New Microglial Model May Aid Genetic Studies, Drug Screening

05 Jan 2018

Human Microglia in a Dish? 

Monocyte-derived microglia-like (MDMi) cells (green) appear to mimic the brain-resident cells.  [Courtesy of Science Translational Medicine/AAAS.] caption

Emerging data suggest that much of the genetic risk of Alzheimer’s disease plays out in microglia, the brain’s resident immune cells. If true, then studying their gene expression might pinpoint functional disease variants and identify pathological mechanisms. Obtaining enough microglia for such studies has been hard, but an article in Science Translational Medicine last December 20 offers a solution. Researchers led by Elizabeth Bradshaw and Philip De Jager at Columbia University Medical School, New York City, generated microglia-like cells from circulating human monocytes, which are plentiful in blood samples. The resulting cells behaved like microglia and expressed key microglial genes, the authors report, suggesting they might serve as a model for these ephemeral brain cells.

• A new protocol for turning monocytes into microglia-like cells generates large enough quantities for genetic studies.
• The cells closely match the behavior and gene expression of brain-derived microglia.
• This model system could help identify functional genetic variants for AD risk genes.

The authors made microglia-like cells from a large cohort of healthy volunteers, then examined the cells for genetic variants that affected the expression of risk genes for AD and other disease. These types of variants are known as expression quantitative trait loci. The authors found 141 of these eQTLs, then compared those findings to eQTLs identified in their previous study of monocytes (May 2014 news). While most of the associations between polymorphisms and gene expression were the same in the two cell types, the authors found differences as well, with some associations only showing up in the microglial model. This highlights the importance of studying disease risk in the right cell type, Bradshaw noted. “The [cellular] context is crucial,” she said.

Terrence Town at the University of Southern California, Los Angeles, said the model impressed him. “The authors painstakingly characterized the cells. They have a robust system for evaluating genetic risk factors in human-derived microglia. I think it will be a valuable resource for the scientific community,” he told Alzforum.

Microglia have been grabbing headlines in AD research ever since the discovery of variants in the microglial receptor TREM2 as some of the strongest genetic risk factors for the disease. In fact, in 2017, geneticists reported that the majority of AD genes are controlled by a master regulator of microglial gene expression, suggesting these immune cells might be responsible for the lion’s share of AD risk (Jun 2017 news). Studying risk genes in microglia has been challenging, however. Isolated directly from the brain, the cells rapidly change in cell culture (Jun 2017 news). Although scientists have generated microglia from induced pluripotent stem cells, these procedures take months and produce few cells, prohibiting large-scale association studies (Jul 2016 conference news). 

Bradshaw wondered if using monocytes as starting material might work better. Some controversial research suggests that these cells can enter the brain during disease states and differentiate into macrophages that are then difficult to distinguish from resident microglia. First author Katie Ryan started with existing protocols that expose human monocytes to a cocktail of cytokines to induce them to become phagocytic cells. Combining two of the protocols, she found that the resulting monocyte-derived microglia-like (MDMi) cells resembled brain microglia more than they resembled monocytes or monocyte-derived macrophages (MDMs) generated in vitro. Intriguingly, MDMi turned up expression of signature microglial genes and responded to inflammatory stimuli in the way microglia do, for example by expressing the cytokine IL10 (Durafourt et al., 2012). Also like microglia, they lived longer in culture than did monocytes or macrophages (Aug 2017 news). 

The authors took advantage of recent microglia gene-expression studies to extensively characterize the MDMis (Jan 2016 news). Of the 368 genes that were turned up or down in MDMi cells compared to their baseline expression in human prefrontal cortex, 55 percent were also similarly regulated in microglia made from iPSCs, and 32 percent in microglia isolated from human brain (Muffat et al., 2016; Zhang et al., 2016). “The MDMi cells seem to be as good a model as the induced microglia,” Bradshaw said.

Is a 32 percent overlap of microglia-specific genes between MDMi and brain microglia good enough? Town said the gene-expression differences did not trouble him. He noted that gene expression in microglia varies with their activation state and environment, and that cultured cells might never completely mimic those in the brain (Jul 2016 conference news). Bradshaw and collaborators will culture the MDMi cells with induced neurons and astrocytes as well as in three-dimensional culture systems to see if that nudges gene expression closer to that of brain-resident cells (Oct 2014 news). 

In the meantime, the authors asked whether disease-associated genes were regulated differently in these MDMi compared with other cells. They generated MDMi from frozen blood samples donated by 95 healthy young people who had participated in a genotyping study. Then they searched for genetic variants that correlated with expression changes in 94 genes linked to either AD, Parkinson’s disease, or multiple sclerosis, turning up the 141 eQTLs. But do these variants actually cause disease? To test this, the authors ran a co-localization analysis with known risk SNPs at those loci. For six genes, an eQTL matched up with a known risk SNP, suggesting that the expression changes might drive risk.

Notably, for two of these genes, LRRK2 and PILRB, risk factors for PD and AD, respectively, the authors found eQTLs in the MDMi cells that did not turn up in their previous study on monocytes. This finding underlines the idea that the genetic risk may be microglia-specific, Bradshaw said. For both LRRK2 and PILRB, increased expression in the MDMi cells associated with SNPs for higher risk. For LRRK2, a major PD risk gene, the association appeared particularly robust, agreeing with previous findings from neurons that too much LRRK2 can harm cells (Jan 2014 news). 

The scientists plan to repeat this genetic correlation study in microglia isolated from brain tissue samples removed during surgery. If the eQTL/gene expression levels match, it will further validate the MDMi cells as a model. Bradshaw believes the cells could be useful for drug screening, and is starting a company to pursue this.

The authors will perform further gene-expression studies. They are obtaining blood samples from Parkinson’s patients to look for gene expression changes in the disease, which may be more informative than looking at healthy controls alone. Larger populations might also uncover more associations. TREM2 variants did not show up in this study, perhaps because they are rare and may not have been present in this small cohort, Bradshaw said.—Madolyn Bowman Rogers

https://www.alzforum.org/news/research-news/new-microglial-model-may-aid-genetic-studies-drug-screening