In England, the M1 and the M2 motorways take you in almost opposite directions. Does the same happen in the brain? In-vitro, microglia can be driven to adopt M1 and M2 phenotypes, with M1 believed primed to drive inflammation and M2 to drive phagocytosis of debris. Researchers have tried to tie M1 and M2 status to pathology in various neurodegenerative diseases, but ironically, just as the concept has caught on in the field more broadly, leaders in neuroinflammation are beginning to question their usefulness. That was one of the main trends coming out of the 3rd Venusberg Meeting on Neuroinflammation, held 28 February to 2 March at the Biomedical Center, University of Bonn, Germany meeting. "It seems we need to place less emphasis on M1 and M2 now," said David Morgan, University of South Florida, Tampa. Many researchers at the meeting echoed that sentiment. "I think a major conclusion to come from this meeting is that we need to look more broadly at microglial gene expression," summed up Michael Heneka, University of Bonn, on the last evening. Others were less inclined to dismiss the classification so easily. "I think the jury is still out with regard to M1/M2. More functional studies need to be done to get a definitive answer," said Joe El Khoury, Massachusetts General Hospital, Charlestown. Heneka organized this conference, which drew researchers from Europe, North America, and Japan who study various neurodegenerative diseases, including Alzheimer's, multiple sclerosis, and motor neuron disease.
That microglia can both help and hinder the brain in Alzheimer's and other diseases (see ARF related news story) has created some confusion in the field. To try to sort it out, researchers have looked at these cells in terms of their M1 and M2 phenotypes. M1 genes include interleukin (IL) 1β, tumor necrosis factor α, interferon γ, and inducible nitric oxide synthase, while YM1, and IL-4 and -10 fall into the M2 bin. In Venusberg, however, researchers questioned the value of this distinction. Richard Ransohoff, Cleveland Clinic, Ohio, showed that, at least in some cases, M1 and M2 expression cannot distinguish good cells from bad.
Ransohoff studies multiple sclerosis, a neurodegenerative disease in which myelin membranes sustain selective injury. His group uses block face scanning electron microscopy to obtain detailed three-dimensional pictures of glia that initiate myelin damage, and contrasts them with glia that clear myelin from injured regions in the central nervous system of mice. Clearance of debris precedes repair and is therefore beneficial. To figure out which cells are responsible for the clearance, Ransohoff and colleagues crossed mice that express red fluorescent protein in their circulating monocytes with others making green fluorescent protein in brain-resident microglia. The promoters of the monocyte-specific CC-chemokine type 2 receptor gene and the microglial fractalkine receptor gene drive production of the fluorescent proteins. In the double-transgenic mice, monocytes always turn up red, microglia green. The researchers induced experimental autoimmune encephalomyelitis (EAE) in the mice to mimic MS pathology, and then determined which cells went about causing injury and which were mopping up the damage.
Ransohoff reported that the monocytes were invariably the cells that seemed to yank myelin off axons, and often seemed to begin the process at the node of Ranvier, where myelin thins out. Microglia, on the other hand, seem to clear up myelin that has already detached from neurons. Without their red and green labels, could these cells be distinguished in any other way, for example, by M1 and M2 gene expression? Apparently not. Researchers in Ransohoff's lab separated monocytes and microglia by fluorescence-activated cell sorting and then carried out microarray expression analysis in collaboration with Oleg Butovsky at Brigham and Women's Hospital, Boston. The patterns overlapped. "The cells do express individual M1 and M2 genes, which are also nonspecific indicators of myeloid cell activation, but they can express both together or not at all, and when they do express them, it's not necessarily the full panel of genes," said Ransohoff. In short, M1 or M2 gene expression did not differentiate the cells. "I think the categorization has little value for studying microglia," said Ransohoff.
That seemed to be echoed by Monica Carson's research at the University of California, Riverside. Carson studies microglia and macrophages together in cell culture with neurons. She saw that the microglia remain quiescent when neurons in the culture are one week old and inactive, but change their morphology a week later, when neurons start to transmit across synapses. The microglia go from being flat and amorphous to assuming a more defined structure, showing that they can react to their environment and to the differentiation state of the co-cultured neurons.
Carson used these mixed cell cultures to study inflammation. She injected lipopolysaccharide into the brains of mice to induce an immune response and then co-cultured cells from those brains. She reported that macrophages that had infiltrated the brain began to kill the neurons, whereas microglia did not. However, she was surprised to find that the toxic macrophages were predominantly of the M2 phenotype, showing high expression of classic M2 markers, including YM-1, TGFb, and IL-1 receptor antagonist. This runs contrary to the view that M1 states are more toxic. Carson emphasized that other factors besides the cells’ activation state determine toxicity. "We should not ignore the M1/M2 paradigm completely, as it may still be useful, but perhaps more important is the cells’ provenance and context," she said. "Infiltrating macrophages were much more toxic, yet they correlated with M2 activation states."
Terrence Town, who recently moved across town from the Cedars Sinai Medical Center in Los Angeles to the University of Southern California, echoed the idea that context determines neuroinflammation in neurodegenerative diseases such as Alzheimer’s. The M1/M2 concept is oversimplified, Town said. He favors a microglial "continuum" of activation, where M1 and M2 represent extreme ends of the spectrum. He also argued that the field desperately needs functional markers. "Markers for phagocytosis, or pro- and anti-inflammatory states, would really be valuable," he said. Others agreed. "Research in the T cell field took off when people were able to associate markers with function," noted Greg Cole, University of California, Los Angeles. Morgan argued for microarray-based data to get a better picture of expression phenotypes.
El Khoury has taken steps in that regard using direct RNA sequencing (DRS). Developed originally for yeast, El Khoury's group applied this relatively new technique to mouse microglia. The technique requires very little mRNA, and avoids bias introduced by amplification steps or making cDNA. DRS captures mRNA using an oligodT chip that binds mRNA polyA tails, then sequences the captured RNA directly, while using a highly sensitive camera to record the addition of nucleotides one at a time. Thousands of transcripts are sequenced together. The only disadvantage, said El Khoury, is that DRS cannot measure alternative splice variants. "You can only sequence what's at the 3' end of the transcript," he said.
El Khoury used DRS to sample what he calls the microglial "sensome." Microglia are constantly sensing their environment by sending out extensive processes (see ARF related news story), which he believes may have cell surface receptors unique to these surveying cells. He developed magnetic beads to capture the microglial surface marker CD11b, and with those beads, isolated microglia from mouse brain tissue by flow cytometry. He then compared the DRS transcriptome of the brain microglia to that of peritoneal macrophages.
From around 22,000 transcripts, El Khoury identified 2,300 that were specific to the glia. He narrowed that down with bioinformatics to a subset of transcripts that he believes represent the "sensome." This panel of markers, El Khoury claims, distinguishes microglia from astrocytes, macrophages, and other cells in the brain. It covers a broad array of different genes, including chemokines and purinergic receptors. El Khoury used proteomic analysis of 2-D polyacrylamide gels and dual in-situ hybridization to check the DRS data. The latter, for example, showed that sensome genes co-localized with CD11b, indicating they were specifically expressed in microglia.
Eventually, this new information could be used to sample changes during disease or immune challenge, said El Khoury. For now, he applies it to study aging in mice. Comparing the "sensome" of young and old mice, he found genes that appear to be upregulated as the animals age, and seem to be involved in sensing microbes. Neither monocytes nor macrophages show such expression pattern changes with age, he said. Some of those genes fall into the M2 classification. El Khoury speculated that microglia assume a more neuroprotective phenotype with age.
Other scientists were impressed with these findings. "This is groundbreaking work," said Heneka, who emphasized that it will be important to relate these expression patterns to specific microglial functions. Others noted that this type of broad approach will be more informative than relying on M1 and M2 expression patterns. "This work illustrates what is now possible in the microglial field," said Ransohoff.—Tom Fagan.
- Hickman SE, Kingery ND, Ohsumi TK, Borowsky ML, Wang LC, Means TK, El Khoury J. The microglial sensome revealed by direct RNA sequencing. Nat Neurosci. 2013 Dec;16(12):1896-905. Epub 2013 Oct 27 PubMed.
- Microglia and AD—Does the Inflammasome Drive Aβ Pathology?
- Soothing Neuroinflammation Quells Plaques in Mice
- Enter the New Alzheimer’s Gene: TREM2 Variant Triples Risk
- The Good, the Bad, and the Fractious? Ligand’s Role in Brain
- Microglia—Medics or Meddlers in Dementia
- No Rest for Microglia: These Immune Cells Manage Healthy Synapses