The protein Npas4 dampens activated excitatory brain circuits by recruiting inhibitory signals to excitatory neurons. It emerges that this protein has the opposite role in some inhibitory neurons, promoting their activity.
The astounding abilities of the mammalian brain arise from a few core circuit 'motifs'. One such motif is positive feedback1, in which the mutual excitation of pyramidal neurons amplifies small signals. Now, fans of rock legend Jimi Hendrix will immediately recognize the problem this raises: positive-feedback amplification can easily get out of control, and an effect that is awesome in 'Voodoo Child' can lead to epilepsy in brain circuits. Our brains must therefore counteract positive feedback with inhibitory circuit motifs — pyramidal neurons excite several subtypes of inhibitory neuron, which then inhibit those same pyramidal neurons through negative feedback (Fig. 1a). One mystery is how these circuits are adjusted to maintain the excitation–inhibition balance in the brain2. Writing in Cell, Spiegel et al.3 provide insight into this homeostatic balancing act, showing how gene-expression pathways that regulate neuronal circuits are differentially tuned to the function of inhibitory and excitatory motifs.
During development, neuronal identity is determined by the restriction of gene expression to a subtype-specific pattern4. However, gene expression does not then remain static. For our brains to learn and adapt, neurons must respond to changes in the environment, and this dynamism arises in part through activity-dependent changes in gene-expression pathways5. These changes are thought to control activity by, for example, adjusting the effectiveness of excitatory and inhibitory synaptic connections (junctions between neurons that transmit information) in a manner that is specific to both cell and synapse type6. For instance, too much activity boosts the effectiveness of inhibitory synapses acting on excitatory neurons, dampening excitation. Conversely, too little activity increases the effectiveness of excitatory synapses acting on excitatory neurons. Thus, homeostatic plasticity follows a 'circuit logic' that coordinately adjusts excitatory and inhibitory feedback loops to stabilize neuron firing6.
Spiegel and colleagues set out to identify genes that contribute to such neuronal-subtype-specific adjustments. To do this, they generated neuronal cultures that were enriched in either inhibitory or excitatory neurons. When the authors depolarized the cultures (which mimics excitation), the two cell types displayed similar early changes in gene expression. In particular, the expression of several early-response genes, including Npas4, was increased in both cultures.
Things got interesting when Spiegel and co-workers turned their attention to the late response to depolarization. After six hours, there was a substantial increase in the number of genes whose expression was modified, but the fraction of modified genes that was shared by inhibitory and excitatory neurons was smaller than during the early response. The authors then confirmed these results in vivo using an approach that allowed them to probe gene expression in a cell-type-specific manner. Taken together, their results suggest that enhanced activity triggers a shared early transcriptional program in excitatory and inhibitory neurons, which then sets in motion distinct downstream signalling pathways.
The early-response gene Npas4 caught Spiegel and colleagues' attention because the transcription factor that it encodes7 acts to promote homeostasis in excitatory pyramidal neurons by regulating the number of inhibitory synapses they receive8. The authors wondered whether Npas4 might have a different function in inhibitory neurons, because enhancing inhibition onto inhibitory neurons would have the paradoxical effect of activating pyramidal neurons — a counterproductive effect for homeostasis.
To test this, Spiegel et al. manipulated Npas4 expression in somatostatin-positive (SST) inhibitory neurons, which mediate a type of feedback inhibition in the brain. Selectively removing Npas4 from SST neurons in brain slices or in cultures containing both inhibitory and excitatory cell types had no effect on the number of inhibitory synapses to SST neurons, but decreased excitatory synapses. Conversely, overexpressing Npas4 in SST neurons increased excitatory synapses to those neurons. Furthermore, the authors found that Npas4 deletion compromised the expression of a subset of late-response genes in SST neurons, but that Npas4 overexpression promoted expression of these same genes.
Spiegel et al. therefore conclude that enhanced neuronal activity activates Npas4 in both cell types. This sets in motion different late-response transcriptional programs that have distinct outcomes — increased excitation of SST neurons and increased inhibition of pyramidal neurons. These two Npas4-mediated gene programs would be expected to synergize, overall inducing increased inhibition of pyramidal neurons and thus counteracting a rise in activity (Fig. 1b).
Although the model is appealing, it is important to bear in mind that brain circuits contain several subtypes of inhibitory neuron, and that the SST–pyramidal circuit is only one of many feedback loops that regulate excitability1. Whether the changes measured here contribute significantly to circuit homeostasis remains unknown.
A second caveat is that, although the model predicts that raising activity should increase excitatory synapses to SST neurons in an Npas4-dependent manner, Speigel and colleagues did not test this prediction directly. Despite the fact that directly reducing or increasing Npas4 expression does modulate synapse number, the effects of Npas4 when manipulated alone may be different from its effects in the context of other activity-induced genes. As such, experiments that confirm the authors' model seem key.
Finally, the study raises the fundamental question of how Npas4 regulates distinct genes in different cell types. Spiegel et al. find a partial answer — regulatory DNA elements that control the expression of Npas4 target genes are in different epigenetic states in the two cell types (epigenetic regulation changes gene expression without altering DNA sequence). This suggests that gene programs underlying homeostasis are epigenetically tuned to the function of each neuron within a neural circuit.
So if listening to Hendrix amps your brain circuits up to 11, don't worry. Dynamic negative feedback loops, working through cell-type-specific effectors such as Npas4, are there to keep a lid on things.
Hangya, B., Pi, H.-J., Kvitsiani, D., Ranade, S. P. & Kepecs, A. Curr. Opin. Neurobiol. 26, 117–124 (2014).
Turrigiano, G. G. & Nelson, S. B. Nature Rev. Neurosci. 5, 97–107 (2004).
Spiegel, I. et al. Cell 157, 1216–1229 (2014).
Edlund, T. & Jessell, T. M. Cell 96, 211–224 (1999).
Lyons, M. R. & West, A. E. Prog. Neurobiol. 94, 259–295 (2011).
Turrigiano, G. Cold Spring Harb. Perspect. Biol. 4, a005736 (2012).
Maya-Vetencourt, J. F. Neural Plast. 2013, 683909 (2013).
Lin, Y. et al. Nature 455, 1198–1204 (2008).