Transcriptional Properties of Mammalian Elongin A and Its Role in Stress Response*

Background: Transcriptional elongation is a rate-limiting step in activation of stress response genes. Results: Optimal expression of stress response regulator ATF3 requires the elongation activity but not the ubiquitination activity of Elongin A. Conclusion: Elongin A plays a key role for the adequate expression of ATF3 in vivo. Significance: RNAPII ubiquitination and transcriptional elongation are independent activities of Elongin A. Elongin A was shown previously to be capable of potently activating the rate of RNA polymerase II (RNAPII) transcription elongation in vitro by suppressing transient pausing by the enzyme at many sites along DNA templates. The role of Elongin A in RNAPII transcription in mammalian cells, however, has not been clearly established. In this report, we investigate the function of Elongin A in RNAPII transcription. We present evidence that Elongin A associates with the IIO form of RNAPII at sites of newly transcribed RNA and is relocated to dotlike domains distinct from those containing RNAPII when cells are treated with the kinase inhibitor 5,6-dichloro-1-β-d-ribofuranosylbenzimidazole. Significantly, Elongin A is required for maximal induction of transcription of the stress response genes ATF3 and p21 in response to several stimuli. Evidence from structure-function studies argues that Elongin A transcription elongation activity, but not its ubiquitination activity, is most important for its function in induction of transcription of ATF3 and p21. Taken together, our data provide new insights into the function of Elongin A in RNAPII transcription and bring to light a previously unrecognized role for Elongin A in the regulation of stress response genes.

Elongin A was shown previously to be capable of potently activating the rate of RNA polymerase II (RNAPII) transcription elongation in vitro by suppressing transient pausing by the enzyme at many sites along DNA templates. The role of Elongin A in RNAPII transcription in mammalian cells, however, has not been clearly established. In this report, we investigate the function of Elongin A in RNAPII transcription. We present evidence that Elongin A associates with the IIO form of RNAPII at sites of newly transcribed RNA and is relocated to dotlike domains distinct from those containing RNAPII when cells are treated with the kinase inhibitor 5,6-dichloro-1-␤-D-ribofuranosylbenzimidazole. Significantly, Elongin A is required for maximal induction of transcription of the stress response genes ATF3 and p21 in response to several stimuli. Evidence from structure-function studies argues that Elongin A transcription elongation activity, but not its ubiquitination activity, is most important for its function in induction of transcription of ATF3 and p21. Taken together, our data provide new insights into the function of Elongin A in RNAPII transcription and bring to light a previously unrecognized role for Elongin A in the regulation of stress response genes.
Transcription initiation by mammalian RNA polymerase II (RNAPII) 3 requires a set of general initiation factors TFIID (or TBP), -IIB, -IIF, -IIE and -IIH, which all assemble with RNAPII to form a preinitiation complex (PIC). After initiation, RNAPII completes transcribing successfully at a fairly constant rate of 3-4 kb/min even in the case of an extraordinarily long (e.g. megabase) transcript. However, in some cases, transcribing RNAPII is subject to promoter-proximal pausing. In addition, subsequent to release from the promoter-proximal pause site, RNAPII is subject to frequent pausing, resulting in premature termination or inefficient elongation of nascent transcripts.
Transcript elongation by RNAPII can be regulated by a collection of elongation factors of which there are at least 20 in mammals (1,2). Negative elongation factors, such as DSIF and NELF, are required for promoter-proximal pausing. Reactivation of paused RNAPII depends on a multiprotein complex called the super elongation complex, which contains the positive transcription elongation factor P-TEFb (Cdk9/CyclinT), elongation factor ELL/EAF (eleven-nineteen lysine-rich in leukemia/ELL-associated factor), and a collection of additional proteins (3)(4)(5)(6). Some factors, such as FACT (7) and Elongator (8), function in a chromatin-dependent manner. ELL/EAF (9,10), TFIIF (11)(12)(13), Cockayne syndrome protein B (14), and Elongin (15,16) are all capable of activating the overall rate of elongation by suppressing transient pausing or by arrest of RNAPII, whereas SII family members reactivate arrested RNAPII after partial cleavage of the 3Ј-end of the transcript (17,18). Although the biochemical activities of these factors have been well documented, their contributions to gene regulation in cells are only now beginning to emerge. For example, an RNAPII elongation factor, Elongin A, plays an essential role in mouse development especially in neuronal differentiation (19).
Elongin is a multimeric elongation factor comprising three subunits, Elongins A, B, and C (15,16,20,21). Elongin A inter-acts directly with RNAPII in vitro (22) and carries the elongation stimulatory activity of Elongin (21); at least three isoforms of Elongin A (A, A2, and A3) are present in human (23,24). Elongin A contains an N-terminal region (residues 1-120) similar to the N terminus of SII that is dispensable for its elongation activity and a C-terminal elongation stimulatory domain that falls between residues 400 and 773 (25). Within this C-terminal domain is a SOCS box containing the Elongin B and C binding site at residues 550 -588 that together with Elongins B and C is required for maximal transcriptional activity in vitro (25). Recently, we showed that the C-terminal elongation activation domain contains a binding site for RNAPII between residues 590 and 690; this region of Elongin A is crucial for in vitro elongation activity (22).
Elongin A acts not only as a component of a transcript elongation factor but also as the substrate recognition subunit of an Elongin BC-containing ubiquitin ligase complex in which Elongins B and C link Elongin A to Cullin 5 and the RING finger protein Rbx2. Several recent studies including our own have suggested that the Elongin ABC-Cul5/Rbx2 ubiquitin ligase contributes to ubiquitination and degradation of RNAPII stalled at sites of DNA damage (26,27).
We previously carried out a comprehensive microarray analysis that revealed that expression of only a small fraction of genes (ϳ3-4%) is affected in Elongin A-deficient mouse ES cells, raising the possibility that Elongin A is not a global regulator of gene expression but rather controls the expression of select genes (28). More recently, we have generated Elongin A-deficient mice (29) and showed that Elongin A plays a crucial role in expression of a variety of genes for neural development (19). We also observed that mouse embryonic fibroblasts (MEFs) derived from Elongin A-null embryos displayed both increased apoptosis and senescence-like growth defects (29). These phenotypes were accompanied by activation of p38 mitogen-activated protein kinase and p53, suggesting a possible role for Elongin A in stress responses.
Typically, stress response genes must be activated rapidly in reaction to stimuli. To guarantee the swiftness of activation, RNAPIIs that are paused near the 5Ј-ends of genes are in a "poised" state. Increasing evidence indicates that release from this pausing is the rate-limiting step for the regulation of gene activation in vivo (30). For example, P-TEFb (31,32), NELF and DSIF (33), FACT (34), and Elongin A (35) are reported to be involved during the activation of HSP70 gene. Furthermore, P-TEFb and NELF are implicated in epidermal growth factormediated c-fos (36,37) and junB induction (38), respectively. However, functional details of Elongin A in in vivo stress response are still elusive.
In this study, we explored further the role of Elongin A in stress responses. Our results provide evidence for a role for Elongin A in controlling expression of the leucine zipper transcription factor and stress response regulator activating transcription factor 3 (ATF3). Arguing that Elongin A directly regulates ATF3 transcription, we show that it is recruited to the transcribed region of ATF3 gene in response to various stimuli. Furthermore, domains of Elongin A essential for maximal induction of the stress response were also identified.
Cell Culture, Cell Extract Preparation, and Western Blot-HeLa cells were cultured in DMEM supplemented with 10% fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin in 5% CO 2 at 37°C. Cells treated as indicated were harvested, washed in phosphate-buffered saline (PBS), and resuspended in lysis buffer (50 mM Hepes-KOH, pH 7.5, 150 mM NaCl, 1% TritonX-100, 1.5 mM MgCl 2 , 1 mM EGTA, 0.1 mM PMSF, 10 g/ml each leupeptin and aprotinin, 200 M sodium vanadate, 100 mM NaF, 10% glycerol). After incubation on ice for 10 min, the cells were centrifuged at 10,000 rpm for 10 min, and the supernatants were taken as whole cell extract. Protein concentrations were measured by the Lowry method using bovine serum albumin as standard (39). Cell extracts (20 g of protein) were separated by SDS-PAGE, transferred onto PVDF membranes, subjected to Western blotting, and detected using ImmunoCruz (Santa Cruz Biotechnologies, Inc.) according to the manufacturer's instructions.
Preparation of MEFs-MEFs were prepared from E10.5 embryos as described (29). Briefly, fetal tissue samples were rinsed in PBS and mechanically dissociated in DMEM with 10% FBS. All animal experiments were done with the approval of the Institutional Animal Care Committees of the Tokyo Medical and Dental University and Kochi Medical School.
HeLa Cell Lines Stably Expressing Elongin A or RAP74 Subunit of TFIIF-HeLa cells were transfected with pCI-neo-hElongin A using Superfect (Qiagen) and cultured in the presence of 800 g/ml neomycin for 9 days. Drug-resistant clones were isolated and assayed for FLAG-tagged Elongin A expression using reverse transcription (RT)-PCR and Western blot. To estab-lish cell lines expressing FLAG-RAP74, HeLa cells stably expressing mouse ecotropic retrovirus receptor were infected with retrovirus vector pMX-puro encoding FLAG-tagged human RAP74, and clonal cell lines positive for FLAG-RAP74 expression were selected in medium containing 20 g/ml puromycin.
Immunoprecipitation and Complex Analysis-Nuclear extract of HeLa cells stably expressing FLAG-Elongin A or FLAG-RAP74 or control cells (1 ϫ 10 7 cells) were prepared as described by Dignam et al. (41). Extracts were then incubated with 0.5 g of anti-FLAG antibody or control IgG at 4°C for 3 h in buffer containing 20 mM Hepes-KOH, pH 7.9, 0.2 mM EDTA, 20% glycerol, 0.2% Nonidet P-40, 1 mM PMSF, 0.5 mM DTT, 80 mM NaCl or 150 mM NaCl followed by incubation with 30 l of Protein G-agarose (Roche Applied Science) for 2 h. The resulting immunocomplexes were washed with the same buffer and subjected to Western blot analysis.
Immunostaining and Labeling of RNA-Cells were fixed with 2% formaldehyde in PBS for 20 min and 0.5% Triton-X for 20 min and then incubated with 100 mM glycine for 10 min. After blocking in PBS containing 3% BSA, cells were incubated with primary antibody at room temperature for 1.5 h or at 4°C overnight, washed, and sequentially incubated with anti-mouse or anti-rabbit IgG antibody conjugated with Alexa Fluor 488 or Alexa Fluor 568 (Molecular Probes). Cell nuclei were stained by treatment with DAPI (Nacalai Tesque). For in vivo RNA labeling, cells were incubated in the presence of 2 mM bromouridine at 37°C for 20 min and then subjected to immunostaining using mouse monoclonal anti-BrdU antibody (Roche Applied Science). Images were captured with an LSM510 confocal microscope (Carl Zeiss, Jena, Germany). For the evaluation of the colocalization, Pearson's correlation coefficient and van Steensel's cross-correlation coefficient (correlation coefficient and dx) were computed using ImageJ software with the JaCoP plug-in (42).

Isolation of HeLa Cell Lines Stably Expressing Elongin A siRNAs-HeLa cells were transfected with piGENE-neo plasmids (Clontech) encoding siRNA sequences for human Elongin
A under the control of the human tRNA Val gene promoter. After selection of cells with 800 g/ml neomycin for 14 days, each colony was assayed for Elongin A expression by Western blot. Sequences for siRNAs that target nucleotides ϩ1420 to 1439 (siElA6) and nucleotides 2216 to 2235 (siElA4) were 5Ј-AGgAAGGTGCtTGgTGTGTTCTTCCTGTCAAACACAT-CAGGCACCTTTCT-3Ј and 5Ј-GtAGCAtTGTTTCtTATGg-TCTTCCTGTCAATCATAGGAAACAGTGCTGC-3Ј, respectively. For rescue experiments, FLAG-tagged rat Elongin A was introduced into these cell lines by transfecting pcDNA3.1 vector encoding rat FLAG-Elongin A and selected by hygromycin. Elongin A knockdown and introduction of rescue constructs were also accomplished using retroviral vectors. Production of retrovirus vector using Plat E cells and pMX-puroII-U6 vector was described elsewhere (43).
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assays were performed as described (44) according to the protocol supplied by Upstate. Briefly, HeLa cells (4 ϫ 10 7 cells) were crosslinked with 1% formaldehyde for 10 min at room temperature. After washing twice with PBS, the cells were collected, lysed with 2 ml of SDS lysis buffer (50 mM Tris-HCl, pH 8.1, 10 mM EDTA, 1% SDS) containing a protease inhibitor mixture (Roche Applied Science), and sonicated to DNA lengths of ϳ250 -500 bp. After centrifugation at 13,000 rpm for 10 min, supernatants were diluted 10-fold in ChIP dilution buffer (16.7 mM Tris-HCl, pH 8.1, 167 mM NaCl, 0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA). Immunoprecipitations were then performed with the control IgG or the indicated antibodies, and the chromatin immunoprecipitated DNA was assessed by quantitative PCR (qPCR) using primer sets (Ref. 45 and Table 1). Fluorescence intensities were calculated by (Immunoprecipitate average Ϫ IgG average)/(Input average). Averaged results are presented as mean Ϯ S.E.
Luciferase Assay-pLuc-ATF3 (Ϫ384) containing the human ATF3 gene promoter was as described (46,47). HeLa cells (5 ϫ 10 4 cells), either wild type or siElA6 cells, were co-transfected with pLuc-ATF3 (Ϫ384) and 1 g each of expression vectors encoding rat Elongin A deletion mutants as well as sea pansy control vector. Cells were treated with 1 M doxorubicin and incubated for 8 h. Cells were then harvested, and their extracts were assayed for luciferase activity as described (40). Both firefly and Renilla luciferase activities were measured using the Dual-Luciferase Reporter Assay System according to the manufacturer's protocol (Promega). pRL-TK (Toyo Ink, Tokyo, Japan) containing the sea pansy luciferase gene was used as an internal control of transfection and expression. Averaged results are presented as mean Ϯ S.E.
Microarray Analysis-siElA or control (siGFP) HeLa cells were treated with 1 M doxorubicin for 2 h. Total RNA was extracted using an RNeasy kit (Qiagen) according to the manufacturer's instructions. RNA was transcribed to Cy3-labeled cRNA using a Low Input Quick Amp Labeling kit (Agilent). Hybridization, processing, imaging, and analysis were performed according to the Agilent protocols using an Agilent DNA Microarray Scanner and GeneSpring GX software (Platform GPL4133). Data were deposited in the Gene Expression Omnibus (GEO; www.ncbi.nlm.nih.gov/geo/) under accession number GSE30432.

RESULTS
Elongin A Associates with RNAPII-Results of a previous study suggested that the SII-like N-terminal region of Elongin A when present at high concentration as the bait protein during protein affinity chromatography can bind a complex or complexes containing RNAPII and general initiation factors (48). Furthermore, TFIIS has been reported to function not only as a transcription elongation factor but also as a component of RNAPII preinitiation complexes (49,50). A domain similar to the Elongin A SII-like region is also found in the human Mediator subunit MED26 and was recently reported to play a role as a docking site for both TFIID and P-TEFb-and ELL/EAF-containing super elongation complexes (51). We therefore began these studies by investigating the in vivo association between the transcription apparatus and Elongin A expressed at physiologically relevant levels. To do so, we established HeLa cells stably expressing FLAG-tagged human Elongin A. As shown in Fig. 1A, FLAG-tagged Elongin A is expressed in these cells at FIGURE 1. Elongin A complexes immunopurified under different salt conditions. A, nuclear extracts were prepared from HeLa cells stably transfected with pCI-neo-FLAG-hElongin A (F-Elongin A) or with empty pCI-neo (mock) and subjected to Western blotting using anti-FLAG, anti-Elongin A, or anti-RAP74 antibody. Note that dual bands (arrowheads) in FLAG-hElongin A cell extract in the Elongin A blot indicate both endogenous and FLAG-tagged hElongin A. B, nuclear extracts from hElongin A-expressing cells were subjected to immunoprecipitation (IP) using the control IgG or anti-FLAG antibody, and the resultant immunocomplexes were assayed for the Elongin trimer using anti-FLAG, anti-Elongin A, anti-Elongin B, or anti-Elongin C antibody. C, nuclear extracts from HeLa cells stably expressing FLAG-hElongin A, FLAG-RAP74, or the control mock vector were immunoprecipitated by anti-FLAG antibody or control IgG under low (80 mM) or high (150 mM) salt conditions. Resulting immunocomplexes containing Elongin A or RAP74 were assayed by Western blotting using anti-Rpb1, -TBP, -TFIIB, -TFIIE␣, and -FCP1 antibodies. D, immunocomplexes in C were also analyzed by Western blot using monoclonal H14 or H5 antibody recognizing the RNAPII CTD phosphorylated on Ser 5 or Ser 2 , respectively. levels similar to the endogenous protein. Nuclear extracts from this cell line and from control cells not expressing a FLAGtagged protein were subjected to immunoprecipitation with anti-FLAG antibody, and immunoprecipitated proteins were detected by Western blotting. As expected, FLAG-Elongin A bound Elongins B and C in vivo, suggesting that FLAG-tagged Elongin A is functional in this cell line (Fig. 1B). To determine whether Elongin A bound RNAPII, general initiation factors, or other transcriptional regulatory proteins, we performed additional immunoprecipitations at two salt conditions (80 and 150 mM). As a control, proteins associated with the general initiation and elongation factor TFIIF were studied in parallel using cells stably expressing FLAG-tagged TFIIF subunit RAP74. As shown in Fig. 1C, RAP74 copurified with RNAPII and initiation factors TBP, TFIIB, and TFIIE and with the CTD phosphatase FCP1. These interactions were stable in both low and high salt conditions. Elongin A could not be detected in the TFIIF complex at these conditions. On the other hand, an immunoprecipitation of FLAG-Elongin A revealed that Elongin A bound RNAPII but not the other initiation factors examined (Fig. 1C). We next addressed whether these complexes had different binding affinity to RNAPIIO. Fig. 1D shows that the amount of RNAPIIO phosphorylated at Ser 5 or Ser 2 in the TFIIF complex was significantly reduced in the high salt condition. In contrast, RNAPIIO (Ser 5 or Ser 2 ) in the Elongin A complex was more stable under the same salt condition, indicating that Elongin A has stronger affinity to bind RNAPIIO. Thus, our data suggest that Elongin A forms relatively stable complexes with phosphorylated RNAPIIO; however, we could not detect the interaction of Elongin A with components of the PIC when it was expressed at physiologically relevant levels in vivo.
Colocalization of Elongin A and RNAPIIO during Active Transcription-The above data are most consistent with the idea that Elongin A associates not with components of the PIC but rather with actively transcribing RNAPIIO. In cells undergoing active transcription, RNAPII is stained broadly throughout the nucleoplasm in a meshwork pattern (52,53). To compare the localization of Elongin A and RNAPII in vivo, we performed an immunocytochemical study. We found the majority of Elongin A staining to be localized in the nucleoplasm in a meshwork pattern that exhibits partial overlap with those of Ser 5 -and Ser 2 -phosphorylated RNAPIIO ( Fig. 2A, upper and lower panels). The degree of overlap between Elongin A and Ser 5 -phosphorylated RNAPII appeared to be greater than the overlap between Elongin A and Ser 2 -phosphorylated RNAPII (Pearson's correlation coefficient, 0.439 versus 0.095). Notably, Ser 5 phosphorylation occurs concomitantly with transcription initiation and continues during the early elongation stage. Thus, Elongin A may associate more stably with Ser 5 -phosphorylated RNAPII and play a role in the release from transient pausing rather than in the early elongation stage. Next, we cultured cells in the presence of DRB, an inhibitor of P-TEFb kinase and transcriptional elongation. As in Fig. 2A (lower panel), RNAPIIO (Ser 2 ) relocated to dotlike domains, consistent with the previous report that it localizes to speckles in the nucleus not actively involved in the transcription (54). Remarkably, Elongin A also relocated to speckle-like domains distinct from those associated with RNAPIIO (Ser 2 ). Note that cross-correlation analysis indicated a mutually exclusive pattern of RANPIIO (Ser 2 ) and Elongin A signals in the presence of DRB (42). RAP74 also exhibited partial colocalization with RNAPIIO, but its distribution did not appear to be significantly altered in DRB-treated cells perhaps because TFIIF appears to function predominantly during and shortly after initiation at steps that are not affected by DRB (55-57) (Fig. 2B).
Mammalian Elongin A Partly Colocalizes with Nascent RNA, and Global RNA Synthesis Is Not Impaired in Elongin A-null MEFs-To further explore the role of Elongin A in transcription in cells, we compared the localization of FLAG-tagged Elongin A in HeLa cells or endogenous Elongin A in MEFs with that of newly synthesized RNA using an in vivo RNA labeling technique in which nascent transcripts are labeled with bromouridine (53). As shown in Fig. 3, A and B (lower panel), Elongin A colocalized in both cell types with bromouridine incorporated into newly transcribed RNA at some but not all loci, consistent with the idea that Elongin A contributes to transcription of only a subset of genes. Bromouridine staining was abolished after treatment by RNase or 0.1 M ␣-amanitin (data not shown), indicating that most of the detectable signals were indeed derived from nascent RNA. To directly test the contribution of Elongin A to global transcription, Elongin A-null MEFs were immunostained with antibodies against Elongin A and RNAPIIO (phospho-Ser 2 ) (Fig. 3B, lower left panel). As expected, there was no apparent difference in the level or distribution of RNAPIIO in wild type or Elongin A-null MEFs. Moreover, the level of bromouridine incorporated into nascent RNA in Elongin A-null cells appeared to be similar to that in wild type cells (Fig. 3B, lower right panel). Taken together, these data support the idea that mammalian Elongin A is involved in transcription in vivo but is not essential for transcription of all genes.
Mammalian Elongin A Promotes Efficient Induction of Stress Genes-Our previous study showed that mammalian Elongin A deficiency causes sensitization of cells to stress stimuli (29), and Gerber et al. (35) demonstrated that Elongin A is essential for HSP70 gene induction in Drosophila. Thus, we examined whether mammalian Elongin A is also involved in the heat shock response using HeLa cells in which expression of Elongin A is stably down-regulated by either of two distinct shRNAs, siElA4 or siElA6 (Fig. 4A, upper panel). Induction of HSP70 protein after heat shock was significantly reduced in these two cell lines (Fig. 4B). To confirm that this reduction was not due to off-target effects of the shRNAs, rat Elongin A, the nucleotide sequence of which differs in six of 20 target nucleotides from human Elongin A, was introduced into siElA6-expressing HeLa cells. In this cell line, rat Elongin A was well expressed (Fig. 4A, lower panel) and restored the heat shock response (Fig. 4B), confirming that HSP70 gene activation depends on Elongin A. In addition, consistent with previous evidence that Elongin A is required for DNA damage-induced polyubiquitination of Rpb1 (26,27), we observed that depletion of Elongin A led to a substantial decrease in Rpb1 polyubiquitination in cells treated with the DNA-damaging agent doxorubicin (Fig. 4C and Ref. 26).
To identify other stress-induced genes whose expression depends on Elongin A, we performed genome-wide microarray analysis of doxorubicin-treated siElA cells (GEO accession number GSE30432). A total of 431 genes were up-regulated, and 1061 genes were down-regulated. Among the down-regulated genes (Table 2), ATF3, a member of the ATF/cAMP response element family transcription factors, is a stress response gene whose expression is maintained at a low or undetectable level but is rapidly induced in response to several stimuli at the transcriptional level (46,58). Because ATF3 is a target of the tumor suppressor p53 (47), induction by doxorubicin of another p53-dependent, stress-inducible gene, p21, was analyzed as well. As expected, induction of both p21 and ATF3 was significantly reduced at both the mRNA (Fig. 4D) and protein FIGURE 4. Mammalian Elongin A is required for the induction of several stress response genes. A, whole cell extract of Elongin A, siElA4, or siElA6 was subjected to Western blotting using anti-Elongin A antibody (upper panel). In the lower panel, an expression vector encoding FLAG-tagged rat Elongin A (F-rat Elongin A) was reintroduced into the siElA4 or siElA6 cells, and cells stably expressing rat Elongin A were selected by hygromycin. Cell extracts were analyzed by Western blotting using anti-Elongin A or anti-FLAG antibody. B, wild type, siElA6, siElA4, or siElA6 cells with FLAG-rat Elongin A were subjected to heat shock at 43°C for 30 min and returned to 37°C. After the indicated times of incubation, whole cell extracts were prepared and subjected to Western blot analysis. Relative intensities of each band are indicated. C, HeLa cells stably expressing siElA6 were treated with MG132 for 1 h followed by 1 M doxorubicin stimulation. Cells were harvested at the indicated time points, and polyubiquitination (poly Ub) of Rpb1 was assessed by Western blot with H14 antibody. The amount of ␤-actin is shown as a loading control. D, Elongin A-knocked down cells were treated with 1 M doxorubicin, and total RNA was extracted after 2-h stimulation. levels (Fig. 4E) in HeLa cells transduced with retrovirus encoding siElA6. To determine whether Elongin A is generally required for ATF3 induction or whether it functions specifically in the response to DNA damage, various reagents were used to stimulate HeLa cells, and the induction of ATF3 was examined. As shown in Fig. 4F, the induction of ATF3 protein in response to oxidative (H 2 O 2 ), DNA damage (doxorubicin), or endoplasmic reticulum stress stimuli (tunicamycin and thapsigargin) was significantly reduced in Elongin A-knocked down cells stably expressing siElA6, indicating that ATF3 induction is Elongin A-dependent. This repression of ATF3 expression was restored by reintroduction of rat Elongin A (Fig. 4F) as observed for HSP70 gene expression (Fig. 4B). Collectively, these data demonstrate that Elongin A plays a significant role in promoting fully efficient induction of ATF3 in response to multiple stresses.

RNAPII-transcribed ATF3 Gene Is Associated with Elongin A-
To determine whether Elongin A affects ATF3 transcription directly or via an indirect mechanism, we used ChIP combined with qPCR (Fig. 5A) to assess the occupancy of Elongin A, RNAPII, and other transcription regulators at the ATF3 gene before and after doxorubicin treatment. As shown in Fig. 5B  (upper left panel), Elongin A ChIP signals were increased across the entire transcribed region of the ATF3 gene after stimulation but remained at near background levels upstream of the ATF3 gene promoter (Ϫ5 kb), indicating stress-induced recruitment of Elongin A onto the ATF3 gene. We also observed an increase in the association of total RNAPII as well as CTD Ser 2 -phosphorylated RNAPII across the ATF3 gene after stimulation. It should be noted here that some poised RNAPIIs appeared to be present at the P1 promoter before treatment, consistent with the recent report by Lin et al. (59). Moreover, Leo1 and RAP74 were also recruited to the ATF3 gene along with Elongin A; however, their ChIP signals were different from one another, implying that they may play different roles in ATF3 gene expression. Gerber et al. (35) reported previously that Elongin A and RNAPII colocalize at heat shock puffs on Drosophila polytene chromosomes; however, they did not test recruitment of Elongin A to heat shock genes by ChIP. As shown in Fig. 5D, following heat shock, we observed a substantial increase in ChIP of Elongin A and CTD Ser 2 -phosphorylated RNAPII at the transcribed region of the human HSP70 gene and increased RAP74 near the promoter. These observations support a direct role of Elongin A in regulating heat shock-induced transcription of HSP70. We further assessed the recruitment of Elongin A to a collection of stress-induced and constitutively active genes, all of which have been reported to be associated with promoter-proximally paused RNAPII (60,61). The p21, c-myc, and c-jun genes are all activated by doxorubicin in an Elongindependent manner ( Table 2 and Fig. 4, D and E). As shown in Fig. 6, Elongin A occupancy was increased on the p21, c-myc, and c-jun genes after doxorubicin treatment. Although it is well established that promoter-proximal pausing is a key step in regulation of these three genes (60), the increase in ChIP signal was much more pronounced in more 3Ј-regions of these genes than at their promoters. Substantially lower levels of Elongin A ChIP signal were detected at the highly expressed, Elongin A-independent RPL29 and GAPDH genes again with a 3Ј bias; however, Elongin A ChIP on these genes was not increased by doxorubicin. Taken together, these data imply (i) that Elongin A contributes preferentially to the expression of stress response genes and (ii) that it associates broadly with the transcribed region of genes with higher recruitment to 3Ј-regions.

Both the C-terminal and SII Similarity Regions of Mammalian Elongin A Are Required for the Proper Response of the ATF3
Gene Promoter-Because Elongin A appeared to play a critical role in the rapid induction of ATF3, we sought to identify the Elongin A functional domain(s) required for induction of the ATF3 gene in vivo. Colgan and Manley (62) previously used luciferase reporter assays to evaluate the function of TFIID, one of the general transcription factors. Thus, we used this assay to assess the transcriptional activity of several mutants of Elongin A using pLuc-ATF3 (Ϫ384) in which ATF3 promoter sequences from Ϫ384 to ϩ34 drive expression of a luciferase reporter gene (47). We first confirmed that (i) doxorubicin-dependent induction of this reporter gene was impaired in siElA6 cells and that (ii) it could be rescued by reintroduction of wild type rat Elongin A (Fig. 7C, empty and 1-773). Thus, ATF3-luciferase reporter gene behaves similarly to the endogenous ATF3 gene and provides a good model with which to study Elongin A-dependent transcription.
As described earlier, Elongin A has an N-terminal SII similarity region that is dispensable for Elongin A transcriptional activity in vitro and a C-terminal elongation stimulatory domain that is required for Elongin A transcriptional activity in vitro. The latter C-terminal domain includes a SOCS box composed of a BC box (Elongin BC binding site) and a Cul5 box (required for Cul5 binding) (Figs. 7A and 8A). We explored whether these domains have a functional role in the stress-

Genome-wide RNA expression analysis comparing control and Elongin A knockdown HeLa cells under doxorubicin treatment
siElA or control (siGFP) HeLa cells were treated with 1 M doxorubicin for 2 h. Microarray analysis was carried out as described under "Experimental Procedures." Representative genes that are up-regulated and down-regulated by Elongin A depletion are shown. Indicated values are the relative expression levels in siEIA cells compared with the control. ATF3 expression was decreased to 0.555 in siElA cells. Data were deposited in the Gene Expression Omnibus (GEO; (www.ncbi.nlm.nih. gov/geo/) under accession number GSE30432.

Up-regulated genes
Glucosidase induced activation of the ATF3 gene promoter. As shown in Fig.  7C, mutants 1-439 and 1-120 NLS (nuclear localization signal), which lack the C-terminal elongation stimulatory domain, did not support doxorubicin-induced activity of the ATF3 promoter, indicating that the C-terminal elongation stimulatory domain is essential for in vivo transcription activity. The Elongin A mutant lacking the BC domain (⌬BC) was also inactive for the induction of the ATF3 reporter. By contrast, both mutants 400 -773 and 120 -773, which contain the C-terminal elongation stimulatory domain but lack the SII similarity region, only weakly stimulated reporter activity. These mutants behaved very similarly when assayed for their ability to stimu- late an HSP70 promoter after heat shock (data not shown).
These data indicate that the C-terminal elongation stimulatory domain including the Elongin BC interaction region is essential for efficient transcription in vivo, whereas the N-terminal SII similarity region is not required but has a positive regulatory role.
The E3 Ligase Activity of Elongin A Is Dispensable for Sufficient Inductions of ATF3-From the results of the reporter assay, the BC box is likely to be essential for the sufficient induction of ATF3. The BC box, which provides a platform for binding of Elongins B and C, has been shown to be essential for the ability of Elongin A to stimulate transcriptional elongation in vitro (25) and to support its E3 ubiquitin ligase activity (26). On the other hand, an intact Cul5 box, which is adjacent to the BC box (Fig. 8A), is reported to be required for E3 ligase activity but to be dispensable for stimulation of transcript elongation. To further define regions of Elongin A needed for rapid activation of the ATF3 gene, we reintroduced several mutants of mouse or rat Elongin A into the siElA6 HeLa cells. In addition to those mutants, we introduced mouse Elongin A deletion mutant 1-674, which contains a complete SOCS box but lacks more C-terminal sequences. This mutant is active in RNAPII ubiquitination but does not stimulate transcript elongation in vitro (19). Each mutant was expressed in siElA6 cells (Fig. 8B), and the induction of the endogenous ATF3 gene was analyzed after doxorubicin treatment by Western blotting using anti-ATF3 antibody. As shown in Fig. 8C, full-length Elongin A(1-773) and the Cul5 box mutant (V571A/P572A) partly but significantly rescued the induction of ATF3. By contrast, BC box mutant (L550P/C554F) and Elongin A(1-674) did not promote ATF3 induction. Similar results were obtained when the abilities of wild type and mutant Elongin A to support ATF3 promoter activity were assessed using the luciferase reporter assay (Fig. 8D). As expected, Elongin A(1-674)-reintroduced cells exhibit the same degree of ubiquitination of Rpb1 after doxorubicin treatment as full-length Elongin A-reintroduced cells. On the other hand, Cul5 box mutant (V571A/P572A) and BC box mutant (L550P/C554F) exhibit only background level of ubiquitination activity, which likely is due to the presence of residual Elongin A or another E3 ubiquitin ligase that targets Rpb1 (Fig. 8E). Taken together, these experiments argue that proper activation of ATF3 in vivo depends on the presence of an intact Elongin A elongation stimulatory domain that can bind Elongins B and C but that the E3 ubiquitin ligase activity of Elongin A is dispensable.

DISCUSSION
The results presented in this study indicate that mammalian Elongin A is essential for rapid and full induction of the immediate early response gene ATF3 and of the heat shock-induced HSP70 gene. Previous studies suggested that Elongin A is not generally required for transcription in vivo (19,28). Supporting this idea, shRNA-mediated depletion of Elongin A or deletion of the Elongin A gene neither altered the distribution of RNAPII in cell nuclei nor impaired production of the majority of nascent RNA (Fig. 3B), indicating that Elongin A is not required for transcription of all genes but rather regulates expression of select genes. With regard to the in vivo function of Elongin A, Gerber et al. (35) reported that Drosophila Elongin A plays a key role for HSP70 gene induction, whereas Yasukawa et al. (19) very recently demonstrated that Elongin A is needed for retinoic acid-induced expression of several genes during neuronal differentiation. In neither case, however, did the authors demonstrate by ChIP that these genes are direct, rather than indirect, targets for regulation by Elongin A. In this study, for the first time, we have provided evidence that Elongin A directly binds to the coding regions of target genes ATF3 and HSP70 along with elongating RNAPII, consistent with the model that it directly regulates their transcription during elongation. Moreover, we identified domains of Elongin A essential for fully efficient activation of stress-induced gene expression in vivo.
Characteristic Feature of Elongin A in Transcriptional Event in Mammalian Cells-Upon doxorubicin treatment, Elongin A becomes physically associated with the entire ATF3 gene (Fig.  5). Moreover, Elongin A clearly binds to RNAPIIO phosphorylated at Ser 5 or Ser 2 in vivo (Fig. 1). Importantly, this binding does not appear to be due to direct interactions with the CTD of RNAPII because we were unable to detect appreciable in vitro binding of Elongin A to a GST-CTD fusion protein that was either unphosphorylated or phosphorylated by MAPK (data not shown). Accordingly, it seems most likely that Elongin A binds to RNAPIIO through sequences in its C-terminal elongation activation domain between residues 590 and 690 ( Fig. 1C and Ref. 23).
It is intriguing to note that Elongin A relocated to large speckle-like domains in the presence of DRB ( Fig. 2A). Although RNAPII is also known to alter its nuclear localization in transcriptionally inactive cells (52,53), Elongin A and RNAPII relocalized to distinct domains. Because DRB is a well known inhibitor of protein kinases such as casein kinase II (61) and P-TEFb (62) and strongly inhibits transcriptional elongation, relocation following DRB treatment may suggest that phosphorylation of Elongin A controls its activity during transcription elongation or that decreased CTD phosphorylation interferes with the ability of Elongin A to bind RNAPII. Alternatively, Elongin A could have a unique unknown activity when transcription is repressed.
Our reporter assays using various Elongin A mutants demonstrated that the C-terminal elongation activation domain including the BC box is crucial for the response of ATF3 gene, consistent with the activity in vitro (22,25,63). In addition, the N-terminal SII similarity region has a mild but positive regulatory role for the Elongin A-driven transcription activity in vivo (Fig. 7C). This region is well conserved among human, rodents, and Drosophila (64) and shares sequence similarity with the N-terminal domains of SII and Mediator subunit MED26 (65). The N-terminal region of yeast SII has been reported to interact with Med13 and Spt8, which belong to the Cdk8 module of the Mediator and to a subform of the SAGA (Spt-Ada-Gcn5acetyltransferase) co-activator, respectively (66). Thus, this region of Elongin A may also interact with Mediator(s) or coactivator(s) in the PIC or poised RNAPII complex, thereby helping to determine the Elongin A target genes. Although we did not detect PIC components in our immunoprecipitations (Fig. 1C), our results do not exclude a role for Elongin A in the PIC; indeed, although low, the Elongin A ChIP signal around the transcription start site is detectable (Fig. 5).
These analyses altogether support a view that when cells are subjected to stress Elongin A, through interaction with Ser 5and/or Ser 2 -phosphorylated RNAPII, is recruited to elongation complexes where it can enhance transcription throughout the entire coding region (Figs. 1 and 5). On the other hand, our immunocytochemical assay indicated that Elongin A colocalizes with RNAPII (Ser 5 ) more than RNAPII (Ser 2 ) ( Fig. 2A). Considering that Elongin A is more strongly associated with more 3Ј-regions than with the promoters of all the genes we tested (Figs. 5 and 6), the elongation activity of Elongin A seems to be exerted mainly after the release of promoter-proximal pausing and thus differs from the super elongation complex (51). Further biochemical analysis of Elongin A after the tran- sition from the paused to elongating RNAPIIO will be required to characterize the active elongation complex in vivo.
Elongin A Has Two Properties in Vivo: Elongation and Ubiquitination-Elongin A has at least two biochemical activities. (i) It stimulates transcriptional elongation by RNAPII, and (ii) it is a substrate recognition subunit of a ubiquitin ligase that supports RNAPIIO ubiquitination. Although several roles of ubiquitination have been reported, including proteolysis, regulation of protein activity, and regulation of protein localization (67), Elongin A-dependent Rpb1 ubiquitination promotes degradation of Rpb1. Moreover, assembly of an active Elongin A ubiquitin ligase that supports this degradation requires Elongin A to bind both Elongin BC and Cul5 (26). Our rescue experiments provide strong evidence that Elongin BC binding is required, but Cul5 binding is dispensable for the activation of ATF3 transcription (Fig. 8). Thus, during rapid induction of stress genes, the elongation activity of Elongin A is totally independent of its ubiquitination activity in vivo as is the case for retinoic acid induction of several genes for neuronal development in ES cells (19). Therefore, the elongation and ubiquitination activity might be more generally independent functions.
Interestingly, these two activities are apparently opposing forces in terms of the regulation of the transcript amount. Polyubiquitination is the early step to degrade the RNAPII FIGURE 8. The C terminus of mammalian Elongin A is essential, but the Cul5 box is dispensable for ATF3 induction. A, amino acid sequence of SOCS box in human Elongin A from 546 to 590. Locations of BC box and Cul5 box are depicted. Arrows indicate the positions of amino acid substitution to abolish each functional domain. B, structures and expressions of several variants of Elongin A are shown. Elongin A full length and L550P/C554F were derived from rat Elongin A. Elongin A V571A/P572A and 1-674 were derived from mouse Elongin A. Each construct was reintroduced into the Elongin A-knocked down cells with retrovirus vectors. The levels of each variant expression were assessed by Western blot using anti-FLAG antibody. The amount of ␤-actin is shown as a loading control. The asterisks indicate the positions of amino acid substitution detailed in A. C, inductions of ATF3 by doxorubicin (Dox) were analyzed by Western blot. Elongin A mutants shown in B were reintroduced into Elongin A-knocked down HeLa cells. Cells were harvested at 3 h following treatment with 1 M doxorubicin. As a control, HeLa cells stably expressing siGFP with empty vector reintroduction were also investigated. D, Elongin A mutants shown in C were transfected with ATF3 reporter into HeLa cells expressing siElA6, and then cells were treated with doxorubicin. Reporter activity was measured as in Fig.  6C. Data represent the means of three independent experiments with error bars representing S.E. E, cells as in B were treated with MG132 followed by 1 M doxorubicin stimulation. After 8 h, polyubiquitination (poly Ub) of Rpb1 was assessed by Western blot as in Fig. 4C. itself. By contrast, elongation is the activity that assists the primary role of RNAPII. One of the possible explanations for this issue is that ubiquitination and degradation of paused or stalled RNAPII free the DNA template of RNAPII that would otherwise block passage of the next polymerase, resulting in an overall increase of the amount of fully elongated transcript. In this regard, however, our new data do not completely fit this model because the ubiquitination activity of Elongin A is unnecessary for the rapid induction of ATF3. In parallel with this, E3 ligase activity of Elongin A is not essential for the neurogenesis-related gene activation upon retinoic acid stimulation in mouse embryonic stem cells (19). We predict that this might be due to the RNAPII pausing feature. In the case of physiological pausing during mid to late elongation phase, Elongin A might facilitate the RNAPII elongation without ubiquitination activity. However, when RNAPII is arrested for a prolonged time because of profound DNA damage, the ubiquitination activity of Elongin A might be indispensable by cooperating with Cul5 to modify or degrade arrested RNAPIIO to ensure efficient gene expression (68).
Finally, rapid and sufficient induction of stress response genes is unambiguously an essential process for the survival of living cells. The role of Elongin A in activation of stress response genes is therefore likely to be of extraordinary importance.