Elongin C contributes to RNA polymerase II degradation by the interferon antagonist NSs of La Crosse orthobunyavirus.

Mosquito-borne La Crosse virus (LACV; genus Orthobunyavirus, family Peribunyaviridae, order Bunyavirales) causes up to 100 annual cases of severe meningoencephalitis in children and young adults in the United States. A major virulence factor of LACV is the non-structural protein NSs which inhibits host cell mRNA synthesis to prevent the induction of antiviral type I interferons (IFN-alpha/beta). To achieve this host transcriptional shutoff, LACV NSs drives the proteasomal degradation of RPB1, the large subunit of mammalian RNA polymerase II. Here, we show that NSs acts in a surprisingly rapid manner, as RPB1 degradation was commencing already at 1 hour post infection. The RPB1 degradation was partially dependent on the cellular E3 ubiquitin ligase subunit Elongin C. Consequently, removal of Elongin C, but also of the subunits Elongin A or B by siRNA transfection partially rescued general RNAP II transcription and IFN-beta mRNA synthesis from the blockade by NSs. In line with these results, LACV NSs was found to trigger the redistribution of Elongin C out of nucleolar speckles, which however is an epiphenomenon rather than part of the NSs mechanism. Our study also shows that the molecular phenotype of LACV NSs is different from RNA polymerase II inhibitors like alpha-amanitin or Rift Valley fever virus NSs, indicating that LACV is unique in involving the Elongin complex to shut off host transcription and IFN response.SIGNIFICANCE The mosquito-borne La Crosse virus (LACV; genus Orthobunyavirus, family Peribunyaviridae, order Bunyavirales) is prevalent in the United States and can cause severe childhood meningoencephalitis. Its main virulence factor, the non-structural protein NSs, is a strong inhibitor of the antiviral type I interferon (IFN) system. NSs acts by imposing a global host mRNA synthesis shutoff, mediated by NSs-driven proteasomal degradation of the RPB1 subunit of RNA polymerase II. Here, we show that RPB1 degradation commences as early as 1 hour post infection, and identify the E3 ubiquitin ligase subunit Elongin C (and its binding partners Elongin A and B) as an NSs cofactor involved in RPB1 degradation and in suppression of global as well as IFN-related mRNA synthesis.

more than half a million people in Latin America (5,6). Also, Maguari-like viruses, associated with febrile illness, are infecting humans all over South America (7). Members of the Maputta serogroup are responsible for epidemics of an acute polyarthritislike disease in Papua-New Guinea and Australia (8). La Crosse virus (LACV) is the causative agent of a severe meningoencephalitis that mostly (but not exclusively) affects children and young adults in the United States (9)(10)(11)(12). Per year, up to 100 cases have required hospitalization or even intensive care, exceeding West Nile virus in numbers of pediatric neuroinvasive arboviral infection (13). A substantial proportion of patients are suffering from long-lasting neurological problems (11). Since most infections, especially in adults, are, however, mild or inapparent, the number of subclinical infections was estimated to be around 300,000 annually (14).
The group of orthobunyaviruses is taxonomically defined as a genus within the family Peribunyaviridae, order Bunyavirales (15). The pleomorphic virions are enveloped and have a diameter of approximately 100 nm. As is typical for bunyaviruses (16), their genome consists of three segments of negative-strand RNA that are named L (large; ca. 7,000 nucleotides [nt]), M (medium; ca. 4,500 nt), and S (small; ca 950 nt). The L segment encodes the RNA-dependent RNA polymerase (RdRP), the M segments encodes a polyprotein that is processed to the envelope glycoproteins Gn, NSm (nonstructural, M segment), and Gc, and the S segments encodes the nucleocapsid protein N and the nonstructural protein NSs. All genomic segments are encapsidated by N protein and contain noncoding regions at their 5= and 3= ends that have the potential to anneal to a so-called "panhandle structure" due to partial sequence complementarities. The panhandle sequences are bound by the L RdRP and constitute the promoter for viral mRNA transcription and genome replication (17).
The entire multiplication cycle of bunyaviruses takes place in the cytoplasm. After entering the host cell via clathrin-mediated endocytosis (18,19) and subsequent low pH-driven membrane fusion, mRNAs are transcribed from the incoming genome RNA nucleocapsids by L RdRP ("primary transcription"). The transcription is primed by 12-to 18-nt 5=-capped oligonucleotides that had been cleaved from host mRNA by an endonuclease activity residing in the N terminus of L (20). After translation of the viral proteins, the viral genome RNA (vRNA) is replicated via a positive-sense, encapsidated full-length intermediate, the copy RNA (cRNA). The newly generated vRNA nucleocapsids can give rise to more mRNAs produced by secondary transcription or become packaged by peptidase-processed Gn/Gc on Golgi membranes and leave the cell via the exocytosis pathway (19,21,22).
For orthobunyaviruses, the S segment-encoded protein NSs is a major determinant of pathogenicity, acting as an antagonist of the antiviral type I interferon (IFN) response (19,23). IFNs are cytokines that become produced upon virus detection by host cells and stimulate the expression of genes (ISGs) for proteins with antiviral activity (24,25). In the case of LACV, infection is detected by the IFN system via the RIG-I/MAVS virus sensor axis (26)(27)(28). RIG-I, a cytoplasmic RNA helicase (29,30), is capable of recognizing the panhandle RNA of bunyaviruses, even if packaged by nucleocapsids (31,32). RIG-I-mediated panhandle detection activates the transcription factor IRF-3, leading to the production of IFN-␤ mRNA (33,34). It is known that orthobunyavirus multiplication is affected by IFN (35)(36)(37)(38), and several ISGs were shown to be involved in this antiviral activity (36,(39)(40)(41)(42). To counteract the IFN/ISG induction, the orthobunyavirus NSs, however, massively and rapidly inhibits cellular mRNA transcription, leaving the host unable to appropriately respond to the infection (35,43). We have previously shown that the NSs of the orthobunyaviruses BUNV and LACV directly interfere with mRNA synthesis by the cellular RNA polymerase II (RNAP II) (28,44). While BUNV NSs is reducing the mRNA elongation-relevant phosphorylation of the C-terminal serine 2 residue (part of the 52 times repeated heptapeptide motif in the C-terminal domain [CTD]) (44), LACV NSs additionally degrades the large subunit (RPB1) of RNAP II (28). An RPB1-degradative activity was also described for the NSs of SBV (43,45). Interestingly, the effect of LACV NSs on RPB1 has strong similarities with parts of the DNA damage response (DDR), not only in terms of RPB1 degradation but also by activation of other DDR markers (28).
The impact of orthobunyavirus NSs on IFN induction and its biological relevance are well established (19). Mechanistically, however, far less is known. Here, we further investigated the effect that NSs has on the cells and on RNAP II, and we identified Elongin C and the Elongin complex as a host factor involved in NSs action.

RESULTS
LACV NSs rapidly reduces RNAP IIo. To become transcriptionally active, the 260-kDa large subunit RPB1 of RNAP II gets hyperphosphorylated at the 52 heptad repeat sequences that are situated at the CTD (46,47). This results in a gain of molecular weight and hence in a band shift on immunoblots. The NSs proteins of both the orthobunyavirus type species BUNV and of LACV trigger the disappearance of the high-molecular-weight hyperphosphorylated RNAP II, termed IIo, and with some delay also of the lighter nonphosphorylated, transcriptionally inactive form (IIa) (28,44). Moreover, the RPB1 CTD has two major phosphorylation sites within each of the 52 heptad repeats (consensus sequence YSPTSPS), serine 2 and serine 5. CTD-serine 5 phosphorylation is a hallmark of promoter-bound RNAP II, whereas CTD-serine 2 phosphorylation indicates transcriptional elongation (46). Both BUNV and LACV were shown to preferentially affect CTD-serine 2 phosphorylation, indicating specific inhibition of host mRNA elongation (28,44).
Previously, we demonstrated the degradation of the hyperphosphorylated RNAP IIo by NS-expressing wt LACV in a time course experiment that started at 5 h postinfection (p.i.) (28). Meanwhile, however, it has become clear that the host cell elicits IFN induction even earlier, upon detection of the nucleocapsid-borne double-stranded RNA panhandle by RIG-I immediately after virus entry (32). Moreover, reverse transcriptionquantitative PCR (RT-qPCR) analyses demonstrated that mRNA transcription by LACV is detectable as early as 1 h p.i. (data not shown). To address this immediate early step of infection, we conducted a time course experiment in human HuH-7 cells (infected at a multiplicity of infection [MOI] of 10) that covered 1 to 4 h p.i. Figure 1A shows that even at 1 h p.i. there is a reduced RPB1 IIo signal in wild-type (wt) LACV-infected cells, whereas infection with an NSs-deleted LACV (LACVdelNSs) had no such effect. These differences in RPB1 IIo levels are not due to potential differences in viral replication, as demonstrated by comparisons of the immunoblot signal quantifications for RPB1 IIo and LACV N at 3 and 4 h p.i. (Fig. 1B and C). Curiously, the phosphorylation states of either RPB1 CTD-serine 2 or 5 are not severely diminished at these time points (in contrast to longer infections [28,44]), probably due to short-term activation under infection, as seen for the delNSs virus. Interestingly, the application of ␣-amanitin, a pharmaceutical transcription inhibitor known to induce RPB1 degradation (48), showed a slightly reduced effect than wt LACV. ␣-Amanitin was given at the same time as the virus, but unlike NSs, which first has to be expressed, the full inhibitor dose is present right from the start. Similar results were obtained with the global inhibitor of DNAdependent RNA polymerases, actinomycin D (data not shown). The finding that LACV NSs is on a par with a chemical RNAP II inhibitor underscores the surprisingly fast and efficient destruction of RPB1 by LACV NSs.
Besides LACV and related orthobunyaviruses, also Rift Valley fever virus (RVFV; genus Phlebovirus, family Phenuiviridae, order Bunyavirales) encodes an NSs that strongly inhibits RNAP II activity (49,50). The two NSs proteins play the same biological role (IFN induction antagonism) and target the same cellular function (RNAP II mRNA transcription) but are unrelated with respect to size and amino acid sequence. Using recombinant RVFV encoding the NSs genes of either LACV or RVFV (both equipped with a Flag tag), we compared their RPB1-destructive activities, again in HuH-7 cells within the first 4 h of infection at an MOI of 10. As shown in Fig. 2, the NSs of RVFV is expressed from 1 h p.i. on, but exhibited no RNAP II-destructive activity. RVFV-expressed LACV NSs, in contrast, becomes detectable at 3 h p.i. and diminishes RNAP IIo from this time point on. Thus, taken together, our data show an astonishingly rapid attack of RNAP II by LACV NSs, with an efficiency comparable to a pharmaceutical transcription blocker. Moreover, the RNAP II inhibition profile is different from the other established bunyaviral RNAP II inhibitor, RVFV NSs.
The Elongin complex is involved in NSs-mediated RNAP II suppression and IFN antagonism. The phenotype of RNAP II degradation by LACV NSs resembles the one occurring after DNA damage (28). When RNAP II translocation on DNA is stalled by FIG 1 Rapid and specific RPB1 degradation by wt LACV. HuH-7 cells were infected with wt LACV or LACVdelNSs viruses (MOI of 10) or treated with ␣-amanitin (10 g/ml). (A) Immunoblot analyses for the various RPB1 states and for viral N at 1 to 4 h p.i., using the antibodies indicated on the left. Immunoblot signals of RPB1 IIo (B) and LACV N (C) were quantified and normalized to the tubulin signal. Means, SD, and individual data points from five independent experiments are shown. A nonparametric, one-tailed Wilcoxon's paired signed-rank test was applied to test for a potential difference in signals for RPB1 IIo and LACV N between wt LACV 3-h p.i. and the LACVdelNSs 3 and 4-h p.i. time points. *, P Ͻ 0.05; ns, nonsignificant. RT-qPCR analyses confirmed the similar replication rates and showed that comparable amounts of input virus were used in all cases (data not shown).
genotoxic damages or chemical inhibitors such as ␣-amanitin (46), the RPB1 subunit becomes ubiquitinated and proteasomally degraded (51). The Elongin E3 ubiquitin ligase complex, consisting of the subunits Elongin A, B, and C, was shown to be involved in degradative polyubiquitination of mammalian RPB1 (52-55). In undamaged cells, however, Elongin A/B/C promotes RNAP II transcription (hence its name) by decreasing transient RNAP II pausing (56). We investigated a possible involvement of the Elongin complex in NSs action. First of all, we knocked down mRNAs of the individual Elongins and monitored the effect under LACV infection. Human A459 cells were transfected with siRNA against Elongin A, B, or C; infected with wt LACV or LACVdelNSs (MOI of 10); and then lysed and immunoblotted 16 h later. Figure 3A demonstrates that the small interfering RNAs (siRNAs) were strongly reducing the individual protein levels and that Elongins B and C seem to stabilize each other, since the knockdown of either of these reduced levels of the other one. However, quantification of the immunoblot signals show that Elongin C is most reduced when knocked down by specific siRNA, whereas under conditions of Elongin B knockdown residual amounts of Elongin C remain (Fig. 3B). Regarding RNAP, the levels of the IIo fraction are still largely suppressed by LACV infection when Elongins were depleted by siRNAs (see Fig. 3A). However, quantification and statistics of the immunoblot signals revealed that RPB1 IIo levels partially recovered under conditions of Elongin C knockdown (Fig. 3C). Elongin A or B knockdown, in contrast, had no such effect.
We also measured another hallmark of NSs action on RNAP II, the suppression of ␥-actin intron RNA levels. Introns have a short half-life (57), and blockade of RNAP II elongation by LACV NSs dries out their supply within 4 h of infection (28). As shown in Fig. 3D, wt LACV infection suppressed ␥-actin intron RNA down to approximately 3% of mock levels. Interestingly, siRNA-mediated removals of either Elongin A, B, or C were all able to rescue ␥-actin intron RNA significantly and to a certain extent, but again the Elongin C knockdown reached the highest level. Compared to the control siRNA, Elongin C siRNA alleviated the wt LACV-triggered ␥-actin intron RNA reduction from 3 to 19% of the mock levels, i.e., by a factor of ϳ6.
The biologically relevant reason for LACV NSs to degrade transcribing RNAP II is suppression of IFN induction at the mRNA transcription level (35). When expression of the three Elongins in A549 cells was individually suppressed by siRNA transfection and IFN induction was measured, a similar picture as with the ␥-actin intron RNA reduction emerged. Suppression of Elongin C rescued IFN-␤ mRNA induction in wt LACV-infected cells from ϳ400-fold to Ͼ7,000-fold (compared to the approximately 70,000-fold measured for the delNSs virus), i.e., by more than 1 log 10 (Fig. 4A). siRNAs against Elongin A or B, in contrast, rescued IFN-␤ induction in wt LACV-infected cells at a lower, but still significant level. ISG56 (IFIT1) is another virus-induced gene that plays a role in the immediate early antiviral state (58). Also, ISG56 mRNA induction was rescued best in wt LACV-infected cells when Elongin C expression was impeded (Fig. 4B). Importantly, under these single-step growth conditions (MOI of 10, 16 h of infection) none of the Elongin knockdowns significantly influenced viral RNA levels, neither of wt LACV nor of LACVdelNSs (Fig. 4C). However, since some siRNAs did increase IFN and ISG56 mRNA levels in wt LACV-infected cells, we also tested the effect of the Elongin knockdowns in the IFN-competent A549 cells under multistep growth conditions (MOI of 0.01), at an extended incubation time (24 h) and with virus titrations. Indeed, with this more sensitive setting we observed the expected difference between wt LACV and the LACVdelNSs virus (35) and that the knockdown of Elongins A or C (but not B) reduced the growth of both wt LACV and the LACVdelNSs virus (Fig. 4D). Thus, again Elongin C turned out to be a factor exhibiting a clear impact on LACV.
Taken together, our siRNA experiments establish the Elongin A/B/C complex, and especially Elongin C, as a contributor to RPB1 degradation, RNAP II transcription shutoff, and IFN mRNA downregulation by LACV NSs.
LACV NSs relocalizes Elongin C. Using immunofluorescence analysis, we sought to observe the behavior of the Elongins in LACV-infected cells. To better distinguish between primary effects caused by NSs and possible secondary effects caused by the NSs-mediated RPB1 degradation and block in host mRNA transcription, we treated the HuH-7 cells in parallel with the RNAP II inhibitor ␣-amanitin. In the immunofluorescence experiments, Elongin A exhibited in uninfected cells a nuclear signal with the tendency to form speckles (Fig. 5A). The nuclear pattern did not change in infected cells (MOI of 1), independent of whether it was wt LACV or LACVdelNSs. The Elongin A speckles, however, diminished in signal strength and number when ␣-amanitin was applied. Elongin B produced a nuclear immunofluorescence signal that did not change by any of the treatments (Fig. 5B). Elongin C also exhibited a nuclear speckle pattern in uninfected cells and in cells infected with the LACVdelNSs virus (Fig. 6). In cells infected with wt LACV, however, the speckles were barely visible, whereas under ␣-amanitin treatment the speckles dissolved but the signal remained nuclear. Since on the other hand protein levels of all Elongins remained unchanged under infection (see Fig. 3), the diminished Elongin C signal in wt LACV-infected cells apparently derived from a relocalization and dilution throughout the cell.
Thus, infection with the NSs-expressing wt LACV converted the nuclear Elongin C speckles in a manner that was unique and clearly distinct from ␣-amanitin. The subcellular distribution of elongin A and B, in contrast, remained unchanged by infection, although ␣-amanitin had an influence on Elongin A distribution. From the different phenotypes of infection and ␣-amanitin we conclude that the redistribution of Elongin C is specific for LACV NSs and not simply a consequence of its ability to block RNAP II.
Elongin C relocalization is a secondary effect of NSs action. The subnuclear distribution pattern of Elongin C in uninfected cells is reminiscent of nucleoli, the sites of rRNA synthesis by RNAP I and ribosome biogenesis (59). Indeed, using nucleolin as marker, we detected a clear and 1:1 overlap with the Elongin C signal in HuH-7 cells (Fig. 7). Both nucleolin and Elongin C remained unchanged in LACVdelNSs-infected cells but became evenly distributed in the nucleus upon ␣-amanitin treatment. wt LACV-infected cells (MOI of 1) exhibited an intermediate phenotype of nucleolin distribution, with still distinguishable but weaker nucleolar speckles, whereas the Elongin C signal became undetectable, as shown above. These results are in line with our previous results showing that LACV NSs impedes neither RNAP I transcription nor nucleolar localization of RNAP I transcripts (28).
NSs seems to specifically disassemble the nucleolar Elongin C speckles without entirely destroying the structural integrity of the nucleoli. Since under NSs on one hand Elongin C becomes undetectable in immunofluorescence but, on the other hand, the levels on immunoblots remain unchanged, we investigated the possibility of a redistribution to the cytoplasm. Leptomycin B (LMB) is an inhibitor of CRM1/exportin 1, the major nuclear export factor for proteins. Using LMB, we could indeed trap Elongin C in the nucleoli of HuH-7 cells infected with wt LACV at an MOI of 1 (although the signal was somewhat weaker than in mock infection) (Fig. 8A). This confirms our assumption of an NSs-driven redistribution out of the nucleus, either by expulsion or by cytoplasmic retention. Nonetheless, immunoblots of HuH-7 cells infected with viruses at an MOI of 10 show that blocking nuclear export by LMB could not relieve the destruction of RPB1 by NSs, in contrast to the proteasomal inhibitor MG132 (Fig. 8B). Also, when ␣-amanitin was applied we found that LMB was unable to entirely rescue RPB1 from proteasomal destruction.
LMB had no impact on wt LACV titers (data not shown) or the N protein expression (see Fig. 8B). Thus, NSs removes Elongin C from the nucleus in a Crm1-dependent manner, but this mechanism seems a secondary phenomenon rather than an essential step in the attack of RNAP II. Altogether, our data show that the E3 ubiquitin ligase subunit Elongin C is involved in the NSs-mediated destruction of the RNAP II large subunit RPB1 and the suppression of IFN induction, and gets relocalized as a consequence.

DISCUSSION
Morphological changes of nucleoli were long known as a hallmark of orthobunyavirus infection (60). Moreover, recently a nucleolar localization sequence important for NSs action was described for the orthobunyavirus SBV (45). Our findings that Elongin C (i) colocalizes with nucleolin in uninfected cells, (ii) contributes to the NSs-mediated RNAP II degradation, and (iii) becomes redistributed in the presence of NSs are in line with those earlier observations. Thus, Elongin C appears to be a player in the anti-IFN strategy of LACV, a view further underscored by the increase in IFN induction and RNAP II activity byϾ log 10 step in Elongin C-deficient cells that were infected with wt LACV.
The cellular Elongin complex consisting of Elongin A, B, and C is able to foster both the elongation of mRNA transcription by RNAP II (56) and (the B/C complex) the degradation of RPB1 under conditions of stress (53)(54)(55). These activities are congruent with the action of orthobunyavirus NSs, which targets exactly these activities, as it inhibits mRNA elongation and drives RPB1 degradation (28,44). Moreover, the fact that mostly Elongin C participated in these NSs activities is in accord with our immunofluorescence data which show its relocalization by NSs but not that of Elongin A and B. Moreover, Elongin C was the only Elongin complex subunit that localizes to the nucleoli, which in turn are reorganized by orthobunyaviral NSs proteins (45,60).
Our attempts to robustly demonstrate a direct interaction between LACV NSs and Elongin C in transfected cells have failed (data not shown), in line with the absence of Elongin C in the LACV NSs interactome (61). Moreover, LACV NSs does not exhibit a discernible nucleolar localization but is faintly distributed all over the nucleus (28), and the nucleolar localization sequence detected for SBV NSs is not conserved in LACV NSs (45). Thus, the effect of NSs on Elongin C in the nucleoli is most likely transient or indirect. Also, the relocation of Elongin C might be a secondary effect and not causative, since nuclear export inhibition by LMB partially prevented Elongin C redistribution but not RPB1 degradation by NSs. We therefore hypothesize that LACV NSs recruits an Elongin C-dependent host cell pathway that leads to RPB1 degradation and inhibition of host mRNA transcription. The relocalization of Elongin C is a morphological consequence of this process (but NSs specific, since RNAP II inhibition by ␣-amanitin has a different phenotype), but it is not necessarily causative. Why Elongin A and B seem less important remains unclear. A slight but statistically significant rescue of RNAP II transcription activity was, however, measurable also in those knockdowns.
Despite the significant impact of the Elongin C knockdown, it was not sufficient to entirely rescue RNAP II and IFN induction from NSs-mediated destruction and blockade. This is comparable to the observation with RVFV NSs and its cellular cofactor, the E3 ubiquitin ligase subunit FBXO3. Similar to the situation with LACV NSs and Elongin C, IFN induction by wt RVFV is rescued by one log 10 step in cells depleted of FBXO3, but an NSs-deleted virus mutant still induced IFN-␤ by another log 10 step (62). RVFV NSs is known to counteract IFN induction by several mechanisms, namely, (i) sequestration of subunits p44 and XPB of the general RNAP II transcription factor TFIIH, (ii) destruction of the TFIIH subunit p62 (mediated by mentioned FBXO3), and (iii) recruitment of the IFN promoter suppressor SAP30 (49,50). It is therefore conceivable that LACV NSs, despite having approximately just one-third the size of RVFV NSs, has similarly evolved several independent mechanisms to destroy RNAP II and suppress IFN induction and that the Elongin C-dependent branch is just one of them. Removing Elongin C can help to uncover these additional anti-IFN strategies which may contribute to the degradation of RPB1 and breakdown of RNAP II activity, which is even faster than for RVFV NSs.
An interesting side observation of our studies was the establishment of Elongin C (but not A or B) as a nucleolar protein. We were able to locate Elongin A in a database of the nucleolar proteome (63) but not Elongin C. We therefore hypothesize a transient and weak association, which is probably disrupted upon cell lysis or nucleoli preparation. In support of this, LACV (and SBV) NSs disturbs the nucleolar organization only slightly (and LACV NSs still allows RNAP I activity there [28]), but this might be sufficient to release Elongin C. ␣-Amanitin, in contrast, dissolves nucleoli entirely and also redistributes Elongin C.
Immunofluorescence analysis. Cells, grown on coverslips to 30 to 50% confluence, were either mock treated, infected with wt LACV or LACVdelNSs, or treated with ␣-amanitin for the indicated times. After the cells were fixed with 3% paraformaldehyde, they were permeabilized with 0.5% Triton X-100 in PBS. Unspecific staining was reduced by incubating the cells in blocking and staining solution (2% bovine serum albumin, 5% glycerol, and 0.2% Tween 20 in PBS) for 30 min at room temperature. The following primary antibodies, diluted in blocking and staining solution, were used: rabbit polyclonal anti-LACV N (1:500), mouse monoclonal anti-LACV G C (1:400, kindly provided by Francesco Gonzales-Scarano, Perelman School of Medicine, Philadelphia, PA), Elongin A (1:500, rabbit; Sigma-Aldrich, HPA005910), Elongin B (1:100, rabbit; Santa Cruz, sc-11447), Elongin C (1:250, mouse; BD Transduction Laboratories, 610761), and nucleolin (1:1,000; Abcam, ab22758). The primary antibodies were incubated on cells at room temperature for 1 h, followed by three washes in PBS. The secondary antibody was either Alexa Fluor 488 donkey anti-mouse IgG Statistical analysis. The quantitative data are presented as means Ϯ the standard deviations (SD) for three biological replicates. The statistical significance between two groups was evaluated by using a nonparametric, one-tailed Wilcoxon paired signed-rank test, with a P value of Ͻ0.05 considered statistically significant.