Toll-like receptor 4 in CNS pathologies

Authors

  • Madison M. Buchanan,

    1. Department of Chemistry and Biochemistry, University of Colorado at Boulder, Boulder, Colorado, USA
    2. The Center for Neurosciences, University of Colorado at Boulder, Boulder, Colorado, USA
    Search for more papers by this author
  • Mark Hutchinson,

    1. Discipline of Pharmacology, University of Adelaide, Adelaide, SA, Australia
    Search for more papers by this author
  • Linda R. Watkins,

    1. The Center for Neurosciences, University of Colorado at Boulder, Boulder, Colorado, USA
    2. Department of Psychology, University of Colorado at Boulder, Boulder, Colorado, USA
    Search for more papers by this author
  • Hang Yin

    1. Department of Chemistry and Biochemistry, University of Colorado at Boulder, Boulder, Colorado, USA
    2. The Center for Neurosciences, University of Colorado at Boulder, Boulder, Colorado, USA
    Search for more papers by this author

Address correspondence and reprint requests to Hang Yin, Department of Chemistry and Biochemistry, 215 University of Colorado at Boulder, Boulder, CO 80309-0215, USA. E-mail: hang.yin@colorado.edu

Abstract

J. Neurochem. (2010) 114, 13–27.

Abstract

The responses of the brain to infection, ischemia and trauma share remarkable similarities. These and other conditions of the CNS coordinate an innate immune response marked by activation of microglia, the macrophage-like cells of the nervous system. An important contributor to microglial activation is toll-like receptor 4, a pathogen-associated molecular pattern receptor known to initiate an inflammatory cascade in response to various CNS stimuli. The present review traces new efforts to characterize and control toll-like receptor 4 in inflammatory etiologies of the nervous system.

Abbreviations used:

amyloid-β peptide

AD

Alzheimer’s disease

AP-1

activator protein-1

IFN-β

interferon-β

IKK

inhibitor of NF-κB kinase

IL

interleukin

IR

interleukin receptor

IRAK

interleukin-1 receptor-associated kinase

IRF3

interferon regulatory factor 3

LPS

lipopolysaccharide

Mal

MyD88 adaptor-like protein

MD-2

myeloid differentiation factor 2

NF-κB

nuclear factor kappa-B

TIR

toll/IL-1 receptor signaling domain

TLR

toll-like receptor

TNFα

tumor necrosis factor-α

TRIF

toll/IL-1 receptor-containing adaptor inducing IFN-β

A toll-encoding gene was originally discovered for its role in dorsal-ventral axis development in Drosophila embryos (Anderson et al. 1985a, b). From its sequence, the toll gene product was determined to be a transmembrane receptor with a cytoplasmic domain similar to the interleukin-1 receptor and a large ectodomain with leucine-rich repeat sequences (Hashimoto et al. 1988). A human analogue of Drosophila toll was later identified; its signaling pathway suggested a role in the evolutionarily conserved host defense mechanism (Miyake et al. 1995). Plants, insects and vertebrates all use homologous mechanisms relying on toll recognition to coordinate an immune response (Medzhitov et al. 1997). Based on the discovery of additional toll genes, the toll-like family has grown to include 11 toll-like receptors (TLRs) in humans and 13 in mice (Gangloff et al. 2003). In vertebrates, TLRs recognize patterns indicative of bacteria, fungi and viruses. These patterns are collectively referred to as pathogen-associated molecular patterns. TLR4, for instance, recognizes cell wall components of gram-negative bacteria; other TLRs bind pathogenic or damage-associated molecules (Gangloff et al. 2003). These biological patterns are structurally diverse but well-conserved among pathogens, providing a molecular recognition tool to detect foreign invasion. TLRs were traditionally seen as discriminators of host ‘self’ and ‘non-self,’ but current data suggest TLRs recognize a wide array of ligands, both exogenous and endogenous molecules related not just to pathogen recognition but also to host damage and repair functions. Even within the tightly controlled blood-brain barrier, a myriad of TLR ligands have been reported. As such, the mechanism by which TLRs discern their ligands is a puzzling question. This answer lies in a fragile balance between immune signaling and neurotransmission in the CNS.

TLR4 signaling

TLR4 is well known for its response to lipopolysaccharide (LPS), an outer cell wall component of gram-negative bacteria (Shimazu et al. 1999). Both in vivo and in vitro, TLR4 expression dictates LPS responsiveness (Poltorak et al. 1998; Lehnardt et al. 2003; Hoshino et al. 1999). To confer a signal TLR4 first requires its extracellular binding partner myeloid differentiation factor 2 (MD-2) to associate before ligand-induced signaling takes place (Shimazu et al. 1999; Nagai et al. 2002). In addition to LPS and its variants, a number of surprising molecules have gained attention for their TLR4-binding properties. Exactly how TLR4 recognizes its many ligands is a longstanding question that has recently made progress as a result of breakthrough structural analyses of the TLR4–MD-2-ligand complexes (Kim et al. 2007; Ohto et al. 2007; Park et al. 2009).

Upon ligand binding, the TLR4–MD-2 complex may recruit another TLR4–MD-2 pair to form a homodimeric state (Fig. 2a). TLR4 agonists such as LPS are known to induce homodimerization of the TLR4–MD-2 complex, also referred to as aggregation (Kobayashi et al. 2006; Prohinar et al. 2007). Several TLR4 inhibitors have been reported to disrupt homodimerization in the presence of agonists (Youn et al. 2006; Wong et al. 2009). Agonists induce homodimerization of the TLR4–MD-2 complex, sending an intracellular signal through TLR4’s toll/interleukin-1 receptor (TIR) domain (Gangloff et al. 2003; Kim et al. 2007; Rittirsch et al. 2009). Exactly how the TIR domain coordinates this response is unclear; however, it is known that the heteromeric assembly of TLR4’s TIR domains constitutes the initial step of signal transduction within the cell (Ohnishi et al. 2009).

Figure 2.

 Overall structure of the TLR4/MD-2 complex. (a) Side and top view of an m-shaped receptor multimer composed of two copies of the TLR4/MD-2 complex arranged symmetrically. (b) Close-up view of the LPS binding site on the TLR4/MD-2 interface. LPS interacts with a large hydrophobic pocket in MD-2 and directly bridges the two components of the multimer. The primary interface between TLR4 and MD-2 is formed before binding LPS, and the dimerization interface is induced upon LPS binding. (c) Molecular structure of LPS.

The activation signal diverges, following either of two inflammatory cascades, the MyD88 pathway to nuclear factor kappa-B (NF-κB) activation or the toll/IL-1 receptor-containing adaptor inducing IFN-β (TRIF) pathway, also called the MyD88-independent pathway. All TLRs except TLR3 and 4 rely on the MyD88-dependant cascade (Akira et al. 2006). TLR3 signals solely through the TRIF adaptor. TLR4 is unique in that it can signal through the MyD88-dependent or TRIF-dependent cascades (Fig. 1). It is unclear what criteria TLR4 uses to determine the downstream signaling adaptor and subsequent pathway. How and when TLR4 signaling recruits MyD88 versus TRIF is the subject of much research. To coordinate the maximal inflammatory response, it has been suggested that TLR4 must signal through both pathways (Kawai and Akira 2007).

Figure 1.

 The signaling pathways of TLR4. MyD88-dependant signaling is common to most TLRs including TLRs 1,2,4 (shown) and 5,6,7,8,9,11 (not shown). TRIF-dependant signaling results from TLR4 and TLR3 activation only. Note that TLR3 resides in endosomal vesicles, shown here responding to engulfed foreign RNA. The pentagon shapes denotes proteins that interact through a TIR domain. The TRAM adaptor (lime green) is exclusive to TLR4 and coordinates the TRIF response through TLR4’s TIR domain. TRIF-dependant signaling primarily results in IFN-ß production (red adaptors), but the TRIF pathway also induces ‘late stage’ NF-κB activation through RIP 1 (white) and TRAF 6 (seafoam green). The MyD88-dependant cascade initiates ‘early stage’ NF-κB activation through the IKKs (IKKs α,β,λ) and/or the MAPK pathway, leading to proinflammatory cytokine expression and subsequent amplification through additional immune pathways.

MyD88-dependant pathway

All TLRs except TLR3 and TLR4 use the MyD88 path exclusively (Akira et al. 2006). MyD88 is a cytosolic adaptor protein with a ‘death domain’ and distal TIR domain similar to that of TLR4. The TLR4 pathway (Fig. 1) through MyD88 occurs via TIR-TIR association between TLR4 and MyD88, with the help of Mal, MyD88 adaptor-like protein (also known as TIRAP) (Horng et al. 2002). Mal is dispensable for TLR4 signaling; however, its TIR domain is useful for recruiting MyD88 to the membrane for the crucial TIR-TIR association of MyD88 with TLR4 and MyD88 with TLR2 (Kagan and Medzhitov 2006). MyD88’s signal is conferred to the interleukin-1 receptor-associated kinase (IRAK) family of protein kinases through interaction of the MyD88 and IRAK4 death domains (Kawai and Akira 2007). This triggers a phosphorylation cascade activating NF-κB transcription factors. TAK1, a crucial ubiquitin-activated complex, sends the signal via a mitogen-activated protein kinase (MAPK) cascade and/or the complex involved in NF-κB activation, the inhibitor of NF-κB (IKK) cascade. These paths induce NF-κB activation of the activator protein-1 (AP-1) and the RelA and P50 heterodimer, respectively. The AP-1 and RelA/P50 factors of NF-κB directly regulate pro-inflammatory cytokine transcription (Kawai and Akira 2007). NF-κB controls the expression of genes that regulate a broad range of biological processes in the CNS such as synaptic plasticity, neurogenesis, and differentiation (Sarnico et al. 2009). NF-κB is essential for neuron survival and its activation may protect neurons against oxidative stress or ischemic neurodegeneration (Lehnardt et al. 2003; Glezer et al. 2006; Sarnico et al. 2009). While NF-κB is associated with neuroprotective benefits, it can also contribute to inflammatory reactions and apoptotic cell death after brain injury and stroke (Caso et al. 2007, 2008; Sarnico et al. 2009). The MyD88-dependent signaling pathway is an important activator of NF-κB and the subsequent neuroregulatory effects of NF-κB signaling.

TRIF-dependant (MyD88-independent) pathway

A TRIF-dependant pathway is common to both TLR3 and TLR4 (Fig. 1). However, TLR4 signaling through TRIF requires the adaptor molecule TRAM while TLR3 does not (Rowe et al. 2006). Signaling through TRAM involves endocytosis of the TLR4 receptor complex (Tanimura et al. 2008). TRAM couples this endocytosis to the induction of IFN-β (Kagan et al. 2008). Studies suggest that TLR4 activates TRIF signaling from the endosome rather than the cell membrane (Tanimura et al. 2008; Kagan et al. 2008).

Downstream the TRIF adaptor molecule, TLR3 signalling and MyD88-independent TLR4 signaling function identically. The TRIF signal can recruit a TRAF3- or TRAF6-mediated adaptor molecule, diverging to different transcriptional effectors (Hacker et al. 2006). TRAF6 interacts with RIP to induce NF-κB activation through TAK1. TAK1 behaves the same as in the MyD88-dependent cascade, activating the NF-κBs ReIA/P50 and AP-1. The TRIF-dependant activation of NF-κB is aptly named ‘late phase’ NF-κB activation, while the faster TLR4 route through MyD88 is the ‘early phase’ NF-κB. TLR4 shares the ‘early phase’ NF-κB pathway with all the TLRs except TLR3, which can only affect NF-κB through the RIP 1/TRAF6 ‘late phase’ activation mechanism. The coordination of both ‘early’ and ‘late’ signaling is a capability unique to TLR4.

The TRIF-dependant signal through TRAF3 (as opposed to TRAF6) starts a TRIF-binding kinase IKK cascade terminating in interferon regulatory factor 3 (IRF3) dimerization and translocation into the nucleus (Poikonen et al. 2009). IRF3 induces IFN-β synthesis, which regulates the cellular response to inflammation. IFN-β is both anti-inflammatory and anti-apoptotic, providing an endogenous mechanism to keep the innate immune system in check. Excretion of IFN-β coordinates the production of additional type-1 interferons, further suppressing the immune response.

TLR4 activation ultimately induces the secretion of proinflammatory substances such as reactive oxide species (nitrous oxide, hydrogen peroxide, and superoxides), cytokines such as tumor necrosis factor-α (TNFα) and interleukins (Bowie and O’Neill 2000; Maier and Watkins 2003; Tsan and Gao 2004a; Blanco et al. 2005; Kagan et al. 2008). In contrast, TLR4 also affects IFN-β release, which counteracts inflammation (Kagan et al. 2008). Proinflammatory factors coordinate immune defense, repair and debris removal, but these factors can amplify out of control without regulation by anti-inflammatory substances. Neurons and oligodendroglia are especially fragile under inflammatory conditions (Lehnardt et al. 2003). Neurological stress provokes NF-κB induced release of reactive oxygen species, which in turn cause neuronal vulnerability (Keller et al. 1999; Hua et al. 2007). A comprehensive review of microglia-mediated inflammation and chronic excitation highlights TLR4’s contribution to neurotoxicity (Block et al. 2007).

The pro-inflammatory response is generally amplified by TLR signaling. Pro-inflammatory cytokines coordinate other immune cells, attracting them to the site of invasion or damage, amplifying it until the insult is eliminated or dampened by immune-suppressing feedback mechanisms. TRIF-mediated IFN-β release can counteract inflammation, whereas the MyD88-dependant path is pro-inflammatory. But when and why the IFN-β anti-inflammatory pathway is induced is not well understood. The criteria for TRIF versus MyD88 signaling are unknown, but common TLR4 ligands appear to utilize the same pathway or both pathways consistently.

TLR4 in the CNS

TLR4 expression and activation

The CNS was once thought to be an immune-privileged site, but researchers now recognize the role of immunity in the CNS. Microglia are the resident immune cells of the CNS, comprising about 12% of the cells in the brain and spinal cord (Lawson et al. 1990). It makes sense that innate immune receptors such as TLR4 would be expressed on the immune cells of the nervous system primarily. TLR4 is primarily expressed on glia, primarily microglia (Lehnardt et al. 2003). However, TLR4 expression has been reported on other CNS cells including astrocytes, endothelial cells and neurons (Jou et al. 2006; Tang et al. 2007). It has been disputed whether TLR4 is expressed on neurons under normal conditions (Lehnardt et al. 2003), but current studies leave little question as to whether TLR4 can be expressed on neurons in pathological environments (Tang et al. 2007, 2008).

The TLRs can regulate cellular development in addition to their well-known immunological tasks. TLR4’s evolutionary precursor Toll was originally discovered for its role directing Drosophila development (Anderson et al. 1985b). In the vertebrate nervous system, microglial cells regulate neuronal development, differentiation and survival using immune mechanisms to elicit apoptosis or proliferation. Microglia enforce the programmed elimination of neurons throughout development (Marin-Teva et al. 2004; Wakselman et al. 2008) and they are necessary to elicit differentiation and migration of neural precursors (Aarum et al. 2003).

Microglia can also damage neighboring cells through chronic overstimulation or prolonged inflammatory responses. Microglial inflammation can be erroneous, amplified and progressive (Block et al. 2007). Overstimulated microglia cause oxidative stress and damage to other CNS cells, most notably neurons (Block et al. 2007). Because microglial activation is widely controlled by pathogen-recognition receptors (PRRs), TLR4 is implicated in the microglia-mediated neurotoxicity that occurs in many brain pathologies. Although some reports suggest neuronal TLR4 is directly responsible for neuron death (Tang et al. 2007, 2008), the majority of studies investigate microglial expression of TLR4 and the subsequent neurotoxic effects of TLR4 signaling. Whether or not the TLR4 is activated on neurons or on microglia, it is widely accepted that excreted products of TLR4 signaling alter neuronal functions. It is clear that TLR4’s detection system remains intact within the blood–brain barrier (Zhou et al. 2006); why this system can be both helpful and harmful is the subject of much research.

Sepsis: the classical TLR4-mediated syndrome

TLR4 is well known for its detrimental role in sepsis and endotoxemia, where LPS-induced TLR4 activation contributes to the systemic inflammation that characterizes these serious conditions. TLR4-mutant mice are resistant to LPS-induced inflammation and associated sepsis syndromes (Hoshino et al. 1999). However, without an intact host defense system, LPS-hyporesponsive mice will die from gram-negative bacterial invasion (Poltorak et al. 1998). The innate immune response to LPS is similarly important for host defense by humans (Hoshino et al. 1999; Arbour et al. 2000). Inflammatory amplification in the case of sepsis demonstrates the power of innate immunity to coordinate a disproportionate reaction to its trigger, initiating an inflammatory response so strong that viable cells are damaged. Proinflammatory cytokine responses signal the brain through neuronal and blood-borne routes, altering neural activity and proliferating the systemic immune response (Maier and Watkins 2003). If the TLR4 pathway is erroneously activated, or if a signal is amplified out of control, the cytokine response may have deleterious effects on the nervous system. TLR4-induced inflammatory signaling has the ability to instruct both necrosis and apoptosis in various CNS cell types. But TLR4 signaling can be beneficial, too. TLR4 has shown critical neuroprotective benefits in studies of stroke (Marsh et al. 2009), and amyloid-β clearance is diminished in TLR4-deficient mouse models of Alzheimer’s disease (AD) (Tahara et al. 2006). Altogether, innate immunity and the responses coordinated by PRRs are extremely powerful modulators of the CNS environment.

Sepsis and its associated inflammatory syndromes influence the nervous system through TLR4. In mouse models of bacterial sepsis, LPS administered peripherally induces a chronic proinflammatory response within the CNS. This inflammation requires the expression of TLR4 in the CNS and is independent of systemic cytokine levels (Chakravarty and Herkenham 2005). Furthermore, LPS-induced mouse models of sepsis experience progressive neurodegenerative effects analogous to Alzheimer’s or Parkinson disease. Mice lacking functional TLR4 expression in CNS are exempt from long-term progressive neuron loss. This example illustrates the paradoxical nature of TLR4 signaling: it is necessary for defense, yet it invokes a powerful cascade that can be toxic. The stakes are high in the CNS, where subtle modifications can tip TLR4 signaling over the neurotoxic edge.

Pathological implications of TLR4 signaling

Lehnardt et al. were the first to definitively illustrate the neurotoxic effects of TLR4 signaling. In mixed CNS cultures, LPS-induced neurodegeneration is microglia-dependant, manifesting in neuronal axon and dendrite loss. Similar cultures obtained from TLR4-deficient mice were resistant to neuronal insult from LPS, establishing TLR4 as a requirement for LPS-induced toxicity (Lehnardt et al. 2003). Oligodendroglia also exhibit damage upon LPS administration and subsequent TLR4 activation, but neurotoxicity prevails as the primary detriment to TLR4 activation. TLR4-induced neuron death occurs independent of organism species and on all neuronal subtypes (Lehnardt et al. 2003). Recent reports of TLR4 activation by endogenous ligands link TLR4 to autoimmunity as well as legitimate inflammation (Midwood et al. 2009). As the interactions of the immune and nervous systems gain attention, more and more ligands are reported to bind PRRs. The myriad of structurally diverse TLR4 ligands demonstrates the puzzling diversity and ambiguity of ‘pathogen-associated molecular patterns’. Based on the promiscuous binding activity of TLR4, it is reasonable to implicate TLR4 signaling in several etiologies of the nervous system.

Evolution has produced pattern-like danger signals and highly conserved TLRs to recognize such pattern-associated pathogens and damage signals. The response coordinated by TLR4 is necessary to protect the CNS from foreign invasion. Microglial TLR4 activation also contributes to repair processes, improving remyelination and conferring cerebral tissue protection in the presence of neurotoxic compounds (Glezer et al. 2006). But at what point do the risks of aberrant TLR4 signaling outweigh the benefits conferred from damage repair and pathogen protection? We will investigate this question with respect to relevant diseases of the nervous system.

Because TLRs recognizes pathogenic and damage-associated molecules, TLR4 is intrinsically implied in pathologies of the nervous system. The neuroinflammatory origins of dementia were studied as early as 1889 by Julius Wagner-Jauregg, who earned the 1927 Nobel Prize in Physiology or Medicine for his neuroinflammatory approach to dementia paralytica. He discovered that infection with the malaria parasite mitigated the psychiatric and paralytic impairment associated with long-term syphilis infection (Nobel Lectures, 1965). But the work of Wagner-Jauregg diminished in the following years, most likely because a neurology-immunology link was intangible in the eyes of early twentieth century physicians.

We now know that immunity remains intact within the nervous system, and immune cells are well represented by microglia in the brain. The peripheral lymphatic system itself is innervated and immune-to-brain communication is now well documented (Watkins and Maier 1999). This is underlined by the finding that neurotransmitters, cytokines and their respective receptors are both endogenous to the brain and immune system. The above discoveries substantiate Wagner-Jauregg’s prodigal connection between the brain and immunity. Nevertheless, only in the past 20 years have the interactions between the immune and nervous systems become the subject of intense research. Investigations into microglial signaling and TLR biology have rapidly expanded and eventually the fields have intersected, bringing TLRs into focus for important CNS diseases. Capable of both protective and pathological roles, TLR4 can be helpful or harmful under varying neurological conditions. When and why TLR4 initiates beneficial outcome is still largely unknown, but progress in the field suggests that this question will receive much attention, as TLR4 is a useful and druggable target.

Neurodegenerative conditions and TLR4: Alzheimer’s disease

Alzheimer’s is a progressive neurodegenerative disease marked by neuron loss, aggregation of amyloid beta peptide (Aβ) into plaques, and microglial activation and recruitment. Research suggests neuroinflammation is a major contributor to Alzheimer’s pathology, as Aβ plaques are closely associated with brain inflammation (Akiyama et al. 2000). Accordingly, microglia and astrocytes concentrate in and around Aβ plaques. An increase in complement components, pro-inflammatory factors and proteases suggests that the innate immune response is a crucial contributor to plaque-induced neurotoxicity (Akiyama et al. 2000). As such, TLR4 has been suggested as a mediator of AD and other neurodegenerative conditions (Keller et al. 1997; Hua et al. 2007; Tang et al. 2008; Marta et al. 2009; Zhao et al. 2010).

Aβ plaque deposits are the pathological hallmark of AD. Despite decades of research, the pathways through which neuritic plaques elicit neurodegeneration are poorly understood. An inflammatory mechanism appears to be responsible for local microglial activation and the subsequent pro-inflammatory sensitization and degradation of neurons in AD (Akiyama et al. 2000). The inflammatory nature of Alzheimer-related neurotoxicity is reinforced by data showing reduced risk of AD in patients taking acetaminophen, the anti-inflammatory agent (Stewart et al. 1997).

Paradoxically, TLR4 expression is also associated with increased uptake of Aβ peptide (Tahara et al. 2006). Under normal conditions, Aβ is removed before it accumulates as extracellular amyloid fibrils, suggesting that Aβ uptake by TLR4 is a beneficial mechanism (Akiyama et al. 2000). Inflammatory markers such as heat-shock proteins (HSPs) are also associated with increased uptake and clearance of Aβ (Kakimura et al. 2002) and several HSPs have been reported to activate TLR4 signaling (Triantafilou and Triantafilou 2004; Lehnardt et al. 2008; Hutchinson et al. 2009a). These inflammatory mechanisms may be necessary for normal Aβ clearance (Tahara et al. 2006), but HSPs may also induce neurodegeneration through TLR4 signaling (Kakimura et al. 2002). It is unclear whether TLR4 favors Aβ uptake over neurotoxic inflammation, or if Aβ clearance and inflammatory reactions take place simultaneously or interdependently of TLR4 signaling. This important question stands to be answered, but the quantity of data favors TLR4’s harmful effects over its benefits. Many studies investigate microglial inflammation and TLR4 signaling in neurodegenerative pathologies such as AD.

Several genetic mutations are known to induce amyloid deposition and subsequent AD symptoms, but the majority of AD cases are sporadic and genetically heterogeneous (Akiyama et al. 2000). Minoretti and coworkers investigated the contribution of TLR4 mutations to AD pathology, screening 277 Italian late-onset AD patients and 300 healthy patients for TLR4 polymorphisms. A TLR4 decreased function polymorphism was found to protect against AD (Minoretti et al. 2006). This common TLR4 polymorphism (Asp299Gly) was disproportionately represented in the control cohort, suggesting the mutation protects against late onset AD (Minoretti et al. 2006). The Asp299Gly mutation stunts TLR4 signaling and the associated inflammatory responses. Taken together, this study provides strong evidence that TLR4 signaling negatively contributes to late-stage AD onset (Minoretti et al. 2006). Further studies are needed to determine whether TLR4 signaling is consistently detrimental to AD onset. The benefits of TLR4-related Aβ uptake (Tahara et al. 2006) must be weighed against the neurotoxic effects of TLR4 signaling (Minoretti et al. 2006; Walter et al. 2007).

Another genetic study suggests aberrant TLR signaling contributes to AD neuroinflammation. Tan and coworkers constructed a Drosophila model of AD to express Aβ-42 in the fruit fly CNS. Toll gene activity through NF-κB signaling was identified as the key mediator of neurotoxic inflammation (Tan et al. 2008). The Toll gene is an evolutionary precursor to the toll-like receptors (e.g. TLR4, TLR6) found in vertebrates. In the fruit fly model Alzheimer’s, Toll activity was correlated with shortened lifespan (Tan et al. 2008). Inhibition of Toll signaling lengthened lifespan in the same model. While Drosophila Toll bears similarity to TLR4 and other vertebrate TLRs, it ought to be noted that the Drosophila model simplifies a complex human disease. Nevertheless, the innate immune response is conserved through evolution and furthermore, induction of NF-κB is common to both TLR4 and Drosophila Toll. This study adds to a body of evidence suggesting toll-like receptors contribute to Alzheimer’s neuroinflammation and subsequent neurodegeneration.

TLR4 expression has been reported on both neurons and glia under normal conditions (Tang et al. 2007, 2008), although TLR4 is probably expressed more appreciably on microglia than other CNS cell types (Lehnardt et al. 2003). Tang and colleagues specifically investigated neuronal TLR4 expression and its effects in CNS pathologies (Tang et al. 2007, 2008). Murine neurons were found to increase TLR4 expression when exposed to Aβ peptide or the membrane peroxidation product, 4-hydroxynonenal (Tang et al. 2008). Aβ and 4-hydroxynonenal were found to trigger neuronal apoptosis in wild-type murine cells. Neurons from TLR4-mutant mice, however, displayed resistance to death under the same conditions (Tang et al. 2008). A JNK inhibitor was shown to protect against neurotoxicity in the presence of Aβ or 4-hydroxynonenal, linking neuronal apoptosis to TLR4-induced JNK signaling. Finally, tissue specimens taken from end-stage AD patients showed slightly decreased TLR4 levels compared with aged-matched control subjects (Tang et al. 2008). The authors suggest that this finding results from the explicit loss of TLR4-expressing neurons due to TLR4-mediated neurotoxicity. These studies suggest that neuronal TLR4 expression may pre-dispose neurons to apoptosis in the presence of Aβ and/or 4-hydroxynonenal.

Walter and colleagues further elucidated the role of TLR4 in Alzheimer-related neuroinflammation, focusing on glial-mediated effects upon neurons. Their results indicate that aggregated Aβ induces inflammation through TLR4 on both microglia and macrophages. Microglia from wild-type mice had significantly increased levels of IL-6, TNFα and nitric oxide when compared with TLR4 loss-of-function mutants (Walter et al. 2007). In human embryonic kidney cells (HEK293), it was shown that TLR4’s accessory protein, MD-2, and the coreceptor CD-14 were also required to coordinate a response to aggregated Aβ. Furthermore, Aβ peptide was only recognized in its aggregated conformation; neither scrambled peptide nor non-aggregated peptide elicited an IL-8 response from the HEK293 cells. Finally, neurotoxicity was assessed using the supernatants of microglia exposed to aggregated Aβ peptide. The supernatants were added to primary murine neuronal cells and neurotoxicity assayed. Only 20% of neurons treated with wild-type Aβ-exposed microglial supernatant survived. Neurons incubated with supernatant from TLR4 mutants were much more likely to survive (70% viable) (Walter et al. 2007). TLR4 was also assessed in experimental AD, where amyloid precursor protein-over-expressing mice possessed significantly elevated TLR4 mRNA when compared to their age-matched non-transgenic littermates. Finally, TLR4 expression was markedly increased in post-mortem brains of AD patients (Walter et al. 2007). This finding stands in contrast to the Tang study, where TLR4 was slightly underexpressed in post-mortem AD brains (Tang et al. 2008). The discrepancy could result from experimental differences in methodology, or simply from the small cohort size in both studies. Nevertheless, the data gathered by both Tang and Walters, and concurrent data (Minoretti et al. 2006; Tan et al. 2008) strongly suggest a function for TLR4 in Aβ-induced neurotoxicity.

In looking to the future we must be careful to remember the role TLR4 may play in clearance and uptake of Aβ (Kakimura et al. 2002; Tahara et al. 2006). Although an increasing volume of data favors TLR4-mediated neurotoxicity, TLR4 may also be essential to the uptake and phagocytic removal of Aβ plaques. In addition, it is unclear whether TLR4-mediated neurotoxic effects result from the neurons themselves (Tang et al. 2008), or from microglial signals (Walter et al. 2007). Until researchers can reconcile if and when TLR4 drives neurotoxicity over Aβ clearance, TLR4-targetting drugs will have limited clinical utility in Alzheimer’s treatment.

Ischemic stroke

It is increasingly clear that post-stroke neuroinflammation from TLR4 signaling worsens stroke outcome, as measured by infarct volumes, neurological function and inflammatory markers (Tasaki et al. 1997; Caso et al. 2007; Abate et al. 2009). Several models of cerebral ischemia have suggested a role for TLR4 in neuroinflammation and exacerbated stroke injury. Mice deficient in TLR4 have shown improved neurological and/or behavioral outcomes in various models of cerebral infarction (Cao et al. 2007; Hua et al. 2007).

TLR4 is known to confer immunological tolerance, giving a dampened response upon a second insult by immunogenic stimuli (usually LPS). Similar studies have demonstrated that pre-conditioning with LPS, the classical TLR4 ligand, protects against the cytotoxic damage elicited from ischemic stroke (Tasaki et al. 1997; Rosenzweig et al. 2004; Hickey et al. 2007). Initial LPS exposure signals through TLR4 to affect a cytotoxic TNFα response, but subsequent LPS-induced TLR4 activation may be dominated by IFN-β production (Hickey et al. 2007; Marsh et al. 2009). IFN-β is anti-inflammatory and anti-apoptotic, and systemic administration of IFN-β reduced infarct volume in mouse and rabbit models of cerebral ischemia (Liu et al. 2002a; Veldhuis et al. 2003). The shift from pro-inflammatory TNFα to the anti-inflammatory IFN-β production suggests TLR4 may switch predominant signaling pathways from MyD88 to TRIF/TRAM, responding more mildly to a second LPS exposure.

Marsh et al. also demonstrated a change of pathways after LPS pre-conditioning. They measured IFN-β neuroprotection and its associated genetic expression. Based on evidence that IFN-β reduces ischemic brain damage (Liu et al. 2002a; Veldhuis et al. 2003), Marsh tested the underlying signaling mechanism by which IFN-β confers its neuroprotective benefits. RNA analysis of post-stroke genetic expression reveals up-regulation of IFN-β and its transcriptional regulators in animals pre-conditioned with LPS (Marsh et al. 2009). Ischemic damage in LPS-pretreated animals is minimal compared with those lacking prior insult by LPS. This concurs with the finding that TNFα signaling is favored upon LPS pre-treatment, but pre-treated animals responded to infarction with IFN-β secretion which lends neuroprotective benefits. Control animals signal through the proinflammatory TNFα pathway, leading to worsened stroke outcome in comparison to animals conditioned with LPS prior to infarction. IFN-β does not play a role in the endogenous response of the brain to ischemia (Marsh et al. 2009). Pre-treatment with LPS must change the cellular environment in such a way that subsequent TLR4 activation induces IFN-β neuroprotection, a beneficial change in response. LPS pre-conditioning appears to reprogram TLR4 signaling from MyD88-mediated proinflammatory factors to the TRIF/TRAM pathway leading to IFN-β and subsequent type 1 interferon secretion. TLR4-mediated IFN-β expression, if favored over TNFα proinflammatory signaling, has powerful neuroprotective benefits which could be exploited to minimize post-stroke neurodegeneration.

Heat-shock proteins released from damaged cells are another trigger of TLR4-induced neuroinflammation (Triantafilou and Triantafilou 2004; Lehnardt et al. 2008). Necrotic neuron death produces an immune response characterized by activation of the MyD88-mediated pathway (Pais et al. 2008). Similar to the response from primary LPS-activation of TLR4, HSP60 causes the proinflammatory activation of microglial TLR4, leading to nitrous oxide production and subsequent neurodegeneration in vitro and in vivo (Lehnardt et al. 2008). A vicious cycle ensues as neurons produce HSP60 in response to stress, in turn activating more and more TLR4 and its neurotoxic effectors.

The possibility of TLR4 changing from MyD88 to the TRIF-mediated pathway was not demonstrated in the above HSP60 study. However, uncontrolled proliferation of a toxic signal has no evolutionary support on which to stand. Lehnardt asserts that the ancient defense system of TLR recognition has evolved to recognize endogenous damage signals in addition to exogenous dangers. This conclusion is supported by the variety of endogenous ligands reported to bind TLR4 (Table 1). The role of TLR recognition is also complicated by its developmental functions in lower phylogenetic species, as is in the Toll gene product which determines neuronal differentiation patterns in Drosophila embryos (Anderson et al. 1985b). Clearly, more research is needed to determine when and why TLR4 affects infarctive stroke and other neuroinflammatory states.

Table 1.   A variety of ligands are suggested to affect TLR4. Endogenous ligands are denoted with an asterisk. It is important to note that endotoxin is a very common and potent contaminant in such studies. Especially where recombinant proteins are reported to activate TLR4, contamination is difficult to exclude
Putative TLR4 interactorExplanationReport
Lipopolysaccharide (LPS) and LPS derivatives (see Fig. 2 for molecular structure) Outer cell wall component of gram-negative bacteria; potent initiator of TLR4 signaling. LPS structure varies with bacterial speciesStructure–activity relationship of LPS and TLR4 (Park et al. 2009), of LPS: (Rietschel  et al. 1994)
CurcuminPolyphenol found in the plant Curcuma longa. Inhibits TLR4 by binding MD-2(Youn et al. 2006)
(Gradisar et al. 2007)
Cinnamaldehyde (3-phenyl-2-propenal)Anti-inflammatory, inhibits ligand-induced TLR4 oligomerization and downstream signaling(Youn et al. 2008)
EthanolAppears to redistribute TLR4 complexes on the cellular membrane by preventing receptor association and/or dimerization in the lipid raft(Blanco et al. 2005, 2008;  Szabo et al. 2007; Fernandez-Lizarbe et al. 2008)
E5564 (eritoran)LPS analogue clinically tested for sepsis; inhibits TLR4 signaling(Rossignol et al. 2004; Yamada et al. 2005; Kim et al. 2007)
OpioidsBoth opioid stereoisomers alter downstream TLR4 signaling. Opioid agonists (e.g. morphine) have different effects than antagonists (e.g. naloxone)(Liu et al. 2000; Hutchinson et al. 2007, 2009b; Juni et al. 2007)
TAK-242 (Ethyl(6R)-6-[N-(2-chloro- 4-fluorophenyl) sulfamoyl] cyclohex-1-ene-1-carboxylate)Clinically tested cyclohexene derivative, selectively inhibits intracellular signaling by TLR4(Ii et al. 2006; Sha et al. 2007; Takashima et al. 2009)
Paclitaxel (Taxol)Widely used cancer therapeutic, reported to inhibit MD-2, thereby knocking down TLR4 activity which was found to correlate with drug efficacy(Wang et al. 2009)
Resveratrol (trans-3,5,4-trihydroxystilbene)Antioxidant reported to inhibit TLR4 signaling; found in the skin of grapes, it is known for anti-inflammatory and anti-carcinogenic effects(Youn et al. 2005; Yusuf et al. 2009)
StatinsStatin drugs influence TLR4-mediated cytokine expression through a Rho-protein feedback mechanism(Konat et al. 2008)
Amyloid-β 42 peptideThe peptide hallmark of Alzheimer’s pathogenesis, appears to activate TLR4 directly and also through signals from damaged neurons (e.g. 4-hydroxynonenal)(Liu et al. 2002b; Tang et al. 2008; Balistreri et al. 2007, 2009)
Extracellular matrix proteins
 • Biglycan
• Fibrinogen
• Fibronectin
• Tenascin C
Negatively charged glycoproteins are reported to activate TLR4 signaling(Schaefer et al. 2005)
(Smiley et al. 2001)
(Okamura et al. 2001)
(Midwood et al. 2009)
Fatty acids*Fatty acids are reported to regulate TLR4 receptor dimerization and recruitment into lipid rafts (Weatherill et al. 2005; Wong et al. 2009)
Heat-shock proteins (HSP) 60, 70, 90Released from dead or dying cells. HSP60 mediates neurodegeneration via TLR4 (Lehnardt et al. 2008). HSP90 may influence TLR4 pain amplification (Hutchinson et al. 2009b). LPS contamination is a common problem in HSP studies HSP60 (Lehnardt et al. 2008); HSPs 70, 90 (Triantafilou and Triantafilou 2004; Hutchinson et al. 2009b). Contamination, (Tsan and Gao 2004b)
PolysaccharidesHeparin sulfate and endogenous hyaluronic acid fragmentation products may activate dendritic cells and macrophages through TLR4(Termeer et al. 2002)

Chronic pain pathologies

Researchers have found that pathological pain invokes an inflammatory response within the CNS. In cases where pathological pain persists beyond the resolution of tissue damage, it constitutes neuropathic pain. Pain management, especially pathological or neuropathic pain, is a significant public health issue. The National Institutes of Health reported that pain costs the USA more than $100 billion per year in medical expenses, lost wages and lost productivity (Department of Health and Human Services 1998). Pain research traditionally focused on neuronal mechanisms, but this work has yielded marginally effective therapeutics. New pain research shifts focus to also investigate glial mechanisms. For a complete review on the roles of glia in chronic pain, see Milligan and Watkins (2009).

Microglia, the immunocompetent cells of the CNS, are important contributors to chronic pain pathologies (Milligan and Watkins 2009). The TLR4 receptor is one avenue through which microglia can be activated and primed for the pain response. TLR4 influences the CNS pain response, invoking the production of pro-inflammatory cytokines and reactive oxygen species (DeLeo et al. 2000; Tanga et al. 2005; Hutchinson et al. 2007). Recent studies link TLR4 to pain etiology in animal models and corresponding in vitro and in silico systems.

DeLeo and coworkers first explored this link based on the involvement of TNF in painful neuropathy and other CNS disorders (DeLeo et al. 2000). Because TNF and other inflammatory-mediating cytokines are regulated by TLR4, the group hypothesized that microglial TLR4 influences hypersensitivity in models of neuropathy (Tanga et al. 2005). Using a standard L5 nerve transection procedure to induce chronic pain, they tested hypersensitivity in genetically altered (TLR4 knockout and point-mutant) mice and TLR4 antisense oligodeoxynucleotide treated rats. Both the mice and rats displayed attenuated behavioral hypersensitivity and decreased expression of proinflammatory cytokines relative to respective controls. This work established a role for TLR4 in mice and rat models of neuropathic pain (Milligan and Watkins 2009).

A similar role for innate immunity in pain amplification is suggested by a recent clinical trial that found naltrexone useful in the treatment of fibromyalgia (Younger and Mackey 2009). Fibromyalgia is a common condition characterized by diffuse chronic pain. Concurrent symptoms such as fatigue, sleep disturbance and cognitive impairment have led to the characterization of fibromyalgia as a CNS sensitivity disorder, but the molecular etiology of the condition is unknown. Naltrexone is a competitive antagonist of opioid receptors, and has been used for the clinical treatment of opioid dependence and reversal of opioid overdose. More recently, naltrexone has been used to suppress microglial signaling, thereby decreasing the production of proinflammatory cytokines and reactive oxygen species (Liu et al. 2000, 2002b). This mechanism is likely responsible for decreased fibromyalgia symptoms in response to low doses of naltrexone, as measured by improved pain threshold testing and self-reported symptoms in a single-blind study (Younger and Mackey 2009).

Naltrexone may elicit its microglial effects through a mechanism distinct from classical opioid receptors, as there is no evidence for abnormal endogenous opioid activity in fibromyalgia patients (Younger et al. 2009). In a follow-up to the above clinical trial, Younger and coworkers assessed fibromyalgia patients for opioid withdrawal symptoms upon opioid antagonism with 50 mg naltrexone. No withdrawal symptoms were reported, indicating that endogenous opioid activity is not dysregulated in fibromyalgia pathophysiology (Younger et al. 2009). Fibromyalgia probably elicits hypersensitivity through a pathway distinct from the mu-opioid system, but that responds to low doses of naltrexone.

Opioid receptor knockout studies suggest that an atypical receptor is responsible for the mechanism which mediates hypersensitivity upon morphine administration, as these effects occur independent of traditional opioid receptor expression (Juni et al. 2007). Evidence has accrued suggesting that opioids elicit their secondary effects through an atypical pathway characterized by inflammatory signaling from glial activation (Hutchinson et al. 2007). LPS induces glial activation, and this activation can be ameliorated by naloxone administration (Wu et al. 2006). Based on the ability of naloxone to interfere with LPS signaling, the TLR4 pathway is implicated as a mediator of non-classical opioid responses (Liu et al. 2000; Hutchinson et al. 2009b).

We explored the involvement of TLR4 in opioid-induced microglial signaling due to opioid antagonists (i.e. naltrexone and naloxone) and opioid agonists (morphine, oxycodone, methadone etc.). In vitro TLR4 signaling was observed in response to clinically relevant opioid agonists (Hutchinson et al. 2009b). Consistent with the ability of naltrexone to reduce chronic pain symptoms, opioid antagonists were found to inhibit TLR4 signaling and the production of pro-inflammatory substances. TLR4 inhibition was proportional to morphine analgesia effectiveness. This increase in analgesia (pain relief) coincided with decreased morphine tolerance, and lessened hyperalgesia and opioid withdrawal behaviors (Hutchinson et al. 2009b). Our data suggest opioid agonists and antagonists can affect downstream TLR4 signaling by respectively activating or inhibiting TLR4-mediated proinflammatory release.

The specificity of the TLR4–opioid interaction was addressed using TLR4 knockout mice to observe opioid response (Hutchinson et al. 2009b). Based on the negative side effects associated with TLR4 activation, it was hypothesized that TLR4 knockout mice would respond to lower doses of opioid agonist and display decreased development of tolerance, dependence and hyperalgesic behavior. To date we have observed potentiation of morphine analgesia in TLR4 knockout mice (Hutchinson et al. 2009b). TLR4 knockouts react to morphine with threefold higher analgesia, and acute inhibition of TLR4 signaling can elicit the same response to a lesser degree (Hutchinson et al. 2009b; Slivka et al. 2009). Together this work strongly supports the role of TLR4 in opioid-induced glial dysregulation, leading to pain amplification and the likely development of tolerance.

Intriguingly, while classical opioid receptors respond only to the (−)-opioid stereoisomer, several studies report that glial cells respond to both opioid stereoisomers [e.g. (+)-morphine and (−)-morphine] (Wu et al. 2006; Hutchinson et al. 2008). The ability of opioid antagonists to inhibit TLR4 can therefore be exploited by administration of the unnatural (+)-opioid antagonist, as the (+)-opioid stereoisomers are inactive at classical opioid receptors (Wu et al. 2006; Hutchinson et al. 2009b). This strategy was successfully implemented to increase morphine potency by administration of (+)-naloxone, the stereoisomer of naltrexone that was successfully used to treat fibromyalgia in preliminary studies (Hutchinson et al. 2008; Younger and Mackey 2009). The effectiveness of (−)-naltrexone in the treatment of chronic pain suggests a similar TLR4-mediated mechanism is at work in fibromyalgic pain amplification.

The ability of LPS to induce severe pain and the ability of LPS inhibitors to relieve these painful effects further link TLR4 to pain amplification (Maier et al. 1993; Mason 1993). But opioid-mediated hyperalgesia and tolerance may not function in the same way as LPS induction of hyperalgesia. Binding of opioids to TLR4 has not been directly observed, and so opioid-induced hyperalgesia and tolerance does not necessarily occur in the same manner as LPS, despite their phenotypic similarities. While TLR4 is necessary to induce neuropathic pain, TLR4 alone cannot coordinate a pain response to low-doses of LPS (Hutchinson et al. 2009a). These results are corroborated by an earlier finding that CD40 also perpetuates neuropathic pain in rat models of nerve injury (Cao et al. 2009). Interestingly, CD40 signaling was associated with production of spinal cord proinflammatory cytokines IL-1β, IL-6, IL-12 and TNFα, but not IL-6, which is up-regulated upon induction of neuropathic pain in both CD40 knockouts and wildtype mice. IL-6, TNFα and IL-1β are inducible by TLR4, suggesting a concomitant activation of CD40 and TLR4 may dually account for neuropathic inflammation.

Recent evidence also suggests a secondary mediator is at work in TLR4-mediated pain enhancement, in addition to the factors involved in the TLR4 complex and signaling pathway. HSP90 was found to be critical for LPS-induced pain and sensitivity (Hutchinson et al. 2009a). This work indicates a mediator may be responsible for the antagonistic proinflammatory effects of opioids which occur, at least in part, through TLR4 (Hutchinson et al. 2009b). Clearly, more work is needed to determine the role of TLR4 in opioid-induced glial activation. Nevertheless, these studies define a relationship between activation of the TLR4 pathway and chronic pain pathologies. Interestingly, our later discussion of TLR4 ligands reveals many anti-inflammatory agents act to inhibit TLR4.

Cancer

Resveratrol, a chemopreventive agent, has shown affinity for TLR4 and downstream adaptor molecules of the MyD88-independent pathway (Youn et al. 2005, Yusuf et al. 2009). The widely used chemotherapeutic Paclitaxel is also reported to elicit TLR4 effects through MD-2 (Zimmer et al. 2008; Wang et al. 2009). These TLR4 compound interactions have implications for understanding aspects of cancer treatment and pain. However, such examples of beneficial TLR4 regulation must be tempered with the negative contribution of TLR4 to cancer-related inflammation.

Researchers have long worked to determine the relationship between chronic inflammation and cancer. Reactive oxygen species such as those produced upon TLR4 activation may be responsible, at least in part, for cancerous proliferation as a result of inflammation. Especially when cells are dividing rapidly, reactive oxygen species can promote carcinogenic genetic mutations (Marx 2004a). NF-κB is involved in important carcinogenic responses; it promotes cancer by inhibiting apoptosis and by encouraging cancerous cells to spread and proliferate in other parts of the body (Marx 2004a,b). Innate immune activation has profound influences on tumor growth, as evidenced by experiments reporting significant increase in tumor size upon LPS injection (Marx 2004a). Although NF-κB is activated by TLR4, it is also influenced by a myriad of other signaling cascades. Why inflammation sometimes promotes cancer development and at other times keeps cancer in check remains unclear.

TLR4 and MD-2 structure and ligand activity

Three studies have been instrumental in establishing the structure-based function of TLR4 and its accessory protein MD-2. The first of these determined two crystal structures of human MD-2, one with and one without its ligand lipid IVa, an LPS derivative (Ohto et al. 2007). MD-2 takes a clamshell shape with a deep hydrophobic cleft flanked by two β-sheets. The structure with lipid IVa shows it sandwiched deep within the hydrophobic cleft of MD-2 (Ohto et al. 2007). Other ligands have since been shown to fit similarly inside MD-2 when complexed with TLR4 (Kim et al. 2007; Park et al. 2009).

The first high-resolution crystal structure of TLR4 itself showed murine TLR4 with its accessory protein MD-2 (Kim et al. 2007). Most recently, Park and colleagues established the basis for ligand recognition by the human TLR4 complex, showing that TLR4 agonist ligands cause two dimers of TLR4–MD-2 to associate, forming a multimeric complex (Fig. 2). These multimers are associated with active TLR4 signaling, while a single unit of TLR4–MD-2 does not necessarily elicit a signal (Kim et al. 2007; Park et al. 2009).

The methods by which TLR4 recognizes its ligands have been studied through LPS structure–activity and TLR4 homodimerization structures (Rietschel et al. 1994; Park et al. 2009). There are many ligands with alleged TLR4 effects; however, LPS contamination may be to blame in some of these cases. TLR4’s binding activity is perplexing nevertheless, as no obvious pattern describes the structures of TLR4 ligands. Further characterization of TLR4 ligands will be necessary to paint a more complete picture of ‘pattern’ recognition receptors. Biophysical assays and direct evaluation of ligand binding, in particular, will help narrow the list of reported TLR4 ligands. This may in turn lead to a more concise repertoire of TLR-affecting agents, and clearer themes of TLR recognition. These themes will be crucial if we are to create more effective TLR-targeting therapeutic agents in the future.

Perspectives in drug discovery

Because of the pathological implications of aberrant TLR signaling, the ability to control TLR4 recognition and activation is a therapeutic topic of much interest. Given the ineffective therapeutics for Alzheimer’s, neuropathic pain and other central nervous system pathologies, drug candidates for these diseases are in urgent need. We believe that the traditional emphasis on neuronal targets is, at least in part, to blame for the void in effective pharmacotherapies to treat these conditions. Microglia are a dynamic, promising target in the aforementioned diseases, and TLR4 controls many microglia-specific responses known to dysregulate neuronal actions. TLR4 represents one of many receptors expressed primarily on microglia, and TLR4 is known to invoke a proinflammatory response to many stimuli.

In addition, TLR4 is a feasible drug target because of its well-characterized structure and downstream signaling pathway. Two clinically relevant TLR4 inhibitors, TAK-242 and Eritoran, stand as justification that small molecules can target TLR4 with reasonably high affinity and specificity. Eritoran exploits LPS structural components; it is therefore large and probably blood–brain barrier impermeable, although this has not been tested to the best of our knowledge. Nevertheless, the blood–brain barrier is compromised as a result of some CNS conditions, suggesting a possible venue for these molecules in the treatment of cerebral infarction and traumatic brain injury, for example.

Naloxone and naltrexone are also clinically relevant TLR4 effectors, but these are limited by short half-life. As interest in microglial targets grows, so too has the use of these opioid antagonists. Intruigingly, naltrexone, the longer-acting of the two antagonists, has recently found a market in the treatment of severe alcoholism. Although cited for its ability to inhibit the reinforcing effects of endogenous opioids (i.e. endorphins), there exists a possibility that naltrexone also works to prevent the action of TLR4-mediated effects, as documented by several studies of TLR4 inhibition (Liu et al. 2000; Wu et al. 2006). The influence of alcohol on TLR4 signaling is reviewed extensively by Szabo and colleagues, who also suggest ethanol as a useful probe of TLR4 signaling with respect to lipid rafts, TLR4 complex association and receptor clustering (Szabo et al. 2007).

TLR4 is also special for its ability to control more than one inflammatory pathway, both the MyD88 and TRIF-mediated pathways. It is the only toll-like receptor known to signal through both the TRIF-dependent and MyD88-dependant pathways. TLR3 is the only other TLR with access to the TRIF/TRAM pathway, and TLR3 is an impractical target because it is expressed on endosomal vesicles (Yamamoto et al. 2003).

Within the TLR4 complex, there are several well-studied interactions that give rise to feasible drug targets. Because both MD-2 and TLR4 are needed to coordinate a signal, inhibiting either one would theoretically inhibit downstream activity. For example, curcumin is suggested to inhibit TLR4 signaling by binding MD-2 (Gradisar et al. 2007). Another way to achieve TLR4 inhibition is to prevent the association of MD-2 and TLR4, a required interaction for signal transduction to occur. This has been achieved using rationally designed peptide inhibitors (Slivka et al. 2009). Finally, the homodimerization interface is an important regulatory site for TLR4 activity. By inhibiting the homodimerization between two TLR4–MD-2 molecules, drugs could prevent TLR4 activation without disrupting native inhibition mechanisms. In other words, the homodimerization interface described by Park and colleagues may be a more benign target to control TLR4 signaling, as it would allow ligands to occupy MD-2 or TLR4 without activating downstream signals through homodimerization.

TLR4 is a complex, dynamic target in structure-based drug design. Both agonists and antagonists have potential therapeutic applications for CNS conditions. The voids in our understanding of TLR4 recognition and subsequent intracellular signaling are balanced by recent advances in TLR4 and MD-2 structure/activity determination. Since the discovery of toll-like receptors, an unprecedented amount of progress has been made to characterize and control these ancient host defense mechanisms. Barring the challenges of blood-brain permeability, researchers are hot on the trail of a TLR4 antagonist for the treatment of the above pathologies. Future TLR4 studies will impact our understanding from basic cellular signaling to the treatment of important neurological diseases. The recent advances in TLR4 and microglial research stand as justification that even greater developments are yet to come.

Acknowledgements

We thank the National Institutes of Health (DA026950, DA025740, NS067425, and DA029119) for support of this work. Dr. Mark Hutchinson is the recipient of an Australian National Health and Medical Research Council CJ Martin Fellowship (ID 465423). We are grateful to Dr Catherine M. Joce and Leslie A. Morton for critical review of this manuscript.

Ancillary