An integrin/MFG-E8 shuttle loads HIV-1 viral-like particles onto follicular dendritic cells in mouse lymph node

During human immunodeficiency virus-1 (HIV-1) infection lymphoid organ follicular dendritic cells (FDCs) serve as a reservoir for infectious virus and an obstacle to curative therapies. Here, we identify a subset of lymphoid organ sinus lining macrophage (SMs) that provide a cell-cell contact portal, which facilitates the uptake of HIV-1 viral-like particles (VLPs) by FDCs and B cells in mouse lymph node. Central for portal function is the bridging glycoprotein MFG-E8. Using a phosphatidylserine binding domain and an RGD motif, MFG-E8 helps target HIV-1 VLPs to αv integrin bearing SMs. Lack of MFG-E8 or integrin blockade severely limits HIV-1 VLP spread onto FDC networks. Direct SM-FDC virion transfer also depends upon short-lived FDC network abutment, likely triggered by SCSM antigen uptake. This provides a mechanism for rapid FDC loading broadening the opportunity for rare, antigen reactive follicular B cells to acquire antigen, and a means for HIV virions to accumulate on the FDC network.


Introduction
Afferent lymphatics deliver local infectious agents and subcutaneously administered antigens via the lymph to nearby lymph nodes (LNs) (Friedlaender and Baer, 1972). As the lymph flows through the subcapsular sinus, macrophages that line the sinus floor (SCSMs) can actively uptake antigenic material (Carrasco and Batista, 2007;Fossum, 1980;Nossal et al., 1965). For example, locally injected ultraviolet-light inactivated vesicular stomatitis virus (VSV) rapidly appeared in the lymph and accumulated at discrete sites on SCSMs. While the mechanism responsible for VSV capture and accumulation on the SCSM was not identified, it did not depend upon natural antibody and complement C3 fixation as no defect in SCSM VSV retention occurred in animals that lacked C3. Furthermore, the capture mechanism was not unique to VSV as SCSMs also retained locally injected adenovirus and vaccinia virus as well as HIV-1 and murine leukemia virus like particles (VLPs) (Junt et al., 2007). Implicating the SCSM scavenger receptor CD169/sialoadhesin in HIV-1 and MLV-VLPs uptake, antibodies directed against CD169 reduced VLP capture. A known CD169 target sialyllactose is found on gangliosides embedded in the HIV-1 and MLV lipid membranes (Sewald et al., 2015). Besides CD169 SCSM likely employ multiple other receptors to capture material flowing in the lymph (Gray and Cyster, 2012).
Once bound to an SCSM, a virus such as VSV rapidly translocate along the SCSM membrane penetrating the SCS floor. SCSM membrane processes bearing virus project into the underlying B cell follicle. This allows migrating virus-specific B cells access to the viral particles. In VSV infection, SCSM macrophage depletion compromised viral retention, worsened viremia, and impaired B cell responses (Junt et al., 2007). Somewhat analogously, captured murine leukemia virus VLPs translocated along the SCSM membrane becoming available to B1 B cells, a target of MLV infection. Once infected the B1 cells migrated to other lymph nodes to spread the infection (Sewald et al., 2015).
Neither the VSV or the MLV-VLP studies identified the mechanism that transports viral particles across the SCS floor facilitating their delivery to B cells in the lymph node follicle. Furthermore, neither study examined the transfer of the virions or VLPs to the underlying FDC network. FDCs are critical for humoral immunity as they serve as a long-term repository for unprocessed antigens and form a framework for the germinal center and B cell memory responses (Mandels et al., 1980).
In patients infected with HIV-1 the ability of FDCs to retain HIV-1 virion for prolonged durations poses a significant challenge to the eradication of the infection. The FDC HIV-1-reservoir contributes to low level viremia. It also provides a potential source of infectious virus for the viral resurgence that typically occurs following treatment interruption (Heath et al., 1995;Schacker et al., 2000;Smith et al., 2001). Important for FDC retention of virions is the complement receptor 2 (CR2, CD21). FDCs prominently express CR2, which binds the complement components C3d and iC3b. FDCs also express a related protein termed CR1 (CD35), which binds C3b, iC3b, C4b, and iC4b (Carroll, 1998). Virions become associated with complement components due to their recognition by natural and specific antibodies that fix complement. This allows FDC complement receptors to retain the HIV-1 virions as immune complexes (IC) on lymphoid organ FDC networks (Armstrong, 1984).
Antibodies that recognize CD21/CD35 are used to identify FDCs in tissue sections. While follicular B cells also express CR2, they can be distinguished from FDCs based on their morphology and lower CR2 expression level. Another marker used to recognize FDCs in tissue sections is FDC-M1 (Haley et al., 1995). The FDC-M1 antibody binds MFG-E8, a bridging molecule that possesses a phosphatidylserine (PS) binding domain and a second domain that contains an RGD motif (Kranich et al., 2008). Human and mouse MFG-E8 have two PS binding domains termed C1 and C2. Human MFG-E8 has a single RGD motif binding domain while the mouse has a short and an extended version, which possesses a solitary, or two RGD motif binding domains, respectively (Hanayama et al., 2002). In lymphoid organs, MFG-E8 binds PS bearing apoptotic cells delivering them to avb3 integrin bearing tingible body macrophages (TBM). Engagement of avb3 integrins promotes the phagocytosis and eventual destruction of the apoptotic cell by the macrophages (Asano et al., 2004). Besides, binding apoptotic cells, MFG-E8 and related proteins can bind PS present in viral envelopes (Morizono and Chen, 2014). As HIV-1 viral particles bud from the host cell, the viral envelope can selectively acquire proteins and lipids from host membranes. While PS is normally located on the plasma membrane inner leaflet during viral budding the PS becomes exposed on the HIV-1 envelope potentially serving as a target for MFG-E8 and related PS binding proteins (Morizono and Chen, 2014;Wu and Shroff, 2018). Because MFG-E8 is widely expressed and found at substantial concentrations in blood and potentially the lymph (Kishi et al., 2017;Yamaguchi et al., 2008); most HIV-1 virions likely bear MFG-E8. This would target HIV-1 virions to monocytes and macrophages that express avb3/5 integrins. This would also provide another mechanism in addition to CD169 for SCSMs to capture HIV-1 virions from the lymph.
In this study, we examined the delivery of HIV-1 VLPs to subcapsular sinus macrophages and the underlying FDC network. In addition, we assessed the uptake of HIV-1 VLPs by naïve lymph node B cells and by gp120-specific transgenic B cells expressing b12 heavy and light chains. During these studies, we found that HIV-1 VLPs accumulated at distinct sites on SCSMs, which co-localized with MFG-E8 and avb3 integrins. Integrin blockade markedly decreased HIV-1 VLP uptake and the appearance of the VLPs on the FDC network. Our results indicate that SCSMs can capture MFG-E8 bound material including HIV-1 VLPs from the lymph, and that entrance into an MFG-E8 enriched compartment facilitates the transfer of this material to the underlying FDC network. Finally, we show a more dynamic FDC network than previously appreciated, which can directly contact the overlying SCSMs.

Results
HIV-1 and murine leukemia virus (MLV) VLPs accumulate on discrete site on SMs enriched for MFG-E8 prior to translocation to the FDC network The HIV-1 envelope protein gp120 injected near the inguinal LN accumulated on SCSMs located between the LN follicles . To check the localization of HIV-VLPs following subcutaneous injection near the inguinal LN, we used intravital 2-photon imaging to collect large image stacks from the SCS deep into the LN follicle at 6 hr ( Figure 1A) and 20 hr ( Figure 1B) post injection. Suggesting that gp120 does not direct HIV-1 virion uptake, the HIV-1 VLPs (NL4.3-GFP) rapidly localized (1 hr) on the SCSMs that overlie the LN follicle rather than those over the interfollicular region (Figure 1-figure supplement 1A). To clarify the uptake patterns, we co-injected differentially labeled gp120 and HIV-1 VLPs. Both intravital imaging and thick lymph node section imaging (SIGN-R1 immunostained to delineate the interfollicular channels) confirmed the differential uptake ( Figure 1-figure supplement 1B &C). At the 6 hr time point some of the VLPs had moved inside the lymph node consistent with early FDC loading ( Figure 1A and Video 1). By 20 hr post injection many of the VLPs had transited deep into the B cell follicle, consistent with extensive FDC loading ( Figure 1B and Video 1). To begin to assess the mechanism of the FDC loading, we first asked whether another viral particle behaved similarly to the HIV-1 VLPs. We co-injected Murine leukemia virus (MLV) VLPs and 3 hr later prepared thick LN slices, which we immunostained for CD21/ 35 and CD169 to outline FDCs and SCSMs, respectively. Following injection of the VLPs, both VLP sets appeared on overlapping sites on the SCSMs (~88% colocalization index, Figure 1C). SCSMs have previously been reported to capture these VLPs (Sewald et al., 2015). We also noted that both sets of VLPs localized on lymph node FDCs that had directly contacted the overlying SCSMs. These contacts persisted as they were readily observed at 8 hr post injection ( Figure 1D).
To confirm the CD21/35 immunostaining, we used another antibody that reacts with FDCs termed FDC-M1. This antibody is known to recognizes MFG-E8 (Gray et al., 1991). Again, we coinjected NL4.3-GFP and MLV-RFP and made lymph node sections. Surprisingly, we found that besides identifying the lymph node FDCs, the FDC-M1 antibody precisely delineated the SCSM membrane sites where the VLPs had localized ( Figure 1E). This is more evident in zoomed images. The FDC-M1 immunostaining often encased the VLPs, and when not confluent appeared vesicular. These FDC-M1 SCSM defined sites extended into the underlying follicle directly abutting FDCs processes. Intensity mapping of the three signals on SCSMs shows the strong overlap between MFG-E8 and the two VLPs ( Figure 1F, far right). 3D-reconstruction of thick section confocal images also shows NL4.3-GFP overlying the SCSM MFG-E8 + sites, which we termed the MFG-E8 + compartment (Video 2). To confirm the lymph node MFG-E8 + SCSMs captured NL4.3-GFP VLPs we used flow cytometry to analyze lymph node cell preparations. The CD169 + macrophages captured most of the NL4.3-GFP + signal ( Figure 1-figure supplement 2A), and when separated based on MFG-E8 expression, the MFG-E8 high subset bound the most ( Figure 1G, Figure 1-figure supplement 2B, C). This was the case after in vivo transfer (3 hr) and following in vitro binding experiments (30 min on ice).
Splenic CD169 + marginal zone macrophages serve a similar function to that of the lymph node SCSM, although they capture pathogens and antigenic material from the blood rather than from the lymph. To check whether blood-borne HIV-1 VLPs would be captured by marginal zone macrophages and transferred to the splenic white pulp FDC networks, we injected NL4-3-GFP into the blood. After 6 hr we collected the spleens of the injected animals and made thick splenic sections for immunohistochemistry ( Figure 1H). We found a similar scenario as we had noted in the lymph nodes, the accumulation of HIV-1 VLPs on marginal zone macrophages at MFG-E8 rich sites and the appearance of VLPs on the underlying FDC networks. Thus, sinus lining macrophages in contact with the blood or the lymph collect HIV-1 VLPs at sites on the macrophages enriched for MFG-E8. This raised several questions. How does the MFG-E8 become co-localized with the VLPs on the SCSM and the marginal zone macrophages? Are the HIV-1 VLPs coated with MFG-E8? If so, does MFG-E8 target the VLPs to avb3 integrins present on the macrophages?
HIV-1 VLPs bind MFG-E8, which can interact with avb3 integrins First, we tested whether the HIV-1 VLPs could bind MFG-E8. HIV-1 envelopes are known to expose PS (Aloia et al., 1993) providing a potential mechanism for MFG-E8 binding. To determine whether MFG-E8 can bind the HIV-1 VLPs, we used stimulated emission depletion (STED) microscopy to examine NL4.3-GFP VLPs incubated with directly labeled recombinant MFG-E8. The images show MFG-E8 encasing the VLPs (Figure 2A, Figure 2-figure supplement 1A-F). As expected MFG-E8 bound PS, but not phosphatidylcholine (PC) liposomes ( Figure 2A). Signal intensity analysis showed MFG-E8 surrounding the NL4.3-GFP core ( Figure 2A, right panels). Due to their strong, condensed GFP signal, we could detect the VLPs by flow cytometry. This provided another method to test whether NL4.3-GFP VLPs bound MFG-E8. We confirmed that the VLPs expressed gp120 by incubating them with the gp120-specific antibody VRC01 ( Figure 2B Histogram). Next, we incubated the VLPs with unlabeled MFG-E8 and detected bound MFG-E8 with an MFG-E8-specific antibody. A low level MFG-E8 antibody signal in the absence of added MFG-E8 suggests that the VLPs may have bound some MFG-E8 during their preparation. However, the addition of recombinant MFG-E8 to the VLPs resulted in a strong shift in the intensity of the MFG-E8 antibody staining (~90% of VLPs were MFG-E8 + ) ( Figure 2B Figure 2D). Together these results indicate that MFG-E8 bound VLPs likely bind to integrin avb3 on SCSM.
Since most blood-borne HIV-1 is likely MFG-E8 bound, the MFG-E8/virion complex may target avb3 expressing peripheral blood leukocytes in infected individuals. As expected, human peripheral blood monocytes expressed avb3 integrins, while CD4 and CD8 T cells did not ( Figure 3C), nor did peripheral blood B cells (data not shown). To test mononuclear cell MFG-E8/VLP binding we incubated human peripheral  blood mononuclear cells on ice with NL4.3-GFP in the presence or absence of recombinant human MFG-E8. In the absence of added MFG-E8 we observed very low-level VLP binding to some donor's monocytes ( Figure 3D), however, the addition of MFG-E8 resulted in a substantial increase in NL4.3-GFP binding as approximately 25% of the cells had both an MFG-E8 and a VLP signal ( Figure 3E). Another donor's monocytes had high levels of constitutive NL4.3-GFP binding, which was enhanced by adding recombinant MFG-E8 ( Figure 3F). PS containing liposomes, but not PC liposomes competitively inhibited the MFG-E8/VLP binding ( Figure 3F). Thus, MFG-E8 bound HIV-1 virions likely target human peripheral blood monocytes by binding to avb3 integrins.
Blocking SCSM av integrins inhibits NL4.3-GFP VLP SCSM capture and transfer to the FDC network, while the lack of MFG-E8 predominately inhibits VLP transfer to the FDC network To determine whether RGD binding integrins participate in SCSM NL4.3-GFP capture, prior to injecting NL4.3-GFP, we injected a circularized RGD peptide as a competitive inhibitor or an av blocking antibody (aCD51). Both approaches reduced the SCSM NL4.3-GFP signal, but only by 20-30% (data not shown); however double blockade (RGDyK + aCD51) strongly reduced SSCM uptake ( Figure 4A, Figure 4-figure supplement 1A,B). To visualize SCSMs in the absence of CD169 antibody injection, we employed CX 3 CR-1-GFP mice as recipients. Additionally, to provide some insight into the kinetics of the double blockade, we simultaneously injected NL4.3-mCherry VLPs with the blocking agents, thereby impacting the later uptake more than the initial uptake. Isotype control and Fc blocking antibody injected animals exhibited no difference in NL4.3-mCherry uptake ( Next, we tested the impact of MFG-E8 deficiency on VLP uptake. In contrast to our expectation, the injected NL4.3-GFP still accumulated on the Mfg-e8 -/-SCSMs at 20 hr post injection. Yet, the MFG-E8 deficiency markedly impaired FDC NL4.3-GFP loading ( Figure 4C, Figure 4-figure supplement 4). Perhaps due to the poor FDC delivery, the Mfg-e8 -/-SCSMs accumulated more VLPs than did the controls ( Figure 4D). As we noted numerous CD169 + blebs on LN follicle FDCs in both the control and KO mice (Figure 4-figure supplements 4 and 5), we normalized the VLP signal to the level of CD169 + blebs. Injecting recombinant MFG-E8 along with NL4.3-GFP into Mfg-e8 -/mice partially reconstituted the MFG-E8 high compartment on the SCSMs and helped transfer some of the VLPs to the FDC network ( Figure 4E and Figure 4-figure supplement 6). In addition, injecting labeled MFG-E8 into the Mfg-e8 -/mice together with NL4.3-GFP allowed visualization of MFG-E8/ NL4.3-GFP complexes in the subcapsular sinus and on the SCSMs (Video 3). Together this data suggests that SCSM avb3/5 integrins can shuttle MFG-E8 bound material to the MFG-E8 + compartment. This compartment is open to lymph node FDCs and B cells (see below). SCSMs that lack functional av integrins to bind MFG-E8/NL4.3-GFP complexes cannot load the MFG-E8 + compartment or FDCs.
Video 2. 3D-reconstruction of thick section images shows NL4.3-GFP overlying the SCSM MFG-E8 + sites, which we termed the MFG-E8 + compartment. Animation shows volume image of thick section confocal image and its 3D-reconstruction. Thick section from inguinal LN collected 3 hr after VLPs injection was stained with FDC-M1 (MFG-E8, blue). NL4.3-GFP (green) and MLV-RFP (red) in MFG-E8 + compartment of SCSMs are indicated with arrows. We remained perplexed by the dissociation between the integrin blockade experiments and the results with the MFG-E8 deficient animals. Perhaps the NL4.3-GFP VLPs acquired low amounts of MFG-E8 during their preparation, which helped promote their capture. The producing cell line HEK293T are routinely cultured in media containing fetal calf serum, a potential MFG-E8 source. To verify we could image an antibody bound to the VLPs, we imaged the HEK293T cells expressing NL4.3-GFP after adding directly labeled VRC01 antibody to the culture. In contrast to negative controls, we found that the majority of the VLPs in the cell culture had bound the VRC01 antibody ( Figure 5A . Next, we added directly labeled antibody against MFG-E8 to a HEK293T culture expressing NL4.3-GFP. Again, many of the NL4.3-GFPs bound the MFG-E8 antibody indicating that had sufficient bound MFG-E8 to allow their detection ( Figure 5B, Figure 5-figure supplement 1E and Video 4). Alternatively, the MFG-E8 related protein EDIL-3 Hidai et al., 1998) or other envelope integrin binding proteins could compensate for the loss of MFG-E8 in the deficient animals. To test whether EDIL-3 might substitute for MFG-E8 we tested whether recombinant EDIL-3 competed for the binding of MFG-E8 to human monocytes. Adding a 1:1 molar ratio of EDIL-3 to MFG-E8 reduced MFG-E8 binding to human monocytes by approximately 50% ( Figure 5C,

Immune complexes co-localize with HIV-1 VLPs on SCSMs and, also trigger the transient abutment of FDCs and SCSMs
Previous studies have shown that SCSM also captures ICs for delivery to FDCs (Phan et al., 2009;Phan et al., 2007;Szakal et al., 1983). To investigate MFG-E8 + compartment involvement in IC uptake and transfer to FDCs, we co-injected preformed phycoerythrin (ICPE) with NL4.3-GFP. Surprisingly, we found a strong overlap between the NL4.3-GFP and ICPE SCSM acquisition sites ( Figure 6A). Presumably, the SCSMs use different mechanisms for capture, but both types of antigens may move through the MFG-E8 + compartment. Previous studies have not reported direct SCSM-FDCs contact perhaps due to their fleeting nature. Intravital imaging the FDC network beginning 45 min after injection of ICPE showed a dynamic FDC network shift towards SCSMs ( Figure 6B,    Figure 6C and Video 6). Hypothesizing that the movement of the FDC network depended upon G-protein coupled receptor signaling we inhibited Ga i signaling by injecting pertussis toxin (PTx) near the inguinal LN 2 hr before NL4.3-GFP VLPs. Examining lymph node sections prepared 6 hr later revealed that SCSM had collected the NL4.3-GFP VLPs, but the FDC network had not shifted toward the SSCMs nor had VLPs accumulated on the FDCs ( Figure 6D, Figure 6-figure supplement 2). Furthermore, we tested that PTx impact on FDC movement in real time by intravital imaging (Video 7). FDCs and adaptively transferred B cells motility was abolished about 1 hr after PTx treatment. An established abutment of SCSM/FDC by ICPE injection started to dissociate about 2 hr after PTx treatment. These results demonstrate that the antigen established transient abutments of FDCs and SCSMs are temporal and controlled by Ga i signaling.

gp120-specific B cells extract HIV-1 VLPs from SCSMs
Finally, to test whether the SCSM MFG-E8 + compartment supported NL4.3-GFP transfer to follicular B cells, we adoptively transferred fluorescently labeled wild type and transgenic HIV-1 gp120-      specific b12 B cells (Ota et al., 2013) the day prior to injecting NL4.3-GFP VLPs. Lymph node section prepared 3 hr after VLP injection revealed that the transgenic B cells localized at the B-T border, along the inter-follicular channel, and near the SCSMs. In contrast, the transferred wild type B cells tended more towards the center of the follicle, showing little sustained interest in the VLP bearing SCSMs ( Figure 7A and Video 8). Intravital imaging showed b12 B cells extending filopodia to actively probe the NL4.3-GFP bearing SCSMs. Following VLP contact, the B cells reverse direction pulling the VLP now localized on its uropod (Video 9) perhaps explaining why B cells carry antigen on their uropod. Zooming in revealed b12 B cells directly contacting the SCSMs bearing NL4.3-GFP VLPs ( Figure 7B).
Reconstruction of the imaging data shows a b12 B cell extending a filopodium to directly contact an MFG-E8 + SCSM bearing NL4.3-GFP VLPs ( Figure 7C and Video 10). We also added b12 B cells to the NL4.3-GFP transfect HEK293T cell cultures along with directly labeled MFG-E8, this allowed us to visualize the transgenic b12 cells capture MFG-E8 coated NL4.3-GFP VLPs ( Figure 7D) Intravital imaging results indicated that the b12 B cells were attracted to, and in contact with NL4.3-GFP bearing SCSMs, while wild type B cells failed to develop sustained interactions ( Figure 7E and Video 8). We compared the B cells behavior patterns by examining the intravital imaging generated cell tracks. Although the WT and b12 B cell tracks did not have significantly different displacements, the b12 B cells had a reduced mean speed, and turned more frequently (moved less straight).When we focused on those tracks that contacted SCSMs we found that the b12 B cell interacted with the VLP bearing SCSMs for significantly longer times than did the WT B cells ( Figure 7F). These results demonstrate that B cells with antigen reactive receptors can use the SCSM MFG-E8 + compartment to retrieve viruses.

Discussion
Although FDCs are a well-known HIV-1 reservoir, how FDCs initially acquire and preserve HIV-1 remains poorly understood. In this study we used HIV-1 VLPs and mice as a model system to visualize the likely early events in the uptake of HIV-1 virions transiting the lymph and blood by sinus lining macrophages and their subsequent transfer to underlying FDC networks. We identified a mechanism by which an avb3 integrin/MFG-E8 shuttle transfers lymph born HIV-1 from SCSM onto FDCs. Furthermore, we suggest that the SCSM MFG-E8 enriched compartment functions as a general antigen transfer portal for B cells and the underlying FDCs. In this paradigm FDCs dynamically react to the delivery of antigen to the overlying SCSMs.
The afferent lymphatics carry lymph from distal sites to the SCS. The lymph flows through the sinus eventually entering the medullary sinuses before exiting the lymph node via the efferent lymphatics. Some lymph-borne material enters the lymph node by exiting the SCS via the conduits. The molecular mass of antigen is a critical factor for how the microarchitecture handles and distributes the antigen (Gerner et al., 2017). Lymph-borne viral particles and ICs are too large to enter the conduits. They can be captured by the SCSMs that overlie the lymph node follicle for delivery to underlying cells (Carrasco and Batista, 2007;Fossum, 1980;Friedlaender and Baer, 1972;Nossal et al., 1965). Fc and complement receptors assist SCSM in IC capture (Junt et al., 2007). CD169 expression helps SCSM capture of HIV-1 and MLV VLPs. Treating mice with an anti-CD169 antibody significantly reduced VLP capture (Sewald et al., 2015). The later study found the captured VLPs had accumulated in a virus containing compartment (VCC) contiguous with the SCSM plasma membranes. Consistently, our study found that lymph and blood borne VLPs accumulated on SC and MZ sinus macrophages, respectively, at sites enriched for MFG-E8. This finding contrasts with our previous report that locally injected gp120 preferentially localized on SIGN-R1 + macrophages  and indicates that the VLP gp120 expression does not mediate their uptake by SCSMs. Confirming that VLP gp120 expression does not affect their SCSM uptake, NL4.3-GFP VLPs lacking gp120 exhibited a similar uptake pattern as did the gp120 expressing VLPs (data not shown). This also agrees with the above study, which noted HIV-1 VLP SCSM acquisition did not depend upon gp120 (Sewald et al., 2015). Interestingly, a study using human genetic diversity found that natural CD169 null individuals infected with HIV-1 had no measurable impact on HIV-1 acquisition and progression (Martinez-Picado et al., 2016). This argues that early myeloid HIV-1 capture can likely occur by other mechanisms. Together these studies indicate that SCSMs employ a variety of strategies to capture viral particles and that CD169 and MFG-E8 have overlapping roles perhaps functioning at different stages of the capture process.
Ultrastructural studies using electron microscopy to study HIV-1-infected macrophages revealed immature and mature virions seemingly in intracellular vesicles, which resembled late endosomes (LEs) or multivesicular bodies (MVBs) (Gendelman et al., 1988;Wu and Shroff, 2018). These correspond to the VCC described above and despite their similarity to LEs and MVBs, they are distinct structures that have a complex three-dimensional form, a unique set of protein markers, and several identifying features including a near-neutral pH and the presence of connections to the extracellular milieu. While initially described in response to HIV-1 infection, VCC-like compartments exist in uninfected macrophages (Tan and Sattentau, 2013). They often contain other vesicular structures including exosomes (Wu and Shroff, 2018). The MFG-E8 enriched compartment described in this study likely correspond to the macrophage VCC. While additional studies are needed to confirm their identity, the SCSM MFG-E8 + compartment functioned as a niche for VLP accumulation. Our findings also suggest a constitutive, ongoing transport of MFG-E8 bound material from the lymph through this compartment. Supporting this supposition, MFG-E8 bearing SCSMs are evident in the absence of exogenous immune stimulus. The direct transfer of MFG-E8 bearing VLPs to FDCs occurred. The local injection of recombinant MFG-E8 into Mfg-e8 -/animals led to the accumulation of MFG-E8 on SCSMs and the eventual appearance of MFG-E8 on the FDC network. MFG-E8 may also be transported in a retrograde fashion from FDCs to the SCSM. Charging the SCSM avb3 receptors with MFG-E8 would provide a mechanism to capture HIV-1 virions that lacked bound MFG-E8.
In infected patients, HIV-1 virions accumulate and persist on FDCs for months often as ICs. FDCs endocytose HIV-1 ICs into non-degradative compartments, thereby contributing to the persistence of the virions (Hart et al., 1991;Heesters et al., 2015;Smith et al., 2001;Smith-Franklin et al., 2002). An established mechanism to deliver SCSM captured ICs to the underlying central FDC network is via the trafficking of noncognate B cell, which capture the ICs using complement receptors (Phan et al., 2009;Phan et al., 2007). However, early in the infection prior to IC complex formation other mechanisms likely exist to help transfer antigens from the SCSMs to FDCs. Furthermore, B cell-mediated IC transfer may be inadequate to transfer large particulate antigens such as a virion. In our studies we failed to observe non-cognate B cell uptake and transfer of HIV-1 VLPs to the FDC network (Park, unpublished observation). Rather we consistently observed an early, but transient abutment of the FDC network to the overlying SCSMs, thereby allowing the direct transfer of HIV-1 VLPs to FDCs. Within 24 hr the FDC network had retracted back to its usual location in the center of the follicle accompanied by the captured VLPs. Notably, interfering with the SCSM/FDC abutment by pre-injecting pertussis toxin severely reduced the rapid VLP loading on the FDCs. The reduced SCSM/FDC abutment by pertussis toxin argues that heterotrimeric Gi protein-mediated signaling contributes to the FDC movement. The most intriguing possibility is a chemoattractant directed movement of the FDC network towards the SCSMs triggered by VLP uptake. However, since the local pertussis toxin injection will affect all the LN cells near the SCS, we cannot conclude that the FDC movement and severe reduction in FDC/VLP loading results solely from reducing the SCSM/ FDC abutment. Interestingly, a similar transient FDC-SCSM abutment occurred following the administration of ICs. Identification of the mechanisms underpinning the FDC network movements may provide a means to control the loading of antigen onto FDCs.  In contrast to our observations, a study using bacteriophage Qb VLPs demonstrated that noncognate B cells captured the VLPs and transported them to splenic FDCs. B cell uptake depended upon natural IgM antibody and complement (Link et al., 2012). The authors did not examine macrophage uptake and confined their analysis to the spleen 24 hr after VLP injection. Bacteriophage Qb VLPs are composed of a T = 3 icosahedral lattice of coat proteins assembled with genomic RNA in absent of lipid membrane with a diameter of 28 nm (Gorzelnik et al., 2016). These VLPs differ substantially from HIV-1 VLPs as they lack lipid components, are unlikely to bind MFG-E8, and are much smaller (150-200 nm versus 28 nm). These differences likely result in alternative mechanisms of VLP capture and transfer to FDCs. Nevertheless, it would be of interest to determine whether the Qb VLPs are captured by LN SCSMs and to visualize how they are transported to LN FDCs.
While SCSM use different strategies to capture ICs and the HIV-1 VLPs, both entered the MFG-E8 + compartment on SCSM. This raised the possibility that this compartment functions as a portal for the transfer of SCSM captured antigens to FDCs and cognate B cells. That MFG-E8 may have a role in the transfer is supported by the poor loading of VLPs on to FDCs in the MFG-E8 deficient mice despite the adequate capture by the SCSMs. Our imaging data clearly shows the direct transfer of both ICs and VLPs from SCSMs to FDCs. An obvious question is how the FDCs acquire the ICs and VLPs from the SCSMs. In the case of ICs FDC complement receptors likely contribute, however, for VLPs the answer is unknown. Still to be determined is whether the VLPs remains MFG-E8 bound and what the role of the avb3 integrins play in the MFG-E8 + compartment. Since avb3 integrins support macrophage phagocytosis the SCSM VLPs may internalize the VLPs prior to delivery to the VCC much like exosomes are delivered to the VCC. Furthermore, the binding of the MFG-E8 virion complex to the SCSM avb3 integrins may signal the cells via its known role in triggering Rac1 activation (Mao and Finnemann, 2012).
Cognate B cells with gp120 reactive B cell antigen receptors (BCRs) readily acquired NL4.3-GFP VLPs directly from the SCSMs. We found b12 B cells extending filopodia to directly contact MFG- https://elifesciences.org/articles/47776#video6 E8 + SCSM bearing NL4.3-GFP VLPs. Eventually, the b12 B cells reversed direction pulling the VLP away from the macrophage. Individual gp120 reactive B cells could acquire and simultaneously transported multiple HIV-1 VLPs on their uropod. This indicates that VLPs in MFG-E8 + compartment retain gp120 recognizable by the b12 B cell BCR. Our in vitro studies suggest that the b12 B cell BCRs can recognize gp120 even when the VLPs are MGF-E8 bound. Yet whether the binding of MFG-E8 to the HIV-1 virions can shape the viral epitopes available to trigger a humoral response remains an open question. While noncognate B cells transiently interacted with SCSMs bearing NL4.3-GFP VLPs, they rarely extracted VLPs from them. BCR triggered F-actin remodeling and myosin-and dynein-mediated contractility are known to contribute to the traction force generation needed for antigen extraction (Wang et al., 2018). BCR engagement and antigen acquisition likely altered the behavior of the cognate B cells as they slowed and focused their interest on the SCSM bearing VLPs as compared to B cells that lacked gp120 reactive BCRs. Beside engaging the gp120-specific BCRs the MFG-E8 bound VLPs are likely also recognized by avb3 integrins expressed by B cells. The engagement of these integrin receptors can be immunosuppressive. The deletion of av or b3 in B cells enhances the humoral response to antigens containing TLR ligands and to the influenza virus. The enhancement results from the loss of an integrin-mediated negative regulatory signal (Raso et al., 2018;Wang et al., 2014).
VLPs have been employed as a vaccine strategy in the hope that could advance the elicitation of broadly neutralizing antibodies (bNAbs) against HIV-1 envelope proteins. Multiple reports have demonstrated that HIV-1 VLPs can induce strong antibody responses in small animal models and macaques (Tong et al., 2014), (Buonaguro et al., 2012). Furthermore, expertise in the generation and usage of VLPs is evolving towards the goal of producing a safe mimic of real HIV-1 virions (Gonelli et al., 2019). However, care in the evaluation of HIV-1 VLPs vaccine studies is needed. Variation in packaging cell lines and animal models chosen may result in significant variation and crossspecies-specific immune responses. Our study utilized non cell-associated VLPs as a mimic of the early stages of HIV-1 virion dissemination in lymphoid organs. While plasma contains significant levels of cell free virus there is little information on the amount of free virus or cell associated virus in the lymph. Yet as much of the plasma virus is derived from tissue sources, it seems likely that it has transited through the lymph and lymph nodes. Limitations of our study include a reliance on a nonreplicating VLP rather than a replicating virus. As such to mimic the transit of infectious virions in the blood and lymph, we had to directly inject the VLPs. Another limitation was the choice of a mouse model. Yet this also provided us with enhanced imaging capability and the ability to use B cells with a defined antigen receptor specificity.
Based on our observations, we propose that MFG-E8 blood levels and lymphoid organ sources of MFG-E8 ensure that most cell-free HIV-1 virus in infected individuals is exposed to MFG-E8. HIV-1 envelope PS provides a means for MFG-E8 binding and MFG-E8 links the virions to av integrins on host cells. Whether this MFG-E8 opsonization benefits the host or the virus needs further study. MFG-E8 opsonization does promote the delivery of fluid borne virions to SMs resulting in their accumulation in an MFG-E8 + compartment. This compartment is accessible to antigen-reactive B cells and transiently to FDC processes. Direct SM-FDC virion transfer depends upon a short-lived FDC Video 7. The impact of pertussis toxin on FDC network and B cell movement in a follicle. In part 1, TP-LSM live imaging movies in upper row show the inguinal LN at the subcapsular sinus (SCS) area for 40 min from 1.5 hr after ICPE (red) injection. FDC (green) contacted and fluctuated around ICPE (red) bearing SCSM. B cells (blue) adoptively transferred the previous day outline a B cell follicle. In the lower row movies, the same subcapsular sinus (SCS) area is shown for 40 min 1 hr after pertussis toxin (PTx, 500 ng) injection. Evans blue (red) was co-injected with PTx to visualize the SCS. Indicated numbers in each movie show the distance from SCSM floor to imaged area. All movies have 15 mm of thickness. In part 2, two comparable videos from part one were enlarged and to show the movement of FDCs and B cells. Arrow in PTx treated video shows loss of transient abutments. Time counter is hour: minute:second.
https://elifesciences.org/articles/47776#video7 network abutment, likely triggered by SCSM antigen uptake. Subsequently, the FDCs network contracts toward the follicle center along with its acquired antigens. This provides a mechanism for rapid FDC loading broadening the opportunity for rare, antigen reactive follicular B cells to acquire antigen, but also a means for HIV-1 virions to accumulate on the FDC network. Continued on next page