Tyro3, Axl, and Mertk are famed receptor tyrosine kinases (RTKs), establishing TAM receptor family (Weinger et al. 2011; Vouri and Hafizi 2017; Shafit-Zagardo et al. 2018). TAM receptors have been reported to participate in many physiological and pathological processes, including phagocytic clearance of apoptotic cells (Seitz et al. 2007; Weinger et al. 2011), regulating immune privilege and homeostasis (Sun et al. 2010; Rothlin et al. 2015), maintaining blood–brain barrier (BBB) integrity and central nervous system (CNS) homeostasis (Weinger et al. 2011; Li et al. 2013; Goudarzi et al. 2016; Ray et al. 2017), promoting tumorigenicity, malignancy and chemoresistance of tumors (Schoumacher and Burbridge 2017; Uribe et al. 2017; Vouri and Hafizi 2017), regulating multiform programmed cell death (PCD) pathways (Rothlin et al. 2015; Han et al. 2016; Najafov et al. 2018) and etc.

TAM receptors play important roles in viral infection. They have been reported to act as entry receptor/cofactor for many enveloped viruses, such as filoviruses and flaviviruses, and regulate immune and inflammatory responses during viral infection (Hunt et al. 2011; Morizono et al. 2011; Shibata et al. 2014; Hamel et al. 2015; Meertens et al. 2017; Chen et al. 2018; Fedeli et al. 2018; Persaud et al. 2018). Recently, TAM receptors have been proved to protect CNS from infection of neuroinvasive viruses through maintaining BBB integrity (Miner et al. 2015).

In this review, we provide detailed information about the multifaceted roles of TAM receptors in viral infection, and also discuss the possible mechanisms underlying TAM receptors' functions.

TAM receptors belong to type Ⅰ receptor tyrosine kinase and are composed of three homologous members, namely, Tyro3, Axl and Mertk, which are broadly expressed in many cell types throughout the body, in particular, the sentinel cells of the immune system, such as macrophages and dendritic cells (Shimojima et al. 2006). TAM receptors share similar structural organization of their extracellular domains, comprising of two tandem N-terminus immunoglobulin-like domains (Ig1–2) that mediate the receptor-ligand interaction and are followed by two tandem membrane-proximal fibronectin type Ⅲ-like (FNⅢ) domains (F1–2) (Fig. 1). After linking to the F1 and F2 domains, it becomes a single transmembrane (TM) domain. The extracellular moiety of the TM domain contains cleavage sites (CS) for ADAM10 and ADAM17 (Fig. 1). ADAM10 cleaves Axl, while ADAM17 cleaves both Axl and Mertk, resulting in soluble TAM receptors (O'Bryan et al. 1995; Thorp et al. 2011). The intracellular region of TAM receptors contains a tyrosine kinase domain (KD) that is highly conserved and includes a TAM-specific sequence KW(I/L)A(I/L)ES in the catalytic domain (Shafit-Zagardo et al. 2018) (Fig. 1).

Figure 1.  The topological structures of TAM receptors and Gas6/Pros1. TAM (Tyro3, Axl, and Mertk) receptors are homologous and have similar topological organization. Gas6 and Pros1 are the two specific ligands that activate TAM receptors. They are also homologous and share similar topological structure.

Growth arrest-specific gene 6 (Gas6) and protein S (Pros1) are the two main ligands that activate TAM receptors. Gas6 and Pros1 are homologous and have similar spatial topological structure, including an N-terminus γ-carboxyglutamic acid (Gla) domain that depends on vitamin K for γ-carboxylation (Fig. 1). The Gla domain mediates the binding of TAM receptors to phosphatidylserines (PtdSers) (Huang et al. 2003; Geng et al. 2017) (Fig. 1). After the Gla domain, it appears four tandem epidermal growth factor (EGF)-like repeats (E1–4) (Fig. 1). Both Gla domain and EGF-like domain are required to efficiently activate TAM receptors (Geng et al. 2017). Following the four EGF-like domains, it presents the two laminin G domains (LG1–2), which directly bind to the Ig1 and Ig2 domains of TAM receptors to launch receptor activation (Fig. 1). Although there is a high degree of homology among different TAMs and between their ligands, the ligand-receptor interactions are seemingly selective. Gas6 exclusively activates Axl with 100–1, 000 folds higher binding affinity over Mertk and Tyro3 (Kd in nmol/L range), while Pros1 mainly revitalizes Tyro3. Interestingly, Mertk shows lower affinity to both ligands (Kd in the lmol/L range) (Nagata et al. 1996). Both ligands can launch TAM receptor activation in a phosphatidylserine dependent and independent manner. In the presence of phosphatidylserines, Gas6 and Pros1 revitalize TAM receptors in a much more efficient way than do in the absence of phosphatidylserines, indicating the importance of TAM signaling in clearing apoptotic cells (Geng et al. 2017). Gas6 and Pros1 act as a bridge concatenating phosphatidylserines exposed on the surface of apoptotic cells to TAM receptors on the surface of macrophages and other phagocytes to enhance phagocytic clearance of apoptotic cells (Geng et al. 2017).

TAM receptors are reported to act as entry receptors for diverse enveloped viruses, such as filoviruses and flaviviruses. Shimojima et al. reported that Axl, Tyro3, and Mertk promote the entry of live ebolavirus (EBOV) and pseudotyped virus having filovirus glycoproteins on its envelope (Shimojima et al. 2006). Interestingly, they found a lack of an exon 10-corresponding region (nine amino acids) and most of the cytoplasmic tyrosine kinase domain in Axl impairs its ability to promote pseudotyped filovirus entry. While loss of only exon 10-corresponding region in Axl displays enhanced effects on filovirus glycoproteinmediated infection, which is similar to those of full-length Axl. These data suggest that the cytoplasmic tyrosine kinase activity of Axl is critically required for mediating filovirus entry. Further investigations demonstrated that TAM receptors mediate filovirus cell entry by facilitating the transfer of viral particles to endosomes, thus allowing invasion into the cytoplasm (Shimojima et al. 2006). Brindley et al. (2011) further stated that despite the conspicuous filovirus entry promoting effect of Axl, there seems to be no direct interaction between ectodomain of Axl and Zaire ebolavirus glycoprotein (ZEBOV-GP). Silence of surface-expressed Axl by RNAi does not affect ZEBOV-GP binding to cells, but significantly reduces ZEBOV pseudovirus internalization and fusion with cellular membranes (Brindley et al. 2011). These findings suggest that Axl may not bind to the glycoproteins on filovirus envelop, but instead may promote the viral envelope fusion with plasma membrane, internalization and endosomal transfer.

In addition to promoting filovirus entry into cytoplasm, TAM receptors also act as entry receptors or facilitators for flaviviruses. Meertens et al. first reported that T cell immunoglobulin domain and mucin domain (TIM) and TAM receptors are the two distinct transmembrane RTKs that mediate dengue virus (DENV) entry. Unlike TIM proteins, which bind directly to the phosphatidylserines on the surface of DENV particles, TAM proteins bind indirectly to viral phosphatidylserines via bridging by their natural ligands Gas6 and Pros1 (Meertens et al. 2012). In 2015, Hamel et al. reported Axl and Tyro3 mediate Zika virus (ZIKV) entry into various host cells (Hamel et al. 2015). In 2015 and 2016, Zika outbreaks took place all around the world and caused great concerns and panics on global public health, which also boomed the studies on ZIKV infection and treatment (Wang et al. 2016). Nowakowski et al. studied the expression of receptors present in cell entry of several enveloped viruses including ZIKV over various cell types in the developing brain, finding that Axl is abundantly expressed in human radial glial cells, astrocytes, endothelial cells, and microglia in developing human cortex and in progenitor cells in developing retina. They also demonstrated that Axl expression in radial glia is conserved in developing mouse and ferret cortex and in human stem cell-derived cerebral organoids, which suggests Axl may be a conserved entry receptor for ZIKV infection in diverse species (Nowakowski et al. 2016). Savidis et al. also proposed that Axl plays important roles in ZIKV and DENV infection through orthologous functional genomic screens using RNAi and CRISPR/Cas9 approaches (Savidis et al. 2016). Liu et al.'s work gave deep insights into the detailed roles of Axl in ZIKV infection (Liu et al. 2016). They found that a polyclonal anti-human Axl antibody blocks ZIKV infection of human cerebral microvascular endothelial cells (hCMEC/D3) and human umbilical vein endothelial cells (HUVECs), while not affecting viral binding to cells. They transfected 293T cells with wild type Axl, kinase dead Axl (K567R mutant, which lost kinase activity), and an irrelevant kinase PDK1 control, and found that the absence of Axl or loss of Axl kinase activity does not affect virus binding to 293T cells, but significantly reduces viral infection of the cells, and loss of Axl kinase activity partially reduces viral infection. These data suggest that Axl is dispensable for viral binding to cells, but its kinase activity is required for viral infection in cells (Liu et al. 2016). Consistent with results from the Axl kinase dead experiment, Cabozantinib and R428, two known inhibitors of Axl phosphorylation, significantly impair ZIKV infection in hCMEC/D3 and HUVECs in a dose-dependent manner, further proving that Axl kinase activity is required to facilitate ZIKV infection (Liu et al. 2016). Collectively, Liu et al.'s work puts forward a new concept that Axl may not act as an exclusive receptor for ZIKV binding, but instead, it functions in post-binding step, and its kinase activity is significant in mediating ZIKV infection.

Retallack et al. (2016) found that Axl blockage reduces ZIKV infection of astrocytes in vitro, and genetic knockdown of Axl in a glial cell line nearly abolishes infection, suggesting Axl promotes viral infection in human astrocytes and glial cells. Later, Meertens et al. (2017) found that Axl is expressed in human microglia and astrocytes in developing brain and it mediates ZIKV infection of glial cells, which requires Axl's natural ligand Gas6. Gas6 concatenates ZIKV particles to glial cells, followed by ZIKV internalization via clathrin-mediated endocytosis and traffic to Rab5⁺ endosomes to establish productive infection. Additionally, they learned that during viral entry, the ZIKV/Gas6 complex activates Axl kinase activity, which downregulates interferon (IFN) signaling and thus facilitates infection. Chen et al. (2018) reported a similar result, they found genetic ablation of Axl protects primary human astrocytes and astrocytoma cell lines from ZIKV infection. Axl deletion does not block the entry of ZIKV, but instead, presence of Axl attenuates ZIKV-induced activation of type Ⅰ IFN signaling genes, including several type Ⅰ IFN and IFN-stimulating genes. Recently, Persaud et al. (2018) reported that Axl is essential for ZIKV infection of a human fibroblast cell line, which requires clathrin-mediated endocytosis and delivery of the virus to acidified intracellular compartments.

Miner et al. (2016) first challenged the consensus that Axl is essential for ZIKV entry and infection. They found that Axl and/or Mertk knockout has no impact on ZIKV replication in brain, eyes, spleen or serum in mice treated with anti-Ifnar1 mAb. Based on these findings, they concluded that in mice deficient in IFN signaling, Axl and Mertk are not required for ZIKV infection. Given that there are intricate interactions between IFN signaling and Axl signaling, the roles of Axl in ZIKV infection remain disputable. Later, Wells et al. (2016) provided more direct and relevant evidence to prove Axl is not essential for ZIKV infection in certain cells. They found that genetic ablation of Axl has no effect on ZIKV entry or ZIKV-mediated cell death in human induced pluripotent stem cell (iPSC)-derived neural progenitor cells (NPCs) or cerebral organoids. Li et al. inoculated various dose of ZIKV to the brains of Axl knockout and wild type mice, and found that Axl knockout does not affect viral replication in brain at any dose of infection. Wang et al. (2017) also proposed that Axl is not the exclusive entry receptor for ZIKV infection in brain, spleen, liver, heart, kidney, serum or neuronal cells in mice. Later, Hastings et al. (2017) further reported none of TAM receptors (Axl, Tyro3, and Mertk) is required for ZIKV entry or infection in various tissues in mice.

In summary, most of in vitro investigations suggest that TAM receptors act as entry receptors or at least entry facilitators to promote viral endocytosis and transfer to endosomes during enveloped virus infection, however there are still in vitro systems or in vivo studies suggesting TAM receptors are not indispensable in enveloped virus infection. We hold the opinion that the roles of TAM receptors in viral infection are not all or nothing, instead they may function in viral infection of certain cell types, while not in other cell types. The different living milieu between in vitro and in vivo lived cells may affect certain characteristics of cognate cells, which may also result in differential roles of TAM receptors in viral infection.

Besides filoviruses and flaviviruses, arenaviruses are also reported to use TAM receptors to promote cell entry and infection. Shimojima et al. (2012) reported Axl and Tyro3 are entry receptors for Lassa virus (LASV), a causative agent of Lassa fever. They found that the binding of LASV to Axl needs assistance from an unidentified molecule(s) on the cell surface, while binding of LASV to Tyro3 is directly. Additionally, the two extracellular immunoglobulin domains are necessary for viral binding, and the intracellular kinase domain is necessary for the infection post binding. Fedeli et al. (2018) declared that Axl fulfills different functions in LASV entry in distinct cell types, depending on the extent of the post-translational modification of dystroglycan, which is a primary receptor for LASV cell entry (Oppliger et al. 2016). In the absence of functional dystroglycan, Axl mediates LASV entry via apoptotic mimicry, followed by macropinocytosis and fusion to late endosomes, and productive viral infection requires Axl kinase activity. Endothelial cells express low level of glycosylated dystroglycan, under this condition, both Axl and dystroglycan act as the receptor for LASV cell entry (Fedeli et al. 2018). Lymphocytic choriomeningitis virus (LCMV), the prototype of arenavirus, is also reported to use Axl and Tyro3 as its entry receptors (Shimojima and Kawaoka 2012).

Collectively, TAM receptors function as entry receptors or facilitators for a myriad of enveloped viruses in a manner termed "apoptotic mimicry" (Moller-Tank and Maury 2014). It suggests that the phosphatidylserines exposed on viral envelope resemble the phosphatidylserines on the surface of apoptotic cells, which bind to TAM receptors on the surface of macrophages, dendritic cells, and other phagocytes. The binding of phosphatidylserines to TAM receptors needs the bridging of Gas6 or Pros1 and significantly facilitates the phagocytosis of apoptotic cells by phagocytes. Similarly, TAM receptors may facilitate viral entry in the same manner. Meanwhile, accumulating evidences suggest that the much more important functions of TAM receptors may not exist in virus-TAM binding step, but in post-binding step. After virus-Gas6/Pros1-TAM binding, TAM receptors are revitalized, which promotes clathrin-mediated endocytosis or macropinocytosis of viruses. There are also vast evidences suggesting that TAM receptor activation during viral infection exerts strong inhibition on the innate immune and inflammatory responses to facilitate viral replication and immune evasion, which we will discuss in next section.

Viral infection can initiate TAM receptor activation through direct and indirect pathways. Some enveloped viruses bind to Gas6/Pros1, forming virus-Gas6/Pros1 complex. The virus-Gas6/Pros1 complex binds and activates TAM receptors, which is the direct pathway for viral infection to launch TAM signaling. Interestingly, TAM receptor activation can further promote the expression of itself, forming a positive feedback regulation to enhance TAM signaling (Sun et al. 2010). Viral infection stimulates certain Toll-like receptors (TLRs), which upregulates TAM receptors and induces type Ⅰ IFN (Sun et al. 2010; Grabiec et al. 2018). The activation of IFN signaling also promotes TAM receptor synthesis and expression on the cell surface (Grabiec et al. 2018). Moreover, Axl physically associates with the type Ⅰ IFN alpha receptor (IFNAR). Type Ⅰ IFN signaling via the IFNAR-STAT1 pathway also stimulate the TAM signaling, resulting in the induction of suppressor of cytokine signaling (SOCS) 1 and SOCS3 upon cytokine engagement with the IFNAR (Rothlin et al. 2007). The above are the indirect pathways for viral infection to stimulate TAM signaling.

Bhattacharyya et al. (2013) found that Gas6 or Pros1-coated viruses bind to TAM receptors and directly activate TAM signaling in dendritic cells, promoting expression of SOCS1, and downregulating expression of IFN-α and IFN-β. TAM-deficient dendritic cells are relatively resistant to he infection of flaviviruses and pseudotyped retroviruses, but the infection can be restored with neutralizing type Ⅰ IFN antibodies. In addition, a TAM kinase inhibitor also antagonizes viral infection of wild type dendritic cells, but has no effect on viral infection of IFNAR deficient dendritic cells. These data suggest that enveloped viruses evade innate immunity through directly activating TAM/SOCS signaling. Meertens et al. (2017) reported that during ZIKV infection in human glial cells, ZIKV-Gas6 complex activates Axl signaling, which results in downregulation of various IFNs, such as IFN-β, IFN-λ1, IFN-λ2, and IFN-stimulated gene 20 (ISG20) and pro-inflammatory cytokines, such as tumor necrosis factor a (TNF-α), interleukin-6 (IL-6), and IL-1β. SOCSs may mediate the global suppression of innate immune and inflammatory response. Chen et al. (2018) reported similar findings, they discovered that during ZIKV infection of human astrocytes, activation of Axl signaling by ZIKV impedes the induction of IFNs and ISGs. Deleting type Ⅰ IFN receptor alpha chain (IFNAR1) restores the vulnerability of Axl knockout astrocytes to ZIKV infection. Further experiments suggest that Axl suppresses type Ⅰ IFN signaling through upregulating the expression of SOCS1, a downstream signal molecule of Axl signaling, in a STAT1/STAT2-dependent manner.

In addition to enveloped viruses, viruses lacking envelope may also directly activate TAM receptors to oppress innate immunity. Nidetz et al. (2018) discovered that Gas6 binds to the fiber proteins (rather than the pentons or hexons) of common adenovirus (AdV) serotypes, e.g. HAdV-5C, with higher affinity than rare HAdV-28D fibers and the binding needs Gla domain in Gas6. HAdV-5C associated Gas6 suppresses IFN induction during viral infection. Given that Gas6 activates Axl signaling, HAdV-5C-Gas6 complex may also activate Axl signaling, resulting in IFN signaling suppression. Ahtiainen et al. (2010) reported that there is the constant upregulation of all TAM receptors and SOCS1 in breast cancer initiating cells upon an oncolytic adenovirus infection, which may be caused partially by the interaction of TAM receptors and viral protein, resulting in strong suppression of type Ⅰ IFN signaling to facilitate oncolytic AdV infection.

Sun et al. figured out a paradigm explaining the inhibitory effects of TAM signaling on the induction of type Ⅰ IFNs and pro-inflammatory cytokines in Sertoli cells (Sun et al. 2010), and this paradigm broadly exists in various kinds of cells (Fig. 2). Toll-like receptor 3 (TLR3) is a pattern recognition receptor that recognizes pathogen derived double strand RNA (dsRNA), followed by activation of TLR3-TRIF-IRF3 signaling, which results in induction of type Ⅰ IFNs, as well as activation of TLR3-TRIF-NF-κB signaling, which elicits transcription of pro-inflammatory cytokines (Fig. 2). Interestingly, TLR3 activation also promotes the expression of Axl, which later translocates to the plasma membrane (Fig. 2). Gas6/Pros1 ligation to TAM receptors launches activation of TAM receptors, followed by activation of transcription factor STAT1 and/or STAT2, which translocate into nucleus and promote transcription of SOCS1 and/or SOCS3, followed by inhibition of TLR3-TRIF-IRF3 signaling and TLR3-TRIF-NF-κB signaling, which impedes the production of type Ⅰ IFNs and pro-inflammatory cytokines (Fig. 2). Noticeably, TAM receptor activation can promote their own expression, together with the induction of TAM receptors by TLR3, these form a positive regulation loop to effectively and immediately shut down IFN and pro-inflammatory cytokine generation to relieve the untoward effects of overactive innate immunity and inflammatory responses (Fig. 2).

Figure 2.  The regulation of TAM receptors on the innate immune and inflammatory responses. TAM receptors are activated by virus-Gas6/Pros1 complex, resulting in endocytosis of viruses and phosphorylation of STAT1/2. The phosphorylated STAT1/2 translocates into nucleus and promotes the transcription of TAM receptors and SOCSs. Virus-derived double strand RNA (dsRNA) activates TLR3, which stimulates TRIF-IRF3 signaling and TRIF-NF-κB signaling to launch innate immune and inflammatory responses. SOCSs have a strong inhibition on TLR3-TRIF-IRF3 and TLR3-TRIF-NF-κB signaling. Moreover, activated TLR3 also promotes the expression of TAM receptors.

The in vivo studies further validate the inhibitory effect of TAM signaling on immune and inflammatory response. Lu and Lemke (2001) first generated TAM receptor single, double, and triple knockout mice and studied the in vivo roles of TAM receptors in immune homeostasis. They found that the mutant mice's lack of TAM receptors developed a severe lymphoproliferative disorder accompanied by broad-spectrum autoimmunity. Li et al. (2013) also reported Tyro3, Axl, and Mertk triple knockout (TKO) mice show systemic autoimmunity, characterized by increased pro-inflammatory cytokine production, autoantibody deposition and autoreactive lymphocyte infiltration into a variety of tissues. TAM TKO engenders higher TNF-α content in serum and higher IL-1β and IL-6 content in lymph nodes. The peritoneal macrophages isolated from TAM TKO mice produce more TNF-α after LPS stimulation than the peritoneal macrophages isolated from wild type mice.

In summary, many enveloped viruses and even some non-enveloped viruses can bind to TAM receptors. After engagement with TAM receptors, the viruses can directly activate TAM signaling, followed by inhibition of innate immunity and inflammatory response. In addition to direct activation of TAM signaling, the viruses can also promote TAM receptor expression through TLR mediated TAM generation and IFN stimulated TAM induction. The suppression of host innate immunity and inflammation are conducive to viral infection and immune evasion.

TAM receptors are expressed in vascular endothelial cells, pericytes, and vascular smooth muscle cells (VSMCs). The expression pattern of TAM receptors is associated with vascular types. In capillaries, Axl is expressed in endothelial cells, while in larger blood vessels, both arterioles and veins, Axl is mainly expressed in VSMCs (O'Donnell et al. 1999). TAM receptors participate in multiple angiogenic processes including survival, proliferation, migration, invasion and aggregation of endothelial cells as well as pericyte adhesion (Li et al. 2017b). Gas6/Axl signaling activation promotes endothelial cell survival under serum starvation via PI3K-Akt signaling, which upregulates anti-apoptotic Bcl-2 and downregulates proapoptotic caspase-3 (Hasanbasic et al. 2004). O'Donnell reported (1999) that Gas6 binds to HUVECs, and soluble Axl inhibits this binding. Exogenous Gas6 protects HUVECs from apoptosis in response to growth factor withdrawal and TNF-α mediated cytotoxicity.

TAM receptors also fulfil some important physiological functions for vascular endothelial cells. Happonen et al. (2016) found that both human aortic endothelial cells (HAECs) and HUVECs can phagocytose plasma membrane-derived microparticles (PMPs), which are released by activated platelets and are pro-coagulant and potentially pro-inflammatory. Thus, rapid cleaning PMPs from the circulation is essential for averting thrombotic diseases. Using TAM-blocking antibodies or siRNA knockdown of individual TAMs, they showed that the endothelial cells' uptake of PMPs depends on Axl and Gas6.

The BBB is mainly composed of brain microvascular endothelial cells, pericytes, basement membrane, and astrocyte end-feet, and it is the pivotal physiological barrier that safeguards CNS homeostasis, allowing for tight regulation of the transit of ions, molecules, and cells between the blood and the brain, and also for rejection of toxic matters and pathogens. The dysfunction of BBB causes various diseases (Daneman and Prat 2015). Many neuroinvasive viruses induce none or only mild symptoms. Notwithstanding, once the viruses invade brain and establish CNS infection, they usually cause severe encephalitis, resulting in mortality or disability. TAM receptors have been shown to maintain BBB integrity under naïve and infected conditions to hinder viral entry into the brain. Li et al. (2013) found that TAM receptors TKO results in increased BBB permeability for Evans blue and fluorescent-dextran, indicating the breakdown of BBB. Miner et al. (2015) reported that Mertk and Axl, but not Tyro3, protect mice against CNS infection of West Nile virus (WNV) and La Crosse fever virus through maintaining BBB integrity. They measured the BBB permeability of wild type (WT), Tyro3-KO, Axl-KO, Mertk-KO, and Axl-Mertk-DKO mice under naïve and WNV-infected condition, finding even under naïve condition, Axl-KO, Mertk-KO, and Axl-Mertk-DKO mice show slightly higher BBB permeability than WT mice, but only the BBB permeability of Axl-Mertk-DKO mice significantly increased. After WNV infection, BBB permeability of all the genotyped mice increased, which may result from BBB damage by viral infection and neuroinflammation. And consistent with BBB permeability tendency under naïve condition, Axl-KO, Mertk-KO, and Axl-Mertk-DKO mice, but not Tyro3-KO mice, show significantly increased BBB permeability than WT mice (Miner et al. 2015). These data suggest that Mertk and Axl may be more crucial to maintain BBB integrity than Tyro3, and Mertk seems to be more effective in maintaining BBB integrity than Axl under both naïve and viral infection conditions. The increased BBB permeability may promote neuroinvasive virus entry into CNS, causing encephalitis, which may further destroy BBB, and allow for entry of more viruses, noxious matters, as well as inflammatory immunocytes, eventually leading to death.

The molecular basis underlying the regulation of BBB integrity by TAM receptors remains unclear. BBB mainly consists of brain microvascular endothelial cells (BMECs), pericytes, and astrocytes, of which BMECs play a paramount role. Miner et al. (2015) studied the regulation of TAM receptors on BBB integrity and permeability by means of an in vitro BBB model, which is composed of primary mouse BMECs cultured in the upper chamber of a transwell, and primary astrocytes in the lower chamber. The transendothelial electrical resistance (TEER) across the BMEC monolayer denotes the barrier integrity, with higher TEER indicating a tighter barrier. They found that Axl-Mertk-DKO BMEC monolayer shows lower TEER than wild type (WT) monolayer in baseline. During WNV infection, both TEERs of Axl-Mertk-DKO and WT BMEC monolayer increase, but Axl-Mertk-DKO BMEC barrier fails to tighten as much as WT BMECs. These findings validate the protective roles of TAM receptors in maintaining BBB integrity in vivo. Miner next studied whether TAM receptors regulate WNV transit across BBB, still using the in vitro BBB model. They discovered that TAM receptors slightly reduce the binding of WNV to BMECs, but significantly impedes WNV transit across the BMEC monolayer (Miner et al. 2015).

TAM receptors pose inhibition on pro-inflammatory cytokine generation, some pro-inflammatory cytokines compromise BBB integrity. Whether TAM receptors maintain BBB integrity through curbing brain pro-inflammatory cytokine production during viral infection is yet to be determined. Miner et al. (2015) found during WNV infection, compared to WT mice, IL-1β, TNF-α, and IFN-γ are significantly increased in supernatant of Axl-Mertk-DKO BMECs. They tested the possible impact of IL-1β and TNF-α on BMEC monolayer integrity, finding that pretreating BMEC monolayer with anti-IL-1β, anti-TNF-α or anti-IL-1β plus anti-TNF-α (most potent) blocking antibodies minimally tightens both WT and Axl-Mertk-DKO BMEC barriers. Addition of TNF-α to medium of BMEC monolayer significantly reduces barrier integrity in both WT and Axl-Mertk-DKO BMEC barriers. These data suggest that the upregulated IL-1β and TNF-α in TAM receptor knockout mouse brain may contribute to BBB breakdown during WNV infection. Li et al. (2015) proved that IFN-γ can untighten BBB via reducing the expression of tight junction proteins during Japanese encephalitis virus (JEV) infection, and blocking IFN-γ with neutralizing antibodies can significantly reduce the BBB permeability through relieving the impaired expression of tight junction proteins. Chai et al. (2014) reported infection with only a laboratory-attenuated rabies virus (RABV), rather than wild type RABV, enhances BBB permeability, which is associated with the reduction of tight junction protein expression during RABV infection. Further investigations demonstrate that the impaired expression of tight junction proteins is caused by the cytokines induced by RABV infection, rather than RABV infection per se. Of the cytokines induced by laboratory-attenuated RABV, IFN-γ is located in the hub of the cytokine network in brain, and neutralization of IFN-γ reduces both the disruption of BBB permeability in vivo and the downregulation of tight junction protein expression in vitro.

Daniels et al. (2014) did a systemic work to elucidate the influences of viral infection-induced innate cytokines on BBB integrity and found that during WNV infection, the induction of type Ⅰ IFN regulates endothelial barrier permeability and tight junction formation via directly regulating small GTPases Rac1 and RhoA, and indirectly antagonizing the barrier-disrupting effects of TNF-α and IL-1β. Type Ⅰ IFN significantly hinders WNV transit across an intact barrier in vitro, while TNF-α and IL-1β promote WNV crossing barrier. In vivo, genetic attenuation of type Ⅰ IFN signaling results in enhanced BBB permeability, which is associated with tight junction disruption following peripheral or intracranial infection with WNV.

There are also some viral infections in brain that increase the virus transit across BBB without compromising BBB integrity. Alimonti et al. (2018) used a human induced pluripotent stem cell (iPSC)-derived BBB model and demonstrated that ZIKV infects brain endothelial cells (i-BECs) without compromising the BBB barrier integrity or permeability. Although no disruption to the BBB is observed during infection, ZIKV particles are released from the abluminal side of the BBB model and infect underlying iPSC-derived neural progenitor cells (i-NPCs). Palus et al. also reported that in an in vitro BBB model, tick-borne encephalitis virus (TBEV) crosses the BBB via a transcellular pathway without compromising the integrity of the cell monolayer (Palus et al. 2017). These data suggest that neuroinvasive virus can also cross BBB through the active transportation by BMECs. TAM receptors have been previously reported to mediate the uptake of plasma membrane-derived microparticles (PMPs), which are released by activated platelets (Happonen et al. 2016). Viruses, in particular, enveloped viruses are likely to be actively transported across BBB by BMECs using TAM receptors.

In sum, TAM receptors safeguard the BBB integrity through maintaining the intact function of tight junction and inhibiting BBB-disrupting cytokine production during viral infection. Given some neuroinvasive viruses pass the BBB through transcellular transit, and TAM receptors may play roles in this process, further studies may focus on the potential impacts of TAM receptors on virus transendothelial transportation into brain.

TAM receptors are expressed in neurons, astrocytes, microglia, neural progenitor cells, BMECs as well as other neural cells. The specific expression and expression level of individual TAM receptor depend on the neural cell type, developmental stage, and pathophysiological status (Shafit-Zagardo et al. 2018). TAM receptors confer very excellent protection for diverse neural cells and keep CNS homeostasis under physiological and pathological conditions.

Li et al. found simultaneous knockout of Tyro3, Axl and Mertk causes BBB disruption, increased production of pro-inflammatory cytokines, autoantibodies and autoreactive lymphocytes infiltrating into brain parenchyma. This chronic neuroinflammation causes damage of the hippocampal mossy fibers and induces neuronal apoptotic death (Li et al. 2013). Li et al.'s data suggest physiologically, TAM receptors are critically important for maintaining BBB integrity, alleviating inflammation, and favoring neural cell survival, which together contribute to CNS homeostasis. Tyro3 is critically important for CNS myelination (Akkermann et al. 2017; Blades et al. 2018). TAM receptors promote the survival, proliferation, and differentiation of neural stem cells by regulating the expression of neurotrophins, especially the nerve growth factor (Ji et al. 2014). TAM receptors promote adult hippocampal neurogenesis and lack of TAM receptors impairs hippocampal neurogenesis, which is largely attributed to exaggerated inflammatory responses by microglia featured by increased mitogen-activated protein kinase (MAPK) and NF-κB activation and elevated production of pro-inflammatory cytokines that hinder neural stem cell proliferation and neuronal differentiation (Ji et al. 2014). Weinger et al. (2011) reported during experimental autoimmune encephalitis (EAE) in mice, Axl-KO mice show a more severe acute phase of EAE than wild type (WT) mice. AxlKO mice manifest more spinal cord lesions with larger inflammatory cuffs, more demyelination, and more axonal damage than WT mice during EAE. Additionally, Axl-KO mice show inefficient clearance of myelin debris in CNS, which may result from fewer activated microglia/macrophages in and/or surrounding lesions in Axl-KO mice relative to WT mice. These data suggest that Axl mitigates spinal cord lesions during EAE through alleviating neuroinflammation and promoting clearance of myelin debris. Goudarzi et al. (2016) found that Gas6/TAM signaling directly stimulates the generation of oligodendrocytes and myelin via Tyro3 receptor in the adult murine CNS, including repair of demyelinating lesions. Gas6/TAM signaling also launches STAT3 activation in murine optic nerves, followed by downregulation of various genes involved in multiple sclerosis. Ray et al. specifically studied the roles of Gas6/Axl signal axis in regulating CNS homeostasis after cuprizone exposure (Ray et al. 2017). They discovered that naive adult Gas6/Axl double knockout (Gas6-Axl-DKO) and WT mice have comparable myelination and equal numbers of axons and oligodendrocytes in the murine corpus callosum. However, after cuprizone exposure, Gas6-Axl-DKO mice display more extensive axonal swellings containing autophagolysosomes and multivesicular bodies, fewer myelinated axons, and less myelin in brain than WT mice, suggesting that Gas6/Axl signaling promotes axon integrity, reducing axon demyelination after cuprizone exposure. While Gas6-Axl-DKO does not affect the number of astrocytes or microglia in brain, it significantly increases the pro-inflammatory cytokine production by regulating SOCSs in brain, suggesting Gas6/Axl signaling can attenuate neuroinflammation induced by cuprizone.

There is only one retrievable paper discussing the roles of TAM receptors in regulating neuroinflammation during viral infection. Miner et al. (2015) reported although TAM receptors curb the production of many pro-inflammatory cytokines in serum during WNV infection, they fail to hinder the induction of pro-inflammatory cytokines in brain. Miner did not show whether TAM receptors have an effect on brain damage during WNV infection. We ourselves tested the role of Axl in regulating neuroinflammation and brain lesions during JEV infection, and found that Axl significantly attenuates the production of some pro-inflammatory cytokines, reduces the infiltration of cytotoxic lymphocytes, and alleviates neuron death in brain during JEV infection (Wang et al. 2020). Given that many viruses infect neural cells and TAM receptors inhibit the inflammatory responses of these cells, TAM receptors may also confer protection against viral encephalitis through maintaining BBB integrity, mitigating neuroinflammation, and promoting neural cell survival.

TAM receptors have been reported to inhibit the apoptosis of multitudinous cancer cells (Keating et al. 2010; Kim et al. 2015; Oien et al. 2017; Smart et al. 2018; Dufour et al. 2019). Najafov et al. (2019) discovered that TAM receptors promote necroptosis in various cellular models via regulating oligomerization of MLKL. Macrophages are major target cells for flaviviruses. Han et al. (2016) reported that Axl activation by Gas6 promotes the autophagy via activating MAPK 14, while inhibiting the pyroptosis of peritoneal macrophages treated with LPS and ATP via sequestering NLRP3 (NLR family, pyrin domain containing 3) inflammasome activation.

Currently, only very few studies have investigated the regulation of TAM receptors on the cell death of virusinfected cells. Wells et al. (2016) reported that genetic ablation of Axl has no effect on ZIKV-induced cell death in human induced pluripotent stem cell (iPSC)-derived neural progenitor cells or cerebral organoids. Hastings et al. (2019) reported that Axl promotes microglia apoptosis in ZIKV-infected IFNAR knockout murine brains. However, the finding may be disputable since Axl physically associates with IFNAR and there are obvious interactions between Axl signaling and type Ⅰ IFN signaling. Liu et al. (2010) reported that Axl is induced in Kaposi sarcoma herpesvirus (KSHV) transformed endothelial cells. Moreover, expression of KSHV latency protein vFLIP is sufficient to upregulate Axl expression in endothelial cells. Blockage of Axl by a specific monoclonal antibody MAb173 which induces Axl degradation results in increased apoptosis of KSHV-infected endothelial cells. Mechanistically, Axl activation stimulates PI3K-Akt signaling pathway, which has long been reported to inhibit cell apoptosis (Liu et al. 2010). Beier et al. (2015) reported that lymphocytic choriomeningitis virus (LCMV) infection in mouse liver induces Axl expression, and this expression is correlated with hepatocyte proliferation. We have discovered that Axl impedes the pyroptosis and apoptosis of JEV-infected macrophages through PI3K-Akt signaling pathway, and PI3K inhibitor LY294002 can significantly increase the pyroptosis and apoptosis of JEV-infected macrophages (Wang et al. 2020).

In sum, in spite of the abundant evidence showing that TAM receptors are pro-survival, their effects on the cell fate of virus-infected cells remain to be elucidated. Different death modes of virus-infected cells may result in different infection outcomes. For instance, apoptosis usually blocks viral replication and avoids inflammatory responses, while pyroptosis and necroptosis usually release progeny viruses and promote inflammatory response, which contributes to viral pathogenesis. So, the regulation of TAM receptors on cell death modes during viral infection needs to be further studied.

TAM receptors play multifaceted roles in viral infection. Figure 3 is a mind map that summarizes the major functions of TAM receptors during viral infection. However, there are still some enigmatic and interesting problems remaining to be solved. For instance, some viruses cross BBB through transendothelial active transportation, and what are the roles of TAM receptors in this process? TAM receptors are broadly expressed in various immune cells and regulate their functions, and what are the roles of these immune cells in viral infection? Given that TAM receptors confer potent protection against neuroinvasion and CNS lesions during neuroinvasive virus infection, their natural agonist Gas6/Pros1, as well as other small-molecule agonists, may have potential application for preventing severe viral encephalitis.

Figure 3.  The major functions of TAM receptors in viral infection. TAM receptors have five major functions in viral infection, including mediating viral entry, suppressing host innate immune and inflammatory responses, maintaining blood–brain barrier integrity and inhibiting viral neuroinvasion, alleviating neuroinflammation and brain lesions, and regulating cell death modes of virus-infected cells.

We thank Dr. Huang Wei Hsun from the International School of Capital Medical University, China for his kind and professional help in language editing of this manuscript. This work is supported by the National Natural Science Foundation of China (81671971, 81871641, 81972979, U1902210 and U1602223), the Scientific Research Plan of the Beijing Municipal Education Committee (KM201710025002), the Key Project of Beijing Natural Science Foundation B (KZ201810025035), and the Support Project of High-level Teachers in Beijing Municipal Universities in the Period of 13th Five-year Plan (IDHT20190510).

The authors declare that they have no conflict of interest.

This article does not contain any studies with human or animal subjects performed by any of the authors.