17 Apr 2025

Ever since the discovery of AD risk-stoking variants of TREM2, the microglial receptor has given scientists seeking to nail down its function a run for their money. It binds many ligands, supports manifold microglial functions, and seems to exert different effects throughout a person’s many years of disease progression. As if that weren’t enough, scientists at the AD/PD meeting, held April 1-5 in Vienna, presented new data revealing unexpected behavior for the microglial receptor.

For one, microglia hovering around plaques—the type that express the most TREM2—reportedly become desensitized to stimulation by TREM2 agonist antibodies. This complicates TREM2 as a therapeutic target, as these microglia are the ones whose functions scientists most hope to boost with agonists. Curiously, these same microglia had no problem springing into action when tickled by an anti-Aβ antibody, suggesting they are recalcitrant specifically to TREM2 agonism. Separately, two independent reports at AD/PD proposed that TREM2 helps quell neuronal hyperexcitability, implying that tweaking TREM2 signaling could do more than support plaque clean-up.

Christian Haass’s group at the German Center for Neurodegenerative Diseases, Munich, investigated how TREM2 expression level influences microglial function, as well as the cells’ response to TREM2 agonist antibodies. Alzforum covered some of this work when it appeared in a preprint (Aug 2024 conference news on Feiten et al., 2024). In brief, first author Astrid Feitan and colleagues generated a TREM2 reporter mouse in which the fluorophore mKate reflects TREM2 expression. This allowed her to distinguish microglia expressing varying levels of the receptor.

In amyloidosis models, it turned out that the closer microglia sat to an amyloid plaque, the more TREM2 they made. When the scientists removed and sorted these microglia based on TREM2 expression, and fed them fluorescently labeled myelin to test their phagocytic prowess, they were surprised to find that the cells expressing low or high TREM2 had little appetite, while the mid-TREM2 microglia gobbled up the myelin with gusto. This suggested to Haass that TREM2-high microglia might be maxed out in their capacity for phagocytosis.

Would TREM2 agonism change this picture? To test this, the scientists treated mice with 4D9:ATV. Developed by Denali Therapeutics, this TREM2 agonist antibody is equipped with an antibody transport vehicle (ATV) that uses the transferrin receptor (TfR) to ferry the antibody across the blood-brain border. In APP knock-in mice expressing human TfR, the antibody incited microglia to mop up the halo around Aβ plaques, yet the cells stopped short of full plaque removal.

Polar Opposites. The samemetabolites whose concentration rose in microglia expressing a “middle” amount of TREM2 (red) in response to TREM2 agonism showed a drop in microglia expressing much more TREM2 (blue). [Courtesy of Christian Haass, DZNE, 2025.]

Next, the scientists treated the mice once a month from 8-12 months of age, then sorted their microglia by TREM2 expression. When they looked if the cells’ metabolites and lipids had changed with this course of TREM2 agonist therapy, they were in for another surprise, Haass said. Microglia expressing mid- or high levels of TREM2 mounted a robust response to TREM2 agonism, but in opposite directions. “Whatever went up in TREM2-mid microglia, went down in TREM2-high cells,” Haass said. For example, metabolites reflecting glycolysis, glucose uptake, and processing in lysosomes ramped up following TREM2 agonism in TREM2-mid microglia, but diminished in TREM2-high cells.

Analysis of microglial gene expression profiles in untreated mice pointed to a potential culprit behind the opposing responses. While both TREM2-mid and -high microglia expressed a cadre of disease-associated microglial (DAM) genes in amyloidosis mice relative to wild-type, the scientists spotted a notable exception. The gene encoding the tyrosine kinase Syk, which mediates TREM2 signaling, was expressed less in TREM2-high microglia. “As soon as TREM2 goes too high, the cells start to downregulate their own TREM2 signal,” Haass believes. This likely happens soon after microglia start congregating around plaques and consuming them, i.e., early in the process of amyloidosis.

To Haass’s mind, these results spell trouble. “For those who want to target this receptor with agonists, this is a translational nightmare,” he said. For TREM2 agonists to be safe and effective, the right population of microglia must be targeted at the right time. “This is really, really difficult to do,” Haass said. Especially in light of the brutally negative results of Alector’s Phase 2 trial of its antibody, AL002, Haass urged scientists to learn more about the biology of TREM2 before putting more agonists into people. In August 2023, Denali scrapped clinical development of its TREM2 agonist antibody, DNL919; other contenders are moving through early trials (see Part 3 of this series).

Fiona Ducotterd of University College London heads UCL’s Drug Discovery Institute. She agreed finding the right window to turn on TREM2 will be key. “At what point in the disease is activation of TREM2 going to be beneficial, and at what point is it too late, or even detrimental?” she said. Despite these unknowns, Ducotterd thinks these questions can be answered only by cautiously moving forward with clinical studies.

Commending Haass for beautiful data, Oleg Butovsky said it evoked the Queen song “Too Much Love Will Kill You.” Butovsky, of Brigham and Women’s Hospital in Boston, interpreted the findings as a sign of immune exhaustion of the plaque-adjacent microglia. He advocated for a different approach. Butovsky thinks blocking checkpoint inhibitors will unfetter microglia, allowing them to carry out neuroprotective functions. In Vienna, Butovsky presented findings relating to two such checkpoint inhibitors, Itgb8 and TIM3 (AD/PD story to come).

David Holtzman of Washington University, St. Louis, told Alzforum that he shares Haass’ concern about targeting TREM2. Previous work from his lab suggests that while TREM2 may help microglia respond appropriately to amyloidosis, the cells sometimes do the opposite in response to tau pathology, where they egg on neurodegeneration (Oct 2017 news; Oct 2022 news). Holtzman asked what advantages TREM2 agonism has over anti-Aβ immunotherapy, which induces microglia to mop up plaques. TREM2 agonism can cause ARIA, too.

Anti-Aβ Antibodies: Microglial Energizers?
To get at this question, Lis de Weerd, a graduate student in the Haass lab, probed molecular details of how microglia react to anti-Aβ immunotherapies. De Weerd injected APP-SAA knock-in mice with three doses of a mouse version of aducanumab, or isotype control antibody, for four months starting in the mice’s early phase of amyloidosis. As expected, the amyloid plaque burden dropped dose-dependently, as gauged by both mouse amyloid-PET scans and immunohistochemistry. More interestingly, CSF proteomics turned up subtle changes in response to treatment, De Weerd reported in a study posted on BioRxiv today. Most changes in the proteome were due to aging over the four-month treatment period, and treatment with aducanumab nudged these changes back to a more youthful baseline. This happened to CSF tau, Cst7, α-synuclein, Gap43—and none other than TREM2. All of these proteins dropped dose-dependently in response to anti-Aβ treatment.

De Weerd and colleagues next focused on microglia, conducting transcriptomics and lipidomics on microglia from the APP-SAA mice following chronic dosing with aducanumab. Overall, expression of DAM signature genes, including TREM2, dwindled, while expression of homeostatic microglial genes increased. This waning of the DAM signature correlated with the extent of Aβ clearance, suggesting that, as Aβ is cleared, microglia largely return to their homeostatic state. However, some genes bucked this trend. In de Weerd’s study, these included galectin-3, SPP1, IFN-γ, MHC II, Cxcl10, and others involved in antigen presentation and immune responses. They increased in microglia with aducanumab treatment. Afterward, they did not return to baseline.

DAM’d if You Do, DAM’d if You Don’t. Microglia huddling closest to plaques expressed high levels of TREM2 (top). This was independent of anti-Aβ treatment, and dropped farther from plaques (top right graph). Plaque-adjacent microglia also expressed high MHC II (bottom), but did so in an anti-Aβ antibody dose-dependent manner (bottom right). [Courtesy of Lis De Weerd, DZNE, 2025.]

Finally, De Weerd used immunofluorescence to compare levels of select proteins expressed by microglia closest to the remaining plaques versus those expressed by microglia further afield (image above). She found that, in contrast to a brain-wide drop in the DAM signature after anti-Aβ treatment, DAM was brimming in close proximity to plaques. There, microglia within 10 microns of a plaque had revved up DAM signature proteins, including TREM2 and CD68, which were expressed far less in microglia stationed 20 or 30 microns away.

Notably, DAM huddled around plaques regardless of anti-Aβ treatment. The other squad of immune proteins—including IFN-γ, galectin-3, and MHC II—were also ramped up in microglia near plaques. This microglial state change, however, only happened in response to anti-Aβ treatment. “This suggests that the [immune] signature we identified in our transcriptomic analysis comes from microglia that are at plaques,” De Weerd said.

To De Weerd, the findings hint that while chronic anti-Aβ antibody treatment promoted a brain-wide return toward homeostasis, it continued to support microglial responses at the frontlines around remaining plaques. “At least with early intervention, we find no indication of microglial ‘exhaustion’ upon chronic anti-Aβ treatment,” De Weerd said. She contrasted this situation to the maxed-out microglia found in response to TREM2 agonism.

The findings mesh with those of a recent human study led by David Gate of Northwestern University. It studied microglia in postmortem brain samples from people who had been immunized with an anti-Aβ vaccine years earlier, and a woman who had died after receiving TPA treatment while on lecanemab (Mar 2025 news). In people who had mounted antibody responses to the vaccine, microglia showed signs of returning to a homeostatic state, while cells surrounding remaining plaques expressed immune response genes. In unimmunized people, or in those whose immune system had mounted a sluggish response to the vaccine, microglia appeared to exist in a state of inflammatory frustration, expressing DAM genes but failing to clear plaques. In the lecanemab case, the ongoing microglial response included DAM genes as well as complement and other immune mediators.

Besides parsing TREM2’s behavior near plaques, in Vienna scientists also proposed a new role of becalmer of hopped-up neurons for this multifarious receptor. Two independent presenters reported that the transmembrane and soluble forms of TREM2, respectively, quelled hyperexcitability, adding yet another plot twist to the unfolding mystery of TREM2 signaling. Read Part 4 in Alzforum’s AD/PD meeting series to find out more.—Jessica Shugart

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

News Citations

1. I’m Open! Exposed Enhancers Reveal Masters of Microglial Moods 16 Aug 2024
2. Changing With the Times: Disease Stage Alters TREM2 Effect on Tau 14 Oct 2017
3. In Mice, TREM2 Antibody Mobilizes Microglia, Yet Worsens Tangles 28 Oct 2022
4. After Immunotherapy, Amyloid Clearance Comes Down to Microglial Moods 14 Mar 2025
5. TREM2: A Chill Pill for Neurons? 17 Apr 2025

Therapeutics Citations

1. AL002
2. DNL919

Paper Citations

1. Feiten AF, Dahm K, vanLengerich B, Suh JH, Reifschneider A, Wefers B, Bartos LM, Wind-Mark K, Schlepckow K, Ulas T, De-Domenico E, Becker M, Khalin I, Davis SS, Wurst W, Plesnila N, Neher JJ, Brendel M, Lewcock JW, DiPaolo G, Capell A, Monroe KM, Schultze JL, Haass C. TREM2 expression level is critical for microglial state, metabolic capacity and efficacy of TREM2 agonism. 2024 Jul 22 10.1101/2024.07.18.604115 (version 1) bioRxiv.

Other Citations

1. Part 3

Further Reading

Papers

1. Das M, Mao W, Shao E, Tamhankar S, Yu GQ, Yu X, Ho K, Wang X, Wang J, Mucke L. Interdependence of neural network dysfunction and microglial alterations in Alzheimer's disease-related models. iScience. 2021 Nov 19;24(11):103245. Epub 2021 Oct 7 PubMed.