S100 Proteins in Alzheimer’s Disease

1Alzheimer’s disease (AD) is a chronic and progressive neurodegenerative disorder that affects wide areas of the cerebral cortex and hippocampus. Most AD patients (>95%) are idiopathic and disease is characterized by late onset (80–90 years of age) with failure in the clearance of amyloid-β peptide (Aβ) from the brain. The main symptoms of the disease are progressive memory deficits, cognitive impairment, and personality changes. The neuropathological and neurochemical hallmarks of AD include selective neuronal death, synaptic loss and the presence of proteinaceous deposits in the extracellular space (known as diffuse and neuritic plaques) as well as inside neurons [known as neurofibrillary tangles (NFTs)]. Neuroinflammation, oxidative stress, and calcium dysregulation are also important features implicated in AD pathology.

2Diffuse and neuritic plaques, most commonly known as amyloid plaques, are mainly constituted by Aβ deposits, surrounded by degenerative presynaptic endings, astrocytes and microglial cells. Aβ peptides are formed from proteolytic cleavage of the amyloid precursor protein (APP) by the γ- and β-secretases (BACE1). Even though the normal function of APP is not known, it is possibly related to regulation of neurite outgrowth, cell adhesion, and neuron migration. APP processing can involve non-amyloidogenic or amyloidogenic pathways. When APP is cleaved by α-secretase and subsequently by γ-secretase, sAPPα is predominantly formed, which has an important role in neuronal plasticity and survival. However, in the amyloidogenic pathway, APP is cleaved by β- and γ-secretase producing sAPPβ, C-terminal fragments and Aβ peptides, which promote a range of detrimental effects in neurons and in the brain. Aβ40 and Aβ42 are the predominant accumulating peptides, whose aggregation into fibrillar cross-β structures is a central feature in AD pathogenesis. Aβ aggregation is naturally heterogeneous and monomers assemble and polymerize into structurally distinct forms, including protofibers, polymorphic oligomers and amyloid fibrils, all found within plaques. Extracellular accumulation of Aβ fibrils is not necessarily intrinsically cytotoxic and emerging evidence suggests precursor oligomers as the key toxic agents, also because of their seeding potential. Moreover, Aβ peptides can also be deposited intracellularly.

3The presence of neurofibrillar tangles, formed by neuronal intracellular deposition of hyperphosphorylated tau protein, is also a major AD hallmark. It has also been suggested that NFT may not be the major player in neurotoxicity, and that tau oligomers are in fact the major toxic forms promoting synaptic impairment. In agreement, it has been described that propagation of tau pathology occurs trans-synaptically. Other factors that contribute to tauophaties, besides tau hyperphosphorylation, are tau truncation, glycosylation, glycation, nitration, and ubiquitination.

4Additionally, metal ion homeostasis and calcium signaling are also implicated in disease pathogenesis. In the early stages of AD, calcium imbalance promotes Aβ formation and tau hyperphosphorylation, as reviewed previously in. Aβ destabilizes neuronal calcium homeostasis generally leading to an increase in cytosolic calcium levels and formation of calcium-permeable pores. Calcium alterations lead to cytoskeletal modifications, triggering neuronal apoptosis and formation of free radicals through mitochondrial dysfunction. Moreover, familial AD mutations in presenilins are linked to altered synaptic Ca2+ signaling that imbalance the activities of Ca2+-calmodulin-dependent kinase II (CaMKII) and Ca2+-dependent phosphatase calcineurin (CaN), increasing the long-term depression and causing memory loss. Transition metals such as Zn2+, Cu2+, and Fe2+ have well-established roles as chemical modulators of protein folding, amyloid aggregation and toxicity and are found to accumulate at protein deposits.

5Neuroinflammation is another cellular process linked to AD pathogenesis. Senile plaques are often closely associated with activated microglial cells and surrounded by activated astrocytes that have abundant filaments. In response to Aβ deposition, activated microglia upregulate the expression of cell-surface proteins and cytokines such as the tumor-necrosis factor (TNF), interleukine-6 (IL-6), interleukine-1 (IL-1), S100 proteins, and chemokines. The presence of Aβ activates different cell receptors and intracellular signaling pathways, mainly those related to the receptor of advanced glycation end products (RAGE)/nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) pathway, that is responsible for the transcription of pro-inflammatory cytokines and chemokines in astrocytes. Additionally, astrogliosis appears as an early manifestation of AD. The migration of astrocytes into Aβ plaques is promoted by chemokines CCL2 and CCL3, that are released by activated microglial cells surrounding amyloid plaques. Astrocytes recruited to Aβ plaques have the potential to mediate both neurotoxicity and participate in the clearance of Aβ. S100 proteins are among the alarmins that are upregulated and are highly secreted by astrocytes during this process, which results in their accumulation within Aβ deposits and brain tissues, as overviewed in the following sections.

6S100 proteins are a family of low-molecular weight EF-hand Ca2+ proteins that are expressed in distinct organs and tissues. They are involved in multiple intracellular functions, including cell proliferation, differentiation, protein phosphorylation, cytoskeletal assembly, and disassembly and intracellular calcium homeostasis. In pathological conditions S100 proteins can be expressed in a cell type where they are not expressed under normal conditions. Additionally, some S100 proteins are secreted and regulate cell functions in an autocrine or paracrine manner by activation of surface receptors, such as the RAGE receptor, thereby promoting NF-κB signaling, an important trigger of inflammatory processes, recruiting and activating cellular pro-inflammatory effectors. Albeit S100 proteins are not cytokines in stricto sensu, in these cases, they have such functions, and act as extracellular alarmins or as damage-associated molecular pattern (DAMP) factors, that can either be beneficial or detrimental depending on concentration and molecular and cellular moiety. From the 25 S100 proteins described so far, several are present in the brain and from those, seven have been implicated in AD pathways: S100B, S100A1, S100A6, S100A7, S100A8, S100A9, and S100A12.

7S100 proteins occur mainly as homodimers. Specifically, it is known that calcium binding to S100 proteins triggers conformational changes that expose a hydrophobic cleft that is crucial to interaction with partners to their activation, regulation and signaling functions. Several S100 proteins also bind zinc and copper, which interestingly are highly abundant in senile plaques. A few S100 proteins are also found as heterodimers, including S100A8/A9, S100B/A1 and S100A6/B. S100 proteins interconvert into functional oligomers, including tetramers, hexamers, and octamers and formation of these species can be promoted by Ca2+ or Zn2+ binding. The functions described for these S100 oligomers involve a tighter interaction with RAGE, assistance in microtubule formation, neurite outgrowth, and tumor suppression.

8S100 proteins influence cognitive processes in the healthy brain and play roles in development and neuronal maintenance. Depending on the study, antisera to neurospecific S100 protein and antibodies directed against S100A1 and S100B either impair or improve learning and memory functions in rat brains. Recent studies suggested that elevated S100B levels have deleterious effects during the neurodevelopmental period through RAGE-dependent processes. Additionally, S100B has been associated with Down Syndrome, a genetic variation where the most profound neurological features are mental retardation, seizures and early onset AD. Levels of S100B are increased in neuronal progenitor cells of patients with Down Syndrome and in human induced pluripotent stem cells derived from Down Syndrome patients. S100B overexpression causes toxicity to neurons, reduces neurogenesis, and increases the production of reactive oxygen species.

9There are several studies using clinical specimens and animal models implicating S100 proteins in AD pathophysiology. As overviewed in the following sections, different S100 proteins seem to be involved in several processes related to APP processing, influencing Aβ levels, tau post-translational modifications, formation of protein inclusions, and multiple signaling pathways. The ability of acting as Ca2+ sensors, regulating the activity of channels/pumps involved in Ca2+ release/uptake also provide feedback and feedforward mechanisms for sustaining aberrant Ca2+ signaling in AD. Therefore, involvement in all these processes makes a strong case for the importance of the S100 protein family in AD development (Table 1).

10S100B is the most studied S100 protein in the scope of AD, as reviewed in and. S100B acts as a pro-inflammatory cytokine and a DAMP molecule depending on its concentration. S100B secreted from astrocytes can have trophic and toxic effect on neurons. At nanomolar concentration, S100B displays neurotrophic effects, leading to promotion of neuronal survival and neurite outgrowth. At micromolar concentrations S100B has deleterious effects inducing neuronal apoptosis. Upregulation of S100B leads to behavioral abnormalities and loss of dendritic density in mice. S100B also regulates the intracellular levels of free calcium in several central nervous system cell types, such as neurons and astrocytes. Recently we demonstrated that S100B acts as a sensor and regulator of elevated zinc levels in the brain and that this metal-buffering activity is tied to a neuroprotective role, through an indirect effect on calcium levels and in inhibition of excitotoxicity.

11Several studies point to high levels of S100B in AD patients and in AD mouse models. The largest increase in S100B levels is observed in the hippocampus, temporal lobe and in the layer I of the cortex. It is demonstrated that sera of AD patients with moderate and severe dementia have a 60- and 37-fold increase in S100B, respectively. In AD patients with moderate dementia, the increase of S100B levels is followed by a 10-fold increase in auto-antibodies: however, in AD patients with severe dementia the levels of auto-antibodies remain identical to controls, indicating that there is no immune-protection against elevated S100B levels in AD patients with severe dementia. S100B is also elevated in the cerebrospinal fluid (CSF) of AD patients. S100B is involved in APP cleavage processes: high levels of S100B increase BACE1 activity resulting in higher levels of toxic APPβ and C-terminal fragments, including the amyloidogenic β-CTF (C99).

12S100B surrounding amyloid plaques is mostly produced by astrocytes, but can also originate from oligodendrocytes and microglia. It was also observed that S100B positive astrocytes are present in the diffuse non-neuritic amyloid plaques, suggesting an early, yet unclear, action of S100B in the formation of senile plaques. Evidence suggests that S100B may regulate plaque formation as the knockout of S100B in the PS/APP AD mouse model selectively decreases plaque load in the cortical region and the overexpression of S100B increases Aβ levels and deposits, at early stages. Even though, it is established that elevated levels of S100B have deleterious effects that promote AD features, nanomolar concentrations of S100B effectively protect cells against Aβ-mediated cytotoxicity. Additionally, we have recently found that, in vitro, S100B binds to Aβ42 monomers, oligomers, and fibrils resulting in a calcium-tuned suppression of Aβ42 aggregation and cellular toxicity in SH-SY5Y cells. S100B was found both in normal and in AD brains in various oligomeric states; however, the protective or pathological functions of S100B oligomers are still unclear.

13Overexpression of S100B in Tg2576 AD transgenic mice is also linked with neuroinflammation, promoting astrogliosis, microgliosis, and neurite proliferation. However, knockout of S100B in PS/APP mouse model decreases cortical gliosis. IL-1 regulates the expression and secretion of S100B from astrocytes. Treatment of the 3XTg-AD mice with an antibody against IL-1 reduces S100B levels and results in attenuation of tau pathology and in partial reduction of certain fibrillar and oligomeric forms of Aβ Therefore, S100B seems to be tied to different processes related to AD pathology as in addition to its ability to promote brain inflammatory response and tau pathology it may play roles in directly promoting amyloidogenic APP processing, as proposed by.

14Inhibition of S100B using pentamidine in AD mouse models, lead to a reduction in the levels of proinflammatory mediators such as nitrite, MDA, PGE2 and IL-1, followed by an inhibition of Aβ-induced gliosis. Indeed, S100B-overexpressing mice that were infused with oligomeric Aβ exhibited enhanced glial activation. Neuroinflammation and loss of synaptic markers were noted, however there was no difference in the amyloid plaque burden, in comparison to controls. These results suggest a relationship between S100B and other cytokines that are also implicated in AD pathways. Indeed, the TNF-α cytokine, which is present at high levels in the AD brain, decreases both GFAP and S100B intracellular levels in astrocytes, while increasing their extracellular levels. This crosstalk suggests a relationship between TNF-α and the increase of these two proteins in CSF and sera of AD patients. Other reports showed that S100B enhances IL-6 mRNA and IL-1 mRNA levels in microglia and in neurons, via Sp1 and NF-κB signaling pathways. Additionally, knockout of S100B in the Tg2576 AD mouse model background results in a decrease in GFAP-positive astrocytes and in Iba-1 positive microglia, while its overexpression has opposite effects. Overall, these results suggest that S100B can influence and be influenced by the levels of other cytokines involved in AD pathogenesis.

15Regarding tau pathology, high S100B levels in AD patients positively correlate to tau tangles with which S100B was found to be clustered. Knockout of S100B in the PS/APP mouse model decreases phosphorylated-tau positive dystrophic neurons and in mouse models expressing tau, S100B levels are upregulated. Indeed, it has been demonstrated, in vitro, that S100B directly binds tau inhibiting its phosphorylation by yet unclear non-covalent interactions in a process that is Ca2+ or Zn2+-dependent. However, contradictory results show that S100B promotes tau hyperphosphorylation by inducing GSK-3β activation and disrupting Wnt signaling, an important pathway to regulate synaptic transmission and plasticity. Indeed, S100B promotes the expression of the Dickkopf-related protein 1 (Dkk1), an antagonist of Wnt signaling that has previously been suggested to play a role in AD.

16As previously mentioned, calcium dysregulation contributes to AD pathology and S100B is a key factor in the Ca2+ homeostasis of astrocytes. It was demonstrated that S100B knockout leads to a decrease of induced-Ca2+ transients such as those induced by Aβ. In what could be a potentially protective mechanism, S100B levels were found to be up-regulated in astrocytes upon Aβ induced Ca2+ intracellular waves.

17The investigation of the role of S100A1 in AD is encouraged by the fact that some of its targets are altered in the disease, such as the ryanodine receptor (RyR), an intracellular calcium release channel, tau and RAGE. S100A1 is primarily expressed in neurons and, as reviewed in is implicated in tau phosphorylation, neuronal cell sensitivity to Aβ and in the regulation of APP expression. In respect to the latter, available data indicates that βAPP steady-state mRNA and intracellular protein levels are down-regulated in response to ablation of S100A1 expression. In the PS/APP mouse model, knockout of S100A1 decreases inflammatory processes, such as astrocytosis and microgliosis, diminishing 3.7-fold the number of cortical plaques and 1.5-fold the number of hippocampal plaques. Decreased S100A1 levels in PC12 cells increase tubulin levels and the number of neurites. Additionally, knockout of S100A1 in PC12 cells increases the resistance to Aβ-induced cell death.

18S100A1 induces Glycogen synthase kinase 3 (GSK3) phosphorylation, that is involved in several processes such as glycogen metabolism and gene transcription. GSK3 over-activation is also related to memory impairment and other AD related features. In human and mouse AD brain tissue, S100A1:RyR complexes are present and their formation is Ca2+-dependent. RyR is a receptor with altered levels in AD that is associated with APP processing and Aβ production, however it is not known if it exerts a protective or pathogenic role in AD. S100A1 also binds to stress-inducible phosphoprotein 1 (STIP1), a Hsp90 cochaperone that is reported to be present in the vicinity of Aβ oligomers, preventing Aβ-induced synaptic loss and neuronal death in primary neurons. S100A1 and S100B also have the ability to cause microtubule disassembly in glioma cells and myoblasts in a Ca2+-dependent manner, suggesting a possible role of S100A1 in tau pathology. Moreover, in human AD patients and in the PS/APP mouse model, extracellular S100A1 has been observed in plaque-like deposits.

19S100A6 was identified in the AD gene signature as one of the most significantly positively correlated proteins with AD phenotype. As other S100 proteins, S100A6 is upregulated in AD patients and in AD mouse models and is found in astrocyte-positive clusters that surround Aβ amyloid deposits in the brain’s gray matter. In PS/APP mouse brains, S100A6 localizes in the peripheral region of amyloid plaques and exogenous S100A6 treatment in mouse brain sections reduces Aβ levels and plaque burden.

20Zinc ions are colocalized with senile plaques in AD patients and there is evidence that AD related-cognitive decline is dependent on extracellular zinc levels. In particular, one study suggested that zinc-binding S100A6 exerts a zinc sequestering function, identical to what has been proposed for S100B, thus preventing zinc-induced toxicity in COS-7 cells. Additionally, PS/APP mice treated with a high-zinc diet have increased S100A6 levels and Aβ deposits. These studies point to a correlation between S100A6, zinc ions and decrease in Aβ plaque load.

21The heterodimer S100A6/B is also implicated in pathological signal transduction in melanoma. It is possible that the formation of the heterodimer also occurs in AD since S100A6 is colocalized with S100B and astrocytic glial fibrillary acidic protein (GFAP), a marker of astrogliosis, near amyloid plaques. Additionally, it is reported that S100A6 binds to the CacyBP/SIP complex, a complex known to participate in the organization of microtubules. Overexpression of S100A6 in neuroblastoma NB2a cells inhibits CacyBP/SIP complex activity, and consequently lowers the rate of tau dephosphorylation.

22There is scarce evidence regarding the role of S100A7 on AD pathways. It is reported that S100A7 is increased in mildly cognitively impaired patients and in the brain and CSF of AD dementia patients. S100A7 mRNA expression is regulated in the brain as a function of AD dementia and amyloid neuropathology. Exogenous S100A7 in primary hippocampal neurons of Tg2576 AD transgenic embryos inhibits the generation of Aβ42 and Aβ40 peptides and promotes the activity of “non- amyloidogenic” α-secretase, via upregulation of ADAM-10 (a disintegrin and metalloproteinase) and phosphorylation of Erk1/2 and PKC. Therefore, a beneficial role of S100A7 on APP processing is suggested, albeit other studies are required to more extensively support this possibility.

23S100A8 was found to be upregulated in the sera of AD patients and in the hippocampus of Tg2576 and TgAPParc AD mice. Indeed, several studies establish a correlation between Aβ and S100A8 expression. An increase in S100A8 mRNA levels was induced when aggregated Aβ was added to a microglia culture isolated from post-mortem AD brain tissues. Subsequent culture growth suggested that chronic secretion of S100A8 can lead to chronic activation of microglia. In rat primary astrocytes, Aβ42 treatment induces a significant increase in S100A8 mRNA levels. Treatment of SH-SY5Y neuroblastoma cells with recombinant S100A8 increased Aβ42 and decreased Aβ40 production. In PS/APP mice, the S100A8/A9 heterodimer is found to be upregulated in microglial cells surrounding amyloid plaques. Additionally, it is reported that S100A8/A9 binds directly to Aβ40 and that it interferes with amyloid formation but no effect was observed over Aβ42 aggregation. There is also a link between S100A8/A9 and the “non-amyloidogenic” α-secretase ADAM-10, since S100A8/A9 has lower expression in AD mice models overexpressing ADAM-10.

24S100A8 assemblies were found in the hippocampus of Tg2576 and TgAPParc AD mice brains, distinct from corpora amylacea, that are formed independently of Aβ plaques. These S100A8 aggregates are likely not amyloidogenic as no staining was observed with thioflavin-S.

25S100A9 was found to be strongly increased in brain lysates of AD patients and AD mice compared to healthy, age-matched controls and in familial PS-1 AD tissues. S100A9 is present in activated glia and neurons positive for tau neurofibrillary tangles. Additionally, there is a strong correlation between S100A9 and Aβ. In in vitro cell assays, Aβ42 reduces extracellular release of S100A9 in human THP-1 monocytes and induces S100A9 expression in microglia BV2 cells. However, in CSF from AD patients with mild cognitive impairment and vascular dementia, the levels of S100A9 and Aβ42 are decreased. Knockdown of S100A9 decreases cognition decline on Tg2576 mice and reduces amyloid plaque burden. S100A9 was found within amyloid plaques of sporadic and familial PS-1 AD brains with distinct Braak stages from III to VI. Indeed, in some studies it was possible to observe Aβ42 plaques and also isolated S100A9 plaques that are not colocalized, forming separate tissue deposits.

26Regarding the formation of S100A9 puncta in AD brain, a recent study reported that, in vitro, S100A9 is able to form polymeric structures that resemble amyloid structures and bind amyloid fluorophores. The polymerization reaction occurs through a two-step nucleation with initial misfolding of S100A9 and β-sheet formation. The impact of S100A9 oligomers on memory was studied through intranasal administration of S100A9 (dimers, oligomers, and fibrillar states) in aged mice. S100A9 oligomers and fibrils, but not dimers, evoked amnestic activity which is correlated with disruption of dopaminergic and glutamate neurochemistry in the prefrontal cortical and hippocampal regions. Additionally, intranasal administration of S100A9 induces cellular stress in the frontal lobe, hippocampus, and cerebellum of aged mice, as well as impaired learning. However, co-treatment with S100A9 fibrillar species and glutamate antibodies increases locomotor activity. In Tg2576 AD mice, knockdown or knockout of S100A9 significantly reduced the neuropathology and greatly improved learning and memory, suggesting a link between S100A9 and AD pathology. Indeed, knockout of S100A9 in the PS/APP AD mouse model led to increased phagocytosis of fibrillar Aβ and to decreased Aβ deposition.

27However, several studies focused on the relationship between S100A9 and Aβ42. In vitro biophysical approaches showed that S100A9 binds to Aβ40 through hydrophobic interactions. Kinetic assays suggested that S100A9 co-aggregates with Aβ40, promoting the formation of amyloid fibrils. The co-aggregation of S100A9 with Aβ42 was also referred to inhibit Aβ42 cytotoxicity.

28Regarding APP processing, it was found that C-terminal fragments of amyloid precursor-like protein 2 (APLP2) upregulate S100A9 protein and mRNA expression in BV2 cell and that inhibitor of γ-secretase promotes downregulation of S100A9 protein levels. AD mice deficient in S100A9 have decreased levels of key cytokines involved in APP processing and a reduction of BACE1 expression and activity. S100A9 knockout also reduced overall levels of Aβ and APP C-terminal fragments in Tg2576 AD mice, due to increase in neprilysin levels and decreased BACE activity. Knockdown of S100A9 significantly attenuated the increase of Ca2+ levels provoked by C-terminals of APP or by Aβ treatment; however, others observed that a reduction of S100A9 extracellular release is followed by an increase in intracellular Ca2+ levels, evidencing a correlation between S100A9 and calcium dysregulation in AD. In the AD brain and in mouse models, S100A9 is present in its native form but also as large complexes ranging from 90 to 190 kDa. Indeed, after intranasal administration of S100A9 fibrils in aged mice, S100A9 plaques were observed in the brain which resulted in an exacerbation of cell stress. Overall there are solid evidences regarding S100A9 as a potential regulator of AD pathways.

29S100A12 is the least studied S100 protein in the context of AD. To date, a single study reported S100A12 hexamers associated with senile plaques, reactive glia and neurons in brains of sporadic and PS-1 AD patients.

30Considering the involvement of S100 proteins in multiple regulatory functions in the brain, the fact that they have age- and damage- related expression, and a direct involvement in neuroinflammation, it is not surprising that they are implicated in molecular processes associated with AD pathogenesis (Figure 1).

31Most of the available studies report essentially deleterious effects of S100 proteins in AD pathological processes. Indeed, elevated levels of S100 proteins around amyloid plaques and neurofibrillary tangles, exacerbate neuroinflammation and interfere with APP processing and with several AD-related proteins and signaling pathways. This scenario is compatible with an aggravating role as part of downstream inflammatory processes in later stages of AD neurodegeneration. However, some protective roles of S100 proteins are emerging as equally important. Indeed, it is now proposed that early inflammatory responses start at very early stages of the AD neurodegenerative process, most likely before amyloid plaques are formed. This process would involve an elevation of inflammatory molecules, including S100, at the earliest stages of amyloid pathology, even before the onset of the disease phenotypes. Interestingly, several recent studies suggest that low levels of S100B protect cells against Aβ42-mediated cytotoxicity and that low levels of S100A9 inhibit Aβ42 cytotoxicity. This has led to the recent proposal of a new chaperone-like function for S100B, which seems to be extensible to a metal buffering activity, both of which are certainly relevant in the context of AD. Therefore, also in AD, S100 proteins may exert different functions according to their (extra)cellular concentrations. At early inflammatory stages and relatively low concentrations they play protective roles, while later in pathology, at higher concentrations, they play essentially disease aggravating roles. This is in line with prior evidence that suggests that decreasing the levels of S100 proteins may be a strategy to mitigate AD progression. The development of S100 neutralizing antibodies and small molecules is already a current therapeutic approach in cancer, autoimmune diseases and chronic inflammatory disorders, as reviewed in, and could present a promising strategy for AD as well. Ongoing research in our laboratory will shed light into this subject in the near future.

32All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

33The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.