The Dihydroquinolizinone Compound RG7834 Inhibits the Polyadenylase Function of PAPD5 and PAPD7 and Accelerates the Degradation of Matured Hepatitis B Virus Surface Protein mRNA

Hepatitis B virus (HBV) mRNA metabolism is dependent upon host proteins PAPD5 and PAPD7 (PAPD5/7). PAPD5/7 are cellular, noncanonical, poly(A) polymerases (PAPs) whose main function is to oligoadenylate the 3′ end of noncoding RNA (ncRNA) for exosome degradation. HBV seems to exploit these two ncRNA quality-control factors for viral mRNA stabilization, rather than degradation. RG7834 is a small-molecule compound that binds PAPD5/7 and inhibits HBV gene production in both tissue culture and animal study.

ABSTRACT

Hepatitis B virus (HBV) mRNA metabolism is dependent upon host proteins PAPD5 and PAPD7 (PAPD5/7). PAPD5/7 are cellular, noncanonical, poly(A) polymerases (PAPs) whose main function is to oligoadenylate the 3′ end of noncoding RNA (ncRNA) for exosome degradation. HBV seems to exploit these two ncRNA quality-control factors for viral mRNA stabilization, rather than degradation. RG7834 is a small-molecule compound that binds PAPD5/7 and inhibits HBV gene production in both tissue culture and animal study. We reported that RG7834 was able to destabilize multiple HBV mRNA species, ranging from the 3.5-kb pregenomic/precore mRNAs to the 2.4/2.1-kb hepatitis B virus surface protein (HBs) mRNAs, except for the smallest 0.7-kb X protein (HBx) mRNA. Compound-induced HBV mRNA destabilization was initiated by a shortening of the poly(A) tail, followed by an accelerated degradation process in both the nucleus and cytoplasm. In cells expressing HBV mRNA, both PAPD5/7 were found to be physically associated with the viral RNA, and the polyadenylating activities of PAPD5/7 were susceptible to RG7834 repression in a biochemical assay. Moreover, in PAPD5/7 double-knockout cells, viral transcripts with a regular length of the poly(A) sequence could be initially synthesized but became shortened in hours, suggesting that participation of PAPD5/7 in RNA 3′ end processing, either during adenosine oligomerization or afterward, is crucial for RNA stabilization.

INTRODUCTION

Chronic hepatitis B (CHB) remains a major public health threat which affects more than 260 million people worldwide (1, 2). Without an effective treatment, CHB can lead to liver cirrhosis and hepatocellular carcinoma (HCC) and is responsible for nearly one million deaths annually (3). Currently, there are two classes of drugs to manage CHB, namely, nucleos(t)ide analogue (NA) and interferon alpha (IFN-α). NAs are inhibitors of the viral polymerase and usually require lifelong administration. The application of IFN-α, a nonspecific immune modulator, is limited by side effects (4–6). Both modalities have disappointing cure rates and only suppress viral DNA replication in selected patients but are impotent to induce a true clearance of the viral genome in the liver; therefore, the viral antigen especially the surface protein (HBs) level remains high in the blood, even after long-term therapy (7, 8). Circulating HBs, in the form of bilipid particles, is derived from covalently closed circular DNA (cccDNA) and viral genome fragments integrated into the host chromosome. The persistent high level of serum HBs is believed to have a role in mediating chronicity, possibly by causing an “exhaustion” of immune cells (9–11).

Hepatitis B virus (HBV) messenger RNAs coding for surface proteins are transcribed by host RNA polymerase II (RNAPII) (12, 13). They are composed of the 2.4-kb large surface protein RNA (LHBs) and 2.1-kb middle/small surface protein RNA (MHBs/SHBs). Similar to eukaryotic mRNAs, HBs mRNA transcription and maturation take place in the nucleus (14, 15). The maturation process includes the addition of a 5′ 7-methylguanosine cap and a 3′ poly(A) tail that are necessary for viral transcript stabilization, transportation, and translation (12, 16). How the steady-state level of matured HBs mRNA is regulated remains largely unclear. Given that the HBV mRNAs are generated by the host transcription machinery, it is likely that the host RNA surveillance and degradation machinery, which influence the turnover of cellular mRNA, also contribute to HBs mRNA turnover. In both the nucleus and cytoplasm, degradation of host mRNA in mammalian cells is known to be carried out by several elaborate mechanisms. The common route of host mRNA decay starts with the removal of the poly(A) tail, followed by decapping at the 5′ end. Once the 5′ cap is removed, a ribonucleolytic process takes place from both the 5′ and 3′ ends of the RNA. In addition, mRNA can be cleaved by endoribonucleases, followed by exonuclease digestion from both ends (17, 18).

Recently, compounds that affect the stability of viral RNA and host noncoding RNAs have attracted interest for their biochemical and therapeutic potential (19–22). It has been reported that the stability of HBV mRNA, especially the SHBs mRNA, is regulated by the host noncanonical poly(A) polymerase-associated domain 5 and 7 (PAPD5 and PAPD7; also called TENT4B and TENT4A, respectively). Knockdown of both PAPD5 and PAPD7 (depicted as PAPD5/7 when referred together hereafter) with small interfering RNAs (siRNAs) causes a significant reduction of HBs in HepaRG cells and primary human hepatocytes (23). Moreover, a dihydroquinolizinone (DHQ) small-molecule compound called RG7834, which bound to, and presumably inhibited, PAPD5/7 enzymatic activity, was identified to be able to reduce SHBs mRNA (24).

The function of human PAPD5/7 polypeptides was initially studied in yeast by their orthologs Trf4 and Trf5 as mediators for RNA quality control (25–27). In Saccharomyces cerevisiae, Trf4 is the key component of the Trf4-Air1/2-Mtr4 polyadenylation (TRAMP) complex that promotes aberrant polymerase I (Pol I)-transcribed RNA decay in the nucleolus (28, 29). Similarly, either PAPD5 or PAPD7 is the key component of the human TRAMP (PAPD5, ZCCHC7, and SKIV2L2) complex (25, 30). The role of PAPD5/7 in mammalian cell RNA metabolism is to add oligoadenosines at the 3′ ends of aberrant, noncoding RNA for exosome degradation (23, 24). In addition, mammalian PAPD5 is found to be important for the maturation of snoRNA, micro RNA (miRNA), and rRNA. Knockdown of PAPD5 disturbs the production of these ncRNAs (26, 27, 31, 32).

HBV seems to hijack PAPD5/7’s RNA adenylating function to maintain its mRNA tail length and subsequently stabilizes the transcript. We reported that PAPD5/7 polypeptides were physically associated with SHBs transcripts. Suppression of PAPD5/7 polyadenylating function with small-molecule RG7834 accelerated viral mRNA turnover. RG7834-induced reduction of HBV mRNA was initially manifested by the appearance of transcripts with a shortened 3′ end. The tail-shortening process could occur after the viral mRNA sequence reached its full length, suggesting that PAPD5/7 may play a role in stabilizing mature viral mRNAs. PAPD5/7-mediated HBV mRNA protection occurred in both the nucleus and cytoplasm where PAPD5 and PAPD7 were preferentially localized, respectively. Interestingly, in a PAPD5/7 double-knockout cell line, HBV mRNA with a regular length could be initially synthesized but was subjected to be tail trimmed in hours, suggesting that HBV hijacks PAPD5/7 to protect its mRNA terminal from being attacked by host exoribonucleases.

RESULTS

RG7834 accelerated the turnover of multiple species of HBV RNAs after their maturation.

We and others have previously reported that incubation of HBV-producing cells, in culture, with the noncanonical polyadenylase PAPD5/7 binding compound RG7834 causes significant decreases in HBV antigen and virion production. This is the result of posttranscriptional reduction of multiple species of HBV mRNA (24, 33, 34). However, it is not clear if RG7834’s destabilizing function takes place during viral RNA poly(A) processing or afterward. To determine the influence of RG7834 on the stability of mature HBV RNA, we infected HepG2-tTA25 cells with adenovectors expressing the 3.5-kb pregenomic RNA (pgRNA), 2.4-kb large surface protein mRNA, 2.1-kb middle/small surface protein mRNA, and the 0.7-HBx mRNA under the tet-off promoter (see Fig. S1 in the supplemental material). Doxycycline (dox) was added to the culture media 3 days after adenovirus infection to stop viral transcription, and the amount of HBV mRNA in the cells was determined as a function of time following the stop of transcription by Northern blotting. We used adenovectors to learn if different HBV mRNAs are sensitive to RG7834, after their transcription is terminated. The reason to use an adenovector approach is because mRNAs transcribed from plasmids quite often render RG7834 less potent and sometimes even fail to downregulate HBV mRNA.

Both the 3.5-kb pgRNA and 2.4/2.1-kb surface protein mRNAs of HBV showed sensitivity to the application of RG7834, even after the mRNAs were at their full lengths (Fig. 1). Viral RNA derived from RG7834-treated samples diminished faster than RNA from cells receiving dimethyl sulfoxide (DMSO) (Fig. 1A). Reduction of viral mRNA became apparent as early as 4 hours after RG7834 treatment. Using the 2.1-kb SHBs mRNA for half-life calculation, we found that RG7834 decreased the RNA half-life by approximately 2 hours (Fig. 1B). After 24 hours, RG7834-treated cells had undetectable levels of viral RNA, whereas DMSO-treated cells still retained 20% to 30% of SHBs mRNA. On the other hand, the stability of the 0.7-kb HBx mRNA was not obviously changed under RG7834 treatment. In our expressing system, HBx had a half-life shorter than 4 hours. The action of RG7834 on mRNA stability required around 4 hours to have a detectable change, so any effect might not be apparent on HBx mRNA. However, in cells constitutively expressing HBx mRNA, treatment of RG7834 for 3 days did not alter the level of this smallest viral RNA significantly, indicating that RG7834 did not affect the steady-state level of the HBx transcript (Fig. 1C).

FIG 1.

PAPD5/7 inhibitor RG7834 accelerated the turnover of multiple species of HBV mRNA. (A) HepG2-tTA25 cells were infected with adenoviruses that expressed HBV pgRNA, LHBs, M/SHBs, or HBx in a doxycycline (dox)-controllable manner (vector maps were shown in Fig. S1). Three days after adenovirus infection, dox was added back to the culture media at time 0 to terminate viral gene transcription. RG7834 treatment, at 1 μM, was initiated at the same time when dox was added. Total cellular RNA was extracted at indicated time points for Northern blot. (B) Similar to the experimental design in A, the change of 2.1-kb SHBs mRNA under RG7834 treatment was monitored at additional time points for half-life calculation. RNA signal was measured using ImageJ software. Average and standard deviation values shown in the plot were based on 3 independent experiments. (C) HepG2-tTA25 cells were infected with adenovirus expressing a CMV-IE-driven HBx gene. Twenty-four hours after adenovirus infection, cells were treated with 1 μM RG7834 for 3 days and HBx mRNA was detected in Northern blot. All the experiments were conducted at least three times, and representative images were presented. Dox(+), doxycycline was always present in culture medium.

RG7834-induced HBV mRNA decay had two catalytic phases.

HBV mRNA degradation induced by RG7834 occurs as a two-phase process. Within the first 4 hours after compound treatment, viral RNA levels were not drastically decreased; however, the size of mRNA was reduced first (Fig. 1A). The reduction in viral RNA size was evident by the faster mobility of RG7834-treated RNA samples in Northern blots at 4 hours. Significant reductions in viral RNA levels were observed 6 to 8 hours after compound treatment for both the 3.5-kb RNA and SHBs mRNA (Fig. 1A and B). In HepAD38, a stable cell line that supports HBV replication in a tet-off manner (35), initiation of RG7834 treatment at the time of dox addition reduced the viral mRNA size 4 hours later (Fig. 2A). Since RG7834 is believed to be an RNA poly(A) polymerase inhibitor, we speculated that poly(A) tail trimming was responsible for the elevated mobility of the viral mRNA that preceded the mRNA degradation. We extracted total RNA from cells treated with DMSO or RG7834 for 6 hours and ligated the mRNA with an RNA adaptor at the 3′ end. The hybrid RNA was next reverse transcribed, followed by PCR amplification with a set of primers specific for HBV and the adaptor (primer set of Pf1 and Pr1). While DMSO treatment yielded a distinct band, samples from cells treated with RG7834 presented smeared bands with reduced sizes (Fig. 2B). On the other hand, when forward primer Pf2 paired with reverse primer Pr2, which were both located upstream of the HBV mRNA poly(A) tail (starting from nucleotide [nt] 1935), the PCR product gave rise to bands with identical size, suggesting that the viral poly(A) tail was truncated after the cell was exposed to RG7834. It is likely that the two-phase decay phenomenon is caused by two consecutive events, namely, a rapid 3′ tail removal and a delayed mRNA degradation.

FIG 2.

RG7834 induced a shortening of the HBV mRNA 3′ tail. (A) HBV transcription in HepAD38 cells was terminated by the addition of dox at time 0. Treatment of the cells with 1 μM RG7834 was simultaneously initiated at time 0, as well. Cellular RNA was extracted at indicated time points and analyzed with Northern blot. (B) RNA derived from the 6-hour time point in A was ligated to an RNA adaptor and was used for RT-PCR to measure the 3′ tail length using an HBV-specific primer and a primer located in the adaptor (Pf1 and Pr1). RNA size between the poly(A) signal and 5′ transcription start site was monitored with PCR primers Pf2 and Pr2. PCR products were resolved in agarose gel, and Lambda/HindIII fragments were used as DNA ladders. *, PCR was carried out on RNA samples without reverse transcription.

HBV mRNA was associated with PAPD5/7 polypeptides.

RNA polyadenylases PAPD5/7 need to be associated with HBV mRNA to extend or protect the poly(A) tail. To study the enzyme-substrate relationship, we introduced both pCMV-FlagD5 and pCMV-S plasmids into HEK293 cells by transient transfection. The use of HEK293 rather than HepG2 for the association assay is because HEK293 can be more efficiently transfected and the yield of PAPD5/7 in HEK293 is significantly higher than that in HepG2 cells. Two days posttransfection, cellular lysates were subjected to immune precipitation (IP) with either M2 anti-Flag antibody or nonspecific mouse IgG. Western blotting of the IP products revealed that PAPD5 was present only in the anti-Flag-precipitated complex, and the amount of control IgG used in the experiment was comparable (Fig. 3A). The immunoprecipitated complexes were next subjected to proteinase K digestion and phenol-chloroform extraction to purify RNA that complexed with the PAPD5 protein. Reverse transcriptase quantitative PCR (RT-qPCR) signals for SHBs mRNA in the Flag antibody-precipitated preparation, after normalization with both spiked-in tetracycline transactivator (tTA) mRNA and β-actin mRNA, were significantly stronger than those from IgG immunoprecipitated complexes (Fig. 3B). As a control, in cells transfected with pCMV-S alone, the M2 anti-Flag antibody was unable to preferentially pull down viral RNA, indicating that the association between SHBs mRNA and PAPD5 was specific. The association between PAPD7 and SHBs mRNA was confirmed similarly (Fig. 3A and B).

FIG 3.

SHBs mRNA was associated with PAPD5/7. (A) HEK293 cells were transfected with pCMV-FlagD5 or pCMV-FlagD7 plus SHBs-expressing vector pCMV-S. Cells transfected with pCMV-S alone were used as a specificity control for anti-Flag antibody. Two days after transfection, the cell lysate was subjected to immune precipitation with an equal amount of Flag antibody or nonspecific mouse IgG. A fraction of the cell lysate and immune precipitates were denatured in Laemmli buffer and resolved in an SDS-PAGE gel, followed by detection with anti-Flag antibody. (B) RNA bound to precipitates was extracted with phenol-chloroform after proteinase K digestion, followed by DNase I clearance. Before RNA extraction, in vitro-transcribed tetracycline transactivator (tTA) RNA was spiked into the RNP precipitates for normalizing RNA extraction variation. Purified RNA was RT-qPCR quantified using gene-specific primers against HBV, tTA, and β-actin. RNA detection in the precipitated RNP complex was carried out in triplicates for statistical analysis. qPCR measurements of SHBs mRNA were consecutively normalized with readings from tTA and β-actin detection. Mock, cells transfected with empty vector; *, Student’s t test, P < 0.01.

RG7834 suppressed poly(A) polymerase function of PAPD5/7.

To determine if the observed shortening of the HBV mRNA 3′ tail by RG7834 was the result of the inactivation of the poly(A) polymerase of PAPD5/7, we transfected HEK293 cells with a plasmid expressing Flag-tagged PAPD5. The expressed polypeptides were then precipitated with beads conjugated with an anti-Flag antibody. Immune-precipitated beads were directly used for the polyadenylation assay as described previously (36). The purified PAPD5 complexes were able to extend the 3′ tail of the P³²-labeled RNA oligonucleotides in a time-dependent manner (see Fig. S2A in the supplemental material). After 2.5 minutes of incubation, the 30-nucleotide-long RNA oligonucleotides were significantly elongated and continued extending up to 60 minutes. The elongation signal reached peak level at 20 minutes and decreased afterward, suggesting that there might be RNase present in the precipitates (Fig. S2A). The question of whether this ribonucleolytic cleavage is polyadenylation dependent or is mediated by nonspecific RNase is worthy of further investigation. On the other hand, the complex precipitated from HEK293 cells transfected with a control vector failed to elongate RNA oligonucleotides, indicating that the polyadenylating function of the anti-Flag-precipitated complex was indeed PAPD5 specific (Fig. S2A, lane 8).

PAPD5 polyadenylase can use not only the ATP but also, to a lesser extent, other nucleotides for RNA tailing (37, 38). Intermittent incorporation of GTP into the RNA poly(A) tail promotes stabilization of the transcript (36). After 20 minutes of incubation, the PAPD5 complex precipitated from HEK293 cells incorporated GTP, but not UTP and CTP, into the RNA oligonucleotides (Fig. 4A, lane 7). As a tailing substrate, GTP has a much lower incorporating efficacy than that of ATP. When a mixture of ATP, UTP, CTP, and GTP at 1 mM each was used, the elongation capacity was elevated but remained significantly lower (120 nt versus greater than 270 nt) (Fig. 4A, lanes 3 to 4) than that of ATP alone. This observation suggests that the addition of guanosine into the RNA tail may pause the polymerizing activity of PAPD5.

FIG 4.

RG7834 inhibited the polyadenylase activities of PAPD5/7. (A) HEK293 cells were transfected with plasmid pCMV-FlagD5. Two days posttransfection, the tagged PAPD5 polypeptides were precipitated with anti-Flag antibody. The precipitated beads, mixed with 1 mM ATP, UTP, CTP, GTP, or NTP (mixture of 1 mM of each nucleotide), were incubated with P³²-labeled RNA oligonucleotides at 37°C for 20 minutes. The enzymatic reaction was stopped by the addition of 2× RNA loading buffer followed by Urea-TBE gel electrophoresis. Radioactive signals were detected with phosphorimager. (B) Serially diluted RG7834 was applied to the RNA tailing reaction mixture containing 1 mM ATP as the nucleotide substrate. After 20 minutes of incubation at 37°C, the end product was resolved in a Urea-TBE gel for phosphorimager quantification. (C and D) Similar to A and B, flag-tagged PAPD7 was incubated with 1 mM of individual nucleotides or a mixture of 4 nucleotides at concentration of 1 mM each at 37°C for RNA elongation. Serially diluted RG7834 was also applied in the in vitro tailing assay using ATP as the substrate. Note, the incubation time for the PAPD7-catalyzed reaction lasted for 3 h. Arrowheads indicated elongated RNA oligonucleotides containing guanosines. Experiments were conducted at least three times, and representative images were presented.

Next, we asked if the polyadenylation and polyadenylation activities of PAPD5 were subjected to RG7834 inhibition. Figure 4B showed that the addition of RG7834 in the polyadenylation reactions at or above 12 nM resulted in an apparent reduction of both the length and signal intensity of the poly(A) tail. However, a significant decrease of polyguanylation required the concentration of RG7834 to be higher than 37 nM, indicating that the inhibitory potency of a PAPD5 antagonist might be related to the composition of the nucleotide substrates (Fig. S2B).

Since the context of the PAPD7 N terminus is known to be crucial for its enzymatic activity, we exploited the N-terminal extended isoform of PAPD7 (792 amino acid [aa]) for in vitro assay development (36). Forty-eight hours after plasmid transfection in HEK293 cells, a tagged PAPD7 protein was precipitated with an anti-Flag antibody. However, compared with PAPD5, bead-bound PAPD7 was relatively less active in elongating the RNA oligonucleotides, although both proteins had comparable expression levels and pulldown efficiencies (Fig. 3A and B, Fig. 4C). In contrast to PAPD5-catalyzed RNA bands, PAPD7-generated signals showed a smeared pattern even after the 3-hour incubation, suggesting that its catalytic activity is not robust under our assay conditions (Fig. 4C). Moreover, the incorporation of guanosine into the RNA tail could barely take place and only a residual amount of elongated tail, probably a single 1- to 2-nucleotide extension, was seen (Fig. 4C, arrowhead site). Since PAPD7 had an extremely low preference to use GTP, this enzyme, in contrast to PAPD5, could elongate RNA with ATP regardless of the presence of other nucleoside triphosphates (NTPs) (Fig. 4C, lane 3). The relative impotency of PAPD7 to tail RNA oligonucleotides in vitro is also likely due to its intrinsic low enzymatic activity compared with its paralog PAPD5, although both polyadenylases share over 80% amino acid homology in their active domains (29, 31, 37). Titration of RG7834 in an RNA polyadenylating assay revealed that PAPD7’s function was significantly inhibited when the compound concentration reached 100 nM and above, indicating that PAPD7 is less sensitive than PAPD5 to RG7834 (Fig. 4D). Since PAPD7-mediated G tailing was barely detectable, its susceptibility to RG7834 inhibition was not evaluated in the current study.

The polyadenylase function of PAPD5/7 was regulated by viral HSLα structure and ZCCHC14.

As shown here and elsewhere, RG7834 selectively destabilizes HBV mRNA. The compound did not interfere with the stability of RNA from other DNA and RNA viruses, and the specificity of RG7834 seems to be determined by the presence of the HBV posttranscriptional regulatory element (PRE) at the 3′ untranslated region (UTR) of mRNA (33). HSLα is a conserved stem-loop structure in the PRE and is believed to be important for mediating the PRE-protein interaction for RNA transportation and stabilization (39, 40). PAPD5/7 appear to lack defined RNA recognition motifs (RRMs) (38, 41). In the TRAMP complex, they require ZCCHC7 to interact with RNA substrates (25, 30). However, PAPD5/7’s HBV RNA-stabilizing function is TRAMP complex-independent, as neither ZCCHC7 nor SKIV2L2 is involved in HBV mRNA metabolism (23). Recently, it was demonstrated that ZCCHC14, a zinc finger CCHC-type 14 protein with an RNA binding domain, was able to bind HSLα for the maintenance of viral mRNA stability (34, 42). This finding is related to our previous report that the HSLα structure is essential for the antiviral activity of RG7834 (33). To determine if the interaction between compound RG7834, PAPD5/7, and ZCCHC14 is mediated by the presence of the HSLα structure, we knocked down PAPD5/7 and ZCCHC14 with siRNAs in HepG2 cells (Fig. S3) and infected the cells with an adenovirus that expresses the SHBs gene with either a wild-type or a mutated HSLα stem-loop. The HSLα mutant contains 4-nucleotide mutations in the apical CNGGN pentaloop that is believed to be important for host protein recruitment (Fig. 5A) (34, 40).

FIG 5.

The polyadenylase function of PAPD5/7 was regulated by HSLα structure. (A) Adenovirus expressing SHBs mRNA with wild-type HSLα (AdHSLα-WT) or mutated HSLα (AdHSLα-MUT) were used for HepG2 cell infection. Mutations in the loop and adjacent stem region of HSLα were marked in gray circles. (B) HepG2 cells were transfected with 20 nM siRNAs against PAPD5/7 or ZCCHC14, followed by AdHSLα-WT and AdHSLα-MUT infection. Twelve hours after adenovirus infection, treatment with DMSO or 1 μM RG7834 was initiated, and cells were harvested 3 days later for Northern detection using a P³²-labeled HBV riboprobe. A representative blot of three duplicates was shown. (C) RNA samples derived from B were subjected to RT-qPCR quantification using HBV-specific primers and normalized with GAPDH. *, Student’s t test, P < 0.01.

Results in Fig. 5 showed that compound RG7834 significantly inhibited the level of SHBs mRNA transcribed from AdHSLα-wild type (WT) but not from AdHSLα-mutant (MUT) (Fig. 5B and C). Knockdown of ZCCHC14 decreased wild-type mRNA but had less of an effect on the mutant, suggesting that both RG7834 and ZCCHC14 require HSLα to regulate viral mRNA expression, probably through interaction with PAPD5/7.

Interestingly, viral transcripts from AdHSLα-MUT remained to be sensitive to PAPD5/7 knockdown (Fig. 5B, lanes 7 and 8 versus 9 and 10) and the difference is statistically significant (Fig. 5C). This observation implies that PAPD5/7 may have an HSLα-independent function for pentaloop-mutated viral RNA maintenance. Whether mutated viral mRNA remains to be accessible to PAPD5/7 needs to be further investigated.

RG7834-induced HBV RNA 3′ trimming occurred in both the nucleus and cytoplasm.

It is known that RG7834 treatment causes a decrease of HBV mRNA in both the nucleus and cytoplasm, and the degradation in the cytoplasm is regarded as a secondary consequence of nuclear RNA decay (33). This was investigated further, and the possibility that RG7834 could induce HBV mRNA decay in the cytoplasm, independent of nuclear degradation, was specifically examined. After the termination of transcription, the nuclear and cytoplasmic distribution of SHBs mRNA was monitored in HepG2-tTA25 cells. HBV transcription was again stopped by addition of dox to the cell cultures, and viral mRNAs in the nuclear and cytoplasmic compartments were quantified by Northern blotting at various time points. Based on the HBV mRNA intensity which was proportioned to the total RNA yields from the nuclear and cytoplasmic extractions, it was apparent that at the time of transcription termination, greater than 90% of viral transcripts were present in the cytoplasm. Six hours after dox addition, approximately 94% (48.4/[48.4 + 2.7]) of the SHBs mRNAs were present in the cytoplasm (Fig. 6A and B). Application of 1 μM RG7834 for an additional 4 hours caused the appearance of a faster migrating band, suggesting that inhibition of PAPD5/7 could elicit an RNA trimming effect in the cytoplasm after the 3′ tailing process of viral mRNA was complete. Clearly, this tail-shortening event predominantly took place in the cytoplasm since 100% of detectable viral transcripts had their band size reduced (Fig. 6C). It is worth noting that in Fig. 6C, a slight shortening and enhanced SHBs mRNA signal was seen after the use of RG7834 for 4 hours (lanes 4, 8, and 12). Quite often, we observe a mild accumulation of the shortened band at early time points after compound application. The exact cause is unknown. Likely, RG7834-induced decay is different than natural viral RNA decay in that the tail-shortening phase is a quick reaction and the degradation phase may need time to adapt.

FIG 6.

RG7834-induced SHBs mRNA tail trimming occurred in both the nucleus and cytoplasm. (A) HepG2-tTA25 cells were infected with adenovirus expressing doxycycline (dox)-controlled SHBs (AdTRE-SHBs, tet-off manner). Three days after adenovirus infection, dox was added to the culture medium to stop viral transcription. Cells were next harvested and fractionated at the indicated time points for Northern analysis. Lanes labeled with “Dox(+)” meant dox was always present during cell culture. (B) SHBs mRNA signals from A were quantified with the ImageJ program and proportioned to nuclear and cytoplasmic compositions based on total RNA yield from each compartment. (C) HepG2-tTA25 cells infected with AdTRE-SHBs were treated with dox for 0, 6, or 6 hours followed by an additional 4-hour of 1 μM RG7834. Cells were then fractionated and analyzed with Northern blotting. Small nucleolar RNA63 (snoRA63) was used as a nuclear fraction marker. Numbers listed beneath the blot were values of SHBs mRNA measured with ImageJ software. Representative blots from 3 duplicated experiments were shown. HPD, hours post-dox addition.

PAPD5 and PAPD7 were preferentially distributed in the nucleus and cytoplasm, respectively.

The capability of RG7834 to downregulate HBV mRNA throughout the cell indicates that for extending/protecting the viral RNA tail either PAPD5 or PAPD7 alone or both must be present in each subcellular compartment. As the key component of the TRAMP complex, PAPD5 is reported to be predominantly localized in the nucleus (43–45). However, in human glioblastoma and Caenorhabditis elegans germ cells, PAPD5 is found exclusively in the cytoplasm (46–48). On the other hand, information about the cellular location of PAPD7 is much less documented. Ogami et al. reported that an enzymatically active PAPD7 was mainly found in the nuclear compartment, and deletion of the nucleotidyltransferase domain caused the polypeptide to be retained in the cytoplasm (41). A clear distribution of PAPD5/7 in the hepatocyte has not be documented. We therefore transfected HepG2 cells with plasmids expressing Flag-tagged PAPD5 or PAPD7 and performed immune fluorescent staining with an anti-Flag antibody. PAPD5 was found to be majorly localized in the nucleus and PAPD7 was present throughout the cell with a moderate enrichment in the cytoplasm (Fig. 7A).

FIG 7.

PAPD5 and PAPD7 were preferentially localized in the nucleus and cytoplasm, respectively. (A) HepG2 cells were transfected with plasmids pCMV-FlagD5 and pCMV-FlagD7, respectively. Transfected cells were reseeded onto the coverslip for an additional 2 days, followed by immune fluorescent staining with anti-Flag antibody. (B) HBV replicating HepG2-2.2.15 cells were employed for PAPD5/7 immune fluorescent staining as described in A. A representative image from five duplicated experiments was shown.

We also examined whether HBV infection could alter the subcellular locations of PAPD5/7. HepG2-2.2.15 cells (49) which support constitutive HBV replication were transfected with plasmids expressing Flag-tagged PAPD5 or PAPD7. HBV replication did not change the subcellular distributions of PAPD5/7 in the transfected cells (Fig. 7B).

DISCUSSION

The recent finding that noncanonical poly(A) polymerases PAPD5/7 are crucial to HBV mRNA stabilization sheds light on how HBV mRNA exploits a cellular RNA quality-control machinery for its biogenesis. PAPD5/7 have been reported to be involved in stabilization for a small number of host mRNA, especially those with long 3′ UTRs encoding secreted proteins. The stabilizing effect on host mRNA is very moderate and the majority of these genes have a less than 2-fold change when PAPD5/7 are silenced with siRNAs (36, 38, 47). HBV mRNAs except for the smallest HBx mRNA seem to be addictive to PAPD5/7’s stabilizing function. The insensitivity of HBx mRNA to RG7834 probably can be attributed to the limited space between the transcriptional start site (TSS) and HSLα structure. The TSS of HBx is located at nucleotide 1287 which is only 5 nucleotides apart from the HSLα position. It is possible that the formation of the 5′ Cap complex may interfere with HSLα recruiting its cognate binding protein(s), such as ZCCHC14. Alternatively, it is also possible that additional sequences upstream of HSLα are required for the function of PAPD5/7.

Although our in vitro data showed that both PAPD5/7 were able to add more than 200 adenosines onto the RNA oligonucleotides (Fig. 4), it is possible that PAPD5/7 are dispensable to de novo poly(A) tail synthesis of SHBs mRNA. As noncanonical polyadenylases, PAPD5/7 do not need poly(A) signals such as AAUAAA for the addition of adenosine. Despite the fact that HBV uses an uncommon UAUAAA sequence for polyadenylation (13), this signal did not play a role for the utilization of PAPD5/7 since replacement of the HBV UAUAAA with the SV40 poly(A) sequence, which contained canonical AAUAAA (50), did not interfere with the potency of RG7834 in tissue culture (see Fig. S5 in the supplemental material). Therefore, it is plausible that the presence of a polyadenylation signal (PAS) in HBs mRNA is to recruit host canonical polyadenylase alpha (PAPα) for poly(A) tail processing.

In order to learn if PAPD5/7 and ZCCHC14 are required for nascent poly(A) tail synthesis, we employed an adenovector that expresses dox-inducible (in tet-on manner) iLov-SHBs fusion protein ilovS (Fig. 8A) to infect cells either knocked out of PAPD5/7 or ZCCHC14. Addition of dox quickly initiated viral mRNA transcription; in the first 1 to 2 hours, the transcripts had size and expression levels comparable to those of the wild-type cells. An obvious shortening of RNA size was seen within 4 hours after the initiation of gene expression in PAPD5/7 and ZCCHC14 knockout cells, but the RNA intensity was not significantly changed (Fig. 8B and C). This result was probably because the slow RNA decaying speed could not catch up the velocity of RNA shortening, which is a phenomenon we also observed when cells were treated with RG7834 (Fig. 6C). Twenty-four hours posttranscription, the majority of viral mRNA in the knockout cell lines was not only shortened but also less accumulated compared with that in the wild-type Huh7.5.1 cells (Fig. 8B). This finding indicates that in the absence of PAPD5/7, or when PAPD5/7’s partner ZCCHC14 is not available, the cell can produce a regular level of viral transcripts that have the normal length of a poly(A) tail. However, these transcripts are not stable (Fig. 8 and Fig. S4 in the supplemental material).

FIG 8.

PAPD5/7 were not required for the de novo synthesis of viral mRNA poly(A) tail. (A) Plasmid map of adenovector Ad3G-ilovS. Fluorescent protein iLov (111 amino acid [aa]; accession no. QCO69681) was fused with SHBs, and the expression cassette was placed under tetracycline regulatory promoter (tet-on manner). Sequence numbering of the HBV genome was based on gene annotation of variant U95551. Underlined hexamer TATAAA depicted a polyadenylation signal in the HBV genome. Cassette harboring a 3rd generation tetracycline transactivator (tTA-3G) was placed downstream of the ilovS gene. (B) Huh7.5.1 and derived PAPD5/7 and ZCCHC14 knockout cell lines were infected with Ad3G-ilovS. Two days postadenoinfection, doxycycline (dox) was added into the culture media to induce ilovS expression. RNA was harvested at indicated time points and detected with a P³²-labeled HBV probe. Numbers beneath the loading gel were ImageJ quantifications of ilovS mRNA. The signal at time 0 of Huh7.5.1 cells was set as the background. (C) Selected RNA samples at time points 1 and 8 hours from B were loaded onto a Northern gel with adjusted amounts. The numbers beneath the loading gel represented proportions of RNA amount that were used in B. A representative blot of three duplicates was shown. (D) Proposed model for PAPD5/7-mediated HBV RNA stabilization. HBV mRNA is transcribed by host RNA polymerase II (RNAPII). After the first 20 to 30 nts are synthesized, the 5′ end of the transcripts is 7-methyl-guanosine capped and protected by the cap binding protein complex (black circle). The transcription of a hexanucleotide UAUAAA polyadenylation signal (PAS; green rectangle) at the 3′ UTR recruits canonical polyadenylating polymerase alpha (PAPα). The RNA sequence 20 nts downstream of the PAS will be endolytically cleaved (read arrowhead) followed by PAPα-mediated polyadenylation. During the extending of poly(A) tail, noncanonical PAPs PAPD5/7 are recruited by ZCCHC14 to incorporate guanosine (red G) into the poly(A) tail for RNA stabilization. However, it is also possible that PAPD5/7 can be retained on the viral mRNA that has a finished tail to protect it from being shortened by exoribonucleases (Exo). RG7834 blocks guanosine incorporation mediated by PAPD5/7 and consequently destabilizes viral RNA.

As the major noncoding RNA modification enzyme, substrates of PAPD5, such as snoRNA, pre-miRNA, telomerase RNA component (TERC) RNA, and VRNA2.7 from human cytomegalovirus (HCMV), are reported to have highly structured stem-loops near their 3′ ends (27, 42, 51, 52). Unfortunately, structural features at the 3′ UTR of host-polyadenylated mRNA stabilized by PAPD5/7 have not been defined. In hepatitis A virus (HAV)-infected cells, PAPD5/7’s polyadenylating function is correlated with an elevated translational efficiency of viral RNA. Nevertheless, the characteristics of the HAV RNA structure important for mediating PAPD5/7’s proviral activity are unknown (53). HBV mRNA is the only coding RNA so far that has a known stem-loop structure associated with PAPD5/7’s tailing function. We found that mutations introduced into the HSLα pentaloop indeed rendered the SHBs mRNA less sensitive to polyadenylase inhibitor RG7834 (33). Intriguingly, when we knocked down PAPD5/7, we unexpectedly found that RNA levels of the HSLα mutant were reduced significantly (60% decrease) compared with their siRNA controls (Fig. 5B, lanes 9 and 10 versus 7 and 8). This finding may imply that the stability of the HSLα mutant relies on PAPD5/7 in an enzymatic activity-independent manner since the mutant is not responsive to RG7834. The question of how the viral RNA transcribed from the HSLα mutant that can no longer interact with ZCCHC14 is related with PAPD5/7 warrants future study.

Our data, therefore, suggest that when RG7834 is applied to a tissue culture that has HBV replication, the compound mainly affects the half-life of mature HBV RNA that has a regular length of the poly(A) sequence. The compound first causes a shortening of viral RNA poly(A) followed by an increased degradation. We hypothesize that HBV RNA, once it has its polyadenylation signal (PAS) transcribed by RNAPII, will employ canonical polyadenylase alpha (PAPα) to start poly(A) synthesis. RNA-binding protein ZCCHC14, associated with HSLα, recruits PAPD5/7 to the growing or completed poly(A) tail to incorporate guanosine into the RNA terminal and consequently shields the 3′ end from being trimmed by exonucleases. However, in the absence of PAPD5/7, or when PAPD5/7 are inhibited by RG7834, viral mRNA with a regular tail length can nevertheless be formed but cannot be stably maintained (Fig. 8D).

It should be pointed out that although RG7834-induced HBV mRNA degradation could take place in the both the nucleus and cytoplasm, RNA turnover in the cytoplasm might be a limited contributor to overall HBV mRNA reduction. Although nuclear HBV mRNA comprises less than 10% of the total viral RNA, mRNA degradation induced by RG7834 in the nucleus might be sufficient to prevent cytoplasmic mRNA from accumulating. Since ZCCHC14 is a cytoplasmic protein (34, 42), whether the same accessory factors from the cytoplasm are employed to interact with PAPD5/7 in the nucleus for viral mRNA protection is unclear. The relative contribution of RG7834-induced HBV mRNA decay in different cellular compartments should be hinged on the balance between the abundance of viral RNA and availability of PAPD5/7 in their corresponding cellular locations.

In terms of subcellular localization of PAPD5/7 in HepG2 cells, we demonstrated that PAPD5 was predominantly localized in the nucleus, whereas PAPD7 was distributed throughout the cell with a moderate enrichment in the cytoplasm (Fig. 7A). This observation, although obtained via an overexpression of tagged protein, is in accordance with the most recent report regarding endogenous PAPD5/7 subcellular localizations in HeLa cells (42). Moreover, the presence of HBV RNA does not change PAPD5/7’s cellular compartmentalization, as evidenced by both proteins having an identical staining pattern between HepG2 and HepG2-2.2.15 cells. This is different than Rift Valley fever virus (RVFV)-infected cells in which the host SKIV2L2 and ZCCHC7, two key components of the human TRAMP complex, are mobilized by viral replication to the cytoplasm. The relocated host SKIV2L2 and ZCCHC7 act as innate immune factors to bind to the 3′ UTR of RVFV mRNA and deliver the transcripts to the cytoplasmic exosome for degradation (54). In contrast, binding of the HBV mRNA with PAPD5/7 stabilizes the viral transcript in both the nucleus and cytoplasm. Whether PAPD5/7 protect HBV mRNA in separate subcellular compartments and whether they can shuttle back and forth for mutual compensation are worthy topics to further investigate.

In conclusion, dihydroquinolizinone compound RG7834 inhibits the polyadenylating function of host RNA quality-control factors PAPD5/7 and subsequently destabilizes HBV SHBs mRNA. Suppression of the tailing activity of both PAPD5/7 shortens the poly(A) length of the matured SHBs mRNA in both the nucleus and cytoplasm where PAPD5 and PAPD7 are preferentially localized, respectively. RG7834 seemed to be able to mediate whole cellular destabilization of SHBs mRNA and, therefore, is expected to reduce the bulk of the surface protein product and may provide a novel modality to break immune tolerance in CHB patients. Further information about host pathways regulating the interaction between PAPD5/7 and HBV mRNA and the nature of RG7834 selectivity for viral transcripts will be necessary to support full clinical development.

MATERIALS AND METHODS

Cells, plasmid, adenovirus, and siRNA.

HepG2 (catalog [cat.] no. HB8065; ATCC) and HEK293 (cat. no. CRL-1573; ATCC) cells were cultured in Dulbecco’s modified Eagle medium (DMEM)/F12 containing 10% fetal bovine serum and 100 U/ml penicillin plus 100 μg/ml streptomycin. The HepG2-tTA25 stable cell line was produced by stable transfection of HepG2 cells with a CMV-immediate-early (IE)-driven cassette of tetracycline transactivator (tTA) for tet-off regulation of gene expression as described previously (33). The cell line was cultured in medium identical to the one for HepG2 cells. The dihydroquinolizinone compound RG7834 was provided by Arbutus Biopharma.

The Huh7.5.1 parental and subsequent knockout cell lines were generous gifts from the Chan Zuckerberg Biohub (53). The PAPD5/7 double-knockout cell line D5/7-DKO and ZCCHC14 knockout cell line ZC14-KO were generated by transfecting Huh7.5.1 cells (harboring a CRISPR-Cas9-expressing cassette) with plasmid pX458 that contained target genomic RNA (gRNA) (Addgene no. 48138; gift from Feng Zhang). Two days posttransfection, GFP-positive cells were single-cell sorted into 96-well plates using a Sony SH800 cell sorter. Genomic DNA was later isolated from obtained clones, and the gRNA-targeted sites were PCR amplified and sequenced.

Adenoviral vectors containing an HBV gene (accession no. U95551) were constructed using the ViraPower system (cat. no. K4930; Invitrogen). Briefly, HBV-expressing cassettes of pgRNA, LHBs mRNA, SHBs mRNA, and HBx mRNA under the drive of either a tet-CMV promoter or CMV-IE promoter, were cloned into the pENTR4 plasmid, sequencing verified, and recombined with the pAd-DEST vector. Adenoviruses were packaged in 293A cells, and the viral titer was determined with qPCR. Infection of HepG2 and HepG2-tTA25 cells with adenovirus was carried out at a multiplicity of infection (MOI) of 500 and lasted for 12 hours.

Plasmids expressing Flag-tagged PAPD5 and PAPD7 were kindly provided by Narry Kim (36). Small interfering RNAs (siRNAs) against host genes, in the format of SmartPool, were purchased from Horizon (cat. no. E-010011-00, L-009807-00, and L-014086-01 for PAPD5, PAPD7, and ZCCHC14, respectively). For temporarily knocking down cellular genes, 20 nM siRNAs were transfected into cells with RNAiMax following instructions from the manufacturer (cat. no. 13778150; Thermofisher).

Nucleic acid detection.

Total RNA was extracted with TRIzol reagents (cat. no. 15596018; Thermofisher) and was resuspended in diethyl pyrocarbonate (DEPC) water. For RNA extraction from subcellular compartments, cells were lysed in ice-cold PARIS fractionation buffer at 1 × 10⁷ cells/ml (cat. no. AM1921; Thermofisher). Cell lysates were next transferred into Eppendorf tubes and centrifuged at 4°C to separate nuclei from cytosol, followed by TRIzol-LS (cat. no. 10296028; Thermofisher) extraction of RNA from each fraction. Purified RNA was electrophoresed in formaldehyde agarose gel and blotted onto a Hybond-N+ membrane (cat. no. 45-000-850; GE HealthCare), followed by hybridization in Ekono buffer (cat. no. 82021-338; VWR) containing a P³²-labeled HBV riboprobe as described previously (55). Hybridization signals were detected with a Typhoon Phosphorimager scanner. Densitometry signals from Northern blots were quantified with ImageJ software (https://imagej.nih.gov/ij/).

To define SHBs mRNA poly(A) tail length, the viral RNA 3′ end was ligated with the RNA adaptor Toll-like receptor 1 (TLR1). The ligated viral transcript was reverse transcribed with primer Pr1 located within the RNA adaptor, followed by PCR amplification with forward HBV specific primer Pf1. The length of SHBs mRNA upstream of the poly(A) site was determined with primers Pf2 and Pr2 (sequence information is listed in Table S1 in the supplemental material).

The quantification of viral and host gene expression was performed using the Luna one-step RT-qPCR kit (cat. no. E3005L; New England BioLabs [NEB]). Forward and reverse primers for HBV, PAPD5, PAPD7, ZCCHC14, GAPDH, tTA, and β-actin genes are listed in Table S1.

RNA-protein (RNP) complex immune precipitation and in vitro tailing assay.

HEK293 cells were transfected with plasmids expressing Flag-tagged PAPD5 or PAPD7. Two days posttransfection, cells were washed with cold phosphate-buffered saline (PBS) and lysed with immune precipitating buffer (cat. no. 8903S; Cell Signaling Technology [CST]). The cell lysate was clarified with a brief centrifugation and incubated with anti-Flag beads (cat. no. A2220; Sigma) for 3 hours at 4°C. Precipitated RNP complexes were washed 3 times with a buffer containing 10 mM Tris-HCl (pH 7.4), 500 mM NaCl, and 0.05% NP-40. Precipitated PAPD5 and PAPD7 were resuspended in a buffer containing 10 mM Tris-HCl (pH 8.0), 100 mM KCl, 3.2 mM MgCl₂, and 1 mM ATP. P³²-labeled RNA oligonucleotides were used as the substrate for 3′ tailing (36). In vitro RNA elongation was carried out at 37°C for 20 minutes and 3 hours for the PAPD5 and PAPD7 assays, respectively. The in vitro tailing reaction was terminated by the addition of 2× RNA loading buffer (cat. no. B0363S; NEB). A fraction of elongated product was resolved on a 6% or 15% urea-Tris-borate-EDTA (TBE) gel (cat. no. EC6865BOX; Thermofisher), and the radioactive signals were scanned with a Typhoon Phosphorimager.

Immune fluorescent staining.

HepG2- and HBV-producing HepG2-2.2.15 cells were transfected with plasmids expressing Flag-tagged PAPD5 or PAPD7. Two days posttransfection, cells were reseeded onto coverslips and cultured for an additional 2 days. The cells were then fixed in 5% formaldehyde and rinsed in PBS, followed by a 1-h treatment of 0.5% Triton X-100. Fixed cells were incubated with 1:500 anti-Flag antibody (cat. no. 14739; CST) followed by the detection with goat anti-rabbit antibody conjugated with AlexaFluro488 (cat. no. ab150077; abcam).