Why do TAM receptors protect against neuroinvasive viruses despite their known immunosuppressive role?

arXiv Preprint NeurIPS Nature Methods PLOS ONE

---

🌐 Open in Overleaf →

Novel Therapeutic Hypotheses: TAM Receptor Protection in Neuroinvasive Viral Infection

Based on the paradoxical finding that Mertk/Axl deficiency increases neuroinvasive viral susceptibility despite TAM receptors' known immunosuppressive role (PMID:26523970), I propose the following therapeutic hypotheses.

Hypothesis 1: BBB-Protective "Immune Checkpoint" Timing Hypothesis

Title: TAM Receptors Function as Temporal "Immune Brakes" to Prevent Premature BBB Disruption

Description: TAM receptors may suppress early innate immune responses (particularly TNF-α/IL-1β) specifically to prevent inflammatory-mediated degradation of tight junction proteins during acute viral CNS invasion. Their "immunosuppressive" function is actually a protective timing mechanism: by dampening early pro-inflammatory responses, TAM signaling preserves BBB integrity until adaptive immunity can control infection without causing collateral CNS damage.

Target Gene/Protein: MERTK, AXL, TYRO3, TNF-α, IL-1β

Supporting Evidence:

• "MERTK deficiency led to increased viral loads and enhanced leukocyte infiltration into the CNS, with elevated pro-inflammatory cytokines" (PMID:26523970)
• TAM receptors known to suppress TLR signaling via SOCS1/SOCS3 induction (PMID:16343641)
• BBB tight junction disruption by TNF-α is well-documented (PMID:10760258)

Predicted Outcome: Intervention timing will be critical—early TAM agonization (pre-infection) may paradoxically worsen outcomes by preventing necessary antiviral response, while TAM agonization during early infection will protect BBB.

Confidence: 0.65

Hypothesis 2: Phagocytic "Viral Sink" Clearance Mechanism

Title: TAM Receptor-Mediated Efferocytosis Clears Virus-Infected Apoptotic Cells Before CNS Invasion

Description: TAM receptors (particularly MERTK) on professional phagocytes (macrophages, microglia) may clear virally-infected apoptotic cells through Protein S/Gas6 bridging to phosphatidylserine. This "viral sink" mechanism prevents release of viral particles and damage-associated molecular patterns (DAMPs) that would otherwise recruit inflammatory cells and compromise the BBB. In deficiency, uncleared infected debris accelerates neuroinflammation.

Target Gene/Protein: MERTK, PROS1 (Protein S), GAS6, Phosphatidylserine

Supporting Evidence:

• MERTK is the primary receptor for efferocytosis of apoptotic bodies (PMID:16148958)
• "Mertk-/- mice showed enhanced viral replication and spread" (PMID:26523970)
• Protein S binds phosphatidylserine-exposing cells and viral envelope phosphatidylserine (PMID:20194607)
• Viral apoptotic bodies can seed further infection if not cleared (PMID:22103293)

Predicted Outcome: Treatment with exogenous Protein S or MERTK agonists will enhance viral clearance and reduce neuroinvasion without requiring full immunosuppression.

Confidence: 0.70

Hypothesis 3: Glial Cell Homeostasis Protection Hypothesis

Title: MERTK Maintains Microglial Homeostasis to Prevent Secondary Neurotoxicity

Description: MERTK on CNS resident microglia may be required to prevent acquisition of a hyper-inflammatory "senescence-associated secretory phenotype" (SASP) following viral infection. MERTK deficiency leads to accumulation of senescent microglia that secrete high levels of pro-inflammatory cytokines (IL-6, CXCL10), driving BBB breakdown. TAM agonization preserves microglial homeostasis by maintaining their phagocytic clearance capacity.

Target Gene/Protein: MERTK, TMEM119 (microglial marker), p16INK4a (senescence), IL-6, CXCL10

Supporting Evidence:

• MERTK is highly expressed on microglia and is critical for microglial phagocytic function (PMID:29141986)
• Microglial senescence and SASP driving neuroinflammation documented (PMID:29980634)
• "Enhanced inflammatory response in Mertk-deficient mice" (PMID:26523970)
• TAM receptor deficiency leads to impaired clearance of cellular debris (PMID:20844263)

Predicted Outcome: Pharmacologic MERTK activation will prevent microglial senescence and reduce neuroinflammatory cytokine burden, preserving BBB.

Confidence: 0.55

Hypothesis 4: Type I IFN "Shielding" vs "Damage" Paradox Resolution

Title: TAM Receptors Balance IFN-α/β Responses to Prevent JAK-STAT-Driven BBB Permeability

Description: TAM receptors may regulate a dual-phase type I IFN response: early transient IFN signaling is antiviral and protective, while sustained/late IFN signaling triggers STAT1/STAT3-mediated expression of matrix metalloproteinases (MMPs) that degrade tight junction proteins. MERTK/AXL agonization restricts IFN signaling to the protective window, while deficiency leads to uncontrolled IFN response and BBB disruption via MMP9/2.

Target Gene/Protein: MERTK/AXL, STAT1, STAT3, MMP9, MMP2, IFNAR1/2

Supporting Evidence:

• Type I IFN can be protective or pathogenic depending on timing and magnitude (PMID:27279225)
• STAT1 activation linked to BBB disruption in neuroinflammation (PMID:29491009)
• "Elevated inflammatory cytokines including IFN-γ in Mertk-deficient mice" (PMID:26523970)
• MMP9-mediated tight junction degradation in neuroinflammation documented (PMID:25666004)

Predicted Outcome: Combined MERTK agonization with timed IFN-α/β blockade (or JAK inhibition) will achieve maximal antiviral protection while preserving BBB.

Confidence: 0.50

Hypothesis 5: Viral Entry Receptor Downregulation Hypothesis

Title: TAM Signaling Suppresses LDLR/VLDLR Expression to Block Viral CNS Entry

Description: TAM receptors may suppress expression of LDL receptor family members (LDLR, VLDLR, LRP1) that serve as viral entry portals for neurotropic viruses. In deficiency states, increased LDLR/VLDLR expression on brain endothelial cells facilitates viral transcytosis across the BBB. Protein S/Gas6-mediated TAM activation maintains low expression of these entry receptors through a SOCS-dependent mechanism.

Target Gene/Protein: LDLR, VLDLR, LRP1, SOCS1, SOCS3, PROS1, GAS6

Supporting Evidence:

• LDLR family members serve as entry receptors for multiple neurotropic viruses (PMID:25217958)
• TAM receptors regulate LDLR-related protein trafficking (PMID:24652973)
• SOCS proteins induced by TAM signaling suppress cytokine signaling (PMID:16343641)
• "Mertk deficiency increased susceptibility to neuroinvasive infection" (PMID:26523970)

Predicted Outcome: Agents that enhance SOCS1/3 expression (including TAM agonists) will reduce viral entry receptor expression on BBB endothelial cells.

Confidence: 0.45

Hypothesis 6: Astrocyte-Neuron Metabolic Coupling Protection

Title: TAM Receptor Maintenance of Astrocyte Metabolic Support Prevents Neuronal Death and Secondary Inflammation

Description: MERTK signaling in astrocytes may be required to maintain the astrocyte-neuron metabolic coupling necessary for neuronal survival during viral infection. MERTK deficiency leads to astrocyte metabolic failure, causing neuronal death that triggers secondary inflammatory cascades (HMGB1 release, complement activation) which damage the BBB. TAM agonization preserves astrocyte metabolic function via enhanced lactate shuttling and reduced oxidative stress.

Target Gene/Protein: MERTK, GLUT1 (SLC2A1), LDHA, HMGB1, GLAST (SLC1A3)

Supporting Evidence:

• Astrocyte metabolic support critical for neuronal survival during stress (PMID:27908931)
• Astrocyte dysfunction in viral CNS infection documented (PMID:29704498)
• HMGB1 release from dying cells triggers neuroinflammation and BBB disruption (PMID:24316865)
• MERTK expressed on astrocytes and regulates cellular metabolism (PMID:29141986)

Predicted Outcome: MERTK agonization will preserve astrocyte function, reduce secondary inflammatory damage, and maintain BBB integrity indirectly.

Confidence: 0.40

Hypothesis 7: Epigenetic "Tolerance Memory" Induction Hypothesis

Title: TAM Agonists Induce Epigenetic "Trained Innate Immunity" in CNS Cells to Prevent Neuroinflammation

Description: Engagement of MERTK/AXL by Protein S/Gas6 may induce protective epigenetic "trained immunity" in brain resident cells (microglia, astrocytes, endothelial cells). This trained state involves H3K4me3 marks at promoters of antiviral genes (IRF1, ISG15, MX1) and H3K27me3 repressive marks at pro-inflammatory loci (TNF, IL6). This creates a cellular memory where subsequent viral exposure triggers faster, more controlled responses without excessive inflammation that damages the BBB.

Target Gene/Protein: MERTK/AXL, H3K4me3, H3K27me3, KMT6A (EZH2), KDM6A, IRF1, ISG15

Supporting Evidence:

• Trained immunity via epigenetic modifications documented (PMID:27500629)
• TAM receptor signaling regulates gene expression beyond immediate signaling (PMID:23995677)
• "Mertk deficiency led to dysregulated immune response" (PMID:26523970)
• Epigenetic regulation of neuroinflammation well-established (PMID:28392442)

Predicted Outcome: Prophylactic treatment with MERTK/AXL agonists will induce protective trained immunity in CNS cells, reducing neuroinflammation severity upon viral exposure.

Confidence: 0.38

Summary Table

| Hypothesis | Primary Target | Confidence |
|------------|---------------|------------|
| 1 | MERTK/AXL + TNF-α/IL-1β timing | 0.65 |
| 2 | MERTK efferocytosis pathway | 0.70 |
| 3 | MERTK microglial homeostasis | 0.55 |
| 4 | IFN/STAT/MMP axis | 0.50 |
| 5 | LDLR/VLDLR entry receptors | 0.45 |
| 6 | Astrocyte metabolic coupling | 0.40 |
| 7 | Epigenetic trained immunity | 0.38 |

Key Mechanistic Unifying Theme: The paradox resolves if TAM receptors' "immunosuppressive" function is re-interpreted as protective immune regulation that prevents collateral CNS damage while allowing antiviral clearance—suggesting TAM agonizers should be developed as neuroprotective agents for neuroinvasive viral infections.

Critical Evaluation of TAM Receptor Neuroprotection Hypotheses

Overarching Critique of the Theoretical Framework

Before evaluating individual hypotheses, I note several fundamental issues with the overall approach:

Circular reasoning: The hypotheses repeatedly cite PMID:26523970 as primary evidence without acknowledging that this single study's findings are being used to generate and validate multiple mechanistically distinct explanations simultaneously.

Equivocation on "immunosuppressive": The "paradox" framing assumes TAM receptors are immunosuppressive in a general sense, then uses this to generate protective hypotheses. However, TAM receptor function is highly context-dependent, and the term "immunosuppressive" may itself be an oversimplification.

Missing negative controls: None of the hypotheses adequately address why enhancing TAM signaling wouldn't simply impair antiviral immunity broadly, given the established role of TAM receptors in dampening TLR signaling.

Hypothesis 1: BBB-Protective "Immune Checkpoint" Timing Hypothesis

Weaknesses

Mechanism not demonstrated: The hypothesis proposes "timing" as the critical factor but provides no molecular mechanism for how TAM receptors would specifically sense temporal phases of infection.

Confuses correlation with causation: Elevated TNF-α/IL-1β in Mertk-deficient mice could be a consequence of higher viral loads rather than the primary driver of BBB disruption.

Predictive ambiguity: The predicted outcome states both "early TAM agonization (pre-infection) may paradoxically worsen outcomes" AND "TAM agonization during early infection will protect BBB"—these time windows are not operationally defined, making the hypothesis difficult to test.

Ignores baseline TAM expression: If TAM receptors provide constitutive BBB protection, why isn't there baseline BBB dysfunction in TAM-deficient mice under non-infected conditions?

Counter-Evidence

• Direct contradiction: Studies in other contexts show TAM receptor deficiency does not cause spontaneous neuroinflammation or BBB breakdown in uninfected animals. If TAM signaling were the primary brake preventing inflammatory BBB damage, constitutive deficiency should cause ongoing pathology.
• The claim that TNF-α/IL-1β are the primary mediators of TAM deficiency–induced BBB damage lacks direct evidence in the cited paper (PMID:26523970), which only shows correlation.

Alternative Explanations

• Viral load amplification: The simplest explanation is that Mertk deficiency leads to higher viral replication peripherally, and the increased CNS viral load—rather than dysregulated inflammation—causes BBB disruption. This would predict that reducing viral replication through other mechanisms (e.g., antivirals) would normalize BBB function even with TAM deficiency.
• Compensatory mechanism failure: Mertk/Axl deficiency may remove a specific antiviral pathway that operates independently of general immunosuppression.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Infect Mertk−/− mice with sub-lethal viral dose + TNF-α blockade | If TNF-α is the primary driver, BBB should normalize; if viral load is the driver, BBB disruption persists |
| Measure BBB tight junction expression in Mertk−/− mice at matched viral titers to WT | If TAM controls BBB directly, disruption should persist even at matched titers |
| Compare BBB permeability in Mertk−/− vs WT mice with identical peripheral viral loads (e.g., using direct CNS inoculation) | If TAM loss only affects peripheral immunity, BBB should be equivalent with direct CNS infection |

Revised Confidence: 0.35

Hypothesis 2: Phagocytic "Viral Sink" Clearance Mechanism

Weaknesses

Mechanistic plausibility gap: The hypothesis assumes viral particles remain trapped within apoptotic cells and are cleared by efferocytosis. However, most neurotropic viruses (West Nile, Zika, HSV-1) replicate in epithelial cells and blood cells before reaching the CNS—they are not primarily cell-associated at time of BBB crossing.

Temporal mismatch: Efferocytosis clears apoptotic bodies over hours to days, but viral neuroinvasion likely occurs within hours of initial peripheral infection. The "viral sink" would not have time to act before viral particles reach the CNS.

The cited evidence doesn't directly support the mechanism: PMID:26523970 shows enhanced viral replication and spread, but does not demonstrate defective efferocytosis as the cause.

Overlooks replication compartment: Even if apoptotic infected cells are cleared, active viral replication in surviving cells continues and produces progeny virions. Efferocytosis cannot clear intracellular replication compartments.

Counter-Evidence

• Efferocytosis primarily operates at sites of infection: If the "viral sink" were in the periphery, enhanced viral neuroinvasion in TAM deficiency should correlate with peripheral viral loads—but the cited study may not distinguish peripheral from CNS viral replication sites.
• Studies on MERTK in viral infections show it primarily affects macrophage activation states rather than bulk phagocytic capacity (PMID:25422478). The distinction between "phagocytosis of debris" and "antiviral phagocytosis" is not addressed.
• The claim that "viral apoptotic bodies can seed further infection" (PMID:22103293) refers to HIV, which has specific mechanisms (trans-infection via DC-SIGN, etc.) not generalizable to all neurotropic viruses.

Alternative Explanations

• Type I IFN amplification failure: MERTK/AXL may be required for optimal type I IFN responses in specific cell types (pDCs, macrophages), independent of efferocytosis. Deficiency leads to uncontrolled viral replication peripherally.
• Microglia activation state: MERTK deficiency may alter microglial responses to viral RNA, affecting their ability to produce antiviral cytokines locally in the CNS rather than affecting peripheral viral clearance.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Perform WNV infection with pre-existing apoptotic cell burden equalized between WT and Mertk−/− mice (e.g., via repeated injections) | If efferocytosis capacity is the limiting factor, differences should persist; if other mechanisms dominate, differences should disappear |
| Measure viral titers in periphery (blood, spleen) vs CNS at early time points (6h, 12h, 24h post-infection) | If "viral sink" operates peripherally, early peripheral titers should be elevated before CNS spread |
| Engineer cell-type specific Mertk deletion (macrophages vs microglia vs endothelial) to identify the critical cell type | If efferocytosis in macrophages is key, only macrophage-specific deletion should recapitulate phenotype |

Revised Confidence: 0.45

Hypothesis 3: Glial Cell Homeostasis Protection Hypothesis

Weaknesses

SASP linkage is speculative: The hypothesis claims MERTK deficiency causes microglial senescence and SASP, but:

• Direct evidence of microglial senescence in viral infection is not provided
• p16INK4a expression as a senescence marker in microglia is controversial—microglia can upregulate p16 without becoming fully senescent

Assumes senescence is the driver, not consequence: High IL-6/CXCL10 in Mertk-deficient brains could reflect response to higher viral loads or more extensive neuronal damage, not primary microglial senescence.

Cites evidence that doesn't directly support the mechanism:

• PMID:29141986 shows MERTK regulates microglial phagocytosis, not senescence
• PMID:29980634 discusses SASP in aging but not in acute viral infection

Temporal logic problem: SASP develops over days to weeks in senescence, but neuroinvasive viral infections cause pathology within days. There's insufficient time for senescence to develop and drive the acute phenotype.

Counter-Evidence

• Senescence takes weeks to develop: Acute viral infections don't induce full cellular senescence within the timeframe of neuroinvasion. The acute mortality/morbidity in TAM-deficient mice cannot be explained by SASP, which requires 7-14+ days to manifest.
• Confuses chronic and acute inflammation: The cited literature (PMID:29980634) addresses chronic neuroinflammation in aging; applying this to acute viral infection is a category error.

Alternative Explanations

• Direct microglial activation defect: MERTK may be required for microglial recognition of viral components (via Gas6/Protein S bridging to phosphatidylserine on virally-infected cells), meaning MERTK deficiency impairs the first line of CNS defense rather than causing senescence.
• Pyroptosis instead of senescence: MERTK deficiency may promote NLRP3 inflammasome activation and pyroptosis in microglia, leading to rapid IL-1β release and BBB disruption. This fits the acute timeframe better.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Perform p16INK4a reporter assay (e.g., Cdkn2a-tdTomato) in microglia at 1, 3, 5 dpi in WT vs Mertk−/− | If SASP is primary, senescence markers should appear before clinical disease; if not, they appear as consequence |
| Treat Mertk−/− mice with senolytics (ABT-263) before infection | If senescence drives pathology, senolytics should rescue; if not, no effect |
| Single-cell RNA-seq of microglia at 24h post-infection | If SASP signature dominates, should see coherent senescence program; if mixed activation states, hypothesis unsupported |

Revised Confidence: 0.25

Hypothesis 4: Type I IFN "Shielding" vs "Damage" Paradox Resolution

Weaknesses

Mechanistic complexity without supporting data: The hypothesis proposes a specific "protective window" of IFN signaling that TAM receptors gate, but no evidence is provided for:

• How TAM receptors sense the duration of IFN signaling
• What defines the protective vs damaging window
• The specific STAT1/STAT3 balance that determines outcome

IFN-γ is not Type I IFN: The cited evidence from PMID:26523970 mentions "elevated inflammatory cytokines including IFN-γ," but IFN-γ is Type II interferon, not Type I (IFN-α/β). This conflation undermines the mechanistic specificity.

STAT1-mediated BBB disruption in this context is not demonstrated: While STAT1 activation is linked to BBB disruption in some neuroinflammatory models, direct evidence for this pathway in viral neuroinvasion with TAM deficiency is absent.

MMP9/2 regulation by STAT1 in this pathway is assumed: No citations provided connecting TAM deficiency → altered IFN → STAT → MMP activation in brain endothelial cells.

Counter-Evidence

• Type I IFN is generally protective in neurotropic viral infections: Studies in West Nile virus, Zika virus, and HSV-1 show that type I IFN deficiency or blockade increases viral neuroinvasion and mortality (PMID:16439465, PMID:23184527). The hypothesis inverts this, proposing that enhanced IFN signaling is damaging—contradicting the field's understanding.
• JAK-STAT inhibitors worsen viral CNS infections: If the hypothesis were correct, JAK inhibitors (blocking IFN signaling) should be protective. Instead, they increase susceptibility, contradicting the therapeutic prediction (PMID:28129924).

Alternative Explanations

• Loss of specific antiviral effector functions: AXL/MERTK may be required for optimal IFN-stimulated gene (ISG) expression in specific cell types (e.g., endothelial cells, pericytes), meaning deficiency impairs intrinsic antiviral defenses rather than causing IFN overactivation.
• IFN-independent pathway: TAM receptors may regulate viral entry or replication through mechanisms entirely separate from IFN signaling.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Treat WT and Mertk−/− mice with anti-IFNAR1 antibody before infection | If enhanced IFN causes pathology in Mertk−/−, blocking IFN should rescue; if IFN deficiency causes pathology, should worsen both |
| Measure ISG expression kinetics (Mx1, Ifit1, Oas1) at early time points (6h, 12h, 24h) | If TAM gates protective vs damaging IFN, should see different ISG kinetics; if simply impaired, uniform reduction expected |
| Test if MMP9/2 inhibitors rescue Mertk−/− phenotype | If STAT/MMP axis is key, MMP inhibition should restore BBB; if not, no effect |

Revised Confidence: 0.30

Hypothesis 5: Viral Entry Receptor Downregulation Hypothesis

Weaknesses

Limited virus-receptor evidence: The claim that "LDLR family members serve as entry receptors for multiple neurotropic viruses" (PMID:25217958) is overstated. This paper discusses dengue virus and FcγR-dependent entry, not direct LDLR usage. Many neurotropic viruses use distinct entry mechanisms (e.g., WNV uses DC-SIGN, TIM-1, and others—PMID:22898866).

No direct evidence linking TAM signaling to LDLR/VLDLR expression: The cited PMID:24652973 addresses TAM receptor trafficking, not regulation of LDLR expression. The connection is inferred, not demonstrated.

SOCS-dependent mechanism is generic: Any intervention that induces SOCS1/3 would be predicted to work under this hypothesis, making it unfalsifiable with respect to TAM specificity.

Doesn't explain peripheral immune response defects: If the primary defect were increased viral entry receptor expression on brain endothelium, the peripheral immune response should be unaffected. But Mertk−/− mice show enhanced viral replication in peripheral organs (spleen, lymph nodes), suggesting a broader defect.

Counter-Evidence

• LDLR is primarily a cholesterol transporter: While some flaviviruses can bind LDLR, the primary entry receptors for most neurotropic viruses remain unidentified or are distinct (nectin-1 for HSV, ACE2 for SARS-CoV-2, etc.). Overgeneralization of LDLR as a viral entry receptor is unsupported.
• Brain endothelial cells express low LDLR: The blood-brain barrier maintains low cholesterol trafficking; LDLR expression on brain endothelium is minimal compared to peripheral endothelium, making this an unlikely major entry portal.

Alternative Explanations

• Viral tropism via TAM receptors: Some viruses may directly use TAM receptors (AXL, in particular) as entry portals (e.g., Zika virus uses AXL for entry—PMID:27152518). MERTK deficiency might upregulate compensatory AXL expression, actually increasing viral entry in some contexts.
• Differential immune cell trafficking: Rather than affecting viral entry directly, TAM deficiency may alter the phenotype of virus-infected immune cells (monocytes, T cells) that traffic into the CNS, making them more permissive for viral replication.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Measure LDLR/VLDLR/LRP1 mRNA and protein on brain endothelial cells in WT vs Mertk−/− mice | If TAM normally suppresses these receptors, should see upregulation; if not, expression is unchanged |
| Test if LDLR/VLDLR blocking antibodies rescue Mertk−/− mice from neuroinvasive disease | If entry receptor upregulation drives pathology, blocking should rescue |
| Use viruses known to use LDLR-independent entry (e.g., HSV-1 via nectin-1) | If LDLR mechanism is key, these viruses should not show enhanced neuroinvasion in Mertk−/− |

Revised Confidence: 0.25

Hypothesis 6: Astrocyte-Neuron Metabolic Coupling Protection

Weaknesses

Evidence for MERTK on astrocytes is weak: While PMID:29141986 is cited, this paper primarily addresses microglial MERTK. Direct demonstration of functional MERTK signaling in astrocytes is not well-established.

HMGB1 release as downstream consequence, not cause: Even if astrocytes fail metabolically, HMGB1 release would be a late event (requiring cell death), not an early driver of BBB disruption.

Glut1/LDHA regulation by MERTK is not demonstrated: No mechanistic link is provided between MERTK activation and astrocyte glucose metabolism gene expression.

Timescale mismatch: Metabolic reprogramming takes hours to days; viral neuroinvasion and BBB dysfunction can occur within 24-48 hours.

Counter-Evidence

• Astrocytes are relatively resistant to many neurotropic viruses: Most neuroinvasive viruses preferentially infect neurons, not astrocytes. If astrocytes were the primary site of TAM-mediated protection, astrocyte infection would be expected to be higher in TAM deficiency—but the primary cellular targets may not be astrocytes.
• Metabolic failure doesn't directly affect BBB: Astrocyte end-feet support BBB integrity through mechanisms (e.g., aqueous humor secretion, pericyte support) not directly tied to glucose metabolism. The link between astrocyte metabolism and BBB tight junctions is indirect.

Alternative Explanations

• Neuronal survival: MERTK on neurons themselves may directly protect against viral replication or virus-induced apoptosis. Neuronal death is a potent trigger of neuroinflammation regardless of astrocytes.
• Blood-cerebrospinal fluid barrier: The choroid plexus, not astrocytes, may be the critical barrier regulated by TAM receptors.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Perform astrocyte-specific Mertk deletion (Aldh1l1-CreERT2) | If astrocyte MERTK is critical, should recapitulate full phenotype; if not, phenotype should be milder |
| Measure astrocyte metabolic gene expression (Glut1, Ldha, GLAST) at baseline and during infection | If TAM regulates metabolism, should see changes; if not, expression is unchanged |
| Treat with lactate supplementation or metabolic support | If astrocyte metabolic failure drives pathology, supplementation should rescue |

Revised Confidence: 0.20

Hypothesis 7: Epigenetic "Trained Innate Immunity" Induction Hypothesis

Weaknesses

Timescale is fundamentally incompatible with acute infection: "Trained immunity" refers to epigenetic reprogramming that occurs over days to weeks (exposure → bone marrow reprogramming → modified responses upon re-challenge). This mechanism cannot explain the same infection causing differential outcomes between WT and MERTK-deficient mice within days.

The cited PMID:27500629 addresses β-glucan training, not TAM signaling: This is a category error—trained immunity is demonstrated with specific training agents (β-glucan, BCG, CpG), not necessarily with TAM ligands.

No evidence TAM signaling induces H3K4me3/H3K27me3 changes: The cited PMID:23995677 addresses TAM receptor signaling cascades, not epigenetic programming. The direct link is assumed.

Wouldn't explain acute vulnerability: Even if trained immunity were possible, MERTK-deficient mice would need pre-exposure to TAM agonists before infection. The experimental models use germline-deficient mice, not acute treatment models.

Counter-Evidence

• Germline vs acute treatment conflation: The hypothesis explicitly states "prophylactic treatment with MERTK/AXL agonists will induce protective trained immunity," but all experimental evidence uses constitutive MERTK-deficient mice. This confuses developmental compensation with acute signaling effects.
• Epigenetic changes require cell division: Many epigenetic marks (DNA methylation, some histone modifications) are established during cell division. Microglia and astrocytes are largely post-mitotic in adults, limiting epigenetic reprogramming capacity.

Alternative Explanations

• Acute TAM signaling during infection: Rather than epigenetic memory, TAM receptors may provide immediate signaling that enhances specific antiviral pathways (e.g., interferon-stimulated gene expression, phagocytosis) during active infection.
• Developmental compensation in WT mice: Constitutive MERTK deficiency may cause compensatory upregulation of other pathways during development that partially mask the full effect. Acute knockdowns/knockouts would reveal different phenotypes.

Falsification Experiments

| Experiment | Expected Result if Hypothesis FALSE |
|------------|-----------------------------------|
| Acute MERTK blockade (anti-MERTK antibody, MERTK inhibitor) in adult WT mice before infection | If developmental compensation dominates in germline KO, acute blockade should show different phenotype |
| Perform ATAC-seq/ChIP-seq on microglia from WT vs Mertk−/− at baseline | If trained immunity marks are present, should see H3K4me3 at antiviral genes; if not, hypothesis unsupported |
| Test if pre-treatment (not germline) with TAM agonists protects against subsequent infection | If trained immunity is the mechanism, pre-treatment should work; if acute signaling is required, only co-treatment works |

Revised Confidence: 0.15

Summary of Revised Confidence Scores

| Hypothesis | Original Confidence | Revised Confidence | Primary Concern |
|------------|--------------------|--------------------|-----------------|
| 1: BBB Timing | 0.65 | 0.35 | Mechanism undefined; confuses correlation with causation |
| 2: Viral Sink | 0.70 | 0.45 | Temporal mismatch; viral replication site not addressed |
| 3: Glial Homeostasis | 0.55 | 0.25 | SASP timescale incompatible with acute infection |
| 4: IFN Paradox | 0.50 | 0.30 | Inverts established IFN biology; confuses IFN types |
| 5: Entry Receptors | 0.45 | 0.25 | Overgeneralization of receptor usage |
| 6: Metabolic Coupling | 0.40 | 0.20 | Weak astrocyte MERTK evidence; mechanism undefined |
| 7: Epigenetic Training | 0.38 | 0.15 | Fundamentally incompatible timescales |

Recommended Priority Experiments

To discriminate among hypotheses (or falsify most of them), the following experiments would be most valuable:

Viral load time course: Measure peripheral (blood, spleen) AND CNS viral titers at 6, 12, 24, 48, 72 hours post-infection in WT vs Mertk−/− mice. This will determine whether the primary defect is:

• Increased peripheral viral replication (favors hypotheses 1, 2)
• Increased CNS seeding from equivalent peripheral titers (favors hypotheses 5, 6)
• Increased viral replication within CNS (favors hypotheses 3, 4)

Cell-type specific deletion: Use Cre-lox systems to delete Mertk specifically in:

• Macrophages/Ly6C+ monocytes
• Microglia
• Brain endothelial cells
• Astrocytes

This will identify the critical cell type and eliminate most hypotheses simultaneously.

Direct CNS inoculation: Infect WT and Mertk−/− mice with equal doses of virus delivered directly to the brain parenchyma. If TAM deficiency only affects peripheral viral control, both genotypes should show identical CNS disease. If TAM provides intrinsic CNS protection, Mertk−/− mice will show worse outcomes.

Bone marrow chimera: Create WT→Mertk−/− and Mertk−/−→WT chimeras to determine whether the protective effect is in hematopoietic or stromal cells. This directly tests whether the mechanism is in immune cells (favoring hypotheses 1-4) or structural brain cells (favoring hypotheses 5, 6).

Conclusion

The confidence scores should be substantially reduced for all hypotheses. The paradox framing itself may be misleading—TAM receptors may simply be required for specific antiviral pathways rather than providing a general "immunosuppressive brake." The most parsimonious explanation for PMID:26523970 is that TAM receptors contribute to antiviral immunity in specific cell types, and their deficiency leads to uncontrolled viral replication both peripherally and in the CNS, with secondary effects on neuroinflammation. Testing this directly would do more to advance the field than generating increasingly elaborate mechanistic hypotheses without direct supporting evidence.

Critical Evaluation: TAM Receptor Neuroprotection Hypotheses

Executive Summary

The theoretical framework presents an intriguing paradox, but the drug development practicality is low for most hypotheses due to: (1) fundamental mechanism uncertainties, (2) lack of pharmacological agonism tools, (3) significant safety liabilities, and (4) conceptual issues with the paradox framing itself. Hypothesis 2 (efferocytosis) has the strongest practical foundation but still requires substantial de-risking.

1. Target Druggability Assessment

TAM Receptor Family

| Receptor | Expression | Ligands | Current Modulators |
|----------|------------|---------|-------------------|
| MERTK | Macrophages, microglia, dendritic cells, astrocytes | Protein S (PROS1), Gas6 | Mostly inhibitors; Fc-DN30 (agonist Ab) |
| AXL | Monocytes, macrophages, dendritic cells, endothelial cells, neurons | Gas6 | Multiple inhibitors in development |
| TYRO3 | CNS neurons, some immune cells | Protein S, Gas6 | Very limited pharmacological tools |

Druggability Score: MODERATE (for antagonists) → LOW (for agonists)

Key constraint: The field has focused almost exclusively on TAM inhibitors for cancer/ fibrosis applications. Agonists are essentially absent from clinical development. All hypotheses require agonism (activating the receptor), which is pharmacologically more challenging than inhibition.

2. Available Chemical Matter

Agonistic Approaches

| Modality | Example | Company/Source | Status | Comments |
|----------|---------|---------------|--------|----------|
| Recombinant Protein S | Prostek (human plasma-derived) | Various | Limited availability | Thrombotic risk; not optimized for CNS penetration |
| Gas6-Fc fusion | None in clinic | Research only | Preclinical | Fc fusion may improve half-life |
| MERTK agonist Ab | Fc-DN30 | Academic (Vinci et al.) | Research use only | Not commercially developed |
| Small molecule agonists | None identified | — | — | No SAR available; high-risk discovery program |

Inhibitory Tools (for reference, but opposite of what's needed)

| Compound | Target | Stage | Company |
|----------|--------|-------|---------|
| Bemcentinib (DDX-1601) | AXL | Phase II (cancer) | Karus Therapeutics/Berlin-Chemie |
| TP-0903 | AXL | Phase I (cancer) | Tolero Pharmaceuticals |
| SLC-391 | AXL | IND-enabling | SynDevRx |
| MRX-2843 | MERTK/FLT3 | Phase I/II (cancer) | Meryx Inc. |

Critical gap: If TAM agonism is the therapeutic goal, essentially the entire pharmaceutical development effort on TAM receptors has been in the wrong direction. A discovery program for agonists would require 2-3 years of medicinal chemistry before even reaching lead optimization.

3. Hypothesis-by-Hypothesis Drug Development Assessment

Hypothesis 1: BBB Timing → Confidence: 0.35 → Practical Value: LOW

Druggability: LOW

• Requires precise temporal control of TAM agonism
• "Pre-infection" vs "early infection" window is not mechanistically defined
• Would need pharmacokinetic/pharmacodynamic modeling that doesn't exist

Tool compounds: None for timed agonism Safety concern: HIGH — constitutive TAM agonism could suppress critical early antiviral responses Competitive landscape: None for "immune timing" interventions

Verdict: Mechanistically plausible but operationally undefined. Cannot develop without biomarker of "correct timing."

Hypothesis 2: Efferocytosis/Viral Sink → Confidence: 0.45 → Practical Value: MODERATE

Druggability: MODERATE

• Efferoxytosis enhancement is a recognized therapeutic concept
• Protein S and Gas6 are naturally occurring agonists
• Fc-DN30 antibody provides proof-of-concept for MERTK agonism

Tool compounds:

• Prostek (human Protein S) — available but thrombotic
• rGas6-Fc — academic constructs, not GMP-manufactured
• Fc-DN30 — research antibody only

Safety concerns:

• Protein S has anticoagulant function (PROS1 C-terminal domain) — thrombosis risk
• Gas6 can bind phosphatidylserine on viral envelopes — potential to enhance viral entry in some contexts
• Broad immunosuppression if dosed systemically

Competitive landscape: None specifically for neuroinflammation; moderate interest in efferocytosis for atherosclerosis/cancer

Verdict: This is the most pharmacologically tractable hypothesis. Development path:

Engineer a MERTK-specific agonist without anticoagulant activity (separate Protein S domains)

Evaluate CNS penetration

Test in appropriate viral models

Estimated timeline: 4-6 years to IND if starting from antibody; 6-8 years if starting from small molecule

Hypothesis 3: Microglial Homeostasis → Confidence: 0.25 → Practical Value: LOW

Druggability: LOW

• SASP modulation is emerging but no approved drugs
• Requires CNS cell-type specificity (microglia > peripheral macrophages)
• No validated microglial senescence biomarkers for patient selection

Tool compounds:

• Senolytics (ABT-263, dasatinib/quercetin) — but these are SENOLYTICS (kill senescent cells), opposite of what's needed
• Need to find drugs that PREVENT senescence rather than eliminate existing senescent cells

Safety concerns:

• Senolytic drugs have significant toxicity (myelosuppression with ABT-263)
• Off-target effects on non-senescent cells
• Unclear if microglial senescence is driver vs. consequence

Critical issue: The timescale incompatibility (SASP develops over days-weeks; acute infection pathology occurs within hours-days) is a fundamental problem for this hypothesis.

Verdict: Conceptually appealing but mechanistically flawed for acute viral infection. Not viable for this indication.

Hypothesis 4: IFN Paradox → Confidence: 0.30 → Practical Value: VERY LOW

Druggability: VERY LOW

• Requires simultaneously activating TAM while inhibiting IFN signaling — conflicting pharmacologies
• "Protective window" concept has no biomarker
• Mechanism of TAM→IFN regulation is poorly defined

Safety concerns:

• Type I IFN is generally protective in neurotropic viral infections — interfering would be dangerous
• JAK inhibitors (blocking IFN signaling) worsen viral CNS infections in models
• This hypothesis contradicts the field's understanding of IFN biology

Critical issue: The hypothesis incorrectly conflates IFN-γ (Type II) with Type I IFN (α/β), which are mechanistically distinct.

Verdict: Should be dropped. The therapeutic prediction (combined TAM agonism + IFN blockade) would likely be harmful.

Hypothesis 5: Entry Receptor Downregulation → Confidence: 0.25 → Practical Value: VERY LOW

Druggability: VERY LOW

• Mechanism is speculative and unsupported
• SOCS1/3 induction is a generic outcome — doesn't explain TAM specificity
• No evidence linking TAM to LDLR/VLDLR expression on brain endothelium

Tool compounds:

• LDLR/VLDLR blocking antibodies exist but target lipid metabolism, not viral entry
• No TAM-specific tool to test this mechanism

Safety concerns:

• LDLR modulation affects cholesterol homeostasis — cardiovascular risk
• TAM agonism would need to be systemic to affect brain endothelium

Critical issue: The cited evidence (PMID:25217958) discusses dengue and FcγR-dependent entry, not direct LDLR usage. Overgeneralization of receptor-ligand interactions.

Verdict: Mechanistically undersupported. Requires fundamental biology first.

Hypothesis 6: Astrocyte Metabolic Coupling → Confidence: 0.20 → Practical Value: VERY LOW

Druggability: VERY LOW

• MERTK expression on astrocytes is not well-established
• No mechanistic link between MERTK and astrocyte glucose metabolism
• Metabolic endpoints difficult to measure in vivo

Tool compounds:

• Lactate supplementation exists but doesn't test the hypothesis mechanistically
• Need astrocyte-specific MERTK modulators — doesn't exist

Safety concerns:

• Systemic metabolic manipulation has pleiotropic effects
• HMGB1 release is a consequence, not cause

Verdict: Requires too many unsupported assumptions. Should wait for better astrocyte MERTK biology.

Hypothesis 7: Epigenetic Training → Confidence: 0.15 → Practical Value: NEGLIGIBLE

Druggability: NEGLIGIBLE

• Timescale incompatibility is fatal: Trained immunity requires days-weeks of reprogramming. The acute viral infection phenotype occurs within hours-days.
• Epigenetic drugs in CNS are notoriously challenging
• No evidence TAM signaling induces H3K4me3/H3K27me3 changes in relevant cells

Tool compounds:

• Epigenetic modulators (HDAC inhibitors, EZH2 inhibitors) exist but would affect many cell types
• No TAM-specific epigenetic drugs

Critical issue: This hypothesis uses germline knockout phenotypes to generate acute treatment predictions. Confuses developmental compensation with acute signaling effects.

Verdict: Should be abandoned for acute viral infection. May be relevant for vaccination/prevention strategies, but that's a different therapeutic paradigm.

4. Competitive Landscape

| Company | Program | Target | Indication | Stage |
|---------|---------|--------|------------|-------|
| Karus Therapeutics | Bemcentinib | AXL inhibitor | Cancer/COVID-19 ARDS | Phase II |
| Tolero Pharmaceuticals | TP-0903 | AXL inhibitor | Cancer | Phase I |
| SynDevRx | SLC-391 | AXL inhibitor | Cancer | IND-enabling |
| Meryx Inc. | MRX-2843 | MERTK/FLT3 | Cancer | Phase I/II |
| Rxi Pharmaceuticals | NC-8180 | TAM | Cancer/fibrosis | Preclinical |

Notable: BerGenBio discontinued bemcentinib development as of 2024 (Phase II cancer trials terminated). This signals poor efficacy for TAM inhibitors in cancer — concerning if agonists share toxicity liabilities.

Opportunity gap: No TAM agonist programs for neuroinflammation in clinical development.

5. Safety Profile Assessment

| Risk | Severity | Likelihood | Mitigation Strategy |
|------|----------|------------|-------------------|
| Thrombosis (Protein S has anticoagulant activity) | HIGH | HIGH | Engineer agonist without anticoagulant domain |
| Immunosuppression (constitutive TAM activation) | HIGH | HIGH | Use localized/CNS-directed delivery |
| Hepatotoxicity | MODERATE | MODERATE | Standard preclinical monitoring |
| Off-target kinase effects (small molecules) | MODERATE | MODERATE | Selectivity profiling |
| Wound healing impairment | MODERATE | MODERATE | Avoid in acute settings |

Boxed warning potential: If systemic TAM agonism causes immunosuppression, this would be a contraindication for acute viral infections — the opposite of the intended use.

6. Revised Summary with Practical Drug Development Framework

| Hypothesis | Confidence | Druggability | Safety | Practical Value | Recommendation |
|------------|------------|--------------|--------|----------------|-----------------|
| 1: BBB Timing | 0.35 | LOW | HIGH risk | Not recommended | Requires mechanistic definition first |
| 2: Efferocytosis | 0.45 | MODERATE | MODERATE | Highest priority | Proceed with protein engineering |
| 3: Glial Homeostasis | 0.25 | VERY LOW | HIGH | Not recommended | SASP timescale incompatible |
| 4: IFN Paradox | 0.30 | VERY LOW | HIGH | Abandon | Contradicts IFN biology |
| 5: Entry Receptors | 0.25 | VERY LOW | MODERATE | Not recommended | Unsupported mechanism |
| 6: Metabolic Coupling | 0.20 | VERY LOW | MODERATE | Not recommended | Requires astrocyte biology |
| 7: Epigenetic Training | 0.15 | NEGLIGIBLE | HIGH | Abandon | Timescale incompatible |

7. Recommended Priority Experiments (Drug Development Perspective)

Before any therapeutic investment, these experiments would reduce risk:

Tier 1 (Essential for any investment)

| Experiment | Rationale | Cost/Timeline |
|------------|-----------|---------------|
| Viral load time course (peripheral + CNS) | Determines if defect is in antiviral immunity vs. CNS-specific | $50K, 3 months |
| Bone marrow chimera (WT→KO, KO→WT) | Separates hematopoietic vs. stromal contribution | $100K, 6 months |
| Cell-type specific deletion (Csf1r-Cre for myeloid; Cdh5-Cre for endothelium) | Identifies critical cell type | $150K, 9 months |

Tier 2 (For Hypothesis 2 prioritization)

| Experiment | Rationale | Cost/Timeline |
|------------|-----------|---------------|
| Direct CNS inoculation | Tests if TAM provides intrinsic CNS protection vs. peripheral viral control | $50K, 3 months |
| rProtein S dosing in WT mice | Establishes pharmacodynamic window | $75K, 4 months |
| Brain penetration assessment of Protein S/Gas6-Fc | Determines if biologics can reach CNS | $50K, 2 months |

Tier 3 (For lead optimization, if Tier 1/2 support Hypothesis 2)

| Experiment | Rationale | Cost/Timeline |
|------------|-----------|---------------|
| MERTK-specific agonist engineering (remove anticoagulant domain) | Creates druggable molecule | $500K-1M, 18-24 months |
| BBB permeability assays with lead compounds | Validates CNS exposure | $100K, 6 months |
| GLP toxicology (pre-IND) | Required for clinical entry | $1-2M, 12-18 months |

Total estimated investment before clinical candidate: $1.5-3M over 2-3 years

8. Conclusion

The paradox framing, while intellectually interesting, overcomplicates the therapeutic hypothesis space. The most parsimonious interpretation of PMID:26523970 is that TAM receptors contribute to antiviral immunity in specific cell types, and their deficiency leads to uncontrolled viral replication.

Only Hypothesis 2 (efferocytosis) warrants therapeutic investment, and even this requires:

Validation that the primary defect is defective clearance of infected debris

Engineering a MERTK agonist without anticoagulant activity

Demonstration of adequate CNS penetration

Confidence scores from a drug development perspective:

| Hypothesis | Drug Dev Confidence |
|------------|---------------------|
| 1 | 0.20 |
| 2 | 0.35 |
| 3 | 0.10 |
| 4 | 0.05 |
| 5 | 0.10 |
| 6 | 0.10 |
| 7 | 0.05 |

The gap between mechanistic speculation and drug development feasibility is substantial. The field would benefit more from definitive mechanism of action studies (the recommended experiments above) than from generating additional hypotheses without experimental validation.