This Week in The Journal

Remodeling Nucleosomes for Long-Term Memory

Miran Yoo, Seongwan Park, Inkyung Jung, and Jin-Hee Han

(see pages 7133–7141)

Long-term memory results from persistent changes in neuronal structure and function. These modifications rely on activity-dependent changes in gene expression mediated by transcription factors that bind to specific DNA sequences. But not all possible binding sites for a given transcription factor are accessible at any time, because DNA is wrapped tightly around histone proteins and compacted into chromatin. Which genes are accessible is determined by epigenetic mechanisms that regulate where histones are positioned and how tightly the DNA is wrapped. Through such regulation, epigenetic modifications help specify cell fate and contribute to long-term memory formation.

Histone positioning and chromatin structure are regulated by ATP-dependent chromatin remodeling proteins including members of the BAF family. BAF proteins assemble into large complexes whose specific composition varies across cell types. Baf53b is a neuron-specific subunit that is required for activity-dependent dendrite growth, spine remodeling, and long-term potentiation in the hippocampus (Choi et al., 2015, Exp Mol Med 47:e155). In addition, Baf53b is upregulated in the lateral amygdala after auditory fear conditioning, and knocking down Baf53b in this region before conditioning reduces freezing responses to the shock-associated auditory cue the day after conditioning. Because Baf53b levels remain elevated in the lateral amygdala for at least 24 h after conditioning, Yoo et al. asked whether this persistent activity is required for longer-term memory.

Knocking down Baf53b levels by ∼50% starting the day after conditioning had no effect on freezing levels the day after that, but it reduced freezing in response to the auditory cue 6 d later. Fear conditioning led to changes in the expression of numerous other genes, and as expected, knocking out Baf53b prevented changes in a subset of these. One gene that underwent Baf53b-dependent upregulation after fear conditioning was fibroblast growth factor 1 (FGF1). Remarkably, injecting FGF1 into the lateral amygdala after fear conditioning and Baf53b knockdown restored fear expression 6 d later.

These results suggest that persistent upregulation of Baf53b in the amygdala contributes to the establishment of long-lasting fear memory by increasing expression of FGF1. Future work should investigate how persistent increases in Baf53b are maintained, how Baf53b alters chromatin structure, and whether Baf53b plays similar roles in other forms of long-term memory.

• Download figure
• Open in new tab
• Download powerpoint

SGIRT can be used to inactivate Mecp2 (blue) in corticofugal projection neurons (red) and restore expression in a subset of corticothalamic projection neurons (green). See Liu et al. for details.

Turning Mecp2 On and Off in Specific Cells

Xue Liu, Liang Ma, Hongzhi Liu, Jingwen Gan, Yidan Xu, et al.

(see pages 7169–7186)

Rett syndrome is a neurodevelopmental disorder caused by mutations in the transcriptional regulator methyl-CpG-binding protein 2 (MECP2). Infants with Rett syndrome develop normally for several months, but then begin to show signs of developmental delays followed by regression in skilled motor function. Studies in mouse models have found that restoring Mecp2 function after symptom onset reverses many Rett-related phenotypes, raising hopes for effective therapy in humans (Guy et al., 2007, Science 315:1143). Understanding when and in what cell types loss of MECP2 leads to specific symptoms might help researchers develop such therapies. Hence, Liu et al. developed a new technique for turning gene function on and off in defined cell types and time windows. They dubbed this technique “single-allele conditional gene inactivation and restoration via recombinase-based flipping of a targeted genomic region,” or SGIRT.

SGIRT combines two established methods for gene recombination: Cre-lox and Flp-FRT. FRT sites were inserted inside loxP sites in introns flanking an essential exon of Mecp2. This arrangement ensured that in the absence of Cre or Flp, wild-type Mecp2 was produced under endogenous regulatory control. Cre or Flp was then expressed at defined times in specific cell types using various methods, such as cell type-specific promoters, inducible promoters, and viral infection. When only one recombinase was expressed in a cell, the flanked exon was stably inverted, leading to production of inactive Mecp2. But when both recombinases were expressed, the exon was restored to its original orientation, allowing the production of wild-type protein.

Using this technique, the authors confirmed that many phenotypes of Mecp2-null mice are replicated when Mecp2 is inactivated selectively in parvalbumin-expressing neurons, and that some of these phenotypes can be rescued by reactivating Mecp2 after symptom onset. Notably, a seizure phenotype seen in Mecp2-null mice did not occur when Mecp2 was inactivated selectively in parvalbumin-expressing neurons, but instead was produced by inactivating Mecp2 selectively in somatostatin-expressing neurons. Future work using these mice should delve deeper into the cell types responsible for specific Rett syndrome phenotypes and determine the extent to which these phenotypes can be reversed by reactivating Mecp2. The SGIRT technique should also be useful for defining cell type- and time-dependent effects of numerous other genes.