Authors: Benjamin C. Shaw (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA), Victoria R. Anders (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA), Rachel A. Tinkey (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA; 2School of Biomedical Sciences, Kent State University, Kent, OH, USA; 3Brain Health Research Institute, Kent State University, Kent, OH, USA), Maria L. Habean (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA; 4Department of Neuroscience, Case Western Reserve University, Cleveland, OH, USA), Orion D. Brock (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA; 5Molecular Medicine, Lerner Research Institute, Cleveland Clinic and Case Western Reserve University, Cleveland, OH, USA), Benjamin J. Frostino (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA; 6College of Science, University of Notre Dame, South Bend, IN, USA), Jessica L. Williams (1Department of Neurosciences, Lerner Research Institute, Cleveland Clinic, Cleveland, OH, USA; 2School of Biomedical Sciences, Kent State University, Kent, OH, USA; 3Brain Health Research Institute, Kent State University, Kent, OH, USA; 4Department of Neuroscience, Case Western Reserve University, Cleveland, OH, USA; 5Molecular Medicine, Lerner Research Institute, Cleveland Clinic and Case Western Reserve University, Cleveland, OH, USA)
Categories: Article
Source: Journal of neurochemistry
Doi: 10.1111/jnc.15999
Authors: Benjamin C. Shaw, Victoria R. Anders, Rachel A. Tinkey, Maria L. Habean, Orion D. Brock, Benjamin J. Frostino, Jessica L. Williams
Cognitive deficits are a common comorbidity with neurological disorders and normal aging. Inflammation is associated with multiple diseases including classical neurodegenerative dementias such as Alzheimer’s Disease (AD) and autoimmune disorders such as multiple sclerosis (MS), in which over half of all patients experience some form of cognitive deficits. Other degenerative diseases of the central nervous system (CNS) including frontotemporal lobe dementia (FTLD), and Parkinson's disease (PD) as well as traumatic brain injury (TBI) and psychological disorders like major depressive disorder (MDD), and even normal aging all have cytokine-associated reductions in cognitive function. Thus, there is likely commonality between these secondary cognitive deficits and inflammation. Neurological disorders are increasingly associated with substantial neuroinflammation, in which CNS-resident cells secrete cytokines and chemokines such as tumor necrosis factor (TNF)α and interleukins (IL) such as IL-1β and IL-6. CNS resident cells also respond to a wide variety of cytokines and chemokines, which can have both direct effects on neurons by changing the expression of ion channels and perturbing electrical properties, as well as indirect effects through glia-glia and immune-glia crosstalk. There is significant overlap in these cytokine and chemokine expression profiles across diseases, with TNFα and IL-6 strongly associated with cognitive deficits in multiple disorders. Here, we review the involvement of various cytokines and chemokines in AD, MS, FTLD, PD, TBI, MDD, and normal aging in the absence of dementia. We propose that the neuropsychiatric phenotypes observed in these disorders may be at least partially attributable to a dysregulation of immunity resulting in pathological cytokine and chemokine expression from both CNS-resident and nonresident cells.
Neurological disorders are a leading cause of physical and cognitive disability worldwide (Feigin et al., 2020). This burden is increasing due to longer life expectancy and population growth. Cognitive deficits associated with neurological disorders include, but are not limited to, verbal and spatial memory recall issues, processing speed deficits, attention and concentration deficits, and impaired executive function (Figure 1). Some neurological disorders, such as Alzheimer’s Disease (AD) and other dementias, rely on the manifestation of cognitive deficits as a defining disease feature; for others, like multiple sclerosis (MS) and traumatic brain injury (TBI), cognitive deficits are a frequent, but not diagnostic, comorbidity (Benedict et al., 2006; McMahon et al., 2014).
While cognitive deficits are a consistent complication of neurological disorders, in many cases, the molecular causes of these clinical impairments remain unknown. Neurological diseases are also associated with alterations in concentrations of immune signaling molecules such as cytokines and chemokines both systemically and within the central nervous system (CNS). These molecules, in addition to their roles in the immune response, are also neuromodulatory and can affect global CNS function. This raises the possibility that, however starkly different these disease etiologies may be, neurological disorders may share commonalities in the development of secondary cognitive deficits.
One commonality appears to be neuroimmune interactions mediated through secreted molecules such as cytokines and chemokines. Here, we primarily discuss the interleukin (IL), interferon (IFN), and tumor necrosis factor (TNF) families. These families have well-described roles in neuroinflammation and are secreted by meningeal and CNS infiltrating T cells (Kocur et al., 2015; Ribeiro et al., 2019; Tzartos et al., 2008), microglia (Klintworth et al., 2009; Liddelow et al., 2017; Noda et al., 2013; Wißfeld et al., 2021), and astrocytes (Couturier et al., 2016; Guo et al., 2007; Liao et al., 2011; Lieberman et al., 1989; Tedeschi et al., 1986; Tzartos et al., 2008; Y. Yan et al., 2012), among others. Many of these cytokines alter neuronal electrophysiology and can change action potential frequency or threshold potential (Cuevas-Olguin et al., 2017; Donnelly et al., 2021; Laumet et al., 2020; J. Yan et al., 2012). Further, these immune signaling molecules have also been shown to trigger neuron or neural progenitor cell (NPC) cell death (Guadagno et al., 2015; Jayaraman et al., 2021).
In this review, we consider four neurodegenerative diseases (MS, AD, frontotemporal lobar dementia; FTLD, and Parkinson’s disease; PD), physical trauma (TBI), a neuropsychiatric disorder (major depressive disorder; MDD), and normal aging. We highlight several secreted immunity-associated molecules which are involved in each disease and are associated with their respective cognitive impairments focusing on the hypothesis that glia-to-neuron and immune cell-to-neuron signaling via cytokines and chemokines is an underappreciated hallmark of cognitive deficits across several neurological disorders using these selected molecules and disorders as illustrative examples.
MS is a neurodegenerative disease characterized by peripheral immune cell infiltration and demyelination across the CNS. MS has three primary clinical relapsing remitting MS (RRMS), characterized by periods of recovery between clinical exacerbations, secondary progressive MS, in which RRMS patients no longer recover between exacerbations, and primary progressive MS where there are no periods of recovery from the start of disease. While white matter lesions are most common, lesions and atrophy in the grey matter, such as in the hippocampus, frequently occur (Dutta et al., 2011; Dutta et al., 2013; Sicotte et al., 2008). Despite most preclinical and clinical research focusing on the improvement of motor deficits in MS, cognitive deficits are an understudied and chief complaint of MS patients. Cognitive deficits affect approximately 65% of MS patients (Benedict et al., 2006; Peyser et al., 1990; Rao et al., 1991) and can occur at any point in the disease (Pelosi et al., 1997; Piras et al., 2003). These deficits impact processing speed and recall (Benedict et al., 2006), long-term memory (Rao et al., 1993), and executive function and decision making (Chiaravalloti & DeLuca, 2008; Schultheis et al., 2001). Interestingly, hippocampal lesions do not appear to lead to axonal damage unlike white matter lesions such as those in the spinal cord (Dutta et al., 2013). Thus, the cognitive deficits from which MS patients suffer are not explained in the same way as motor deficits via neuronal loss. In both MS patients (Ruggieri et al., 2003) and in an animal model of MS, experimental autoimmune encephalomyelitis (EAE) (Acharjee et al., 2013; Habbas et al., 2015), cognitive deficits manifest prior to overt motor function loss. Neuroinflammation is thus a plausible precursor to these cognitive deficits.
TNFα is a pro-inflammatory cytokine with deleterious effects on memory in both humans (El-Salem et al., 2021) and in animal models (Habbas et al., 2015). TNFα is expressed locally within the CNS, and its receptors are ubiquitously expressed in the brain (Zhang et al., 2016). In MS patients, serum TNFα concentrations inversely correlate with cognitive performance (El-Salem et al., 2021). In one study, in addition to TNFα measurement, patients were screened for cognitive deficits using the Montreal Cognitive Assessment (MoCA) and Brief International Cognitive Assessment for Multiple Sclerosis (BICAMS), as well as for overall motor disability using the Enhanced Disability Status Scale (EDSS). Motor disability (EDSS) and cognitive deficits (MoCA and BICAMS) positively correlated with TNFα serum level. Animal models corroborate these findings using EAE. As early as day 7 following EAE induction, prior to the onset of motor decline, peripheral immune cell infiltration, or hippocampal demyelination, mice exhibited deficits in the Morris Water Maze (MWM), a measure of spatial memory, and fear conditioning (Acharjee et al., 2013). Authors also observed a striking but not statistically significant increase in Tnfa mRNA in the hippocampus of these mice. Further evidence comes from Habbas et al., who performed elegant studies dissecting TNFα signaling in the hippocampus (Habbas et al., 2015). Using adoptive transfer EAE, substantial increases in TNFα protein in the hippocampus as well as fear conditioning deficits were observed corroborating previous findings (Acharjee et al., 2013). This increase in TNFα protein seemed to be specific to the dorsal hippocampus which is integral to the contextual learning and memory circuitry (Anagnostaras et al., 2001). Studies using constitutive deletion of Tnfr1 and astrocyte-selective re-expression demonstrated that physiologically relevant (600 pM) TNFα application elicits presynaptic neuronal hyperactivity which is mediated by astrocytic TNFα signaling, possibly through astrocytic glutamate release. Authors posit that TNFα in the hippocampus acts on astrocyte-to-neuron signaling to inhibit long-term potentiation (LTP), thereby inducing memory and learning deficits (Habbas et al., 2015). Taken together, TNFα has a substantial and detrimental role in cognition in MS and EAE.
IL-17 is another pro-inflammatory cytokine implicated in MS pathogenesis with a possible pleiotropic role in cognition, whose receptors are ubiquitously expressed in the CNS (Zhang et al., 2016). IL-17 increases blood-brain barrier (BBB) permeability, and the adoptive transfer of IL-17-producing T helper type 17 (Th17) cells is sufficient to induce EAE (Kebir et al., 2007). Serum IL-17 concentration has a weak but significant inverse correlation with performance on cognitive tests in MS (Trenova et al., 2018). Increased IL-17 concentration correlated with poorer performance on tests for information processing speed (paced auditory serial addition test; PASAT) and rapid visual processing (symbol digit modalities test; SDMT). This correlation was significant even after accounting for potential covariates such as EDSS, age, and disease duration (Trenova et al., 2018) . While Th17 cells are a potent source of IL-17 during neuroinflammation, IL-17 can also be produced locally by resident microglia within the hippocampus during EAE (Di Filippo et al., 2021). It was demonstrated that this IL-17 can directly repress LTP, likely through neuronal expression of IL-17 receptor. Interestingly, Il17a^−/−^ mice with EAE retain their ability to learn a novel environment using the Hole-Board Test (Di Filippo et al., 2021). Importantly, Ribeiro et al. found that IL-17 produced by meningeal γδ T cells is required for short-term memory development (Ribeiro et al., 2019). Using either IL-17- or TCRδ-deficient mice, significant deficits in short-term paradigms of both the Y Maze and MWM, hippocampal-dependent tests for spatial memory, were observed. When using a long-term learning paradigm, however, mice did not display deficits in either test (Ribeiro et al., 2019). Since loss of IL-17 led to significant decreases in LTP, IL-17 may have a paradoxical role in learning and memory. This suggests that some basal level of IL-17 is required for short-term memory and LTP, but that higher concentrations of IL-17, which occur during diseases like EAE and MS, is pathological and abrogates LTP. Whether this abrogation is due to direct effects via neuronal IL-17 receptor signaling or through astrocytic IL17RA activation is unknown.
IFNβ is a pleiotropic cytokine, but has a primarily anti-inflammatory role in MS (Kasper & Reder, 2014). Regarding CNS resident cells, its receptor is expressed in astrocytes, oligodendrocytes, neurons, and microglia. IFNβ clinical trials for MS were successfully completed in 1993 (The IFNB Multiple Sclerosis Study Group, 1993). Patients given IFNβ exhibited a significant decrease in relapse rate over the two-year trial, indicating fewer periods of disability. An early clinical trial for IFNβ found significant improvement in the Wechsler Memory Scale Visual Reproduction-Delayed Recall test, which assays immediate and delayed recall memory, in a dose-dependent manner (Pliskin et al., 1996). This effect remained after accounting for changes in magnetic resonance imaging (MRI), indicating beneficial effects of IFNβ independent of disease activity. There was also no difference in EDSS progression between the placebo, low dose, and high dose groups. Another trial followed patients on IFNβ for 2 years and found a significantly lengthened time to onset of sustained deterioration in the PASAT processing rate (Fischer et al., 2000). Likewise, a long-term clinical trial investigated the effect of IFNβ on cognition by following a cohort of 16 subjects for 16 years (Lacy et al., 2013). Authors found significant benefit in immediate and delayed memory using the same tests, corroborating results from Pliskin et al. In the BENEFIT trial, patients were monitored for up to 8.7 years after initially starting IFNβ therapy (Edan et al., 2014). Patients who started IFNβ treatment earlier in disease performed better on the PASAT test, indicating better information processing speed (Edan et al., 2014). However, in a study which followed MS patients for up to 3 years, known as the CONFIDENCE trial, authors found no significant association between IFNβ medication adherence and cognitive performance using a number of neuropsychiatric batteries including the SDMT and PASAT (van Ballegooijen et al., 2022). Of note, patients in IFNβ trials also had a lower annualized relapse rate, so it is not possible to determine whether this improved performance was due to a direct benefit of IFNβ on the brain or a result of decreased disease activity.
AD is the most common form of dementia, accounting for 60-80% of all cases (Alzheimer's Association, 2021). Patients with AD experience significant cognitive deficits including loss of memory and executive function. Histologically, AD is characterized by the accumulation of neurofibrillary tangles and amyloid-β (Aβ) plaques (Nelson et al., 2009). While a full description of AD pathophysiology is beyond the scope of this review, readers are referred to a recent comprehensive review (Long & Holtzman, 2019). AD plaques typically form in the entorhinal cortex, hippocampus, and frontotemporal lobe. Recent evidence implicates immune cells in the pathology of AD as genome-wide association studies have identified at least 40 risk factors for late-onset AD and over half of these are either primarily or exclusively expressed in microglia (Hansen et al., 2017; Srinivasan et al., 2016). Additionally, the Aβ plaques can be seeded by cores of aggregated apoptosis spec-like protein containing a caspase recruitment domain (ASC) protein (Ising et al., 2019; Venegas et al., 2017). ASC is an adaptor protein in the inflammasome complex assembled within myeloid cells, most prominently described in microglia. Further, the adaptive immune system may also be implicated in AD pathology. A recent study by Gate et al. found a signature of CD8^+^ effector memory T cells in both peripheral blood and the cerebrospinal fluid (CSF) of AD patients (Gate et al., 2020). This signature was inversely correlated with cognitive performance, implying a detrimental effect of these CNS T cells (Gate et al., 2020). These data corroborate a previous report which demonstrated an association between activated T cells, either in the peripheral blood or CSF, and structural MRI changes, specifically microstructural parahippocampal damage (Lueg et al., 2015). CD8^+^ T cells have also been reported in close proximity to neurons undergoing necroptosis in AD brain (Jayaraman et al., 2021). Further, Monson et al. demonstrated that AD and MS patients have similar amounts of CD8^+^ T cells in their CSF (Monson et al., 2014). Thus, investigating the molecules secreted by microglia and peripheral immune cells in AD will likely lead to a better understanding of the disease.
AD patients also exhibit increased TNFα in peripheral blood and CSF (Culjak et al., 2020; Swardfager et al., 2010; Tarkowski et al., 2003). Plasma TNFα concentration correlates with plasma Aβ42 and total plasma tau concentration in African Americans, suggesting TNFα may be a useful biomarker for AD (Deniz et al., 2021). Systemic TNFα correlates with the rate of cognitive decline in AD patients as well (Holmes et al., 2009). TNFα is also highly expressed in the CA3 hippocampal region of AD brains analyzed post-mortem (Zhao et al., 2003). Furthermore, AD brain has increased expression of the genes encoding the receptors for TNFα, TNFR1 and TNFR2 (Jayaraman et al., 2021).
Indeed, preclinical studies indicate a potential for TNFα inhibition as a possible therapeutic target for AD (Bhaskar et al., 2014; Cavanagh et al., 2016; Chang et al., 2017; McAlpine et al., 2009). In one study, authors administered two different TNFα scavenger proteins to APP/PS1 mice. These scavenger proteins consist of a TNF receptor fused to stabilizing protein domains. Administration of either of these TNFα scavengers resulted in substantial decreases in amyloid deposition and better discrimination in the novel object recognition test in the APP/PS1 mice (Chang et al., 2017). Other studies using humanized mice showed TNFα secreted by microglia regulates the neuronal cell-cycle which in turn leads to neuronal death in vivo (Bhaskar et al., 2014), and inhibiting TNFα signaling reduces Aβ pathology (McAlpine et al., 2009) and improves performance on memory tests with concomitant recovery of LTP (Cavanagh et al., 2016). Studies using cultured rat neurons demonstrate that TNFα receptor expression may differ between young and old neurons, suggesting a possible age-related pro-apoptotic effect of TNFα (Patel & Brewer, 2008).
This preclinical evidence has spurred an interest in TNFα inhibition for the treatment of AD patients (Butchart et al., 2015; Tobinick & Gross, 2008a, 2008b). A single-patient case report showed that perispinal administration of etanercept led to significant clinical benefit (improved lucidity and performance on the MoCA) within 2 hours (Tobinick & Gross, 2008a). A subsequent pilot study followed 15 patients who were given perispinal etanercept weekly for 6 months (Tobinick & Gross, 2008b). Twelve of the 15 patients in this study saw significant improvement across a battery of verbal cognitive tests at the 6-month endpoint (Tobinick & Gross, 2008b). Importantly, in a trial using a 24-week regimen of weekly subcutaneous, rather than perispinal, etanercept, there was significant improvement in the Bristol Activities of Daily Living Scale, a measure of dementia patients’ ability to complete daily tasks such as dressing, and the Neuropsychiatric Inventory, which assesses behavioral disturbances such as delusions and hallucinations (Butchart et al., 2015). Overall, anti-TNFα modalities may be a promising avenue for AD treatment.
Inflammasomes are multiprotein complexes, canonically comprised of a sensor, which can include Nod-like receptors (NLRs, including NLRP3), caspase-1, and the adaptor ASC. Pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs) bind to pattern recognition receptors, including Toll-like receptors (TLRs). This TLR engagement leads to downstream activation of the nuclear factor kappa B (NF-κB) pathway and induces transcription of both inflammasome components and substrates. After this priming, secondary activation of the complex through danger signals, such as lysosomal permeabilization, leads to assembly and activation of the inflammasome and cleavage of pro-IL-1β or pro-IL-18 into their mature forms for secretion. Interestingly, Aβ is both a ligand for TLR4 and a danger signal. When cells sense Aβ, inflammasome components such as NLRP3 (a well-described sensor for the NLRP3 inflammasome), caspase-1 (the active enzyme which cleaves pro-IL-18 and pro-IL-1β), and ASC, along with Il1b mRNA, and mature IL-1β in vitro are upregulated and the inflammasome assembles (Halle et al., 2008; Liu et al., 2020; Reed-Geaghan et al., 2009; Stewart et al., 2010; Vollmar et al., 2010). For a full review of inflammasome structure and function, readers are directed to a recent and comprehensive inflammasome review (Singh et al., 2023). This TLR-4-dependent mechanism has been demonstrated in vivo as well (Vollmar et al., 2010). Inhibition or genetic ablation of NLRP3 or caspase-1 reduces IL-1β production and improves memory behaviors in the APP/PS1 mouse model of AD (Gu et al., 2021; Heneka et al., 2013). At the molecular level, activation of the inflammasome using systemically injected lipopolysaccharide (LPS), a prototypical ligand for TLR4, increases hippocampal IL-1β while decreasing LTP (Lynch et al., 2004). In AD patients, both IL-1β and IL-18 are increased in peripheral blood (Swardfager et al., 2010). Intrathecal IL-1β concentration has been shown to correlate with cognitive deficits indicated by Mini-Mental Status Exam (MMSE) scores, a neuropsychiatric test for cognitive deficits commonly used for AD patients (Tarkowski et al., 2003).
However, the inflammasome is involved in AD beyond IL-1β. The adaptor protein ASC has a detrimental effect in seeding Aβ plaques. Studies have demonstrated the ability of ASC to seed Aβ oligomers in vitro and in vivo (Venegas et al., 2017). Importantly, they show that ASC also co-sediments with Aβ from human AD brain and that ASC increases the rate at which Aβ oligomerizes in a cell-free aggregation assay (Venegas et al., 2017). This plaque seeding has also been demonstrated in vivo using the APP/PS1 mouse model where injections of ASC protein led to increased amyloid deposition which is blocked by anti-ASC IgG treatment. Behaviorally, ASC-deficient mice crossed with either the APP/PS1 or 5xFAD line have less pronounced memory deficits than ASC-competent mice (Couturier et al., 2016; Venegas et al., 2017). In a subsequent study, authors observed that ASC-seeded Aβ aggregates induce caspase-1 cleavage to a greater extent than either ASC or Aβ alone (Friker et al., 2020). This in turn led to increased cleavage of gasdermin D and increased secretion of IL-1β, highlighting again that ASC-Aβ aggregates more efficiently seed than either ASC or Aβ alone. Microglia are also responsible for phagocytosis and clearance of Aβ, however, this process is significantly impeded when Aβ is bound to ASC (Friker et al., 2020), indicating a potentially deleterious feedback loop. ASC seeds amyloid plaques, and these ASC-Aβ aggregates can then act through TLR4 to activate the inflammasome. This in turn leads to the induction and secretion of more ASC and IL-1β, while simultaneously inhibiting Aβ clearance such that more Aβ can oligomerize with ASC. Additionally, aggregated tau activates the NLRP3 inflammasome in microglia (Jiang et al., 2021; Stancu et al., 2019). Inhibition of NLRP3 or genetic ablation of ASC reduced tau pathology in humanized tau mice in vivo (Stancu et al., 2019). Using a conditional knockout of ASC or NLRP3 in microglia, loss of inflammasome function also leads to better performance in the MWM and decreased hippocampal mature IL-1β expression (Ising et al., 2019; Jiang et al., 2021). The inflammasome is also required for Aβ-induced tau pathology in mouse models. Injection of APP/PS1 brain homogenate into the hippocampus of single-transgenic Tau22 mice, but not inflammasome-deficient Tau22 mice, led to increased tau hyperphosphorylation (Ising et al., 2019). Thus, the inflammasome is both activated by and contributes to tau pathology in addition to Aβ pathology. The inflammasome therefore has both an IL-1β-dependent and –independent mechanism in the progression of AD cognitive deficits.
FTLD is a neurodegenerative disease characterized by frontal and temporal lobe gray matter atrophy in conjunction with cognitive, language, and emotional impairments. FTLD is divided into three behavioral variant (bvFTLD), semantic dementia, and non-fluent variant primary progressive aphasia. The most common subtype, bvFTLD, is predominantly driven by disease-causing mutations including those in TAR DNA binding protein (TDP)-43 (Kokiko-Cochran et al., 2018; Vieira et al., 2021; Yin, Banerjee, et al., 2010; Yin, Dumont, et al., 2010). FTLD-TDP-43 shares mutational commonalities with amyotrophic lateral sclerosis (ALS), specifically those in open reading frame 72 on chromosome 9 (C9ORF72) and TDP-43, where the severity of ALS correlates with the onset of FTLD and vice versa, but each has a unique neuroinflammatory profile (DeJesus-Hernandez et al., 2011; Dutta et al., 2020; R. Kobayashi et al., 2022; McCauley & Baloh, 2019; Oeckl et al., 2019). Patients with ALS develop cognitive and language deficits, such as apathy, bland affect, reduced emotional insight, depression, and impaired sentence comprehension and grammar, in addition to motor deficits. Similarly, FTLD patients have greater cognitive impairment as motor neurons are lost (De Silva et al., 2016; DeJesus-Hernandez et al., 2011; McCauley & Baloh, 2019; Samra et al., 2023; Saxon et al., 2020; Saxon et al., 2017; Tan et al., 2015). Although there are various subtypes of FTLD, cognitive impairments remain similar between subtypes. Cognitive impairments in bvFTLD are noted as reduced empathy, altered dietary preferences, loss of inhibitory control, obsessive compulsive behavior, progressive change in personality, and apathy. However, there is no impairment in language-motor function (Chen et al., 2018; Krabbe et al., 2017; Phan et al., 2020; Samra et al., 2023). Corresponding with these deficits, gray matter atrophy occurs bilaterally in the frontal and temporal lobes as well as most lobules in the cerebellum (Chen et al., 2018; Pini et al., 2022). Increased cytokine and complement production are present in FTLD patients (Phan et al., 2020) where increased cytokine production is correlated with changes in behavior, presumably by altering neurotransmission (A. H. Miller et al., 2013). The cognitive impairments caused from common mutations in bvFTLD and cytokine production are mirrored in animal models as well (Krabbe et al., 2017; Yin, Dumont, et al., 2010). Gray matter loss is thought to be a result of inflammatory responses during FTLD, driven by secretion of cytokines such as TNFα, ILs, and chemokines from immune cells.
TNFα is a potent pleotropic cytokine involved in the modulation of inflammation during disease pathology. TNFα is produced by several resident CNS cells such as macrophages, astrocytes, and microglia, and exerts varying effects on neighboring CNS cells (Righi et al., 1989; Sjogren et al., 2004). Like TNFα, transforming growth factor (TGF)β is a pleiotropic cytokine with primarily anti-inflammatory effects. During FTLD pathology, TGFβ inhibits the production of pro-inflammatory cytokines including IL family members and TNFα, specifically by acting as a negative feedback mechanism for TNFα production (Sjogren et al., 2004). Measuring cytokine levels during FTLD, an increase in TNFα and TGFβ in the CSF compared to healthy controls was observed (Bossu et al., 2011; Krabbe et al., 2017; Martens et al., 2012; Z. A. Miller et al., 2013; Sjogren et al., 2004; Vieira et al., 2021). Importantly, there was no increase in serum TNFα or TGFβ and patients had an albumin quotient within normal range, indicating an intact BBB and local CNS production of cytokines (Sjogren et al., 2004). The concomitant increase in TGFβ with TNFα may indicate that TGFβ works to dampen excessive TNFα production as FTLD pathology progresses. Compared to healthy controls, patient-centered research focusing on bvFTLD uncovered many correlations between TNFα levels and cognitive deficits in social cognition, abstract reasoning, and problem-solving. Additionally, there was a correlation between the presence of TNF family members and brain atrophy. Patients with bvFTLD had increased cerebral and temporal lobe atrophy which positively correlated with soluble TNF receptor II implicating a role for this receptor in cognition and social behavior (Vieira et al., 2021). Separate studies investigated the involvement of a progranulin mutation, which often occurs in TDP-43 FTLD, on cognitive deficits and cytokine levels. Progranulin (Grn) deficiency led to increased serum TNFα levels signifying a dependency on progranulin for inhibition of TNFα secretion (Galimberti et al., 2015; Z. A. Miller et al., 2013; Yin, Banerjee, et al., 2010). In murine models of FTLD, specific deletion of Tnf and Grn abolished the excessive grooming phenotype, an example of obsessive compulsive-like behavior. Grn single knockouts exhibited elevated TNFα secretion that corresponded to impaired social interactions that were not improved in Tnf-deficient mice (Krabbe et al., 2017). Together this suggests that in FTLD, progranulin expression suppresses TNFα secretion thereby preserving cognition.
Both serum and CSF concentrations of many ILs correlate with the progression of FTLD pathology. Variants of FTLD are associated with increased plasma IL-6 (Gibbons et al., 2015; Phan et al., 2020), CSF IL-8 (Galimberti et al., 2006; Rainero et al., 2009), and CSF IL-11 (Galimberti et al., 2008; Rainero et al., 2009). By contrast, CSF IL-12 (Rentzos et al., 2006) is decreased. Some ILs seem to differ only when stratified for FTLD mutations. For instance, CSF IL-15 expression decreases in patients with GRN mutations, but not in sporadic FTLD (Rainero et al., 2009). Discrepant data exists, however, as patients with FTLD-TDP-43 and GRN mutations exhibit an increase in CSF IL-15 according to another study (Galimberti et al., 2015). CSF IL-17 decreases in FTLD-TDP43, while CSF IL-23 increases in FTLD with tau mutations (Hu et al., 2010). This suggests that GRN mutations lead to a pathologically divergent disease course, with increased microglial activation (Rainero et al., 2009) and systemic cytokine dysregulation (Bossu et al., 2011) compared to other subtypes. MMSE scores of FTLD patients positively correlate with CSF IL-11 concentration indicating involvement of this cytokine in early disease stages (Galimberti et al., 2008). While ILs are clearly altered in FTLD and correlate with cognitive changes, their mechanisms remain unknown. It would be beneficial to address this gap in knowledge in FTLD cognitive deficits using both animal models and human data given the current discrepancies and sparse literature.
Chemokines and complement are linked to frontal and temporal lobe atrophy in FTLD. During FTLD pathology, many studies have demonstrated increases in C-C motif chemokine ligand (CCL)2 and 3 (Sogorb-Esteve et al., 2021), C-X-C motif chemokine ligand (CXCL)1 and 6 (Sogorb-Esteve et al., 2021; Yin, Banerjee, et al., 2010), complement (C)3 (Lui et al., 2016; Phan et al., 2020), C1qa, C1qb, and C1qc (Lui et al., 2016). By contrast, decreased levels of CCL5, CCL19, and CXCL6 (Galimberti et al., 2015; Sogorb-Esteve et al., 2021) have also been observed during FTLD. These fluctuations have been associated with specific FTLD-causing genetic mutations. Other studies have focused on the FTLD-TDP-43 subtype with progranulin mutations. In Grn^−/−^ mice, microglia had increased levels of of C1qa, C1qb, C1qc, and C3 transcript corresponding to increased synaptic pruning and microglial activation during aging (Lui et al., 2016). Interestingly, Grn^−/−^ C1qa^−/−^ double knockout mice displayed a complete mitigation of the characteristic obsessive compulsive-like behaviors and a restoration of the thalamic microcircuit function when compared to Grn^−/−^ mice (Lui et al., 2016). This suggests a potential role for complement in the development of obsessive compulsive-like behaviors and learning impairment in FTLD-TDP-43 subtypes with progranulin mutations. Further, a case study demonstrating increased C3 levels in serum substantiated the role for C3 in cognitive decline during FTLD which had previously only been shown in mice (Lui et al., 2016; Phan et al., 2020). Interestingly, another human-based study that included patients with and without progranulin mutations found that decreased CCL5 levels were independent of progranulin mutations (Galimberti et al., 2015; Sogorb-Esteve et al., 2021). The ubiquitous effect of CCL5 in FTLD patients indicates a pathophysiological role independent of mutational subtypes.
PD is the second most common neurodegenerative disease worldwide and projected to progressively increase in global incidence and prevalence (Bloem et al., 2021; Menza et al., 2010). PD symptoms are caused by the loss of dopaminergic neurons in the substantia nigra and the accumulation of misfolded α-synuclein found in Lewy bodies (Bloem et al., 2021; Emamzadeh & Surguchov, 2018; Kehagia et al., 2013; Menza et al., 2010; A. H. Miller et al., 2013). PD is an aging-related disease, with typical onset around 65 years of age, and is characterized by progressive disability (Bloem et al., 2021; Fang et al., 2020). Symptoms range from tremors, loss of balance, bradykinesia, and cognitive deficits (Bloem et al., 2021; Isella et al., 2013; Menza et al., 2010; Verbaan et al., 2007). These cognitive deficits, present in 1 of every 4 patients, can manifest in various ways, including executive dysfunction, attention deficit, and lack of visuospatial awareness due to disruption of the frontostriatal pathway (Fang et al., 2020). Aberrant α-synuclein is a primary mediator of this cognitive impairment, as it is known to contribute to the disruption of homeostasis and perpetuate neuronal death. The effects of α-synuclein on cognition in PD patients has been recently reviewed (Aarsland et al., 2021). Comparatively, cytokines also have a substantial impact on the regulation of neurotransmitters and neural circuits, playing a key role in both the management of motor activity and cognitive dysfunction (A. H. Miller et al., 2013). Specifically, chronic secretion of pro-inflammatory cytokines by resident glial cells has been identified as a plausible catalyst for age-related, exacerbated inflammation associated with PD onset and its accompanying cognitive deficits (Chinta et al., 2013; Krizhanovsky et al., 2008; Mukhara et al., 2020; Thomas et al., 2021).
TNFα is elevated in the CSF and in post-mortem brain samples from PD patients and has been linked to cognitive deficits (Menza et al., 2010). In a double-blinded study observing 52 PD patients with clinically diagnosed depression, blood was drawn at 0 and 8 weeks and assayed for a panel of cytokines, including TNFα (Menza et al., 2010). Participants were screened for general cognitive function through an array of assessments, such as the MMSE, the Wechsler Adult Intelligence Test, and the Wechsler Memory Scale. Notably, increased levels of TNFα at the time of observation were correlated with decreased measures of cognition (Menza et al., 2010). Addenbrooke’s Cognitive Examination (ACE) is another useful test for quantifying cognitive ability in PD patients in which performance has been associated with TNFα levels. El-Kattan et al. found that serum levels of TNFα were significantly greater in PD patients who had significantly worse ACE visuospatial function and ACE cognitive scores compared to control subjects (El-Kattan et al., 2022). Clinically, TNFα inhibitors, such as etanercept, have been shown to effectively minimize PD incidence and improve cognitive test scores (Gagne & Power, 2010) as anti-TNFα therapy showed a 78% reduction in PD incidence when compared with patients given the current standard of care (Peter et al., 2018).
PD patients assessed for cognitive function using the MMSE also showed a negative correlation between blood concentration of IL-1β and cognition (El-Kattan et al., 2022). Although the specific mechanism by which IL-1β contributes to cognitive impairment has yet to be fully uncovered in the context of PD, it has been shown to have detrimental effects with respect to dopaminergic neuron loss when increased above physiological levels. This suggests that potential cognitive deficits in PD patients may be linked to an increase in neuronal cell death through excessive IL-1β production. Additionally, the synergistic effects of IL-1β with other cytokines have been demonstrated in animal models of PD (Carvey et al., 2005). Notably, Carvey et al. demonstrated that when TNFα and IL-1β are concomitantly injected into the median forebrain bundle of a rat model of PD, the resultant loss of dopaminergic neurons within the lateral substantia nigra was significantly greater than either cytokine injected alone. While cognition was not specifically assessed in this study, these data may suggest the need to consider interactions between cytokines in determining cognitive outcomes of PD patients.
The development of animal models for PD cognitive deficits is relatively recent. Unfortunately, no well-established PD animal models exist that also elicit cognitive deficits. While the classic models, 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) or 6-hydroxydopamine (6-OHDA) injection, are acute and largely mediate motor deficits, they are useful in analysis of TNFα and/or IL-1β levels in response to dopaminergic neuron loss. Currently, 3 new PD models exist that allow for the study of cognitive the α-synuclein preformed fibril aged mouse model (Iba et al., 2022), the gut-injected α-synuclein preformed fibril mouse model (Kim et al., 2019), and the humanized SNCA A53T mouse model (Paumier et al., 2013). However, these new models are still in development and require further research to validate cognitive deficits and PD-like pathology.
TBI occurs when there is an external force to the head which affects the functionality of the brain. In the United States, there are nearly 1.7 million cases of TBI that occur each year and it is one of the leading causes of death and disability worldwide (Faul & Coronado, 2015; Jassam et al., 2017; Nguyen et al., 2016). Primary mechanical insults during TBI result in secondary injury, which refers to the biochemical and inflammatory cascades that perpetuate further tissue damage (Bramlett & Dietrich, 2015; Kaur & Sharma, 2018). Secondary injury following primary TBI is associated with a variety of comorbidities, including cognitive impairment, that can vary and include deficits in speech, attention, concentration, as well as in learning and memory. In mild TBI (mTBI), cognitive impairment can typically be fully resolved in 3-6 months post injury for 80-85% of patients without brain lesions (Belanger & Vanderploeg, 2005); however, there are subsets of mTBI patients that experience extended cognitive deficits (McMahon et al., 2014). In contrast, 65% of patients with moderate to severe TBI report long-term cognitive issues (Whiteneck et al., 2004). Specifically with regard to inflammation, the overproduction of cytokines by resident glia contributes to adverse outcomes following TBI (Kumar & Loane, 2012; Woodcock & Morganti-Kossmann, 2013) and has been identified as a potential mechanism for cognitive dysfunction in TBI patients (Corps et al., 2015; Xu et al., 2022).
TNFα expression has also been studied in connection with cognitive function in TBI patients and was found to be elevated in patient serum 24 hours following injury (Goodman et al., 1990; Morganti-Kossman et al., 1997). TNFα transcript is also present in post-mortem TBI patient brains that succumbed as early as 17 minutes post-injury (Frugier et al., 2010). Notably, sustained TNFα levels in patient serum (between 6- and 12-months post-injury) are associated with decreased MoCA scores and have a positive correlation with impaired memory (Xu et al., 2022). Additionally, like IL-6, elevated TNFα levels in mTBI with loss of consciousness in military personnel also correlated with an increase in post-traumatic stress disorder symptoms, further demonstrating a connection between neuroinflammation and cognitive dysfunction (Kanefsky et al., 2019).
Various murine models of TBI have demonstrated TNFα-associated cognitive deficits following pharmacological inhibition or genetic deletion of TNFα. TNFα^−/−^ mice which underwent controlled cortical impact brain injury exhibited less severe memory deficit via the MWM test compared to wildtype controls (Scherbel et al., 1999). Inhibition of TNFα synthesis using 3,6′-dithiothalidomide 1 or 12 hours post-TBI fully prevented neuronal cell loss, gliosis, as well as improved spatial and short-term recognition memory assessed using the Y-Maze and novel object recognition tests, respectively, at day 7 post-TBI. Interestingly, the benefits of this treatment were not sustained when given 18 hours post-injury, after peak levels of TNFα had subsided, potentially suggesting a time-dependent efficacy of TNFα suppression (Baratz et al., 2015). Similarly, treatment of rats with etanercept has also been shown to reduce brain TNFα levels and attenuate both cognitive and motor deficits through stimulation of neurogenesis at day 7 post-TBI when administered every 12 hours for 3 days post-injury (Cheong et al., 2013). Likewise, selective inhibition of soluble TNFα using XPro1595 following controlled cortical impact injury improved learning and memory deficits as well as prevented loss of synaptic plasticity and neuronal degeneration in the mouse hippocampus (Larson et al., 2022). Collectively, these studies demonstrate a significant contribution of TNFα to both the development and maintenance of cognitive deficits in TBI.
IL-1β cytokine levels are elevated in acute TBI and affect long-term cognitive outcomes in patients. Serum IL-1β concentrations measured in patients with acute mTBI are associated with poor working memory as determined by the Digit Span Backward Task (Sun et al., 2019). Researchers observed a positive correlation between IL-1β levels and improvement in working memory 3 months post-injury potentially suggesting that low, chronic levels of IL-1β may be beneficial; however, they attributed this to patients having a lower cognitive function at baseline that allowed for increased improvement later in recovery (Sun et al., 2019). In adults with moderate to severe TBI, increased serum IL-1β was associated with poor outcome in memory scores using both the California Verbal Learning Test II-Long Delay Free Recall and the Rey-Osterrieth Complex Figure Delayed Recall Test 6 to 12 months post-TBI (Milleville et al., 2021).
In murine models, intracerebroventricular administration of an anti-IL-1β neutralizing antibody demonstrated improved visuospatial learning in the MMW 20 days post-injury, but was not effective at improving memory deficits when administered during acute stages of injury (48 hours post-TBI) (Clausen et al., 2011). Further, blockade of IL-1β via global IL-1R deletion in mice resulted in improved cognitive function using the Barnes Maze Test. These results were also confirmed through the administration of anakinra, a recombinant IL-1R antagonist, which led to similar memory improvements following TBI (Newell et al., 2018). In addition, inhibition of the NLRP3 inflammasome using the selective inhibitor MCC950 improved neurological function and cognitive outcomes through suppressed caspase-1 activity and IL-1β production post-TBI (Ismael et al., 2018; Xu et al., 2018). In all, these data suggest that IL-1β plays a critical role in perpetuating cognitive deficits following TBI, particularly in chronic stages.
IL-6 is a hallmark cytokine that is associated with inflammation-related secondary injury in TBI patients. It is one of the most widely studied cytokines in TBI and is utilized as a biomarker due to its abundant upregulation following initial trauma. Under normal conditions, blood plasma levels of IL-6 fluctuate between 0-42 pg/mL, while expression in the CSF is rarely detected (Kossmann et al., 1996; Maier et al., 2005). Following TBI, IL-6 levels in the CSF can increase to concentrations as high as 124,000 pg/mL (Hillman et al., 2007) and within the first 24 hours post-injury, increased plasma levels (>100 pg/mL) aid in predicting TBI severity and mortality (Woiciechowsky et al., 2002). Sustained serum levels of IL-6 have also been associated with cognitive deficits in TBI patients. A study conducted by Xu et al., demonstrated increased IL-6 concentrations in TBI patient serum which remained elevated 3 months post-injury. Patients assessed for cognitive function using the MoCA exhibited an inverse correlation between serum IL-6 levels and cognitive performance, suggesting a role for systemic inflammation as a predictor of cognitive deficits following TBI (Xu et al., 2022). Additionally, IL-6 trans-signaling via soluble IL-6 receptor has been linked with potentiation of neuroinflammation via activation of Janus kinase/signal transducer and activator of transcription (JAK/STAT) pathways in resident glia. This occurs through formation of a complex with the membrane receptor glycoprotein 130 (Campbell et al., 2014). IL-6 trans-signaling is also negatively associated with overall cognition and verbal fluency of TBI patients between 6- and 12-months post-injury (Milleville et al., 2021). Additionally, elevated serum IL-6 levels were observed in military personnel that have sustained TBIs. These levels correlated with an increased incidence in post-traumatic stress disorder and depression (Devoto et al., 2017), highlighting a prevalence of cognition-associated co-morbidities following TBI within select patient groups.
While a deleterious link between IL-6 levels and cognition has been reported, other studies suggest a beneficial role for IL-6 with respect to cognitive outcome. In murine models of TBI, compared to controls, IL-6-deficient mice exhibited impaired behavioral performance and decreased exploratory activity and rearing in the open field test, a measure of locomotion and anxiety-related behavior. They also had compromised performance on tasks assessing orientation to visual and olfactory cues and startle responses 24 hours post-injury (Ley et al., 2011). IL-6 trans-signaling in microglia has also been shown to be important for supporting the development of new neurons that directly impact cognitive function after TBI. Following pharmacological depletion of microglia using the colony stimulating factor-1 receptor antagonist PLX5622, repopulating microglia improved deficits in spatial learning after TBI via stimulation of neurogenesis in an IL-6-dependent manner (E. F. Willis et al., 2020). Moreover, elevated parenchymal IL-6 levels in TBI patients correlated with increased Glasgow Coma scores (Winter et al., 2004), a measure of consciousness and TBI severity in which higher scores indicate a more alert patient. This suggests a nuanced role for IL-6 that may be acutely neuroprotective but chronically detrimental.
MDD is a common, chronic, and complex psychiatric disorder that is characterized by mood disturbances such as feelings of hopelessness and decreased motivation, cognitive impairments such as concentration difficulties and memory loss, physical changes such as weight loss, and increased risk of suicide (Caraci et al., 2018). Current treatment strategies of MDD focus on neurotransmitter signaling and neglect an important component of MDD: immune dysregulation. Both peripheral and neuroinflammation are strong risk factors for the development and progression of MDD (Frodl & Amico, 2014; Miller et al., 2009). Cognition is an important link between inflammation and MDD, as many of the same pathways that are involved with inflammation-induced depression also affect neuronal integrity, leading to learning and memory impairments (Miller & Raison, 2016). MDD has a strong link to both acute and chronic inflammation through secretion of pro-inflammatory cytokines (Ali et al., 2020; Richardson et al., 2022). Though pro-inflammatory cytokines are involved in the immunopathophysiology of MDD (N. Kobayashi et al., 2022), the mechanisms for how they affect the cognitive deficits seen in MDD patients vary.
Pro-inflammatory cytokines are highly involved in the pathophysiology of MDD and have been associated with the cognitive impairments observed (Bourgognon & Cavanagh, 2020; Strawbridge et al., 2015). Proposed mechanisms for how these cytokines lead to depressive symptoms include alterations in neurotransmitter metabolism, neuroendocrine function, neuroplasticity, and neurocircuitry (Miller & Raison, 2016). Current research supports a bidirectional feedback loop perspective in which inflammation reduces monoamine availability, evoking neuroplastic changes that increase depressive symptoms and further increase inflammatory responses (Richardson et al., 2022). According to multiple studies and meta-analyses, IL-1β, IL-6, and TNFα are the most common pro-inflammatory biomarkers associated with the cognitive impairments seen in MDD (Dowlati et al., 2010; Milenkovic et al., 2019; Miller et al., 2009; Strawbridge et al., 2015).
IL-1β affects memory formation in the hippocampus by repressing LTP (Vereker et al., 2000), possibly by suppressing brain-derived neurotrophic factor (BDNF) signaling cascades (Tong et al., 2012). While there is evidence that low IL-1β concentrations support long-term learning and memory by promoting the production of neurotrophic factors (Bourgognon & Cavanagh, 2020), high amounts are detrimental to memory as measured by tests like the MWM (Bourgognon & Cavanagh, 2020). The specific effects on spatial navigation memory appear to vary based on the dosage and age of the test subject (Bourgognon & Cavanagh, 2020). Hippocampal expression of IL-1β is positively correlated with memory impairment in fear conditioning tests in mice (Cibelli et al., 2010), and elevated levels block LTP memory encoding processes by altering the maintenance of neuronal viability (Tong et al., 2012). IL-1β also reduces neural plasticity through NF-κβ signaling activation, glutamate excitotoxicity, and reduced levels of BDNF (Dantzer & Walker, 2014). Interestingly, IL-1β plays a critical role in the development of anxiety by increasing the recruitment of brain macrophages following social stress (Wohleb et al., 2013), suggesting a role for IL-1β in the modulation of immune cell localizing cues during stress. Furthermore, its plasma levels are negatively correlated with overall cognitive function on the MATRICS consensus cognitive battery standardized measurement in MDD patients (Jin et al., 2020). During MDD, there are elevated IL-1β plasma levels, but surprisingly, concentrations are higher in early onset compared to late onset MDD (Anzolin et al., 2022) suggesting a potential age-dependent effect of IL-1β on cognition during MDD.
IL-6 is also elevated in the serum of MDD patients with decreased hippocampal volumes compared to healthy controls. The authors concluded that IL-6 may be a predictor of hippocampal atrophy in MDD patients (Frodl et al., 2012). Patients with MDD had higher ratios of IL-6 compared to its anti-inflammatory regulator IL-10 (Dhabhar et al., 2009) which may suggest that there is significant cytokine dysregulation during MDD that contributes to pathology as this increase in IL-6 is associated with impairments in psychomotor speed, learning, and memory (Zhou et al., 2021). Importantly, IL-6 levels decreased in depressed patients after antidepressant treatment, suggesting that tempering IL-6 concentrations is critical in alleviating the symptoms of MDD (Strawbridge et al., 2015).
Recent evidence suggests that TNFα is also involved in depressive disorders as levels decreased in individuals with depression who were treatment-responsive (Strawbridge et al., 2015). TNFα, along with other pro-inflammatory cytokines, may contribute to depressive behaviors by increasing the reuptake of monoamine transmitters along with decreasing their availability which would inhibit neuron responses (Tran et al., 2021). Furthermore, TNFα can affect dopamine metabolism and reduce synaptic serotonin levels (Anzolin et al., 2022) suggesting that TNFα has a large role in mood modulation in depressed patients. Similar to IL-1β, TNFα levels were found to be higher in early onset MDD compared to late onset MDD (Anzolin et al., 2022), again suggesting a possible age-related mechanism in the pathophysiology of MDD.
While elevated pro-inflammatory cytokine serum levels are often seen in MDD patients who have impaired memory and learning processes (Worthen et al., 2020), anti-inflammatory cytokines also have a major role in disease. IL-10 is a hallmark anti-inflammatory cytokine involved in the pathophysiology of cognitive deficits in MDD. Anti-inflammatory cytokines such as IL-10 act as pro-inflammatory cytokine regulators (Dhabhar et al., 2009) and there is evidence to suggest that imbalances between various pro- and anti-inflammatory cytokine levels can increase disease symptomology and progression in psychiatric disorders like MDD (Gazal et al., 2015; Langhein et al., 2022). IL-10 has not only been found to protect against cognitive loss, but also reverse learning and memory impairments induced after learned helplessness in mice (Worthen et al., 2020). Interestingly, decreased levels of anti-inflammatory cytokines are specifically seen in treatment-resistant depression (Caraci et al., 2018). Studies have found differing levels of IL-10 expression in MDD patients, however (Dhabhar et al., 2009; Zou et al., 2018). It is hypothesized that increased IL-10 may only be present in patients who have developed MDD later in life (Gazal et al., 2015) or in patients who simultaneously have higher levels of pro-inflammatory IL-8 (Langhein et al., 2022). Moreover, In MS patients with comorbid depression, depressive symptoms have an inverse relationship with plasma IL-10: as IL-10 concentrations decreased, patients exhibited worse depressive symptoms (do Sacramento et al., 2022), further highlighting a protective role for IL-10.
Other cytokines with anti-inflammatory properties involved in MDD include TGFβ and BDNF. Not only were plasma levels of TGFβ found to be reduced in MDD patients, but the reduction correlated with both depression severity and treatment resistance (Caraci et al., 2018). The study also found that TGFβ deficiency contributed to the progression of mild cognitive impairment in long-lasting depression (Caraci et al., 2018). BDNF is a member of the neurotrophin growth factor family and has anti-inflammatory properties that have been associated with verbal memory retention in MDD (Grassi-Oliveira et al., 2008; Zhao et al., 2022). It is thought that the antidepressant characteristics of BDNF are due to its involvement in increasing hippocampal neurogenesis, a vital process for synaptic plasticity (Zhao et al., 2022). Intriguingly, compared to healthy control subjects, MDD patients were found to have lower plasma levels of BDNF, which negatively correlated with memory encoding processes, particularly those involving short-term verbal memory (Grassi-Oliveira et al., 2008). Furthermore, supplementing BDNF in the hippocampus showed benefit in treating MDD (Zhao et al., 2022). Together, these studies suggest that deficiencies in anti-inflammatory cytokines like IL-10, TGFβ, and BDNF likely contribute to the cognitive decline seen in long-lasting depression and MDD.
Advances in healthcare have extended human life expectancy, increasing the proportion of geriatric individuals in society. Despite delays in mortality, aging continues to be a strong risk factor for cognitive decline, even in the absence of active disease (Legdeur et al., 2018; Murman, 2015; Nebes et al., 2013; Ofori-Asenso et al., 2019). It is estimated that only 1 in 1000 geriatric individuals will not develop some form of mild cognitive impairment (Petersen, 2011). Understanding the age-dependent factors that contribute to cognitive decline is essential to develop effective prophylactics for those at risk of pre-clinical dementia. A possible hypothesis for the relationship between aging and cognitive decline is inflammaging, a phenomenon that describes an age-related increase in low-grade, sterile, chronic inflammation (Franceschi et al., 2000; Wyss-Coray, 2016). One clinical marker of inflammaging is the elevation of pro-inflammatory cytokines not only in circulation (Marsland et al., 2015) but importantly, in brain regions responsible for cognition (Koelman et al., 2019; Porcher et al., 2021; Wyss-Coray, 2016). While chronic inflammation has previously been associated with brain atrophy (Gunning-Dixon et al., 2009; Markov et al., 2022; Walker et al., 2017), impaired neurogenesis (Chesnokova et al., 2016; Ekdahl et al., 2003), and neurodegeneration (Mohamed et al., 2023; Walker, 2018), the mechanisms for how specific cytokines contribute to cognitive decline during normal aging is currently unclear.
Cytokines play a significant role in regulating various aspects of cognitive development including neurogenesis, synapse formation, and synaptic plasticity. To understand how changes in cytokines can lead to cognitive decline, it is important to describe their role in normal cognitive development. Post-natal murine NPCs express IL-6 and IL-6 receptor to self-regulate proliferation and self-renewal (Storer et al., 2018) and Il6^−/−^ mice have a decreased number of proliferating NPCs (Bowen et al., 2011). IL-1β has been shown to promote NPC differentiation and growth (Park et al., 2018) while TNFα has been shown to inhibit these processes (Lan et al., 2012). IL-6, but not IL-1β or TNFα, has been shown to induce transcriptional changes in fetal hippocampal neurons related to synaptogenesis (Mirabella et al., 2021). Balanced cytokine production and signaling are crucial for maintaining proper cognitive development.
Adult neurogenesis occurs in the hippocampal dentate gyrus and is key for memory processing (Aimone et al., 2006; Deng et al., 2010; Valero et al., 2017; Van Praag et al., 2002). Decline of hippocampal neurogenesis is normal with age but can be greatly accelerated by neuroinflammation driven by astrocytes and microglia (Valero et al., 2017). In the normal aging brain, astrocytes and microglia undergo a senescence-associated secretory phenotype (Barrientos et al., 2015; Clarke et al., 2018; Holtman et al., 2015; Soreq et al., 2017) specifically within the hippocampus (Matias et al., 2022; Suda et al., 2021). Both aged astrocytes and microglia have increased expression of IL-6 (Robinson et al., 2020), IL-1β (Ritzel et al., 2019; C. M. Willis et al., 2020), and TNFα (Mariani et al., 2006). Serum IL-6 is negatively associated with executive function and processing speed, but had no association with verbal episodic memory (Tegeler et al., 2016). High levels of serum IL-6 were predictive of psychomotor speed decline (Palta et al., 2015) indicating that IL-6 may signal to and affect neurons in a region-dependent manner.
One possible link between glial reactivity and cognition is through BBB breakdown. Glial cells, along with endothelial cells, maintain CNS homeostasis by surrounding the neurovasculature (Smith et al., 2022). BBB leakage is positively associated with age and cognitive impairment (Hussain et al., 2021; Verheggen et al., 2020). Likewise, the choroid plexus of the blood-CSF barrier exhibits age-associated degeneration (Serot et al., 2003). The choroid plexus in aged mice has reduced CSF production (Masseguin et al., 2005) and a type I IFN responsive transcriptional profile (Baruch et al., 2014). Type I IFNα/β is ubiquitously expressed and is typically upregulated in response to innate recognition of pathogens, though type I IFNs are occasionally secreted during sterile inflammation such as during aging. Injection of IFNβ in aged naïve mice decreased memory recall and spatial learning whereas transgenic mice with either a global (Ifnar1^−/−^) or brain endothelium-specific (Ifnar1^fl/fl^Slco1c1Cre^ERT2^) deletion of the IFNα/β receptor were protected from these behavioral changes (Blank et al., 2016). This suggests that IFNα/β signaling plays an important role in behavior, potentially through regulation of the BBB. Hyper-responsiveness to IFNα/β signaling has been reported by others (Baruch et al., 2014; Cao, 2022; Deczkowska et al., 2016) and is a possible link between normal aging and cognitive decline. Interestingly, this contrasts with the protective effects of IFNβ used to treat MS patients.
A consequence of BBB dysfunction is the infiltration of adaptive immune T and B cells, significant sources of cytokines, into the CNS. Interestingly, Rag1^−/−^ mice, which lack T and B lymphocytes exhibit spatial learning and memory deficits. This cognitive dysfunction is rescued by the adoptive transfer of CD4^+^ T cells from naïve transgenic mice in which the T cell receptors are specific to a self-antigen, myelin oligodendrocyte glycoprotein (Radjavi et al., 2014). Chronic antigenic exposure over time exhausts naïve CD4^+^ T cells and results in the accumulation of memory and effector CD4^+^ T cell subsets primarily responsible for the secretion of cytokines like IL-4, IFNγ, and IL-10. IL-4 producing Th2 cells accumulate in the meninges in response to memory and learning tasks (Derecki et al., 2010; Gadani et al., 2012). Hippocampal concentrations of IL-4 decreases with age and is associated with impairments in synaptic plasticity (Nolan et al., 2005). IL-4 maintains the balance of T helper type 1 (Th1) and T helper type 2 (Th2) cell responses by suppressing the differentiation of IFNγ-producing Th1 cells. IFNγ is elevated in the brain during social behavior (Filiano et al., 2016) and is associated with slower cognitive decline in the absence of disease (Yang et al., 2022). In contrast, mice with a constitutive deletion of IFNγ had enhanced cognitive performance and increased hippocampal neurogenesis (Monteiro et al., 2016). While IFNγ is canonically considered a pro-inflammatory cytokine, whether IFNγ is deleterious or protective is likely contextual. Similarly, the anti-inflammatory cytokine IL-10, which is secreted by T regulatory cells to suppress inflammation, can have varying effects. For instance, overexpression of IL-10 in a young adult transgenic mouse model resulted in spatial learning and memory deficits similar to those in normally aged mice (Sanchez-Molina et al., 2022). While the secretion of cytokines by both immune and resident brain cells has a clear connection to cognition, further studies are necessary to determine whether the levels of specific cytokines are causative for normal aging-associated cognitive decline.
Without pharmacological interventions, current strategies to ameliorate normal aging-associated cognitive decline are to reduce systemic inflammation through life-style modifications (Muscat & Barrientos, 2020) or to strengthen cognition through programs like perceptual training (Berry et al., 2010). To develop specific treatments for cognitive decline, the use of animal models is essential in understanding the physiological mechanisms underlying aging and cognitive dysfunction. Mouse and non-human primate models are advantageous as they resemble human aging but are resistant to developing age-related cognition deficits (Folgueras et al., 2018; Vanhunsel et al., 2020; Wang et al., 2011).
In addition to animal models, correlative studies of cognitively resilient populations can provide insight to physiological targets to prevent or reverse cognitive decline. Soluble TNF receptor I and II (sTNF-RI and -RII) are anti-inflammatory through inhibition of TNFα, similar to the effects of etanercept in mouse models of AD and TBI. Both sTNF-RI and sTNF-RII were found to be elevated in healthy aged and centenarian populations, a group of individuals that have reached 100 years of age and may be a model for healthy aging (Franceschi & Bonafè, 2003). One study reported that hematopoietic cells derived from centenarians had decreased production of granulocyte-macrophage colony-stimulating factor and IL-3 (Bagnara et al., 2000). Increased expression of IL-4 and IL-10 alongside decreased expression of IL-2 and IFNγ in centenarians could contribute to their resistance to significant cognitive deficits (Cossarizza et al., 1996; Pawelec et al., 2002). While centenarians have lower rates of cognitive decline, analysis of centenarian post-mortem tissue shows varying levels of neuropathology (Beker et al., 2021). If neuropathological changes in brain morphology occur in individuals with and without cognitive decline, altered cytokine profiles in centenarians may be important for resistance to the development of cognition-associated dysfunction.
Each of the disorders discussed above have significantly different etiologies; however, all share common cognitive deficits. MS, AD, FTLD, PD, TBI, and MDD patients all experience recall deficits while MS and AD patients frequently have comorbid depressive disorders (Feinstein et al., 2014; Taragano et al., 2009). One common thread among all of these disorders is a dysregulation of cytokine and chemokine release, detected both systemically and within the CNS. These cytokines and chemokines may act directly on neurons by modulating activity or even causing death, or indirectly through glia-glia or immune-glia communication. This glia-glia and immune-glia communication can lead to cascades of cytokine. The impact of each cytokine on cognitive deficits is likely context-dependent as IL-6 and IFNs can exert either beneficial or detrimental effects, while IL-1β and IL-17 are developmentally necessary (Figure 2).
The disorders described here include a role for TNFα as a strong contributor to cognitive deficits, both in human patients and in animal models. TNFα is a highly pleiotropic cytokine, however, as recently reviewed (Gough & Myles, 2020); as such, simply blocking TNFα may not be a viable therapeutic approach. In mouse models of MS, for instance, blocking all TNFα receptor activity is detrimental, as TNF receptor 2 is beneficial for recovery and remyelination (Fiedler et al., 2023; Patel et al., 2012; Pegoretti et al., 2023). Thus, more investigation into the cell-specific effects of signaling by specific TNFα receptors is needed.
Other cytokines, such as IL-6 and IL-1β, have more nuanced roles, given their apparent requirements for proper development and temporal differences in effect on cognition during disease. These cytokines are both associated with impaired working memory in many disorders; however, there appears to be a temporal component, as observed in TBI where IL-6 may be acutely neuroprotective. Likewise, IL-1β production may have an undiscovered physiological—rather than purely maladaptive—role, as its transcription and secretion are tightly regulated particularly during development. Like aging, in which IL-6 and IL-1β promote NPC proliferation and differentiation, these cytokines may promote the survival of NPCs in disease at physiological, rather than pathological, concentrations as well. A better understanding of neuroimmune signaling resulting from these cytokines is critical to move closer to translational benefit. Currently, meta-analyses investigating the contribution of these inflammatory biomarkers and cognitive deficits is lacking within these diseases and across many pathologies in general. Completion of such an analysis would strengthen the link between these inflammatory factors and cognitive decline.
Other discussed cytokines and chemokines, such as those in the CCL and CXCL families, may be more specific to individual diseases rather than general neuroinflammation. However, it is possible that chemokines have a yet undefined role in cognitive decline. There is limited literature regarding the CCL family and its role in cognitive deficits in either patients or animal models, though there is evidence for its impact in FTLD. Complement, however, is heavily implicated in multiple disorders, including MS (Ramaglia et al., 2021), AD (Hong et al., 2016; Shi et al., 2017), FTLD (Lui et al., 2016), TBI (Mallah et al., 2021), and even developmental disorders such as autism (Odell et al., 2005). Overall, this complex milieu of secreted molecules and their downstream signaling components may represent promising targets for the treatment of cognitive dysfunction.