The first rule of triage is do no harm, but try telling that to microglia. Though these cellular paramedics may be invaluable first responders to pathological damage in the brain, including accumulation of amyloid-β (Aβ), there is increasing evidence their overzealous attempts to dispose of unwanted debris is accompanied by a torrent of dangerous proinflammatory signals and toxins, such as TNFα and reactive oxygen species. Perhaps these concepts are not mutually exclusive. As demonstrated by two recent papers, the beneficial and detrimental properties of microglia may depend on when and why they are recruited.

Writing in the March 11 Nature Medicine online, Joseph El Khoury and colleagues at Harvard Medical School and the Okayama University Medical School, Japan, report that microglia protect transgenic mice expressing amyloidogenic human Aβ precursor protein (APP). In contrast, William Van Nostrand and colleagues at Stony Brook University, New York, report in the March 21 Journal of Neuroscience that blocking microglial activation in a model of cerebral amyloid angiopathy rescues behavioral deficits. Together, the two papers demonstrate how the relationship between microglia and pathology may be far from simple.

El Khoury and colleagues studied mice lacking chemokine receptor 2 (Ccr2), a microglial cell-surface receptor that mediates chemotaxis and recruitment of microglia to sites of injury. “We figured one way to test whether microglia are toxic is to prevent them from coming into the brain,” El Khoury told ARF. There is good evidence to suggest that microglia infiltrate the AD brain from the blood. Chemotactic cytokines, including monocyte chemotactic protein 1 (Mcp-1), are upregulated in AD brain (see Ishizuka et al., 1997) and are produced by microglia and astrocytes after challenge with Aβ (see Smits et al., 2002 and El Khoury et al., 2003). Recent studies have also identified bone marrow-derived microglia in the brain of APP transgenic mice (see ARF related news story).

To test the role of migratory microglia, El Khoury and colleagues bred transgenic APP mice (Tg2576) with Ccr2-/- animals, generating APP mice both heterozygous and homozygous-negative for Ccr2. The researchers found that the loss of the chemokine receptor had a dramatic effect on survival in the APP transgene background. Only about 15 percent of mice lacking Ccr2 survived to 130 days, at which point about 70 percent of APP animals were still alive. Mice with one copy of Ccr2 faired a bit better—about 40 percent were still alive at day 130. The poor survival was directly related to APP pathology because lack of Ccr2 had no effect on survival in a wild-type background. To investigate this more deeply, El Khoury and colleagues examined pathological changes in the brain. They found that the Ccr2-negative animals had about 1.5- and twofold increases in levels of Aβ40 and Aβ42, respectively. Levels of β-secretase, presenilin 1, and insulin degrading enzyme were unchanged, suggesting that the elevated Aβ was not due to increased production or decreased enzymatic degradation. Aβ deposits were clearly visible in 65-day-old APP Ccr2-/- animals, but not in age-matched APP or wild-type animals.

The pathological changes in these mice seem related to loss of brain microglia. In the absence of Ccr2, the number of microglia in the brain of 65-day-old APP transgenic mice was reduced by sixfold, as judged by staining for the microglial marker Cd11b. Furthermore, using flow cytometry to distinguish resident brain microglia (Cd11 positive and low expression of Cd45) from migratory cells from the blood (Cd11 positive and Cd45 high expression), the authors found that the APP mice had about fivefold more migratory microglia in the brain than APP, Ccr2 +/- animals. In addition, while stereotactic injection of Aβ into the brain of wild-type mice led to robust infiltration of microglia, it had a minimal effect on microglial recruitment in Ccr2-negative animals. The data indicates that migratory microglia may play a crucial role in ridding the brain of toxic amyloid.

The paper from Van Nostrand and colleagues paints microglia in a slightly different light. First author Rong Fan and colleagues used the anti-inflammatory drug minocycline in Tg-SwDI transgenic mice, which express human APP with Swedish, Dutch, and Iowa mutations. These animals accumulate abundant Aβ fibrils in the microvasculature of the brain. Because minocycline readily crosses the blood-brain barrier and suppresses activation of microglia by Aβ (see Familian et al., 2006 and Seabrook et al., 2006), the authors used it determine how tempering microglial activation might affect the transgenic mice.

Fan and colleagues gave minocycline every other day for 28 days to 12-month-old mice. They found that this short treatment had no effect on Aβ accumulation, but it did reduce activation of microglia, particularly in the hippocampus where the treatment reduced the numbers of activated cells by about one half, as judged by the number of MHCII- and collagen type IV-positive cells. The treatment also reduced levels of the inflammatory cytokine interleukin 6 (IL6) by about 25 percent. The number of reactive astrocytes, however, was unchanged by the drug.

If microglia are protective in this model, then minocycline might be expected to exacerbate the disease phenotype. Instead, Fan and colleagues found that treatment with the anti-inflammatory led to significant improvements in performance in the Barnes maze test, where the animals must correctly identify one of eight holes, equally spaced around the circumference of a circular platform, that leads to an escape box. Over a 5-day testing period, transgenic mice that received the drug showed steady improvement in time taken to find refuge, and by the end of testing they were performing almost on par with wild-type animals. Untreated APP mice, however, performed progressively worse. By the end of testing they took on average 100 seconds to find the escape hole, as opposed to about 40 seconds for treated animals. “These data further support the hypothesis that microglial activation is involved in promoting cognitive impairment in this transgenic mouse model of cerebral microvascular amyloid,” write the authors.

Curiously, when El Khoury and colleagues examined the Ccr2-negative animals, they noted Aβ accumulated primarily around blood vessels and that several animals had signs of intracranial hemorrhage on autopsy. These findings suggest that microglia are needed to prevent CAA. So how does one reconcile the two studies? The crucial aspect might be timing, El Khoury suggested. “Microglia in AD may be behaving like monocytes in atherosclerosis. They are initially recruited to clear and get rid of Aβ, but as the disease progresses (and the mice grow older), microglial accumulation and their subsequent activation becomes detrimental.” (See also El Khoury’s recent ARF comment.) Van Nostrand echoed that sentiment in a comment to ARF. Early on, activation of microglia may be helpful, “but if Aβ is not cleared and microglia remain activated this may lead to the chronic neuroinflammation and behavioral deficits that we observed in our model,” he wrote (see full text of his comment below). In short, while their triage attempts may at first prove successful, over the long term these cells may end up doing more harm than good, in which case, a dose of anti-inflammatory medicine may prove beneficial.—Tom Fagan


  1. The paper from Joseph El Khoury and colleagues presents convincing evidence that the absence of activated microglia is detrimental in the Tg2576 model. On the surface, from our study in J. Neuroscience, one may conclude that microglial activation is harmful. It likely depends on the context of how you are viewing the problem. Early on, microglial activation may be helpful by facilitating clearance of Aβ from brain; in their absence more Aβ accumulates (El Khoury). On the other hand, if Aβ is not cleared and microglia remain activated, this may lead to the chronic neuroinflammation and behavioral deficits that we observed in our model.

    Another caveat that we must all recognize is what are the specific features of the models we work with. Each has its own strengths and weaknesses for studying specific aspects of Aβ pathology. For example, the widely used Tg2576 mouse expresses high amounts of Swedish mutant human APP in many cell types, producing high amounts of wild-type Aβ peptides and parenchymal amyloid plaques. The Tg-SwDI mouse expresses low levels of Swedish/Dutch/Iowa mutant human APP only in neurons producing low levels of vasculotropic Dutch/Iowa mutant Aβ peptides and microvascular amyloid deposits. In light of these differences in the models, some variations in results may be attributed to the sites of amyloid deposition and possibly due to differences in microglial responses to wild-type and vasculotropic mutant Aβ peptides and amyloid deposits.

  2. El Khoury et al. have produced a dataset that adds to those indicating a beneficial role for monocytic phagocytes (either activated microglia or hematogenous macrophages) with respect to the development of Alzheimer-related pathology. Some data have indicated that inflammation-related events elaborated by microglia contribute to AD pathology. This includes the overexpression of interleukin-1-β in APP-transgenic mouse models of AD, as well as attenuation of Aβ accumulation in these mice by anti-inflammatory agents such as ibuprofen and, more recently, minocycline (see Fan et al., 2007). But beginning with paradigms in which such mice are immunized against Aβ, increasing evidence has suggested that monocyte-derived cells can help to clear Aβ from the brain through phagocytosis and/or expression of Aβ-degrading proteases. For instance, Morgan and colleagues have shown that injection of the powerful inflammatory agent lipopolysaccharide into APP-transgenic mice results in Aβ clearance (DiCarlo et al., 2006), and the clearance or prevention of Aβ deposits in immunized mice is associated with some signs that microglia are more active.

    An important question has been whether these beneficial roles of monocytic phagocytes are operative in the basal condition (and eventually overwhelmed in the development of disease) or are instead induced only by extraordinary manipulation, such as immunization or injections of lipopolysaccharide. El Khoury’s approach was to remove or reduce a chemokine receptor (CCR2) responsible for trafficking microglia and/or peripheral macrophages to sites of inflammation, which would include amyloid plaques in the APP transgenic mice. The resulting increase in Aβ accumulation (both soluble and deposits), coupled with an absence of the accumulation of monocytic phagocytes that normally arises in APP transgenics, suggests that monocyte-derived cells tonically participate in the removal of Aβ; microglia from the CCR2-knockout mice still reacted to Aβ in culture. This specific requirement for chemotaxis, then, is consistent with recent studies showing the homing of bone marrow-derived monocytic cells to plaques in APP transgenics (Simard et al., 2006). Microglia are so extensively distributed throughout the cortex that one should imagine they scarcely need to migrate if they were the primary mediators of Aβ clearance.

    Of course, the caveat that an APP transgenic mouse is not a human with AD goes without saying. And that may be most relevant to the interpretation of what happens downstream of Aβ clearance. El Khoury et al. reported a decrease in lifespan in the CCR2-knockout animals, but this may have been due to cerebrovascular hemorrhage. It is possible that well-intentioned clearance of Aβ, regardless of how successful, may produce byproducts that interfere with neurophysiology. To wit, the application of the anti-inflammatory antibiotic minocycline by Fan et al. protected against behavioral deficits in APP transgenic mice without altering Aβ levels or deposition, and ibuprofen treatment is associated with a decrease in a marker of apoptosis per plaque rather than a reduction in plaques themselves (Lim et al., 2001). Thus, strategies aimed at optimizing the impact of inflammatory processes or monocytic phagocytes on AD pathogenesis should take into account the potential requirement of a balance between the benefits of Aβ clearance and the maladaptive consequences of inflammatory sequelae on neuronal function and viability.

    It is somewhat unfortunate that El Khoury et al. utilized an APP-transgenic strain that has a mixed genetic background (SJL x C57BL/6). Aβ deposition is notoriously strain-dependent, with the relevant alleles remaining unknown. Any cross of a mixed background creates the opportunity for genetic variability in the progeny, even in littermates. This concern can be mitigated by analyzing sufficient numbers. El Khoury et al. used as few as three or four animals per group, which seems low except for the fact that techniques were applied which precluded the use of the same animals for some of the techniques (e.g., immunohistochemistry vs. FACS); thus, the true numbers of animals over which dramatic differences were seen was actually six or seven per genotype.


    . Minocycline reduces microglial activation and improves behavioral deficits in a transgenic model of cerebral microvascular amyloid. J Neurosci. 2007 Mar 21;27(12):3057-63. PubMed.

    . Intrahippocampal LPS injections reduce Abeta load in APP+PS1 transgenic mice. Neurobiol Aging. 2001 Nov-Dec;22(6):1007-12. PubMed.

    . Bone marrow-derived microglia play a critical role in restricting senile plaque formation in Alzheimer's disease. Neuron. 2006 Feb 16;49(4):489-502. PubMed.

    . Ibuprofen effects on Alzheimer pathology and open field activity in APPsw transgenic mice. Neurobiol Aging. 2001 Nov-Dec;22(6):983-91. PubMed.

  3. It is odd that an effect was noted by El Khoury et al. in a Ccr2 knockout. Cedric Raines showed in a landmark paper that Ccr2 was so redundant that it made no impact on trafficking of monocyte-related cells in EAE (experimental autoimmune encephalomyelitis).

    Gaupp S, Pitt D, Kuziel WA, Cannella B, Raine CS. Experimental autoimmune encephalomyelitis (EAE) in CCR2(-/-) mice: susceptibility in multiple strains. Am J Pathol. 2003;162:139-50. Abstract

  4. The report by El Khoury and colleagues shows that recruitment of macrophage-like cells to the brains of Tg2576 mice via Ccr2 plays an important role in limiting AD-like pathology. This is a very interesting finding and extends the work of Stalder et al. (2005), who noted the presence of round, non-process-bearing, macrophage-like cells in APP23 mice with appreciable amyloid deposits.

    El Khoury et al. have gone further by establishing that Ccr2-dependent recruitment of microglia/macrophage-like cells is important in limiting progression of cerebral amyloidosis. If taken to the logical endpoint, this would mean that microglia and/or macrophages serve to limit amyloidosis by phagocytosing/clearing amyloid deposits in AD mice in the absence of genetic manipulation (and perhaps something similar may occur in AD patients). However, careful 3D reconstruction of microglia and amyloid in APP23 or Tg2576 mice fails to show this (Stalder et al., 2001; Wegiel et al., 2004).

    An alternate explanation is that microglia/macrophages secrete a soluble factor (e.g., a cytokine or protease) that limits cerebral amyloidosis. Yet, the converse—that reactive glia produce acute phase reactants/cytokines such as ApoE, ACT and IL-1 that promote amyloidosis—has been shown (Potter et al., 2001; Nilsson et al., 2001). In light of these reports, what is the authors’ take on the mechanism responsible for their finding?

    El Khoury et al. also report that Ccr2 deletion limits the lifespan of Tg2576 animals, and suggest that there is a connection between increased AD-like pathology in Ccr2-deficient Tg2576 mice and their premature death. This conclusion should be taken with caution. Although not often pointed out, Tg2576 mice actually overexpress the mutant human APP transgene in regions other than the brain (for example, peripheral vascular smooth muscle cells and endothelial cells), and it is well-established that transgene-derived Aβ is easily detected systemically in these mice, so early death of Ccr2-deficient Tg2576 mice may be CNS-independent.

    The paper by Fan and colleagues presents an interesting set of results that suggest dampening microglial activation via minocycline treatment is beneficial in their mouse model of vascular amyloidosis. Interestingly, they found reduction in “activated” microglia that corresponded with mitigation of behavioral impairment. Their results fit well with the work of Greg Cole’s group, who showed that treatment of Tg2576 mice with the non-steroidal anti-inflammatory drug (NSAID) ibuprofen or the naturally occurring NSAID curcumin reduces microglial activation concomitant with reduced cerebral amyloidosis and behavioral impairment (Lim et al., 2000; Lim et al, 2001a; Lim et al., 2001b). Fan and colleagues’ data also fit well with our previous results showing that genetic or pharmacologic interruption of CD40-CD40 ligand interaction mitigates microglial activation in response to Aβ peptides, and reduces microgliosis, cerebral amyloidosis, and behavioral impairment in AD mouse models (Tan, Town et al., 1999; Tan, Town et al., 2002; Town et al., 2001; Todd Roach et al., 2004).

    When taken together, the studies suggest that “activation” of microglia/macrophages is not simply one phenotype. We have suggested that these innate immune cells may respond with a range of responses from pro-phagocytic/anti-inflammatory to anti-phagocytic/proinflammatory (Town et al., 2005). Understanding the molecular underpinnings of these various responses of microglia/macrophages will likely be key in targeting these cells for therapeutic intervention in neurodegenerative diseases (particularly AD).


    . Ccr2 deficiency impairs microglial accumulation and accelerates progression of Alzheimer-like disease. Nat Med. 2007 Apr;13(4):432-8. PubMed.

    . Minocycline reduces microglial activation and improves behavioral deficits in a transgenic model of cerebral microvascular amyloid. J Neurosci. 2007 Mar 21;27(12):3057-63. PubMed.

    . Ibuprofen suppresses plaque pathology and inflammation in a mouse model for Alzheimer's disease. J Neurosci. 2000 Aug 1;20(15):5709-14. PubMed.

    . Ibuprofen effects on Alzheimer pathology and open field activity in APPsw transgenic mice. Neurobiol Aging. 2001 Nov-Dec;22(6):983-91. PubMed.

    . The curry spice curcumin reduces oxidative damage and amyloid pathology in an Alzheimer transgenic mouse. J Neurosci. 2001 Nov 1;21(21):8370-7. PubMed.

    . Alpha-1-antichymotrypsin promotes beta-sheet amyloid plaque deposition in a transgenic mouse model of Alzheimer's disease. J Neurosci. 2001 Mar 1;21(5):1444-51. PubMed.

    . The inflammation-induced pathological chaperones ACT and apo-E are necessary catalysts of Alzheimer amyloid formation. Neurobiol Aging. 2001 Nov-Dec;22(6):923-30. PubMed.

    . 3D-Reconstruction of microglia and amyloid in APP23 transgenic mice: no evidence of intracellular amyloid. Neurobiol Aging. 2001 May-Jun;22(3):427-34. PubMed.

    . Invasion of hematopoietic cells into the brain of amyloid precursor protein transgenic mice. J Neurosci. 2005 Nov 30;25(48):11125-32. PubMed.

    . Microglial activation resulting from CD40-CD40L interaction after beta-amyloid stimulation. Science. 1999 Dec 17;286(5448):2352-5. PubMed.

    . Role of CD40 ligand in amyloidosis in transgenic Alzheimer's mice. Nat Neurosci. 2002 Dec;5(12):1288-93. PubMed.

    . Behavioral effects of CD40-CD40L pathway disruption in aged PSAPP mice. Brain Res. 2004 Jul 23;1015(1-2):161-8. PubMed.

    . CD40 signaling and Alzheimer's disease pathogenesis. Neurochem Int. 2001 Nov-Dec;39(5-6):371-80. PubMed.

    . The microglial "activation" continuum: from innate to adaptive responses. J Neuroinflammation. 2005 Oct 31;2:24. PubMed.

    . Cells of monocyte/microglial lineage are involved in both microvessel amyloidosis and fibrillar plaque formation in APPsw tg mice. Brain Res. 2004 Oct 1;1022(1-2):19-29. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Calling for Backup: Microglia from Bone Marrow Fight Plaques in AD Mice

Paper Citations

  1. . Identification of monocyte chemoattractant protein-1 in senile plaques and reactive microglia of Alzheimer's disease. Psychiatry Clin Neurosci. 1997 Jun;51(3):135-8. PubMed.
  2. . Amyloid-beta-induced chemokine production in primary human macrophages and astrocytes. J Neuroimmunol. 2002 Jun;127(1-2):160-8. PubMed.
  3. . CD36 mediates the innate host response to beta-amyloid. J Exp Med. 2003 Jun 16;197(12):1657-66. PubMed.
  4. . Inhibitory effect of minocycline on amyloid beta fibril formation and human microglial activation. Glia. 2006 Feb;53(3):233-40. PubMed.
  5. . Minocycline affects microglia activation, Abeta deposition, and behavior in APP-tg mice. Glia. 2006 May;53(7):776-82. PubMed.

Other Citations

  1. ARF comment

Further Reading

Primary Papers

  1. . Ccr2 deficiency impairs microglial accumulation and accelerates progression of Alzheimer-like disease. Nat Med. 2007 Apr;13(4):432-8. PubMed.
  2. . Minocycline reduces microglial activation and improves behavioral deficits in a transgenic model of cerebral microvascular amyloid. J Neurosci. 2007 Mar 21;27(12):3057-63. PubMed.