Dual roles of CXCR4 (C-X-C motif chemokine receptor 4) in promoting entry of ebolavirus and targeting excessive glycoprotein for reticulophagic degradation to facilitate viral fitness

ABSTRACT

Ebola virus disease (EVD) caused by Zaire Ebolavirus (EBOV) infection is a major threat to public health in Africa and even worldwide, due to its extremely high mortality rate. However, there are still no effective antiviral therapies that can completely cure EVD. A comprehensive understanding of virus-host interactions would be beneficial for developing new antiviral agents. Here, we showed that CXCR4-induced macroautophagy/autophagy and was internalized to endosomes by interacting with glycoprotein (GP) on viral particles during EBOV infection; this promoted the EBOV attachment and entry, which was reduced by CXCR4 antagonist and neutralizing antibody. We also found that CXCR4 increased EBOV replication by downregulating cytotoxic GP to promote viral fitness instead of influencing the assembly of viral factory. Mechanistically, excessive EBOV GP could hijack CXCR4 sorting and transporting pathways by their interactions with HGS, one of the key components of the ESCRT machinery; subsequently GP could be carried back to the endoplasmic reticulum by CXCR4, where the E3 ubiquitin ligase RNF185 was recruited to polyubiquitinate GP in a K27- and K63-linked manner. Finally, polyubiquitinated GP was degraded in lysosomes via reticulophagy by interacting with RETREG1 (reticulophagy regulator 1), in an ATG3- and ATG5-dependent manner. Our findings revealed dual roles of CXCR4 in regulation of EBOV life cycle, either acting as an entry factor by interacting with GP on viral particles to facilitate viral entry or targeting excessive GP for reticulophagic degradation, providing new evidence that EBOV hijacked the host vesicular transportation system through efficient virus-host interactions to facilitate viral fitness.

Abbreviations: Baf A1: bafilomycin A₁; BDBV: Bundibugyo Ebolavirus; CHX: cycloheximide; CXCR4: C-X-C motif chemokine receptor 4; CLEC4M/DC-SIGNR: C type lectin domain family 4 member M; EBOV: Zaire Ebolavirus; EEA1: early endosome antigen 1; ER: endoplasmic reticulum; ERAD: ER-associated degradation; ESCRT: endosomal sorting complex required for transport; EVD: Ebolavirus disease; HAVCR1/TIM-1: hepatitis A virus cellular receptor 1; GP: glycoprotein; HGS: hepatocyte growth factor-regulated tyrosine kinase substrate; HIV: human immunodeficiency virus; IFL: internal fusion loop; ITCH/AIP4: itchy E3 ubiquitin protein ligase; LAMP: lysosomal associated membrane protein; LC-MS/MS: liquid chromatography mass spectrometry; PDIs: protein disulfide isomerases; RBD: receptor binding domain; RESTV: Reston Ebolavirus; RETREG1: reticulophagy regulator 1; RNF185: ring finger protein 185; SQSTM1/p62: sequestosome 1; SUDV: Sudan Ebolavirus; TAFV: Taï Forest Ebolavirus; TRIM21: tripartite motif containing 21; trVLPs: transcription- and replication-competent virus-like particles; Ub: ubiquitin.

Introduction

The genus of ebolavirus includes five main species, Zaire Ebolavirus (EBOV), Sudan Ebolavirus (SUDV), Taï Forest Ebolavirus (TAFV), Bundibugyo Ebolavirus (BDBV) and Reston Ebolavirus (RESTV) [1]. EBOV infection is responsible for Ebola virus disease (EVD), which is also known as Ebola hemorrhagic fever with the mortality rate of up to 66% during its outbreak in West Africa in 2013–2016 [2,3]. However, there are still no effective antiviral therapies that can completely cure EVD. Although two monoclonal antibody drugs (mAb114 and REGN-EB3) were approved by the U.S. Food & Drug Administration (FDA) and showed a significant reduction in the mortality rate of EVD patients, they could not prevent 34% of the EVD patients from death [4,5]. Thus, new and more efficient antiviral strategies are needed to be developed, which relies on further investigation of EBOV life cycle.

The genome of EBOV is negative-sense, single-stranded RNA encoding 7 viral proteins, including glycoprotein (GP), nucleoprotein (NP), viral protein (VP) 40, VP35, VP30, VP24 and large protein (L) [6]. Mature GP consists of two subunits, GP1 and GP2, presenting as a trimer of heterodimers linked by a disulfide bond, which are responsible for receptor binding and membrane fusion, respectively [7,8]. Apart from the well-accepted intracellular receptor NPC1 (NPC intracellular cholesterol transporter 1), two kinds of cell surface molecules including T-cell immunoglobulin and HAVCR1/TIM-1 (hepatitis A virus cellular receptor 1) and CLEC4M/DC-SIGNR (C type lectin domain family 4 member M) were reported to interact with GP, facilitating its initial attachment [9–11]. However, HAVCR1 was reported to be expressed in only a few kinds of cell types, which was not consistent with the broad tropism of EBOV [12]. Besides, it cannot mediate the internalization of EBOV in the non-permissive cells even in the presence of CLEC4M [13]. Thus, searching for new entry factors of EBOV is of significance for developing new antiviral therapies.

Apart from its role in viral attachment and entry, mature EBOV GP is toxic to host cells, leading to the detachment and cell death [14–16]. To balance its cytotoxicity and the ability of transmission and replication, it has evolved several mechanisms downregulating the expression of GP at both mRNA and protein levels to promote viral fitness. The transcripts of GP could be edited by a + 1 and + 2 frameshift, producing two shorter transcripts that encode soluble GP (sGP) and small soluble GP (ssGP), respectively [17,18]. Macroautophagy/autophagy plays a fundamental role in keeping cell homeostasis by degrading dysfunctional proteins and organelles [19], and different kinds of autophagy including reticulophagy and chaperone-mediated autophagy (CMA) have distinct roles during the EBOV life cycle [20,21]. Autophagy can either promote EBOV internalization during macropinocytosis, and inhibit EBOV replication and egress via RETREG1 (reticulophagy regulator 1)-mediated reticulophagy and BAG3 (BAG cochaperone 3)-mediated CMA, respectively [20–22]. Moreover, it has been reported that PDIs (protein disulfide isomerases) downregulate GP expression via the autophagy-lysosomal pathway, and RNF185 (ring finger protein 185) regulates the proteostasis of GP among the CANX (calnexin) cycle, endoplasmic reticulum (ER)-associated degradation (ERAD) and reticulophagy [23,24]. However, various host E3 ubiquitin ligases are shown to degrade GP, leading to the fact that regulation of GP proteostasis is complicated and remains to be further elucidated [24–26].

CXCR4 (C-X-C motif chemokine receptor 4) belongs to the superfamily of G protein-coupled receptor/GPCR, which is widely expressed in bone marrow, liver, lung, heart, kidney and intestine, and participates in embryo development, tissue damage repair and host immune response under physiological conditions [27–30]. Apart from binding to its natural ligand, CXCR4 is also widely known as the co-receptor of human immunodeficiency virus (HIV) by interacting with gp120, making it an important target for the development of anti-HIV therapies [31,32]. However, few reports indicate the ability of CXCR4 to mediate the attachment and entry of other viruses. Recent studies also revealed that CXCR4 can induce autophagy and maintain Epstein-Barr virus (EBV) infection [33]. However, little is known about the roles of CXCR4-induced autophagy during EBOV life cycle. Besides, after being activated by binding to its ligand, CXCR4 can be internalized to endosomes, either recycling back to the membrane or entering lysosomes by the endosomal sorting complex required for transport (ESCRT) pathway for degradation [34,35]. During its sorting and degradation process, CXCR4 was reported to interact with HGS (hepatocyte growth factor-regulated tyrosine kinase substrate) and their complex can be polyubiquitinated by ITCH/AIP4 (itchy E3 ubiquitin protein ligase) [36,37]. However, little is known about whether CXCR4 could mediate the degradation of other proteins via its sorting and degradation pathway simultaneously.

Here in this study, we found that CXCR4 induced autophagy and was internalized to endosomes by interacting with EBOV GP on viral particles during EBOV infection, and then promoted the EBOV attachment and entry, which was reduced by CXCR4 antagonist and neutralizing antibody, making CXCR4 possess the potential to be an entry factor of EBOV. In addition, CXCR4 increased EBOV replication by downregulating GP levels and promoting viral fitness instead of influencing the assembly of viral factory. Mechanistically, excessive EBOV GP was found to hijack CXCR4 sorting and transporting pathways by their interactions with HGS, one of the key components of the ESCRT machinery for lysosomal degradation, which occurred in an ATG3 (autophagy related 3)- and ATG5-dependent manner. In addition, the CXCR4-mediated GP degradation was also dependent on RETREG1 anchored on the endoplasmic reticulum (ER) membrane and the soluble autophagy receptor SQSTM1/p62 (sequestosome 1) that bound to the ubiquitinated cargo for degradation in lysosomes. Finally, we confirmed that the E3 ubiquitin ligase RNF185 participated in CXCR4-induced polyubiquitination of GP in a K27- and K63-linked manner, which was however independent of the previously reported K673 residue of GP [24]. Thus, our findings indicated the dual roles of CXCR4 in facilitating EBOV entry by interacting with GP on viral particles, and mediating intracellular excessive GP degradation via CXCR4 sorting and transporting pathways as well as reticulophagy, providing new evidence that EBOV hijacked host vesicular transportation systems through efficient virus-host interactions to promote viral fitness.

Results

CXCR4 is internalized after EBOV infection

Since autophagy was important for EBOV internalization, and CXCR4, the co-receptor for human immunodeficiency virus (HIV) entry, was reported to induce autophagy to maintain Epstein-Barr virus (EBV) infection [22,33], we determined the role of CXCR4 in EBOV entry and internalization. We confirmed that CXCR4 could induce autophagy by increasing the formation of autophagosomes indicated by the number of GFP-LC3 puncta (Figure 1A). Then we wondered whether the expression level of CXCR4 was increased after EBOV infection. With the usage of a transcription- and replication-competent virus-like particles (trVLPs) system, we found that neither mRNA nor protein level of CXCR4 was observed to be changed after EBOV trVLP (10⁶ copies/mL) infection (Figure 1B-D). Instead, the subcellular localization of CXCR4 was changed from the cell membrane to cytosol (Figure 1E, S1A and S1B), suggesting that CXCR4 was internalized after EBOV infection. As HAVCR1 was known to interact with EBOV GP, we further detected whether HAVCR1 was involved in CXCR4 internalization after EBOV infection. It showed that no interactions existed between HAVCR1 and CXCR4 (Fig. S1C and S1D), suggesting that HAVCR1 was not involved in CXCR4 internalization after EBOV infection. All above indicated that EBOV induced CXCR4 internalization.

Figure 1.

CXCR4 is internalized into endosomes by interacting with EBOV GP. (A) GFP-LC3 was co-transfected with 2 μg HA-tagged CXCR4 (or vector control) into HEK 293T cells, and the formation of autophagosomes indicated by GFP-LC3 puncta was determined by confocal microscopy. (B) HEK 293T cells was pre-transfected with NP, VP35, VP30, L and HAVCR1, and were infected with EBOV trVLPs (10⁶ copies/mL) for 24 h, and dual luciferase reporter assay was carried out to evaluate the success of infection indicated by relative luciferase activity of Renilla to firefly. (C) and (D) HEK 293T cells was treated the same as (B), and levels of CXCR4 mRNA (C) and CXCR4 protein (D) 48 h and 72 h post infection were measured by RT-qPCR and western blotting (WB). The ACTB gene and TUBA/α-tubulin were used as internal controls. (E) HEK 293T cells were treated the same as (B), and the subcellular distributions of CXCR4 (red) and cell membrane (indicated by DiD staining, far-red) were determined by confocal microscopy. (F), (G) and (H) two μg flag-tagged EBOV GP (F), EBOV GP1 (G) and EBOV GP2 (H) was individually expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells. EBOV GP (F), GP1 (G) and GP2 (H) proteins were immunoprecipitated by anti-flag antibodies and their interactions with CXCR4 were analyzed by WB. (I) two μg HA-tagged CXCR4 was expressed with 2 μg either GP1 or GP1 with receptor binding domain (RBD) deletion in HEK 293T cells. GP1 and GP1ΔRBD proteins were immunoprecipitated by anti-flag antibodies and their interactions with CXCR4 were analyzed by WB. GP, GP1, GP2 and GP1ΔRBD were detected by anti-flag, CXCR4 was detected by anti-HA. TUBA/α-tubulin was used as an internal control. (J) two μg HA-tagged CXCR4 was either expressed with 2 μg flag-tagged eGFP or flag-tagged GP fused with eGFP in HEK 293T cells. Subcellular localization of eGFP (green), HA-CXCR4 (red), EBOV GP (green) were determined by confocal microscopy. (K) HEK 293T cells were treated as (J), and whole cell lysate, membrane and cytoplasmic proteins were isolated and detected by WB. (L) HEK 293T cells were transfected with either 2 μg flag-tagged eGFP or flag-tagged EBOV GP fused with eGFP plasmids. Subcellular localization of eGFP (green), EBOV GP (green) and endogenous CXCR4 (red) were determined by confocal microscopy. (M) Huh7 cells were treated as (L), and whole cell lysate, membrane and cytoplasmic proteins were isolated and detected by WB. TUBA/α-tubulin was used as an internal control. (N) flag-tagged GP fused with eGFP was expressed alone or with a growing amount of HA-tagged CXCR4 (100 ng, 400 ng and 1.6 μg) in HEK 293T cells. Subcellular localization of eGFP (green), HA-CXCR4 (red), EBOV GP (green) were determined by confocal microscopy. (O) two μg flag-tagged GP fused with eGFP was expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells. Subcellular localization of HA-CXCR4 (red), EBOV GP (green), and EEA1 (far-red) were determined by confocal microscopy. (P) two μg flag-tagged GP fused with eGFP was expressed in wild-type and CXCR4 gene knockout HEK 293T cells. Subcellular localization of CXCR4 (red), EBOV GP (green), and EEA1 (far-red) were determined by confocal microscopy. DAPI indicated the nucleus and scale bars for (A), (E), (J), (L), (N), (O) and (P): 5 μm. Data are presented as means±standard error of measurements (SEMs) of three independent experiments. Two-tailed Student’s t test was used. ****p < 0.0001, NS, non-significant.

CXCR4 is internalized into early endosomes by its interaction with EBOV GP

Internalization was the first step of EBOV infection after initial attachment between GP and its receptors or membrane molecules [9–11], we further determined whether the internalization of CXCR4 was the result of its interaction with EBOV GP. After constructing an HA-tagged CXCR4 and a Flag-tagged EBOV GP plasmids (Fig. S1C), we determined the interaction between CXCR4 and EBOV GP by co-immunoprecipitation. It was found that HA-tagged CXCR4 could be immunoprecipitated by anti-Flag antibody (Figure 1F) and vice versa (Fig. S1E), demonstrating that the interaction between CXCR4 and EBOV GP did exist. Given that the mature GP was catalyzed into functional GP1 and GP2 (Fig. S1F), we therefore detected which functional domain of GP was involved in such interaction. However, both GP1 and GP2 could individually interact with HA-tagged CXCR4 (Figure 1G,H). Since the receptor binding domain (RBD) plays a significant role in mediating the interaction between EBOV GP and receptors or membrane molecules [38], we further determined whether the RBD domain of GP was involved in its interaction with CXCR4. It showed that the interaction between GP1 and CXCR4 was significantly reduced while the RBD was deleted (Figure 1I and Fig. S1G), suggesting that CXCR4 had the potential to be an entry factor of EBOV.

Considering that the successful entry of EBOV into host cells included both attachment and internalization of the virus into endosomes [39], we further explored whether the complex of EBOV GP and CXCR4 could be internalized into endosomes after their initial interaction. We fused eGFP protein with EBOV GP at its C terminus to show its subcellular localization (Fig. S1H), and found that the subcellular location of ectopic-expressed CXCR4 were changed from membrane to cytosol by interacting with EBOV GP (Figure 1J,K, S2A and S2B). A similar phenomenon was also observed between GP and endogenous CXCR4 (Figure 1L and S2C). As the endogenous level of CXCR4 in HEK 293T cells was relatively low (Fig. S2D), Huh7 cells were chosen for investigating the internalization of endogenous CXCR4. It was found that the GP could promote internalization of CXCR4 (Figure 1M and S2E), and the internalization of the GP-CXCR4 complex presented in a CXCR4 dose-dependent manner (Figure 1N and S2F). Furthermore, we confirmed that the CXCR4-GP protein complex was internalized into the early endosomes indicated by their colocalization with the EEA1 (early endosome antigen 1) (Figure 1O, S3A and S3B), whereas knocking out of the CXCR4 gene could significantly decrease the interactions between GP and EEA1 (Figure 1P, S3C-S3E). The above results demonstrated that CXCR4 possessed the potential to be an entry factor of EBOV.

CXCR4 acts as an entry factor of EBOV and increases EBOV replication without influencing viral factory assembly

To validate the potential role of CXCR4 as an entry factor of EBOV, HEK 293T cells and HeLa cells were selected as representative cell types using for EBOV trVLPs infections (10⁶ copies/mL) and the quantification of GP and VP40 RNA could indicate the amount of virus attached to and entered host cells (Figure 2A). It was found that overexpression of CXCR4 in HEK 293T cells could promote both EBOV attachment (Figure 2B,C) and entry (Figure 2D,E), as indicated by elevated levels of GP and VP40 RNA in the CXCR4-overexpressing group. Whereas knocking out of the CXCR4 gene could significantly reduce both EBOV attachment (Figure 2F) and entry (Figure 2G) in HEK 293T cells, and similar phenomenons were also observed in HeLa cells (Figure 2H for attachment, 2I for entry, and S3F). Besides, specific CXCR4 antagonist (motixafortide) and neutralizing antibody (12G5) were chosen to further illustrate the role of CXCR4 in promoting EBOV attachment and entry. It showed that motixafortide treatment could effectively reduce the attachment (Figure 2J,K) and entry (Figure 2L,M) of EBOV to HEK 293T and HeLa cells, compared to the control group. Also, pretreatment with CXCR4 neutralizing antibody 12G5 could inhibit the attachment of EBOV to both HEK 293T (Figure 2N) and HeLa cells (Figure 2O), compared to the control group. All above results suggested that CXCR4 could act as an entry factor of EBOV.

Figure 2.

CXCR4 acts as an entry factor of EBOV. (A) schematic representation on HEK 293T and HeLa cells infected with EBOV trVLPs for attachment and entry detection by quantifying GP and/or VP40 RNA levels. (B) and (C) HEK 293T cells stably overexpressing CXCR4 and wild-type HEK 293T cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 4°C. After washed with prechilled PBS, total RNAs were isolated and reverse transcribed. Relative levels of GP RNA (B) and VP40 RNA (C) were detected by RT-qPCR. (D) and (E) HEK 293T cells stably overexpressing CXCR4 and wild-type HEK 293T cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 37°C and cells were collected by trypsin digestion and centrifugation. After washed with PBS, total RNAs were isolated and reverse transcribed. Relative levels of GP RNA (D) and VP40 RNA (E) were detected by RT-qPCR. (F) CXCR4 KO HEK 293T cells and control cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 4°C for attachment detection. (G) CXCR4 KO HEK 293T cells and control cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 37°C for entry detection. (H) CXCR4 KO HeLa cells and control cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 4°C for attachment detection. (I) CXCR4 KO HEK 293T cells and control cells were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 37°C for entry detection. (J) and (K) HEK 293T cells (J) and HeLa cells (K) were pretreated with 20 μmol/L motixafortide and were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 4°C for attachment detection. (L) and (M) HEK 293T cells (L) and HeLa cells (M) were pretreated with 20 μmol/L motixafortide and were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 37°C for entry detection. (N) and (O) HEK 293T cells (J) and HeLa cells (K) were pretreated with 15 μg/mL 12G5 and were infected with EBOV trVLPs (10⁶ copies/mL) for 2 h at 4°C for attachment detection. The ACTB gene was used as an internal control. Data are presented as means±standard error of measurements (SEMs) of three independent experiments. Two tailed Student’s t test for (B) to (E), one-way ANOVA and post hoc Tukey test for (F) to (O). **p < 0.01, ***p < 0.001, ****p < 0.0001.

Apart from attachment and entry, we further explored the role of CXCR4 in EBOV replication. It was found that ectopic expression of CXCR4 could increase the EBOV VP40 expression (Figure 3A), whereas knocking out of the CXCR4 gene could significantly decrease VP40 expression (Figure 3B), suggesting that CXCR4 May increase EBOV replication levels. Besides, we also determined the mRNA level of IL6 (interleukin 6) which was reported to be increased after CXCR4 was activated by its natural ligand [40], and found that the mRNA level of IL6 induced by CXCR4 was significantly decreased after EBOV infection (Fig. S3G and S3H), suggesting that the interactions between GP and CXCR4 during EBOV infection may decrease the antiviral cytokine levels and synergistically enhance EBOV replication. Meanwhile, we also found that CXCR4 increased EBOV replication levels but decreased the GP levels (Figure 3A), whereas knocking out of the CXCR4 gene decreased EBOV replication levels but increased the GP levels (Figure 3B). Given that GP was cytotoxic to host cells, which was not beneficial for viral replication, and several mechanisms had evolved to decrease GP levels [14–16], we further speculate that whether the role of CXCR4 in promoting EBOV replication may be due to its downregulation of cytotoxic GP. It was found that EBOV replication levels indicated by the VP40 protein level was decreased while ectopically expressed cytotoxic GP (Figure 3C, lane 1 and lane 2), which could be further inhibited by knocking out of the endogenous CXCR4 gene to alleviate its downregulation of cytotoxic GP levels (Figure 3C, lane 2 and lane 4), indicating that the role of CXCR4 in promoting EBOV replication was due to its downregulation of cytotoxic GP. Consistently, while analyzing the fluorescence intensity of GP in Figure 1N, we also found that CXCR4 could dose-dependently decrease the fluorescence intensity of GP (Figure 3D). However, no influence was observed on the transcription levels of GP by CXCR4 (Figure 3E), suggesting that the downregulation of GP by CXCR4 was at a post-transcription level.

Figure 3.

CXCR4 promotes EBOV replication without influencing viral factory assembly. (A) HEK 293T cells stably expressing CXCR4-HA or control cells were pre-transfected with NP, VP35, VP30, L and HAVCR1, and were infected with EBOV trVLPs (10⁶ copies/mL) for 48 h. Levels of VP40, GP and HA-CXCR4 were analyzed by western blotting (WB). (B) Wild-type and CXCR4 gene knockout HEK 293T cells were treated as in (A), and levels of VP40, GP and CXCR4 were analyzed by WB. (C) two μg flagged-tagged GP was transfected into CXCR4 gene knockout HEK 293T cells or control cells and both groups of cells were infected with EBOV trVLPs (10⁶ copies/mL) for 48 h. Levels of VP40, GP, CXCR4 and HA-CXCR4 were analyzed by WB. (D) two μg flag-tagged GP fused with eGFP was expressed alone or with a growing amount of HA-tagged CXCR4 (100 ng, 400 ng and 1.6 μg) in HEK 293T cells. The fluorescence intensity of GP was analyzed using image pro plus 6.0 software. (E) two μg HA-tagged CXCR4 was expressed with 2 μg flag-tagged GP (or vector control) in HEK 293T cells. Levels of CXCR4 and GP mRNA were measured by RT-qPCR. The ACTB gene was used as an internal control. (F) one μg flag-tagged NP and 1 μg MYC-tagged VP35 were expressed with 1 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells. NP was immunoprecipitated with anti-flag. Interactions between NP and VP35 were determined by WB. (G) one μg flag-tagged NP and 1 μg MYC-tagged VP24 were expressed with 1 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells. Interactions between NP and VP24 were determined by WB. CXCR4 was detected by anti-HA or anti-CXCR4, VP40 was detected by anti-VP40, GP was detected by anti-GP, NP was detected by anti-flag, and VP35 and VP24 were detected by anti-MYC. TUBA/α-tubulin was used as an internal control. Data are presented as means±standard error of measurements (SEMs) of three independent experiments. One-way ANOVA and post hoc Tukey test for (D) and two tailed Student’s t test for (E). ****p < 0.0001, NS, non-significant.

A viral factory was formed by the interactions between NP, VP35 and VP24 after internalization, which played an important role in viral replication and assembly [41]. We therefore explored the influence of CXCR4 on the formation of the viral factory. However, CXCR4 did not increase the interactions between NP and VP35 (Figure 3F), as well as NP and VP24 (Figure 3G), indicating that CXCR4 did not influence the formation and assembly of viral factory. The above results indicated that CXCR4 could promote EBOV replication by decreasing GP expression at the post-transcription level and thus increasing viral fitness.

CXCR4 mediates GP degradation by reducing its stability

Then we evaluated whether CXCR4 could decrease GP expression at the protein level, and found that CXCR4 could decrease the GP protein expression levels in a dose-dependent manner in both HEK 293T and HeLa cells (Figure 4A,B, S4A and S4B). To further verify the role of CXCR4 in decreasing GP levels, an EBOV minigenome plasmid containing GP, VP40 and VP24 open reading frames was transfected into HEK 293T cells and HeLa cells, and the levels of VP40 could be used to bring the different transfection efficiency between the two cell lines into consistency. It was found that relative level of GP to VP40 was lower in HeLa cells expressing higher levels of endogenous CXCR4 than that in HEK 293T cells (Fig. S4C), indicating that the role of CXCR4 in decreasing GP levels was also appropriate for endogenous CXCR4. Moreover, it was also found that knocking out of the CXCR4 gene could increase the GP protein levels (Figure 4C and S4D), and the higher increasing level of GP in CXCR4-knockout HeLa cells than that in CXCR4-knockout HEK 293T cells further confirmed that both exogenous and endogenous CXCR4 could downregulate the GP expression at protein levels. Considering the highly cytotoxic effect of GP to host cells [14,15], we wondered whether CXCR4-mediated downregulation of GP could alleviate its cytotoxicity and promote viral fitness. We found out that simultaneous ectopic expression of CXCR4 could increase cell viability which was decreased by GP overexpression alone in HEK 293T cells and HeLa cells (Figure 4D and S4E).

Figure 4.

CXCR4 induces EBOV GP degradation by reducing its protein stability. (A) two μg flag-tagged EBOV GP was expressed with 2 μg HA-tagged CXCR4 (or vector control) in HEK 293T and HeLa cells. Protein levels of GP and CXCR4 were analyzed by western blotting (WB). (B) two μg flag-tagged EBOV GP was expressed with a growing amount of HA-tagged CXCR4 in HEK 293T and HeLa cells. (C) two μg flag-tagged GP was expressed in CXCR4 gene knockout (KO) HEK 293T and HeLa cells. Protein levels of GP and CXCR4 were analyzed by WB. GP was detected by anti-flag, CXCR4 was detected by anti-HA in (A) and (B), and by anti-CXCR4 in (C). TUBA/α-tubulin was used as an internal control. (D) two μg flag-tagged GP was expressed with 2 μg CXCR4 (or vector control) in HEK 293T cells, and cell viability was measured by OD value at 450 nm of different groups treated with CCK-8 for 1 h, 24 h, 48 h and 72 h. (E) two μg flag-tagged GP, GP1 and GP2 were expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells. (F) two μg flag-tagged GP and its truncates were with 2 μg HA-tagged CXCR4 in HEK 293T cells. (G) two μg flag-tagged GP from various ebolaviruses was expressed with HA-tagged CXCR4 in HEK 293T cells. (H) gray value analysis of indicated bands of GP in (G) was analyzed using imageJ software (version 1.53a). (I) two μg flag-tagged GP and its mutants were expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells. (J) gray value analysis of indicated bands of GP in (I) was analyzed using imageJ software (version 1.53a). (K) and (L) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells (K) or HeLa cells (L) treated with 100 μg/mL cycloheximide (CHX) for 0 to 8 h. Gray value analysis of indicated bands of GP was analyzed using imageJ software (version 1.53a). GP and its truncates or mutants were detected by anti-flag, CXCR4 was detected by anti-HA, and TUBA/α-tubulin was used as an internal control in (E), (F), (G), (I), (K) and (L). Data are presented as means±standard error of measurements (SEMs) of three independent experiments. One-way ANOVA and post hoc Tukey test for (D), (H) and (J). *p < 0.05, **p < 0.01, ***p < 0.001.

To determine the key domain of GP in CXCR4-mediated downregulation, we first detected whether it was in GP1 or GP2. However, the result showed that full-length GP, GP1 and GP2 could all be downregulated by ectopic expression of CXCR4 (Figure 4E and S4F), indicating that more than one domain may be involved. Therefore, several truncates of GP were constructed based on GP structure (Fig. S1E), and we found out that depletion of either glycan cap or internal fusion loop (IFL) domain of GP could partially alleviate the CXCR4-mediated GP downregulation (Figure 4F and S4G). As both GP1 and GP2 were involved in the interactions between GP and CXCR4 (Figure 1G,H), and RBD domain located within GP1 was proved to interact with CXCR4 (Figure 1I), we further investigated whether IFL domain was also involved in the interaction between GP2 and CXCR4. It was found that the interaction between GP2 and CXCR4 was significantly reduced while IFL domain was deleted (Fig. S4H), suggesting that IFL domain played a role in mediating the interaction between GP2 and CXCR4, which may participate in the CXCR4-mediated GP downregulation. To further investigate whether such downregulation of GP protein was ubiquitous within different ebolaviruses, we also constructed the GP expression plasmids of Sudan Ebolavirus (SUDV), Taï Forest Ebolavirus (TAFV), Bundibugyo Ebolavirus (BDBV), and Reston Ebolavirus (RESTV), and found that CXCR4 could induce the downregulation of all these GP proteins from above ebolaviruses with different pathogenicity (Figure 4G), with the most inhibition in Zaire Ebolavirus (EBOV) (Figure 4H). Meanwhile, we observed that CXCR4 could also inhibit expression of EBOV GP with R64A and A82V mutations (Figure 4I), and the inhibition of A82V mutant was to a greater extent (Figure 4J). The results above suggested that CXCR4 could downregulate the GP expression at protein levels, which may be related to balancing the pathogenicity, transmission and viral fitness.

To further understand the CXCR4-mediated downregulation of GP, we detected the half-life of GP before and after ectopic expression of CXCR4 by blocking intracellular translation processes with cycloheximide (CHX). It showed that the half-life of GP decreased from over 8 h to nearly 2 h by CXCR4 in both HEK 293T (Figure 4K) and HeLa cells (Figure 4L), indicating that CXCR4 could induce degradation of GP by reducing its protein stability.

CXCR4 targets GP for degradation via lysosomes

To illustrate the pathway of CXCR4-mediated GP degradation, we blocked protein degradation pathways with either proteasome inhibitor (MG132) or autophagosome-lysosome fusion inhibitor (bafilomycin A₁, Baf A1). It showed that the degradation of GP mediated by CXCR4 was blocked by Baf A1, but not MG132 (Figure 5A,B), indicating that GP was degraded in lysosomes. To further verify the degradation of GP in lysosomes, we detected the subcellular localization of GP, CXCR4 and LAMP1 (lysosomal associated membrane protein 1), and found that CXCR4 could promote the colocalization of GP and LAMP1 (Figure 5C, S5A and S5B), which further confirmed that CXCR4 targeted GP for degradation via lysosomes.

Figure 5.

The CXCR4 sorting and degradation pathway is partially involved in lysosomal degradation of EBOV GP. (A) and (B) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells and HeLa cells treated with 20 μmol/L MG132 (A) or 100 nmol/L bafilomycin A₁ (baf A1) (B). Protein levels of GP and CXCR4 were analyzed by western blotting (WB). GP was detected by anti-flag, CXCR4 was detected by anti-HA. TUBA/α-tubulin was used as an internal control. (C) two μg flag-tagged GP fused with eGFP was expressed with 2 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells. Subcellular localization of HA-CXCR4 (red), EBOV GP (green), and LAMP1 (lysosomal associated membrane protein 1; far-red) were determined by confocal microscopy. Scale bars: 5 μm. (D) liquid chromatography mass spectrometry (LC-MS/MS) was used to identify proteins that interacted with EBOV GP in HEK 293T cells. Numbers of proteins interacted with flag-tagged GP (or control IgG) were indicated by Venn diagram. (E), (F) and (G) gene ontology enrichment analysis of proteins potentially interacting with flag-tagged GP according to their biological process (E), molecular function (F) and cellular component (G). (H) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in HGS knockout (KO) HEK 293T cells or control cells. (I) one μg flag-tagged GP was expressed with 1 μg HA-tagged CXCR4 and/or 1 μg MYC-tagged HGS in HEK 293T cells. (J) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 in ITCH/AIP4 KO HEK 293T cells or control cells. (K) one μg flag-tagged GP was expressed with 1 μg HA-tagged CXCR4 and/or 1 μg MYC-tagged ITCH/AIP4 in HEK 293T cells. (L) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in ITCH/AIP4 KO HEK 293T cells or control cells. (M) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 in TRIM21 KO HEK 293T cells or control cells. (N) one μg flag-tagged GP was expressed with 1 μg HA-tagged CXCR4 and/or 1 μg MYC-tagged TRIM21 in HEK 293T cells. (O) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in TRIM21 KO HEK 293T cells or control cells. Protein expression was analyzed by WB. GP was detected by anti-flag, CXCR4 was detected by anti-HA. HGS, ITCH/AIP4 and TRIM21 were detected by anti-MYC or specific antibodies. TUBA/α-tubulin was used as an internal control.

CXCR4 sorting pathway is partially responsible for GP degradation

To uncover the underlying mechanisms of CXCR4-induced EBOV GP degradation, we conducted liquid chromatography mass spectrometry (LC-MS/MS) to identify proteins that interacted with GP in HEK 293T and Huh7 cells. In total, 1781 and 1073 proteins interacting with GP potentially were identified in HEK 293T and Huh7 cells, respectively (Figure 5D and S5C), which were further classified by Gene Ontology enrichment analysis. Similar enrichment results were observed between HEK 293T (Figure 5E-G) and Huh7 cells (Fig. S5D-S5F). CXCR4 was identified in the proteins interacting with GP both in HEK 293T cells and Huh7 cells (mass spectrometry results). Besides, we noticed that several items included key words of “vesicle-mediated transport”, “ubiquitin protein ligase binding” and “autophagosome membrane”. As is known that CXCR4 could be sorted into recycling endosomes and was transported either to the membrane or to lysosomes for degradation, in which HGS and the E3 ubiquitin ligase ITCH/AIP4 were involved [35,36]. Therefore, we detected whether the CXCR4 sorting and degradation pathways were involved in CXCR4-mediated GP degradation. It was found that knocking out of the HGS gene could partially alleviate the CXCR4-mediated GP degradation (Figure 5H, S6A and S6B), whereas ectopic expression of HGS could further promote CXCR4-mediate GP degradation (Figure 5I and S6C). As HGS was a component of ESCRT [42], its involvement suggested that the CXCR4 sorting pathway played a role in mediating GP degradation. However, although knocking out of the ITCH/AIP4 gene could lead to turnover of GP levels (Figure 5J, S6D and S6E) and ectopic expression of ITCH/AIP4 could enhance CXCR4-mediated GP degradation (Figure 5K and S6F), knocking out of the ITCH/AIP4 gene could not even partially block CXCR4-mediated GP degradation (Figure 5L and S5G). Similar phenomena were observed when the gene encoding another GP-interacting E3 ubiquitin ligase, TRIM21 (tripartite motif containing 21), identified by LC-MS/MS was knocked out (Figure 5M-O, S6H-S6K). The results above indicated that the CXCR4 sorting pathway was involved in and partially responsible for GP degradation.

CXCR4 promotes GP polyubiquitination and targets GP for reticulophagic degradation

Previous studies have reported that reticulophagy participated in regulating proteostasis of GP [23,24]. And CXCR4 was proved to induce autophagy without (Figure 1A) or with the presence of GP (Figure 6A). Besides, CXCR4 could promote the colocalization of GP with endogenous LC3 (Figure 6B, S7A and S7B), indicating that CXCR4-related autophagy may be involved in GP degradation. Furthermore, we determined whether key molecules in reticulophagy were involved in CXCR4-mediated GP degradation. We first generated endogenous ATG3 gene- and ATG5 gene-knockout HEK 293T cells (Fig. S7C-S7F) and found that knocking out of the either ATG3 or ATG5 gene could partially alleviate CXCR4-mediated GP degradation (Figure 6C,D, S7G and S7H). Moreover, the genes for two autophagic receptors anchored either in the inner membrane of autophagosomes (SQSTM1/p62) or on the surface of the endoplasmic reticulum membrane (RETREG1) were knocked out (Fig. S7I-S7L), and it showed that CXCR4-mediated GP degradation was partially blocked by knocking out of the SQSTM1/p62 (Figure 6E and S7M) and RETREG1 genes (Figure 6F and S7N). Given that autophagosomes would fuse with lysosomes to degrade its cargoes, we further knocked out the LAMP2 (lysosomal-associated membrane protein 2) gene (Fig. S7O and S7P), which was also found to partially block the CXCR4-mediated GP degradation (Figure 6G and S7Q). To confirm the autophagic degradation of GP mediated by CXCR4, we finally treated HEK 293T cells with rapamycin, an autophagy inducer, and found that it can further promote CXCR4-mediated GP degradation (Figure 6H). We also found that the protein levels of RETREG1 remained unchanged during EBOV infection (Fig. S7R), and knocking out of the RETREG1 gene did not influence the CXCR4 expression level (Fig. S7S), which further indicated the specific role of reticulophagy that was hijacked by EBOV GP to mediate GP degradation and promote viral fitness.

Figure 6.

CXCR4 targets EBOV GP for reticulophagic degradation. (A) one μg GFP-LC3 was co-transfected with 1 μg flag-tagged GP and 1 μg HA-tagged CXCR4 (or vector control) into HEK 293T cells, and the formation of autophagosomes indicated by GFP-LC3 puncta was determined by confocal microscopy. (B) two μg flag-tagged GP fused with eGFP was expressed with 2 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells. Subcellular localization of HA-CXCR4 (red), EBOV GP (green), and LC3 (far-red) were determined by confocal microscopy. DAPI indicated the nucleus and scale bars: 5 μm. (C) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in ATG3 knockout (KO) HEK 293T cells or control cells. (D) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in ATG5 KO HEK 293T cells or control cells. (E) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in SQSTM1/p62 KO HEK 293T cells or control cells. (F) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in RETREG1 KO HEK 293T cells or control cells. (G) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in LAMP2 KO HEK 293T cells or control cells. (H) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells and HeLa cells treated with 10 nmol/L rapamycin. Protein expression was analyzed by western blotting. GP was detected by anti-flag, CXCR4 was detected by anti-HA. ATG3, ATG5, SQSTM1/p62, RETREG1 and LAMP2 were detected by their specific antibodies. TUBA/α-tubulin was used as an internal control. Data are presented as means±standard error of measurements (SEMs) of three independent experiments. Two tailed Student’s t test was used. ****p < 0.0001.

Since the polyubiquitination of substrate was important for reticulophagic degradation [43], we next determined whether CXCR4 could promote polyubiquitination of GP after expressing ubiquitin in HEK 293T cells with co-immunoprecipitation. The results showed that GP was likely polyubiquitinated as indicated by high-molecular-mass ubiquitin-GP (over 180 kDa) (Figure 7A, input samples), and the polyubiquitination of GP was further promoted by ectopic expression of CXCR4 (Figure 7A, IP samples). To understand how GP was polyubiquitinated, we next investigated the ubiquitin chain linkage on GP. As totally 7 lysine residues could be used for polyubiquitination including K6, K11, K27, K29, K33, K48 and K63 [44], we thus generated mutants of ubiquitin that only possessed one lysine residue, with other six residues mutated to arginine, and also a ubiquitin mutant with all seven lysine residues mutated to arginine (7K/7 R). It was found that CXCR4-mediated GP polyubiquitination only in the presence of K27 and K63 mutants, compared with the 7K/7 R mutated group (Figure 7B), indicating that CXCR4-mediated GP polyubiquitination was via K27- and K63-linked ubiquitin chains.

Figure 7.

RNF185 participates in the CXCR4-mediated EBOV GP polyubiquitination and degradation. (A) one μg flag-tagged GP was expressed with 1 μg his-tagged ubiquitin (ub) in the presence or absence of 1 μg HA-tagged CXCR4 in HEK 293T cells. (B) one μg flag-tagged GP was expressed with 1 μg ub, its mutants with indicated lysine residue (K6, K11, K27, K29, K33, K48 and K63) reserved, or all seven lysine residues mutated to arginine (7K/7 R) in the presence or absence of 1 μg HA-tagged CXCR4 in HEK 293T cells. GP was pulled down by anti-flag, and protein expression in whole cell lysate (input) and immunoprecipitation (IP) samples were analyzed by western blotting (WB) in (A) and (B). GP was detected with anti-flag, CXCR4 was detected with anti-HA, and ubiquitin was detected with its specific antibody. (C) one μg flag-tagged GP was expressed with 1 μg HA-tagged CXCR4 and/or 1 μg MYC-tagged RNF185 in HEK 293T cells. (D) two μg flag-tagged GP was expressed alone or with 2 μg HA-tagged CXCR4 in RNF185 knockout (KO) HEK 293T cells or control cells. (E) two μg flag-tagged GP was expressed with 2 μg HA-tagged CXCR4 (or vector control) in HEK 293T cells. GP was pulled down by anti-flag and protein expression in whole cell lysate (input) and IP samples were analyzed by WB. (F) one μg flag-tagged GP was expressed with 1 μg MYC-tagged RNF185 in the presence or absence of 1 μg HA-tagged CXCR4 in HEK 293T cells. RNF185 was pulled down by anti-MYC and protein expression in whole cell lysate (input) and IP samples were analyzed by WB. (G) one μg flag-tagged GP was expressed with 1 μg MYC-tagged HGS in the presence or absence of 1 μg HA-tagged CXCR4 in HEK 293T cells. HGS was pulled down by anti-MYC and protein expression in whole cell lysate (input) and IP samples were analyzed by WB. (H) two μg MYC-tagged RNF185 and 2 μg flag-tagged GP were expressed in CXCR4 KO cells or control cells. RNF185 was pulled down by anti-MYC and protein expression in whole cell lysate (input) and IP samples were analyzed by WB. (I) two μg flag-tagged GP was expressed in CXCR4 KO cells or control cells. GP was pulled down by anti-flag and protein expression in whole cell lysate (input) and IP samples were analyzed by WB. (J) two μg flag-tagged wild-type or K673A mutated GP was expressed with 2 μg HA-tagged CXCR4 in HEK 293T cells. GP was detected by anti-flag, CXCR4 was detected by anti-HA. RNF185 and HGS were detected by anti-MYC or their specific antibodies. SQSTM1/p62 and RETREG1 were detected by their specific antibodies. TUBA/α-tubulin was used as an internal control.

RNF185 participates in CXCR4-mediated GP polyubiquitination and degradation

Previous study has reported that an E3 ubiquitin ligase, RNF185 (ring finger protein 185), was responsible for regulating proteostasis of GP, and catalyzed its K27-linked polyubiquitination [24]. Besides, other reports also indicated that RNF185 could catalyze the K63-linked polyubiquitination [45]. Therefore, we detected whether RNF185 was involved in the CXCR4-mediated GP degradation. We found that ectopic expression of RNF185 could promote GP degradation (Figure 7C and S8A), whereas knocking out of the RNF185 gene could partially block the CXCR4-mediated GP degradation (Figure 7D, S8B-S8D), suggesting that RNF185 participated in CXCR4-mediated GP polyubiquitination and degradation. To investigate whether RNF185 was recruited to EBOV GP by CXCR4, we detected the interactions between GP and RNF185 with or without ectopic expression of CXCR4. It showed that CXCR4 could promote the interaction between GP and RNF185 (Figure 7E,F), indicating the recruitment of RNF185 to GP for polyubiquitination and degradation by CXCR4. Meanwhile, CXCR4 was also found to promote the interaction between EBOV GP and HGS (Figure 7E), and HGS was shown to connect the GP and RNF185 complex to the reticulophagic receptor RETREG1 (Figure 7G). However, the interactions between GP, RNF185, HGS and RETREG1 were inhibited when the CXCR4 gene was knocked out (Figure 7H,I), which further proved that CXCR4-mediated GP degradation via reticulophagy. RNF185 was reported to be responsible for GP degradation via the K673 residue of GP in its cytoplasmic tail [24]. However, K673A mutation of GP could not block the CXCR4-mediated GP degradation (Figure 7J), suggesting that RNF185 participated in CXCR4-mediated GP polyubiquitination and degradation in a K673-independent manner. Collectively, all the results above demonstrated that EBOV GP was able to hijack the intracellular endosome sorting pathway through interacting with CXCR4 and could be carried back to endoplasmic reticulum for further reticulophagic degradation (Figure 8).

Figure 8.

The graphical abstract on the dual roles of CXCR4 in promoting EBOV entry and mediating excessively expressed mature GP degradation via reticulophagy to facilitate viral fitness. CXCR4 interacts with EBOV GP, and such interaction could mediate internalization of protein complex into early endosomes, where GP further interacts with NPC1 to release its genome to cytosol. Mature GP is transported to the cell membrane for release of the mature viral particles, whereas mature GP anchored on the cell membrane could hijack CXCR4 sorting and transporting pathway by their interactions with HGS, and then GP could be carried back to endoplasmic reticulum by CXCR4, where the E3 ubiquitin ligase RNF185 was recruited to polyubiquitinate GP in a K27- and K63-linked manner. Finally, polyubiquitinated GP was degraded in lysosomes via reticulophagy by interacting with the reticulophagic receptor RETREG1, in an ATG3- and ATG5-dependent manner. Thus, our study uncovered dual roles of CXCR4 in EBOV life cycle and provided new evidence that EBOV hijacked the host machinery through efficient virus-host interactions to promote viral fitness. (this figure was created by BioRender.com.) [the nucleus is double membrane.].

Discussion

Ebola virus disease is lethal to the EBOV-infected patients with the mortality rate up to 66% during the 2013–2016 outbreak in West Africa, however, the most efficient antibody drugs approved by FDA could not prevent over one third patients enrolled in the clinical trials from death [3–5]. Therefore, it is necessary to search for new antiviral strategies, which relies on deep understanding of EBOV life cycle and its interactions with the host. During EBOV life cycle, the initial step is the interaction between GP and cell surface molecules mediating viral attachment and internalization, which are key targets for the development of antiviral agents. HAVCR1 and CLEC4M were two kinds of cell surface molecules reported to interact with GP, facilitating its initial attachment [10,11]. However, HAVCR1 was expressed in only a few kinds of cell types and CLEC4M cannot support the internalization of EBOV in the non-permissive cells, there must be other cell surface molecules that can mediate EBOV attachment and entry considering the broad tropism of EBOV [12,13]. Autophagy was important for EBOV internalization, and CXCR4, the co-receptor for human immunodeficiency virus (HIV) entry, was reported to induce autophagy to maintain Epstein-Barr virus (EBV) infection [22,31,33]. Therefore, the role of CXCR4 and its-induced autophagy in EBOV life cycle were explored. We found that CXCR4 induced autophagy and EBOV infection promoted the internalization of CXCR4 without influencing its expression level, which prompted us to investigated whether there were interactions between CXCR4 and EBOV GP. We observed that CXCR4 interacted with GP on the EBOV particles and such interactions existed both in GP1 and GP2 subdomains. Most importantly, we found that CXCR4 interacted with RBD located in GP1, which revealed its potential to be an entry factor. Besides, interactions between GP and CXCR4 mediated the internalization of the protein complex into endosomes. The above two characters of CXCR4 prompted us to further explore its roles in EBOV life cycle.

We found that CXCR4 promoted the attachment and entry of EBOV both in HEK 293T and HeLa cells, which could be further reduced by CXCR4 antagonist and neutralizing antibody, confirming that CXCR4 can act as an entry factor of EBOV, except for its role in HIV. However, such conclusions need further confirmation on authentic virus in our future study. Apart from attachment and entry, we further explored the role of CXCR4 in EBOV replication. It was found that CXCR4 can increase EBOV replication, and the mRNA level of IL6 induced by CXCR4 was significantly decreased after EBOV infection, indicating that the competitive interactions between GP and CXCR4 instead of its natural ligand during EBOV infection may decrease the antiviral immune responses and synergistically enhance EBOV replication. Besides, as GP was toxic to host cells and we found that CXCR4 could facilitate EBOV replication by dose-dependently downregulating cytotoxic GP levels, which played significant roles in increasing viral fitness. Furthermore, as viral factory was formed by the interactions between NP, VP35 and VP24 after internalization, which played an important role in viral replication and assembly [41]. We first explored the influence of CXCR4 on the formation of viral factory. However, CXCR4 did not increase the interactions between NP and VP35, as well as NP and VP24, indicating that CXCR4 did not influence the formation and assembly of viral factory.

EBOV has evolved several mechanisms to reduce GP expression both in RNA and post-translation levels [17,18,23,24]. A previous study indicated that PDIA3 (protein disulfide isomerase family A member 3) triggered the GP misfolding, induced ER stress and targeted the misfolded GP to ER-associated degradation (ERAD) machinery [23]. In our study, we found out that CXCR4 could dose-dependently reduce the protein level of GP without influencing its transcriptional level, which raised our interest that how can a cell membrane-anchored receptor mediate the degradation of another protein. After further investigation, we found that CXCR4-mediated downregulation of GP was observed in all five species of ebolavirus, with the most inhibition in EBOV and its A82V mutant to a greater extent, which may indicate that such degradation may be evolved during EBOV-host interactions to balance the pathogenicity, replication and transmission of EBOV.

Several degradation pathways exist in cells, including proteasomal, endosomal and lysosomal pathways [46]. By treatment with either the proteasomal inhibitor MG132 or autophagosome-lysosome fusion inhibitor Baf A1 as well as the subsequent confocal analysis, we confirmed that CXCR4 mediated GP degradation via lysosomes. It was reported that CXCR4 could be internalized into endosomes and transported either to cell membrane for recycling or to lysosomes for degradation once activated by binding to its ligand [34,35]. During its degradation, CXCR4 interacted with HGS, the component of ESCRT, and recruited E3 ubiquitin ligase ITCH/AIP4 for ubiquitination-dependent degradation [36]. Therefore, we sought to investigate the roles of HGS and ITCH/AIP4 in CXCR4-mediated GP degradation, due to their presence in GP-interacting protein list by liquid chromatography mass spectrometry (LC-MS/MS). We found that HGS was involved in CXCR4-mediated GP degradation. ITCH/AIP4 degraded GP alone, however, it was not responsible for CXCR4-mediated GP degradation, which was the same to TRIM21, another E3 ubiquitin ligase screened out by LC-MS/MS.

Autophagy was reported to promote EBOV internalization by interactions of LC3B with macropinosomes [22]. We found that CXCR4 induced autophagy at the presence or absence of GP. As internalization happened after initial attachment, the roles of CXCR4 within the EBOV life cycle may be mainly as an entry factor by increasing the attachment process. Besides, reticulophagy is one kind of selective autophagy that serves as a quality control mechanism in the ER to degrade misfolded proteins by delivering them to phagophores, which subsequently form autophagosomes for further lysosomal degradation; this was also reported to participate in GP degradation [23,43,45,47]. We found that knocking out of the key ATG genes for phagophore formation and the reticulophagic receptor RETREG1 gene could alleviate CXCR4-mediated GP degradation, and such degradation was further promoted by treatment with rapamycin, an autophagy inducer. The above results indicated that GP hijacked the CXCR4 sorting pathway for further reticulophagic degradation to alleviate its cytotoxicity, thereby enhancing viral fitness. Moreover, we observed that the CXCR4-mediated GP degradation was reduced when the glycan cap of the gene encoding GP was deleted. GP is heavily glycosylated, which is important for its maturation, and numerous N-linked sites were identified in its glycan cap domain [48,49]. N-glycans were reported to target ERAD-resistant misfolded proteins for reticulophagic degradation [50]. Therefore, it was possible that deletion of the glycan cap of GP may inhibit the N-glycans linked to GP, influence its maturation and membrane localization, and finally block its reticulophagic degradation pathway, which was important for keeping GP proteostasis. Apart from glycan cap subdomain, deletion of internal fusion loop (IFL) subdomain can also reduce CXCR4-medaited GP degradation. IFL was important for membrane fusion between endosomes and virus, and was proved to participate in the interaction of GP and CXCR4, how it involved in CXCR4-mediated GP degradation remained to be elucidated [51].

The substrates undergoing reticulophagic degradation were always polyubiquitinated [43], and we found that CXCR4 induced GP polyubiquitination in a K27- and K63-linked manner. RNF185 was reported to catalyze K27-linked polyubiquitination of EBOV GP [24]. Meanwhile, it can also catalyze K63-linked polyubiquitination of BNIP1 (BCL2 interacting protein 1) [45]. Here, we confirmed that RNF185 was recruited by CXCR4 to the CXCR4-GP complex and was involved in CXCR4-mediated GP polyubiquitination and degradation, however, it cannot completely block the CXCR4-medaited GP degradation when the RNF185 gene was knocked out. Besides, RNF185 was reported to polyubiquitinate GP via its K673 residue [24], but K673A mutation of GP cannot block the CXCR4-mediated GP degradation, indicating that such degradation was in a K673 residue-independent manner, and other E3 ubiquitin ligases that catalyzed K27- and K63-linked polyubiquitination may be involved synergistically.

In conclusion, we reported the dual roles of CXCR4 in regulation of EBOV life cycle. It can act as an entry factor by interacting with GP on viral particles to facilitate viral attachment and entry without influencing viral factory assembly. Also, it can mediate the reticulophagic degradation of intracellular mature GP anchored on the cell membrane via its sorting and transporting pathways, which was important for balancing the pathogenicity, replication and transmission of EBOV, and could further promote viral fitness in host cells. Considering the technique limitations while utilizing pseudo-virus to study EBOV, our conclusions need further confirmation on authentic virus in our future study.

Materials and methods

Cell lines

The human embryonic kidney (HEK) 293 cell line (Pricella, CL-0005) transformed with SV40 large T antigen (HEK 293T), human cervical cancer cell line HeLa (Pricella, CL-0101) and human hepatocellular carcinoma cell line Huh7 (Pricella, CL-0120) were stored in our lab and were used in this study. Both cell lines were maintained in Dulbecco’s modified Eagle medium (DMEM; Thermo Fisher Scientific 11,965,092) supplemented with 10% fetal bovine serum (FBS; ExCell Bio, FSD500) and 1% penicillin-streptomycin (Thermo Fisher Scientific 15,070,063), and cultivated at 37°C in a humidified atmosphere in an incubator with 5% CO₂.

HEK 293T CXCR4 KO, HGS KO, ITCH/AIP4 KO, TRIM21 KO, ATG3 KO, ATG5 KO, SQSTM1/p62 KO, RETREG1 KO, LAMP2 KO, RNF185 KO, and HeLa CXCR4 KO cell lines were generated by CRISPR-Cas9. HEK 293T cells stably expressing CXCR4 were constructed for this study. All the cells were maintained in DMEM supplemented with 10% FBS, 1% penicillin-streptomycin and 1 μg/mL puromycin (Thermo Fisher Scientific, A1113803), and cultivated at 37°C in humidified atmosphere in an incubator with 5% CO₂. All cell lines used in our research were tested negative for mycoplasma contamination (Vazyme Biotechnology, D101–01).

Small molecule reagents including MG132 (MCE, HY-13259), BafA1 (MCE, HY-100558) and rapamycin (MCE, HY-10219) were dissolved in dimethyl sulfoxide (DMSO; Acmec, DMSO-D54260-500 mL) and maintained at −20°C.

Plasmids construction

Plasmids for EBOV trVLPs assay were kindly provided by Professor Thomas Hoenen (Friedrich-Loeffler Institute, Riems, Germany) and were maintained in our lab. pCAGGS-GP-3×flag, pCAGGS-GPΔRBD-3×flag, pCAGGS-GPΔglycan cap, pCAGGS-GPΔMLD-3×flag, pCAGGS-GPΔIFL-3×flag, pCAGGS-GPΔNHR-3×flag, pCAGGS-GPΔTM-3×flag were constructed from EBOV minigenome p4cis-vRNA-RLuc by PCR and EcoRI/XhoI digestion followed by homologous recombination, respectively. pCAGGS-GP1-3×flag, pCAGGS-GP2-3×flag, pCAGGS-GP1ΔRBD-3×flag, pCAGGS-SUDV GP-3×flag, pCAGGS-TAFV GP-3×flag, pCAGGS-BDBV GP-3×flag, pCAGGS-RESTV GP-3×flag and pCAGGS-3×flag-eGFP were constructed by directly gene synthesis and subsequent homologous recombination with pCAGGS-vector digested by EcoRI/XhoI. pCAGGS-CXCR4-HA, pCAGGS-HGS-3×MYC, pCAGGS-ITCH/AIP4–3×MYC, pCAGGS-TRIM21-3×MYC and pCAGGS-RNF185-3×MYC were constructed from HeLa cDNA by PCR and EcoRI/XhoI digestion followed by homologous recombination (Vazyme Biotechnology, C112–01 and C115–01). pCAGGS-GP^R64A-3×flag, pCAGGS-GP^A82V-3×flag and pCAGGS-GP^K673A-3×flag were constructed by site-directed point mutation (Takara, R010Q). All vectors were sequenced by Sanger DNA sequencing, whose maps were all generated by SnapGene 4.3.6. Plasmids were extracted by Qiagen plasmid Midi kit (Qiagen 12,143). Primers for plasmids construction are listed in Table S1.

Construction of CRISPR-Cas9 knockout (KO) cell lines

HEK 293T CXCR4 KO, HGS KO, ITCH/AIP4 KO, TRIM21 KO, ATG3 KO, ATG5 KO, SQSTM1/p62 KO, RETREG1 KO, LAMP2 KO, RNF185 KO, and HeLa CXCR4 KO cell lines were generated in this study, which was described previously [52]. Briefly, 1 × 10⁶ HEK 293T or HeLa cells were seeded into 6-well plates and transduced with lentivirus that was produced by HEK 293T cells transfected with lenti-CRISPR plasmids containing sgRNAs targeting each gene (designed by zlab guide-design-resources, https://portals. broadinstitute.org/gpp/public/analysis-tools/sgrnadesign), and together with 10 μg/mL polybrene (Santa Cruz Biotechnology, sc -134,220). After 48 h of infection, the infected HEK 293T cells were maintained in DMEM supplemented with 10% FBS and 2 μg/mL puromycin for another 48 h, and then 100 cells were counted and subsequently seeded into a 96-well plate. Single cell clone was subsequently identified by sequencing of PCR product and western blotting. Sequences of sgRNAs used in this study were listed in Table S2.

Construction of CXCR4 stably expressing cell lines

HEK 293T cells stably expressing CXCR4 was generated in this study. Briefly, 1 × 10⁶ HEK 293T cells were seeded into 6-well plates and transduced with lentivirus that were produced by HEK 293T cells transfected with pCDH-CXCR4 plasmids, and together with 10 μg/mL polybrene (Santa Cruz Biotechnology, sc -134,220). After 48 h of infection, the infected HEK 293T cells were maintained in DMEM supplemented with 10% FBS and 2 μg/mL puromycin for another 48 h. The expression of CXCR4 was confirmed by western blotting.

Transfection

Both HEK 293T and HeLa cells were transfected using the Lipofectamine 3000 kit (Thermo Fisher Scientific, L3000075) according to the manufacturer’s instructions. Briefly, 1 × 10⁶ HEK 293T cells or HeLa cells were seeded into 6-well plates with 70% confluency, 4 μg plasmids were mixed with 8 μL P3000 Reagent in 500 μL Opti-MEM medium (Thermo Fisher Scientific 31,985,070), and 8 μL Lipofectamine 3000 diluted in another 500 μL Opti-MEM. After mixing by vortex and incubation at room temperature for 5 min, they were mixed and incubated at room temperature for another 15 min. Then, the mixture was added to the cell culture. After 24 h of transfection, cells were washed with PBS and fresh media was added for culturing another 24 h before cells were harvested.

Production of EBOV trVLPs and infection

Production of EBOV trVLPs were described previously [53]. Briefly, 1 × 10⁶ HEK 293T cells (producer cells) were seeded into 6-well plates with 70% confluency, and were transfected with 125 ng pCAGGS-NP, 125 ng pCAGGS-VP35, 75 ng pCAGGS-VP30, 1 μg pCAGGS-L, 250 ng pCAGGS-p4cis-vRNA-RLuc, 250 ng pCAGGS-T7, and 25 ng pCAGGS-firefly by Lipofactamine 3000 kit. After 24 h of transfection, supernatant was discarded, and fresh medium was added. After another 48 h, supernatants were collected, centrifuged at 800×g for 5 min, and 2000×g for 10 min, followed by filtration through a 0.45-μm filter (Millipore, SLHVR33RB), which was stored at −80°C. For EBOV trVLPs infection, 1 × 10⁶ HEK 293T cells (target cells) were seeded into 6-well plates with 70% confluency, and transfected with 125 ng pCAGGS-NP, 125 ng pCAGGS-VP35, 75 ng pCAGGS-VP30, 1 μg pCAGGS-L, 250 ng pCAGGS-HAVCR1 and 25 ng pCAGGS-firefly for 24 h. Then, supernatant was discarded and EBOV trVLPs were added for infection for 24 h, and the supernatant was discarded and replaced with fresh medium for another 48 h.

Evaluation of viral attachment and entry

Protocols used for evaluation of viral attachment and entry were as previously described [54]. Briefly, HEK 293T, HEK 293T CXCR4 KO, HEK 293T stably-expressing CXCR4, HeLa, and HeLa CXCR4 KO cells were infected by EBOV trVLPs (1 × 10⁶ copies/mL) at 4°C for 2 h and were washed with pre-chilled PBS to remove unattached EBOV viral particles for attachment assay. For internalization assays, cells mentioned above were infected by EBOV trVLPs (1 × 10⁶ copies/mL) and maintained at 37°C for 2 h, followed by PBS washing and 0.05% trypsin-EDTA (Gibco 25,200,056) digestion to remove the membrane-bound viruses. Cells underwent attachment and entry assays were lysed by TRIzol ® Reagent (Thermo Fisher Scientific, 15596018CN) for further RNA isolation and quantification.

RNA isolation, reverse transcription and quantitative PCR

Total RNA was extracted using TRIzol® Reagent according to the manufacturer’s instructions. Reverse transcription was performed using M-MLV reverse transcriptase kit (Thermo Fisher Scientific 28,025,013) with random primer. The mRNA levels of CXCR4 and ACTB as well as RNA levels of GP and VP40 were detected by qPCR (CWBIO, CW3008H) using an QuantStudio 5 Real-time PCR machine (Thermo Fisher Scientific). The primers used to detect the mRNA levels of CXCR4 and ACTB as well as RNA levels of GP and VP40 are listed in Table S3.

Dual luciferase reporter assay

At 48 h post transfection, cells in 6-well dishes were washed in PBS and lysed with passive lysate buffer (Promega Corporation, E1941), and then the relative luciferase activity (Renilla to Firefly) was measured using the dual luciferase assay kit (Promega Corporation, E2940) following the manufacturer’s instructions.

Cell viability assay

HEK 293T cells were preciously transfected with pCAGGS-GP-3×flag and/or pCAGGS-CXCR4-HA plasmids (or vector control) for 24 h. Then, 1 × 10⁵ HEK 293T were seeded into a 96-well plate for 24 h, and 10 μL CCK-8 (Beyotime Biotechnology, C0038) was added into each well. Then, the absorbance at OD 450 and OD 630 was measured by a multimode reader synergy H1 (BioTek) at the indicated time point.

Western blotting

Cells were harvested and lysed in RIPA lysis buffer (Beyotime, P0013J) supplemented with 1×protease inhibitor cocktail (Beyotime, P1005) on ice for 30 min, centrifuged at 4°C 12000×g, then the supernatant was harvested, boiled and was subsequently subjected to SDS-PAGE. Proteins in gel were then transferred onto a nitrocellulose membrane (Merck Millipore, HATF00010) by the wet electrophoretic transfer method. The membrane was then blocked in 5% skimmed milk for 1 h and was incubated with primary antibodies at 4°C overnight. After washing in TBST (Solarbio Science & Technology Corporation, T1081) for 3 times, the membrane was incubated with the secondary antibodies for 1 h at room temperature. Then levels of proteins were detected by Odessey infrared imaging system (LICOR) or chemiluminescence imaging analysis system (Tanon). Antibodies used in this study are as follows: anti-Flag M2 mouse antibody (Merck, F1804), anti-HA-tag mAb (MBL Beijing Biotechnology, M180–3), anti-TUBA/α-tubulin mAb (MBL Beijing Biotechnology, M175–3), CXCR4 monoclonal antibody (Proteintech 60,042–1-Ig), ATG3 Rabbit polyclonal antibody (Proteintech 11,262–2-AP), ATG5 Rabbit polyclonal antibody (Proteintech 10,181–2-AP), RETREG1/FAM134B Rabbit polyclonal antibody (Proteintech 21,537–1-AP), ubiquitin recombinant antibody (Proteintech 80,992–11-RR), HGS Rabbit polyclonal antibody (Proteintech 10,390–1-AP), TRIM21 Rabbit polyclonal antibody (Proteintech 12,108–1-AP), LAMP2 Mouse polyclonal antibody (Proteintech 66,301–1-Ig), anti-ITCH/AIP4 Rabbit antibody (abcam, ab108515), anti-MYC antibody (MBL, 562), RNF185 polyclonal antibody (Immunoway, YN2489), CDH2/N-cadherin (Proteintech 66,219–1-Ig), IRDye 680LT donkey anti-rabbit IgG (LICOR, 926–68023), IRDye 800CW donkey anti-mouse IgG (LICOR, 926–32212), Goat anti-mouse IgG H&L (HRP; abcam, ab6789), Goat anti-rabbit IgG H&L (HRP; abcam, ab97051). Extraction of membrane and cytoplasmic proteins were conducted by Surface and Cytoplasmic Protein Reagent Kit (Beyotime, P0033). The gray value of bands in indicated experiments was analyzed using imageJ software (version 1.53a).

Immunoprecipitation

HEK 293T cells were seeded in 10-cm dishes, and were transfected with the indicated plasmids for 48 h. Then, 300 μL RIPA lysis buffer supplemented with 1×protease inhibitor cocktail was added into cell lysate and was incubated on ice for 30 min. After centrifugation at 4°C 12000×g for 10 min, 100 μL supernatant were collected as input, whereas the remaining 200 μL supernatant was incubated with the indicated agarose beads as follows: anti-DDDDK-tag pAb agarose (MBL Beijing Biotechnology, PM020–8), anti-HA-tag pAb-agarose (MBL Beijing Biotechnology, 561–8), anti-MYC-tag mAb agarose (MBL Beijing Biotechnology, M047–8) and rotated at 4°C for 2 h. After washing with pre-chilled RIPA lysis buffer for 5 times, proteins were detached from the beads after boiling in 50 μL 1×loading buffer and analyzed by western blotting.

Immunofluorescence and confocal microscopy

Cells were washed with pre-chilled PBS 3 times and were fixed with 4% paraformaldehyde at room temperature for 20 min. Cells were permeabilized with 0.5% Triton X-100 (Solarbio, T8200–100 mL) at room temperature for 15 min (not for detection of membrane-anchored proteins). Then, cells were blocked in 20% goat serum at room temperature for 1 h and incubated with the primary antibody at 4°C overnight. After washed with pre-chilled PBS for 3 times, cells were then incubated with the secondary antibodies at room temperature for 1 h, followed by staining with DAPI (Solarbio Science & Technology Corporation, C0065). The Treble-Fluorescence immunohistochemical mouse/rabbit kit (Immunoway Biotechnology, RS0035) was used for the colocalization of three proteins. Finally, cells were examined under a LSM780 laser scanning confocal microscope (Zeiss). Antibodies used in the immunofluorescence assay were as follows: anti-HA tag antibody ChIP grade (abcam, ab9110), CXCR4 monoclonal antibody (Proteintech 60,042–1-Ig), anti-EEA1 antibody (abcam, ab109110), LAMP1/CD107a Rabbit mAb (Abclonal Technology, A23947), LC3 (abcam, ab192890). Cell membrane was stained with cell plasma membrane staining kit with DiD far-red fluorescence (Beyotime, C1995S). Colocalization analysis was carried out using image pro plus 6.0 software and imageJ software (version 1.53a).

Mass spectrometry

To identify GP-interacting proteins, HEK 293T and Huh7 cells were seeded in a 10-cm dish, and were transfected with pCAGGS-GP-3×flag plasmids for 48 h. Then cells were lysed in 300 μL RIPA lysis buffer supplemented with 1×protease inhibitor cocktail on ice for 30 min. After centrifugation at 4°C, supernatant was collected and incubated with anti-DDDDK agarose beads (or control beads), after which proteins were separated by gel. The protein gel strips were sent to BGI corporation and MajorBio corporation for further LC-MS/MS analysis. Briefly, the protein gel strips were soaked in pure water, decolorized and dehydrated, and were digested with 25 mm NH₄HCO₃, followed by peptide samples were freeze-dried. The dried peptide samples were reconstituted with mobile phase A (2% ACN, 0.1% FA), centrifuged at 20,000×g for 10 min, and the supernatant was taken for injection. Separation was performed by Thermo UltiMate 3000 UHPLC. The peptides separated by liquid phase chromatography were ionized by a nanoESI source and then passed to a tandem mass spectrometer Orbitrap Fusion Lumos (Thermo Fisher Scientific) for DDA (Data Dependent Acquisition) mode detection. Finally, data were aligned with the UniProt protein database. The GP-interacting proteins were listed and uploaded as a single Excel file. Further functional annotation and enrichment analysis were carried out using the Database for Annotation, Visualization and Integrated Discovery (DAVID, https://david.ncifcrf.gov/) and MajorBio online analyzing platform (https://analysis.majorbio.com).

Statistical analysis

All data presented in this study were as mean±standard error of mean (SEM). Statistical analysis was preformed using GraphPad prism 10 software (GraphPad prism, v10.2.3). Comparisons between two groups were analyzed by two-tailed Student’s t tests. Comparisons among groups were analyzed by one-way ANOVA and post hoc Tukey tests, and a P-value less than 0.05 was considered statistically significant.

Funding Statement

This work was supported by the Shenzhen Science and Technology Program [ZDSYS20230626091203007] to Qian, J., the Introduced innovative R&D team project of “The Pearl River Talent Recruitment Program” [No. 2021ZT09Y544] to Qian, J.; Science and Technology Planning Project of Guangdong Province [No. 2021B1212040017] to Qian, J.; Guangzhou Basic and Applied Basic Research Program [No. 2024A04J4559] to Huang, H.

Disclosure statement

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

Data availability statement

All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. The GP-interacted proteins were listed and uploaded as a single Excel file. Additional data related to this paper are available on reasonable request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2025.2492877