Collaborator of alternative reading frame protein (CARF) regulates early processing of pre-ribosomal RNA by retaining XRN2 (5′-3′ exoribonuclease) in the nucleoplasm

Collaborator of alternative reading frame protein (CARF) associates directly with ARF, p53, and/or human double minute 2 protein (HDM2), a ubiquitin-protein ligase, without cofactors and regulates cell proliferation by forming a negative feedback loop. Although ARF, p53, and HDM2 also participate in the regulation of ribosome biogenesis, the involvement of CARF in this process remains unexplored. In this study, we demonstrate that CARF associates with 5′-3′ exoribonuclease 2 (XRN2), which plays a major role in both the maturation of rRNA and the degradation of a variety of discarded pre-rRNA species. We show that overexpression of CARF increases the localization of XRN2 in the nucleoplasm and a concomitant suppression of pre-rRNA processing that leads to accumulation of the 5′ extended from of 45S/47S pre-rRNA and 5′-01, A0-1 and E-2 fragments of pre-rRNA transcript in the nucleolus. This was also observed upon XRN2 knockdown. Knockdown of CARF increased the amount of XRN2 in the nucleolar fraction as determined by cell fractionation and by immnocytochemical analysis. These observations suggest that CARF regulates early steps of pre-rRNA processing during ribosome biogenesis by controlling spatial distribution of XRN2 between the nucleoplasm and nucleolus.

Collaborator of ARF (CARF), which was identified as an ARF-interacting protein based on yeast two-hybrid screening (15,16), is found ubiquitously in almost all human tissues. CARF is mainly localized in the nucleoplasm and colocalizes with ARF in the periphery (granular region) of nucleoli (15), where ribosome biogenesis takes place. Therefore, CARF may interfere with the role of ARF in ribosome biogenesis by interacting with ARF. CARF is involved not only in the ARF-dependent p53 pathway but also in the ARF-independent p53 pathway, both of which regulate tumor cell proliferation (15,17). In the ARF-dependent p53 pathway, CARF directly interacts with the ubiquitinprotein ligase Mdm2 in a complex with ARF (18,19) and thus cooperates with ARF in activating p53 (18). In the ARF-independent p53 pathway, CARF directly interacts with p53, stabilizing and functionally activating p53 (17); however, when the amounts of CARF and the p53 complex are elevated, these complexes are ubiquitinylated by the action of Mdm2 and subsequently proteolytically degraded (17). Thus, a feedback loop appears to exist in the CARF-p53 pathway in the absence of ARF, i.e. CARF activates p53, p53 activates Mdm2, and Mdm2 degrades CARF and p53 (15)(16)(17)(18)(19). In this feedback loop, CARF can also act as a transcriptional repressor of HDM2, the human counterpart of Mdm2 (19). An Mdm2 inhibitor interferes with this feedback network (20).
Overexpression of CARF induces premature senescence in human fibroblasts (21). Similarly, replicative and stressinduced senescence triggers an increase in CARF expression and activates the p53/p21 WAF (cyclin-dependent kinase inhibitor 1A) pathway (21). In contrast, CARF depletion induces apoptosis and abnormal cell division in cultured cells (21) and suppresses tumor growth in a human tumor xenograft mouse model (22). CARF depletion also affects various cell death and survival pathways, such as those involved in mitochondrial stress, ataxia telangiectasia mutated-ataxia telangiectasia and Rad3-related, Rasmitogen-activated protein kinase, and retinoblastoma cascades (22). CARF is regulated by neuronal PAS domain protein 2 (NPAS2), a product of the circadian NPAS2 gene in MCF-10A cells (23). However, the molecular mechanisms by which CARF is involved in premature senescence, cell growth, and cell death remain unclear.
In this study, we examined CARF-interacting proteins using a proteomics approach to gain insight into the role of CARF. We show that CARF interact with 5 -3 exoribonuclease 2 (XRN2) and may be implicated in the early steps of pre-rRNA processing.

Immunoblot (IB) analysis
Proteins were separated with SDS-PAGE and electrophoretically transferred to a polyvinylidene difluoride membrane (85 mm × 55 mm; Millipore, Billerica, MA). The membranes were blocked with 5% non-fat dried skim milk in TBS for 1 h at 25 • C and incubated with an appropriate primary antibody in 1% non-fat dried skim milk in TBS overnight at 4 • C. After washing three times with TBS containing 0.1% (w/v) Tween 20 for 10 min, the membranes were incubated with a secondary antibody conjugated to alkaline phosphatase in 1% non-fat dried skim milk in TBS for 1 h, washed three times in TBS containing 0.1% (w/v) Tween 20 for 10 min, and washed once in TBS alone for 5 min. Membranes were stained in staining solution, which was prepared as a 1:50 dilution of nitro-blue tetrazolium chloride/5-bromo-4-chloro-3 -indolylphosphatase p-toluidine salt stock solution (Roche Diagnostics GmbH, Mannheim, Germany) in alkaline phosphatase buffer (100 mM Tris-HCl, pH 9.5, 100 mM NaCl, 50 mM MgCl 2 ). Quantification of visualized protein bands was carried out using ImageJ software (http://rsbweb.nih.gov/ij/).

Isolation of CARF-associated complexes
TOCARF cells were cultured to 80% confluency in four 150-mm cell culture dishes in normal medium with or without 1 ng/ml doxycycline for 48 h. After the cells were washed with phosphate buffer [PBS(−)], they were lysed in buffer A (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 0.5% IGEPAL CA-630; 1 mM Na 3 VO 4 ; and 1 mM PMSF) and incubated on ice for 30 min. The cell extracts were obtained as the supernatant following centrifugation of the lysate at 4 • C for 30 min at 22,180 × g. Cell extracts (15 mg) were incubated with anti-FLAG M2 agarose beads (Sigma) for 4 h at 4 • C. The beads were then washed five times in buffer A and then once with buffer B (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 1 mM Na 3 VO 4 ; and 1 mM PMSF). The FLAG-CARF-associated complexes were released from the anti-FLAG M2 agarose beads by adding 40 l of 0.5 mg/ml FLAG peptide (Sigma) on ice for 30 min twice.

In-gel protease digestion and liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis (GeLC-MS/MS)
These analyses were done by methods described by Fujiyama-Nakamura et al. (24) and Yanagida et al. (25), and described briefly in Supplementary Materials.

Isolation of HA-XRN2-associated proteins
TOCARF cells were cultured to 25% confluency in two 90mm cell culture dishes in normal medium with or without 100 ng/ml doxycycline. These cells were transfected with Nucleic Acids Research, 2015, Vol. 43, No. 21 10399 10 g pcDNA3.1 (+)-HA-XRN2 using the calcium phosphate method (26), cultured for 24 h, and incubated further in new medium for 24 h. After washing with Phosphate buffered saline (PBS), the cells were lysed in buffer A and incubated on ice for 30 min. Cell extracts were obtained as the supernatant following centrifugation of the lysate at 4 • C for 30 min at 22,180 × g. The cell extracts from 1 × 10 7 cells were incubated with 1 g anti-HA and 15 l protein G-Sepharose 4 Fast Flow beads (GE Healthcare) for 4 h at 4 • C. The beads were washed six times in buffer A, and the HA-XRN2-associated complexes were released from the beads with 50 l SDS sample buffer (50 mM Tris-HCl, pH 6.8; 2% SDS; 6% ␤-mercaptoethanol; 10% glycerol; and 0.05% bromophenol blue; BPB).

Isolation of endogenous CARF-and XRN2-associated proteins
Antibody-conjugated affinity beads were incubated with 1 g of normal rabbit IgG (Millipore), anti-CARF (Bethyl) or anti-XRN2 (Bethyl) with 15 l Dynabeads protein G (Thermo) for 1 h at room temperature. 293T cells were cultured 90-mm cell culture dish in normal medium. After washing with PBS, the cells were lysed in buffer A on ice for 30 min and centrifuged at 22,180 × g at 4 • C for 30 min. The supernatant (1 mg) was incubated with antibodyconjugated affinity beads for 3 h at 4 • C. The beads were washed with buffer A 5 times, and the endogenous CARF or XRN2-associated proteins were released from the beads with 50 l SDS sample buffer.

Cell fractionation
After washing with PBS, the cells were lysed in 1 ml buffer C (16.7 mM Tris-HCl, pH 8.0; 50 mM NaCl; 1.67 mM MgCl 2 ; 1 mM PMSF; 0.05% Triton X-100) and incubated on ice for 5 min. The cytosol was obtained as the supernatant following centrifugation of the lysate at 4 • C for 5 min at 1,000 × g. The precipitate was washed with 1 ml buffer C, lysed in 0.5 ml buffer D (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 5 mM MgCl 2 ; 1 mM PMSF), and sonicated twice for 20 s at an interval of 90 s with a Bioruptor (Cosmo Bio, Tokyo, Japan) at the highest setting. The nuclear extract fraction was obtained as the supernatant following centrifugation of this lysate at 4 • C for 15 min at 15,000 × g. The pellet was lysed in 0.5 ml buffer D and sonicated 10 times for 20 s at intervals of 90 s with a Bioruptor at the highest setting. The pellet from centrifugation of this lysate at 4 • C for 30 min at 15,000 × g was lysed in 0.5 ml buffer E (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 10 mM EDTA; 10 mM DTT; 1 mM PMSF) and sonicated 10 times as described above. The extract fraction representing the nucleolar/Cajal bodies was obtained as the supernatant following centrifugation of the latter lysate at 4 • C for 30 min at 16,000&nbsp× g.

Cell fractionation for immunoprecipitation (IP)
After washing with PBS, 1.2 × 10 7 cells were lysed in 1 ml buffer F (16.7 mM Tris-HCl, pH 8.0; 50 mM NaCl; 1.67 mM MgCl 2 ; 1 mM PMSF; and 0.1% Triton X-100) and incubated on ice for 3 min. The cytosol was obtained as the supernatant following centrifugation of the lysate at 4 • C for 5 min at 1,000 × g. The precipitate was washed with 1 ml buffer F, lysed in 0.5 ml of 50 mM Tris-HCl, pH 8.0, containing 150 mM NaCl; 5 mM MgCl 2 ; 1 mM PMSF; and 0.5% IGEPAL-CA630, and sonicated 10 times for 20 s at an interval of 90 s with a Bioruptor at the highest setting. Nuclear extract fraction-1 was obtained as the supernatant following centrifugation of this lysate at 4 • C for 30 min at 16,000 × g. The precipitate was lysed in 0.5 ml buffer E and sonicated 10 times as described above. Nuclear extract fraction-2 was obtained as the supernatant following centrifugation of the latter lysate at 4 • C for 30 min at 16,000 × g.

Immunocytochemical staining
Collagen-coated culture slides were prepared by adding 100 l of 50 g/ml rat tail collagen type I (Becton, Dickinson and Company, Franklin Lakes, NJ), which was dissolved in 0.02 N acetic acid, into each well of an 8-well culture slide (Becton, Dickinson and Company). Cells were grown on collagen-coated culture slides that were washed with PBS. After the culture medium was removed and the slide was washed with PBS, the cells were fixed with 3.7% formaldehyde in PBS for 10 min at 25 • C. The cells were washed with PBS containing 0.05% (w/v) Tween 20 (PBST), and permeabilized with PBS containing 0.5% (w/v) Triton X-100 for 5 min at 25 • C. The cells were then blocked with 3% (w/v) nonfat dried skim milk in PBS for 1 h, incubated with the appropriate primary antibody for 1 h at 25 • C, washed three times with PBST for 10 min, and incubated with a fluorochromeconjugated secondary antibody for 1 h at 25 • C. Finally, after being washed three times with PBST for 10 min, the cells were mounted using VECTASHIELD Mounting Medium with DAPI (Vector Laboratories, Burlingame, CA) and visualized with an Axiovert 200 M microscope (Carl Zeiss, Oberkochen, Germany).

Detection of protein-protein interactions using the mKG reporter system
Protein-protein interactions were detected in cells using the mKG reporter system (MBL, Nagoya, Japan) according to the manufacturer's instructions. Flp-In T-Rex 293 cells were cultured in collagen-coated culture slides and at a confluency of ∼60% (visually estimated based on viewing through a microscope) were transfected with 1.5 l Lipofectamine 2000 and 750 ng of the following plasmids: phmKGN-MC-FLAG-CARF and phmKGC-MC-XRN2, phmKGC-MC-FLAG-CARF and phmKGN-MC-XRN2, without plasmid as a negative control, or with pCONT-1 and pCONT-2 (MBL) as a positive control. The cells were fixed and permeabilized 48 h after transfection as described above and then were washed three times with PBST. The cells were mounted using VECTASHIELD Mounting Medium with DAPI and visualized with an Axiovert 200 M microscope. The cells were immunostained as described to detect transfected proteins. Imaging experiments using the mKG reporter system were repeated at least twice using independent transfections.

Northern blot hybridization
About 293T cells were cultured until 70% confluency in 35mm cell culture dish. Cells were transfected with scRNA (control) or siRNA for XRN2 knockdown using Lipofectamin RNAiMAX reagent (Invitrogen). After 24 h transfection, the cells were cultured in 100-mm cell culture dish for 48 h. TRex or TOCARF cells were cultured until 80% confluency in 100-mm cell culture dish in the presence (10 ng/ml) or absence of doxycycline for 72 h.
The following oligonucleotides were used as probes to detect human pre-rRNAs with northern blot hybridization: 5 external transcribed spacer (5 ETS)-1 probe

CARF associates with XRN2
To gain insight into the role of CARF in cellular function, we examined proteins associated with CARF using a combination of an epitope-tagged pull-down methodology and LC-MS/MS (28). We used a site-directed (Flp-In) recombinase-based system to generate isogenic cell lines (27) in which a cytomegalovirus promoter was used to drive the expression of CARF that was integrated at a common locus in the Flp-In T-Rex 293 genome (28). A product of a single-copy transgene of CARF was tagged at the amino terminus with a FLAG affinity purification tag that is used for visualization and purification. Using this approach, we obtained isogenic cell lines expressing FLAG-tagged CARF (FLAG-CARF) with a molecular weight of ∼70 kDa, which corresponds to that estimated from its amino acid sequence. FLAG-CARF was localized mostly in the nucleoplasm (Supplementary Figure S1A) as reported (17). We analyzed FLAG-CARF-associated proteins from TO-CARF cell lysate using anti-FLAG mAb coupled beads. Associated proteins were separated by SDS-PAGE, visualized by silver staining and several candidate proteins were identified by LC-MS/MS (28). Prominent amongst these was XRN2 (Mascot protein score 1.168, 35 peptides representing 36.5% sequence coverage, Supplementary Table  S1). Immunoblot analysis with anti-XRN2 indicated that XRN2 corresponded to a protein band with a molecular weight of ∼100 kDa and most strongly stained among the proteins associated with FLAG-CARF on the SDS-PAGE gel ( Figure 1A, arrow). Although mass-based analysis did not identify ARF, it does not necessarily mean that ARF is not there; in a complex mixture, ARF-related peptide ion signals may be masked/sequested by more abundant peptide ion signal, or ARF may be of very low abundance. We know minimally that the structural integrity of recombinant CARF overexpressed in Flp-In T-Rex 293 cells is correct, given that FLAG-CARF can interact with HA-tagged ARF when co-expressed in Flp-In T-Rex 293 cells (Supplementary Figure S1B) (15).
To further ascertain the interaction between CARF and XRN2, we performed an additional IP using TOCARF cells. FLAG-CARF binding to XRN2 was increased in a doxycycline dose-dependent manner ( Figure 1B). In addition, we performed reverse pull-down analysis using HAtagged XRN2 (HA-XRN2) as bait and observed a reciprocal interaction ( Figure 1C). Because CARF has a putative double-stranded RNA-binding domain (http://www. ebi.ac.uk/interpro/protein/Q9NXV6) and XRN2 is the homolog of yeast XRN2/Rat1, which processes RNAs with its 5 -3 exoribonuclease activity (29-32), we tested whether they associated with each other in the presence of RNA. We treated the immunoprecipitate obtained from the Flp-In T-Rex 293 cells that stably expressed FLAG-CARF with RNase A and observed that XRN2 remained associated with CARF even after RNase treatment ( Figure 1D), indicating that CARF associates with XRN2 independently of RNA. The RNase treatment released ribosomal protein S7 from fibrillarin, confirming the RNase activity. To rule out the possibility that the interactions were due to overexpression of HA-XRN2, we next immunoprecipitated endogenous XRN2 from 293T cells using an antibody against endogenous XRN2 and showed the interaction between the two proteins ( Figure 1E). We also immunoprecipitated endogenous CARF from 293T cells using an antibody against endogenous CARF, and showed reciprocal interaction between the two proteins ( Figure 1E).

CARF binds directly to XRN2 in the nucleoplasm
To identify the region in CARF that was responsible for binding to XRN2, we first constructed expression vectors for five FLAG-tagged truncated CARF mutants (N1, N2, C1, C2 and NC) (Figure 2A). These mutants were con-  6 and 7). Controls were performed with Flp-In T-Rex 293 cells that were untransfected (lanes 4 and 5) and that stably expressed FLAG-fibrillarin (lanes 8 and 9). Input fractions (10 g protein/lane) and each complex were analyzed with immunoblotting using the antibodies indicated on the left. (E) CARF-and XRN2-associated complexes were immunoprecipitated with anti-CARF and anti-XRN2 from 293T cell lysate (1 mg/1 ml), respectively (lanes 3 and 4). Normal rabbit IgG was used as a control (lane 2). Input fraction (2%) and complexes were analyzed with immunoblotting using the antibodies indicated on the left. structed based on the exon-intron boundaries of CARF (http://www.ncbi.nlm.nih.gov/nuccore/NM 017632.2) and the domain structure of human CARF (http://www.ncbi. nlm.nih.gov/protein/NP 060102.1). The expression and cellular localization of the truncated proteins were examined with SDS-PAGE and immunocytochemistry, respectively ( Figure 2B, asterisks). All truncated CARF mutants were expressed in the cells and showed the expected molecular sizes as estimated from their amino acid sequences ( Figure  2B). IB with anti-XRN2 showed that domain mutants N1 and N2 were associated with XRN2 ( Figure 2B). Those two mutants were localized mainly in the nucleoplasm, similar to wild-type CARF (Supplementary Figure S2A), suggesting that the interaction between CARF and XRN2 occurs in the nucleoplasm. In contrast, all other mutants lacking the amino-terminal region corresponding to amino acids 1-175 did not associate with XRN2 ( Figure 2B). Because those mutants were not localized in the nucleoplasm (Supplementary Figure S2A), we constructed additional expression vectors for C1 and C2 mutants fused to nuclear localization signal (NLS) and confirmed their nuclear localization (Supplementary Figure S2B). Those mutants, however, did not associate with XRN2 (Supplementary Figure S2C). Thus, N-terminal region of CARF (amino acids 1-175) is critical for its interaction with XRN2. It is interesting that mutant CARFs (N1 and N2), which can bind XRN2, seem to keep the ability to localize XRN2 in the nucleoplasm (Supplementary Figure S2A).
We next constructed expression vectors for three HAtagged truncated XRN2 mutants (xC, xN1 and xN2) to identify the region in XRN2 that was responsible for binding to CARF ( Figure 2C). These truncated mutants were constructed based on the exon-intron boundaries of XRN2 (http://www.ncbi.nlm.nih.gov/nuccore/NM 012255. 3) and the domain structure of human XRN2 (http://www. ncbi.nlm.nih.gov/protein/NP 036387.2). All truncated mutants were localized in the nucleoplasm of their expressing cells (Supplementary Figure S2D) and showed the expected molecular sizes as estimated from their amino acid sequences ( Figure 2D, asterisks). IB with anti-HA showed that domain mutant xN1 was associated with CARF (Figure 2D). Thus, the minimal region of XRN2 corresponding to xN1 (amino acids 1-680) is necessary for the interaction with CARF.
In addition, we used the mKG reporter system to detect an in vivo interaction between XRN2 and CARF in Flp-In T-Rex 293 cells. Flp-In T-Rex 293 cells were transfected with a combination of phmKGN-MC-FLAG-CARF and phmKGC-MC-XRN2 vectors, two empty vectors as a negative control, or pCONT-1 and pCONT-2 (MBL) as a positive control. In this system, mKG fluorescence is reconstituted in transfected cells when CARF and XRN2 interact with each other ( Figure 2E). Fluorescence was detected in the nucleoplasm of Flp-In T-Rex 293 cells that were transfected with phmKGN-MC-FLAG-CARF and phmKGC-MC-XRN2 vectors, but not with empty vectors ( Figure 2F). The expression of FLAG-CARF and XRN2 in the Flp-In T-Rex 293 cells that were transfected with phmKGN-MC-FLAG-CARF and phmKGC-MC-XRN2 vectors was confirmed with immunocytochemistry using anti-FLAG and anti-XRN2 antibodies (Fig-ure 2G). The transfection efficiency of phmKGN-MC-FLAG-CARF and phmKGC-MC-XRN2 vectors was about 30.9% as calculated by ratio of the number of FLAG-CARF-positive cells versus that of DAPI-positive cells (data not shown). This efficiency was very similar to that obtained using Flp-In T-Rex 293 cells transfected with the positive control pCONT-1 and pCONT-2 vectors ( Figure  2H). Finally, we prepared recombinant CARF-FLAG and showed its interaction with XRN2 ( Figure 2I). Coupled with the report that human XRN2 interacts directly with CARF (33), these data suggest that CARF interacted with XRN2 in the nucleoplasm and that the binding does not require cofactors.

CARF suppresses early steps of pre-rRNA processing and the degradation of 5 -ETS fragments
XRN2 plays a major role in both the maturation of rRNA and the degradation of the 5 -extended form of 34.5S-and 45.5S-pre-rRNAs, and in the degradation of 5 -A and 19S segments, and it has also a role in degradation of the E-2 fragment, which is generated by two endonucleolytic cleavages at site E and site 2 in the ITS1 region in mouse and human cells (34)(35)(36). Thus, we postulated that CARF plays a role in pre-rRNA processing by interacting with XRN2. To test this idea, we first examined the effects of transient overexpression of CARF on the processing of 5 -ETS regions of pre-rRNA. We used four probes, 5 -ETS-1, 5 -ETS-2, 5 -ETS-3 and 5 -ETS-4, to detect the 5 -ETS region with northern blotting ( Figure 3A). Overexpression of FLAG-CARF resulted in the accumulation of 5 -01 fragments (detected only with 5 -ETS-1 and 5 -ETS-2), A0-1 fragments (detected only with 5 -ETS-4), and 47S (detected with 5 -ETS-1 and 5 -ETS-2) relative to 28S rRNA methylene blue staining as compared with controls that expressed only the FLAG epitope (Supplementary Figure S3A-S3D). Consistent with these results, overexpression of FLAG-CARF significantly reduced 43S/45S/47S pre-rRNA detected with 5 -ETS-4, whereas it did not affect the processing of 30S pre-rRNA detected with 5 -ETS-3 and -4 (Supplementary Figure S3B). Although 30SL5 pre-rRNA was only faintly detected with 5 -ETS-2 and was not significantly increased upon the transient expression of FLAG-CARF (Supplementary Figure S3A), we detected significant increase of 30SL5 pre-rRNA with 5 -ETS-1 and 5 -ETS-2 upon the increased expression of FLAG-CARF in Flp-In T-Rex cells treated with doxycycline for 72 h ( Figure 3C). In addition, we observed that the increased expression of FLAG-CARF resulted in the accumulation of 5 -01, and 45S/47S pre-rRNA ( Figure 3B-C). We also used two additional probes (ITS1-1 and ITS2-1) to detect ITS1 and ITS2 regions by northern blotting, respectively ( Figure 3A), and showed that overexpression of FLAG-CARF also accumulated 36S/30SL5 and E-2 fragment ( Figure 3D). Thus, CARF suppresses the degradation of 5 -01, A0-1, and the 5 -extended form of 45S-pre-rRNA as well as the degradation of E-2; thus, it also affects not only the processing of 5 ETS but also the processing of ITS1. These effects of overexpression of FLAG-CARF on the processing of pre-rRNA were very similar to those obtained by knockdown of XRN2 with stealth siRNAs in HeLa cells (Figures 3B-  (ITS1-1, ITS2-1 see A). D), though we did not detect the processing of the 5 end maturation of the 28S rRNA that was also accumulated by the knockdown of XRN2 (34). Thus, CARF plays a role in suppressing the action of XRN2 during the processing of pre-rRNA.

CARF retains XRN2 in the nucleoplasm
To gain insight into the mechanism by which CARF suppresses the action of XRN2 during the early processing of pre-rRNA, we first examined whether CARF directly inhibits the nucleolytic activity of XRN2 by interacting with XRN2. We constructed expression vectors encoding FLAG-TEV-HA-tagged XRN2 (WT) and FLAG-TEV-HA-tagged XRN2 lacking the amino-terminal exonuclease domain ( N278) and transiently expressed them in Flp-In T-Rex 293 cells. WT and N278 fragments were prepared by pull-down using anti-FLAG-fixed beads (Supplementary Figure S4A). The WT protein showed nuclease activity when mixed with the synthetic RNA substrate (pSTP19 fragment), whereas the N278 protein did not (Supplementary Figure S4B). We then added recombinant GST-FLAG-CARF (Supplementary Figure S4C) to this assay system, but we found no evidence that CARF affected the RNase activity of XRN2 (Supplementary Figure S4B). Although the construct was different, minimally the recombinant TF-CARF was able to bind to XRN2 ( Figure 2I). We next examined whether knockdown of CARF affects the expression level of endogenous XRN2 or XRN2 mRNA in cells. We observed no change in the expression level of XRN2 or XRN2 mRNA with knockdown of CARF as examined with IB using anti-XRN2 (Supplementary Figure  S4D). We also observed no change in the expression level of endogenous XRN2 or XRN2 mRNA upon doxycyclineinduced overexpression of FLAG-CARF (Supplementary Figure S4E).
We finally considered the possibility that CARF affects the localization of XRN2 in the cell. Thus, we examined the cellular localization of XRN2 with immunocytochemistry before and after induction of FLAG-CARF in TO-CARF cells. In the absence of induction of FLAG-CARF, XRN2 was dispersed throughout the nucleus without a clear boundary of its staining between the nucleoplasm and the nucleolus as shown by co-localization images of XRN2 with nucleolar markers specific for the 3 sub-region of the nucleolus in TOCARF cells ( Figure 4A and B). In contrast, induction of FLAG-CARF reduced the nucleolar staining of XRN2 and revealed a clear boundary of XRN2 staining between the nucleoplasm and nucleolus ( Figure 4A-C). We treated Flp-In T-Rex 293 cells expressing FLAG-CARF with stealth siRNA for CARF and observed dispersed localization of XRN2 in the nucleus of the siRNAtreated cells when compared with the scRNA treated cells expressing FLAG-CARF ( Figure 4D). To assess the ability of CARF to further affect the localization of XRN2 in the nucleolus, we prepared the nucleolar fraction (NoE) from the nuclear extract of siRNA-or scRNA-treated HeLa cells using cell fractionation and analyzed the fractions with IB using anti-CARF and anti-XRN2. Knockdown with siRNA reduced CARF by 70% as compared with its level in the scRNA-treated cells and increased the proportion of XRN2 in the NoE fraction when compared with that prepared from the scRNA-treated cells ( Figure 4E). The total amount of XRN2 did not differ between scRNA-and siRNA-treated cells. Conversely, overexpression of FLAG-CARF reduced the proportion of XRN2 in the NoE fraction ( Figure 4F). Thus, CARF can affect the localization of XRN2 in the nucleolus.
In this study, we showed that XRN2 is another protein that binds directly to CARF, in addition to ARF, p53 and HDM2. XRN2 is a member of the eukaryotic 5PX family of exonucleases (37) that carry out 5 to 3 degradation of 5 -monophosphate-terminating RNA substrates and that are inhibited by a 5 -triphosphate or secondary structures in RNA (30,38). XRN2 is involved in pre-rRNA processing (34,35), and we also showed that CARF suppresses the processing of 45S/47S pre-rRNA and the degradation of 5 -01, A0-1 and E-2 fragments via its ability to retain XRN2 in the nucleoplasm. The present data demonstrate that CARF participates in pre-rRNA processing independently of its action on ARF, which regulates the stability of the rRNA-processing factor B23. Our observations confirmed the role of XRN2 in the processing of pre-rRNA and the degradation of its discarded fragments as reported by Wang & Pestov (34) and by others (35,36) and provided more details regarding the regulatory mechanism underlying mammalian pre-rRNA maturation and decay. CARF is expressed in all tissues (including brain, kidney, liver, lung, pancreas, placenta, colon and ovary) and in all cell lines (including U2OS, Saos-2, HeLa, C33A, H1299, WI38, WET, MRC5) tested so far, except for the cell line MCF7, and its expression level varies greatly among tissues and cell lines (16,17). The ubiquitous presence of CARF is consistent with its role in ribosome biogenesis, a fundamental function of the cell. Because the expression level of CARF regulates the availability of XRN2 in the nucleolus, different expression levels of CARF are expected to differently affect pre-rRNA processing in different tissues and cell lines. Thus, we have identified a new role for CARF in the regulation of the early processing of pre-rRNA and, possibly, in the regulation of cell growth.
Previous evidence indicates that CARF has a role in both cell growth and cell arrest. In addition, CARF overexpression results in growth arrest with no apparent signs of apoptosis in a cellular background lacking ARF, suggesting that overexpression of CARF supports p53-mediated growth arrest of cells independently of ARF (15). Our results provide another possible explanation for growth arrest, in which overexpressed CARF retains XRN2 in the nucleoplasm and suppresses the early steps of pre-rRNA processing. Perturbation of ribosome biogenesis activates p53, which leads to cell cycle arrest and apoptosis as reported in animal models of Treacher-Collins syndrome, a congenital disorder of craniofacial development arising from mutations in TCOF1 (39,40). Therefore, overexpression of CARF may perturb ribosome biogenesis, leading to p53-dependent cell cycle arrest. CARF interacts with p53 in the nucleoplasm, stabilizing and functionally activating p53 in the absence of ARF (15,17). Thus, determining whether XRN2 competes with p53 to interact with CARF in the nucleoplasm will be interesting. An additional feedback loop may regulate cell growth and arrest.