To ensure homeostasis in the brain, microglia are constantly feeling about with their finger-like processes, making sure that all is well, or sensing problems and rearing up to help. A study in today’s Science suggests that these processes directly touch the soma, i.e., the stable control center of neurons. Most previous studies had focused on microglial interaction with synapses, which are far away from the soma and more changeable.

  • Microglia contact the cell bodies of 90 percent of neurons.
  • Hotspots on the neurons’ plasma membranes attract microglial processes.
  • These junctions allow constant signaling between microglia and neurons.

An international group of scientists led by Ádám Dénes, Institute of Experimental Medicine, Budapest, Hungary, reported that the soma of neurons in both humans and mice is dotted with docking areas built to receive microglial processes (see video below). Called somatic microglial junctions, these areas are equipped with special machinery to constantly pass signals from the neuron’s mitochondria to the microglia. Key for the signals flying between the two cell types are microglial ATP/ADP-sensing receptors. In essence, the study identifies a new communication site between microglia and neurons, built to maintain neuronal function.

“We think at least in the noninjured brain under physiological conditions, these somatic junctions are one of the main sites of microglia-neuron interactions,” Dénes told Alzforum. “These hotspots provide an ideal platform for microglia to monitor and influence neurons over longer periods.”

Cserép et al., Science, 2019.

“This paper shows beautiful images of interactions between microglia and neurons and provides evidence that microglia and neurons may communicate via purinergic signaling,” Tara Spires-Jones, University of Edinburgh, Scotland, wrote to Alzforum. 

“This is an exciting paper,” wrote Jonas Neher, German Centre for Neurodegenerative Diseases, Tübingen, Germany. “This novel form of microglial-neuronal communication could be very relevant for neurodegenerative diseases with strong implications of mitochondrial involvement, such as Parkinson’s disease.”

Most previous studies of microglia-neuron communication focused on interactions with neuronal synapses, for example synaptic pruning (for a review see Wu et al., 2015; Weinhard et al., 2018). Synapses are located far away from the cell body, can number in the thousands to a large neuron, and are highly dynamic, disappearing and re-forming all the time. Individual microglia live for years, during which time they steadfastly monitor neurons, detect signs of injury, and distinguish between neurons that can be repaired and those that need to be eaten.

It seemed to Dénes that monitoring a more central, stable part of the neuron would make more sense for microglia. In zebrafish and mice, microglia had already been reported to get close to neuronal cell bodies, but it was unclear if they made direct contact (Li et al., 2012; Stowell et al., 2018). 

To find out, first authors Csaba Cserép and Balázs Pósfai used in vivo two-photon imaging to examine brain cell interactions in mice that had fluorescently labeled microglia and neurons. They observed microglial processes essentially laying hands on the soma of cortical neurons, with each “touch” lasting an average of 25 minutes. This contact was more than three times longer than the average 7.5-minute connection microglia also made with dendrites.

Pat Down. Three-dimensional reconstruction of two-photon micrograph stacks shows a microglia (green) “touching” (arrows) the body of a neocortical neuron (red). [Courtesy of Cserép et al., Science/AAAS.]

“This impressive new paper […] shows that microglia spend a disproportionate fraction of their time surveying neuronal somata, as compared with synapses,” David Attwell at University College, London, wrote to Alzforum (see full comment below).

This new type of handsiness is common in the brain, the scientists report. Analyzing brain tissue with three-dimensional confocal laser scanning microscopy revealed that most neurons have such touch points, which the authors named somatic microglial-neuron junctions. More than 90 percent of 443 cortical, neocortical, subcortical, and interneuron cell bodies examined in four mice did, whereas only 10 percent of 1,183 synapses examined did. In human postmortem tissue from three patients examined in this study, 87 percent of 170 neocortical neurons had similar microglial-neuronal soma connections.

The authors used high-resolution electron-tomography to scrutinize these contact sites. They saw a specialized architecture. The spots comprised a unique combination of proteins and organelles that was absent from surrounding membranes (see image below). For instance, the membrane proteins Kv2.1 and Kv2.2 were massing together on the neuronal side. When clustered, these proteins enable exocytosis by anchoring the requisite vesicle fusion proteins and intracellular organelles to the plasma membrane. Gathered just behind these proteins were mitochondria, reticular membrane structures, intracellular tethers, and vesicle-like membrane structures.

In between microglia and neurons, the researchers spotted cell-adhesion molecules connecting the two membranes. They brought the neuronal hotspots face-to-face with clusters of the P2Y12 receptor on the microglial processes. Known to be instrumental for microglia sensing injury or pathological states, this receptor is attracted by ATP and ADP signals.

Unique Build. At somatic microglia-neuron junctions, both neuronal and mitochondrial membranes contain a specific set of proteins and organelles. [Cserép et al., Science, 2019.]

Could neurons signal to microglia at these junctions via ATP/ADP? At least in cultured mouse neurons, ATP-containing vesicles dispatched by neuronal mitochondria appeared between mitochondria and the plasma membrane at these junctions, and ATP-containing vesicles were released from neuronal cell bodies upon stimulation with KCl.

Asking what injury might do to these junctions, Cserép and colleagues modeled stroke in mice by occluding their carotid arteries, then monitored microglia-neuron interactions near the lesion. Stroke caused mitochondria to fragment at somatic microglial junctions, which led the overlying Kv2.1 protein cluster to spread out (see image below). In response, the surface area of microglial processes touching neurons quadrupled. However, when the researchers blocked microglial sensing using the P2Y12R inhibitor PSB0739, microglia no longer responded, leading to a larger lesion and disrupted functional connectivity.

Injury TLC. In mice, a stroke leads to fragmented mitochondria, which signal for more microglial processes to contact the neuronal soma. Electron-tomographic three-dimensional reconstruction. [Courtesy of Science/AAAS.]

Together, the data suggest that somatic microglial junctions pass an ongoing readout of neuronal activity to microglia. Junctions falling silent in response to injury triggers microglia to surround the damaged neurons and activate protective responses.

Dénes emphasized that somatic microglia junctions may well enable microglia to sense molecules beyond ATP/ADP that come from the neuron, including other exocytic or signaling molecules. “In a way these are windows on the neuronal membrane, where neurons communicate with their environment,” Dénes said, noting that he will further define their cellular and protein makeup.

The scientists will also assess how this mode of communication is altered in diseases, both in animal models and human postmortem tissue. “We hypothesize that in brain diseases such as Alzheimer’s or Parkinson’s, these interactions could be changed, which means they could represent a potential therapeutic target,” Dénes said. Fully understanding the molecular composition of these sites might ultimately enable researchers to influence microglia to save degenerating neurons, or change how they phagocytose injured cells.

Spires-Jones agreed the work could have implications for AD research. For instance, it may complement recent work suggesting pathological tau causes alterations in the neuronal membrane that trigger microglia to eat them (Aug 2018 news).—Gwyneth Dickey Zakaib and Gabrielle Strobel


  1. This impressive new paper from Adam Dénes’s lab shows that microglia spend a disproportionate fraction of their time surveying neuronal somata, as compared with synapses. This is intriguing, because microglia are generally assumed to play a major role in regulating synapse number and strength, both developmentally and in pathology (Paolicelli et al., 2011; Schafer et al., 2012; Pfeiffer et al., 2016Sipe et al., 2016; Hong et al, 2016).

    Prolonged somatic contacts may be a consequence of attraction to the soma of microglial processes, which highly express ADP-activated P2Y12 receptors at their tips (Dissing-Olesen, 2014), by release of ATP/ADP at specialised neuronal sites. Such sites were found to be associated in the neuronal soma with Kv2.1 channels, proximity to mitochondria, and expression of the vesicular nucleotide transporter. Intriguingly, blocking P2Y12 receptors shortened the attachment time of microglial processes to somata, but did not affect the time microglia spent surveying synapses.

    On formation of microglial-neuronal soma contacts, NADH concentration was observed to rise in neuronal mitochondria, implying some kind of signalling from the microglial cell to the neuron. This could perhaps relate to the assembly of cell adhesion molecules, as observed in this paper at the junction, or it may reflect signalling mediating the previously reported calming action of microglia on neuronal firing (Li et al, 2012; Peng et al., 2019). 

    One “metabolic” possibility is that activation of microglial P2Y12 receptors, which are known to activate THIK-1 K+ channels in their membrane (Madry et al, 2018), may raise the extracellular [K+] locally and thus stimulate neuronal Na/K pumps, increasing neuronal ATP consumption.

    These microglial-somata contacts may also play an important role in pathology. After ischaemia more such contacts formed, and block of P2Y12 receptors to inhibit junction formation showed that they reduced neuronal Ca2+ overload and neuronal damage.

    This fascinating paper raises numerous questions, including:

    (i) Does the apparently disproportionate time microglia spent on surveilling neuronal somata, as opposed to synapses, reflect the fact that there are thousands of synapses to be surveyed within a microglial cell’s volume of influence, so each one needs to be “inspected” relatively briefly, or does it simply reflect more ATP/ADP release from somata?

    (ii) Why is such a long microglial process dwell time needed to survey and influence somatic function?

    (iii) Apart from fostering interactions with microglia, are these novel ATP release sites in neuronal somata the source of the ATP which activates astrocytes to mediate neurovascular coupling at the capillary level (Mishra et al., 2016)? 

    (iv) How does the positive effect of microglial contact on neuronal health in pathological conditions, reported here, relate to the removal of neurons by microglia in some situations (Brown and Neher, 2014)?

    There is clearly far more going on in the interactions between microglia and neurons than was previously suspected!


    . Synaptic pruning by microglia is necessary for normal brain development. Science. 2011 Sep 9;333(6048):1456-8. PubMed.

    . Microglia sculpt postnatal neural circuits in an activity and complement-dependent manner. Neuron. 2012 May 24;74(4):691-705. PubMed.

    . Induction of hippocampal long-term potentiation increases the morphological dynamics of microglial processes and prolongs their contacts with dendritic spines. Sci Rep. 2016 Sep 8;6:32422. PubMed.

    . Microglial P2Y12 is necessary for synaptic plasticity in mouse visual cortex. Nat Commun. 2016 Mar 7;7:10905. PubMed.

    . Activation of neuronal NMDA receptors triggers transient ATP-mediated microglial process outgrowth. J Neurosci. 2014 Aug 6;34(32):10511-27. PubMed.

    . Reciprocal regulation between resting microglial dynamics and neuronal activity in vivo. Dev Cell. 2012 Dec 11;23(6):1189-202. Epub 2012 Nov 29 PubMed.

    . Microglial P2Y12 receptor regulates ventral hippocampal CA1 neuronal excitability and innate fear in mice. Mol Brain. 2019 Aug 19;12(1):71. PubMed.

    . Microglial Ramification, Surveillance, and Interleukin-1β Release Are Regulated by the Two-Pore Domain K+ Channel THIK-1. Neuron. 2018 Jan 17;97(2):299-312.e6. Epub 2017 Dec 28 PubMed.

    . Astrocytes mediate neurovascular signaling to capillary pericytes but not to arterioles. Nat Neurosci. 2016 Dec;19(12):1619-1627. Epub 2016 Oct 24 PubMed.

    . Microglial phagocytosis of live neurons. Nat Rev Neurosci. 2014 Apr;15(4):209-16. PubMed.

  2. This interesting and superbly executed paper homes in on one of the least-understood areas of microglial-neuron interactions, i.e., the unique molecular nature and functional importance of microglial-neuronal soma contacts (named somatic microglial junctions (SMJ)). Not only are the proteins that form these contacts elucidated (Kv2.1 channels on neurons, ADP-sensing P2Y12R on microglia), but their possible importance in preserving neuronal health is explored in a stroke model.

    Since it appears that the same type of junctions occur in the human brain, a major question is whether loss of such contacts occurs in neurodegenerative diseases, and whether this has functional consequences. Interestingly, single-cell microglial transcriptomic analysis shows that the diversity of microglial types during development is reduced to a uniform microglial signature in adult mouse brain. Since microglia contact 80-90 percent of all neurons visualized in the mouse brain via SMJs at steady state, perhaps this one-type signature is due to the somal contacts elucidated here.

    The authors’ focus on delivery of ATP/ADP, and SMJ spreading when mitochondria are dysfunctional in potentially viable neurons, together with exacerbation of stroke damage by inhibition of P2Y12R or in P2Y12R knockout mice, highlights the important neuroprotective role of microglial P2Y12R in stroke. However, the role of mitochondrial dysfunction in neurodegenerative disease is still debated.

    Since HEK cells expressing Kv2.1 clusters still form P2Y12R-rich microglial contacts in co-culture, it would be interesting to see whether mitochondrial substructures, and mitochondrial disruption, play a role here, too. We have noted that neuronal soma containing tau aggregates in the facial nucleus of P301S tau transgenic mice show a peculiar microglial phenotype with much closer proximity compared with wild-type neurons (Brelstaff et al., 2018; Aug 2018 news). These motor neurons also show a redistribution of mitochondria (unpublished data) so similar SMJ adaptations may occur.

    A study of cultured hippocampal neurons from 3xTg mice has implicated loss of function of Kv2.1 channels through increased ROS production leading to hyperexcitability (Frazzini et al., 2016). It will be interesting to test whether microglia would form SMJs with these tau-bearing neurons preferentially, and whether these would be neuroprotective, especially as other studies in the PS19 transgenic tauopathy model suggest that microglia drive neurodegeneration (Shi et al., 2019). 

    Clearly, a whole new set of molecular tools and functional connections between neuronal soma and microglia have been established in this landmark paper that may shed new light on the role of microglia in neurodegeneration.


    . Living Neurons with Tau Filaments Aberrantly Expose Phosphatidylserine and Are Phagocytosed by Microglia. Cell Rep. 2018 Aug 21;24(8):1939-1948.e4. PubMed.

    . Altered Kv2.1 functioning promotes increased excitability in hippocampal neurons of an Alzheimer's disease mouse model. Cell Death Dis. 2016 Feb 18;7:e2100. PubMed.

    . Microglia drive APOE-dependent neurodegeneration in a tauopathy mouse model. J Exp Med. 2019 Nov 4;216(11):2546-2561. Epub 2019 Oct 10 PubMed.

  3. This is an exciting paper, which describes a novel hub for communication between microglia and neurons at the level of the neuronal cell body in mouse and human brains.

    I find it particularly striking that somatic microglial-neuron contacts increase, on the one hand, in response to enhanced neuronal activity in the healthy brain and, on the other hand, upon ischemic neuronal damage, where a breakdown of the described somatic microglial-neuron junctions seems to occur (based on disintegration of Kv2.1 clusters).

    Nevertheless, in both cases the authors found that the enhanced microglial-neuron interaction was mediated by P2Y12R signaling, indicating that different ATP release mechanisms might be at play during health and disease. As the authors note, most previous reports have linked P2Y12R signaling to phagocytic uptake (for review see, e.g., Brown and Neher, 2014). Therefore, I wonder whether the authors’ reported neuroprotective somatic-microglial contact, relatively early after an ischemic insult, would eventually turn into phagocytic engulfment if one were to investigate this interaction for longer periods of time as neuronal loss in the penumbra can occur after the maximum of 24h studied here. 

    Interestingly, as the somatic microglial-neuron contact appears to be responsive to mitochondrial signals and/or dysfunction, this novel form of microglial-neuronal communication could be very relevant for neurodegenerative diseases with strong implications of mitochondrial involvement, e.g., Parkinson’s disease. Also, it will be very interesting to see if and how this interaction changes when neurons bear intracellular amyloid inclusions, e.g., α-synuclein aggregates or tau tangles.


    . Microglial phagocytosis of live neurons. Nat Rev Neurosci. 2014 Apr;15(4):209-16. PubMed.

  4. Previous research on interactions between microglia and neurons has focused on synapses as the targets for microglia. That’s because microglia have receptors for neurotransmitters that are released from synapses, including ATP/ADP, enabling them to detect synaptic function and dysfunction, and because communication through neural circuits will be altered by microglia pruning of these connections.

    Neuroscientists focus so intently on synaptic connections, they can forget that all cells in the body communicate with each other by releasing signaling molecules through many different mechanisms. This paper details a new mechanism for release of signaling molecules from the cell body of the neuron.

    This is important because microglia have a key role in protecting neurons in distress and in removing dead neurons, so a mechanism of intercellular communication located on the cell body is particularly important in this process, whereas synapses are small and located long distances away from the neuron’s cell body.  

    Moreover, the molecule ATP is both a neurotransmitter and an energy source for cells. This newly identified signaling mechanism between the neuron’s cell body and microglia is sensitive to the energy demands of the neuron as reflected in changes in ATP concentration in the organelles that generate ATP, the mitochondria. 

    This enables microglia to monitor the level of neuronal firing and detect metabolic stress, to marshal a protective response to preserve the neuron or begin the process of removing it.

    ATP/ADP have been known to be important neuron-microglial signaling molecules for some time. Most of that research was associated with synaptic signaling. However, our research implicated ATP in communication between axons firing action potentials and glia that make myelin (oligodendrocytes and Schwann cells) and astrocytes, and we identified several mechanisms for how these signaling molecules are released outside of synapses. We have shown that both vesicular and non-vesicular release mechanisms are involved, but this new paper reports a new mechanism of ATP release from the cell body—one that links signaling to mitochondria and thus is modulated by metabolic state.

    In a review article I pointed out that purinergic signaling provides a universal means of cell-cell communication in the brain, because all cells generate ATP and non-synaptic release mechanisms enable communication among neurons and all other cell types in the brain (Fields et al., 2011). 

    This new study greatly advances that general notion and expands our understanding of microglial function.


    . Nonsynaptic and nonvesicular ATP release from neurons and relevance to neuron-glia signaling. Semin Cell Dev Biol. 2011 Apr;22(2):214-9. Epub 2011 Feb 12 PubMed.

Make a Comment

To make a comment you must login or register.


No Available References

Further Reading


  1. . Microglia in steady state. J Clin Invest. 2017 Sep 1;127(9):3201-3209. Epub 2017 Jul 17 PubMed.
  2. . Cerebellar microglia are dynamically unique and survey Purkinje neurons in vivo. Dev Neurobiol. 2018 Jun;78(6):627-644. Epub 2018 Jan 18 PubMed.

Primary Papers

  1. . Microglia monitor and protect neuronal function through specialized somatic purinergic junctions. Science. 2020 Jan 31;367(6477):528-537. Epub 2019 Dec 12 PubMed.