Neuroprotective versus Neuroinflammatory Roles of Complement: From Development to Disease

Abstract

Complement proteins are ancient components of innate immunity that have emerged as critical regulators of neural networks. In this review, we discuss these roles in the context of CNS development, acute viral infections of the central nervous system (CNS), and post-infectious and noninfectious CNS disorders, with an emphasis on microglial-mediated loss of synapses. Despite extensive examples that implicate classical complement proteins and their receptors in CNS dysfunction, recent data suggest they exert neuroprotective roles in CNS homeostasis through continued refinement of synaptic connections. Thorough understanding of the mechanisms involved in these processes may lead to novel targets for the treatment of CNS diseases involving aberrant complement-mediated synapse loss.

Neuroinflammation is Multifaceted

Neuroinflammation, especially the expression of innate immune molecules, has been historically defined as an inflammatory response within the brain or spinal cord that is associated with maladaptive pathological outcomes [1]. However, there is increasing evidence that neuroinflammatory responses participate in many homeostatic and neuroprotective processes involved in normal brain function [2–4]. Protective versus pathological outcomes are dependent on factors such as disease state, levels and locations of inflammatory mediators, and the temporality of the process (acute vs chronic) [5–6]. Innate immune responses induce the expression of pro-inflammatory cytokines, such as chemoattractant signals that recruit peripheral leukocytes or activate resident central nervous system (CNS) immune cells. There is growing evidence that innate immune molecules also impact the function of neural networks via synapse pruning during development or synapse elimination during disease. In this review we discuss the role of classical complement proteins, innate immune molecules expressed within the CNS, in the context of homeostatic, neurodegenerative, and neuroinfectious processes, as well as potential links between viral and non-infectious processes. We will discuss recent lines of evidence that identify sources of CNS complement, complement signaling in glial interactions, and complement mediated synaptic pruning/elimination.

Complement is an Ancient Component of the Neuroimmune system

The complement system consists of a number of proteins that normally circulate as inactive precursors and can be cleaved into smaller activated fragments. This system is composed of more than 50 different plasma proteins that participate in three major pathways: The lectin pathway, the classical pathway and the alternative pathway, which each, through initiation by sugar moieties, antibody complexes, or spontaneous activity (respectively), leads to the formation of active protein components [7–8] (Fig 1). Classical complement proteins are found in invertebrates where they contribute to the opsonization of pathogens, induce chemotaxis of immune cells, and contribute to the function and repair of the nervous system [9–10]. This latter function indicates that CNS complement evolved as an essential component of the normal nervous system. In mammals, complement proteins are classically thought of as peripheral proteins, produced in the liver, circulating in the bloodstream, and activated upon challenge by pathogens. The classical pathway is initiated by complement component C1q (and associated proteases, C1r and C1s), which is composed of six identical subunits with globular heads and acts as an initial pathogen sensor. C1q can directly bind pathogens or bind to antibody-antigen complexes, leading to the enzymatic activity of C1r and C1s [11]. This enzymatic activity leads to the cleavage of the C4 protein into the additional active products C4b and C4a, in which C4b functions as an opsonin. C4b covalently binds to pathogens targeted for recognition and is necessary to induce the formation of the C3 convertase, which is also known as C4b2b. Activation of the classical pathway through C1q also leads to C4b2b-mediated-cleavage of C3 into active products C3b and C3a. C3b also functions to opsonize pathogens and leads to further downstream enzymatic activity. Specifically, C3b can initiate C5 convertase (C4b2b3b), leading to cleavage of C5 into active components C5a and C5b. C5b combines with proteins C6–9 to ultimately produce the formation of a pore forming membrane attack complex (MAC) that promotes lysis of pathogen-infected cells. C3a and C5a are anaphylatoxins and strong chemoattractants for immune cells, particularly macrophages and monocytes. Attraction of these cells helps to eliminate pathogens via phagocytosis, as well as enhance the humoral response, through initiating proteins such as C1q binding directly to antibodies [11]. Through these various mechanisms, complement promotes acute inflammatory responses, including increases in vascular permeability, which together promote the extravasation of leukocytes into peripheral tissues where they lyse or phagocytose infected cells [12]. Complement receptors mediate chemotaxis of immune cells to sites of infection and injury. These include complement receptor type 1 and type 3 (CR1 and CR3), whose activation also counter regulates the complement system. Thus, CR1 activation inhibits C3 convertase activity and converts complement proteins into inactive fragments [7–8]. Complement component 3a and 5a receptors (C3aR/C5aR) are G protein coupled receptors that are generally expressed by antigen presenting cells. Historically, the antimicrobial, autoimmune and other immune actions of complement proteins were believed to be their primary functions.

Figure 1: Complement Pathways.

The complement system is divided into three pathways: Classical, lectin, and alternative. The classical pathway is initiated by antigen-antibody complexes by C1qrs. The lectin pathway is initiated by mannose binding lectin (MBL) binding to sugar monomers (mannose) on pathogen surfaces. The alternative pathway is initiated by the spontaneous hydrolysis of C3 into active components. Through a series of enzymatic steps that produces the C3 convertase (C4b2b or C3bBb), each pathway converges to the cleavage of C3 into active components C3a and C3b. C3b then produces C5 convertases (C4b2b3b or C3bBb3b), which cleaves C5 into active components C5a, and C5b. Through the combination of C5b-C9 the membrane attack complex is formed and is able to produce pores in targeted cells or pathogens, resulting in lysis. Furthermore, anaphylatoxins (C3a and C5a) function as chemoattractants and recruit immune cells to sites of injury.

Since the foundational discoveries of complement proteins regulating synapse elimination during CNS development [13–15], the complement cascade has been found to function in homeostatic and pathological neurological processes in mammals, just as observed in invertebrates. Thus, complement proteins have been implicated in the maintenance or disruption of neural networks, and in processes involving network reorganization such as normal forgetting [16], as well as several neurological and psychiatric diseases including traumatic brain injury, Alzheimer’s disease, post-viral encephalitis cognitive dysfunction, and schizophrenia [17–20]. The upregulation of classical complement protein members during CNS disorders suggests they are involved in generalizable mechanisms that contribute to pathological forgetting and other forms of cognitive impairment. Importantly, while the majority of complement proteins are produced in the liver, various complement proteins are also made within the normal and diseased CNS. For example, while C1q and CR3 are expressed throughout life in the healthy CNS, the C3a receptor, C3aR, is expressed by astrocytes, microglia and infiltrating immune cells during neuroinflammation [21–26]. Altogether, these lines of evidence indicate that complement activation is vital for normal CNS function, but may also drive CNS disease states.

Homeostatic Roles for Complement in the CNS

There is increasing evidence of immune components produced within the CNS that contribute to its development and maintenance. Neuroimmune interactions are also known to play key roles in repair following injury and CNS disorders. These various roles have led to intense investigation of the localization and temporal patterns of neuroimmune modulators in both homeostatic and disease states. In this section we will discuss emerging evidence of the role of complement in driving homeostatic processes during development and aging.

Complement in the Developing CNS

Sculpting of neural networks during development is partly driven by synaptic pruning in response to differential activation of neural networks [13–14, 28, 30]. Several studies have demonstrated that complement proteins critically mediate synaptic pruning in a process that involves interactions between neurons, microglia and astrocytes and results in the removal of inappropriate synapses and selective maintenance of synaptic connections that serve the eventual circuit’s function. For instance, using a mouse model of visual development, C1q and C3 were demonstrated to be required for synaptic pruning by providing phagocytic signals to microglia [13–14]. C1q, the initiator protein for the classical complement cascade, and C3 are localized to developing synapses in the visual system, while CR3 is localized to microglia during visual system development. C1q is upregulated in retinal ganglion cells (RGC) in response to TGF-β secreted from astrocytes [13, 31]. C1q and C3 both tag inactive retinogeniculate synapses, and CR3-expressing microglia engulf these synaptic elements via a process dependent on C3-CR3 signaling [14]. Non-engulfed synapses may be tagged with CD47, a self-associated molecular pattern transmembrane protein. CD47 is enriched in synapses in the dorsal lateral geniculate nucleus during peak pruning periods, and its receptor, SIRPα, is upregulated on microglia during these periods [32]. Both are required for neuroprotection and reducing engulfment of synapses [32]. Similar studies focused on the developing spinal motor circuit show that C1q and C3 also localize to VGlut+ excitatory synapses [30]. C1q tagged synapses are pruned by microglia, suggesting complement plays a role in aspects of CNS development outside of the visual system [30]. While these studies demonstrate that C1q is vital for formation of appropriate neural networks during early stages of development, studies have shown this does not occur during later stages of development. Specifically, deleting C1q does not alter experience-dependent synaptic plasticity in the binocular zone of the primary visual cortex during a later critical period when ocular dominance occurs [33]. These studies suggest that the role of complement proteins during synaptic pruning is dependent on brain region and developmental stage.

Importantly, mice deficient in C1q exhibit increased synaptic connectivity and display atypical seizure behaviors, further suggesting the molecule is necessary to maintain normal synapse numbers in the CNS [34]. Complement mediated alterations in synaptic connectivity during development may underlie psychiatric disorders in which genetic alterations in complement genes have been identified. Genetic variation of the C4b allele is linked to schizophrenia and autism spectrum disorders [18, 35–36]. Schizophrenia risk-associated variants within the human complement component 4 locus are associated with increased neuronal complement deposition and synapse uptake in induced microglial-like cells in vitro [37], suggesting increases in phagocytic elimination of synapses. C4 mediates synapse elimination during development in a microglial dependent manner, and overexpression of C4 results in hypoconnectivity of cortical regions [38]. In sum, classical complement proteins are found on different CNS cell types during development and, by mediating interactions between glial cells and neurons, play a vital role in shaping neural circuits during development.

Complement proteins are also involved in developmental processes outside of synaptic pruning. Furthermore, complement proteins are produced in the CNS at time points before liver development, suggesting these proteins have roles in neurodevelopment outside of immunity [15]. C5aR1 is expressed on mouse embryonic neural progenitor cells and human embryonic stem cell derived neural progenitors [39]. Furthermore, addition of C5a in utero increased the number of apical progenitor cells, whereas blockade of the C5a-C5aR1 axis in utero resulted in alterations in behavior and microstructural changes in the CNS of adult mice, indicating these developmental changes have long term consequences [39]. The C3a-C3aR axis has also been shown to participate in neural crest cell migration in a zebrafish neurodevelopmental model [40]. Relevant to both anaphylatoxins, in utero inhibition of the classical complement pathway via use of the C1 inhibitor, Serping1, leads to deficits in neuronal migration that was rescued by addition of agonists to both C3aR and C5aR (C3aR agonist alone had minimal effect) [41]. This demonstrates cooperativity between both signaling pathways in ensuring proper neuronal development.

Complement in the Aging CNS

As in development, there is increasing evidence that classical complement signaling mediates neuroinflammatory and neuronal network changes during aging. During aging, increased C1q/C3 deposition and increased levels of C1q are found at synapses in the cortex and hippocampus [42–44], with microglia as the dominant source of C1q [45]. Using ultrastructural approaches researchers have found C1q within glial processes and near synapses, particularly associated with the post-synaptic density protein (PSD) 95 in aged non-human primates [42], suggesting that age-related changes in C1q display specific intra-neuronal distribution patterns. Somewhat relatedly, loss of C3 in mice attenuates age related hippocampal decline (i.e. synapse and neuronal loss, deficits in spatial learning and memory) [44]. Unfortunately, there are limited data on cell specific detection of activated complement proteins in the human CNS, due to lack of reliable antibodies, which are also needed to distinguish similarities and differences in complement production between human and murine studies.

Complement proteins also mediate glia-neuronal interactions that impact neural networks during aging. For example, neurotoxic reactive astrocytes, which produce complement components and lose some of their normal functions were described in aging mice [46]. Further studies are needed for determining the similarities and differences between complement activation during development and in aging, as well as assess other homeostatic roles complement plays during aging.

Increased Levels of Complement are Found in Neuroinflammatory Diseases

Complement in the Neurodegenerating CNS

A hallmark of neurodegenerative diseases is activation of resident immune and glial cells with infiltration of leukocytes from the periphery [1]. In this section we will discuss evidence that implicates classical complement proteins in these neuroinflammatory processes, including effects on glial communication, and synaptic elimination. Complement expression and activation increases in CNS disease states, including in neurodegenerative diseases such as Alzheimer’s disease (AD) [17, 47–49], and other tauopathies [50–51], as defined by the presence of abnormal tau proteins in the CNS [52]. Complement (C1q, C3, C4) deposition in AD senile plaques was documented decades ago using immunoperoxidase techniques [53]. Complement activation by β-amyloid and increases in complement secretion by microglia in CNS tissues of AD patients were also demonstrated over 20 years ago [54–56]. More recent studies have provided more cell specific and mechanistic insights into the role of complement in neurodegenerative disease. In AD and other tauopathies, C1q colocalizes with hippocampal PSD95 [17,50], C3 is expressed by astrocytes and microglia [47–48, 51, 57], and C3R, a complement protein receptor typically expressed in neuroinflammatory states, by microglia [48–51]. Current distinctions between these disorders include the findings that C4 is expressed by oligodendrocytes [47] and CR1 by astrocytes in AD [49], which has not been observed in non-AD tauopathies. Expression of CR1 in the brain is controversial, as detection has varied between studies, which may be a result of antibody effectiveness [49, 58]. Further studies in tauopathy models are needed to determine whether these impact clinical differences between these diseases, which would provide more insight into the pathology of the diseases, as well as offer putative disease-specific therapeutic targets.

As in development, glia-neuronal interactions during neurodegenerative disease can impact neuroinflammation and disease outcomes. Neurotoxic astrocytes that upregulate genes involved in antigen presentation [46, 59] are implicated in neurodegenerative diseases [46, 60]. Studies in mice indicate that C1q secreted from microglia can also induce an inflammatory phenotype in astrocytes [46] that may contribute to astrocyte derived C3, which, via C3R signaling in microglial cells, leads to increased microglial phagocytosis in AD [48]. C3-C3R signaling between astrocytes and neurons in AD also disrupts dendritic morphology [61]. C5a-C5aR signaling also plays a prominent role in neuroinflammation during CNS injury. In a mouse model of spinal cord injury, C5aR1KO mice exhibited significantly improved motor function early following SCI, but this recovery was not sustained and resulted in poorer long-term outcomes [62]. However, administration of a C5aR1 antagonist acutely led to improved outcomes sustained throughout chronic phases of injury [62]. C5aRKO mice also had decreased levels of proinflammatory cytokines and inflammatory macrophages during the acute SCI phase. Treatment of Tg2576 and 3xTg mice (models of AD) with a C5aR inhibitor (PMX205) resulted in decreased fibrillar and total β-amyloid deposition, and lower detection of hyperphosphorylated tau [63]. Loss of C5aR signaling also decreases CCR2+ monocyte accumulation around plaques and reactivity of CD45 and GFAP positive cells in the cortex and hippocampus [63–64]. Deletion of C3 in various mouse models of β-amyloid deposition and tauopathy (i.e. APP/PS1 and PS2APP) also confers protection from glial activation, synapse loss and downstream cognitive effects [57, 65]. Interestingly C3 deletion also increased plaque load in these studies, suggesting the resulting complement meditated inflammatory response to plaques plays a more important role than plaque accumulation [57, 65]. In a mouse model of amyotrophic lateral sclerosis, pharmacological blockage of C5aR led to improved hindlimb grip strength and slower disease progression, providing further evidence to the role of C5a-C5aR signaling in multiple neurodegenerative diseases [66].

Increased microglia-mediated synapse elimination has also been observed in AD and other tauopathies [17, 50–51], and has been previously reviewed [67]. Expression of classical complement proteins that influence synaptic elimination during neurodegeneration appear to be mediated in part by mitochondrial stress, apoptosis, and cytokines. A recent study demonstrated that C1q tagged synaptosomes from APP/PS1 mice, which overexpress β-amyloid, had elevated levels of mitochondrial superoxide and accumulations of septin proteins (Sept3), which are normally involved in synaptic vesicle exocytosis [68]. Increases mitochondrial superoxide suggest leakage from the mitochondria and/or mitochondrial dysfunction, which may trigger complement activity. Studies also suggest apoptotic processes may lead to tagging of synapses via complement. Using proteomics, researchers demonstrated that C1q tagged synaptosomes are upregulated in apoptotic proteins, and C1q colocalized with Caspase 3 in synapses [69]. Altogether these studies suggest indicators of cellular damage may mediate complement expression. Other CNS secreted proteins such as progranulin (a mediator of inflammation) are also potential upstream modulators of C1q and classical complement activation [70].

Cytokine release from resident immune cells may also induce neuronal damage, promote gliosis, and enhance activation of infiltrating immune cells, which all increase complement production in the CNS [71]. Intracranial inoculation of type I interferons (IFN) in a murine model of AD increased C3 activation in vivo [72]. Increased mouse IFNγ expression via a recombinant adenovirus vector in AD TgCRND8 mice (transgenic mice over-expressing human amyloid precursor protein) was also shown to underlie increased C3 in the hippocampus and result in suppressed amyloid deposition and increased microgliosis and astrogliosis [73]. Future studies will help determine whether cytokine mediated complement expression leads directly to pathological changes in AD, or simply adds to the overall inflammatory milieu.

Complement in CNS Viral Infections

Neurotropic viruses continue to emerge and spread worldwide, leading to increased cases of acute infectious and autoimmune diseases of the CNS, and post-infectious neurologic sequelae [74]. Infection with neurotropic viruses can result in severe neuroinvasive syndromes that manifest as meningitis, encephalitis, or acute paralysis depending on the type of virus and/or CNS site of infection [74–75]. As mentioned above, via an efficient cascade of enzymatic steps, the complement system can identify, bind, and lyse infected cells outside the CNS (Fig 1). This pattern holds true for a myriad of viruses, where the complement system has been shown to enhance antibody mediated viral clearance in the periphery and limit neuroinvasion [76–77]. To illustrate these processes, we will briefly describe the roles of complement in peripheral virologic control during infections with encephalitic arboviruses that cycle between vertebrate reservoirs and invertebrate vectors of mammalian diseases.

Arboviral members of the Togaviridae and Flaviviridae families of viruses are enveloped viruses consisting of a positive, single-strand RNA genome. The encephalitic members of the Togaviridae family include alphaviruses Eastern, Western, and Venezuelan equine encephalitic viruses (EEEV, WEEV, and VEEV, respectively). The epidemiology and neuropathogenesis of encephalitic alphaviruses has been previously reviewed [74]. Encephalitic alphaviruses naturally cycle between mosquitoes and birds (EEEV and WEEV), mosquitoes and rodents (VEEV enzootic cycle), mosquitoes and horses (VEEV epizootic cycle), or between humans and mosquitoes (VEEV, epidemic infections). Peripheral inoculation leads to an initial round of replication in the skin, followed by dissemination into the blood and other peripheral organs, similar to other encephalitic arboviruses. Shortly after infection, VEEV may enter the CNS, leading to extensive neuronal damage, infiltrating of mononuclear cells, and resulting neurological sequelae. The complement system exerts peripheral virologic control that may limit neuroinvasion. C3−/− mice peripherally inoculated with a mutant version of VEEV exhibit delayed clearance of infectious virions in the serum, more rapid invasion of the CNS and more severe encephalitis [76]. Importantly, C3−/− mice intracranially inoculated with the mutant strain had similar weight loss and disease score to WT controls, suggesting the complement system may mainly function to influence disease outcomes in the periphery that limit VEEV infection of the CNS.

The encephalitic members of the Flaviviridae family include Japanese encephalitis virus (JEV), Tick-borne encephalitis virus (TBEV), West Nile virus (WNV), and Zika virus (ZIKV). The epidemiology and neuropathogenesis of encephalitic flaviviruses has also been previously reviewed [74]. Encephalitic flaviviruses circulate between birds and arthropod vectors (mosquitoes and ticks), which may directly infect mammals [78]. WNV is the leading cause of domestically acquired arboviral disease in the United States [75]. The virus enters permissive cells such as keratinocytes, dendritic cells, and neurons via clathrin-mediated endocytosis, and replicates in the cytoplasm [79]. The virus can then spread via exocytosis or cell lysis. Innate immune cells such as macrophages and humoral immune responses are important for controlling peripheral viral replication, while cell-mediated immunity is critical for viral clearance within the CNS [80–81]. Within the CNS, neurons are the main cellular target of the virus [82]. Similar to alphaviruses, complement activation is required to control peripheral viral replication and limit spread to the CNS, mainly via induction of protective antiviral antibody responses. C3−/− mice exhibit increased peripheral viral titers and decreased WNV specific antibodies after footpad inoculation with WNV compared to similarly infected wildtype animals [77]. Early neuroinvasion and increased viral titers in the CNS of C3−/− mice was also demonstrated, again suggesting C3 is necessary to decrease neuroinvasion of virus via antibody-mediated clearance in the periphery [77]. In the case of ZIKV infection, C1q binds envelope proteins and non-structural proteins, and lysis of virion occurs via MAC formation [83]. Although studies indicate that complement proteins are increased in the CNS of mice during WNV and ZIKV encephalitis [20, 84] their role in virologic control or spread within the brain has not been extensively studied.

Viral Infections and Complement-mediated Pathological Forgetting

As in neurodegenerative processes, complement activation is implicated in mechanisms of neuroinflammation and synaptic elimination following neurotropic viral infection. Complement mediated antiviral immunity minimizes viral spread from the periphery to the CNS and limits the overall numbers and level of activation of infiltrating immune cells and inflammation-amplifying resident neural cells, such as microglia and astrocytes, upon neuroinvasion. However, increases in complement activation in the CNS may also lead to neurologic sequelae, including cognitive dysfunction. Recent studies have shown that certain viral infections, including DNA viruses, may be risk factors for pathology associated with various neurodegenerative diseases including AD, although a relationship with complement-mediated memory dysfunction has not been established [85–88].

Neurotropic RNA viruses are associated with certain hallmarks of neurodegeneration, such as deficient autophagic processes, synaptopathies, and neuronal death [20, 89]. Using a murine model of post-infectious cognitive dysfunction after WNV encephalitis, persistent loss of pre-synaptic termini was observed in the CA3 region of the hippocampus [20], a region important for spatial learning and memory [90]. Decreases in presynaptic markers were also observed in the hippocampi of post-mortem samples from WNV encephalitis patients [20]. Loss in synapse numbers has also been demonstrated in murine models of ZIKV and HIV-1 encephalitis [84, 91]. During recovery from ZIKV, microglia engulfed post-synaptic termini in the CA3 region of the hippocampus, followed by complete phagocytosis of neuronal perikarya [89]. ZIKV-infected animals also exhibit deficits in spatial learning. In the cases of WNV and ZIKV encephalitis, classical complement proteins C1q and C3 were increased in the CNS, and microglia were shown to mediate synapse elimination [20, 84]. Importantly, ZIKV infected WT mice treated with a C1q neutralizing antibody also exhibited decreased synapse elimination [84]. Of interest, loss of synaptic termini within the CA3 region, as observed post-recovery following viral CNS infections, also occurs in mouse models of β-amyloid overexpression and tauopathy [50, 57]. This suggests generalizable mechanisms of synapse loss occur in neuroinfectious and neurodegenerative processes. Furthermore, through genetic deletion or pharmacological blockade, synapse elimination was prevented in both models. WNV infected C3−/− and C3aR−/−, but not C3R or C1/2R, mice exhibit lack of synapse elimination, in conjunction with decreased colocalization of presynaptic markers in activated microglia during acute infection [20]. The requirement for C3aR is in contrast with various other studies of non-infectious CNS complement mediated synapse elimination, where synapse elimination occurs via C3-CR3 signaling, suggesting anaphylatoxin signaling and infiltrating immune cells may play a larger role during synapse elimination following viral infections. A murine model of HIV-1 associated neurocognitive disorders (HAND) demonstrated that the HIV-1 Tat protein induces increases in C1q and C3 levels, as well as synapse loss in the cortex. However, C1qKO mice administered Tat protein also exhibit loss of synapses, indicating Tat induced synapse elimination is C1q independent [91]. Thus, complement activation may not underlie synapse elimination in all cases of CNS viral infections, or may exhibit CNS region-specific effects. Further studies assessing complement mediated synapse elimination in other encephalitic viral models, evaluating both cortical and hippocampal regions, are needed to address this.

While our discussion here focuses primarily on encephalitic arboviruses, it should be mentioned that there are reports of neurological consequences as a result of infection from SARS-CoV-2 [92–94]. Patients have experienced anosmia, encephalopathy, acute necrotizing encephalitis, and ischemic events. Indeed, post-COVID-19 clinics have arisen in anticipation of long-term neurological sequalae following viral infection. SARS-CoV-2 infection increases complement activation, yet appears to be protected from complement-mediated viral restriction strategies during acute infection [95], suggesting the virus can adequately evade or even exploit host immune responses. However, whether CNS complement activation occurs within the CNS of COVID-19 patients remains to be determined. With so many questions left to be answered about the pathogenesis of SARS-CoV-2 infection, monitoring long term neurological outcomes will be important.

Complement as a biomarker for post-infectious and noninfectious cognitive disorders

In addition to the acute neuroinvasive syndromes caused by neurotropic viruses, patients that recover may experience significant long-term cognitive dysfunction [74–75, 79]. Acutely, these infections can be diagnosed via virus-specific IgM ELISA or PCR analyses of cerebrospinal fluid (CSF) [80]. However, there are currently no methods to identify patients at risk for the development of cognitive deficits. Complement proteins have emerged as potential biomarkers of CNS viral infections and resulting post cognitive sequelae. A study of CSF samples obtained from patients with the flavivirus tick-borne encephalitis virus (TBEV) reported increased C1q, C3a, C3b, and C5a concentrations compared to healthy controls. Of interest, there was a marked increase in C1q compared to other complement components [96]. Similarly, study of CSF samples from HIV-1+ neurocognitively impaired individuals reported increases in C1q compared to CSF from neurocognitively normal controls [97]. These studies suggest that detection of classical complement proteins in the CSF may be utilized to predict or follow long-term neurocognitive deficits from viral infections.

The onset of pathology during degenerative neurological diseases often occurs years before symptoms are evident, necessitating adequate biomarkers that provide early diagnosis and predict disease progression. Due to its extensive involvement in CNS disorders, researchers have also evaluated the use of complement proteins as potential biomarkers for cognitive dysfunction in a variety of neurologic disease contexts. Increased levels of C1q and C3 are detected in the CSF of patients after traumatic brain injury (TBI), a major cause of cognitive deficits, compared to controls [98–99]. Complement proteins have also been examined in biomarker studies of AD [100–101]. C3, in particular, is a key differentiator of AD, with low plasma levels of C3 shown to be associated with AD, and lower CSF levels of C3 associated with more severe disease progression [102–103]. Differences in complement CSF levels are also observed in autoimmune CNS disorders such as multiple sclerosis with significant increases of C1q in relapsing remitting MS and secondary progressive MS CSF samples compared to controls [104]. Levels of CSF C3 also correlate with increased disability in MS patients, with highest levels found in patients with progressive MS [105]. In most of these diseases, fluctuations in complement deposition signify changes in disease stage, making it a potential biomarker of disease progression (Fig 3).

Figure 3: Complement as a biomarker of human CNS disease.

Complement proteins increase or decrease in CSF levels during CNS disease progression. In AD patients CSF C3 levels decrease in more severe disease [103–104]. After TBI injury, studies show increases in C1q and C3 in human CSF [98–99]. Similar trends are seen in other autoimmune CNS diseases such as MS, where C1q and C3 increases correspond with more severe disease and disability [104–105]. After infection from TBEV, and HIV patients have increased CSF complement expression (TBEV-C1q, C3a, C3b, C5a) (HIV-C1q) compared to healthy controls [96–97].

Concluding Remarks and Future Perspectives

The study of complement in developing, normal and diseased CNS states has revolutionized our understanding of neuroimmune regulation of neuronal networks. Under normal conditions, this process refines networks for instance by the maintenance of certain synapses and removal of others. In various neurologic diseases, complement levels and microglial activation are increased, in some cases via unknown mechanisms, leading to disruption of networks, with subsequent memory loss and cognitive dysfunction. Importantly, the complement components and receptors involved in this process are often similar between neuroinfectious and non-infectious neurodegenerative diseases, suggesting generalizable mechanisms contribute to cognitive decline and pathological forgetting.

Translating these recent advances to therapeutics for treating memory disorders, however, will require extensive investigation to determine how levels of complement are regulated during neuroinflammatory diseases and whether complement is a suitable target to reverse pathological forgetting (see Outstanding Questions). Less studied, with regard to complement, are diseases involving other forms of pathological memory processes, such as post-traumatic stress disorder (PTSD) or TBI, the latter of which may present with persistent confabulation, a type of memory error in which gaps in a person’s memory are unconsciously filled with fabricated, misinterpreted, or distorted information [106]. This is especially relevant for PTSD, where progression of disease is positively correlated with CNS increase in master regulators of innate immune pathways [107]. Correlation and kinetics of altered neuronal activity with regard to complement expression and localization, and innate immune-mediated microglial activation are needed to better define therapeutic targets in various diseases of pathological forgetting. Use of transgenic animal models for neural-circuit tracing and manipulation, inducible conditional complement mouse models, single cell ‘omics, microglial ablation approaches, and novel inhibitors of complement may be used in combination to allow more thorough comprehension of the molecular and physiological mechanisms underlying memory disorders.

Outstanding Questions.

• Complement signaling has been shown to contribute to synaptic pruning in the visual system and spinal motor system during development. Is complement signaling a generalizable mechanism for sculpting of all neural systems during development?

• Does complement contribute to synaptic plasticity during adulthood, and if so, how?

• There is evidence of complement expression in neurons, microglia, infiltrating macrophages, astrocytes, and oligodendrocytes in the CNS. Global deletion of complement leads to decreased synapse elimination following WNV infection, AD, and other neurological disorders. What cells are producing complement leading to improvements in synapse numbers in these disease models? How are levels of complement regulated in the CNS?

• Complement proteins have been shown to be potential biomarkers of CNS disorders, and contribute to mechanisms of CNS cognitive dysfunction. Is there potential for the use of complement inhibitors (for instance ones used in certain rheumatologic disorders, or being evaluated in clinical trials) as therapeutics for neurodegenerative and neuroinfectious etiologies of cognitive dysfunction?

Figure 2: Complement signaling in neurodegeneration and neuro-infection.

There is evidence of complement expression on astrocytes, microglia, neurons, and oligodendrocytes following either neuro-infection or neurodegeneration. In neurodegeneration, classical complement components are expressed on all 4 cell types, and activation leads to synapse elimination, increased neuroinflammation and changes in dendritic morphology. In neuro-infection, components are expressed on neurons and microglia, and activation leads to synapse elimination. Interestingly, C3-CR3 signaling contributes to synapse elimination in neurodegeneration, whereas C3-C3aR signaling leads to synapse elimination in neuro-infection. Modulators of complement during neurodegeneration include cytokines, progranulin, TREM2, ApoE [70, 73, 108–109].

Highlights.

• Classical complement proteins mediate many physiological and pathological processes both in the periphery and in the central nervous system (CNS). Within the CNS, an established function of these proteins is mediating microglial phagocytosis of synapses during development and disease. The pro- and anti-inflammatory signals leading to these processes in the CNS are incompletely understood.

• CNS injury resulting from both neurodegeneration and viral infections is associated with increased complement expression and activation. Activation of classical complement proteins alter microglial responses in a context specific manner, and synaptic elimination occurs via similar complement mediated pathways in neurodegeneration versus viral infection.

• Various viruses are neutralized in the periphery via mechanisms involving classical complement proteins in order to limit viral spread. Some of these proteins are upregulated in the CNS after neurotropic viral infection. Less known is whether CNS complement limits viral spread in this organ.