key: cord-0968498-pfnikcuw
authors: Yuan, Zhen; Hu, Bing; Wang, Yulei; Tan, Xuan; Xiao, Hurong; Yue, Mengzhen; Cai, Kun; Tang, Ke; Ding, Binbin
title: The E3 Ubiquitin Ligase RNF5 Facilitates SARS-CoV-2 Membrane Protein-Mediated Virion Release
date: 2021-03-01
journal: bioRxiv
DOI: 10.1101/2021.02.28.433287
sha: d6aa36c504428e3e6d303abbbd14207cc58a0fe0
doc_id: 968498
cord_uid: pfnikcuw

As enveloped virus, SARS-CoV-2 membrane protein (M) mediates viral release from cellular membranes, but the molecular mechanisms of SARS-CoV-2 virions release remain poorly understood. Here, we performed RNAi screening and identified the E3 ligase RNF5 which mediates ubiquitination of SARS-CoV-2 M at residue K15 to enhance the interaction of viral envelope (E) with M. M-E complex ensures the uniform size of viral particles for viral maturation and mediates viral release. Moreover, overexpression of M induces complete autophagy which is dependent on RNF5-mediated ubiquitin modification. M inhibits the activity of lysosome protease, and uses autolysosomes for virion release. Consequently, all these results demonstrate that RNF5 mediates ubiquitin modification of SARS-CoV-2 M to stabilize the M-E complex and induce autophagy for virion release.

In this study, we identified RNF5 as the E3 ligase of SARS-CoV-2 M by using RNAi 84 screening. A mechanistic study demonstrated that RNF5 regulates virion release 85 by enhancing the interaction of M with E. Furthermore, we showed that M induces 86 autophagy which is dependent on RNF5-mediated ubiquitin modification. We also 87 found that M inhibits the activity of lysosome protease to block the degradation of 88 autolysosomes, and uses autolysosomes for egress. All in all, our findings draw out 89 the formation and regulation mechanisms of SARS-CoV-2 VLPs and provide 90 molecular details of SARS-CoV-2 virion release. We identify RNF5 as the E3 ligase 91 for ubiquitination of M which will be helpful in the development of novel therapeutic 92 approaches. 95 We began investigating the mechanisms of SARS-CoV-2 assembly and release by 96 using VLPs, as VLPs systems had been proved to be useful tools for studying the 97 viral assembly and release processes of many enveloped viruses. We first 98 transiently expressed M alone, M/E and M/N of SARS-CoV-2, and culture medium 99 were collected and subjected to ultracentrifugation to pellet VLPs. We found that 100 only M/E co-expression resulted in a significant VLPs formation, not M alone (Fig.   101   1A) . To further confirm that the pellet VLPs were indeed the membrane-bound 102 VLPs, we then treated VLPs with trypsin or/and Triton X-100. No significant 103 digestion of M was observed either in trypsin or Triton X-100, in contrast, under 104 trypsin plus Triton X-100 treatment, M was completely degraded (Fig. 1B) . These 105 data suggested that SARS-CoV-2 E is required for M mediated VLPs release. Then 106 we used this convenient assay to investigate the mechanism(s) of SARS-CoV-2 107 release. To identify host factors essential for virion release, we performed a small 108 scale of RNAi screening targeting candidates which are on the list from SARS-CoV-109 2 M IP/MS (34), and found that knockdown of RNF5 (Ring Finger Protein 5), an ER-110 localized E3 ligase, significantly reduced virion release (Fig. S1A) . We confirmed 111 the interaction of M with RNF5 via co-IP: endogenous RNF5 coimmunoprecipitated 112 with M (Fig. 1C) . RNF5 colocalized well with M in cytoplasm (Fig. 1D) . TMD of 113 RNF5 and CTD of M were critical for M-RNF5 interaction ( Fig. S1B and S1C) . Next, 114 we confirmed that knockdown of RNF5 did decrease the virion release (Fig. 1E) . 115 Over-expression of RNF5 significantly increased VLPs release while its catalytic 116 dead mutant RNF5-C42S reduced VLPs release may be due to the domain 117 negative effect (Fig. 1F) , and RNF5-C42S still interacted with M ( Fig. S1D) , 118 suggesting that RNF5 promotes virion release which is dependent on its E3 ligase 119 activity. Furthermore, to exclude potential off-target effects of siRNA, we performed 120 rescue experiments in rnf5 KO cells and found that wild-type RNF5, but not C42S 121 rescued the reduction of VLPs release (Fig. 1G) . Next, we asked whether RNF5 122 could modulate the maturation and release of SARS-CoV-2. We infected wild type 123 and RNF5 KD Vero cells with SARS-CoV-2 and performed plaque assay. The 124 extracellular viral production was lower in RNF5 KD cells than wild type cells (Fig. 125 1H and S1E). To exclude the possibility of infectibility defect in RNF5 KD cells, we 126 evaluated the viral gene expression in extracellular and intracellular via real-time 127 PCR and found that intracellular SARS-CoV-2 ORF1ab gene expression was 128 slightly decreased while extracellular ORF1ab expression was significantly 129 decreased in RNF5 KD cells than wild type cells (Fig. 1I) showed several peaks, suggesting M alone fails to efficiently release as complete 139 VLPs, and the size of VLPs from M-E co-expression were uniform with average 140 diameter 144 nm ( Fig. 2A) , suggesting that E interacts with M to ensure the uniform 141 size of viral particles. The kinetics of E expression parallels the increasement of M-142 mediated VLPs formation (Fig. 2B) . M self-interaction was required for VLPs 143 formation, as mutant M△CTD failed to interact with M ( Fig. 2C ) and lost the ability to 144 release as VLPs (Fig. 2D ). E expression parallels the increasement of M self-145 interaction in co-immunoprecipitation assay (Fig. 2E) . In the presence of the (Fig. 2G) . C-terminal of M was required for its 151 interaction with E, as mutant M △CTD failed to bind to E (Fig. 2G ) and release as 152 VLPs (Fig. 2D) , and CTD of M alone was sufficient to interact with E ( Fig. 2H) , 153 suggesting that M interacts with E via its CTD and this interaction is critical for VLPs impact on M self-interaction, M-E interaction and VLPs release (Fig. 2C, 2G and  2I ). Remarkably, M△CTD, TMD deleted mutant (△20-100aa) and CTD plus NTD 158 deleted mutant (20-100aa) completely lost their ability to form VLPs (Fig. 2D) . Thus, 159 M uses its CTD for M-M and M-E interaction which are critical for VLPs release. 160 Previous study showed that SARS-CoV E forms ion channel on membrane via 161 homo-oligomerization (35). Thus, we sought to determine whether SARS-CoV-2 E 162 could exist homo-oligomerization and this oligomerization plays an important role 163 in VLPs release. We generated several truncations and found that E△CTD showed 164 less self-interaction in a co-IP assay (Fig. 2J) . To our surprise, E△CTD still promoted 165 VLPs release (Fig. 2K) , suggesting that E homo-oligomerization is not required for 166 VLPs release. Remarkably, E△NTD mutant failed to interact with M ( Fig. 2L ) and lost 167 its ability to promote VLPs release (Fig. 2K) . Furthermore, we found that E△NTD 168 failed to enhance the self-interaction of M (Fig. 2M) . Thus, E-M interaction, not E 169 homo-oligomerization is essential for VLPs release. Taken together, these data (Fig. 4I) . These data suggested that RNF5 ubiquitinates M at the residue 213 of K15 to enhance the interaction of M and E. 214 We had shown that: 1) M only used its CTD for self-interaction and M-E interaction; to VLPs assay. We found that Monensin treatment slightly decreased VLPs release 240 and Kifunensine or Brefeldin A treatment had no effect on VLPs release (Fig. 5A) , 241 suggesting that virion release is not dependent on ERAD and ER-Golgi trafficking. (RFP + GFP -) (Fig. 5E) . We further found that M targeted red LC3-positive dots ( Fig.   254 5E), suggesting that M targets autolysosomes (red LC3 positive dots). 255 Next, we sought to determine whether M induces autophagy for virion release. For 256 this purpose, we gradually increased the expression of M. To our surprise, we found 257 that LC3-II levels were notably increased in M-expressed cells, whereas we did not 258 observe significant degradation of p62 (Fig. 5F) . Under CHX treatment, protein 259 translations were inhibited and we then observed the significant increased LC3-II 260 level and degradation of p62 in M-expressed cells (Fig. 5F) , suggesting that M 261 induces autophagy and autophagic degradation was inhibited. Furthermore, 262 ubiquitination defect mutant K15R failed to induce autophagy (Fig. 5G) . Thus, these 263 results indicated that M expression induces autophagy which is dependent on 264 RNF5-mediated ubiquitin modification.

Next, we sought to determine why M can use autolysosomes for release, not for 266 degradation. We did not observe any change of lysosome pH (Fig. 5H) . Galectin3 267 is specifically localized on damaged endosomes or lysosomes (38). Similar with 268 control cells, we found that GFP-Gal was also diffusely localized in M expressed 269 cells (Fig. 5I) , suggesting that M does not cause lysosome damage. We further 270 used LAMP2 to track lysosomes and found that the number of LAMP2 positive 271 puncta in M-expressed cells was slightly decreased compared with control cells 272 (Fig. 5J) . Furthermore, we observed the significant decrease in maturated 273 Cathepsin D level in M-expressed cells compared to control cells (Fig. 5K) , 274 suggesting that M expression inhibits the degraded activity of lysosome by 

A novel coronavirus outbreak of global health concern

A Novel Coronavirus from Patients with Pneumonia in China

Coronavirus M proteins accumulate in the Golgi complex beyond the site of virion 471 budding

The intracellular sites of early replication and budding of SARS-coronavirus

Assembly of the coronavirus envelope: homotypic interactions 475 between the M proteins

Spike protein assembly into the coronavirion: exploring 477 the limits of its sequence requirements

The M, E, and N structural proteins of the severe acute respiratory syndrome coronavirus are 479 required for efficient assembly, trafficking, and release of virus-like particles

Late assembly motifs of human T-cell leukemia 482 virus type 1 and their relative roles in particle release

Functional replacement of a retroviral late 484 domain by ubiquitin fusion

Ubiquitin-dependent virus particle budding 486 without viral protein ubiquitination

Role of the human T-cell leukemia virus 489 type 1 PTAP motif in Gag targeting and particle release

A leucine residue in the C terminus of human parainfluenza virus type 3 matrix protein is 491 essential for efficient virus-like particle and virion release

Role of ubiquitin in parainfluenza virus 5 particle 493 formation

Evidence for a new viral late-domain core 495 sequence, FPIV, necessary for budding of a paramyxovirus

Ubiquitin-regulated nuclear-cytoplasmic trafficking of the Nipah virus matrix protein is 497 important for viral budding

Cell-specific inhibition of paramyxovirus maturation by proteasome inhibitors

Measles virus M and F proteins 501 associate with detergent-resistant membrane fractions and promote formation of virus-like particles. The 502

Structure-based in silico identification of ubiquitin-binding domains provides insights 504 into the ALIX-V:ubiquitin complex and retrovirus budding

The Host E3-Ubiquitin Ligase TRIM6 Ubiquitinates the Ebola Virus VP35 Protein and 506 Promotes Virus Replication

Interferon-stimulated TRIM69 interrupts dengue virus replication by ubiquitinating viral 508 nonstructural protein 3

TRIM26 negatively regulates interferon-β production and 510 antiviral response through polyubiquitination and degradation of nuclear IRF3

MARCH8 Ubiquitinates the Hepatitis C Virus Nonstructural 2 Protein and Mediates Viral 513 Envelopment

Ring finger protein 121 is a potent regulator of adeno-associated viral genome 515 transcription

E3 ubiquitin ligase RNF128 promotes innate antiviral immunity through K63-linked 517 ubiquitination of TBK1

RNF90 negatively regulates cellular antiviral responses by targeting MITA for degradation

RING finger protein 11 targets 521 TBK1/IKKi kinases to inhibit antiviral signaling

Inactivation of the Caenorhabditis 523 elegans RNF-5 E3 ligase promotes IRE-1-independent ER functions

Lysates and corresponding purified VLPs were analyzed via WB. (G) rnf5 KO 600 HEK293T cells were transfected with M-Flag, E-HA and RNF5-HA or C42S-HA for 601 36 h. Lysates and corresponding purified VLPs were analyzed via WB

WBP-1) for 24 h and 603 then mediums were collected and analyzed via Plaque Assay as described in 604 Materials and Methods. (I) The level of SARS-CoV-2 ORF1ab mRNA of intracellular 605 and extracellular from (H) were measured via real-time PCR. Error bars, mean ± 606 SD of three experiments (n = 3). Student's t test

617 E-Myc for 36 h, and subjected to HA IP and analyzed via WB. (F) HEK293T cells 618 were transfected with M-Flag and E-HA for 36 h, then cells were cross-linking by 619 treating with DSS for 30 min and then lysates were analyzed via WB. (G) HEK293T 620 cells were transfected with E-HA and M-Flag or its mutants for 36 h, and subjected 621 to Flag IP and analyzed via WB

624 TMD2 or TMD3 for 36 h, and the VLP budding assay was performed and analyzed 625 via WB. (J) HEK293T cells were transfected with E-Myc and E-HA or its mutants 626 for 36 h, and subjected to Myc IP and analyzed via WB. (K) HEK293T cells were 627 transfected with M-Flag, E-HA and mutants for 36 h. Lysates and corresponding 628 purified VLPs were analyzed via WB. (L) HEK293T cells were transfected with M-629 Flag and E-HA or its mutants for 36 h, and subjected to Flag IP and analyzed via 630 WB

Figure 3. RNF5 Promotes the Interaction of M with E for Viral Release

M-HA expression stable HEK293T cells were transfected with or without RNF5-635 HA and treated with CHX for indicated hours, and cells lysis were analyzed via WB

or mutant 637 C42S-HA for 36 h. Lysates were subjected to Flag IP and analyzed via WB

HEK293T cells were transfected with indicated plasmids for 36 h, and subjected to

D) rnf5 KO HEK293T cells were transfected with 640 indicated plasmids for 36 h, and subjected to Flag IP and analyzed via WB

KO HEK293T cells were transfected with M-Flag and E-HA with or without RNF5-642 HA for 36 h, then size of purified VLPs were analyzed via NanoSight NS300 643 (Malvern)

Ubiquitination of SARS-CoV-2 M Mediated by RNF5 is Critical for

HEK293T cells were transfected with HA-Ub and M-Flag for 36 h, and subjected 648 to Flag IP and analyzed via WB. (B) HEK293T cells were transfected with M-Flag

or mutant C42S-HA for 36 h. Lysates were subjected to Flag 650 IP and analyzed via WB

Lysates were subjected to Flag IP and 652 analyzed via WB. (D) HEK293T cells were transfected M-Flag with HA-UbK48 Only 653 or HA-UbK63 Only with or without RNF5-HA for 36 h

E) HEK293T cells were transfected M-Flag or mutants with HA-Ub. Lysates were subjected to Flag IP and analyzed via WB

HEK293T cells were transfected with HA-Ub and M-Flag or its mutant for 36 h, and 657 subjected to Flag IP and analyzed via WB. (G) Amino acid sequence of SARS-CoV-658

Lysates 659 were subjected to Flag IP and analyzed via WB. (I) HEK293T cells were transfected 660 with E-HA, M-Flag and mutants for 36 h. Lysates were subjected to Flag IP and 661 corresponding purified VLPs analyzed via WB. (J) HEK293T cells were transfected 662 with indicated plasmids for 36 h, and subjected to Flag IP and analyzed via WB

Model of RNF5 regulates M-E interaction