A small molecule blocking TLR4 reduced brain inflammation in female mice with Alzheimer's risk

APOE (apolipoprotein E) genotype is a major risk factor for sporadic Alzheimer’s disease (AD), with APOE4 increasing risk up to 12-fold compared to APOE3 (reviewed in [1, 2]). Therefore, an important challenge is identifying therapeutic targets for APOE4 carriers in AD. The role of APOE in AD is complex, as APOE has been shown to modulate multiple functions and pathways in humans and transgenic mouse models that overproduce amyloid-beta (Aβ) via familial AD mutations (FAD) (reviewed in [3, 4]). For example, compared to APOE3, APOE4 is associated with higher soluble Aβ levels and amyloid plaques, altered metabolism, and neurovascular function (reviewed in [3–5]). In addition, increasing evidence suggests that APOE-modulated neuroinflammation contributes to AD progression [6]. In AD patients, compared to APOE3, APOE4 is associated with earlier onset of cognitive deficits and increased neuroinflammation [7–10]. These human data are recapitulated in vivo, as APOE4 is associated with greater microgliosis [11–14], astrogliosis [13–15], and altered cytokine levels [12, 14, 16, 17] in FAD mice and in APOE-knock in mice after induction of peripheral inflammation. In addition, there is greater neuroinflammation in female APOE4 FAD mice compared to males [11, 15, 18, 19], consistent with higher AD risk and pathology in female APOE4 carriers. Therefore, identifying pathways that contribute to APOE4-associated neuroinflammation is an important approach for developing AD therapeutics.

Neuroinflammation is complex and involves multiple cell-types, receptors, signaling pathways, cytokines, and chemokines. One approach to identify the contribution of neuroinflammatory pathways to AD progression is to evaluate activity of FDA approved drugs with known anti-inflammatory activities such as NSAIDs [20]. Although some studies supported that NSAIDs may be efficacious as an AD therapeutic, including in APOE4 carriers, others have shown no beneficial effects [21–23]. Ongoing research is defining if these established anti-inflammatories will be beneficial for specific patient groups at certain stages of AD and the optimal treatment regime. An alternative approach is to determine if compounds that target pathways modulated by APOE4 can impact neuroinflammation and behavior. The innate immune system is one of the most highly conserved immune responses across plants, drosophila, and animals to defend against invading pathogens and is also activated by endogenous stress-related molecules. Toll-like receptor 4 (TLR4) is a key component of the innate immune response is expressed by microglia (and to a lesser extent by astrocytes) [24, 25] and is activated by lipopolysaccharide (LPS) and other stress-associated ligands (reviewed in [26, 27]). Evidence suggests that TLR4 could contribute to neuroinflammation in AD. For example, there is higher TLR4 expression in AD patients’ [28, 29] and FAD mouse brains [29, 30] and single nucleotide polymorphisms (SNPs) in TLR4 have been shown to modulate AD risk [31–33]. There is also evidence that TLR4 is involved in APOE4-associated neuroinflammation. In vitro, LPS-induced inflammatory response is greater with APOE4 compared to APOE3 [14, 16, 17]. Furthermore, Aβ-induced cytokine production is greater with APOE4 in glial cultures, an effect that is blocked by TLR4 antagonists, and expression of TLR4-related genes are greater in APOE4-FAD mice [34]. Based on these data, it has been proposed that with APOE4, greater TLR4 activation in the brain may lead to higher neuroinflammatory responses and contribute to behavioral deficits. However, to date, no studies have directly evaluated the effect of blocking TLR4 on AD-relevant pathology and behavior in mice that express APOE4.

The goal of this study was to determine whether TLR4 antagonism can modulate neuroinflammation, Aβ pathology, and behavior in mice that express APOE4. To address this goal, we used a novel small molecule TLR4 antagonist IAX0-101, which we have previously shown to inhibit Aβ-induced cytokine release in APOE4-mixed glial cells [34]. Therefore, we treated mice that express APOE4 (E4FAD) and overproduce Aβ with IAXO-101 in prevention and reversal paradigms and evaluated the impact on neuroinflammation, Aβ pathology, and behavior.

The goal of this study was to evaluate the effect of TLR4 antagonism on neuroinflammation and other markers of AD-relevant pathology in mice that express human APOE. To address this goal, we treated EFAD mice from 4 to 6 months of age (Prevention paradigm; PVT) or from 6 to 7 months of age (reversal paradigm; RVS) with either vehicle or IAXO-101 (10 mg/kg ~ 0.3 mg/mouse, 3 subcutaneous injections/week). We used EFAD mice as they overproduce Aβ42 and express human APOE4 (E4FAD) or APOE3. Previous data have demonstrated that E4FAD mice have higher levels of astrogliosis, microgliosis, and Aβ pathology at 6 months of age compared to E3FAD mice [12, 35]. Therefore, in our prevention paradigm, we treated mice from 4 to 6 months or age, and in our reversal paradigm, treatment was from 6 to 7 months. For TLR4 antagonism, we used IAXO-101, as we have previously demonstrated that this compound lowers inflammation in vitro using APOE4 mixed glial cultures. IAXO-101 has been used in vivo at doses ranging from 0.06 to 0.3 mg/mouse/day via various routes [46–48], and we therefore selected the higher dose for this study (~ 0.3 mg/mouse). There was no difference between the body weights of mice treated with vehicle or IAXO-101 within a specific cohort (Additional file 1: Fig. S7). To evaluate the effects of IAXO-101, we measured neuroinflammation (astrogliosis, microgliosis, glial cell morphology), Aβ pathology, and behavior in EFAD mice.

Neuroinflammation mediated by microglia is increasingly recognized as an important component of AD pathogenesis. However, the role of microglia function in AD is complex and is dependent on many factors including activation state and disease stage. For example, it has been proposed that moderately activated microglia are neuroprotective in the early stages of AD, while chronically activated microglia may be detrimental in later stages (reviewed in [64–66]). Genetic AD risk factors and sex may also impact the extent that microglia are beneficial, neutral, or detrimental for cognitive function in AD. Our study supports that reducing microglial neuroinflammation is particularly or specifically beneficial for improving memory in female APOE4 carriers. Although there are limitations (see below), the implication is that neuroinflammation may contribute to higher risk and progression of AD found in APOE4 females. Consistent with this concept, human data demonstrate that compared to APOE3, female APOE4 AD patients have earlier onset of cognitive deficits (reviewed in [67]) and increased neuroinflammation, including increased activation of microglia and secretion of cytokines [7–10]. Similarly, in mice, greater impairment in learning/memory and neuroinflammation is seen in females compared to males in multiple FAD mouse models [68–72], with APOE4 exacerbating these effects [11, 15, 19]. Thus, collectively, the combination of female sex and APOE4 may produce an environment that specifically results in greater microglial inflammatory respond that leads to greater cognitive dysfunction in AD.

The concept that APOE4 and female sex synergistically impact neuroinflammation to impair cognition raises the important discussion of the underlying mechanisms. In general, apoE4 has been proposed to alter cellular function in different ways [73]. At the structural level, apoE isoforms differ by two amino acids at position 112 and 158, which affects protein folding and is thought to result in changes to lipidation (apoE4 < apoE3) stability and levels (apoE4 < apoE3) [74, 75]. These structural changes are proposed to induce differences apoE availability, distribution, aggregation, receptor binding affinities and signaling that modulates peripheral inflammation (via macrophages and other immune cells), cerebrovascular function (via endothelial cells, pericytes and astrocytes), Aβ levels, lipid trafficking, and neuron function directly (reviewed in [3, 76–79]). Therefore, the impact of APOE on other cell types could indirectly impact glial morphology/activation. In addition, apoE can be produced by microglia to impact their activity. For example, microglia expressing APOE4 has altered immune responses and metabolism in vitro [18, 80, 81], and selective ablation of microglial apoE4 in a tauopathy mouse model blocks brain atrophy [82, 83]. Thus, collectively APOE4 could indirectly and directly results in higher potential for microglia reactivity to be exacerbated by female sex.

The underlying mechanisms of how female sex impacts AD progression is unclear. One proposed mechanism is that loss of sex hormones at menopause results in greater AD susceptibility. Indeed, ovariectomy results in behavioral impairments and accelerated aging [84–86] in women as well as in rodents [87, 88]. Importantly, ovariectomy results in greater neuroinflammation in rodents including microglial activation [87–91]. Therefore, as estrogen demonstrates anti-inflammatory activity via the estrogen receptors, the loss of estrogen may promote a microglial response [92]. In addition to direct effects on microglia, the loss of estrogen also modulates AD pathology, as in general, Aβ deposition is increased in women during the menopausal transition [93, 94], in post-mortem brains of ovariectomized women [95], and in brains of ovariectomized rodents with some conflicting results in FAD mice (discussed in [96]). Thus, menopause could indirectly and directly modulate microglial activation in a way that is particularly pronounced with APOE4 carriers. However, in our study, mice were not ovariectomized, suggesting a different mechanism of action for how female sex and APOE4 impacted glial activation. In general, there are sex differences in microglial numbers, function, and response to an acute/chronic insult and age without menopause mimics [62, 97], including in APOE4 FAD mice [11, 34]. Thus, one potential explanation is with APOE4, there is greater age-dependent changes in sex hormones or their receptors that influence the inflammatory state of microglia, thereby modulating its function. Alternatively, the difference between sex hormones and immune-related genes on the X chromosome may result in higher age-related neuroinflammation (reviewed in [98–101]). Future studies could focus on understanding in more detail the precise mechanism(s) though which sex and APOE4 increase neuroinflammation and microglial activation.

Our data supports a specific role of TLR4 in modulating neuroinflammation and cognition in APOE4 females. Data from human, in vivo and mouse studies have led to the idea that blocking TLR4 is a potential AD therapeutic strategy albeit in a non-APOE4 context. Several single nucleotide polymorphisms (SNPs) in TLR4 have been associated with modulating AD risk in humans. For example, the G coding variant of rs4986790(A/G) [31, 32] and the C variant of rs11367(G/C) in TLR4 is associated with decreased AD risk [102]. Consistent with a detrimental role of TLR4 in AD, in vitro and ex vivo studies supports that these variants reduce TLR4 activation, resulting in attenuation of proinflammatory signaling [103, 104]. Several minor alleles of TLR4 SNPs (rs10759930, rs1927914, rs1927911, rs12377632) were also found to increase the risk of AD [33], although their function is yet to be understood. Further evidence for a role of TLR4 in AD include that levels of LPS [105] and TLR4 expression are increased in the brains of AD patients [28, 29] and FAD mice [29, 30]. In addition, Aβ has been shown to activate TLR4 resulting in reactive oxygen species production in microglia in vitro [106], and deletion of the TLR4 co-receptor CD14 resulted in lower amyloid plaques in vivo [107]. An important question for further study is whether SNPs in TLR4, TLR4 expression data, and in vitro functional data are modulated by APOE and sex in AD-relevant context. In terms of APOE alone, LPS (TLR4 agonist) induces APOE-specific neuroinflammatory markers in multiple tissues (APOE4 > APOE3) in vivo [14, 108], ex vivo [108], and in vitro [109] with an inconsistent sex effect [110]. In terms of female sex, TLR4 expression is greater in multiple tissues in females than males in human and rodents [110], and the estrogen receptor may impact TLR signaling [92]. Potentially, therefore, the combination of APOE and female sex may specifically result in altered TLR4-mediated neuroinflammation to impact cognition.

The implication of our findings is that targeting neuroinflammation is a potential AD therapeutic approach; for certain high-risk groups, however, current human studies are conflicted. To date, most data is derived from epidemiological studies, and a few clinical trials focused on established drugs with known anti-inflammatory properties. One class is NSAIDs that inhibit COX1 and 2 activities. Epidemiological and in vivo studies indicated that non-steroidal anti-inflammatory drugs (NSAIDs) lower AD risk and pathology (reviewed in [111, 112]). However, NSAIDs have not been found efficacious in some AD clinical trials [20]. In addition, a recent re-analysis of the NSAID epidemiological data in an ADNI dataset revealed that although some NSAIDs are associated with decreased AD prevalence, they did not affect cognitive decline in AD patients [21]. A second drug used to evaluate neuroinflammation as an AD target is paracetamol that has unknown and likely pleiotropic mechanisms of action. For example, paracetamol is used for its anti-pyretic, anti-inflammatory, and analgesic actions that may be mediated through lowering prostaglandin levels, inhibiting COX, altering cannabinoid signaling, and also blocking neuronal sodium channels [113]. Paracetamol has been shown to lower AD prevalence; however, whether the beneficial effects of paracetamol in AD are due to effects on inflammation or other mechanisms is unclear [21]. In addition to further understanding the mechanism of action of anti-inflammatories, there are several important considerations for clinically targeting neuroinflammation in AD. One is that neuroinflammation is complex and recognition of pathogens or danger-associated molecules occurs in multiple cell types, through a large array of receptors that produce an equally complex signaling and functional response. Thus, targeted one specific pathway may not alone be sufficient to improve all aspects of neuroinflammation in AD. Therefore, a detailed understanding of how different pathways contribute to any proposed detrimental neuroinflammatory phenotype in AD, including which aspects are more proximal to neuronal dysfunction, is important for interpreting clinical trial data. A second is that whether an inflammatory mediator or receptor is beneficial or detrimental for AD progression may depend on the stage of disease. For example, it has proposed that anti-inflammatories may be efficacious as a prophylactic treatment for APOE4 carriers to prevent AD [21]. A third, as highlighted by our study, is that the contribution of neuroinflammation to neural dysfunction may be dependent on risk factors (e.g., sex, APOE genotype, lifestyle, other genetic or environmental factors). Perhaps future studies in larger heterogenous populations could provide an explanation for determining whether NSAIDs have a greater impact in specific groups.

There are limitations in the extent that we can conclude TLR4 and/or neuroinflammation specifically contributes to neural dysfunction in female APOE4 mice. In fact, studies have shown that microglial activation and neuroinflammation are greater with APOE4 than APOE3 in males using APOE knock in mice [14, 114–116], FAD mice [11, 15, 43], and AD patients [117, 118]. Indeed, it is important to note that we found that IAXO-101 lowered microglial number and reactivity in older male E4FAD mice, although performance in the Morris water maze test was not affected. The contribution of TLR4 to microglial activation maybe dependent on age and/or Aβ pathology, which is why IAXO-101 modulated microglia number/reactivity when treatment occurred in older rather than younger male E4FAD mice. The lack of a behavioral benefit in male E4FAD mice as opposed to female E4FAD mice with IAXO-101 treatment could be due multifactorial factors. For example, there could be the same age-dependent consideration as for microglial activation within males, such that beneficial effects in behavior would have been observed if treatment had continued for longer. Alternatively, the contribution of microglia to performance in Morris water maze performance may be greater in females than males. Furthermore, a more comprehensive set of behavioral tests along with expanded markers of neuron function may have revealed that IAXO-101 was beneficial in E4FAD male mice. In addition, APOE3-FAD also shows age-dependent increases in glial activation and cognitive impairment [19, 40]. Therefore, future studies could explore if TLR4, neuroinflammation, and/or microglia activity are more proximal to behavioral dysfunction in older male E4FAD, female E3FAD, and male E3FAD mice. Such studies could reveal if there are different optimal treatment windows for targeting neuroinflammation within each APOE genotype and sex combination or support specificity for female APOE4 carriers.

Our study was a proof-of-concept design centered on TLR4 antagonism. However, TLR4 may not be the main contribution to inflammation for all APOE genotypes and sexes. For example, there is the possibility that TLR4 activation is more pronounced in female APOE4 mice, but other inflammatory pathways are important for APOE4 male mice or APOE3 mice. Indeed, sex hormones and X chromosome genes can differentially regulate TLR4 [98, 99, 119] and testosterone decreases expression and function of pro-inflammatory cytokines [120]. In addition, as mentioned above, different inflammatory pathways may be important at different ages/stages of pathology. Therefore, identifying how APOE genotype, sex, and age impact the neuroinflammatory phenotype, including specific pathways, is important for ultimately advancing our mechanistic understanding of neuroinflammation in AD.

There were also experimental limitations surrounding the way that we targeted TLR4. One aspect is the dose of IAXO-101, which we selected based on previous publications; however, we did not conduct detailed PK studies to determine brain and plasma levels. Therefore, IAXO-101 levels in the plasma and brain may have been different for each group of mice, such as higher in female E4FAD mice, which may explain the beneficial effects. In addition, IAXO-101 inhibits TLR4 activation by two mechanisms: sequestering LPS by forming stable co-aggregates or competing with LPS for binding to CD14 and myeloid differentiation factor 2 [121]. This mechanism of action is based on LPS; however, it is unclear what pathways lead to TLR4 activation in female E4FAD mice. Thus, another approach would be to use competitive inhibitors of TLR4 and/inhibitors of TLR4 signaling down-stream. There is also the question of whether there are cell type specific functions of TLR4. TLR4 is expressed by virtually every cell in the body that could have opposed or synergistic functions. In fact, both activating and inhibiting TLR4 has been shown to decrease AD-related pathology in FAD mice with opposing effects on neuroinflammation [122, 123]. Therefore, evaluating the cell type-specific functions of TLR4 in EFAD mice using detailed mechanistic readouts for neuroinflammation, neuron function, and behavior as well as plasma biomarkers is important for fully evaluating TLR4 as a therapeutic target in AD.

Our study demonstrates that the TLR4 antagonist IAXO-101 improves memory and reduced neuroinflammation with minimal effects of Aβ pathology in female APOE4 mice. Thus, TLR4 inhibition is a potential therapeutic approach for AD, particularly in female APOE4 carriers. In addition, these results support stratification of preclinical and clinical studies that target neuroinflammation by APOE, sex, and treatment window.