Stimulation of RNA Polymerase II Elongation by Hepatitis Delta Antigen

See allHide authors and affiliations

Science  06 Jul 2001:
Vol. 293, Issue 5527, pp. 124-127
DOI: 10.1126/science.1057925


Transcription elongation by RNA polymerase II (RNAPII) is negatively regulated by the human factors DRB-sensitivity inducing factor (DSIF) and negative elongation factor (NELF). A 66-kilodalton subunit of NELF (NELF-A) shows limited sequence similarity to hepatitis delta antigen (HDAg), the viral protein required for replication of hepatitis delta virus (HDV). The host RNAPII has been implicated in HDV replication, but the detailed mechanism and the role of HDAg in this process are not understood. We show that HDAg binds RNAPII directly and stimulates transcription by displacing NELF and promoting RNAPII elongation. These results suggest that HDAg may regulate RNAPII elongation during both cellular messenger RNA synthesis and HDV RNA replication.

The human transcription elongation factors DSIF and NELF bind RNAPII and repress the elongation activity of this polymerase (1–4). This repression is reversed by P-TEFb, a positive elongation factor with a protein kinase activity that phosphorylates the COOH-terminal domain (CTD) of RNAPII and a subunit of DSIF (5), in a manner sensitive to the kinase inhibitors 5,6-dichloro-1-beta-d-ribofuranosylbenzimidazole (DRB) and H8. Because DRB affects the synthesis of most mRNAs, the DRB-sensitive elongation involving DSIF, NELF, and P-TEFb may reflect a general rate-limiting step of RNAPII transcription (6). The yeast homologs of DSIF, transcription factors Spt5 and Spt4, have been shown to affect elongation (7). Purified NELF is composed of five polypeptides, A to E, the smallest of which, NELF-E (46 kD), is identical to a putative RNA-binding protein (4). We microsequenced NELF-A (66 kD) and found it to be encoded by WHSC2 (Fig. 1A), a candidate gene for the Wolf-Hirschhorn syndrome, a multiple malformation syndrome characterized by mental and developmental defects resulting from a deletion in chromosome 4p16.3 (8). Computer analyses identified a weak sequence similarity (27% identity) between the NH2-terminal half of NELF-A/WHSC2 (amino acids 89 to 248) and HDAg, the sole protein encoded by HDV, with the similarity extending to the predicted secondary structures of these proteins (Fig. 1B).

Figure 1

NELF-A is encoded by WHSC2. (A) NELF from the final purification step was visualized by silver staining. Molecular sizes (in kD) of the subunits are indicated in parentheses. (B) Ami- no acid sequence of human NELF-A/WHSC2. The three obtained peptide sequences (underlined) match parts of the predicted human WHSC2 protein (22). (C) Sequence comparison of NELF-A/WHSC2, HDAg-S (genotype IA, P35884), and HDAg-S (genotype IB, 225754). The homology was initially identified through a search against the ProDom database (23) with NELF-A/WHSC2 as a query. Identical residues are boxed. Abbreviations used are c, charged; h, hydrophobic; l, aliphatic; p, polar; s, small; t, turnlike; u, tiny, and +, positive (24). Also shown are the secondary structures as predicted by the program PHD (25). Only positions with the predicted accuracy of greater than 82% are indicated (E, sheet; H, helix; L loop). Single-letter abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

HDV, a satellite of hepatitis B virus, has a ∼1700-nucleotide (nt) circular, single-stranded RNA genome with a rodlike structure (9). Replication of HDV RNA appears to involve the host RNAPII and requires the presence of HDAg (9–11). Two forms of HDAg, HDAg-S (195 amino acids long) and HDAg-L (214 amino acids long), originate from editing of the common mRNA. Both forms bind HDV RNA, but have distinct roles in the viral life cycle. HDAg-S activates HDV replication, whereas HDAg-L, which contains a 19–amino acid COOH-terminal extension, inhibits replication and directs virion assembly (9–12). Earlier reports have implicated HDAgs in both activation and inhibition of RNAPII transcription (13, 14), but the nature of this discrepancy and the mechanism of HDAg action are unknown.

To investigate if, and how, HDAgs regulate RNAPII transcription, we examined the effect of HDAg on DNA-templated transcription using HeLa nuclear extracts (NE). In the presence of DRB, endogenous DSIF/NELF can repress transcription (Fig. 2A) (4). HDAgs reversed this inhibition (HDAg-S was more effective than HDAg-L), with little effect on basal (–DRB) transcription (Fig. 2A). This effect required NELF, because NE immunodepleted of NELF failed to respond to HDAgs (Fig. 2B). These results suggest that HDAg stimulates RNAPII transcription by counteracting the negative effect of DSIF/NELF.

Figure 2

HDAgs stimulate DNA-templated transcription with crude (A and B) or purified (Cand D) transcription systems. (A) Increasing amounts of HDAgs (12, 50, and 200 ng) were added to transcription reactions containing NE, as described (4,26). DRB (50 μM) was included where indicated. Upper and lower bands represent 380- and 270-nt G-less transcripts from the adenovirus E4 and major-late promoters, respectively. (B) HDAgs (200 ng) were added to reactions containing either untreated or NELF-immunodepleted NE. [(C) and (D)] WT or mutant HDAg (100 ng) was incubated with purified RNAPII and dC-tailed templates with or without DSIF and NELF (4, 27). α-Amanitin (ama) (1.0 μg/ml) was included where indicated.

Because HDAg does not affect the kinase activity of P-TEFb or CTD phosphorylation (15), we sought to determine whether it affects association of NELF, DSIF, and RNAPII under transcription conditions. In NE prepared from HeLa cells expressing Flag–NELF-E, antibodies to Flag (anti-Flag) immunoprecipitated the other NELF subunits (15), DSIF, and RNAPII (Fig. 3A) (4). Preincubation of the NE with HDAg-S substantially reduced the levels of DSIF and RNAPII in the precipitate (Fig. 3A), but had little effect on DSIF-RNAPII interaction (15). To determine the direct target of HDAg, we analyzed the NE proteins associated with glutathioneS-transferase (GST)–HDAg. Under saturating concentrations, both GST–HDAg-L and GST–HDAg-S, but not GST alone, bound to substantial amounts of RNAPII (Fig. 3B). In addition, small amounts of DSIF, but not NELF, were found associated with GST-HDAg (Fig. 3B). These results suggest that HDAg directly binds RNAPII and inhibits NELF-RNAPII association, possibly because HDAg competes with NELF-A for a common surface on RNAPII, without substantially affecting the DSIF-RNAPII interaction. Binding of HDAg-S to RNAPII was strongly impaired by a deletion of eight amino acids from its COOH-terminus, whereas the NH2-terminal segment—including the oligomerization domain, NLS, and a part of the RNA-binding domain—was dispensable (Fig. 3B). Thus, amino acids 130 to 195 of HDAg-S are sufficient for RNAPII binding. The COOH-terminus of HDAg-S is conserved among different genotypes of HDV (16), but to date, no function has been assigned to this region.

Figure 3

Direct interaction between HDAgs and RNAPII. (A) HDAg-S inhibits interaction between NELF and DSIF/RNAPII. NE obtained from HeLa cells constitutively expressing Flag–NELF-E was incubated with either HDAg-S or HDAg-L (the amounts correspond to the “1/4×” amounts used in Fig. 2A) for 45 min in transcription buffer containing 100 mM KCl and 0.1% Nonidet P-40. The mixture was immunoprecipitated with anti-Flag, and the precipitates (P) were analyzed by immunoblotting with the indicated antibodies (28). Input (IN) and supernatant (S) lanes represent 2% of the total material. (B) Various GST-HDAg derivatives (∼1.0 μg) were coupled to glutathione-Sepharose and incubated with NE (100 μl). After extensive washing with the transcription buffer containing 100 mM KCl and 0.1% Nonidet P-40, bound materials were analyzed by immunoblotting and by silver- or Coomassie-staining. (C) Structure of HDAg-S. A solid bar corresponds to the minimal region required for RNAPII-binding. NLS, nuclear localization signal, and ARM, arginine-rich motif.

To directly measure the effect of HDAg on elongation, we used reactions containing purified RNAPII and deoxycytidine (dC)-tailed templates. In the presence of 4NTPs, transcripts of increasing lengths appeared with time, and DSIF/NELF inhibited this process (Fig. 2C) (4). The addition of HDAgs strongly stimulated RNAPII elongation, irrespective of the presence of DSIF/NELF, whereas the HDAg-S mutant lacking 31 amino acids at the COOH-terminus (ΔC31) was inactive (Fig. 2, C and D). These results suggest that the HDAg-RNAPII interaction not only counteracts the repression by DSIF/NELF but can also stimulate RNAPII elongation in a DSIF/NELF-independent manner. In crude NE (Fig. 2, A and B), the presence of factors that stimulate transcription more efficiently than HDAg might obscure the DSIF/NELF independent effect of HDAg.

To investigate the role for HDAg in HDV RNA replication, we used a previously described model system in which RNAPII in HeLa NE transcribes a specific segment of HDV RNA (17). This reaction involves cleavage of the RNA template at a unique site followed by extension of the new 3′ end, generating a chimeric template/transcript product (Fig. 4A). Transcription discontinued after copying 41 nt of the template, suggesting that this region of HDV RNA contains a pause signal, and/or that some protein factors required for transcription are limiting in the reaction (17). DRB inhibited this reaction, indicating that DSIF/NELF participates in the RNA-templated transcription (Fig. 4B). The addition of HDAg-S reversed the DRB-mediated inhibition and strongly stimulated basal transcription (Fig. 4B), suggesting that HDAg-S affects the RNA-templated transcription through both DSIF/NELF-dependent and -independent mechanisms, as observed with purified RNAPII (Fig. 2, C and D).

Figure 4

HDAg stimulates HDV RNA-templated transcription. (A) Sequences of the template AG103 (top) and the product Pb (bottom). The 3′ end of Po and Pa is marked with an arrow. RNase A cleavage sites are indicated by triangles. The segment transcribed by RNAPII is boxed. (B) HDV RNA template, AG103, and HeLa NE were incubated in the presence of [α-32P]GTP (17), with (+) or without (–) DRB (50 μM) and HDAg-S (200 ng). (C) The same reaction products were resolved in a 5% polyacrylamide–8 M urea gel. Positions of nonspecifically labeled templates (T) and specific products (Po, Pa, and Pb) are indicated. (D) The reaction products from (C) (lanes 1 and 6) were gel purified, subjected to exhaustive RNase A digestion, and resolved in a 25% polyacrylamide–8 M urea gel. Pa+b is a mixture of Pa and Pb. (E) Plasmids expressing HDV RNA—pSVL(D3) (WT) or pSVL(D2m) (not expressing HDAg)—were introduced into HeLa cells with or without an additional HDAg expression plasmid. Three days after transfection, total RNA was isolated and the level of 1.7-kb genomic HDV RNA was monitored by Northern blotting with an antigenomic-sense oligonucleotide probe.

Upon better resolution of products of the HDAg-S–supplemented reactions, two distinct RNAs were detected (Fig. 4C). Analysis of ribonuclease (RNase) A digests of these RNAs demonstrated that, whereas the slower migrating product (Pa) is identical to Po formed in the absence of HDAg, the faster migrating product (Pb) is a previously unknown species, extended at its 3′ end by at least 15 nt (Fig. 4, A and D). The unusually fast mobility of Pb probably results from the presence of an extended double-stranded segment in its structure. These results suggest that HDAg stimulates HDV RNA replication by suppressing RNAPII pausing at +41. Unexpectedly, this position corresponds to the primary site of DSIF/NELF repression during RNAPII elongation on DNA (4).

To elucidate the physiological relevance of these findings, we compared the ability of wild-type (WT) and mutant HDAgs to support in trans replication of the HDAg-defective HDV mutant. WT HDAg-S, but not ΔC8 or ΔC31, supported replication of the defective virus (Fig. 4E). Therefore, the same region of HDAg is required for efficient RNAPII elongation in vitro and HDV RNA replication in vivo.

These results establish that HDAg directly binds RNAPII and stimulates transcription by two different mechanisms: It reverses the negative effect of DSIF/NELF by displacing NELF from RNAPII and directly stimulates transcription elongation. This is, to our knowledge, the first example of a viral protein that regulates elongation through direct binding to RNAPII. Nonetheless, similarity between HDAg and human immunodeficiency virus (HIV) Tat is intriguing. During HIV transcription, Tat stimulates RNAPII elongation by recruiting P-TEFb and reversing the negative effect of DSIF/NELF immediately downstream of the promoter (18).

HDV RNA resembles the genomes of plant viroids, although viroid RNAs are much smaller (<400 nt) and contain no open reading frames. This has led to a hypothesis that HDV may have evolved from a primitive viroidlike RNA through capture of a cellular transcript (19). Previously, a putative cellular homolog of HDAg was identified (20). This protein, termed DIPA, is 24% identical to HDAg but shows no apparent sequence similarity with NELF-A (15). The low levels of sequence similarity make it difficult to define evolutionary relationships of the three proteins. Regardless of its evolutionary origin, the HDAg-RNAPII interaction is critical for HDV replication and could be used as a therapeutic target against this pathogenic virus.

  • * To whom correspondence should be addressed. E-mail: hhanda{at}


View Abstract

Navigate This Article