DMV biogenesis during β-coronavirus infection requires autophagy proteins VMP1 and TMEM41B

ABSTRACT

Upon entering host cells, β-coronaviruses specifically induce generation of replication organelles (ROs) from the endoplasmic reticulum (ER) through their nonstructural protein 3 (nsp3) and nsp4 for viral genome transcription and replication. The most predominant ROs are double-membrane vesicles (DMVs). The ER-resident proteins VMP1 and TMEM41B, which form a complex to regulate autophagosome and lipid droplet (LD) formation, were recently shown to be essential for β-coronavirus infection. Here we report that VMP1 and TMEM41B contribute to DMV generation but function at different steps. TMEM41B facilitates nsp3-nsp4 interaction and ER zippering, while VMP1 is required for subsequent closing of the paired ER into DMVs. Additionally, inhibition of phosphatidylserine (PS) formation by siPTDSS1 partially reverses the DMV and LD defects in VMP1 KO cells, suggesting that appropriate PS levels also contribute to DMV formation. This work provides clues to the mechanism of how host proteins collaborate with viral proteins for endomembrane reshaping to promote viral infection.

In the past 20 years, β-coronavirus family members, which include severe acute respiratory syndrome coronavirus (SARS-CoV), middle east respiratory syndrome coronavirus (MERS-CoV) and the most recent SARS-CoV-2, have repeatedly caused infectious diseases with severe social and economic burdens worldwide. For therapeutics against the virus, much attention has been given to the development of vaccines and neutralizing antibodies against viral epitopes. However, vaccines may fail to effectively protect people against infection due to rapid viral mutations, while the clinical benefits of currently approved drugs are also limited. Thus, a better understanding of the intracellular mechanism of viral replication will facilitate the identification of novel therapeutic targets to restrict viral amplification and virus-induced host cell death.

One of the foremost changes to host cells caused by β-coronavirus infection is the drastic remodeling of the intracellular endomembrane system to accommodate viral replication. Shortly after β-coronaviruses enter into host cells, the viral genome is translated into two polyproteins, pp1a and pp1ab, which are further processed and give rise to 16 nonstructural proteins (nsps), nsp1-16. Among these proteins, nsp3 and nsp4 are necessary and sufficient to induce the formation of RO-like structures. However, the underlying mechanism of the process, and how host proteins are involved, remain poorly understood.

Whole-genome CRISPR-Cas9 screens identified TMEM41B, and VMP1 to a lesser extent, as host factors essential for SARS-CoV-2 infection. Both VMP1 and TMEM41B are ER-integral multi-transmembrane proteins, and they function as a complex to regulate autophagosome and LD formation. In our recent work [1], we first verified that β-coronavirus infection is almost completely blocked in VMP1 or TMEM41B knockout (KO) cells using the mouse hepatitis virus (MHV) model. VMP1 KO cells proliferate much more slowly and are more vulnerable to stresses, compared to control and TMEM41B KO cells, which may explain the absence of VMP1 from the other genome-wide screens. Normal levels of MHV viral RNAs are detected at early infection stages in both KO cells, and viral particles are occasionally found in the endosomes from transmission electron microscopy (TEM) images of VMP1 KO cells. This suggests that VMP1 and TMEM41B function at a later step after virus entry. Given that DMVs and autophagosomes are both ER-derived double-membrane structures, we speculated that VMP1 and TMEM41B may also be involved in DMV formation. Indeed, DMVs are not observed in MHV-infected VMP1 or TMEM41B KO cells by TEM analysis.

In order to explore the specific roles of VMP1 and TMEM41B in DMV biogenesis, we exploited a simple system to generate DMV-like structures through exogenous expression of SARS-CoV-2 viral proteins. Coexpression of nsp3 and nsp4 induce formation of punctate structures corresponding to numerous DMVs in TEM images, and the two proteins interact with each other through their lumenal domains. Interestingly, in fluorescence protein protease (FPP) assays, the nsp3 signal rapidly disappears upon proteinase K treatment in plasma membrane (PM)-permeabilized cells, while nsp4 largely persists. This indicates that nsp3 and nsp4 mainly localize to the outer and inner membranes of DMVs, respectively. To examine if this unique distribution is required for DMV formation, we artificially linked nsp3 and nsp4 using a streptavidin (strep) and streptavidin-binding peptide (SBP) heterodimerization system. Ectopic expression of GFP-nsp3-SBP and Strep-nsp4-mCherry fail to induce nsp3⁺ nsp4⁺ foci, but separation of nsp3 and nsp4 by addition of biotin to disrupt strep-SBP binding results in formation of nsp3⁺ nsp4⁺ puncta. This distinct arrangement of nsp3 and nsp4 may facilitate the generation of different degrees of membrane curvature on each side of the DMV, which may also explain why cleavage of the nsp3-nsp4 polyprotein is a prerequisite for DMV formation.

Using the above model, we dissected the role of VMP1 and TMEM41B in DMV biogenesis. Both proteins are weakly recruited to the DMV formation sites, and VMP1 shows a stronger binding to the nsp3-nsp4 complex than TMEM41B. However, VMP1 deficiency does not affect nsp3-nsp4 complex formation, while TMEM41B is required for efficient nsp3-nsp4 interaction and hence DMV induction. Ultrastructural analysis demonstrated that compared to control cells, the DMV numbers are greatly reduced in VMP1 or TMEM41B KO cells. An intriguing finding regarding the VMP1 KO cells is accumulation of concentric membrane structures, suggesting that the closure of DMVs may be defective. Consistent with this idea, nsp4 puncta are digested immediately upon proteinase K addition, similar to nsp3 puncta, in VMP1 KO cells. This argues that DMVs are not closed in the absence of VMP1. Similar abnormal membrane structures were previously observed in cells expressing uncleaved nsp3-nsp4 polyprotein. These observations suggest that VMP1 may promote membrane curvature generation possibly through regulating the segregation of nsp3 and nsp4 to opposite sides of the ER membrane at DMV formation sites. Taken together, our data suggest that VMP1 and TMEM41B play distinct roles in DMV formation.

VMP1 was reported to show phosphatidylserine (PS) scramblase activity using in vitro assays. Our genome-wide RNAi screen in C. elegans revealed that knockdown of pssy-1, the worm PS synthetase, rescued the enlarged LDs in epg-3 (the worm homolog of VMP1) mutants. Mammals have two PSSY-1 homologs, PTDSS1 and PTDSS2, which are responsible for converting phosphatidylcholine (PC) and phosphatidylethanolamine (PE) into PS, respectively. Depletion of PTDSS1, but not PTDSS2, partially rescues both the DMV and LD defects in VMP1 KO cells. Thus, these results suggest that excessive PS in VMP1 KO cells may contribute to the defective DMV and LD formation.

In conclusion, our study reveals that macroautophagy/autophagy proteins VMP1 and TMEM41B are required for DMV formation during β-coronavirus infection, and they function at different steps. TMEM41B facilitates ER zippering by promoting the interaction of nsp3 and nsp4. VMP1 is involved in a later step to generate sufficient membrane curvature for the closure of the paired ER into sealed DMVs. Previous work reported that the phosphatidylinositol 3-kinase complex, but not ATG5 or ATG16L1, is also involved in DMV formation. We also found that DMVs are normally induced in RB1CC1/FIP200 KO cells, suggesting that only part of the autophagy machinery is employed by β-coronaviruses. Thus, our study provides molecular insights into how the viral proteins hijack host factors involved in membrane dynamics for viral replication cycles.

Funding Statement

This work was supported by Chinese Ministry of Science and Technology of the People’s Republic of China [National Key Research and Development Program, 2021YFA1300802 to Y.G.Z. and H.D.], the National Natural Science Foundation of China (NSFC) [32170753 to Y.G.Z.], the Chinese Academy of Sciences [XDB37030205 and KJZD-SW-L05 to H.D.], and the Shenzhen–Hong Kong Institute of Brain Science–Shenzhen Fundamental Research Institutions [2022SHIBS0002 to Y.G.Z.].

Disclosure statement

No potential conflict of interest was reported by the author(s).