Background
Microglial cells are brain-resident immune cells that have many protective roles but can also contribute to neurological disease processes. The functional phenotype of microglia depends on the cell types and specific activating factors in their surroundings, and for this reason microglia are said to have an adaptive or acquired phenotype, reflecting their response to a collection of external signals [
1]. Secreted small molecules, such as cytokines and growth factors, allow microglia to communicate with each other, and with other immune and brain cells which express these receptors [
2]. In this way microglia can be recruited to sites of injury, where they can respond by ameliorating the damage [
3,
4]. However, there have also been numerous reports of detrimental microglial activity in response to injury, and neuroinflammation is thought to play a major role in pathogenesis of many neurological disorders, including epilepsy and Alzheimer’s disease [
5‐
9]. The activation of microglia to a beneficial or detrimental phenotype can be described in terms of inducible protein expression, morphology, and functional outcomes such as cytokine production and phagocytosis [
8,
10]. Modulating microglial phenotype towards a protective role is a potential strategy for the treatment of many brain disorders and thus the factors influencing microglial activation require further research.
One molecule that can influence microglial phenotype is the cytokine macrophage colony-stimulating factor (M-CSF). M-CSF is found in the brain and its receptor is expressed by microglia [
11,
12]. M-CSF is produced by a range of cells in the developing and adult brain [
12,
13]. M-CSF mRNA and protein have been found to be constitutively expressed by human fetal astrocytes and low levels of M-CSF mRNA and protein were detected in unstimulated microglia cultures [
14]. Du Yan
et al. (1997) have also demonstrated M-CSF expression in neurons of the adult human brain [
12].
Signaling through the M-CSF receptor (CSF-1R) is required for the development and differentiation of microglia [
15] and M-CSF has been shown to increase division of rodent and human fetal microglia [
14,
16,
17]. M-CSF also has the ability to change microglial morphology [
18,
19] and influence microglial activation [
20,
21]. M-CSF signaling has been shown to be dependent on the transcription factor PU.1 [
22], a vital myeloid transcription factor expressed by microglia in the adult human brain [
23]. Celada
et al. (1996) found that sense PU.1 expression constructs increased M-CSF-dependent proliferation in mouse bone-marrow macrophages, and antisense PU.1 constructs reduced proliferation in response to M-CSF [
24]. They also found that sense PU.1 constructs gave rise to increased M-CSF receptor expression, and it has previously been demonstrated that PU.1 binds to the
c-fms promoter [
25].
While M-CSF is important for normal brain development and function [
26], several studies have found abnormal levels of M-CSF associated with neurological diseases. M-CSF was demonstrated to be upregulated in brain tumors [
27,
28] and a correlation was found between levels of M-CSF and HIV-associated cognitive impairment [
29]. Furthermore, within three months of HIV therapy, levels of both M-CSF and viral RNA in the CSF were reduced [
29]. Despite Boissonneault
et al. (2009) reporting beneficial effects of M-CSF on cognitive impairment and Aβ
1-42 deposition in a mouse model of Alzheimer’s disease, it has been suggested that increased M-CSF expression could contribute to Alzheimer’s disease pathogenesis [
12,
30]. Lue
et al. (2001) cultured glia from adult human brains and found that M-CSF was elevated in Alzheimer’s disease compared to non-demented microglia [
31]. M-CSF receptor expression in microglia in human brain tissue was upregulated in lesions of Alzheimer’s disease and amyotrophic lateral sclerosis [
11]. On the contrary, while microglia are found to be associated with multiple sclerosis lesions, the relative number of microglia expressing M-CSF and its receptor have been found to decrease [
32]. M-CSF is therefore considered a key factor in regulating microglial inflammatory responses in the damaged brain.
Despite this comprehensive body of literature, there are several caveats which may prohibit the linking of functional studies to observations in human brain tissue. Although most of the research on M-CSF and microglia has been carried out using rodent cells and models, it is becoming increasingly clear that there are important differences between rodent microglia and their human counterpart [
33,
34]. Age has also been shown to have an impact on immune cell responses and activation [
35‐
37]. Furthermore, the micro-environment of the brain may affect immune responses in ways that are different to those of the periphery [
4] and while the majority of research on M-CSF has been carried out on peripheral immune cells, the effects of M-CSF on adult human microglia have not been fully investigated. We have thus assessed the phenotypic profile of adult human microglia after exposure to M-CSF and found that this cytokine dramatically influences their phenotype and activation state.
Discussion
M-CSF has numerous interesting effects on adult human microglia, many of which may have significance for a range of neurological diseases.
The numbers of PU.1 and CD45 immunopositive cells were increased by M-CSF treatment and we have shown that this is at least partially through an increase in microglial division. This division effect was immediately apparent as adult human microglia do not frequently divide when cultured in basic medium (DMEM/F12 + FBS + PSG). This observation is in line with previous studies noting a proliferation effect with M-CSF for human fetal microglia [
14], and up-regulation of M-CSF following axotomy of the rat facial nucleus which triggered microglia to proliferate [
17]. Many myeloid cells have been shown to have M-CSF growth dependence [
44]. It may be that adult human microglia require M-CSF for division, but not for survival,
in vitro. The proportion of microglia that undergo division is specific to each individual patient. However, the result of increased microglial division with M-CSF treatment is consistent between cases.
It has previously been shown that microglia in the rat brain, as well as the BV2 rodent microglial cell line, constitutively express a high level of PU.1 in both the ‘normal’ and post-injury state [
45]. We have recently shown that microglia in the adult human brain also express the transcription factor PU.1 [
23]. Previous studies have demonstrated that PU.1 regulates the M-CSF receptor [
25]. We have found here that PU.1 also acts down-stream of M-CSF signaling, as PU.1 protein expression is increased in microglia following M-CSF treatment (Figure
2). PU.1 has been shown to be involved in M-CSF-dependent proliferation of mouse bone-marrow macrophages [
24] and thus it is likely that PU.1 is involved in the M-CSF-induced proliferation of adult human microglia, as well as many of the other processes discussed below.
Phagocytosis is an important innate function of microglia as part of their role to respond to cell injury and regulate the extracellular environment. We found that M-CSF increased adult human microglial phagocytosis of Aβ
1-42 peptide by microglia from different cases, regardless of the starting level of phagocytosis. In the context of Alzheimer’s disease it is thought that microglia may be helpful in clearing the brain of extracellular deposits of Aβ
1-42 protein [
46,
47]. M-CSF has been shown to be beneficial in Alzheimer’s disease mouse models and one mechanism that may be responsible for this effect is increased phagocytosis of Aβ
1-42[
30]. In another recent study using human microglia, M-CSF treatment (in combination with IL-4 and IL-13) increased microglial phagocytic ability for myelin debris compared to microglia stimulated with a combination of GM-CSF, IFNy and LPS [
48].
We have previously found that PU.1 is necessary for basal phagocytosis in human adult microglia [
23]. DAP12 also appears to be involved in the phagocytic process [
49] and associates with proteins which have been found to play a role in phagocytosis, for example CD68, TREM2 and SIRPB1 [
42,
50]. Given the M-CSF-induced increases in PU.1 and DAP12 expression, it is likely that these proteins are involved in M-CSF-dependent phagocytosis in adult human microglia.
Another prominent effect of M-CSF on adult human microglia is their change in morphology to bipolar, elongated (‘ramified’) cells. Microglial morphology is presumed to relate to their function, although exactly how is currently unclear. Round ‘amoeboid’ microglia are traditionally viewed as activated, inflammatory microglia. The M-CSF-induced morphology change could be a sign of microglia being ‘primed’ towards a particular activation state. However, microglial phenotype is multifaceted and M-CSF-induced elongation doesn’t prohibit microglia from changing morphology when exposed to other molecules, for example Aβ1-42 which induces changes to their cytoskeletons necessary for phagocytosis. Even though the M-CSF-treated microglia are more rod-shaped and ‘ramified’ than vehicle-treated microglia, they have an increased propensity to phagocytose compared to control microglia of heterogeneous morphology.
Durafourt
et al. (2012) used the same concentration of M-CSF as part of an ‘alternative’ macrophage/microglia activation polarizing protocol. Whereas macrophages under these conditions had more extended processes, no change was observed for microglia. It may be that the CD45 marker of microglia used here is better for observing morphological differences, or that their addition of IL-4 and IL-13 reduced the morphology effect seen in microglia [
48]. The M-CSF-induced ‘rod’-like effect has also been noted for human fetal microglia [
14].
Elongated rod-shaped microglia have also been reported
in situ in rodent models of neurological injury including ischemia [
51,
52]. Graeber (2010) has reported on microglial rod cells observed in human brain tissue associated with cognitive symptoms and psychopathologies [
10]. Rod-shaped, elongated microglia have been reported in the Huntington’s disease cortex [
6] and in subacute sclerosing panencephalitis, Alzheimer’s disease and Wilson’s disease brains [
53]. In concordance with our findings, Wierzba-Bobrowicz
et al. (2002) also noticed proliferating cell nuclear antigen co-labeling with rod microglia and reduced expression of HLA compared to more rounded microglia [
53].
Rod-shaped microglia are relatively under investigated [
10] and our study presents some of the first functional results relating to the findings of rod-microglia in human brain tissue. If M-CSF-treated adult human microglia are indeed
in vitro correlates of the ‘rod’ microglia reported in diseased adult human brain tissue, this will provide an invaluable tool for investigation into this particular microglial phenotype. Furthermore, our method for quantifying microglial morphology is a quick, high throughput and unbiased way of assessing these changes, compared to laborious and subjective quantification by eye.
To determine the functional relevance of the morphology change and to look further at the activation state of these M-CSF-exposed microglia, we investigated whether their expression of HLA was affected. HLA-DP, DQ, DR is an inducible protein and we find that its expression by microglia varies widely between cases. We asked whether this inducible expression of HLA-DP, DQ, DR was modulated by M-CSF and found that, despite an increased number of microglia, fewer microglia express high levels of HLA-DP, DQ, DR with M-CSF. HLA-DP, DQ, DR expression by microglia is often taken to represent an ‘activated’ or inflammatory microglial phenotype. Our results suggest that M-CSF-treated microglia may be alternatively activated and have reduced antigen presentation capacity.
This M-CSF effect of reduced HLA-DR has also been noted for human fetal microglia [
14]. Furthermore, expression of HLA class II molecules was noted to be less intensive on the surface of microglial rod cells compared to neighboring ramified microglia in neurologically diseased human brain tissue [
53]. In the different context of mouse monocytic precursor cells, Henkel
et al., (2002) found that M-CSF-induced maturation increased MHC class II expression with IFNy [
22]. This finding indicates that the effect of M-CSF on HLA may be dependent on cell differentiation stage and/or species. Conversely, Melief
et al. (2012) have recently shown that the ‘alternative’ activating cytokine IL-4 increases HLA-DR mRNA expression whilst also inducing an elongated morphology in adult human microglia [
54]. Overall, these results demonstrate the wide phenotypic diversity of microglia.
An HLA-DQ-derived peptide has been found to have anti-proliferative effects [
55], suggesting that the decrease in HLA-DP, DQ, DR seen in the present study may be mechanistically related to the increase in microglial proliferation also seen with M-CSF. Thus decreased HLA may not only have implications for antigen presentation, but perhaps for other cellular functions like proliferation.
Of particular interest in our study is that an up-regulation of PU.1 with M-CSF correlates, in general, with a less activated microglial phenotype (for example, reduced HLA-DP, DQ, DR expression), although phagocytosis was stimulated. This contrasts with a recent report by Ponomarev
et al. showing that increased PU.1 was associated with an activated microglial phenotype in rodents [
56]. The most likely reason for these different results is species differences, highlighting the importance of studying adult human brain microglia.
To decipher the mechanisms by which the M-CSF-induced phenotypic changes occur in microglia, we looked for changes in factors known to be associated with the transcription factor PU.1.
CCAAT enhancer-binding protein (C/EBP) transcription factors are expressed throughout the body including the brain [
45,
57]. The C/EBP family has already been shown to have a number of species-specific regulation processes and expression patterns [
57]. From the C/EBP transcription factor family we detected an increase in C/EBPβ expression within microglia in human mixed glial cultures treated with M-CSF.
C/EBPβ plays numerous roles in activation and differentiation of macrophages [
57]. It has been reported to have a role in inflammatory processes in rodents [
58‐
60] and may play a role in differentiated macrophage morphology [
61]. Although our studies find increased C/EBPβ along with increased microglial proliferation, C/EBPβ has been reported to inhibit proliferation of human THP-1 monocytic cells and murine macrophage-like cells [
61]. However, other reports have shown that C/EBPβ can promote proliferation [
57] and furthermore, be involved in M-CSF-directed mechanisms in tumors [
62] and HIV infection [
63]. C/EBP transcription factors including C/EBPβ can transactivate the CSF-1R promoter in mammalian cell line COS-7 cells [
64], and could also be involved in the increase in CSF-1R expression observed in our experiments. C/EBPβ has recently been demonstrated in human spinal cord tissue from amyotrophic lateral sclerosis patients, but rarely in controls, co-localized with the microglial marker CR3 [
58].
C/EBPβ forms heterodimers with members of its own family and interacts with several other transcription factors [
57] including PU.1 [
65,
66]. M-CSF-mediated enhanced PU.1 and C/EBPβ transcription factor protein expression have also been reported for a murine myeloblastic cell line [
67]. Furthermore, co-expression of PU.1 and C/EBPβ in fibroblasts can induce a macrophage phenotype [
44]. PU.1 and C/EBPβ transcription factors together may be responsible for many of the M-CSF induced effects we find in adult human microglia.
DAP12 is a myeloid adapter protein found in microglia in the brain. We found that M-CSF treatment of adult human microglia increased their DAP12 expression. It has been shown to be involved in M-CSF-induced proliferation and survival of mouse bone marrow-derived macrophages [
68]. The concurrent increase in DAP12 protein expression with increased adult human microglia number and proliferation suggest a role for DAP12 in this M-CSF-induced mechanism. DAP12 could also be involved in the process of phagocytosis as primary mouse microglia transduced with mutant DAP12 have reduced phagocytic ability [
50].
Henkel
et al. (2002) demonstrated an upregulation of DAP12 in PU.1-rescued monocytic precursor cells and Weigelt
et al. (2007) have shown that DAP12 expression is dependent on PU.1 via a binding site in the DAP12 proximal promoter [
22,
49]. Thus the M-CSF-induced increase in DAP12 expression may be directly mediated by the increase in PU.1. Furthermore, the role of M-CSF in determining PU.1 and DAP12 expression in microglia may have implications for many neurological diseases [
69].
CSF-1R expression is shown here to be restricted to microglia, and not detected on other cell types in our cultures, both basally and with M-CSF treatment. In addition, we found that M-CSF increased the expression of its receptor on microglia. The increase in PU.1 found with M-CSF treatment is likely to directly increase CSF-1R levels as it has been reported to regulate
c-fms transcript expression [
25]. Yamamoto
et al. (2010) have also found that M-CSF increases microglial CSF-1R expression in the context of the rat axotomized facial nucleus [
17].
Coincidentally, M-CSF treatment also increased microglial expression of the IGF-1 receptor. There have been a number of associations previously reported between the growth factors M-CSF and IGF-1 [
43]. Both factors are mitogenic, play critical roles in development and have similar regulation mechanisms. In a study of mouse macrophage tumor cells, Wessells
et al. (2004) found C/EBPβ to have a critical role in cell survival, in part by regulating expression of IGF-I. Furthermore, M-CSF was found to compensate for IGF-1 and could rescue IGF-1-deficient cells [
70]. The overlapping functions of these ligands may explain the simultaneous increase in both the CSF-1 and IGF-1 receptors in response to M-CSF.
Another CSF-1R ligand, IL-34, has recently been discovered and is expressed throughout the body, including the brain [
71]. Like M-CSF, IL-34 is involved in human monocytic proliferation and viability [
71,
72] but its biological activity and signal activation are not identical to that of M-CSF [
72]. Furthermore, it has been suggested that the
in vivo role of IL-34 may differ between rodents and humans [
72], and research into the effects of this cytokine on microglia in the human brain are warranted.
Authors’ contributions
AMS, HMG and MD conceived and designed the experiments. AMS performed cell isolation and cell culture experiments, immunocytochemistry, image acquisition and analysis. AMS and MD interpreted data and wrote the manuscript. RLO, PSB, EWM, MAC and RLMF contributed materials and revised the manuscript. All authors read and approved the final manuscript.