MCPIP1 reduces HBV RNA
First we determined the correlation between MCPIP expression and HBV infection in the HBV-infected liver, by mining three public datasets. First, we found that the expression of MCPIP1, MCPIP2, and MCPIP4, but not that of MCPIP3, was increased in the liver tissue taken from patients with chronic HBV infection (n = 122), compared to healthy patients (n = 6) (Fig. S1A), based on the expression profiling of chronic hepatitis B (CHB) liver (GSE83148)9. Second, using the GSE52752 dataset, we found that MCPIP1 and MCPIP4 expression was significantly elevated (2.8 and 1.2 fold) in the livers collected from HBV-infected human liver-chimeric mice (8 weeks after infection) (n = 9), compared to mice in the control group (n = 6) (Fig. S1B). Third, MCPIP1 and MCPIP2 mRNA were increased in the primary human hepatocytes (PHHs) infected with HBV (GSE69590, Fig. S1C)10.
To further investigate the role of MCPIP family in HBV infection, the HBV pgRNA reporter system was adopted11,12. In this system, a reporter gene nano-luciferase (NL) is inserted in the core region of HBV genome, and its activity reflects the HBV RNA level in the transfected cells. MCPIP expression vectors, HBV pgRNA reporter plasmid (pCMV1.2xHBV/NL), a helper plasmid (pcDNA-CP), and pCAG-SEAP (to monitor transfection efficiency) were cotransfected into Huh7 cells, followed by western blot and luciferase assay. We found that expression of MCPIP1, but not MCPIP4, significantly decreased the NL reporter activity (Fig. 1A). MCPIP2 and MCPIP3 were poorly expressed in this experimental condition (Fig. S2). In addition, hydrodynamic injection, an in vivo model to reproduce HBV replication in mice, was utilized13. C57BL/6 mice were injected with the pgRNA reporter and the helper plasmid pcDNA-CPtds, along with the MCPIP1 or mock vector. After 2 days, liver samples were harvested and the NL activity was determined. The NL reporter activity was lower in the livers injected with MCPIP1 vector, compared to the mock-injected livers (Fig. 1B,C).
Further to verify whether endogenous MCPIP1 contributes to viral restriction, MCPIP1 was knocked out by CRISPR/Cas9-mediated genome editing in Hep38.7-Tet cells (Fig. 1D), in which HBV replication starts from the chromosomally integrated HBV genome after the removal of tetracycline14. When we depleted tetracycline for 5 days, we found that the levels of HBV RNA, supernatant viral DNA, and HBsAg/HBeAg were higher in the knockout cells, compared with the parental (Fig. 1E–I). When tetracycline was added back to the cells after HBV accumulation (Fig. S3A), the viral RNA decreased in a significantly slower manner in both of the MCPIP1 knockout cells relative to the parental cells (Fig. S3B). Taken together, these results suggested that endogenous MCPIP1 also decreased HBV RNA level.
Domains responsible for RNase activity and oligomerization of MCPIP1 are required for its anti-HBV activity
The MCPIP1 protein has multiple domains (Fig. 2A)2, including an UBA domain (43–89) important for deubiquitination3; a NYN domain (133–270) and a CCCH-type zinc-finger domain (305–325), which are critical for its RNase activity; and a proline-rich domain (458–536), which triggers oligomerization15. To determine the contributions of these domains to the antiviral activity, we cotransfected the HBV pgRNA reporter vector with expression vectors encoding wild type or mutant MCPIP1 (Fig. 2A). The NL activity was lower in the cells overexpressing wild type MCPIP1, compared with that detected in the mock transfectants. In contrast, overexpression of D141N or C306R mutants, each harboring a single amino acid mutation in the RNase domain or CCCH-type zinc-finger, respectively15, did not result in decreased NL activity (Fig. 2B). Furthermore, a C-terminus truncated mutant (1–453) failed to decrease antiviral activity, which was comparable to that of the wild type (Fig. 2C). The D141N mutant is reportedly defective for both deubiquitinase (DUB) and RNase activity16. We further examined the anti-HBV activity of the C157A mutant, defective for DUB but not for RNase activity16, to determine the contribution of each activity in detail (Fig. 2D). We found that it lowered NL activity to a comparable level to that of the wild type, suggesting that RNase, but not DUB activity is important for the anti-HBV activity. Taken together, these results highlighted the importance of domains responsible for RNase activity, RNA-binding, and oligomerization, for the antiviral activity of MCPIP1.
Epsilon structure of HBV RNA is required for MCPIP1-mediated reduction
As described earlier, MCPIP1 recognizes a specific secondary RNA structure, stem-loop, to degrade target mRNAs, and the UAU or UGU loop sequence is the preferential target for MCPIP15. Indeed, HBV epsilon has two UGU sequences within the loop structure (Fig. S4A). Thus, we hypothesized that MCPIP1 targeted the stem-loop structures of HBV RNA to reduce viral RNA. We constructed expression vectors of mutant pgRNA reporters lacking 5′-and/or 3′-epsilon structures (Fig. 3A). They were cotransfected with the expression vectors of GFP or GFP-tagged MCPIP1, and the NL activity was determined (Fig. 3B,C). The result revealed that the RNA levels of 5′-deleted or 3′-deleted epsilon pgRNA reporter were significantly less affected by MCPIP1 overexpression, compared to the wild type pgRNA. Deleting the other additively attenuated the effect, suggesting that MCPIP1 targets both of the epsilon structures to downregulate viral RNA.
Huh7 cells were cotransfected with the expression vector for MCPIP1 and pgRNA reporter vector to verify whether MCPIP1 binds to viral RNA. We found that FLAG-MCPIP1 immunoprecipitation enriched the viral RNA, but not the host HPRT, significantly more than FLAG-GFP, suggesting that MCPIP1 binds with HBV RNA (Fig. 3D,E). Furthermore, an in vitro cleavage assay was performed, as previously done for IL-6 mRNA and a precursor miRNA (pre-miRNA), to demonstrate that their stem-loop structures were targeted by MCPIP115,17. We found that the recombinant MCPIP1 cleaved HBV epsilon RNA in vitro (Fig. S4B). Overall, these results suggested that MCPIP1 reduced HBV RNA via cleavage of the epsilon structures.
IL-1β adopts MCPIP1 for viral RNA downregulation
MCPIP1 is reportedly involved in immune response, such as macrophage activation, regulation of the inflammatory response, and antiviral activity, by cleaving target RNA such as IL-6 mRNA and pre-miRNA2,4, and is induced by proinflammatory cytokines, such as IL-1β, TNFα, and MCP-118,19,20. We found that the expression of IL-1β was higher in the CHB livers than in the healthy controls, and HBV-infected human liver-chimeric mice than those in the control group (Fig. 4A,B). Moreover, IL-1β expression was positively correlated with MCPIP1 expression in HBV-infected patients (Fig. 4C). As we previously reported, IL-1β reduced pgRNA reporter activity in Huh7 cells21. And we found that MCPIP1 protein was also increased (Fig. 4D), and this led us to assess the contribution of MCPIP1 to the IL-1β-mediated anti-HBV activity in hepatocytes. The parental and MCPIP1 knockout Hep38.7-Tet cells were cultured in the presence of IL-1β. Treatment with IL-1β significantly reduced HBV RNA as well as culture supernatant HBV DNA in the parental cells (Fig. 4E,F). In contrast, IL-1β mediated reduction was not observed in the MCPIP1-knockout cells. For further verification, NTCP-overexpressing HepG2 cells22 were transfected with siMCPIP1, infected with HBV, and treated with IL-1β (Fig. S5A). In the absence of IL-1β, the viral RNA level was higher in the siMCPIP1 transfectants, compared to the control siRNA (Fig. S5B), as we found in the MCPIP1 knockout Hep38.7-Tet cells (Fig. 1F). Unexpectedly, IL-1β rather increased the viral RNA level in the siMCPIP1-transfected cells, while it decreased in the parental cells (Fig. S5B). Consistently, IL-1β treatment increased the viral RNA level in the MCPIP1-knocked out HepG2-NTCP cells, infected with HBV (Fig. S5C). The cellular DNA level was comparable between the IL-1β-treated and untreated cells. And when the knock-out HepG2-NTCP cells were transfected with the expression plasmid for MCPIP1, the wild type but not D141N mutant decreased the viral RNA and protein, compared to the mock transfectant (Fig. S5D-H). Taken together, these results suggest that MCPIP1 exerts an antiviral effect downstream of IL-1β.