Exosomal release of the virus-encoded chemokine receptor US28 contributes to chemokine scavenging

Summary The human cytomegalovirus (HCMV)-encoded chemokine receptor US28 contributes to various aspects of the viral life cycle and promotes immune evasion by scavenging chemokines from the microenvironment of HCMV-infected cells. In contrast to the plasma membrane localization of most human chemokine receptors, US28 has a predominant intracellular localization. In this study, we used immunofluorescence and electron microscopy to determine the localization of US28 upon exogenous expression, as well as in HCMV-infected cells. We observed that US28 localizes to late endosomal compartments called multivesicular bodies (MVBs), where it is sorted in intraluminal vesicles. Live-cell total internal reflection fluorescence (TIRF) microscopy revealed that US28-containing MVBs can fuse with the plasma membrane, resulting in the secretion of US28 on exosomes. Exosomal US28 binds the chemokines CX3CL1 and CCL5, and US28-containing exosomes inhibited the CX3CL1-CX3CR1 signaling axis. These findings suggest that exosomal release of US28 contributes to chemokine scavenging and immune evasion by HCMV.


INTRODUCTION
An interesting feature of US28 is its predominant intracellular localization, with only 20% of the receptor localized at the plasma membrane and the majority residing in intracellular compartments. 18 This intracellular localization has been attributed to the constitutive endocytosis and recycling of US28, which results in removal of receptor-bound chemokines from the microenvironment of HCMV-infected cells. [19][20][21] Chemokine scavenging by US28 has been demonstrated to neutralize CCL2-and CCL5-mediated migration of monocytes toward HCMV-infected fibroblasts, thereby contributing to immunosuppressive properties of the virus. 21 Based on its scavenging role, one would expect US28 to primarily localize in early and recycling endosomes. However, electron microscopic analysis by Fraile-Ramos and colleagues revealed that US28 is highly enriched in the intraluminal vesicles (ILVs) of late endosomal compartments called multivesicular bodies (MVBs). 18 When sorted into ILVs, US28 is presumably non-functional as it is physically separated from the cytosolic interactors that mediate downstream signaling. Furthermore, the fusion of US28-containing MVBs with lysosomes would result in substantial degradation of the receptor. This raises the question as to why it is beneficial for HCMV to express a receptor that is so heavily sorted toward this late endosomal compartment.
In the last decades it has become clear that not all MVBs are targeted for lysosomal degradation. Instead, a subset of MVBs can fuse with the plasma membrane, resulting in secretion of their ILVs as small extracellular vesicles (sEVs) called exosomes. 22 Exosomes contain various cargo molecules, including mRNAs, micro-RNAs (miRNAs), and proteins, and play a role in intercellular communication by interacting with, or delivering cargo to, recipient cells. 23 Many viruses, including herpesviruses, have been found to employ components of the host exosome biogenesis machinery for virion production. 24,25 Furthermore, cells infected with the gamma-herpesviruses Epstein-Barr virus (EBV) and Kaposi sarcoma-associated herpesvirus (KSHV) have been found to secrete exosomes harboring viral miRNAs and proteins. [25][26][27][28] In line with this, recent studies have reported that key drivers of exosome biogenesis contribute to HCMV virion production and that envelopment of HCMV virions occurs, in part, at MVB membranes. [29][30][31] In addition, HCMV-infected cells release sEVs containing viral proteins, such as glycoprotein B (gB). 29,30,[32][33][34] Since US28 is mainly expressed intracellularly and highly enriched in MVBs, we hypothesized that US28 sorting into MVBs facilitates its secretion on exosomes and HCMV virions.
Here, we report that US28-containing MVBs can fuse with the plasma membrane, leading to the secretion of exosomes carrying US28. We further show that exosomal US28 retains the ability to bind the human chemokines CX 3 CL1 and CCL5 and that CX 3 CL1 scavenging by exosome-bound US28 modulates the CX 3 CL1-CX 3 CR1 signaling axis.

US28 localizes to MVBs and is sorted into ILVs
US28 localizes to intracellular compartments, both in HCMV-infected cells and when expressed exogenously in non-infected cells. 12,18 Previous work demonstrated that US28 overexpression in HeLa cells results in its sorting into the ILVs of late endosomal compartments. 18 Given the oncomodulatory role of US28 in glioblastoma and the intracellular expression of US28 in glioblastoma cells, we determined its localization in an iHA-US28-U251 glioblastoma cell line that allows for doxycycline-inducible hemagglutinin (HA)-tagged US28 expression. 35 In line with previous findings, 18,36 we observed partial colocalization between US28 and the late endosomal/lysosomal markers cluster of differentiation 63 (CD63) and lysosomal-associated membrane protein 1 (LAMP1) (Figures 1A and S1A). Furthermore, immunoelectron microscopy confirmed sorting of US28 into the ILVs of MVBs ( Figure 1B). In addition, we observed US28 on intracellular membranes that appear to be early endosomes or enlarged Golgi tubules ( Figure 1B). To determine US28 localization in the context of HCMV infection, we infected parental U251 cells with the clinical HCMV strain Merlin. At six days post-infection, the cells express US28, which partially co-localizes with CD63 as well as the cis-Golgi marker GM130 in a perinuclear compartment ( Figures 1C and S1B). These findings confirm that US28 partially localizes to late endosomal compartments where it is sorted into ILVs.
US28-containing MVBs can fuse with the plasma membrane to release exosomal US28 MVB maturation and fusion with lysosomes result in the degradation of its content. However, is has become clear that a subset of MVBs fuse with the plasma membrane instead, leading to secretion of the ILVs as exosomes. 22 We previously developed a live-cell total internal reflection fluorescence (TIRF) microscopy approach using the reporter CD63-pHluorin that enables visualization of MVBplasma membrane fusion. 37 iScience Article membrane, we developed US28-pHluorin by inserting the pH-sensitive green fluorescent protein superecliptic pHluorin 39 into the second extracellular loop of US28 ( Figure 2A). US28-pHluorin retains functionality as it constitutively activates NFAT ( Figures S2A and S2B) and localizes to CD63-positive endosomes ( Figure S2C). We expressed US28-pHluorin in HeLa cells, which have a flat and spread-out plasma membrane that makes them ideal for studying exocytosis with TIRF microscopy. 37 Upon live TIRF imaging of these cells, we detected multiple sudden localized bursts of fluorescence over time, suggesting fusion of US28-pHluorin-positive acidic compartments with the plasma membrane ( Figures 2B and 2C, Video S1). We previously observed that MVB-plasma membrane fusion, as visualized by CD63-pHluorin, results in significantly longer signal duration when compared to plasma membrane deposition and lateral diffusion of vesicle-associated membrane protein 2 (VAMP2)-pHluorin. 38 This prolonged fluorescence signal upon MVB-plasma membrane fusion most likely results from trapping of the released exosomes between the cells and the coverslips, preventing their fast diffusion. Thus, signal duration gives us an additional measure to differentiate exosome release from other exocytic events. For both US28-and CD63-pHluorin we observed several fusion events with a signal duration comparable to VAMP2-pHluorin (mean: 1.8 s), iScience Article which likely result from plasma membrane deposition of US28 and CD63 by recycling endosomes or transport vesicles ( Figure 2D). However, the signal duration of the majority of US28-pHluorin fusion events was longer (mean: 23 s) and similar to the duration of CD63-pHluorin fusion events (mean: 22 s), suggesting that most fusion events result from the fusion of US28-pHluorin-containing MVBs with the plasma membrane. In line with this, we could detect US28 in the sEV pellet from the supernatant of doxycycline-induced iHA-US28 U251 cells ( Figure 2E). iScience Article

HCMV-infected cells secrete US28 on small EVs and virions
To study exosome-mediated secretion of US28 in the context of HCMV infection, we used bacterial artificial chromosome-based recombineering to generate HCMV-US28-pHluorin from the low-passage HCMV strain Merlin 40 ( Figure 3A). In HCMV-US28-pHluorin-infected human fetal foreskin fibroblasts (HFFF-Tet cells 40 ), green fluorescent signal could be observed in the perinuclear virion assembly compartment (VAC) ( Figure 3B), indicating the presence of US28-phluorin in non-acidic compartments in the VAC, such as the Golgi. To assess whether US28-pHluorin also localizes to MVBs in infected fibroblasts, we performed immunoelectron microscopy using an anti-GFP-antibody that recognizes pHluorin. Similar to our observations in cells overexpressing US28, we could detect gold labeling on the small ILVs within MVBs ( Figure 3C). Unfortunately, the high background caused by US28-pHluorin in non-acidic compartments in the VAC hampered the live TIRF imaging of MVB-plasma membrane fusion events in HCMV-infected cells. To investigate the presence of US28 in EVs secreted by HCMV-infected cells, we used ultracentrifugation to pellet both HCMV virions and sEVs from the supernatant of HCMV-infected fibroblasts and performed immunoelectron microscopy using anti-GFP-gold labeling ( Figures 3D and 3E). Consistent with previous findings, 9 we did occasionally observe gold labeling on HCMV virions, which could be identified by their size of G200 nm and the presence of a viral capsid 41 ( Figure 3E, upper left panel). In addition, we observed the association of gold particles with 100-150 nm sEVs, corresponding to the size of MVB-derived exosomes ( Figures 3D and 3E). These results demonstrate that HCMV-infected cells release US28 both on mature virions and on sEVs. The secretion of US28 on exosomes raises the question as to what the functional relevance of this process during HCMV infection could be. It has been reported that mesenchymal stem cell EVs containing the chemokine receptor CCR2 act as chemokine scavengers. 42 Given the proposed scavenging function of US28, we set out to investigate whether exosomal US28 modulates the activity of extracellular chemokines. First, we determined whether exosomal US28 can indeed bind to its ligands CX 3 CL1 and CCL5. For this, we used a NanoBRET approach to measure the binding of fluorescently labeled chemokines to exosomes containing NanoLuc (Nluc)-tagged US28 ( Figure 4A). Exposure of supernatant from Nluc-US28-transfected HEK293T cells to fluorescent CX 3 CL1 or CCL5 resulted in a concentration-dependent increase in bioluminescence resonance energy transfer (BRET) signal, whereas fluorescent CXCL12, which is not a ligand for US28, did not increase the BRET signal ( Figures 4B-4D). Importantly, the increase in BRET signal obtained with fluorescent CX 3 CL1 or CCL5 could be counteracted by competition with a saturating concentration of anti-US28 nanobody (VUN100), which has previously been demonstrated to displace CX 3 CL1 and CCL5 iScience Article from US28 43 (Figures 4C and 4D). To ensure that we specifically measured chemokine binding to exosomeassociated Nluc-US28, we repeated the NanoBRET assay after separation of EVs and soluble proteins from the culture supernatant by size-exclusion chromatography (SEC) and could only observe an increase in BRET signal for the EV fraction ( Figure 4E). These findings show that the increase in BRET is caused by specific interactions of fluorescent CX 3 CL1 and CCL5 with exosomal US28 and demonstrate that exosomal US28 has the potential to scavenge chemokines.
To investigate whether chemokine scavenging by exosomal US28 could influence human chemokine receptor signaling, we determined the effect of exosomal US28 on CX 3 CL1-induced activation of CX 3 CR1. CX 3 CR1 activation was measured using a BRET-based assay that measures recruitment of Mini G i proteins 44 to CX 3 CR1-Nluc ( Figure 4F). US28-containing EVs were isolated from the culture supernatant of doxycycline-stimulated iHA-US28-U251. In the presence of US28-containing EVs, but not US28-negative control EVs, CX 3 CL1-induced recruitment of mini Ga i proteins to CX 3 CR1 was inhibited by over 60% ( Figure 4G). Moreover, preincubation of US28-containing EVs with the anti-US28 nanobody VUN100 rescued CX 3 CR1 activation, showing that this inhibitory effect is indeed US28 dependent. Together these results suggest that exosomal US28 can modulate human chemokine receptor signaling by scavenging soluble chemokines.

DISCUSSION
The viral chemokine receptor US28 is a versatile receptor that contributes to various aspects of HCMV infection, such as chemokine scavenging, cell-to-cell transmission, and the establishment of latency. 8 In contrast to the plasma membrane localization of most human chemokine receptors, the majority of US28 is localized within the ILVs of MVBs. 18 Sorting of US28 into MVBs has been shown to be mediated by the adapter GPCRassociated binding protein 1 (GASP1), 36 which facilitates the incorporation of GPCRs into ILVs by linking them to the endosomal sorting complexes required for transport (ESCRT) protein hepatocyte growth factor regulated tyrosine kinase substrate (HRS). 45 The sorting of US28 into ILVs has been suggested to lead to its degradation in lysosomes, but the reason for such extensive degradation was unclear. In this study, we show that alternatively to fusion with lysosomes, a subset of the US28-containing MVBs can fuse with the plasma membrane and release US28-containing ILVs into the extracellular space as exosomes.
US28 sorting into ILVs and its subsequent secretion on exosomes might serve a number of different purposes that could be beneficial for HCMV. Firstly, US28 is constitutively active and in recent years it has become clear that activated GPCRs can sustain signaling from endosomal compartments. 46 This limits the possibilities for HCMV to control US28 signaling in infected cells and increases the risk of overactivation of US28 signaling pathways, which could be detrimental for the host cell and have undesirable consequences on viral persistence. The sorting of US28 into ILVs and its subsequent release via exosomes or degradation in lysosomes could help to regulate US28 signaling. In fact, another herpesvirus, the Epstein-Barr virus (EBV), employs a similar strategy to control constitutive nuclear factor kB (NF-kB) activation by the viral latent membrane protein-1. 47 Secondly, sorting of US28 to MVBs could be a strategy to enable efficient incorporation of US28 into the viral envelope during virion maturation. Although for a long time HCMV envelopment was believed to occur mostly at early endosome and Golgi membranes, 48,49 a growing body of evidence suggests that the host exosome biogenesis pathway underlies virion production 29,30 and that HCMV envelopment at least partially occurs at the MVB limiting membrane. 31 HCMV infection causes remodeling of the secretory pathway, in part through viral miRNAs that target members of the endocytic pathway, such as ras-related protein rab-5 (RAB5) and synaptosomal-associated protein 23 (SNAP23), which could influence MVB exocytosis dynamics in infected cells. 50 However, multi-color TIRF microscopy using CD63-pHluorin, the viral glycoprotein gM/UL100-mScarlet-I, and the tegument protein pp150-SNAP revealed that virion-containing MVBs can fuse with the plasma membrane. 31 A potential reason for the sorting of US28 into virions became apparent, when direct insertion of US28 from the viral envelope into the plasma membrane of newly infected cells was found to contribute to the establishment of latency. 9 iScience Article that suppress CCL2-induced macrophage migration and activation, 42 suggesting that chemokine scavenging by EV-associated chemokine receptors is a more common feature of chemokine receptors. In the context of HCMV infection, exosomal US28 could reduce chemokine-dependent chemotaxis of immune cells toward the site of infection. In addition to chemokine scavenging by constitutive endocytosis and recycling of cellular US28, the secretion of US28 on exosomes may enable HCMV to disrupt chemokine gradients further away from the infection site. As such, the secretion of US28-containing exosomes could help HCMV-infected cells to escape immune surveillance. Similarly, chemokine scavenging by various human chemokine receptors that are present in cancer cell-derived EVs 51,52 may have significant implications for anti-tumor immune responses. Interestingly, two other HCMV-encoded GPCRs, UL33 and US27, also sorted into ILVs, 53 have recently been identified in densitygradient ultracentrifugation-purified sEVs from the supernatant of HCMV-infected cells. 30 Unlike US28, these receptors are not known to bind chemokines, indicating that chemokine scavenging might not be the only reason for EV-mediated secretion of viral GPCRs.
Besides their interaction with extracellular matrix components, exosomes and other EVs mediate intercellular communication by delivering their cargo to recipient cells. Exosome-mediated transfer of viral material is increasingly recognized as a mechanism by which herpesviruses manipulate non-infected cells in their environment, 25 and HCMV exosomes have been shown to promote viral spread in fibroblasts. 32 EV-mediated transfer of functional receptors has been described for various GPCRs, including the human chemokine receptors CXCR4 52,54,55 and CCR5. 56 Whether exosomal US28 can be functionally transferred to non-infected cells remains to be investigated. However, given the role of virion-associated US28 in the establishment of latency, it would be of interest to see whether functional delivery of exosomal US28 contributes to the immediate MIEP silencing observed at early time points after infection of monocytes and hematopoietic progenitor cells.
Over the years, it has become clear that the viral GPCR US28 regulates a wide variety of viral processes throughout the various stages of HCMV infection. 8 In this study, we demonstrated that the functionality of US28 extends beyond its roles in infected cells or in virions; exosome-associated US28 contributes to chemokine scavenging. These findings provide an explanation for the localization of US28 in MVBs and pave the way to further explore additional functionalities of exosome-mediated viral GPCR secretion by HCMV.

Limitations of the study
In this study we demonstrated that US28-containing MVBs can fuse with the plasma membrane to release US28 on exosomes. Furthermore, we observed US28 within MVBs and in the sEV pellet of HCMV-infected cells. However, the presence of US28-pHluorin in non-acidic compartments hampered the live TIRF imaging of MVB exocytosis in HCMV-US28-pHluorin-infected cells. Therefore, we cannot exclude that (a fraction of) the US28containing sEVs from HCMV-infected cells bud directly from the plasma membrane. Importantly, the size of the US28-containing sEVs is consistent with an endosomal origin and our findings are consistent with a recent study that demonstrated MVB-plasma membrane fusion in HCMV-infected cells using CD63-pHluorin. 31 While it is clear that chemokine scavenging by exosomal US28 inhibits CX 3 CR1 signaling, further studies should be performed to assess the effect of chemokine scavenging by exosomal US28 on immune evasion by HCMVinfected cells. Similarly, the potential horizontal transfer of exosomal US28 to non-infected recipient cells warrants further investigation as this could have implications for the establishment of latent HCMV infection.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Lead contact
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Martine J. Smit (mj.smit@vu.nl).

Materials availability
Plasmids and recombinant HCMV strains generated in this study will be made available upon request. A material transfer agreement will be required prior to sharing of materials.

Data availability
d Data reported in this paper will be shared by the lead contact upon request d This paper does not report original code.
d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.