Ars2 modulates arsenic sensitivity and colony formation
When we isolated Ars2 cDNAs from a murine hematopoietic cell line the clones obtained had an open reading frame coding for a predicted protein of 875 amino acids, in contrast to the 225 amino acids coded for by the open reading frame of the cDNA originally isolated (Rossman and Wang, 1999
). The predicted protein contained an amino-terminal arginine-rich domain and an RNA recognition motif in addition to the carboxy-terminal zinc finger domain originally identified. When expressed in either murine fibroblasts or the human cell line K562 this clone of Ars2 resulted in no increase in resistance to arsenic trioxide compared to control transfected cells at a range of doses (data not shown).
Since the original reported Ars2 clone appeared to be a truncation of the full-length gene product, we reasoned that it may have functioned as a dominant negative. 3T3 MEFs infected with shRNA to Ars2 (shArs2-1) showed effective depletion of the protein after four days in culture as determined by Western blot analysis () compared to cells treated with empty vector control retrovirus (shVec). When Ars2 depleted cells were treated with arsenic trioxide for 48 hours the cells exhibited a slower rate of cell death when compared to control infected cells treated with arsenic (). However, few of the cells infected with Ars2 shRNA were able to proliferate when subsequently passaged in arsenic-free medium (data not shown).
Ars2 is required for cell proliferation
To extend these findings, 3T3 clonal cell lines were generated (hp1–11 and hp2–9) expressing shRNAs targeted to two independent sequences within the Ars2 mRNA. These stable cell line clones showed reduced levels of Ars2 in comparison to two different vector control clonal cell lines (vec-7, vec-12) (). The sensitivity of these cell lines to arsenic trioxide was assessed by colony forming assay. Ars2 knockdown cell lines consistently formed one-half the number of colonies as control cell lines ().
Ars2 is required for cell proliferation
The inability of viable Ars2-deleted cells to recover following treatment with arsenic trioxide suggested the possibility that these cells had a proliferative defect. 3T3 MEFs were treated with Ars2 shRNA and population doublings measured over time in culture. shArs2-1 caused a near complete loss of detectable Ars2 protein () and suppressed proliferation for four days in culture, while shArs2-2 caused a lesser amount of Ars2 depletion and had a more modest effect on cell proliferation (). The proliferative defect of Ars2-deficient cells was associated with a reduced ability of Ars2 knockdown cells to incorporate BrdU following serum stimulation (). Ars2 deficient cells did not arrest at a particular stage in the cell cycle or exhibit a significant amount of cell death until after they had ceased progressing through the cell cycle.
To independently confirm the effects of Ars2 on proliferation, a floxed allele of Ars2 was generated in murine embryonic stem cells by homologous recombination (Supplemental Figure 1
). Injection of these cells into blastocysts resulted in chimeric animals and further breeding established mice homozygous for the targeted allele (Ars2flox/flox
) from which MEFs were generated and immortalized by transfection with SV40 large T antigen.
Genetic deletion of Ars2 in the Ars2flox/flox (fl) MEFs was accomplished by infecting the cells with a retrovirus expressing Cre recombinase. By 72 hours post-infection Ars2 protein levels were significantly diminished compared to wild-type (wt) MEFs infected with Cre (). Prior to this timepoint, there was no difference in the number of cells that had accumulated in culture. However, between 72 and 120 hours post-infection, the number of Ars2flox/flox cells that had accumulated was significantly diminished compared to wt cells (). Over this timecourse, there was an inhibition of cell proliferation comparable to that observed using Ars2 shRNA.
Ars2 genetic deletion causes proliferative arrest and bone marrow hypoplasia
Analysis of Ars2 expression in vivo revealed high expression in hematopoietic tissues and reduced or absent expression in parenchymal organs like liver and kidney ( and data not shown). To examine the effects of Ars2 deletion in proliferating tissues in vivo Ars2flox/−
mice were crossed to mice harboring an MxCre transgene to obtain mice of the compound genotype Ars2fl/−
; MxCre (fl/−; MxCre). The MxCre transgene has been used to mediate inducible recombination in a wide variety of tissues and with high efficiency in the liver and hematopoietic compartments (Kuhn et al., 1995
). Injection of these mice with the nucleotide analog polyinosinic-polycytidylic acid (pIpC) led to induction of Cre recombinase and deletion at the Ars2 locus as evidenced by Western analysis from hematopoietic tissues (, lane 6). Mice harboring one wt allele and one floxed allele of Ars2 (fl/wt) served as controls and did not have depletion of the Ars2 protein, even in the presence of the MxCre transgene (, lanes 4 and 5). Histological examination of Ars2flox/−
;MxCre mice sacrificed 9 days following pIpC treatment revealed decreased bone marrow cellularity compared to control (fl/−) injected mice (). Other hematopoietic organs including the thymus and spleen showed histological evidence of increased apoptosis (not shown). In contrast, the histology of other organs including the liver, heart, brain, kidney and lungs was unaffected in both Ars2flox/−
;MxCre and control mice following pIpC injections ( and data not shown). Despite the ability of MxCre to induce recombination in liver following pIpC (Kuhn et al., 1995
), liver function tests including serum alanine aminotransferase and alkaline phosphatase levels did not show significant differences between fl/−;MxCre and control mice following pIpC injection.
Ars2 interacts with the nuclear cap binding complex
To gain further insight into Ars2 function a biochemical screen for interacting proteins was performed. Flag-Ars2 was transiently transfected into 293T cells and immunopurified with anti-Flag antibody. Co-precipitated proteins were identified by liquid chromatography coupled to mass spectrometry. Two proteins that specifically enriched with Flag-Ars2 immunoprecipitation but not with control immunoprecipitation were identified: the 80 kDa subunit of the nuclear cap binding complex (CBC) and importin α.
Immunoprecipitation of Flag-Ars2 followed by Western blot analysis with specific antibodies confirmed that Ars2 co-precipitates both the 80 kDa subunit of the nuclear cap binding complex (CBP80), as well as the 20 kDa CBC subunit (CBP20) and importin α/β heterodimeric nuclear import receptor (). Importin α has been previously shown to be a constitutive component of the nuclear CBC (Gorlich et al., 1996
). Under the conditions tested Flag-Ars2 did not co-immunoprecipitate other molecules that have been reported to bind the nuclear cap binding complex, including the mRNA export adaptor ALY/REF (Cheng et al., 2006
) or the nonsense-mediated decay factor Upf1 (Hosoda et al., 2005
) (data not shown). Furthermore, Ars2 did not co-immunoprecipitate the cytoplasmic 7-methyl guanosine (7mG)-associated complexes eIF4A or eIF4E (data not shown), nor the highly abundant RNA binding protein hnRNP A1 ().
Ars2 interacts with the nuclear cap binding complex
To verify that Ars2 interacts with the nuclear CBC, Flag-tagged CBP80 was immunopurified from transfected 293T cells using anti-Flag antibody. Ars2 was enriched along with other CBC complex components in this immunoprecipitate (). Furthermore, the sedimentation profile of endogenous Ars2 and CBP80 proteins was compared using sucrose gradient analysis. Ars2 was concentrated in the light fractions, whereas CBP80 was detected throughout the gradient with a significant amount unresolved in the pellet fraction (). Sucrose gradient analysis carried out in the presence of RNase A revealed that both Ars2 and CBP80 sedimentation profiles collapsed to fractions 3 to 5 at the top of the gradient (). Despite the observation that Ars2 sedimentation profiles were sensitive to RNase A treatment, the ability of Ars2 to co-immunoprecipitate with CBP80 was not dependent on RNA. Immunoprecipitation of Flag-Ars2 in the presence of RNase A did not abolish the Ars2-CBC interaction (). In contrast, a small amount of eIF4G immunoprecipitated by Ars2 was found to be sensitive to RNase treatment and served as a positive control for RNase A activity ().
Ars2 interacts with 7-methyl guanosine capped RNAs
To confirm that endogenous Ars2 interacts with the nuclear CBC, the ability of Ars2 antibodies to induce a gel shift in CBC-RNA assembly reactions was examined. Using previously published conditions (Izaurralde et al., 1992
), a 7mG-dependent gel shift of capped and radiolabeled RNA probe was resolved by native gel electrophoresis (, complex, lane 3). This shift was 7mG-dependent as a significantly diminished shift was observed using 2,2,7-trimethylguanosine (TMG)-capped RNAs (, lane 4), for which the nuclear CBC has a low affinity. The 7mG-dependent shift has been reported to contain the nuclear CBC complex proteins CBP80 and CBP20 (Izaurralde et al., 1995
). This was confirmed by incubating assembly reactions with a rabbit antibody raised against full-length CBP20 which resulted in an inhibition of the 7mG-dependent shift (data not shown) as has been previously reported for similar antibodies (Izaurralde et al., 1995
). An Ars2-specific antibody reproducibly supershifted a fraction of the RNA binding complex (, supershift, lane 11). A supershift was not observed with control rabbit antiserum (lanes 5–8), TMG-capped RNA (lanes 2, 4, 6, 8, 10, 12) or in the absence of nuclear extract (lanes 1, 2, 5, 6, 9, and 10).
Like other CBC components, Ars2 interacts with 7mG-capped RNAs and shuttles in a Crm1-dependent fashion
The protein components of the cap binding complex shuttle between the nucleus and the cytoplasm. If Ars2 is associated with the CBC during nuclear export it should also undergo shuttling. Expression of V5-tagged Ars2 in 3T3 MEFs showed Ars2 to be a predominately nucleoplasmic protein, with faint staining of the cytoplasm (data not shown). In heterokaryon shuttling experiments, MEFs stably expressing V5-tagged Ars2 were fused to HeLa cells in the presence of cycloheximide to inhibit new protein synthesis. After two hours, immunofluorescence showed that Ars2 had accumulated in HeLa nuclei of heterokaryons (, upper panels). Addition of leptomycin B to inhibit Crm1-dependent nuclear export strongly decreased the shuttling of Ars2 (, lower panels).
Ars2 and the nuclear CBC are required for miRNA-mediated RNA interference
The data suggest Ars2 binds the nuclear CBC ( and ). In Arabidopsis, mutations in genes encoding the plant homologues of CBP80 (called ABH1) and Ars2 (SERRATE) have similar phenotypes (Bezerra et al., 2004
; Hugouvieux et al., 2001
; Papp et al., 2004
). Mutation of SERRATE
is proposed to result in developmental defects because of its required role in the processing of miRNAs (Grigg et al., 2005
Based on these observations the effect of Ars2 on miRNA-mediated gene silencing in mammalian cells was tested. Two luciferase reporters that are sensitive to let-7-mediated miRNA repression were used. The first (“lin28 site”) contains Renilla luciferase with three let-7 target sites in the 3′UTR derived from the lin28 transcript. The second (“perfect match”) is a similar construct with the lin28 site mutated to be perfectly complementary to let-7 to trigger Argonaute 2 (Ago2)-dependent slicer activity and reporter mRNA cleavage. Transfection of siRNAs was first performed to deplete Ars2 or CBP80 from HeLa cells (Supplemental Figure 2
) and then the cells were transfected with the reporter constructs. siRNAs to Ago2 were used as a control and caused a 1.5-2 fold loss of repression compared to control siRNAs for both reporters (). Two siRNAs targeted to Ars2 (Ars2-1, Ars2-2) or siRNA to CBP80 inhibited miRNA-mediated repression to an equal or greater extent compared to Ago2 siRNA for the lin28 site reporter and had a significant, but lesser, effect on the perfect match reporter (). A mutant reporter with a single base pair insertion in the lin28 site seed region to disrupt let-7 binding led to a loss of repression under all conditions tested and verified that our assay was responsive to let-7 (data not shown). Addition of excess let-7/let-7* duplex to the transfection mix rescued the loss of repression afforded by Ars2, but not Ago2, depletion ().
Ars2 and CBC contribute to miRNA-mediated gene silencing
Depletion of either Ars2 or DGCR8 with two independent siRNAs led to a decrease in let-7 miRNA levels compared to control treated cells (). miR-21, another miRNA whose expression has also been implicated in the regulation of cell proliferation, was also decreased by depletion of Ars2 from cells. Although pre-let-7 was not detectable by Northern blotting, depletion of Ars2 protein also led to a clear reduction of pre-miR-21 (Supplemental Figure 3
). However, not all miRNAs were similarly affected; the levels of several housekeeping miRNAs were not found to be affected by depletion of Ars2 (Supplemental Figure 3
In the nucleus, primary miRNA transcripts are processed by the Microprocessor complex that contains Drosha and DGCR8, whereas Dicer functions to process pre-miRNAs in the cytoplasm. To determine whether Ars2 could interact with either the Dicer or Drosha processing complexes, immunoprecipitation of Flag-Drosha and Flag-Dicer was performed from 293T cells. Western blotting on these immunoprecipitates revealed that endogenous Ars2 and CBP80 co-precipitated with Drosha, but not Dicer (). RNase A treatment failed to disrupt co-precipitation of endogenous Ars2 and CBP80 with Drosha. Similarly when extracts from 293T cells transfected with V5-tagged Ars2 (V5-Ars2) were used for immunoprecipitation, Drosha was co-precipitated ().
Ars2 functions at the level of the primary miRNA transcript
To test whether miRNA maturation downstream of Drosha was preserved in the absence of Ars2 or CBP80, the let-7-responsive luciferase reporters were transfected into cells in the presence or absence of excess pre-let-7 that is not dependent on Drosha for activity. When Ars2, CBP80 or Ago2 were depleted from cells a loss of let-7-mediated repression was observed for both lin28 site and perfect match reporters (). Addition of excess pre-let-7 to the transfection mix rescued the loss of repression in cells that had been depleted of Ars2 or CBP80 for both reporters (). When the levels of endogenous pri-miRNA transcripts were analyzed, depletion of either Ars2 or CBP80 led to a decrease in pri-miR-21 RNA (). In contrast, as previously reported (Gregory et al., 2004
; Han et al., 2004
; Landthaler et al., 2004
), depletion of DGCR8 led to increased levels of pri-miR-21 compared to control treated cells as detected by RT-PCR ().
The above data indicate that the Ars2/CBC complex may be involved in the stability of pri-miRNAs and/or their processing. Recently, the plant Ars2 homolog SERRATE has been implicated in regulating the fidelity of pri-miRNA processing by the DCL1 complex (Dong et al., 2008
). To assess whether Ars2 plays a similar role in mammalian cells, primary miRNA processing assays were performed. As shown in , when pri-miR-155 (pri) was incubated with a 293T extract under processing conditions, a cleavage product of approximately 65 nucleotides (pre) was obtained that was indistinguishable from that generated by incubation with immunopurified Drosha complexes, whereas control immunoprecipitations contained no activity. Extract prepared from MEF cells gave rise to an identical cleavage product as that generated by immunopurified Drosha complexes ().
To determine if Ars2 contributed to the primary miRNA processing activity in MEFs, extracts were generated from Ars2flox/flox cells infected with Cre recombinase (KO). These KO extracts contained diminished processing activity compared to control extracts prepared from wt cells infected with Cre recombinase which generated the correct 65 nucleotide cleavage product (). KO extracts generated a substantial fraction of cleavage products of smaller size than that generated by wt extracts. No processing was observed in the absence of extract. Quantitation of the indicated pri and pre bands revealed that KO extracts contained approximately 3 fold less correct processing activity compared to wt extracts (). Comparable results were obtained with pri-miR-21.
Ars2 and CBC are differentially regulated during cell quiescence and proliferation
To examine whether Ars2 is a constitutive component of the nuclear CBC or a more specialized component, the expression of Ars2 and other components of the CBC were examined as cells underwent transitions between quiescence and proliferation. MEFs undergoing exponential proliferation in 10% serum contain readily detectable protein levels of CBP80 and CBP20, as well as Ars2. However, upon serum starvation in 0.1% serum for 48 hours, a treatment that causes exit from the cell cycle, Ars2 protein levels decreased substantially (). Under the same conditions, levels of nuclear CBC were only modestly affected. As a control, the levels of actin were slightly increased by serum starvation whereas the levels of the RNA binding protein hnRNP A1 were unchanged ().
Non-proliferating cells down-regulate Ars2 and display altered miRNA processing
Regulation of Ars2 protein levels were also examined in an independent cell system. Hematopoietic cells derived from bax−/−
mice have been shown to reliably exit the cell cycle upon IL-3 withdrawal and then re-enter cell proliferation in a synchronized fashion following restimulation with IL-3 (Lum et al., 2005
). Cells maintained in the presence of IL-3 (day 0) had a mean resting cell size of approximately 700 femtoliters (fL) that rapidly declined upon IL-3 withdrawal as the cells exited the cell cycle. In these cultures proliferation reproducibly ceased 3 days later (). Cell size then slowly declined as the cells engaged in macroautophagy to survive over the next 11 days in culture. At day 14 IL-3 was added back to the medium and the cells began to reaccumulate cell mass without proliferating. By day 18 the cells had achieved their original cell size and proliferation resumed one day later (). Ars2 was highly expressed at day 0 in the presence of IL-3 and was depleted by day 3 when proliferation ceased (). Furthermore, Ars2 protein levels were not initially induced during cell growth (days 15–18), but became highly expressed upon resumption of proliferation at day 19 (). In contrast, the levels of CBP20 did not correlate with cell proliferation, cell growth or quiescence and were maintained invariantly throughout the course of the experiment. CBP80 levels were maintained for a longer period following growth factor withdrawal than Ars2 levels, but quickly recovered upon IL-3 readdition during the cell growth phase (days 15–18).
Whether the downregulation of Ars2 protein levels associated with withdrawal from the cell cycle affected the ability to process proliferation-associated miRNAs was investigated using primary miRNA processing assays. Extracts prepared from MEFs undergoing proliferation in 10% serum efficiently processed a miR-155 pri-miRNA substrate to the correct pre-miRNA product (). In contrast, extracts from cells cultured in 0.1% serum for 48 hours gave rise to a decreased proportion of correctly processed pre-miRNA products and increased heterogeneity of the products of primary miRNA processing (). To determine whether the alteration in primary miRNA processing observed during cell cycle arrest could be rescued by addition of excess Ars2 protein, extracts prepared from MEFs cultured in 0.1% serum was mixed with extracts prepared from wt or Ars2-depleted (KO) MEFs grown in 10% serum and primary miRNA processing assays were performed. Mixture of extract from wt cells grown in 0.1% serum with extract from wt cells grown in 10% serum gave rise to correctly processed pre-miRNAs when incubated with pri-miR-155 substrate (Supplemental Figure 4
). In contrast, mixture of extract from wt cells grown in 0.1% serum with extract from KO cells grown in 10% serum failed to generate properly processed pre-miRNAs.