Residues 173–222 of the RABV P protein form an autophagy-inducing domain
Our previous report discovers that the RABV P protein induces incomplete autophagy [10], however, the domain responsible for this incomplete autophagy induction is unknown. To identify the functional domain of the P protein required for autophagy induction, we constructed six truncated P protein mutants (Fig. 1A). 293T cells were transfected with the plasmids pCMV-N-Flag-tagged P, PΔC75, PΔC125, PΔN19 (P2), PΔN52 (P3), PΔN68 (P4), and PΔN82 (P5), respectively. Western blotting assay revealed that the level of endogenous LC3‑phosphatidylethanolamine conjugate (LC3-II) was dramatically increased in cells transfected with all truncated P mutants except for those transfected with PΔC125 compared with the empty vector transfected cells, and chloroquine (CQ), a lysosomal proteolysis inhibitor, as a control for autophagic flux (Fig. 1B; P < 0.05, P < 0.001). Moreover, all truncated P proteins caused no significant increases in autophagy associated proteins autophagy related (ATG)5, ATG7, Unc-51 like autophagy activating kinase 1 (ULK1), BECN1, and autophagic degradation substrate sequestome 1 (also known as P62) levels (Fig. 1B). Consistently, confocal observation also revealed green fluorescent protein (GFP)-LC3B punctas were dramatically increased in cells transfected with all truncated P mutants except for those transfected with PΔC125 compared with the empty vector transfected cells (Fig. 1C and D; P < 0.001), however, there were no significant differences in GFP-LC3B punctas caused by all truncated P proteins after these cells were treated with CQ (Fig. S1). Therefore, these data showed that a P protein containing C-terminal residues aa 173-222 was responsible for autophagic activity.
Phosphoprotein P5 forming ring-like structures induces autophagosomes accumulation
Small phosphoprotein P5 contains residues aa 83-172, 173-222, and 223-297 of full-length P protein. Interestingly, a ring circle-like structure was observed in N2a cells transfected only with the P5 mutant but not in N2a cells either transfected individually with other P protein mutants or co-transfected with P5 and other P protein mutants (Fig. 1C, 2A and S2). Similarly, the ring circle-like structure was observed in N2a cells co‑transfected with both Flag-P5 and Myc-P5 (Fig. 2B), indicating that only P5 could form the ring-like structure. Moreover, the number of GFP-LC3B puncta autophagosomes surrounded by the P5 ring-like structure increased significantly in N2a cells cotransfected with GFP-LC3B and Flag-P5 in comparison with N2a cells cotransfected with Flag-vector and GFP-LC3B (Fig. 2C). Collectively, these data demonstrated that the P5 ring-like structure induced autophagosomes accumulation.
Incomplete autophagic vesicles are induced by P5 protein
To further investigate the relationship between autophagosomes and P5, N2a cells were cotransfected with Flag-P5 and GFP-LC3B, and labeled with LysoTracker Red. As expected, the number of GFP-LC3B puncta autophagosomes increased markedly and did not colocalize with LysoTracker Red in Flag-P5-transfected N2a cells compared with EBSS-treated N2a cells (Fig. 3A), indicating that the autophagosomes did not fuse with acidic compartments after P5 transfection. To rule out the possibility that autophagosomes fused with lysosomes but were not efficiently acidified in the transfected cells, we investigated the colocalization of GFP-LC3B with lysosomal associated membrane protein 1 (LAMP1) in Flag-P5-transfected N2a cells. GFP-LC3B puncta did not colocalize with LAMP1 in Flag-P5-transfected N2a (Fig. S3). These data suggested that autophagosomes did not efficiently fuse with lysosomes in Flag‑P5‑transfected cells.
In addition, we used adenovirus that expressed mCherry-GFP-LC3B, which was used to discriminate autophagosomes (expressing both mCherry and GFP fluorescent) from acidified autolysosomes (expressing red fluorescentonly) to determine the function of P5 in autophagosome maturation. N2a cells were transfected with Flag-P5 plasmids for 12 h, and infected with the adenovirus. In Flag vector transfected cells, few yellow puncta autophagosomes could be detected after adenovirus infection (Fig. 3B). In contrast, in Flag-P5 transfected cells, we observed the accumulation of yellow puncta autophagosomes but a low number of mcherry puncta autophagosomes, suggesting impaired autophagosome fusion with lysosomes. These results implied that P5 protein was responsible for the observed incomplete autophagic induction.
The protein P5 attaches to the BECN1 ring-like structure by interaction with BECN1
We previously demonstrated that the RABV P protein could interact with BECN1 [10]. To identify whether BECN1 binding to the P protein involves P5, N2a cells were cotransfected with Myc-BECN1 and Flag-P5 or Flag-PΔC75, Flag-PΔC125, Flag‑PΔN19, Flag-PΔN52, and Flag-PΔN68, respectively. Confocal microscopy showed that the BECN1 colocalized with the full-length P and the P mutants except for PΔC125P5 mutant, notably, the P5 formed ring-like structure had stronger localization with BECN1 ring-like structure compared with P ring-like structure (Fig. 4A, and S4). Similarly, endogenous P protein colocalized with BECN1 to form the ring-like structure in RABV infected cells, and interestingly, the ring-like structure was not observed after Becn1 gene was knocked down, suggesting that BECN1 was necessary for RABV infection to form the ring-like structure (Fig. 4B). Subsequently, a co-immunoprecipitation assay (Co-IP) was performed to further analyze whether the colocalization involves protein-protein interactions. The Co-IP data demonstrated that full-length and all truncated P proteins except PΔC125 could immunoprecipitate BECN1, and that P5 showed stronger binding to BECN1 than the other truncated P mutants (Fig. 4C). In addition, surprisingly, P5’s binding ability to BECN1 was stronger than that of the full-length protein (Fig. 4C). To identify the P protein binding domain of BECN1, Myc-tagged truncation mutants of BECN1 (1–351aa, 139–351aa, and 139–448aa) were constructed and transfected into 293T cells (Fig. 4D). Confocal microscopy analysis showed that only the 1–351aa BECN1 mutant formed the ring-like structure, and the P protein colocalized with the ring-like structure and the 1–351aa BECN1 mutant (Fig. 4E). Further co-IP experiments showed that only 1–351aa BECN1, but not 139–351aa BECN1 and 139–448aa BECN1, interacted with P protein (Fig. 4F), revealing that first 139 N-terminal residues of BECN1 are responsible for interacting with P. Collectively, these data confirmed that RABV protein P attached to the BECN1 ring-like structure by residues 173–222 of P binding to N-terminal residues 1–139 of BECN1.
P5 binding to BECN1 ring-like structure promoted RABV replication
To determine the effect on RABV replication of P5 binding to the BECN1 ring-like structure, we investigated the dynamics of RABV infection under condition of P5 overexpression. N2a cells transfected with Flag-P5 for 12 h were infected with RABV. We found that the level of viral N protein, viral N mRNA, viral anti-genomic RNA, and infectious RABV progeny were all significantly increased; however, when Becn1 was knocked down using two short interfering RNAs (siBecn1), there was a detectable downregulation of viral N protein, viral N mRNA, viral anti-genomic RNA, and infectious RABV progeny in the absence or presence of P5 (Fig. 5A-E, P < 0.05, 0.01, or 0.001), suggesting a positive role of P5 in regulating RABV infection dependent of BECN1. In addition, to further confirm whether the effect of the ring-like structure on RABV replication was dependent of autophagy induction, we also detected the level of viral N protein in presence of protein P5 together with the autophagy inhibitor 3-methyladenine (3-MA), or wortmannin treatment. The results showed 3-MA or wortmannin treatment significantly inhibited the level of viral N protein compared with that in non-treated P5 group (Fig. 5F and G, P < 0.01 or 0.001). Collectively, these data demonstrated that RABV replication hijacked BECN1 by P5 binding to the BECN1 ring-like structure.
P5 binding to the BECN1 ring-like structure regulates RABV replication via the BECN1-mediated signaling pathway
To further investigate the BECN1-dependent signaling pathway through which P5 regulates RABV replication, we examined whether BECN1, AMP-activated protein kinase (AMPK), CASP2, protein kinase B (AKT), mammalian target of rapamycin (MTOR), and mitogen activated protein kinases MAPKs [extracellular signal-regulated kinase (ERK), P38] levels changed during overexpression of P5. Western blotting analysis showed that P5 dramatically upregulated the phosphorylation (p) level of AKT, MTOR, AMPK, ERK1/2, and P38, and reduced the CASP2 level; however, it did not affect the total amount of these proteins nor BECN1 levels (Fig. 6A). Moreover, we knocked down cellular Becn1 using siBecn1 to further show whether the P5 protein affected the expression of BECN1, AMPK, CASP2, AKT, MTOR, and MAPK (ERK, P38). The results showed that there was a significant downregulation of CASP2 and p-AMPK, p-AKT, p-MTOR, and p-MAPK (ERK1/2, P38) levels, and an insignificant alteration of the total amount of these proteins in Becn1-knockdown cells with viral gene P5 transfection (Fig. 6B).
To investigate whether P5 regulated RABV replication depends on the downstream of BECN1-dependent signaling pathway, next we knocked down cellular Akt, Mtor, Ampk, Mapk respectively, and examined the NP expression levels in the absence or presence of P5. The results showed that there was a significant decrease of NP in absence or presence of P5, suggesting that the RABV replication was dependent of the AKT, MTOR, AMPK, MAPK proteins (Fig. 6C). Collectively, these data demonstrated that the BECN1 binding to P5 was responsible for regulating RABV replication via a BECN1-mediated signaling pathway.