Abstract
Neuronal activity is crucial for adaptive circuit remodelling but poses an inherent risk to the stability of the genome across the long lifespan of postmitotic neurons1,2,3,4,5. Whether neurons have acquired specialized genome protection mechanisms that enable them to withstand decades of potentially damaging stimuli during periods of heightened activity is unknown. Here we identify an activity-dependent DNA repair mechanism in which a new form of the NuA4–TIP60 chromatin modifier assembles in activated neurons around the inducible, neuronal-specific transcription factor NPAS4. We purify this complex from the brain and demonstrate its functions in eliciting activity-dependent changes to neuronal transcriptomes and circuitry. By characterizing the landscape of activity-induced DNA double-strand breaks in the brain, we show that NPAS4–NuA4 binds to recurrently damaged regulatory elements and recruits additional DNA repair machinery to stimulate their repair. Gene regulatory elements bound by NPAS4–NuA4 are partially protected against age-dependent accumulation of somatic mutations. Impaired NPAS4–NuA4 signalling leads to a cascade of cellular defects, including dysregulated activity-dependent transcriptional responses, loss of control over neuronal inhibition and genome instability, which all culminate to reduce organismal lifespan. In addition, mutations in several components of the NuA4 complex are reported to lead to neurodevelopmental and autism spectrum disorders. Together, these findings identify a neuronal-specific complex that couples neuronal activity directly to genome preservation, the disruption of which may contribute to developmental disorders, neurodegeneration and ageing.
Main
Sensory experience is essential for proper neuronal maturation and circuit plasticity1. The signalling cascades initiated by experience-driven neuronal activity culminate in the induction of gene programmes that control diverse processes such as dendrite and synapse growth, synapse elimination, recruitment of inhibitory neurotransmission, and adaptive myelination6,7. However, neuronal activity also threatens the genomic integrity of postmitotic neurons that must survive the lifetime of an organism. For example, heightened metabolic demands during periods of elevated activity may increase oxidative damage to actively transcribed regions of the genome8. Activity-induced transcription itself poses a further threat to genome stability, as it has been linked to the induction of repeated DNA double-strand breaks (DSBs) at regulatory elements, such as the promoters of stimulus-inducible genes2,3,4,5,9,10. Although the coupling of transcription to DNA breaks is observed across cell types, this process poses a specific challenge to long-lived neurons, which cannot use replication-dependent DNA repair pathways and possess limited regenerative mechanisms to replace damaged cells11. Accumulating DNA damage to neuronal genomes is a cardinal feature of neurodegenerative disorders and organismal ageing12,13. Thus, understanding the strategies that neurons use to prevent and repair damage may have direct translation to human longevity and ageing therapies. So far, there are no examples of neuronal-specific repair machinery that mitigate the risks of genome instability during heightened activity. By investigating features of the activity-dependent transcriptional programme specific to neurons, we discover a biochemical coupling of neuronal activity to DNA repair through a previously unknown form of the NuA4 chromatin remodeller–DNA repair complex that assembles around the inducible, neuronal-specific transcription factor NPAS4.
Identification of the NPAS4–NuA4 complex
Unlike most activity-inducible transcription factors, which are broadly expressed and induced by various stimuli, NPAS4 is selectively expressed in neurons following membrane depolarization-induced calcium signalling14. To understand the functions of this factor, which is specifically attuned to neuronal activity, we sought to purify NPAS4-containing protein complexes from the adult mouse brain. We reasoned that NPAS4 might assemble into a multisubunit complex that expands its biochemical activities in activated neurons. Using size-exclusion chromatography and non-denaturing gel electrophoresis, we observed that NPAS4 resides in a high molecular weight complex of around 1 MDa. As the predicted size of NPAS4 with either of its heterodimer partners (ARNT1 and ARNT2) is around 175 kDa, this finding suggests that NPAS4 interacts with multiple unknown protein partners (Extended Data Fig. 1a).
To facilitate the purification of this putative NPAS4 complex, we generated Npas4–Flag-HA and Arnt2–Flag-HA knock-in mouse lines in which the epitope tags Flag and haemagglutinin (HA) are appended to the carboxy termini of NPAS4 and ARNT2. In homozygous knock-in mice, we validated the correct genomic insertion of the tags and verified overlapping immunostaining of endogenous NPAS4 or ARNT2 and the HA epitope (Fig. 1a,b and Extended Data Fig. 1b,c). We also demonstrated that wild-type and tagged NPAS4–Flag-HA (NPAS4–FH) exhibit similar expression levels and induction kinetics (Extended Data Fig. 1d). To generate high levels of NPAS4 required for biochemical purification, we stimulated neurons in the hippocampus of Npas4–FH mice through the injection of low-dose kainic acid (KA), a glutamate receptor agonist that synchronously depolarizes hippocampal neurons. We then immunopurified NPAS4 using anti-Flag antibodies and performed mass spectrometry (Fig. 1c and Extended Data Fig. 1e). The mass spectrometry data revealed interactions between NPAS4 and all reported subunits of a single chromatin modifier, the NuA4 complex, the estimated size of which is approximately 1.0–1.3 MDa (Fig. 1c and Supplementary Table 1)15,16,17.
We first demonstrated co-immunoprecipitation between NPAS4 and several NuA4 subunits (TRRAP, EP400 and DMAP1) in the visual cortex using light exposure as a physiological stimulus to induce neuronal activity (Extended Data Fig. 1f). We further characterized this complex by immunoprecipitating either NPAS4 or a component of the NuA4 complex, TIP60 (also known as KAT5), followed by immunoblotting and mass spectrometry analyses. These experiments confirmed that the interaction between NPAS4 and the NuA4 complex is reciprocal. Moreover, NPAS4–ARNT2 dimers were among the most abundant transcription factors associated with NuA4 in the brain. In addition, a new subunit of the complex, the poorly characterized protein ETL4, was identified (Extended Data Figs. 1g–j and 2a,b and Supplementary Table 1). We observed that several subunits of the NuA4 complex interacted before stimulation, which suggests that following NPAS4 induction, NPAS4–ARNT dimers join a pre-existing complex (Extended Data Fig. 2b). However, we cannot exclude the possibility that other subunits associate with, or post-translational modifications are added to, the NuA4 complex following activity. NPAS4 is the major inducible component of the complex at the RNA level (Extended Data Fig. 2c). Neither the related protein NPAS3 nor another activity-inducible factor, FOS, co-immunoprecipitated with NuA4 components in the brain. This result highlights the specificity of the NPAS4–NuA4 interaction (Extended Data Fig. 2b). Moreover, NPAS4, but not other activity-inducible transcription factors such as FOS and EGR1, interacted with NuA4 components in a heterologous expression system (HEK293T cells) (Extended Data Fig. 2d,e). In both human and mouse cells, expression of these 19 new interactors of NPAS4 were enriched in neurons compared with other brain cell types18 (Fig. 1d and Extended Data Fig. 2f,g). This result suggests that within the brain, the NPAS4–NuA4 complex functions specifically within neurons.
To determine whether NPAS4 and NuA4 co-localize on chromatin, we performed CUT&RUN19 to obtain a map of NPAS4–NuA4 genomic binding in stimulated hippocampal nuclei isolated from mice injected with low doses of KA (Fig. 1e and Supplementary Table 2). We confirmed the specificity of the NPAS4 CUT&RUN signal by performing NPAS4 CUT&RUN on nuclei isolated from unstimulated brains (in which little to no NPAS4 is expressed) and on nuclei isolated from brains of stimulated Npas4 knockout mice. This experiment generated a list of 10,225 high-confidence NPAS4-binding sites (Fig. 1e and Extended Data Fig. 3a–c). These binding sites were highly correlated with NPAS4 signal from chromatin immunoprecipitation assays with sequencing (ChIP–seq) and showed enrichment of the E-box and bHLH–PAS binding motifs (Extended Data Fig. 3d,e). NPAS4, its partner ARNT2, the NuA4 component EP400 and the newly identified NuA4 subunit ETL4 also co-localized across the genome in stimulated neurons (Fig. 1e–g and Extended Data Fig. 3f,g). Moreover, binding of both NPAS4 and ETL4 to the genome was highly inducible at NPAS4 sites (Fig. 1f). By contrast, EP400 was present at NPAS4-binding sites before stimulation, which suggests that it may be retained at these sites in the absence of NPAS4 owing to the ability of NuA4 to bind to acetylated histones20 or by NuA4-independent binding of EP400 (Fig. 1f,g). However, significantly more EP400 CUT&RUN signal was observed at NPAS4 sites that lack FOS than the converse (that is, sites with FOS but no NPAS4). This result demonstrates the specificity of NPAS4 and EP400 co-binding rather than a general recruitment of EP400 by activity-inducible transcription factors (Extended Data Fig. 3h,i). In summary, our biochemical evidence and genomic binding assays identified a neuronal-specific form of the NuA4 complex that assembles with NPAS4 in multiple brain regions in an activity-dependent manner.
NPAS4–NuA4-driven inducible gene programmes
We next investigated the functions of the NPAS4–NuA4 complex in the brain.
