When it comes to the brain, neurons are the superstars. They pass the electrical signals that control our bodies and mind, so they’re pretty important. But have you thought about what neurons would be like without their entourage? Image Kanye without his stylists, makeup artists, assistants, caterers, producers, and drivers – he’d have to worry about doing a lot more work on his own, taking the focus away from his music and his persona. Usually, we don’t think about the people behind Kanye, supporting him and keeping him on track. The same is true when it comes to neurons – they get all the love, and their helpers get left in the shadows.
Your brain is more than its neurons. It also contains glia – the word literally means “glue”. Glial cells make up 90% of your brain matter, and the term covers pretty much anything that isn’t a neuron. Many neuroscientists don’t pay them much attention, but in the last few decades, we’ve begun to learn some exciting new things about the roles played by these glial cells in the brain. For example, astrocytes were once thought to be just “support cells” holding the neurons in place and providing them with nutrients and oxygen, but we now know that astrocytes are essential for helping neurons develop appropriate synaptic connections (1), and many synapses in the brain are enveloped by an astrocytic process that influences neurotransmission and plasticity (2).
This week’s Neurosciences Seminar speaker, Dr. Beth Stevens of Harvard University, studies the ways these under-appreciated cells interact with the immune system, and how those interactions lead to refinement of neural circuits by pruning back unnecessary synapses. The brain has traditionally been considered to be an “immune privileged” site, meaning that foreign antigens could be introduced without inducing inflammation. Because of the limited regenerative ability of the central nervous system, it makes sense that the body would want to reduce inflammatory responses that could damage the brain. That view is changing as we begin to better understand the signalling pathways of the brain’s immune cells, called microglia.
It turns out that microglia are important for synaptic pruning, because that process requires signalling using immune system molecules, part of what is called the “classical complement cascade”. In the body, a protein called C1q will be tagged to the surface of cellular debris. C1q then kicks off a signalling cascade that opsonizes C3 and ultimately results in phagocytosis (or engulfment) by a macrophage, allowing your body to efficiently clear away dead cells. In the brain, these same proteins are expressed at synapses that have been “tagged” for elimination, allowing their identification and phagocytosis by microglia. Dr. Stevens studies these molecules using retinal ganglion cells (RGCs) from the eye, allowing her to look at how changes in synapse elimination affect changes in visual perception. In normal eyes, each retina sends projections to the dorsal lateral geniculate nucleus, and each retina makes connections in distinct, segregate areas within the region – so within a patch of dLGN, you’d only see synapses from the left eye or the right eye, not both. In mice lacking C1q and C3 due to a genetic knock out, Dr. Stevens found that the visual system doesn’t develop properly into its normal, eye-specific territories, and you see a lot of overlap – meaning that without C1q or C3, the RGCs aren’t properly pruned (3). It turns out that expression of C1q is also tied to astrocytes – the presence of astrocytes in the cell culture led to increased levels of C1q in the neurons, implicating some kind of astrocyte-secreted protein factor in C1q expression. Given that astrocytes are known to produce cytokines, and cytokines are known to influence expression of complement genes, this connection is not unexpected – but until recently, the identity of the astrocyte cytokine factor was unknown.
In recent work out of her lab, Dr. Stevens and her team were able to identify this factor through in vitro studies with astrocytes and RGCs: transforming growth factor-β (TGF-β) (4). TGF-β is a cytokine secreted by astrocytes, and when media from those astrocytes was placed on RGCs in culture, C1q increased in expression within 15 minutes (Fig1). The lab decided to focus on cytokine factors, given the connection with the complement pathway, and looked specifically at a few proteins found in the media (Fig2A). Among the candidate protein factors screened, only TGF-β proved to be necessary for upregulation of C1q expression (Fig 2B), as immunodepletion to remove TGF-β from media prevented C1q upregulation. TGF-β selectively upregulated C1q in RGCs and not microglia or astrocytes (Fig 2G), and further investigation implicated a particular isoform of TGF-β as the culprit – TGF-β3 (Fig2D,F).
Figure 1 – C1q is rapidly upregulated in neurons in response to astrocyte-secreted factors
Figure 2 – TGF-β is necessary and sufficient for neuronal C1q upregulation in vitro
Furthermore, Dr. Stevens and her team were able to determine that TGF-β regulation of C1q is linked to the period of synaptic refinement in the retina in vivo, showing its importance in living animals, by looking at TGF-β expression at several time points (Fig3). At five dates after birth (P5), the time at which C1q expression in the mouse retina peaks, they saw a similar peak in TGF-β expression selectively in the RGCs, with a drop off by P10 (Fig3A-C). To make sure this TGF-β expression was directly regulated C1q expression, they knocked out TGF-β in the retina (using a Cre driver to prevent its expression in retinal neurons) and using anti-TGF-β antibody to block TGF-β action in normal, wild-type mouse retinas. The TGF-β-Cre knock out (KO) mice showed a decrease in C1q expression in RGCs at P5 and not in other cell types (Fig4), and injection of the anti-TGF-β antibody led to a 40-50% reduction of C1q staining in RGCs (not shown). This supports the hypothesis that TGF-β directly regulates C1q expression in RGCs.
Figure 3 – TGF-β expression corresponds to synaptic refinement period in the retinogeniculate system
Figure 4 – TGF-β signaling is required for neuronal C1q expression in vivo
C1q and C3 expression is seen at synapses of the dorsal lateral geniculate nucleus (dLGN), the thalamic brain structure downstream of the retina, but the source of C1q was not clear until now. Dr. Steven’s lab used immunohistochemical staining to determine that C1q is expressed in the axons within the optic nerve, and that the TGF-β knock-out mice showed significantly reduced levels of C1q in these axons (Fig 4E,F). Looking at C1q expression directly in the dLGN, they found that the knock-out mice had reduced expression levels not seen in the primary visual cortex (V1) (Fig5A-C). Since the knock-out selectively removed TGF-β from RGCs only, this implicates RGCs as a source of C1q for the dLGN. The expression of C1q in these knock-out animals wasn’t just reduced throughout the dLGN; examining more closely, the lab found a significant decrease in C1q expression localized to synaptic puncta (5D-G), where C1q needs to be expressed to signal phagocytosis. This led the researchers to conclude that the majority of synaptic C1q in the dLGN is supplied by RGC neurons synapsing in the dLGN, and that other local cells can’t compensate for the loss of TGF-β in the knock-out animals.
Figure 5 – Retinal TGF-β signaling is required for complement localization in the dLGN
In the final part of their study, the group examined whether or not TGF-β regulation of C1q expression was necessary for the dLGN to form normal, eye-specific territories. Turns out – it is! By injecting tracer dyes into the retinas of knock-out and wild-type mice, the researchers were able to visualize the territories where the RGCs synapses in the dLGN using two different dyes (one in each eye). In a normal eye, because each retina forms distinct, individual territories, one would not expect to see a high degree of overlap between the dyes in the dLGN. More overlap of the dyes would indicate a decrease in eye-specific territory formation. Dr. Stevens had seen this earlier when she selectively knock-out C1q expression, and a very similar phenotype was seen with the TGF-β retina-specific knock out – segregation of eye-specific territories is reduced in both lines (Fig6A,B). This is perhaps a little easier to visualize in Figure 7, where green and red represent the projections from either eye, and yellow represents the overlap of projections from both eyes. The only change Dr. Stevens saw was an increase in overlap; there was no change in dLGN total area (Fig7C). To demonstrate that TGF-β and C1q are part of the same pathway, the lab injected anti-TGF-β antibodies into the retinas of C1q knock out mice. If these two factors acted by different pathways, one might expect to see a “stacking” of effects, and thus a stronger phenotype – but if they act by the same pathway, the phenotype should be the same. Indeed, this is what the lab saw (Fig7A,B). All of this demonstrates that retinal TGF-β expression is necessary for eye-specific segregation of retinal projections, and that C1q and TGF-β likely act via the same pathway.
Figure 6 – TGF-β signaling and C1q are required for eye specific segregation and microglia-mediated pruning in the retinogeniculate system
Figure 7 – Mice deficient in C1q or retinal TGF-β signaling show increased overlap of contralateral and ipsilateral areas in the dLGN
Finally, to connect all of this work to the microglial phagocytosis I mentioned at the beginning of the post, Dr. Stevens used an assay she had previously developed to demonstrate that TGF-β and C1q knock out mice showed reductions in microglial engulfment of RGC inputs to the dLGN at P5 (Fig6D) but no differences in microglial number or distribution (not shown). As the researchers put it, “Taken together, our findings support a model in which retinal TGF-β signaling controls C1q expression and local release in the dLGN to regulate microglia-mediated, complement-dependent synaptic refinement”.
Going forward, Dr. Stevens lab has several questions to investigate. Recent work has implicated a role for neuronal activity in microglial phagocytosis of synapses as microglia appear to target less-active synapses for elimination (5); however, it’s not known whether or not neuronal activity regulates this complement signaling pathway or how it might do so. How does C1q from the RGCs arrive in the dLGN and localize to dLGN synapses? Does TGF-β/C1q signalling play a role in developmental synapse elimination across the CNS, or is this pathway specific to the visual system? And finally, TGF-β has been implicated in the amyloid-beta plaques seen in Alzheimer’s disease (6), and having a deficiency in C1q has been shown to be neuroprotective (7). Is it possible that this complement pathway plays a role in the synapse loss and/or dysfunction seen in Alzheimer’s disease?
Dr. Beth Stevens will be speaking tomorrow afternoon at 4 PM in the CNCB Marilyn Farquhar Seminar Room here at UCSD. If you’re interested in how she’s moving forward to try and answer some of these questions, please join us for what is sure to be a great presentation on the importance of these little immune cells and the full entourage of mechanisms that keep your Kanye-brain healthy, clean, and ready to rock.
Alie Caldwell is a second-year student in the UCSD Neurosciences Graduate Program. She works under Dr. Nicola Allen studying the roles of astrocyte-secreted factors in synapse formation using mouse models of neurodevelopmental disorders. She creates educational neuroscience YouTube videos on her channel NeuroTransmissions and can be found on Twitter at @alie_astrocyte.
- Chung, W.S. et. al. (2015). “Astrocytes Control Synapse Formation, Function, and Elimination”. Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a020370
- Araque, A. et. al. (1999). “Tripartate synapses: glia, the unacknowledged partner”. Trends Neurosci. 1999 May;22(5):208-15. doi: 10.1016/S0166-2236(98)01349-6
- Stevens, B. et. al. (2007). “The classical complement cascade mediates CNS synapse elimination”. Cell. 2007 Dec 14;131(6):1164-78. doi: 10.1016/j.cell.2007.10.036
- Bialas, A.R. & Stevens, B. (2013). “TGF-β Signaling Regulates Neuronal C1q Expression and Developmental Synaptic Refinement”. Nat Neurosci. 2013 Dec; 16(12): 1773–1782. doi: 10.1038/nn.3560
- Wyss-Coray T. et. al. (1997). “Amyloidgenic role of cytokine TGF-beta1 in transgenic mice and in Alzheimer’s disease.” Nature. 1997 Oct 9;389(6651):603-6. doi: 10.1038/39321
- Fonseca, M.I. et. al. (2004). “Absence of C1q leads to less neuropathology in transgenic mouse models of Alzheimer’s disease”. J Neurosci. 2004 Jul 21;24(29):6457-65. doi: 10.1523/JNEUROSCI.0901-04.2004