TIM-1 Mediates Dystroglycan-Independent Entry of Lassa Virus

PtdSer receptors, such as TIM-1, are emerging as critical entry factors for many enveloped viruses. Most recently, hepatitis C virus and Zika virus have been added to a growing list. PtdSer receptors engage with enveloped viruses through the binding of PtdSer embedded in the viral envelope, defining them as GP-independent receptors. This GP-independent entry mechanism should effectively mediate the entry of all enveloped viruses, yet LASV GP-pseudotyped viruses were previously found to be unresponsive to PtdSer receptor enhancement in HEK 293T cells. Here, we demonstrate that LASV pseudovirions can utilize the PtdSer receptor TIM-1 but only in the absence of appropriately glycosylated α-dystroglycan (αDG), the high-affinity cell surface receptor for LASV. Our studies shed light on LASV receptor utilization and explain why previous studies performed with α-DG-expressing cells did not find that LASV pseudovirions utilize PtdSer receptors for virus uptake.

ABSTRACT

Lassa virus (LASV) is an Old World arenavirus responsible for hundreds of thousands of infections in West Africa every year. LASV entry into a variety of cell types is mediated by interactions with glycosyltransferase LARGE-modified O-linked glycans present on the ubiquitous receptor α-dystroglycan (αDG). However, cells lacking αDG are permissive to LASV infection, suggesting that alternative receptors exist. Previous studies demonstrated that the phosphatidylserine (PtdSer)-binding receptors Axl and Tyro3 along with C-type lectin receptors mediate αDG-independent entry. Here, we demonstrate that another PtdSer receptor, TIM-1, mediates LASV glycoprotein (GP)-pseudotyped virion entry into αDG-knocked-out HEK 293T and wild-type (WT) Vero cells, which express αDG lacking appropriate glycosylation. To investigate the mechanism by which TIM-1 mediates enhancement of entry, we demonstrate that mutagenesis of the TIM-1 IgV domain PtdSer-binding pocket abrogated transduction. Furthermore, the human TIM-1 IgV domain-binding monoclonal antibody ARD5 blocked transduction of pseudovirions bearing LASV GP in a dose-dependent manner. Finally, as we showed previously for other viruses that use TIM-1 for entry, a chimeric TIM-1 protein that substitutes the proline-rich region (PRR) from murine leukemia virus envelope (Env) for the mucin-like domain served as a competent receptor. These studies provide evidence that, in the absence of a functional αDG, TIM-1 mediates the entry of LASV pseudoviral particles through interactions of virions with the IgV PtdSer-binding pocket of TIM-1.

IMPORTANCE PtdSer receptors, such as TIM-1, are emerging as critical entry factors for many enveloped viruses. Most recently, hepatitis C virus and Zika virus have been added to a growing list. PtdSer receptors engage with enveloped viruses through the binding of PtdSer embedded in the viral envelope, defining them as GP-independent receptors. This GP-independent entry mechanism should effectively mediate the entry of all enveloped viruses, yet LASV GP-pseudotyped viruses were previously found to be unresponsive to PtdSer receptor enhancement in HEK 293T cells. Here, we demonstrate that LASV pseudovirions can utilize the PtdSer receptor TIM-1 but only in the absence of appropriately glycosylated α-dystroglycan (αDG), the high-affinity cell surface receptor for LASV. Our studies shed light on LASV receptor utilization and explain why previous studies performed with α-DG-expressing cells did not find that LASV pseudovirions utilize PtdSer receptors for virus uptake.

INTRODUCTION

The Old World arenavirus Lassa virus (LASV) causes hemorrhagic fever in humans, with estimates of 300,000 infections and 5,000 deaths each year in West Africa (Viral Hemorrhagic Fever Consortium). Unfortunately, no vaccine has been approved for human use. Arenaviruses contain four viral open reading frames in two ambisense RNA genomic segments (1), one of which encodes the glycoprotein precursor (for a recent review, see reference 2). Cellular processing of this precursor protein generates the LASV glycoprotein (GP) complex that is found on the surface of arenavirions (3), composed of a stable signal peptide, GP1, and GP2 (4). GP1 on Old World and clade C New World arenaviruses has been shown to bind with high affinity to α-dystroglycan (αDG) on the surface of cells (5, 6). The virus enters into the endosomal compartment through this interaction (7). Under the low-pH conditions found in endosomes, GP1 releases αDG and binds to LAMP1, and this latter interaction is required for productive LASV infection (8; for a recent review, see reference 9).

The ubiquitous dystroglycan (DG) complex consists of two subunits (α and β), which are cotranslated and proteolytically cleaved into two proteins that are associated at the plasma membrane (10, 11). Together, αDG and βDG link the intracellular cytoskeleton with the extracellular matrix (11, 12). βDG contains an N-terminal transmembrane domain and a proline-rich C-terminal cytoplasmic tail, whereas αDG is completely extracellular. The mature αDG contains a highly O-glycosylated mucin domain with a carboxy-terminal globular domain, and αDG is noncovalently associated with βDG (13–15). Many of the O-linked glycans found on αDG are O-mannosyl carbohydrates (16, 17), and modification of these O-glycans by the glycosyltransferase LARGE is essential for αDG to bind to its ligand laminin (18–20). LASV GP binding is known to compete with the αDG ligand laminin (7, 21) for αDG binding, and modification of mannosyl O-glycans by the glycosyltransferase LARGE is critical for interactions with viral glycoprotein (19, 22).

A growing number of enveloped viruses are now appreciated to enter some cell populations through phosphatidylserine (PtdSer) receptor interactions. While a variety of PtdSer receptors are found in mammals (23), the T cell, immunoglobulin, and mucin (TIM) family and the Tyro3, Axl, and Mer (TAM) family of PtdSer receptors have been most strongly implicated in mediating enveloped virus entry through interactions with PtdSer on the outer leaflet of the viral membrane. We have termed these receptors PVEERs (PtdSer-mediated virus entry-enhancing receptors) (24, 25). The TIM receptors directly interact with PtdSer on apoptotic bodies or membrane-associated viruses (24, 26–30), whereas the TAM receptors bind to one of two serum proteins, Gas6 or protein S, which, in turn, bind to PtdSer (31–34). Not surprisingly, given their direct interaction, TIM receptors are thought to have a higher affinity for PtdSer than TAM receptor complexes (35). TIM receptors are type I, cell surface glycoproteins that contain an amino-terminal immunoglobulin variable (IgV)-like domain, a heavily O-glycosylated mucin-like domain (MLD), a transmembrane domain, and a cytoplasmic tail (36). A PtdSer-binding pocket is positioned within the IgV domain (26, 27, 37, 38), and residues within this pocket are required for TIM binding of virions and apoptotic bodies as well as its role in immune cell regulation (27, 30, 39–44).

Recently, the expressions of two TAM receptors, Axl and Tyro3, and two C-type lectins, DC-SIGN and LSECtin, were shown to confer αDG-independent cellular entry of LASV pseudovirions into cell lines that do not have LARGE-modified O-linked glycans on αDG, thereby making αDG unavailable for LASV binding (45). C-type lectin are thought to interact with LASV GP-containing virions through the binding of glycans on the LASV glycoprotein (46). In contrast, it would be predicted that the TAM receptors bind to PtdSer on LASV pseudovirions through interactions with Gas6 or protein S, although this was not investigated in this study. The role of PtdSer receptors for the entry of LASV virions remains controversial. Some investigations have reported that pseudovirions or chimeric viruses bearing LASV GP do not utilize PtdSer receptors (24, 28, 47), yet Shimojima et al. demonstrated that the exogenous expression of Axl and Tyro3 supports LASV pseudovirion entry into Jurkat T cells (45). The importance of TAM receptors for the Old World arenavirus lymphocytic choriomeningitis virus (LCMV) in vivo remains unclear, as Sullivan et al. demonstrated that Axl knockout (Axl-KO) mice are readily susceptible to LCMV (48). A number of the studies indicating that Axl does not mediate LASV pseudovirion entry were performed with cells that expressed wild-type (WT) αDG. Hence, the use of alternative receptors by LASV may occur only when functional αDG is not present. Consistent with this, Fedeli et al. recently demonstrated that Axl serves as a LASV receptor in cells where functional αDG is either absent or present at low levels (49).

In this study, we found that that PtdSer receptor TIM-1 mediates the entry of either LCMV or vesicular stomatitis virus (VSV) pseudovirions bearing LASV GP but only when αDG either is not expressed or does not contain the necessary LARGE-dependent alterations of the O-linked glycans. This is consistent with findings that the high-affinity interactions of LASV GP and αDG prevail over lower-affinity PtdSer/PtdSer receptor interactions (49). Furthermore, we found that the TAM receptor Axl was unable to serve as a receptor for LASV pseudovirions in HEK 293T and Vero cells, irrespective of the status of αDG in these cells.

RESULTS

LASV entry is TIM-1 dependent in Vero cells.

Multiple lines of evidence indicate that αDG is not the only receptor available to Old World arenaviruses (45, 49–51), although when appropriately glycosylated, αDG binds to LASV GP with high affinity and mediates Old World arenavirus entry (21, 22). Although αDG is widely expressed throughout the body, some cell types do not glycosylate αDG in a way that is compatible with LASV GPC engagement and laminin binding (22). As Vero cells are readily permissive to LASV but are not sensitive to laminin-mediated competition (22), we assessed the ability of mAb IIH6 to bind to Vero cells. IIH6 has been previously shown to distinguish if αDG is glycosylated in a LASV GPC-compatible manner (22, 52). Surface staining of cells with IIH6 demonstrated that suitably glycosylated αDG was detected on HEK 293T cells, but not Vero cells, yet both cell types had readily detectable dystroglycan on their surface (Fig. 1A). These findings are consistent with previous studies proposing that αDG is not used by LASV for entry into Vero cells (22, 45).

FIG 1.

LASV pseudovirion entry is TIM-1 dependent in Vero cells. (A) Cell surface detection of endogenous DG expression on Vero or HEK 293T cells. Live cells were stained with polyclonal DG antisera (top) or the mannosylation-dependent anti-αDG mAb IIH6 (bottom). Filled histograms represent cells stained with isotype control antisera or mAb; unfilled histograms represent cells stained with anti-DG polyclonal antisera or IIH6. (B) Cell surface expression of endogenous TIM-1, Axl, Mer, Tyro3, or TIM-4 on WT Vero cells. Also shown are TIM-1 and Axl staining of Axl-KO, TIM-KO₁, TIM-KO₂, or Axl/TIM-KO (DKO) Vero cells. Filled histograms represent cells stained with the isotype control; unfilled histograms represent cells stained with 2 μg/ml anti-TIM-1, anti-Axl, anti-Mer, anti-Tyro3, or anti-TIM-4 polyclonal antibody. (C and D) LASV-VSV (C) or adenovirus-eGFP (D) transduction of WT, TIM-1-KO, Axl-KO, or DKO Vero cells. Serial dilutions of virus were applied to each cell type in duplicate. Transduction is represented as the percentage of cells that are eGFP positive, as assessed by flow cytometry. Data points are representative of results from three independent experiments with the indicated standard errors of the means (SEM). (E) LASV-LCMV transduction (MOI = ∼0.2) of WT, TIM-1-KO, Axl-KO, or DKO Vero cell lines. Data points are representative of results from three independent experiments with the indicated SEM. (F and G) Ability of adenovirus vector (Ad)-delivered TIM-1 to enhance transduction of LASV-VSV in WT, TIM-1-KO, or TIM-1- and Axl-KO (DKO) Vero cells. Forty-eight hours following Ad transduction, cells were divided and assessed independently for TIM-1 expression (F) or LASV-VSV transduction (G). Bars in each panel represent results from one of three independent experiments. (F) Cell surface expression of TIM-1 48 h following transduction of WT or KO Vero cells with Ad-TIM-1 (MOI of 2, 4, or 8). Cells were surface stained with 2 μg/ml anti-TIM-1 polyclonal antibodies. TIM-1 expression is represented as geometric mean fluorescence index (MFI) values. (G) LASV-VSV transductions of empty (E) or TIM-1-expressing (T) adenovirus vector-transduced WT or KO Vero cells. Equivalent volumes of LASV-VSV were applied to cells in duplicate. Transduction is represented as the percentage of eGFP-positive cells, as assessed by flow cytometry.

To determine if PtdSer receptors are involved in LASV entry into Vero cells, we assessed the surface expression of those PtdSer receptors most highly implicated in mediating enveloped virus entry (24, 28, 53–55) (Fig. 1B). Because Vero cells are known to express abundant Axl at the cell surface (53), others have proposed that Axl may mediate the entry of LASV into Vero cells (45). We found that both TIM-1 and Axl were highly expressed on the surface of WT Vero cells, whereas other PtdSer receptors implicated in virus uptake (Tyro3, Mer, or TIM-4) had little or no expression.

To determine if Axl and/or TIM-1 serves as a receptor for LASV-VSV entry in Vero cells, we used CRISPR-Cas9 gene-editing technology to create both AXL and HAVCR1 (TIM-1) knockout (KO) Vero cell lines. We produced one Axl-KO cell line, two independently derived TIM-1-KO cell lines, TIM-KO₁ and TIM-KO₂, and one Axl/TIM-1 double-KO (DKO) cell line. Successful KO was confirmed by cell surface staining with antibodies targeting the respective receptors (Fig. 1B). Altered expression of DG, Mer, Tyro3, or TIM-4 was not observed in any of the KO cell lines, and all cell lines divided at similar rates (data not shown).

The impact of Axl or TIM-1 expression on LASV entry was assessed using two different types of virus particles. Pseudovirions generated from the Old World arenavirus LCMV and from the unrelated rhabdovirus VSV were pseudotyped with LASV GP and transduced into our Vero cell lines. Both pseudovirion systems expressed green fluorescent protein (GFP) as a reporter gene. Regardless of the viral particle bearing LASV GP, the loss of TIM-1 in Vero cells abrogated transduction, whereas the knockout of Axl expression had no effect (Fig. 1C and E). The DKO line that did not express either TIM-1 or Axl gave findings similar to those with TIM-1-KO lines alone, providing additional evidence that Axl is not being used by LASV-VSV for entry into Vero cells. The loss of TIM-1 expression in Vero cells did not affect the entry of all viruses, as the transduction of an adenoviral vector expressing enhanced GFP (Ad-eGFP) was not affected by the loss of TIM-1 (Fig. 1D). To examine if the inability of Axl to support LASV-VSV transduction was due to low concentrations of Gas6 in our fetal calf serum (FCS), we transduced LASV-VSV into the TIM-1-KO lines in the presence of exogenous Gas6. Gas6 had no effect on levels of virus transduction (data not shown).

To confirm that the loss of TIM-1 expression was the cause of LASV transduction resistance in Vero cells, we performed a dose-dependent rescue experiment with adenovirus-delivered TIM-1 (Ad-TIM-1). While WT levels of TIM-1 expression in Ad-TIM-1-transduced cells were not achieved in these studies, increasing levels of TIM-1 expression at the cell surface (Fig. 1F) correlated with increasing LASV-VSV transduction (Fig. 1G) in the Vero TIM-1-KO cell lines and the Axl/TIM-1-KO cell line.

LASV pseudovirion entry is αDG dependent in HEK 293T cells.

We further evaluated the impact of PtdSer receptors on LASV pseudovirion transduction in HEK 293T cells. For these studies, we initially validated the importance of αDG for LASV-VSV entry. In these studies, we used CRISPR-Cas9 editing to knock out αDG expression in HEK 293T cells, generating two independent DG-KO cell lines, DG-KO₁ and DG-KO₂. DG-KO lines were surface stained with anti-DG antibodies to confirm the loss of αDG expression at the cell surface (Fig. 2A), with WT HEK 293T cells being used as a positive control. Increasing concentrations of LASV-VSV or VSV pseudotyped with VSV G (VSV-VSV) were incubated with WT or αDG-KO lines. As expected, the loss of αDG had a profound impact on the transduction of pseudovirions bearing LASV GP (Fig. 2B) but not VSV-VSV (Fig. 2C). To confirm that the loss of αDG was responsible for the reduction of LASV-VSV transduction in the KO lines, we performed dose-dependent rescue experiments with exogenous αDG. The endogenous expression level of αDG was high on WT HEK 293T cells (Fig. 2A), and transfection of an exogenous DAG1-expressing plasmid did not significantly increase either αDG expression or LASV-VSV transduction (data not shown). However, in the DG-KO lines, increasing levels of DG expression at the cell surface (Fig. 2D) correlated with increasing permissivity to LASV-VSV transduction (Fig. 2E), with statistically significant increases in LASV-VSV transduction in the DG-KO₁ line and a similar trend observed in the second KO line. These data support a role for αDG as a critical receptor for LASV-VSV entry in HEK 293T cells.

FIG 2.

LASV-VSV entry is αDG dependent in HEK 293T cells. (A) Cell surface expression of endogenous DG on WT or DG-KO HEK 293T cells. Filled histograms represent cells stained with the isotype control; unfilled histograms represent cells stained with 2 μg/ml anti-DG polyclonal antibodies. (B and C) LASV-VSV (B) or VSV-VSV (C) transduction in WT or DG-KO HEK 293T cells. Serial dilutions of pseudovirus were applied to each cell type in duplicate. Transduction is represented as the percentage of eGFP-positive cells relative to 1 μl of pseudovirus in WT HEK 293T cells, as assessed by flow cytometry. Shown are means and standard errors of the means of data from three to four independent experiments. (D and E) Dose-dependent rescue of LASV-VSV transduction in αDG-transfected DG-KO HEK 293T cells. Cells were transfected with 0, 0.25, 0.5, or 1 μg DG-expressing plasmid DNA (increase in DNA is indicated with unfilled triangles). After 48 h, cells were divided and assessed independently for the surface expression of DG (D) or LASV-VSV transduction (E). Shown are means and standard errors of the means of data from four independent experiments. (D) Cell surface expression of DG on DG-transfected DG-KO HEK 293T cells. Cells were stained with 2 μg/ml anti-DG polyclonal antisera, and values shown are mean fluorescence index (MFI) values. (E) LASV-VSV transduction of αDG-transfected DG-KO HEK 293T cells. Equivalent volumes of LASV-VSV were applied to cells in duplicate. Transduction is represented as the percentage of eGFP-positive cells relative to untransfected cells, as assessed by flow cytometry. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

TIM-1 can serve as a receptor for LASV pseudovirions in the absence of αDG.

We next sought to determine if PtdSer receptors mediate LASV pseudovirion entry into WT or DG-KO HEK 293T cells. Because HEK 293T cells do not endogenously express Axl or TIM-1 (Fig. 3A), we evaluated whether the ectopic expression of either PtdSer receptor mediates LASV pseudovirion entry. With the transfection of increasing concentrations of a TIM-1- or Axl-expressing plasmid, dose-dependent increases in the levels of these PtdSer receptors were observed on the surface of transfected HEK 293T cells, regardless of the αDG status of the cells (Fig. 3B and D). The presence of TIM-1 or Axl on WT HEK 293T cells did not increase levels of LASV-VSV transduction (Fig. 3C and E), indicating that the expression of αDG rendered the addition of these PtdSer receptors superfluous for LASV-VSV entry. In contrast, TIM-1 expression strongly correlated with increasing LASV-VSV transduction in both DG-KO cell lines. These data are consistent with our findings in Vero cells and indicate that in the absence of DG expression, pseudovirions bearing LASV GP use TIM-1 as a receptor. In contrast to our findings with TIM-1, increasing Axl expression at the cell surface (Fig. 2D) had no impact on LASV-VSV transduction in the DG-KO lines (Fig. 2E), despite the ability of Axl expression in HEK 293T cells to enhance Zika virus entry, as others have shown (56, 57), and to enhance Gas6-dependent cell migration (58) (Fig. 3F and G).

FIG 3.

TIM-1 serves as a receptor for LASV-VSV in the absence of αDG. (A) Cell surface expression of endogenous TIM-1 or Axl expression on WT HEK 293T cells. Filled histograms represent cells stained with the isotype control; unfilled histograms represent cells stained with 2 μg/ml anti-Axl or anti-TIM-1 polyclonal antisera. Analyses of Axl and TIM-1 surface expression on HEK 293T cells were performed in parallel with ectopic Axl and TIM-1 surface staining of transfected cells, shown as MFI in panels below. (B to E) Dose-dependent rescue of LASV-VSV transduction in TIM-1- or Axl-transfected WT or DG-KO HEK 293T cells. Each cell type was transfected with 0, 0.25, 0.5, or 1 μg TIM-1- or Axl-expressing plasmid DNA (increase in DNA is indicated with unfilled triangles). After 48 h, cells were divided and assessed independently for TIM-1 or Axl expression (B and D) or LASV-VSV transduction (C and E). Bars in each panel are representative of data from four independent experiments with the indicated SEM. (B and D) Cell surface expression analysis of TIM-1 or Axl expression on TIM-1- or Axl-transfected WT or DG-KO HEK 293T cells, respectively. Cells were stained with 2 μg/ml anti-TIM-1 or anti-Axl polyclonal antibodies. TIM-1 or Axl expression is represented as units of geometric means. (C and E) LASV-VSV transduction of TIM-1- or Axl-transfected WT or DG-KO HEK 293T cells, respectively. Equivalent volumes of LASV-VSV were applied to cells in duplicate. Transduction is represented as the percentage of eGFP-positive cells relative to untransfected cells, as assessed by flow cytometry. Shown are means and standard errors of the means of results from four independent experiments. (F) Axl expression in DG-KO HEK 293T cells enhances Zika virus infection. DG-KO cells were transfected with an empty vector or a vector expressing Axl. Zika virus (Puerto Rico strain) (MOI of ∼0.25) was added to Axl or empty plasmid-transfected DG-KO cells. Infected cells were lysed at day 3 of infection, and viral genome equivalents were determined by qRT-PCR. (G) HEK 293T cells transfected with empty, Axl-expressing, or murine BST-2-expressing plasmids were plated on the insert of a transwell in serum-free medium. Medium containing 10% FCS with or without 0.5 μg/ml of Gas6 was added to the bottom of the transwell apparatus. Cell migration to the bottom of the transwell filter was determined 24 h later. Data shown represent results of one of two studies that were performed in triplicate. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

Pseudovirions bearing LASV GP bind TIM-1 in a PtdSer-dependent manner.

We and others have reported that PtdSer on a variety of enveloped viruses binds to a conserved PtdSer-binding pocket present in the amino-terminal IgV domain of TIM proteins (24, 28, 55, 60). To assess whether the mechanism of TIM-1-mediated LASV pseudovirion transduction is similar to that of other enveloped viruses, we transduced LASV-VSV in TIM-1-expressing WT or DG-KO HEK 293T cells in the presence of anti-human TIM-1 monoclonal antibody (mAb) ARD5. ARD5 had no impact on LASV-VSV transduction in TIM-1-expressing WT HEK 293T cells, as LASV-VSV entry in these cells is αDG dependent, but it inhibited LASV-VSV transduction in TIM-1-expressing DG-KO HEK 293T cells in a dose-dependent manner (Fig. 4A). ARD5 also inhibited LASV-LCMV transduction in both WT and Axl-KO Vero cells (Fig. 4B). These data suggest that LASV pseudovirions interact with TIM-1 similarly to other enveloped viruses.

FIG 4.

TIM-1 mediates LASV pseudovirion transduction through a PtdSer-dependent mechanism. (A and B) Anti-human TIM-1 antibody ARD5 inhibits TIM-1-dependent LASV pseudovirion transduction in HEK 293T cells in a dose-dependent manner. (A) WT or DG-KO HEK 293T cells transfected with a TIM-1-expressing plasmid were incubated with serial dilutions of ARD5 prior to the addition of LASV-VSV. Relative transduction is represented as the percentage of eGFP-positive cells relative to cells incubated without ARD5, as assessed by flow cytometry. (B) WT or Axl-KO Vero cells were incubated with serial dilutions of ARD5 prior to the addition of LASV-LCMV, and data are shown as relative transduction, as described above. Data points are representative of results from three independent experiments with the indicated SEM. (C to F) Ability of two human TIM-1 mutants to support LASV-VSV transduction in DG-KO HEK 293T cells. Plasmids expressing a PtdSer pocket mutant, TIM-1 (ND115DN), or a chimeric TIM-1 that has the MLD replaced with the proline-rich region (PRR) of murine leukemia virus glycoprotein were transfected into WT or DG-KO cells. Cells were transfected with 0, 0.25, 0.5, or 1 μg ND115DN or PRR plasmid (increase in DNA is indicated with unfilled triangles). After 48 h, cells were divided and assessed independently for TIM-1 expression (C and E) or LASV-VSV transduction (D and F). Equivalent volumes of LASV-VSV were applied to cells in duplicate in transduction studies. Shown are fold change values relative to transduction levels of untransfected cells of the same type, as assessed by flow cytometry. Shown in panels A and B are means and standard errors of the means of data from three independent experiments. Shown in panels C to F are means and standard errors of the means of data from 3 to 5 independent experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

To further explore if a PtdSer-dependent mechanism is used by pseudovirions bearing LASV GPC in DG-KO HEK 293T cells, we employed a previously characterized TIM-1 mutant, ND115DN, in which critical residues in the PtdSer-binding pocket (N114 and D115) were mutated (24). As expected, increasing levels of ND115DN-TIM-1 expression at the cell surface (Fig. 4C) had no impact on the permissivity of WT or DG-KO HEK 293T cells to LASV-VSV transduction (Fig. 4D). These data suggest that TIM-1-mediated LASV-VSV transduction requires a functional PtdSer-binding pocket and is likely PtdSer dependent.

Unlike TAM receptors, TIM receptors have an MLD that contains an abundance of O-linked glycans (61, 62). Since αDG also has an MLD, and the interaction between LASV GPC and αDG is dependent on the O-linked glycans within the domain, we wanted to determine if the MLD in TIM-1 is also important for interactions with pseudovirions bearing LASV. To address this, we utilized a previously characterized mutant of TIM-1 in which the O-glycan-rich MLD was replaced with a proline-rich region (PRR) from the amphotropic 4070A murine leukemia virus envelope gene, PRR-TIM-1 (25). To assess whether TIM-1-mediated LASV-VSV entry is MLD dependent, we performed a dose-dependent enhancement experiment with exogenous PRR-TIM-1. The level of PRR-TIM-1 expression at the cell surface was modest but increased with increasing amounts of the expression plasmid transfected (Fig. 4E). PRR-TIM-1 expression had no impact on the permissivity of WT HEK 293T cells to LASV-VSV transduction; however, increasing levels of PRR-TIM-1 expression correlated with increasing LASV-VSV transduction in both DG-KO cell lines (Fig. 4F). These data support PtdSer-dependent entry as the mechanism of TIM-1-mediated LASV-VSV transduction in DG-KO cells and confirm that specific residues within the TIM-1 MLD, despite similarity to that of αDG, are not critical for TIM-1-dependent entry.

DISCUSSION

αDG is a ubiquitous cell surface receptor that, when appropriately glycosylated, interacts with LASV GP with high affinity (7) to allow the internalization of this Old World arenavirus into a broad range of cell types. Here, we show that endogenous TIM-1 expression in Vero cells or exogenous TIM-1 expression in HEK 293T cells mediates LASV pseudovirion transduction. However, TIM-1-dependent entry was evident only in the absence of functional αDG.

LASV pseudovirion entry is mediated by TIM-1 through interactions with the PtdSer-binding pocket within the TIM-1 IgV domain. Mutations in this pocket that are known to block binding and uptake of apoptotic bodies and other enveloped viruses (24, 28, 37) abrogated virus entry. Adding support to LASV pseudovirion PtdSer/TIM-1 interactions, we found that mAb ARD5, which binds to the TIM-1 IgV domain and is thought to conformationally alter the domain, inhibited transduction in a dose-dependent manner. Thus, consistent with previous Ebola virus (EBOV) and dengue virus studies investigating TIM-1-dependent entry (24, 28, 55, 60), our findings provide evidence that TIM-1 enhances LASV pseudovirion entry through interactions with virion lipids rather than with LASV GP.

Our studies evaluated LASV entry using two different pseudovirions bearing LASV GP: LCMV (63) and VSV. Similar findings were observed with both virions, suggesting that sufficient quantities of PtdSer are present and available on these two different viral particles to allow TIM-1 interactions. As both of these viral particles bud from the plasma membrane, VSV and LCMV particles may contain similar lipid compositions.

LASV infects a variety of cell types, from epithelial and endothelial cells to macrophages and dendritic cells (49, 64, 65), and TIM family members are surface receptors on a number of these cells. Thus, the TIM-1/LASV interaction may be relevant to in vivo LASV infection and pathogenesis, serving as a LASV receptor on cells lacking functional DG. We have shown previously that TIM-1 is expressed on mucosal epithelial cell populations (66), and others have shown that TIM-1 is expressed on kidney tubule epithelial cells following ischemia (30). The TIM family member TIM-4, which mediates enveloped virus infection in a manner similar to that of TIM-1 (55), is robustly expressed on some tissue macrophages (27). Future studies examining the role of TIM-4 in mediating Old World arenavirus infection in tissue culture and in vivo are needed.

Our studies support data from recent studies by Fedeli et al. (49), providing insights into why PtdSer receptor expression did not mediate LASV-VSV transduction in previous studies, such as those performed in HEK 293T cells (47). These cells have high levels of αDG on their surface, and only upon the knockout of αDG expression does the addition of exogenous TIM-1 expression affect LASV-VSV entry. Our findings indicate that the alternative entry route made available by TIM-1 expression is unable to compete for virion interactions in the presence of αDG.

By performing complementation studies in HEK 293T and Vero cells, we demonstrated that TIM-1 can serve as a receptor for pseudovirions bearing LASV GP (i) in the absence of αDG or (ii) in the presence of abundant yet unsuitably glycosylated αDG. Vero cells that have been shown to have aberrant αDG glycosylation are highly permissive to LASV-VSV, indicating the existence of an efficient route of LASV entry that is independent of the WT αDG O-linked glycans (22). Beyond our expectations, knockout of TIM-1 in Vero cells rendered the cells refractory to LASV pseudovirions despite the continued abundance of another PtdSer receptor, Axl. While our data support a role for TIM-1 both in the absence of αDG in HEK 293T cells and as a primary route of entry in Vero cells, it is possible that additional surface receptors also mediate LASV-VSV entry into these cells and/or that TIM-1 acts simply as an attachment factor rather than a receptor that internalizes the virus into endosomes. Consistent with a potential role of other receptors for LASV uptake into Vero cells, others have determined that PtdSer binding in the TIM-1 IgV domain requires the incorporation of a divalent cation (30), and while TIM-1-dependent EBOV-VSV transduction in Vero cells is sensitive to chelation with EGTA in a dose-dependent manner, LASV-VSV transduction has been shown to be unaffected by EGTA (24). Additional studies are still needed to elucidate the role of and mechanism by which TIM-1, or other PtdSer receptors, mediates enveloped virus internalization.

While endogenous Axl expression on Vero cells or ectopic expression on transfected HEK 293T cells was robust, Axl did not mediate LASV pseudovirion transduction. Our findings with Axl are consistent with recent work demonstrating in several cell lines that the endogenous expression of Axl does not influence the entry of recombinant LCMV (rLCMV) encoding LASV GP (47). However, it should be noted that the studies by Oppliger et al. were performed with αDG-expressing cells (47), and the use of αDG by LASV likely abrogates the need for Axl. In contrast to our findings with Axl, Shimojima et al. showed that ectopic Axl expression in Jurkat T cells and endogenous Axl expression in HT1080 cells enhanced the entry of HIV or VSV pseudovirions bearing LASV GP (45). Similar studies in HT-1080 cells also demonstrated that Axl could serve as a receptor for LASV (49). LASV pseudovirions may utilize Axl-dependent uptake in a cell-specific manner. Such a conclusion is consistent with data from our previous study with EBOV-VSV pseudovirions demonstrating that some, but not all, cells that express robust levels of Axl at the cell surface support virus entry in an Axl-dependent manner (53).

Since LASV GP interacts with the O-linked glycans present on the αDG mucin domain and both αDG and TIM-1 contain an extracellular MLD, it was possible that LASV GP was interacting with TIM-1 in a manner similar to that of αDG. To test this, we examined if the TIM-1 MLD was critical for LASV pseudovirion transduction. However, we found that TIM-1-dependent transduction was independent of the TIM-1 MLD, as the amphotropic MLV Env PRR could replace it. While previous studies of TIM-1 demonstrated that the MLD is required for enveloped virus uptake into cells, it is the length of the MLD stalk domain that is needed to enhance virus entry (25) rather than specific amino acids or O-linked glycans. The present study along with our previous work (25) provide evidence that the length of the PRR sequence is sufficient to substitute for the TIM-1 MLD and is consistent with an absence of LARGE-mediated glycan modifications associated with this mucin domain. This is not surprising, as O-mannosylation modifications of glycans that are found on αDG are thought to be relatively uncommon in mammals (67), and more conventional O-linked glycans on α-DG do not play a role in arenavirus binding (21).

We have previously proposed that the use of PtdSer receptors expands both the species and cellular tropism of some viruses (68), since specific receptor/viral GP interactions are not required for the interaction and subsequent virion internalization (24, 28). Examples of enveloped viruses that utilize PtdSer receptors have been found to have GPs that interact with receptors within the endosomal compartment, such as EBOV or LASV (8, 69, 70). Alternatively, arboviruses can utilize this GP-independent internalization mechanism, presumably because a fusion-ready conformation of their GP is generated during endosomal acidification, and do not appear to require GP/receptor interactions to trigger fusion (reviewed in references 71 and 72). The use of PtdSer receptors is not limited to enveloped viruses, as it has been shown to mediate the entry of vesicle-bound hepatitis A virus (73) and poliovirus (74). The breadth of viruses able to utilize this mechanism highlights the importance of this pathway in viral uptake. Continued work is needed to understand virus interactions with specific PtdSer receptors and why some of these receptors are preferentially targeted for use by viruses over others when both are expressed in the same cell.

MATERIALS AND METHODS

Cells and CRISPR-Cas9-mediated receptor knockout cell lines.

Vero cells (ATCC CCL-81) and HEK 293T cells (ATCC CRL-3216) were maintained in Dulbecco's modified Eagle medium supplemented with 5 to 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin in a 37°C incubator with 5% CO₂. The TIM-1 knockout Vero lines were generated and characterized as previously described (75). WT and KO cells were evaluated for rates of cell division by measuring cellular ATP in an ATP Lite assay (PerkinElmer) over a 24-h period, the time frame used in our transduction studies. No significant differences were detected between the WT and KO cells in these assays.

To create α-dystroglycan knockout (KO) cell lines, CRISPR-Cas9 sequence targets (GX₂₀GG) were identified on genomic DNA within exons of the gene of interest. These sequences were modified to include BbsI restriction sites on either end and were synthesized as oligonucleotides along with each reverse complement. Complementary sequences were annealed together and subsequently ligated into BbsI-digested pX330-U6-Chimeric_BB-CBh-hSpCas9 (pX330) (a kind gift from Kimberly Leslie, University of Iowa). All cloned constructs were confirmed by Sanger sequencing. KO cell lines were created using paired combinations of genomic sequence targets. For DG-KO 293T cell lines, two pairs of targets in DAG1 were cloned into pX330. For DG-KO₁ (also called D2), plasmids targeting GGA-AGG-AGG-CTT-TGC-CAT-CTT-GG (exon 2, coding strand) and GGA-CTC-ACA-GAG-CCA-CAC-CCT-GG (exon 2, coding strand) were used. For DG-KO₂ (also called F6), plasmids targeting GTG-AAC-CCT-CAG-AGG-CTG-TCA-GG (exon 1, coding strand) and GAG-GCT-GTT-CCC-ACA-GTG-GTT-GG (exon 1, coding strand) were used. For the Axl-KO Vero cell line, a pair of targets of African green monkey (Cercopithecus aethiops) AXL were cloned into pX330. For AXL KO clone 2A that was used in this study, plasmids targeting GGG-TGC-TGT-CCG-CGA-GCT-CCA-GG (exon 2, template strand) and GTA-CCA-GTG-TTT-GGT-GTT-TCT-GG (exon 3; coding strand) were used. For AXL/TIM-1 DKO, the same plasmids were used to target Axl in the TIM-1-KO 1A2 cell line described previously (75). Vero or HEK 293T cells were transfected with paired plasmids; 48 h later, cells were surface stained for the receptor of interest, and unstained cells were sorted to 1 cell per well into 96-well plates with a FACSAria Fusion high-speed cell sorter (HEK 293T cells) or by cell dilution in culture (Vero cells). Clonal populations were expanded and subsequently surface stained to confirm the loss of receptor expression. Only populations that had a complete loss of receptor expression without observable deleterious phenotypes were selected. All KO cell lines were maintained under the same conditions as WT cell lines.

Plasmids.

All viral glycoproteins used to produce VSV pseudovirions, along with human Axl (a kind gift from the University of Iowa Viral Vector Core), human TIM-1 (GenBank accession number NM_012206.2), and murine DG (a kind gift from Kevin Campbell, University of Iowa) cDNA sequences, were cloned into mammalian expression vectors. The coding sequence of the Axl cDNA encodes full-length Axl (GenBank accession number M76125.1). Murine DG is >90% identical to human DG and is a competent receptor for LASV (76). ND115DN-TIM-1 contains two mutations that together have been shown to disrupt the ability of the virus to bind to the PtdSer-binding pocket of TIM-1 (24). N114D and D115N mutations in TIM-1 were introduced as previously described (24). PRR-TIM-1 contains the proline-rich region (PRR) from the amphotropic 4070A murine leukemia virus envelope gene in lieu of the mucin domain of TIM-1. PRR-TIM-1 was cloned as previously described (25). LASV GP expressed in the pCAGG vector was used to generate the LASV-LCMV pseudovirions from constitutively expressing Vero cells (63).

Pseudoviruses and adenoviral expression vectors.

VSV pseudovirions were produced by transfecting HEK 293T cells with plasmids containing either the LASV GP (Josiah; a kind gift from Robert Mandell, NewLink Genetics), VSV G (Indiana; a kind gift from Paul McCray, University of Iowa), or full-length EBOV GP (Mayinga; a kind gift Robert Davey, Texas Biomedical Research Institute) gene 18 to 24 h prior to transduction with LASV GPC-pseudotyped VSVΔG-eGFP at a multiplicity of infection (MOI) of ∼1. At 3h postransduction, cells were washed once with phosphate-buffered saline (PBS) and replenished with fresh medium supplemented with 1.5% fetal bovine serum (FBS) and 1% penicillin-streptomycin. The supernatant containing pseudovirus was collected both 24 h and 48 h after the PBS wash step and strained through a 0.45-μm syringe filter. Collections were then combined, concentrated, and purified through ultracentrifugation with a 20% sucrose cushion at 90,000 × g for 2 h at 4°C before resuspension in PBS. Virus stocks were distributed into aliquots and stored at −80°C.

GP-deficient, GFP-expressing recombinant LCMV (rLCMV) (Armstrong strain) bearing LASV (Josiah) GPC was produced as previously described (63). Briefly, Vero cell lines constitutively expressing LASV GP (63) were infected at a low multiplicity of infection (MOI of 0.01) with rLCMVΔGP/GFP. At 72 h postinfection (p.i.), LASV GP-pseudotyped rLCMVΔGP/GFP was collected from the supernatant, and titers were determined on LASV GP-expressing Vero cells.

Ad-hTIM-1 was created by cloning human TIM-1 (hTIM-1) downstream of the cytomegalovirus (CMV) promoter in the E1 cassette of the Ad shuttle pacAd5CMVmcspA (provided by the University of Iowa Viral Vector Core) with KpnI and NotI. Recombinant adenoviral vectors Ad-TIM-1, Ad-Empty, and Ad-eGFP were made by the University of Iowa Viral Vector Core as previously described (77). Vector titers were ∼10¹⁰ PFU/ml.

Cell surface staining.

The following polyclonal antibodies used in cell surface staining assays were obtained from R&D Systems: sheep anti-hDG (catalog number AF6868), goat anti-hAxl (catalog number AF154), goat anti-hTIM-1 (catalog number AF1750), goat anti-hMer (catalog number AF891), goat anti-hTIM-4 (catalog number AF2929), and goat anti-hTyro3 (catalog number AF859). Mannosylation-dependent mouse anti-rabbit αDG monoclonal antibody IIH6 was developed by Kevin Campbell (University of Iowa) and obtained from the Developmental Studies Hybridoma Bank, created by the NICHD of the NIH and maintained at the University of Iowa Department of Biology, Iowa City, IA. IIH6 was used as the supernatant (29 μg IgG/ml).

For surface staining experiments, cells lifted from plates with 5 mM EDTA in PBS were transferred into a 96-well round-bottom plate at ∼200,000 cells per well. Cells were washed with 5% FBS in PBS three times prior to incubation with primary antibodies for 1 h on ice at concentrations specified in the respective figure legends. Primary-antibody-stained cells were washed three times and incubated with Cy5-conjugated donkey anti-goat (catalog number A21447; Life Technologies), donkey anti-sheep (catalog number 713-605-003; Jackson ImmunoResearch), or donkey anti-mouse (catalog number 715-175-150; Jackson ImmunoResearch) IgG secondary antibody at a 1:1,000 dilution for 15 min on ice. Secondary-antibody-stained cells were washed three times and subsequently analyzed via flow cytometry using a FACSCalibur instrument (Becton, Dickinson). Data were analyzed with single-cell analysis software (FlowJo).

Transductions.

For all transductions, cells were plated at 50,000 cells per well in a 48-well format overnight, and virus stocks, as noted, were then applied to cells and assessed for GFP expression 18 to 24 h later by flow cytometry.

In LASV-VSV pseudovirion, VSV-VSV pseudovirion, or eGFP-encoding adenoviral vector transduction studies, increasing concentrations of virions were added to 15,000 cells in a 48-well tissue culture plate. In LASV-LCMV pseudovirion studies, a single dose (MOI of ∼0.2) of the stock was used in 48-well transduction studies.

In studies where plasmids expressing surface receptors (αDG, WT or mutant TIM-1, or Axl) were transfected prior to transduction experiments, cells were plated overnight in a 12-well format. Wells were transfected with 0, 0.25, 0.5, or 1 μg of plasmid by using a polyethyleneimine (PEI)-based transfection protocol. Cells were split into two groups 24 h following transfection. Forty-eight hours following transfection, one group of the cells was transduced, whereas the other transfected cells were assessed for receptor surface expression by flow cytometry. For these transduction experiments, equivalent amounts of pseudovirions (MOI of ∼0.15) were added, resulting in ∼15 to 20% transduction of WT HEK 293T cells. MOIs of stocks used in these studies were determined by transducing 2-fold serial dilutions of pseudovirus onto WT HEK 293T cells for 24 h and assessing GFP expression by flow cytometry. An MOI in the linear range of the curve was selected for use in these studies. This value was selected because this level of transduction is in the linear range of the transduction curve. The same quantity of pseudovirions applied to untransfected receptor knockout lines gave dramatically lower levels of transduction.

For assessing the ability of anti-human TIM-1 monoclonal antibody ARD5 to inhibit LASV-VSV transduction, WT or αDG CRISPR-Cas9-knocked-out HEK 293T cells were transfected with a human TIM-1-expressing plasmid. ARD5 was added to wells at the concentrations noted above for 1 h prior to LASV-VSV transduction. Equivalent amounts of LASV-VSV pseudovirions were added to the treated or untreated cells. Transduction levels were determined 24 h later by assessing the GFP positivity of cells by flow cytometry. Similar protocols were used for studies that were performed with LASV-LCMV in WT or Axl-KO Vero cells.

For transduction experiments with WT and CRISPR-Cas9-knocked-out Vero cells, cells were plated overnight in a 12-well format prior to transduction with an empty adenovirus (Ad-Empty) or an adenovirus expressing TIM-1 (Ad-TIM-1) at an MOI of 2, 4, or 8. For these transductions, treated cells were lifted with 5 mM EDTA in PBS at 24 h posttransfection/posttransduction and split into two groups. The first group was analyzed for TIM-1 surface expression by flow cytometry, and the second was distributed into separate tissue culture plates for LASV-VSV transduction. For each set of these experiments, the same quantity of LASV-VSV was transduced across all untreated and treated cells.

Zika virus infections.

Six-well plates of HEK 293T cells were transfected with 2.5 μg of an Axl expression plasmid or an empty vector using Lipofectamine 2000 according to the manufacturer's instructions. The next day, cells were lifted, and 10,000 cells were plated in a 96-well format or held in the 6-well format. Twenty-four hours later, cells from the 6-well plate were lifted and evaluated for Axl surface expression on cells by flow cytometry using a 1:200 dilution of goat anti-Axl polyclonal antibody (R&D Systems). The 96-well plate of transfected cells was infected with Zika virus (Puerto Rico strain) at an MOI of 0.25. After 3 days of infection, cells were placed into TRIzol, and RNA was extracted. One microgram of RNA was reverse transcribed, the cDNA was diluted 1:50, and quantitative real-time PCR (qRT-PCR) was performed using primers (forward primer 5′-AARTACACATACCARAACAAAGTGGT-3′ and reverse primer 5′-TCCACTCCCTCTCTGGTCTTG-3′) (78) against the Zika virus genome. The C_T (threshold cycle) values were converted to virus copy numbers using a standard curve of genome equivalents.

Axl migration assay.

HEK 293T cells expressing an empty vector, Axl, or murine bone marrow stromal antigen 2 (mBST-2) were plated on the apical chamber of 24-well cell culture inserts (Merck Millipore) at a concentration of 10⁵ cells/insert in 100 μl of serum-free Opti-MEM. Opti-MEM (600 μl) containing 10% FBS with or without 0.5 μg/ml of Gas6 (R&D Systems) was added to the basal chamber, and cells were allowed to migrate for 24 h at 37°C. Nonmigrated cells were washed off, and migrated cells were fixed, permeabilized, stained, and imaged as previously described (79). Migrated cells from at least five different 10× fields were counted using ImageJ software and averaged.

Statistics.

All studies were performed at least three independent times with two to three replicates in each experiment. Raw data were normalized to control values and are shown as either transduction levels relative to control values or increases in expression or transduction relative to baseline values. Means and standard deviations for all studies were generated and statistical differences were evaluated by two-way analysis of variance (ANOVA). Differences were deemed significant if P values were less than 0.05.