, and protease inhibitor mixture) for 3 h at Ϫ80°C and centrifuged at 10,000 ϫ g for 20 min at 4°C. Then the supernatants were precleared with 20 l of A/G-magnetic beads for 1 h at 4°C. Successively, the supernatants were incubated for 4 h at 4°C with protein A/G-magnetic beads coated either with antibody anti-RALY (A302-069A, Bethyl Laboratories) (3 g) or with normal rabbit IgG polyclonal antibody (Millipore). The beads were then washed three times with NT2 buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 1 mM MgCl 2, 0.05% Nonidet P-40, and 1% urea). The RNA was isolated with TRIzol and processed for qRT-PCR analysis. The primers are shown in supplemental Table S3.

Real-time cell analysis
The cell proliferation assay using the xCELLigence system (Roche Applied Science) was performed according to the manufacturer. Briefly, 4 ϫ 10 4 RALY KO and PX330 cells were seeded into each well of a specific xCELLigence E-plate. Cell proliferation was monitored by the instrument by measuring the impedance at the bottom of the wells for 3 days with detections every 15 min. The proliferation signals were normalized on the signals acquired at 10 h post seeding.

Cell cycle distribution analysis
Cell cycle distribution was analyzed as described by Massey (58). Briefly, 10 4 RALY KO and PX330 cells were seeded into a 96-well plate and grown for 24 h. Cells were incubated successively in DMEM plus 10 M 5-ethynyl-deoxiuridine and successively treated for Click-iT reaction as described above. Processed cells were analyzed successively with the high-content imaging system Operetta, as described by Massey (58).

RNA polymerase II elongation rate
RNA polymerase II elongation was monitored according to Singh and Padgett (60). Briefly, HeLa cells were grown in 6-well plates until 70% confluence and successively transfected with either si-CTRL or si-RALY. After 72 h, the cells were incubated for 3 h in Complete DMEM ϩ 100 M DRB to block RNAPIIdependent transcription. The cells were washed successively twice with ice-cold PBS and incubated with DMEM at 37°C to reactivate transcription. Cells were then lysed at 5-min intervals using the lysis buffer of the High Pure isolation kit (Roche Applied Science), and total RNA was extracted. Reverse transcription was performed using the ImProm-II reverse transcription system (Promega). The analysis of pre-mRNA levels was done by quantitative real-time PCR using the primers listed in supplemental Table S3. For this analysis, the qRT-PCR results of the transfected cells (si-CTRL or si-RALY) not treated with DRB were used as reference values for the normalization of the transfected cells treated with DRB for 3 h and then allowed to recover transcription.

High-content analysis
HeLa cells were plated (1 ϫ 10 4 cells/well) in 96-well plates (Corning) and transfected on the following day with either si-CTRL or si-RALY for 72 h. Then, depending on the experiment, the cells were either immunostained (for the experiment shown in Fig. 11) or incubated for different amounts of time with 0.5 mM 5-EU in warm DMEM to mark the newly synthesized RNA and successively stained with 5-FAM through Click-iT reaction (see "RNA synthesis assay" above) followed by immunocytochemistry, starting from the blocking step, as described above (for the experiment in Fig. 7). Plates were imaged on the highcontent imaging system Operetta TM (PerkinElmer). In each of the 4 wells for every condition, images were acquired in five preselected fields with a LWD (long working distance) 20ϫ objective over three channels, with ϭ 380 nm excitation/ ϭ 445 nm emission for DAPI, ϭ 495 nm excitation/ ϭ 519 nm emission for Alexa Fluor 488 or 5-FAM, and ϭ 535 nm excitation/ ϭ 615 nm emission for Alexa Fluor 594. For feature extraction, the images were analyzed using Harmony software, version 4.1 (PerkinElmer). Individual cell nuclei were segmented based on DAPI staining. For the Click-iT experiment (Fig. 7) the Select Population algorithm allowed us to identify the subpopulation of silenced cells for RALY by setting a fluorescence intensity threshold on the Alexa Fluor 594 signal. The 5-FAM signal was then measured only in the cells evaluated as silenced for RALY.

Microarray analysis
HeLa cells were transfected for 72 h with either si-CTRL or si-RALY as described above, and total RNA was extracted using the High Pure isolation kit (Roche Applied Science). Samples were prepared for hybridization following the Affymetrix WT PLUS reagent kit protocol. The amplified and labeled samples were hybridized to the Affymetrix GeneChip human transcriptome array 2.0 (Affymetrix, Santa Clara, CA). Arrays were washed and stained using the Affymetrix Fluidics Station 450, and scanned in the Affymetrix GeneArray 3000 7G scanner. The experiment was performed in biological triplicate.
The CEL files resulting from the GeneChip analysis were analyzed with Affymetrix transcriptome analysis console (TAC) 3.1 and the Bioconductor library of biostatistical packages (http://www.bioconductor.org/). 4 (82). Raw data were preprocessed and normalized using the robust multichip analysis (RMA) method. Gene average intensities were determined with Tukey's biweight average algorithm using Affymetrix default analysis settings. Differentially expressed genes (DEG) upon RALY silencing were determined adopting a double threshold based on statistical significance (unpaired one-way analysis of variance, FDR Ͻ 0.05) and log2 fold change (Ͼ0.5 for up-regulated genes and ϽϪ0.5 for down-regulated genes).
The frequencies of RALY RIP-seq targets among DEGs were tested with the "test of equal or given proportions" implemented in R. The ClusterProfiler package was used for enrichment analysis of DEG lists using annotations from the Gene Ontology (http://www.geneontology.org), 4 KEGG (http://www. genome.jp/kegg/), 4 and REACTOME (http://www.reactome. org/) 4 databases. The significance of overrepresentation was determined using an FDR threshold of 0.05. GSEA was performed on ranked DEGs against the MSigDB hallmark collection of annotated gene sets (v 5.2) (81).
Author contributions-N. C., A. R., R. A. P., and P. M. designed the experiments. N. C. and A. R. conducted the experiments. L. G. contributed to the production of the constructs and participated in the experiments with recombinant proteins (Fig. 10). R. A. P. and J. S. contributed to the conception of Fig. 8 and designed the primers for the experiments depicted therein. TT analyzed the microarray data presented in Figs. 1 and 3. N. C., A. R., and P. M. wrote the paper. All authors analyzed the results and approved the final version of the manuscript.