Background
Alzheimer's disease (AD) is a neurodegenerative disorder and is the most common cause of dementia among the elderly. Accumulation of amyloid-β (Aβ) peptides in the brain is considered to be a key step in the pathogenesis of the disease and leads to formation of amyloid plaques in brain parenchyma. The Aβ peptides can be truncated at both the C- and N-terminal ends, and also undergo posttranslational modifications. Although Aβ1-40 (40 amino acids long) is the most abundant form, the major focus is on Aβ1-42 which is more prone to aggregate and considered to be the most neurotoxic form. Aβ is found in different aggregation states in the brain ranging from monomers and non-fibrillar aggregates, termed oligomers, to a highly fibrillar form found in the deposits. Recent evidence suggests that diffusible Aβ oligomers have the most toxic properties [
1,
2]. However, it should also be noted that Aβ fibril-containing senile plaques precede the development of dystrophic neurites [
3] and of spinodendritic calcium decompartmentalization that presumably leads to cognitive dysfunction [
4]. In addition to massive neurodegeneration, chronic neuroinflammation is a pathological hallmark of AD, manifested by activated microglia and reactive astrocytes. Accumulation and deposition of Aβ can trigger activation of glial cells, which will set off an inflammatory response that, over time, becomes chronic causing a persistent deleterious condition [
5].
The role of neuroinflammation in the development and progression of AD is, however, not clear. Neuroinflammation is often referred to as a "double-edged sword". On the one hand microglia and astrocytes secrete inflammatory cytokines, chemokines and neurotoxins upon activation, and can thereby promote neuronal degeneration. On the other hand, activated microglia surrounding Aβ plaques may have beneficial effects by phagocytosis of, and thus elimination of, Aβ [
6]. Astrocytes have also been reported to be able to migrate towards Aβ plaques and, upon contact, to degrade Aβ [
7,
8]. This somewhat confusing picture calls for delineation of signaling pathways that may be involved in the beneficial effects of neuroinflammation or that may promote neurodegeneration.
The inflammatory response is, to a large degree, orchestrated by the transcription factor nuclear factor κB (NF-κB). However, NF-κB works in concert with other transcription factors. Of particular interest are members of the CCAAT/enhancer binding protein (C/EBP) family that can amplify the effects of NF-κB and may also form heteromeric complexes with NF-κB [
9‐
11]. C/EBP is a protein family consisting of six members, C/EBPα-ζ (reviewed in [
12]). In order to be active, C/EBPs will form homo- or heterodimers with each other or with other transcription factors. Until recently, C/EBP studies have mainly focused on the liver, where these proteins regulate expression of a variety of genes including acute phase proteins [
13]. However, C/EBPα, β and δ are also all expressed in brain [
14,
15]. Among their target genes are many pro-inflammatory cytokines including interleukin- (IL-) 6 (an early inflammatory marker in AD brain), inducible nitric oxide synthase, complement factors, and cyclooxygenase-2 (COX-2) [
16]. In both brain and liver, inflammatory stimuli have been shown to, in general, down-regulate expression of C/EBPα and up-regulate expression of C/EBPβ and δ [
15]. Several previous studies have suggested that members of the C/EBP transcription factor family are dysregulated in AD and in response to Aβ. mRNA levels of C/EBPα, β and δ have all been shown to be up-regulated in hippocampus of AD patients [
17‐
19] and protein levels of C/EBPδ have also been reported to be up-regulated in AD brain [
20]. However, in a recent study we observed that C/EBPδ DNA binding activity to a C/EBP motif is, instead, completely blocked in IL-1β-stimulated primary astro-microglial cultures from rats after exposure to oligomeric forms of Aβ peptides [
21]. It can be speculated that Aβ peptides cause an imbalance between NF-κB and C/EBP transcription factors that could result in abnormal responses to inflammatory stimuli. To further investigate how the inflammatory response in the AD brain may be affected we have analyzed the effects of different aggregation states of Aβ on NF-κB-C/EBP cross-talk, and also C/EBPα and δ expression levels in activated primary glial cultures. In addition, we have analyzed the expression levels of C/EBPα, β and δ and in mice with high levels of fibrillar Aβ deposits.
Discussion
In this study we observed that IL-1β-induced expression of C/EBPδ was inhibited in rat primary astro-microglial cultures after exposure to Aβ. However, this effect is dependent on the presence of fibrillar forms of Aβ and no effect on protein levels was observed after exposure to an oligomer-enriched Aβ preparation. In general, C/EBPδ is up-regulated in response to inflammatory stimuli. Our results instead indicate an Aβ-dependent decrease in C/EBPδ levels. This decrease of C/EBPδ was further confirmed using an AD transgenic mouse model, tg-ArcSwe mice, characterized by high levels of fibrillar Aβ deposits [
25]. When analyzing brain areas of aged mice carrying a high Aβ load, C/EBPδ was significantly down-regulated compared to the same brain areas from aged non-transgenic littermates. This effect on C/EBPδ levels is not the expected result since up-regulation of inflammatory markers has been detected in other AD transgenic mice models (cf., [
34]). An Aβ-induced inflammatory response in tg-ArcSwe mice is, however, supported by down-regulated C/EBPα levels and up-regulated C/EBPβ levels in forebrain from aged tg-ArcSwe mice.
Differential effects of oligomeric and fibrillar forms of Aβ on glial cells have been reported earlier: Rat astro-microglial cultures show significantly higher levels of inflammatory markers, such as tumor necrosis factor (TNF)-α, when exposed to oligomeric compared to fibrillar Aβ [
35]. In microglial cells TNF-α induction is typical for a classic cytotoxic phenotype, whereas an alternative activation state is not correlated with expression of inflammatory cytokines and is instead characterized by increased Aβ phagocytic capabilities. A recent study in a
PS1xAPP transgenic AD model showed that hippocampal microglial cells switch from the alternative activation state to the classical cytotoxic phenotype during aging [
36]. Interestingly, only microglia not located in the vicinity of Aβ plaques express this classical cytotoxic phenotype in 18-month-old transgenic mice. Microglial cells surrounding plaques are instead TNF-α negative. The same authors also showed that both oligomeric Aβ and soluble extract from 18-month-old transgenic mouse hippocampus produces potent TNF-α induction in astro-microglial cultures from non-transgenic mice [
36]. One possibility is that increased expression of C/EBPδ is a marker for classical cytotoxic glial cells and that the fibrillar form of Aβ actually blocks induction of this phenotype. This could also explain the decreased levels of C/EBPδ that we observed in brain areas with high Aβ load in aged tg-ArcSwe mice, since the Arctic mutation not only increases the rate of Aβ protofibril formation but also fibril formation and senile plaque deposition [
26,
37].
As indicated above, increased levels of C/EBPβ and δ are expected when an inflammatory response is induced. It is believed that during inflammation the rapidly and transiently activated NF-κB pathway is crucial for a primary wave of gene induction and that a second and more long-term wave of gene transcription is mediated by other transcription factors including the C/EBP family (cf., [
12,
38,
39]). Members of the C/EBP family can form heterodimers with NF-κB subunits [
9,
11]. Previous studies have indicated a reciprocal cross-coupling between NF-κB and C/EBPs: NF-κB and C/EBPs seem to synergistically activate promoters with C/EBP sites, and inhibit promoters with κB motifs [
10,
38,
40]. In this study we show that, in astro-microglial cells, IL-1β induces binding of C/EBPδ-containing complexes to a κB site. It could be speculated that this is a part of a negative feed-back mechanism regulating the effect of NF-κB activation. If this is true, the inhibition of C/EBPδ binding to the κB motif that we observe after exposure to Aβ could result in prolonged effects of NF-κB acting at this site.
When the astro-microglial cells were exposed to fibril-enriched Aβ preparations the IL-1β-induced binding of C/EBPδ to both the C/EBP and the κB motifs were more or less abolished. This could be expected since the IL-1β-induced increase of C/EBPδ protein levels was strongly reduced by Aβ fibrils. However, the effect of oligomer-enriched Aβ preparations on C/EBPδ binding activity was not expected. Although no effect could be observed on protein levels, IL-1β-induced binding of C/EBPδ to the κB site was decreased to the same degree as after treatment with Aβ fibrils. In addition, we showed in a previous study that oligomer-enriched Aβ preparations also decrease the binding of C/EBPδ to the C/EBP site [
21]. Taken together these results indicate that oligomer-enriched Aβ affects the functional properties of C/EBPδ rather than expression levels.
The activity of C/EBPs has been shown to be dependent upon phosphorylation. Phosphorylation can occur in a negative regulatory domain of C/EBPs and is believed to relieve the inhibitory effect on their DNA binding and transactivation domains [
41,
42]. C/EBPβ has a conserved MAPK consensus phosphorylation motif [
43]. Phosphorylation induced by the Ras pathway upon inflammatory stimuli will lead to a switch from a repressed to an active conformation [
43‐
45]. In murine embryonic fibroblasts, LPS-induced expression of both C/EBPβ and δ is dependent on NF-κBp65 and IκB kinase (IKK) 2 [
46]. In cells lacking the IKK-related kinase IKKi, LPS will still activate NF-κB and induce C/EBPβ and δ expression, but C/EBPδ-specific DNA binding is absent [
46].
IL-1β, as well as other inflammatory stimuli, has been shown to activate the Ras pathway [
47]. The Ras pathway in turn can result in both extracellular signal-regulated kinase (ERK) 1/2 and IKKi activation [
48]. A recent study showed that LPS-induced upregulation of C/EBPδ in a microglial cell line could be prevented by an ERK inhibitor [
31]. Both LPS-induced C/EBPδ protein expression and DNA binding to a C/EBP motif have been reported to be blocked by a combination of MAP kinase inhibitors (ERK/JNK/p38 inhibitors) [
49]. Interestingly, an age-dependent and significant reduction of the active form of ERK1 (phosphoERK1; pERK1) has been observed in cerebral cortex from the AD-transgenic mouse model tg2576/PS1
P264L [
50]. In addition, in the tg2576 mice model, which shows a slower build-up of amyloid burden, a slight, but significant, increase of pERK2 is observed in CA1 of hippocampus at 13 months of age. This is followed by a significant reduction at 20 months of age [
51]. Thus, it is possible that our results showing decreased levels of C/EBPδ both in astro-microglial cultures exposed to Aβ fibrils and in aged tg-ArcSwe mice could be explained by reduced MAP kinase activity.
Our results showing decreased levels of C/EBPδ both in astro-microglial cultures exposed to Aβ fibrils and in aged tg-ArcSwe mice seems contradictory to a previous study from Rogers' laboratory showing up-regulated C/EBPδ in AD brains [
20]. However, as discussed above, data from transgenic mouse models indicate that the effect on C/EBPδ may correlate with a later phase when build-up of fibrillar Aβ deposits reaches a certain level that will lead to reduced expression of C/EBPδ. It may be assumed that the ratio Aβ fibrils:oligomers may be lower in brain tissue from patients with the sporadic form of AD as compared to aged tg-ArcSwe transgenic mice. However, it cannot be ruled out that DNA binding and transcriptional activity is reduced also in AD brain since we propose that oligomer-enriched Aβ disturbs the functional properties of C/EBPδ rather than expression levels.
The C/EBP family members exert pleiotropic effects in tissues in which they are expressed and are also expressed in a number of different cell types. C/EBPα, β, and δ are all expressed in primary astro-microglial cultures. At least in human AD brains, C/EBPδ immunoreactivity seems to localize primarily to astrocytes [
20]. However, in brain tissue from mice neuronal localization of C/EBPs has also been reported [
20,
52] and this may also explain the high levels of C/EBPδ in young mice. At this point we have not identified the cell types which are responsible for the effects on tissue levels of C/EBPα, β, and δ in tg-ArcSwe mouse brain. It is also possible that the differential effects on the different isoforms (i.e., decreased levels of C/EBPα in both affected and non-affected brain areas, increased levels of C/EBPβ in affected areas and decreased levels of C/EBPδ only in affected areas) mirror the effects of Aβ deposits on different cell types.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
The original design of this study was made by VR, MS, KI. VR coordinated the experiments. VR and LT performed the western blot analyses. EMSA was performed by LT. Streptavidine-agarose pull-down assay and Thioflavin T assays were performed by VR. LNGN maintained the colony of transgenic mice, prepared and dissected brain tissues and performed Aβ ELISA. LNGN and LT participated in the design of the study. VR, LT and KI wrote the first draft of the manuscript. All authors discussed the results and commented on the manuscript. All of the authors have read and approved the final version of the manuscript.