Contribution of CNS and extra-CNS infections to neurodegeneration: a narrative review

Abstract

Central nervous system infections have been suggested as a possible cause for neurodegenerative diseases, particularly sporadic cases. They trigger neuroinflammation which is considered integrally involved in neurodegenerative processes. In this review, we will look at data linking a variety of viral, bacterial, fungal, and protozoan infections to Alzheimer’s disease, Parkinson’s disease, amyotrophic lateral sclerosis, multiple sclerosis and unspecified dementia. This narrative review aims to bring together a broad range of data currently supporting the involvement of central nervous system infections in the development of neurodegenerative diseases. The idea that no single pathogen or pathogen group is responsible for neurodegenerative diseases will be discussed. Instead, we suggest that a wide range of susceptibility factors may make individuals differentially vulnerable to different infectious pathogens and subsequent pathologies.

Background

Central nervous system (CNS) infections have been suggested to act as a possible trigger for neurodegenerative diseases such as Alzheimer’s disease (AD), Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS) and multiple sclerosis (MS). They are particularly interesting when trying to explain the high prevalence of sporadic neurodegenerative diseases which genetic factors alone are unable to explain.

For a while, the field of neurodegeneration has been dominated by the idea that neurodegenerative diseases are caused by the pathological accumulation of toxic aggregating proteins such as amyloid-β (AD), α-synuclein (PD). However, the processes that trigger protein accumulation in various brain areas are still poorly understood, apart from rare familial cases where the accumulation is caused by dominantly inherited mutations. Since research targeting aggregating proteins has been limited in its ability to produce effective treatments, many researchers have turned towards complementary hypotheses [1].

The ‘infection hypothesis’ suggests that infections can push the system to pathology by direct harm caused by the infection or by triggering neuroinflammation [1,2,3]. As we will later see, microbes can harm the host organism in multiple ways, such as via toxic molecules, inhibition of host immune function, disruption of tissue tight junctions, induction of proteotoxic stress, or by directly killing host cells. On the other hand, they trigger an immune response which can be detrimental to host cells (‘bystander effect’) especially if the inflammatory state becomes chronic. However, not all infections lead to neurodegeneration, and genetic susceptibility factors may help explain why some people are more vulnerable to the detrimental effects of infections than others. Furthermore, infections are only one possible cause for neuroinflammation alongside other factors such as traumatic injury, ischemic injury, autoimmunity, metabolic disease, and lifestyle (diet, sleep, exercise, stress, smoking) which we will not cover here.

To assess whether infections are involved in neurodegeneration, we need to determine (a) whether patients suffering from neurodegeneration have a history of CNS infections, (b) whether non-CNS infections can cause neurodegeneration without direct CNS invasion, (c) whether infections increase risk of neurodegeneration, and (d) whether infections are able to induce neuroinflammation and neurodegeneration-related changes at the cellular and tissue level, possibly even after infection clearance. It would also be important to know whether pathogen infections and known risk factors for neurodegenerative diseases interact, and whether eradicating infections is an effective treatment and/or prevention strategy for neurodegenerative diseases. Since aggregating proteins are integral to neurodegenerative diseases, we will look at how aggregating proteins and infections could fit into the same picture.

Main text

Prevalence of CNS infections in the global population

Neurological symptoms such as fatigue, headache, sensory changes, cognitive changes, psychosis, seizures, paresis, and coma are common symptoms of many infectious diseases [4]. For example, around 20–30% of all Covid-19 patients have been reported to have neurological symptoms such as fatigue or cognitive impairment even months after acute respiratory infection [5]. It is often difficult to know whether the neurological symptoms are secondary to the systemic disease state or caused by direct infection of the CNS. However, a variety of pathogens from viruses and bacteria to fungi and protozoa have been detected in the CNS using histochemical and molecular biology methods [6]. In fact, many globally significant diseases, such as malaria, borreliosis/Lyme disease, HIV, diphtheria, tuberculosis, candidiasis, and syphilis have neurological presentations where the otherwise systemic infection enters the brain [4].

Differences in target tissues, infection routes, replication, and release patterns are likely to affect the outcome of the infection [7]. Some CNS-infiltrating pathogens specifically target brain cells such as neurons or glia. For example, rabies virus, Zika virus, tick-borne encephalitis virus and herpes viruses specifically target neurons. They often cause severe disturbance of the CNS homeostasis, such as encephalitis or meningitis, which can cause death or disability. Such severe infections are associated with many disorders including neurodegeneration in survivors [8,9,10,11,12].

Other pathogens are more likely to enter the CNS during disturbance such as immunodeficiency, inflammation, blood–brain barrier (BBB) breakdown, trauma, neurosurgical procedures or other infections. Such pathogens include Candida yeasts, Cryptococcus and Mycorales fungi, the protozoan parasite Toxoplasma gondii, as well as the bacteria Streptococcus pneumoniae, Staphylococcus aureus and Mycoplasma sp. [13]. These microbes can cause encephalitis in immunocompromised patients, but such severe representations are rare in otherwise healthy individuals.

Certain pathogens, such as Helicobacter pylori, periodontal bacteria, and gut microbes, have been linked to neuropathology without a direct infection of the CNS. In fact, changes of the gut-brain axis, such as gut dysbiosis (imbalance of the gut microbiome), are very common in patients of neurodegenerative diseases [14]. It is thought that systemic inflammation caused by extra-CNS infections can jump across the CNS barriers (most notably the BBB) via proinflammatory factors such as cytokines, extracellular vesicles, and small lipid mediators, which triggers neuroinflammation [15,16,17,18,19,20]. Interestingly, this jump can occur prior to BBB disruption [21] which suggests that systemic infections could disturb BBB function from both sides. Furthermore, changes in the gut-brain axis can affect the CNS also via the secretion of neurotransmitters, vitamins, important fatty acids such as butyrate, and amyloid proteins by gut microbes [14].

Finally, most neurodegenerative disease patients do not have a history of severe CNS pathologies such as encephalitis or meningitis. Thus, common pathogens that cause milder infections have gained the interest of scientists. For example, most of us carry latent infections (Table 1) which means that such infections could potentially explain common diseases such as dementias. Latent infections are characterized by alternating active and quiescent periods, where the pathogens hide from the immune system only to reactivate under favorable conditions such as stress-induced immunosuppression. To do this, they use immunomodulatory molecules [22, 23], biofilms [24], cysts [25], encapsulation [23], and integration to the host genome (viruses) [26]. While the reactivation of latent infections can lead to severe pathology such as encephalitis [11], majority of these infections are mild or asymptomatic. However, the chronic presence of these pathogens in the CNS is thought to cause long lasting or reoccurring neuroinflammation [25, 27, 28]. In fact, latent infections have been linked to severe conditions such as cancer [29,30,31] and neurodegenerative diseases.

Table 1 Examples of common pathogens that cause chronic latent infections and their prevalence in different adult human populations

General mechanisms of microbe-induced pathology in the brain

Potential mechanisms for microbial neuropathology are diverse. Many of them are secondary to the microbe’s attempt to infect, survive and replicate inside the host. For example, many viruses and intracellular parasites lyse the host cell as part of their reproductive cycle to release the newly produced pathogens. However, there are other ways in which pathogens are harmful. For example, many molecules that microbes use to survive can be toxic to the host [70]. Furthermore, the infiltration of pathogens into host tissues via pore formation [71], modulation of adhesion proteins [72], and hijacking of the host endocytosis [73] can disrupt important host cell-to-cell contacts [74, 75], cause the fusion of adjacent cells (syncytia) [75,76,77,78], and cause leakage of molecules across host barriers such as the BBB [75].

Similarly, by hijacking the host protein synthesis machinery, cytoskeleton, and intracellular transport system, the pathogen facilitates its own replication and spread. Simultaneously, it can also disrupt the homeostatic function of the host cell, such as normal protein synthesis and axonal transport [79, 80]. This can lead to problems such as proteotoxic stress [81]— a common occurrence in neurodegenerative diseases [82].

Furthermore, the immunomodulatory and immunosuppressive tactics utilized by pathogens to evade the host immune system [22, 23, 83] can lead to secondary infections, or to the disruption of homeostatic functions performed by the immune cells, e.g. waste clearance performed by microglia.

Finally, host immune response to pathogen surface structures or secreted products can become as detrimental as the infection itself. For example, cytokines, antimicrobial molecules, reactive oxygen species (ROS), nitric oxide, phagocytosis, and forced cell death are used by immune cells to destroy pathogens [84]. These same mechanisms can be harmful to nearby host cells if there is no balance between optimal pathogen clearance and excessive inflammatory response [75, 84].

A brief introduction of discussed neurodegenerative diseases

Dementia (unspecified)

World Health Organization estimates that around 55 million people are affected by dementia around the world and the number is thought to increase to 139 million by 2050 [85]. Dementia itself is not a disease but a syndrome (a collection of symptoms) which describes the decline of cognitive abilities such as memory, learning, concentration, planning, motivation, language processing, reasoning, and thinking. It also affects mood and can cause anxiety, depression, or aggression. The most common causes for dementia are AD (70% of all cases), frontotemporal dementia, vascular dementia, and dementia with Lewy bodies. All of them have slightly different main symptoms and disease dynamics, but the overlap between diseases is considerable. Determining which disease causes dementia in each patient is difficult and often based on symptoms alone. Thus, many studies handle dementia as a collective unit.

Alzheimer’s disease

AD causes around 60–70% of all dementia cases [85]. It is characterized by progressive degeneration of the brain parenchyma and the accumulation of proteinaceous amyloid-β and tau inclusions. The pathology is associated with neuroinflammation, glial activation, and dysfunction of the BBB and the brain blood circulation.

Apolipoprotein E (APOE) ξ4 allele is the most prevalent genetic risk factor for late-onset AD (begins after the age of 65). It increases the risk of AD ~ threefold in heterozygotes and ~ 15-fold in homozygotes [86]. APOE is a lipid carrier molecule that facilitates the trafficking of cholesterol and phospholipids between cells. Due to its effect on lipid membranes, APOE is involved in many cellular functions, including the regulation of immune cells such as microglia [87, 88].

Many other risk mutations for AD are enriched in microglia, including mutations in TREM2, ABI3, ABCA7, CD33, and CR1 [89,90,91]. For example, triggering receptor expressed on myeloid cells 2 (TREM2) regulates microglial functions such as phagocytosis, microglial activation, and inflammatory responses, which are crucial for CNS immunity. Rare mutants of TREM2, such as the single nucleotide mutation R47H, have been associated with increased risk of AD and other dementias [92, 93].

Parkinson’s disease

PD is the second most common neurodegenerative disease after AD. In 2019 it was estimated to have affected over 8.5 million individuals globally [94]. PD is the most common cause of parkinsonism, a motor disorder characterized by tremors, rigidity, bradykinesia, and posture changes. PD also causes a variety of gastrointestinal (GI) symptoms. The symptoms are a result of the progressive loss of dopaminergic neurons of substantia nigra. The process is associated with neuroinflammation and the accumulation of amyloid-like α-synuclein.

Many important PD risk genes are involved in endosomal (LRRK2, SNCA) and lysosomal (GBA, TMEM175, CTSB) function, mitophagy (PINK1, PARK2), autophagy (SNCA, KAT8), RNA processing (TARDBP), and antigen presenting (HLA-DRB6, HLA-DQA1) [95]. Many of these functions are important for immune response. Thus, mutations in these pathways can make an individual susceptible to the negative effects associated with infections. For example, the PD risk mutation p.G2019S in the LRRK2 gene lead to higher reovirus mortality in mutant mice compared to wildtype animals. The effect seems to be caused by heightened inflammatory response [96, 97]. Interestingly, the LRRK2 p.G2019S mice were better at controlling septic Salmonella typhimurium infection [96] which hints at a pathogen-specific effect.

Amyotrophic lateral sclerosis (ALS) and ALS-like syndromes

Amyotrophic lateral sclerosis, also known as Lou Gehrig’s Disease, is a fatal motor neuron disease characterized by progressive degeneration of upper and lower motor neurons. It leads to muscle weakness and the loss of motor control, which leads to death when it spreads to respiratory muscles. Intracellular aggregation of proteins such as transitive response DNA-binding protein 43 (TDP-43), fused in sarcoma (FUS) and superoxide dismutase 1 (SOD1) are common.

Noncoding hexanucleotide GGGGCC repeat expansion in C9orf72 is the most common genetic mutation found in ALS patients (40–50% of familial cases, 5–10% of sporadic cases) and frontotemporal dementia [98]. The gene is involved in many cellular functions including lysosomal function, stress granule formation and immune function. Several gain-of-function and loss-of-function mechanisms have been suggested to explain how mutations in C9orf72 increase ALS risk [99, 100].

Several mutations in SOD1 explain around 2% of all ALS cases and 20% of familial cases. SOD1 is an important antioxidant enzyme that protects cells from ROS, which is secreted by immune cells during pathogen infections [101]. A combination of infection and genetic susceptibility to ROS could make an individual vulnerable to neurodegeneration [102].

Multiple sclerosis

MS is the most common inflammatory demyelinating disease in the brain and spinal cord. It affects around 2.8 million people worldwide [103]. It is characterized by defects in sensory, motor, and autonomic functions. The most common clinical course involves alternating relapses and remissions where flare-ups of symptoms are followed by periods of full or partial recovery. Common histopathological findings include focal demyelinated plaques or lesions surrounded by activated T-cells and myeloid cells such as microglia. The leading theory is that MS is caused by autoimmunity against host myelin-producing cells. In fact, many MS risk genes are involved in autoimmunity, such as the human leukocyte antigen class II allele HLA-DBR1*15:01 and the interleukin 2 receptor subunit alpha (IL2RA) [104]. The possible role of infections in MS remains unknown. On the one hand, microbial molecular mimics of myelin-associated proteins such as the myelin basic protein (MBP) could trigger autoimmunity [105, 106]. On the other hand, it has been suggested that excessive hygiene in modern societies leads to deficient training of immune cells which can cause them to attack host tissues (‘the hygiene hypothesis’).

Role of pathogens in neurodegenerative diseases

Many CNS-infiltrating pathogens have been found in neurodegenerative disease patients using histological and PCR-based methods. Many of them show positive associations with neurodegeneration in population level studies (often around 1.3–3.0-fold increase in risk, see Table 2). The strength of these associations is of the same magnitude as other commonly accepted neurodegeneration-associated factors such as cardiovascular disease, stroke, and sleep disorders [107,108,109,110,111,112].

Table 2 An overview of studies investigating the association between herpesviruses and dementia, AD, PD, ALS, or MS

Furthermore, in vitro cell culture studies and in vivo animal models have shown that many pathogens can induce neuroinflammation and neurodegeneration-related changes, such as glial activation, leukocyte infiltration, cytokine production, BBB dysfunction, and the accumulation of neurodegeneration-related aggregating proteins.

The diversity of pathogens that have been implicated in neurodegenerative diseases is striking, even though none of the connections have been conclusively proven. For example, Sipilä et al. screened the association of hospital-treated pathogen infections (925 International Classification of Diseases-10 codes) with dementia using the medical records of over 700 000 participants in Finnish and United Kingdom databanks. The dataset included pathogens of all classes: viruses, bacteria, fungi, and protozoan parasites, and both CNS and non-CNS infections. All types of infections increased the risk of dementia which suggests that there is no single pathogen or infection type that would be solely responsible for dementia. Furthermore, multiple simultaneous or subsequent infections increased the risk of dementia more than single infections (adjusted hazard ratio (aHR) 5.08 vs. 3.04 respectively), particularly in the case of combined virus and bacterial infections (aHR 8.15, 95% CI: 4.73–14.05) [116]. Similar results have since been published by Bohn et al. [139]. However, there may be disease-specific differences. For example, Fang et al. found no association between CNS infections and ALS in a study of 4000 ALS patients and 20 020 matched controls [140].

Below, we will continue exploring this diversity by reviewing the data that supports the involvement of CNS infections by herpesviruses, enteroviruses, HIV, SARS-CoV-2, spirochete bacteria, bacterial pneumonia, fungi, and T. gondii in unspecified dementia, AD, PD, ALS, and MS (Fig. 1). We will also briefly mention extra-CNS infections that have strong link to neurodegeneration, namely periodontal disease, H. pylori infections, and gut dysbiosis. Simultaneously, we will discuss the importance of individual susceptibility, which may explain why neurodegeneration is triggered in some individuals while most are left unaffected. The final section of this review discusses the idea that neurodegeneration-related aggregating proteins could function as innate immune proteins that bind microbes and repress their replication while activating antimicrobial immune pathways.

Fig. 1

A variety of CNS infections are associated with AD, PD, ALS, and MS. IL2RA Interleukin-2 receptor subunit alpha

Herpesviruses

The family of human herpes viruses includes herpes simplex virus 1 and 2 (HSV-1 and HSV-2), varicella-zoster virus/human herpes virus 3 (VZV/HHV-3), Epstein-Barr virus/human herpes virus 4 (EBV/HHV-4), cytomegalovirus/human herpes virus 5 (CMV/HHV-5) and human herpes viruses 6, 7 and 8 (HHV-6, HHV-7, HHV-8). They cause common ailments such as cold sores of the mouth and genitals (HSV-1 and HSV-2), chickenpox/shingles (VZV), and mononucleosis/ ‘the kissing disease’ (EBV). HSV-1 is also the most common cause of encephalitis worldwide.

While many members of this family of viruses have been implicated in neurodegenerative diseases, it is unlikely that any of these viruses would be the sole cause of any neurodegenerative disease. After all, despite the extremely high prevalence of herpesviruses in the global population, most us are not affected by neurodegeneration.

Thus, it is not surprising that studies attempting to find correlation between neurodegenerative diseases and herpesviruses have yielded mixed results (Table 2). However, accounting for genetic risk factors associated with neurodegenerative diseases might help explain this discrepancy. For example, the association between APOE ε4 and HSV-1 in AD was first suggested by Ruth Itzhaki et al. in the late 1990s, when they reported that the odds of finding APOE ε4 allele in HSV-1 positive AD patients was almost 17 times higher compared to HSV-1 negative individuals without AD (OR 16.8, 95% CI: 3.61–77.8) [141].

Linard et al. and Lopatko Lindman et al. have also reported a 3.68–4.55-fold increase in AD risk in HSV-1 positive APOE ε4 carriers [123, 125]. Furthermore, Lopatko Lindman et al. reported an association between AD, HSV-1, and a risk score which was calculated from the presence of mutations in nine AD risk genes (ABCA7, BIN1, CD33, CLU, CR1, EPHA1, MS4A4E, NECTIN2, and PICALM) (OR 2.35, 95% CI: 1.21–4.56, P = 0.01) [125]. Interestingly, the combination of APOE ε4 with other pathogens, including HSV-2, did not increase AD risk [125]. It suggests that the genotype-microbe interactions are somewhat pathogen-specific.

The mechanisms through which genetic risk factors, such as APOE ε4 allele, increase AD risk, are still under investigation. However, it has been suggested that the APOE isoform may affect for example HSV-1 latency and frequency of reactivation [141,142,143] or the number of oral lesions [144].

Cell culture models also support the involvement of HSV-1 in AD. In vitro studies by Cairns et al., Qiao et al. and D’Aiuto et al. have shown that HSV-1 can induce AD-like pathology in human induced pluripotent stem cell (hiPSC) -derived neural cultures and 3D organoids[145,146,147,148]. The signs of AD pathology included changes in neuronal morphology, the formation of syncytia, neuronal loss, amyloid-β and tau accumulation, expression changes in amyloid-β processing genes (PSEN1, PSEN2, BACE1), increased expression of pro-inflammatory cytokines (TNF-α, IFN-γ, IL-1β, IL-6, IL-10, IL-4, CXC3R) as well as gliosis characterized by changed astrocyte morphology, increased expression of genes involved in astrogliosis (e.g. glial fibrillary acidic protein, GFAP) and microglial activation (CD11b, CD68, HLADR) [145,146,147,148]. Fruhwürth et al. have also reported that HSV-1 infection downregulates TREM2 pathway in hiPSC-derived microglia which leads to impaired interferon-β induction and impaired phagocytosis of HSV-1 infected neurons [149]. One possible mechanism for HSV-1-induced neuropathology is through the binding of the virus envelop glycoprotein D and the Aβ42 peptide. The amino acid residues involved in Aβ42-binding are left open, which means that the Aβ42-virus complex can act as a seed for Aβ oligomerization[150]. While this binding property may have evolved as part of the innate immune system (see later section), the process has likely become harmful to the host in conditions such as AD.

VZV is another herpesvirus, which has been researched in terms of neurodegeneration. Some studies have reported a mild increase in the risk of dementia (1.09–1.11-fold) [115, 117, 118] and PD (1.17–1.5-fold) [129, 130] in herpes zoster patients (detectable reactivation of VZV) while others have found either no association or an inverse association [114, 127, 151, 152]. Interestingly, Cairns et al. have reported that VZV induced gliosis and secretion of proinflammatory cytokines in hiPSC -derived neural cultures, but not the accumulation of amyloid-β or tau. Instead, VZV induced the reactivation of quiescent HSV-1 infection which then led to the accumulation of Aβ and phosphorylated tau [145]. Similar reactivation could be induced by other viruses such as SARS-CoV-2 [153,154,155].

Like HSV and VZV, EBV has been linked to many neurodegenerative diseases [156, 157]. The strongest link has been identified between EBV and MS. Up to 32-fold increase in MS risk has been reported following EBV infection [8, 134, 135]. Furthermore, MS is very rare in individuals who are seronegative for EBV [158]. EBV can infect neurons directly, disrupt BBB integrity, and cause neuroinflammation [156]. However, the prevailing theory is that EBV predisposes the host to MS-related autoimmunity through the infection of host B cells as well as through molecular mimicry of host myelin-associated proteins myelin basic protein (MBP) and glial cell adhesion molecule (GlialCAM) by the EBV nuclear antigen 1 (EBNA1) [105, 106, 159]. Since EBV is almost ubiquitously present in the population (90%), the affected individuals likely carry other vulnerabilities to autoimmune diseases. For example, the MS risk factor HLA-DRB1*15:01 can facilitate EBV entry into host B cells [160, 161]. Lifestyle factors, such as smoking, can further modulate this interaction [162]. Furthermore, another EBV protein, EBNA2, binds and alters the expression of MS-associated host genes which may increase MS risk. The effect is dependent on the presence of known protective or risk mutations for MS (listed in Table 3) [163].

Table 3 A list of publications that studied the interactions between infections and known risk genes for AD, PD, ALS, and MS

Antiherpetic drugs such as valacyclovir and acyclovir have been reported to reduce symptoms and slow disease progression in dementia patients. Linard et al. have reported that intake of at least one systemic antiherpetic drug reduced the risk of AD (aHR 0.85, 95% CI: 0.75–0.96, p = 0.009) in a cohort of 6642 subjects over the age of 65. Most subjects had undergone only a single intake of antiherpetic drugs and regular treatment was rare [124]. Antiviral therapy also reduced the risk of dementia in a Taiwanese population cohort study of 39,205 herpes zoster patients (HR 0.55, 95% CI 0.40–0.77) [118] and a South Korean cohort study of 34 505 herpes zoster patients (aHR 0.79, 95% CI: 0.69–0.90)[117]. The typical length of antiviral drug treatment was not specified for these studies. A study by Young-Xu et al. on 87 687 HSV-positive US veterans over the age of 50 years found that antiherpetic medication was associated with lower dementia risk when compared to the untreated group (aHR 0.75, 95% CI: 0.72–0.78). The reduction in dementia risk was associated with lower markers of neuroinflammation [167]. It is noteworthy that these therapies are not specific to any specific herpesvirus and that the positive effect could also be mediated by other viral infections.

In contrast, Schnier et al. found no convincing association between antiherpetic drug use and reduction in dementia risk in a study of 2.5 million individuals aged over 65 years. They only found a small and heterogenous negative effect in an analysis of four European databases from Wales, Germany, Scotland and Denmark. However, the typical length of antiherpetic drug treatment was relatively short: only 1–2 weeks [168]. In fact, it has been suggested that the antiherpetic/antiviral treatment lengths commonly administered in Europe are not long enough to show positive effects in these database-driven association studies. For example, the above-mentioned study by Young-Xu et al. reported that increasing length of antiviral treatment is associated with larger reductions in dementia risk in symptomatic HSV carriers. While any antiherpetic medication reduced the risk of dementia by 25% (HR = 0.75, 95% CI: 0.72–0.78), treatments longer than one year reduced the risk of dementia by 43% (HR = 0.57, 95% CI, 0.53–0.61). In the subgroup receiving medication for less than 30 days, the reduction was negligible (HR = 0.93, 95% CI: 0.87- 0.98) [167]. Thus, more studies testing the efficacy of longer antiherpetic drug treatments are needed. An ongoing clinical study by Columbia University [169] is currently assessing the efficacy of 7–8-week valacyclovir treatment in HSV-1 and HSV-2-positivie patients suffering from mild AD [170].

Enteroviruses

Enteroviruses, such as poliovirus, coxsackievirus, echovirus, enterovirus-A71, and enterovirus-D68 have gained interest in the field of motor neuron diseases due to their ability to infect motor neurons [171]. For example, poliovirus, the most famous enterovirus, attacks motor neurons of the spinal cord and brain stem causing neuroinflammation which can lead to irreversible paralysis (poliomyelitis aka. Polio). Around 28% of poliomyelitis survivors develop motor neuron disease (post-polio syndrome) decades after acute disease suggesting a chronic or reactivated infection [172].This disorder resembles ALS because it is characterized by gradual weakening and atrophy of specific muscles (often the limbs affected by poliomyelitis years earlier) due to the loss of motor neurons in the brainstem and spinal cord. Other symptoms include muscle fasciculations, fatigue, pain, sleep disturbance, and sometimes problems breathing or swallowing. However, post-polio syndrome can often be distinguished by the life history of poliomyelitis as well as slower progression and more generalized fatigue [173].

The relationship between ALS and ALS-like syndromes is still unclear, as is the involvement of enteroviruses in ALS. Some studies have found enteroviral RNA in ALS and motor neuron disease patients more frequently than in controls using PCR methods [174,175,176,177]. In contrast, others have failed to detect enterovirus RNA in the spinal cord of ALS patients [178,179,180]. The discrepancies could be explained by geographical differences (positive results in France, UK, Japan vs. negative in US and Australia) or by random variation introduced by small sample sizes (< 30 samples per group for all studies except one positive study where the number of samples was ten times higher).

Interestingly, Xue et al. have reported that sublethal coxsackievirus B3 infection can cause an increase in proinflammatory gene expression, TDP-43 pathology, neuronal damage, and immune cell infiltration in the CNS of normal C57BL/6J mice. In addition, coxsackievirus B3-infected mice carrying an ALS-related mutation SOD1^G85R displayed also a reduction in their lifespan and earlier start of motor dysfunction than non-infected ALS mice. Their results suggest that while the virus alone is able to cause neurodegenerative changes in these mice, a genetic susceptibility is needed for the onset of motor dysfunction [166].

In contrast, MS is a disease which affects the myelin sheaths around axons instead of the neurons themselves. Based on the limited evidence available, enteroviruses are not involved in the development of MS. For example, Kuusisto et al. found no evidence of enterovirus infection in the serum or cerebrospinal fluid of 17 MS patients [181]. Similarly, Perlejewski et al. detected enterovirus by RT-qPCR in the cerebrospinal fluid of only one MS patient (1 out of 34) [182].

HIV

HIV/AIDS is an infectious viral disease that compromises the host immune system due to the selective tropism of the virus to immune cells such as CD4 positive T helper cells, macrophages, dendritic cells, and microglia. The replication cycle of HIV kills the host cell which results in depletion of host immune cells and generalized immune deficiency. If left untreated, the patients succumb to secondary infections typically within 10 years [183]. In fact, HIV patients are susceptible to other opportunistic infections such as candidiasis, toxoplasmosis and bacterial pneumonia [184,185,186] as well as the reactivation of other latent infections such as herpesvirus infections [187, 188].

Interestingly, HIV has been linked to multiple neurodegenerative diseases. Around half of HIV/AIDS patients develop HIV-associated neurocognitive disorder (HAND) even with effective antiretroviral medication [189, 190]. Many of the symptoms resemble dementia, such as difficulties in learning, memory, decision making, and concentration. Patients also display neuroinflammation, neuronal loss, microglia/macrophage activation, multinucleated giant cells, and diffuse atrophy of many brain areas [191,192,193]. According to Wendelken et al. HAND-associated brain atrophy is exacerbated by the AD risk factor APOE ε4 [194]. Furthermore, neuropathologies such as progressive brain atrophy and microglia/macrophage activation, are present even in those with successful antiviral therapy which suggests that the pathology is mainly driven by neuroinflammation [195,196,197]. However, low-level viral replication in the CNS cannot be completely discounted even with successful suppression of the viral load in the plasma [191]. Interestingly, the use of antiviral therapy in HIV positive individuals is associated with an increase in the CSF levels of Aβ40 and Aβ42 compared to untreated HIV patients. Patients suffering from HIV-associated dementia (HAD) display reduced CSF levels of Aβ40 and Aβ42 compared to neurocognitively unimpaired individuals. The former is also commonly observed in AD patients [198]. These results suggest that HIV impairs Aβ clearance to the CSF, which may be a result of increased Aβ deposition into mature plaques. This process may be exaggerated in HAD patients. Increased CSF Aβ40 and Aβ42 levels following antiviral therapy would then indicate rescued Aβ clearance, which is however not enough to rescue cognitive functions [199].

Parkinsonism is another possible, albeit rare, outcome of HIV infection. There are multiple possible causes for parkinsonism in people living with HIV, including secondary infections, HAND, dopamine-blocking drugs such as neuroleptics, adverse reaction to antiretroviral therapy, and HIV encephalitis. Thus, the mechanistic connection between HIV and parkinsonism is not clear-cut. In general, the emergence of antiretroviral therapy has reduced the occurence of HIV-associated parkinsonism. Amelioration of symptoms following antiretroviral treatment, or the discontinuation of dopamine-blocking drugs, has also been reported [200]. Hence, HIV testing is worth considering in patients with otherwise unexplained parkinsonism.

Several case studies have also reported ALS-like syndrome in HIV patients, including in vivo signs of upper and lower motor neuron involvement [201,202,203,204,205,206,207,208,209,210,211]. However, the authors of this article are unaware whether motor neuron loss has been documented in post mortem or not. The patients often display earlier onset than in classical sporadic ALS [209]. The causation between HIV and ALS-like syndrome is yet unproven, and the putative mechanism is still unknown. However, the mechanism is likely indirect since HIV specifically targets immune cells such as microglia (instead of motor neurons). Similar to HIV and parkinsonism, antiretroviral therapy has been an effective treatment in many cases of HIV-associated ALS-like syndrome [202, 204, 208, 209] which suggests that the possibility of HIV infection should be taken into consideration in patients at high risk of exposure. In contrast, reduced risk of MS and reduced rate of relapsing have been reported in HIV positive individuals. It is currently not known whether the reduction in MS risk is a result of the HIV infection itself or secondary to antiretroviral therapy. Multiple explanations have been suggested. On the one hand, the depletion of CD4 + T cells by HIV could inhibit the development of CD4 + T cell-associated autoimmunity in untreated or late stage HIV patients. On the other hand, antiretroviral therapy could also inhibit other CNS viruses than HIV, including EBV, which could subsequently reduce MS risk. So far, the results have been variable and clear conclusions cannot be drawn. This and many other open questions in the field of HIV and MS have recently been reviewed by Stefanou et al. [212].

SARS-CoV-2

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a respiratory virus which caused the recent COVID-19 pandemic. Interestingly, CNS symptoms are common during and after SARS-CoV-2 infections, and they can last for extended periods of time (‘long covid’). A meta-analysis by Ceban et al. reported that twelve or more weeks after COVID-19 diagnosis 32% of individuals suffered from fatigue (68 included studies), and 22% of cognitive impairments (43 included studies) [5]. In a study by Xu et al. the burden of neurologic sequelae was 70 per 1000 persons after one year [213]. Changes in brain anatomy, such as reduction in grey matter thickness and global brain volume, have also been detected in COVID-19 patients – even milder cases [214]. Microhaemorrhages are also common [215,216,217].

This has triggered the questions whether SARS-CoV-2 can infect the brain, and whether the COVID-19 pandemic could increase the prevalence of neurodegenerative diseases in the future. Particularly, AD has been implicated due to the cognitive changes seen during and after COVID-19 disease. The diseases also share the risk factor APOE ε4, which increases the risk of late onset AD as well as severe COVID-19 disease [217, 218]. Furthermore, severe COVID-19 disease and mortality due to it are more common in AD patients.

To answer the first question: Post mortem studies have detected the presence of viral components in neurons, glia, and brain endothelial cells in deceased COVID-19 patients (the cases were not selected based on neurological symptoms) [219,220,221,222,223]. Consistently, stem cell-derived 2D and 3D models have shown that SARS-CoV-2 can infect the choroid plexus epithelium, astrocytes, subset of neurons, and possibly even microglia [224,225,226,227,228,229,230,231,232,233,234]. Thus, CNS infection by SARS-CoV-2 is possible.

Signs of neuropathology are also present after SARS-CoV-2 infection. Histopathological signs of neuroinflammation such as astrogliosis, microglial activation [219, 220, 235], BBB disturbance [235, 236], and infiltration of peripheral immune cells [219, 235] have been reported in COVID-19 patients. Stem cell studies have also reported microglial activation [237], cytokine production [231, 232, 237, 238], astrogliosis [223, 231, 239], altered neuronal morphology [229, 238, 239], synapse elimination [230], and neuronal loss [223, 230, 232] after SARS-CoV-2 infection. All of these processes are involved in AD. Even tau accumulation has been reported [240, 241], which is a hallmark of AD and other so called ‘tauopathies’.

Taken together, SARS-CoV-2 does display signs of neurodegenerative potential. However, it is possible that in vivo the neurodegenerative potential of SARS-CoV-2 is mediated by peripheral cytokine release and brain barrier disturbance (BBB, blood-CSF barrier) and cerebral hypoperfusion rather than by direct infection [236]. According to Matschke et. al, the severity of the histopathological changes they observed was not correlated with the presence of viral particles in the CNS which suggests that the peripheral inflammatory process is enough to cause CNS pathology [219]. Similarly, Käufer et al. have reported that intranasal SARS-CoV-2 infection caused microgliosis, tau hyperphosphorylation, and α-synuclein pathology in the hamster cortex even without CNS infection. Notably, the pathology persisted beyond virus clearance which could have implications for long-COVID and neurodegeneration [223].

Spirochete bacteria

Spirochetes are a group of spiral-shaped bacteria from the genera Spirochaeta, Treponema, Borrelia and Leptospira. Many members of this group are human pathogens that cause diseases such as leptospirosis, Lyme disease, relapsing fever, syphilis, and periodontitis. Cases for and against the involvement of spirochetes in neurodegenerative diseases have been raised.

For example, Miklossy et al. have suggested the involvement of spirochetes in AD [242,243,244]. They have reported that spirochetes, such as periodontal Treponema sp. and Borrelia burgdorferi, co-localize with Aβ and neurofibrillary tangles in AD brains, form plaque-like colonies, establish latent infections, disturb cerebral blood flow, and induce AD-related lesions and neuroinflammation [243, 244].

In vitro experiments by the same group on rat primary neurons, astrocytes, and microglia have shown that 2–8-week exposure to Borrelia burgdorferi induces AD-like changes, such as overexpression of the amyloid precursor protein (APP), tau hyperphosphorylation, and the accumulation of Aβ inclusions reminiscent of amyloid plaques. Interestingly, the presence of microglia increased Aβ accumulation, which highlights the importance of microglia in the neurodegeneration induced by spirochetes [242].

In contrast, Gutacker et al. found no evidence of Borrelia burgdorferi in the brains of ten AD patients [245]. Similarly, Forrester et al. found no geographic correlation between the incident rate of Lyme disease and death rate of neurodegenerative diseases in the United States [246]. However, neurodegenerative diseases are considered highly multifactorial, which makes such broad correlations difficult to assess. If it turns out multiple pathogen groups can trigger neurodegenerative changes in susceptible individuals, important correlations can get overlooked by studies focusing on one type of infection.

Syphilis is another pathogen that may be relevant to only a subset of patients. Syphilis is a sexually transmitted disease caused by the spirochete Treponema pallidum, which swept through Europe in the 1500s. It has since become rare due to effective antibiotic treatments. CNS presentation is relatively common at the early stages of syphilis [247]. It is characterized by sensorimotor symptoms such as abnormal gait, tremors, and numbness of the lower limbs accompanied by cognitive and mood dysfunction such as confusion, poor concentration, depression, and irritability. Interestingly, several cases of dementia caused by late neurosyphilis have been reported [248,249,250,251,252,253,254,255,256,257,258,259]. Commonly the patients are around 40–60-year-old at the time of diagnosis (early-onset for dementia) and they display a variety of neurological symptoms such as rapid cognitive decline, behavioral changes, and psychosis. In some patients, the symptoms ameliorate after treatment with antibiotics [248,249,250,251,252,253,254,255,256,257,258,259]. Thus, syphilis should be considered as a cause for dementia-like symptoms in infected individuals. Interestingly, asymptomatic neurosyphilis patients display different levels of Aβ42 and tau in the CSF compared to AD patients, which could be used as a biomarker to differentiate between the diseases as well as to determine the stage of neurosyphilis in a patient [260].

A few cases of suspected syphilis-induced ALS or ALS-mimic syndrome have also been reported [261,262,263]. However, determining whether these syndromes truly fill the criteria for ALS, and whether syphilis is causative in these cases, is beyond the expertise of the authors of this article [264]. Furthermore, anecdotal evidence suggests that other spirochete bacteria such as Borrelia burgdorferi can cause ALS-like pathology [265]. However, in general, Lyme disease does not seem to be associated with ALS [266, 267].

As for MS, one study has reported spirochete bacteria in the brains of four MS patients [268], but other evidence of spirochete involvement in MS is currently lacking.

Bacterial pneumonia

A link has been suggested between bacterial pneumonia and dementia. For example, a Taiwanese cohort study has reported a positive association between bacterial pneumonia and subsequent dementia risk (HR 2.83, 95% CI: 2.53–3.18), particularly between Staphylococcus pneumonia and vascular dementia (HR 5.4) and Hemophilus pneumonia and AD (HR 3.85, 95% CI: 1.66–8.96) [269]. Tate et al. have reported that hospitalization for pneumonia increased the risk of later developing dementia (aHR 1.9, 95% CI: 1.4–2.8, P < 0.0001) [270]. Neither study assessed whether the infection affected the CNS directly.

Chlamydia pneumoniae is a bacterial pathogen which causes respiratory infections, including pneumonia. It has been shown to sometimes enter the brain. Interestingly, Gérard et al. have reported C. pneumoniae in the neurons and glia of around 80–90% of late-onset AD patients and only 5–10% of healthy controls. Notably, the bacteria were detected in proximity to AD-related lesions such as amyloid plaques [164, 271, 272]. The authors did not list the cause of death for these patients. Since AD and dementia patients have a higher risk of hospitalization and mortality due to pneumonia than unaffected individuals [273, 274], more information would be needed to rule out the possibility that C. pneumoniae was present in the brains of these patients due to an infection secondary to AD.

However, Chacko et al. have shown that C. pneumoniae can infect the glia in olfactory and trigeminal nerves, olfactory bulb, and other brain areas of mice within 72 h of inoculation. They observed Aβ accumulation adjacent to bacteria in the olfactory system, and changes in neurodegeneration-relevant pathways [275]. Lopatko Lindman et al. have studied whether the Alzheimer’s disease risk allele APOE ε4 mediates the risk between C. pneumoniae infection and AD. They tested the presence of C. pneumoniae in the plasma of 360 AD and 360 matched controls from samples that had been collected on average 9.6 years before the diagnosis. They reported no association between Chlamydia pneumoniae, APOE ε4 allele, and the risk of AD [125]. In contrast, Gérard et al. have reported that C. pneumoniae burden is higher in the brains of AD patients that carry APOE ε4 allele compared to non-carriers [164]. Others have also reported that full length human APOE and derived peptides have direct antimicrobial function against Gram-negative bacteria [276, 277].

C. pneumoniae has also been studied in relation to many diseases with immunological etiology, including MS. However, reports have been published both for and against an association with MS, and clear conclusions cannot be drawn yet [278].

Finally, a study by Turkel et al. assessed the presence of Chlamydia pneumoniae in the serum of 51 PD patients and 37 matched controls. They found no statistically significant correlation between C. pneumoniae and PD [279].

Fungi

Fungi have received less attention in the study of neurodegenerative diseases than viruses and bacteria due to their lower relative abundance in the human microbiome. However, they are relevant human pathogens, particularly in immunocompromised individuals.

Most of the exploratory work on the involvement of fungi in neurodegenerative diseases has been done by a single research group lead by Luis Carrasco. They have detected multiple species of fungi in the brains of AD, PD, ALS, and MS patients.

For example, they detected Candida albicans, Candida ortholopsis, Candida tropicalis, Cladosporium, Malassezia globosa, Malassezia restricta, Neosartorya hiratsukae, Phoma, Saccharomyces cerevisae, and Sclerotinia borealis in the brains of altogether ten AD patients but not in the same number of controls. Not all species were detected in all patients [280]. Similarly, they found antibodies against multiple species of fungi, as well as fungal proteins, and fungal 1,3-β-glucan in the serum of AD patients. Such signs of fungal infection were also present in healthy controls but at lower frequency [281]. Consistently, Yashkin et al. have reported that fungal infections increased the risk of AD (HR 1.98, 95% CI:1.89–1.92) in a dataset collected from a 5% sample of U.S. Medicare beneficiaries aged 65 + years [282] (the number of Medicare beneficiaries was over 65 million in 2023 [283]).

The group of Luis Carrasco have also reported the presence of Botrytis, Candida, Fusarium and Malassezia species in the brains of all six PD patients using nested PCR analysis. When they compared these results to a previously published control group (n = 12) [281], the level of fungal infection was higher in PD patients. They also detected chitin immunopositivity in samples from PD patients but not from controls [284]. Interestingly, they detected bacteria in the spinal cord, medulla, and motor cortex of the same PD patients, which suggests the possibility of polymicrobial infection in PD. While bacteria were also present in healthy controls (n = 9) the results clustered separately in principal component analysis which suggests that the infection profile was different in patients and controls [284].

The same group has also reported fungal intracellular structures such as yeasts and hyphae in the motor cortex, medulla, and spinal cord of all eleven assayed ALS patients. PCR and next generation sequencing revealed fungi from multiple genera: Candida, Malassezia, Fusarium, Botrytis, Trichoderma, and Cryptococcus [285]. Bacteria were present in the spinal cord, medulla, and motor cortex of the ALS patients but not in healthy controls, which suggests the possibility of polymicrobial infection also in ALS patients [286].

Similarly, polymicrobial infections of fungi and bacteria were present in the post mortem CNS tissues of all ten MS patients but not in controls. Particularly, the fungi Trichosporon mucoides was found in most MS patients [287]. Furthermore, the group has detected fungal RNA, antigens, and antibodies against multiple Candida species [288, 289] in the CSF of MS patients [288]. They also detected fungal DNA and β-1,3 glucan in patient blood and serum samples, respectively [289]. The same group reported that antigens for Candida increased the risk of MS 3.0 – 7.3-fold (depending on Candida species) in a cohort of 80 MS patients and 240 matched controls [290].

As for the mechanisms of fungi-induced neurodegeneration, C. albicans has been shown to cause AD-like pathology in cell culture models, including Aβ oligomerization and accumulation [291]. Furthermore, a study by Wu et al. showed that intravenous infection of C. albicans in mice causes mild memory impairment which resolves after fungal clearance. The impairment was associated with the production of proinflammatory cytokines (IL-1β, IL-6, TNF-α) as well as localized accumulation of soluble Aβ peptides, activated microglia, and astrocytes around the yeast cells [291] which suggests that neuroinflammation is a key factor in the observed pathology.

Interestingly, a couple studies have shown Aβ to be protective against Candida infections. Even though the suggested mechanisms differed between studies (reviewed in more detail in later section), the main finding was the same: Aβ peptides can inhibit the growth of C. albicans in vitro [291, 292]. Furthermore, full length APOE and derived peptides may also have a direct antimicrobial function against pathogens such as Candida yeasts [276, 277]. In fact, Vonk et al. have reported that APOE knock-out mice display higher mortality following C. albicans infection than their APOE wildtype controls [293]. Together this suggests a possible interplay between fungal infections and AD risk factors.

The involvement of fungal toxins in neurodegeneration has also been suggested. Fungal toxins such as gliotoxin and fumonisin have been shown to destroy glia such as astrocytes and oligodendrocytes, which could potentially lead to myelin loss in MS [294]. Gliotoxin from Aspergillus fumigatus has been reported to induce neuroinflammation and to aggravate clinical symptoms in experimental autoimmune encephalomyelitis mice (the most common mouse model for MS) [295]. It has also been reported to penetrate hiPSC model of the BBB and to impair the barrier function as measured by transendothelial electrical resistance (TEER) and fluorescein permeability [296].

Together, these data show that fungal pathogens are worth considering when searching for causes for neurodegenerative diseases. However, more research on the topic is needed in the future to better understand fungal CNS infections in seemingly immunocompetent individuals. Furthermore, the important exploratory work done by Carrasco et al. should be replicated by others to confirm and further elucidate the role of polymicrobial infections and different fungal species in neurodegeneration.

Toxoplasma gondii

Toxoplasma gondii is an obligate intracellular protozoan parasite, which is primarily hosted by cats and other small mammals. It infects around one third of the global human population [65, 66]. It forms chronic, cystic infection inside neurons, astrocytes and microglia of the brain. Normally, the host immune response keeps the parasite in latent state, and clinical manifestations (e.g. encephalitis) are typically seen only in immunocompromised patients.

T. gondii has been implicated in various diseases from psychiatric and mood disorders to epilepsy and autoimmune disorders. Reports regarding its involvement in neurodegenerative diseases have been conflicting [297]. For example, a Taiwanese cohort study of 200 dementia patients and 400 controls has reported a positive association between toxoplasmosis and dementia (HR 2.878, 95% CI = 1.71–4.97, P < 0.001). Furthermore, the treatment of toxoplasmosis by sulfadiazine or clindamycin was associated with a reduction in dementia risk [298]. A meta-analysis of eight studies (3239 subjects) suggests that T. gondii may be involved in AD (OR = 1.53, 95% CI: 1.07–2.18). However, the authors were worried about publication bias and called for more research on the subject [299]. In contrast, Wennberg et al. detected no association between T. gondii seropositivity and cognitive performance in 575 adults aged 41–97 years [121].

T. gondii infection has been reported to induce production of proinflammatory cytokines (TNF-α, IL-1β, IL-6), glial nodules, and hyperalgesia in wild type mice [300, 301]. Interestingly, studies of T. gondii infection in AD model mice have reported beneficial effects such as higher levels of anti-inflammatory cytokines, reduced Aβ tissue deposition, increased Aβ phagocytosis and degradation, reduced nitrite production from primary cultured microglial cells, and better performance in memory tests [302, 303].

Multiple studies have suggested that T. gondii infections might be associated with reduced risk of MS [304,305,306,307]. Due to the autoimmune nature of MS, the ‘hygiene hypothesis’ has been suggested as an explanation for reduced disease risk in Toxoplasma carriers. The idea is that parasites promote immunotolerance in the host by suppressing the innate and adaptive immune responses against themselves. The IFN-γ driven immune response is the main host response to T. gondii infection. However, the parasite can modulate transcription factors that regulate this pathway, e.g. signal transducer and activator of transcription (STAT) 1, 3, and 6. It also inhibits multiple apoptosis pathways, ROS production by nicotinamide adenine dinucleotide phosphate (NADPH) oxidases, as well as the nuclear factor kappa B (NF-κΒ) pathway, which regulates the production of many pro-inflammatory cytokines [308]. Activation of these same pathways have been implicated in the development of MS, however, their roles at different stages of the disease remain under debate [309, 310].

The hygiene hypothesis of MS has been criticized because of the link to EBV infections. [158]. However, different pathogens may have a different effect on the immune system. In fact, protozoan parasites have been explored as potential therapies for autoimmune diseases and allergies [311, 312] while the data on herpesviruses points mostly towards increased risk of autoimmunity [313]. In any case, conflicting reports on the involvement of T. gondii in MS have been published [314,315,316,317] and thus, the question remains under debate.

Extra-CNS infections

Patients suffering from neurodegenerative diseases frequently display non-CNS infections such as gastric H. pylori infections [318,319,320,321], periodontal disease [322, 323], and changes in normal gut microbiome (dysbiosis) [324,325,326,327,328,329,330,331,332,333,334]. These changes have been linked to neurodegenerative diseases even in the absence of direct CNS infection [165, 335,336,337,338], and they share common mechanisms through which they can be detrimental to the CNS.

One key mechanism is the ability of these microbes to manipulate the extracellular matrix and the tight junctions of the host tissues. For example, gut microbes modulate the tight junctions of the gut epithelium by a variety of molecules. To illustrate, beneficial short-chain fatty acids (SCFAs), such as acetate, propionate and butyrate, are produced by fiber-digesting bacteria and they are utilized by host epithelial cells as an energy source. As a result, these bacteria facilitate normal epithelial barrier function [339, 340]. Furthermore, SCFAs are also known to cross the BBB where they modulate microglial function via G protein-coupled receptors [339]. Patients of neurodegenerative diseases often display reduction in the abundance of beneficial SCFA-producing bacteria [341, 342]. In turn, harmful gut microbes produce proinflammatory molecules such as LPS, cytokines, and prostaglandins, which disrupt the tightness of the epithelial barriers and can lead to a ‘leaky gut’. These same molecules can transfer to blood circulation and disrupt the BBB or cross into the CNS where they induce neuroinflammation [18,19,20].

Similarly, periodontal pathogens can produce molecules that disrupt BBB function [343]. For example, the Gram-negative bacteria Porphyromonas gingivalis and the spirochete Treponema denticola secrete molecules such as gingipains, toxic fimbriae, and dentilisin which allows them to modulate host tight junctions and to form a suitable habitat in the periodontal pocket [74, 344]. Interestingly, P. gingivalis has been associated with neurodegenerative diseases such as AD, PD and MS [165, 336,337,338, 345,346,347,348,349,350,351].

P. gingivalis has been detected in the post mortem brain tissue of AD patients. It has been reported to produce outer membrane vesicles carrying gingipains, LPS and toxic fimbriae, which can disrupt the BBB [345, 346]. Lei et al. reported that P. gingivalis infection and secreted gingipains increase transcytosis across the BBB by altering the expression of calveolin-1 and major facilitator superfamily domain containing 2a (Mfsd2a). They did not detect changes in occludin levels [344]. In contrast, Nonaka et al. have reported that P. gingivalis/gingipain disrupt tight junctions by direct degradation of zonula occludens-1 and occludin [344].

Once in the brain, gingipains cause AD-related changes such as as increase in Aβ and tau accumulation, modulate APP and tau processing, cause microglial activation and proliferation, induce IL-1β, TNF-α and IL-6 secretion, cleave APOE proteins, and induce neuronal degeneration and synapse loss [336, 346,347,348, 352]. P. gingivalis-derived LPS and inactivated bacteria also cause the secretion of nitric oxide and prostaglandin E2 from rat primary glial cells (95% astrocytes) [337]. Strikingly, peripheral administration of P. gingivalis is enough to cause a variety of neurodegenerative changes, which supports the involvement of indirect mechanisms such as outer membrane vesicles [165, 336,337,338]. For example, subcutaneous injection of P. gingivalis aggravated experimental autoimmune encephalomyelitis in MS model mice [337]. Oral P. gingivalis administration also led to microglial activation and loss of dopaminergic neurons in mice carrying the PD risk mutation LRRK2 R1441G. The mice displayed decreased expression of the tight junction protein zonula occludens-1 and increased expression of proinflammatory factors TNF-α, IL-1β, and α-synuclein in the colon. Increased serum IL-17A and increased brain IL-17A receptor expression implicates T helper (Th) 17 lymphocytes in the observed brain pathology [353]. Finally, P. gingivalis is not the only bacteria to produce harmful extracellular vesicles. Similar function has been attributed to the periodontal bacteria Aggregatibacter actinomycetemcomitans and H. pylori [335, 354,355,356].

Together, these results show that non-CNS infections can be as important for neurodegenerative diseases as CNS infections due to their potential to induce neuroinflammation. Thus, the search for causes for these diseases should not be limited to CNS infections, and non-CNS infections should also be considered when designing future treatments.

Neurodegeneration-related aggregating proteins display antimicrobial properties

Many neurodegenerative diseases display characteristic protein aggregation in the nervous tissue. For example, AD is hallmarked by the accumulation of Aβ plaques and neurofibrillary tangles formed by hyperphosphorylated tau. Similarly, PD is hallmarked by α-synuclein inclusions, ALS by TDP-43 and FUS aggregation, and frontotemporal dementia with tau, TDP-43 and FUS aggregates. TDP-43 aggregates are also reported in some cases of AD. The role of these protein aggregates is still under debate, but traditionally they have been thought to be toxic. However, in recent years many of these aggregating amyloid proteins have been reported to have antimicrobial effects which suggests a role in the innate immune response.

The role of Aβ as an antimicrobial peptide was first suggested by Soscia et al. in 2010 [292]. They showed that synthetic and rodent amyloid-β peptides are effective against Gram -negative and -positive bacteria, and C. albicans. According to them, Aβ42 peptides showed greater efficiency against microbes than the less amyloidogenic Aβ40 peptides. They also showed that Aβ is able to bind anionic bacterial membranes, and that temporal lobe homogenates from AD patients inhibited the growth of C. albicans more than controls. Homogenates from the cerebellum of AD patients did not have the same effect, which makes sense considering that the cerebellum is not commonly affected by AD. Finally, the dose-dependent reduction on C. albicans growth by the temporal lobe homogenates from AD patients was abolished by pre-incubation with antibodies against Aβ [292].

Since then, Eimer et al. have reported that the heparin-binding site on soluble Aβ oligomers bind herpesvirus glycoproteins and leads to Aβ deposition and viral entrapment [357]. Bourgade et al. have reported that Aβ40 and Aβ42 peptides inhibit the replication of HSV-1, an enveloped virus, but not that of non-enveloped human adenovirus [358]. Their results suggest that Aβ may exert its antimicrobial function through the pathogen surface membranes, and it may only be effective against certain pathogens.

The effectivity of Aβ against HSV-1 has been called into question by Bocharova et al. who reported that Aβ failed to protect AD model mice (5xFAD) from HSV-1 infection. These mice carry mutations in APP which leads to Aβ overexpression and early-onset AD in humans. Bocharova et al. did not observe any colocalization of Aβ aggregates and viral particles, or significant differences in survival rate between infected 5xFAD and wildtype mice. However, they did report that HSV-1 invasion was limited in areas of high Aβ burden which they attributed to increased phagocytosis by microglia chronically activated by Aβ [359].

This result is in line with Wu et al. who observed that synthetic and human Aβ40 and Aβ42 peptides inhibited the growth of C. albicans through an interaction with immortalized murine BV-2 microglia. Mainly, they observed that Aβ treatment increased the phagocytosis of yeast particles by BV-2 microglia and induced the secretion of presently unknown fungistatic compounds from the same cells [291]. Unlike Soscia et al. they did not observe a direct fungistatic effect of Aβ [291].

The study by Wu et al. suggests that Aβ facilitates the fast innate immune response. In short, they showed that a knockout of APP in mice impaired the clearance of C. albicans during the early days of the infection (days 4–7). The defect was associated with hypothermia and impaired secretion of proinflammatory cytokines. In contrast, C. albicans clearance was enhanced 5xFAD mice known for the overexpression of APP and Aβ. However, since all mice were able to clear the C. albicans infection by day 10, these data suggest that Aβ is not obligatory for fungal clearance but instead boosts the fast immune response [291]. Interestingly, Bocharova et.al. also observed better survival rate within the first 140 h (day 5) after HSV-1 infection in 7–10-month-old 5xFAD mice compared to wildtype littermates. However, the result was not statistically significant [359].

Taken together, these data support the role of Aβ as an innate immune alarm molecule that mediates acute response to microbial infection. Similar function has been suggested to other neurodegeneration-related aggregating proteins as we will see below.

The microtubule-associated protein Tau is involved in cytoskeletal organization. It has been linked to many neurodegenerative diseases termed ‘tauopathies’ due to its propensity to hyperphosphorylate and to form intracellular inclusions called ‘neurofibrillary tangles’. In AD, the accumulation of neurofibrillary tangles is thought to be secondary to Aβ pathology. Kobayashi et al. have reported that small synthetic peptides carrying microtubule binding sites of the tau 4R isoform display potent antimicrobial effects against Staphylococcus aureus and Escherichia coli. The effect was strengthened by the addition of tandem repeat sequences or sequences for nuclear localization or laminin receptor binding site. The addition of sequences for nuclear localization or laminin receptor binding site also made the peptides effective against C. albicans [360]. Similarly, Kanagasingam et al. have reported that some phosphorylated tau peptides, but not all, reduced the viability of the periodontal bacteria P. gingivalis. The effective peptides were more prone to form β-sheet structures than peptides that did not show antimicrobial effects [361].

Alam et al. report that α-synuclein, which is secreted from enteric nervous system neurons, is needed for normal peritoneal immune response. In their study, α-synuclein knockout mice displayed reduced immune cell invasion of peritoneal cavity and reduced production of proinflammatory cytokines compared to controls following injection with bacterial peptidoglycan. Furthermore, the exogenous administration of α-synuclein in the peritoneal cavity of the knockout mice promoted leukocyte recruitment which suggests that α-synuclein can act as an immune system alarm molecule (alarmin). In addition, α-synuclein aggregation has been found in the gut and it has been suggested that α-synuclein pathology could originate in the gut and travel to the brain via afferent neurons [362].

FUS and TDP-43 are RNA-binding proteins which are known to localize in stress granules [363, 364]: intracellular assemblies formed by RNA, ribosomes and RNA-binding proteins that repress protein translation during stress. Since viruses require the host cell machinery for replication, the repression of host protein synthesis can protect the system from the spread of the infection [365].

Interestingly, it has been reported that infection with the enterovirus coxsackievirus B3 induces the cleavage and translocation of both TDP-43 and FUS. Furthermore, the knockout of TDP-43 or FUS induces an increase in virus titer, suggesting these proteins can repress virus replication [366, 367].

The connection between FUS and the innate immune system is further illustrated by the fact that poly(I:C), stress granule-inducing respiratory syncytial virus, and type I interferon are all able to induce FUS assembly in in vitro cell lines [368]. FUS knockout causes a reduction in the production in type 1 interferon and other pro-inflammatory cytokines following similar infection-mimicking treatments [367].

Similarly, Dunker et al. have reported that TDP-43 is required for the regulation of the accumulation of immunostimulatory double-stranded RNA, and for the inhibition of interferon-mediated necrotic cell death [369]. Interestingly, a recent article by Licht-Murava et al. reports that aberrant TDP-43 accumulation in hippocampal astrocytes causes memory impairment by inducing interferon-mediated antiviral response [370]. It suggests that TDP-43 may be useful or harmful depending on the situation. Cabrera et al. have reported that latent HSV-2 infection in mouse spinal cord was not associated with changes in TDP-43 or FUS despite other signs of immune activation and ALS pathology such as leukocyte infiltration, microglial changes near motor neurons and reduction in C9orf72 levels [371]. Thus, a lot more research on the topic is needed.

Methodologies and future experimental avenues

In the future, population level association studies are expected to move from associations between a single pathogen and a neurodegenerative disease (e.g. HSV-1 and AD) to more complex models that consider a person’s lifetime infection burden and polygenic risk score [372]. Also moving the focus from severe infections towards more prevalent milder infections will further clarify the interaction between infections and neurodegenerative diseases. Such studies would require large-scale screening of the healthy population so that elusive patterns between asymptomatic or mild infections with neurodegeneration could be uncovered. When done consistently over long periods of time, such datasets could also elucidate the temporal patterns of disease development. However, such projects are labor- and cost-intensive, especially when both sexes and ethnically different human populations are represented in studies. The most feasible way to carry out such studies is their execution as part of large operations such as biobanks.

Simultaneously, experimental approaches are needed to advance from mere associations towards more mechanistic insight. Progressing from broad claims such as ‘infections cause neuroinflammation, which leads to neurodegeneration’ to clearer insight on why some individuals are affected while others are not, and to more detailed processes that lead to distinct neurodegenerative diseases, requires more studies where animals or cell culture models carrying different risk factors are infected with different pathogens. Including polymicrobial infections [373] and their combinations with different genetic risk factors into cell and animal models are likely to increase our understanding of possible heterogeneous susceptibility to infection-associated neurodegeneration.

Both animal models and stem-cell based methods have their advantages. For example, hIPSC-derived 2D and 3D models allow us to study infections on human brain cells, to overcome challenges related to species differences, and to elucidate the roles of different human cells types, e.g. neurons, astrocytes and microglia [374]. Furthermore, stem cell lines can be engineered to carry relevant genetic mutations, or they can be directly collected from affected donors e.g. AD patients carrying APOE mutations [374]. The limitation of in vitro stem cell models is that the peripheral immune component and the microbiome are often missing. Stem cell models also do not offer any information on whether the cell or molecular level change will penetrate to the level of behavior, e.g. cognitive function or motor symptoms. Xenotransplantation of human cells into mice (chimeric models) could solve this problem [375]. However, the xenotransplantation of human cells is currently only possible in immunocompromised animals [375] which limits the relevance of these models when studying infections. Thus, traditional in vivo animal models are still needed. For example, animal models of latent and reactivating virus infections have a lot to give to the field [27, 28].

Considering the importance of BBB disturbance in neurodegenerative diseases, improved models and detection methods are needed to elucidate the role of infections in this complex process. In humans, the measurement of BBB integrity in vivo is still emerging. However, imaging techniques such as positron emission tomography (PET) or magnetic resonance imaging (MRI) in combination with small molecule contrast agents show potential [376]. In animal studies, a wider range of dyes, diffusible tagged molecules, and in vivo and ex vivo imaging techniques are available [377]. In cell culture, trans-epithelial electrical resistance (TEER) and immunocytochemistry of tight junction proteins are common ways to study barrier integrity [378]. Stem cell-derived BBB-on-a-chip models are also under development [379].

Finally, the addition of pharmacological interventions into animal and cell culture-based models can be very powerful. Combined use of antimicrobials and anti-inflammatory drugs, such as resveratrol [380,381,382], curcumin [383], or omega-3 fatty acids [384], should tell us more about the role of each component in neurodegeneration. However, some compounds, such as resveratrol [382] and some antifungals [385], display both antiviral and anti-inflammatory properties. Thus, any experimental procedures need to be carefully planned.

Conclusions

The current literature shows that a broad range of pathogenic CNS and extra-CNS infections are linked to AD, PD, ALS, and MS. While some pathogens seem to be more strongly associated with certain diseases than others (e.g. EBV and MS), no direct causality can yet be assigned. Instead, the current data suggests that the neurodegenerative potential of CNS pathogenic infections is tied to the neuroinflammation which follows. The current understanding of the antimicrobial effects of neurodegeneration-associated aggregating proteins supports this view. Furthermore, many of us carry CNS infections without ever developing neurodegenerative disease, which suggests that a combination of genetic susceptibility factors and environmental triggers, such as infections, are needed for the development of neurodegenerative pathologies. Going forward, more research on the interplay between genetic risk factors and CNS infections is needed, as well as on the cumulative effect of multiple simultaneous or subsequent infections. While it is impossible to stop people from ever getting infections again, antimicrobial and anti-inflammatory drugs and dietary supplements have shown some promise in the treatment of neurodegenerative diseases. Thus, better understanding of the role of CNS infections in neurodegenerative diseases could help us detect vulnerable individuals and to target a combination of antimicrobial and anti-inflammatory therapies to those that could most benefit from them.