Background
As part of the innate immune system, macrophages rapidly respond to a large variety of pathological molecular pattern stimuli (PAMPS) such as bacterial coat and viral proteins [
1,
2]. The programmed response to acute stimuli includes the induction of a specific gene profile and the subsequent production of multiple cytoactive factors such as TNFα, NO and IL-1 that protect against tissue invaders. In peripheral macrophages, this first phase of an innate immune response has been described as classical immune activation [
1‐
3]. "Classical activation" is also characterized by the involvement of Th-1 cytokines such as interferon-γ (IFN-γ), a "master" cytokine that orchestrates the coordinated induction and production of the "killing" phase [
4‐
6]. However, the gene profile of macrophages can change, shutting down the production of pro-inflammatory cytokines and increasing the production of factors that participate in tissue repair and wound healing [
6]. Anti-inflammatory, Th-2 cytokines such as IL-10 and TGFβ are associated with broad ranging and potent inhibition of pro-inflammatory activity while other Th-2 cytokines such as IL-4 and IL-13 serve to antagonize IFN-γ, are anti-parasitic, mediate allergic responses and induce tissue matrix reconstruction [
6,
7]. IL-4 and IL-13 -mediated gene induction has been specifically termed alternative activation and includes genes that produce arginase I (
AG1), mannose receptors (
MRC1) and genes associated with tissue remodeling such as Found in Inflammatory Zone 1 (
FIZZ1) and chitinase 3-like 3 (
YM1) [
8‐
11].
The induction of the genes characteristic of alternative activation during an immune response provides an anti-inflammatory balance to an acute, pro-inflammatory response. Consequently, alternatively activated macrophages are viewed as immunosuppressive and involved in tissue repair and extracellular matrix remodeling [
5,
6]. However, alternatively activated macrophages may also contribute to disease processes in a complex way. For example, alternatively activated alveolar macrophages contribute to the fibrotic lesion in idiopathic pulmonary fibrosis [
11] and in the liver fibrosis associated with
Schistosoma mansoni [
12].
Although classical and alternative activation are commonly viewed as the two opposing ends of macrophage activation, additional activation states may also exist [
5,
6]. For example, Anderson and Mosser [
13] have described a third class of macrophages, called Type II macrophages. This state requires a specific two step activation pattern that involves ligation of Fcγ receptors and signaling through Toll receptors, CD40 or CD44 [
5,
13,
14]. The end result is decreased
IL-12 expression concomitant with increased
IL-10 mRNA. As a consequence, the gene profile of Type II macrophages is a mixture of pro-inflammatory and anti-inflammatory genes such as
IL-1, TNF-a, IL-6, IL-10 and
IL-4 [
15]. Arginase, however, is not induced. A "deactivation" state of macrophages that is similar to Type II macrophages has also been described by Gordon [
6].
Alzheimer's disease (AD) is characterized pathologically by extracellular fibrillar deposits in the parenchyma of the brain which are composed of the β-amyloid (Aβ) peptide 1–40 and 1–42 fragments of the amyloid precursor protein (APP) [
16‐
18]. It is generally believed that soluble APP and various forms of Aβ peptides, either alone or in conjunction with other immune factors, serve as activating signals for an innate immune response in the brain [
19]. Using immunocytochemistry, Griffin et al [
20] have demonstrated the presence of IL-1β in microglia and astrocytes surrounding the amyloid deposits. Other investigators have confirmed these findings and have also shown that IL-6, TNFα and MHC expression is increased in AD [
21‐
25]. As a result, AD has been associated with classical immune activation and the production of an acute Th-1 immune response. However, AD is a chronic neurodegenerative disease in which the inflammatory process has not been thoroughly charted over time and with disease progression. It is highly likely that brain macrophages may change their activation state as a function of the disease and time. To assess this possibility, we have determined if genes associated with alternative activation are expressed in cortical samples from individuals with AD compared to cognitively normal aged matched control individuals. Additionally, we investigated the alternative activation-related gene expression profiles in mouse models of AD and of cerebral amyloid angiopathy. Our data demonstrate that genes typical of alternative activation are clearly expressed in AD and in a mouse model of amyloid deposition. However, it is likely that the macrophage activation state in AD represents a novel hybrid state between classical and alternative activation.
Discussion
The regulation of gene transcription in macrophages during immune activation is dependent on multiple factors including the type of phagocytic ligands, the types of non-phagocytic adhesion interactions, the extracellular matrix and the cytokine environment [
6,
42]. Multiple, distinct populations of macrophages have been identified and include classically activated macrophages that are pro-inflammatory and associated with the "killing" phase of the innate immune response and alternatively activated macrophages that are primarily associated with wound healing and tissue repair. Type II macrophages that appear to be a hybrid activation state share some characteristics of each [
5,
6]. Peripheral macrophages cycle between these activation states, and dysregulation of this cycling underlies various forms of chronic disease [
43].
To determine if innate immune cells of the brain exhibit alternative activation, we have used quantitative RT-PCR to identify a specific gene profile. This profile was created by stimulating BV2 and primary murine microglia with anti-inflammatory cytokines used to induce an alternative activation state in peripheral macrophages. Similar to peripheral macrophages, microglia treated with IL-4 or IL-13 significantly increased mRNA expression levels for AG1, MRC1, FIZZ and YM1, alternative activation genes. In contrast, genes for classical activation, NOS2 and TNFα , were not increased by either IL-4 or IL-13 treatment. Thus, our data confirm previously published studies on peripheral murine macrophages and provide a tool to assess the presence of alternative activation gene induction in the brain.
Using this gene expression pattern, we then probed cortical tissue from two transgenic mouse models of AD and from individuals with AD for evidence of alternative activation genes. These results are compared in Table
3. Essentially, cortical tissue from the Tg-2576 mouse and individuals with AD demonstrate a mixed profile of alternative activation and classical activation genes, particularly TNFα. The Tg-SwDI mouse that represents a cerebrovasuclar amyloid model, however, primarily demonstrates classical activation.
Table 3
Comparison of alternative activation gene profile.
TNFα
| Increased | Increased | Increased |
NOS2
| No Change | No Change | No Change |
AG1
| Increased | No Change | Increased |
CHI3L1
| - | - | Increased |
CHI3L2
| - | - | Increased |
YM1
*
| Increased | No Change | - |
FIZZ1
**
| No Change | No Change | - |
MRC1
| Increased | No Change | No Change |
The presence of alternative activation genes in AD necessitates a more complex view of inflammation in neurodegenerative disease. Numerous studies have shown that the immune cells in the vicinity of amyloid deposits in AD express mRNA and proteins for pro-inflammatory cytokines, leading to the hypothesis that AD is primarily associated with classical (Th-1) immune activation [
20,
44‐
47]. Multiplex ribonuclease protection assays and gene micro-array studies have only partially confirmed this hypothesis [
46,
48,
49]. For example, Blalock et al [
48] have examined gene profiles found in AD using gene arrays on brain samples from 22 AD subjects. The primary pro-inflammatory genes represented were MHC class II and IFNγ, although IL-18 mRNA expression was elevated as well as genes for cytokine receptors, particularly IL-6R and IL-10R. A second array analysis by Xu et al [
49] confirmed the increased in MHC class II but did not find changes in other pro-inflammatory genes. Coangelo et al [
25], however, found a 3-fold increase in IL-1β expression using a DNA microarray analysis based on pooled AD brain samples compared to pooled control samples. The discrepancies and variability in gene expression patterns between the published studies are puzzling, especially in view of the fact that acute exposure of microglia to Aβ peptides initiates a classical activation pattern of gene expression [
23,
50‐
52]. There are many possible reasons for the differences between the
in vitro and
in vivo data. Hoozemanns et al[
53] have suggested that the sequence and timing of pathological events in AD is critical. We further suggest that microglia may exhibit specific stages of response during a chronic neuroinflammatory disease such as AD. These stages in microglia in AD may be similar to those stages observed in chronic inflammatory diseases of the lung or liver, including the induction of alternative activation.
A shift to alternative activation in brain innate immune cells is supported by the analysis of mRNA expression for classical and alternative genes in the Tg-2576 mouse model of AD, a well-studied animal model for CNS parenchymal amyloid deposition [
32].
AGI, MR C
1 and
YM1 mRNA levels were significantly increased in the Tg-2576 mouse brains compared to age-matched controls while
TNFα and
NOS2, genes commonly associated with classical activation, increased and did not change, respectively. The increased
TNFα mRNA suggests a mixed activation state reminiscent of Type II macrophage activation [
5]. However, both classical and Type II activated peripheral macrophages exhibit increased
NOS2 mRNA, not decreased NOS2, and no induction of
AG1. Thus, since
NOS2 mRNA induction is not observed in peripheral macrophages that exhibit alternative activation, while
AG1 expression is increased [
5,
54,
55]., the preponderance of the data suggest that alternative activation is a dominant feature of the innate immune response in the APP Tg-2576 mouse. However, we cannot rule out that activation state in the APP Tg-2576 mouse is a novel, hybrid state.
In contrast, the Tg-SwDI mouse model, which represents a localized cerebrovascular amyloid angiopathy [
33,
34]., did not demonstrate the same increase in alternative activation markers. These differences may be due to the predominant cerebrovascular microglial proinflammatory phenotype that is observed in Tg-SwDI mice brains or in humans who express either the Iowa or Dutch mutation [
33,
56]. An increase in
TNFα mRNA observed in the Tg-SwDI mice brains is consistent with this hypothesis.
Both classical and alternative activation markers were also observed in brains from AD patients and resemble the activation pattern found in Tg-2576 mice. In AD brain, mRNAs for TNFα, AGI, CHI3L1 and CHI3L2 were significantly increased in cortical samples compared to age-matched control brains while no significant difference was observed for MRC1 mRNA. The strong presence of the alternative activation genes in cortical tissue samples in AD brain implies that cells, such as microglia or astrocytes, have undergone a shift in functional profile. This finding does not negate immunocytochemical studies that demonstrate discretely localized, pro-inflammatory cytokine expression such as IL-1β in plaque associated microglia or astrocytes. However, the presence of alternative activation markers in plaque associated cells has not been determined. Thus, it is not clear if microglia within the vicinity of plaques show a complex activation state or if some cells express pro-inflammatory genes while others express alternative activation genes in a mosaic-like pattern.
The induction signal(s) for alternative activation in the amyloid mouse models and in AD remains unclear. Although IL-4 and IL-13 have been most closely linked to alternative activation [
6], other Th-2 cytokines such as IL-10 and TGFβ down-regulate inflammation and are involved in repair and matrix remodeling [
6,
57‐
59]. Of these, only TGFβ has been firmly observed in AD brain [
60,
61]. while both TGFβ and IL-10 immunoreactivity have been detected in brains of Tg-2576 mouse [
45]. The effects of anti-inflammatory cytokines in AD are largely unknown. Recently, however, Koenigsknecht-Talboo and Landreth [
62] have shown that IL-4, IL-13, TGFβ or IL-10 enhance uptake of fibrillar Aβ peptides. Interestingly, no effect on Aβ uptake is observed with the anti-inflammatory cytokines alone, but instead, they serve to reduce the suppression of Aβ phagocytosis initiated by pro-inflammatory cytokines. These findings underscore the complexity of the brain's cytokine environment and its role in modifying microglial responses to Aβ peptides. Aβ, itself, may also influence the gene switch from classical towards alternative activation in microglia. Fibrillar Aβ interacts with numerous microglial membrane receptors including scavenger receptors A and B; CD40; an α6/β 1integrin/CD36/CD47 complex and complement receptors [
63‐
67]. Crosslinking of these or other receptors has been associated with macrophage "down-regulation" through multiple mechanisms [
68,
69]. For example, mannose receptor signaling initiates an anti-inflammatory program within macrophages [
68]. MRC-1 (CD206) is a transmembrane glycoprotein that mediates Ca
2+ dependent endocytosis and phagocytosis of mannosylated ligands [
58] whose role in AD is currently unknown.
The induction of alternative activation genes is commonly considered to be a harbinger of repair and extracellular matrix re-organization that may begin during or after the first stages of an acute innate immune response [
6,
70,
71]. Although the exact functions of many of the protein products of these genes are not clear, some alternative activation genes such as
AG1 have been well studied in peripheral macrophages. Both isoforms of arginase utilize arginine as a substrate for biosynthetic pathways that produce polyamines and proline [
72]. Polyamines such as spermine are well known to alter cell proliferation but have wide-ranging physiological effects such as regulation of NMDA channel function, membrane potentials and gene transcription [
73,
74]. Proline is an important component of collagens and is involved in repair of the extracellular matrix. The maintenance of high
AG1 expression, as observed in our studies, is likely to direct arginine utilization toward the production of proline or polyamines and away from the production of nitric oxide. The enzymatic activities of both inducible NOS and arginase are solely dependent on intracellular arginine and these enzymes compete for arginine [
72]. The low expression of
NOS2 mRNA coupled with the increased expression of
CAT2 mRNA, a critical arginine transporter, observed in our AD samples may further promote arginase activity. Interestingly, Hesse et al [
12,
54]. have shown that increased
AG1 expression in schistosome egg-induced granulomas is associated with increased proline and polyamine production and promotes fibrosis in liver. Hesse et al [
54] demonstrated that the re-induction of
NOS2 expression or activity reduced the fibrotic load in the parasite-induced liver granulomatosis model. The upregulation of
AG1 in AD, coupled with the loss of
NOS2 mRNA, then, may have critical relevance to amyloid deposition in the extracellular matrix of the brain.
The
FIZZ1 and
YM1 genes also provide a link between alternatively activated macrophages and repair processes after infection or injury [
10,
70]. The protein product of
YM1 induction is a novel mammalian lectin that binds saccharides and heparin/heparin sulfate on cell surfaces, but whose functions are largely unknown [
70,
75]. Hung et al [
75] have suggested that
YM1 helps to protect the extracellular matrix scaffold at sites of injury by reducing heparin sulfate degradation.
FIZZ1 encodes a 9.4 kDa cysteine rich protein which was originally described in lung lavage fluids in a murine allergic pulmonary inflammation model [
11]. Three
FIZZ family members have been identified and are now known to be part of a new gene family of resistin-like molecules. As such,
FIZZ proteins may contribute to insulin resistance during diabetes but they have also been linked to angiogenesis, to stimulation of collagen production and to inhibition of apoptosis [
10,
11,
76]. Although a human homolog exists for
FIZZ1, no direct human homologs have been identified for
YM1. Our data, however, demonstrates that two closely related chitinase genes, namely
CHI3L1 and
CHI3L2 are overexpressed in AD brain. Both forms of chitinase 3-like proteins do not have enzymatic chitinase activity and instead, inhibit IL-1 and TNFα-mediated responses by blocking cell signaling [
8].
In summary, immune cells within the brain display gene profiles that suggest heterogeneous, functional phenotypes that range from a pro-inflammatory, classical activation state to an alternative activation state involved in repair and extracellular matrix remodeling. These different functional phenotypes not only protect the tissue from invaders, but orchestrate and promote tissue reconstruction resulting in resolution of the injury. Repair processes mediated by alternative activation genes, however, can be associated with maintenance of disease and, in particular, enhanced fibrosis [
6,
12,
77]. For example, diseases in the periphery that have fibrosis as a characteristic feature, such as
Schistosoma japonicum egg induced fibrosis in the lung,
Schistomsoma mansoni infection of the liver or idiopathic lung fibrosis show defective repair that, in fact, favors fibrosis [
54,
55,
77]. Anti-inflammatory cytokine treatment under these circumstances worsens the fibrosis, while the re-establishment of NOS induction and activity reduces the fibrosis [
12,
77]. Neuroinflammation in AD is characterized by both degeneration and regeneration that occurs in a specific pattern of time and locale [
53]. Studies on AD neuropathology implicate the presence of a defective repair process that is linked to the presence of Abeta peptides and amyloid fibrils [[
53];78]. Our data presented here begin to build the case that alternative activated macrophages are present in AD brain and may contribute to a Th-2-linked, rather than a Th-1 linked, pathology. If true, then therapeutic approaches may need to consider this additional alteration of the immune response.
Authors' contributions
CAC designed the study, analyzed data, prepared figures and wrote the manuscript. RM performed Q RT-PCR experiments, analyzed data and contributed to the preparation of the manuscript; HS performed Q RT-PCR; QX prepared brain samples, performed Q RT-PCR and analyzed data; WVN provided transgenic mice and contributed to the preparation of the manuscript; MPV provided transgenic mice, participated in the study design and in the manuscript preparation.