15 March 2010. Shhh! Is silence conducive to learning and memory? Perhaps. But before you pack for the Buddhist temple, we’re talking gene silencing here, and only certain genes. In the March 10 Journal of Neuroscience, researchers report that long-term memory requires methylation of histones, core components of the chromatin spools that wind DNA. According to the research, learning causes histone methylation that prevents DNA from unwinding, thereby silencing genes. But that’s not the full story. Some methylation patterns actually do the opposite, weakening chromatin and activating genes. The work suggests a complex balance between gene silencing and activation in the formation of memories, and the researchers identify some genes that may be involved. These include those for brain-derived neurotrophic factor (BDNF) and Zif268, both known players in memory formation (see ARF related news story).
Transcriptional regulation is part and parcel of long-term memory consolidation. Traditionally, this regulation was seen as the domain of transcription factors such as CREB (see ARF related news story). But beyond that, post-translational modification of histones dramatically alters the strength of the chromatin, either exposing DNA or shielding it from factors such as CREB. Links between long-term memory and acetylation and phosphorylation of these proteins emerged (see, e.g., ARF related news story and Chwang et al., 2006), but whether their methylation is also involved has not been clear.
Led by senior author Farah Lubin at the University of Alabama, Birmingham, first author Swati Gupta and colleagues correlated contextual fear conditioning with histone methylation patterns in the hippocampus of young male rats. In this learning paradigm, animals are trained to associate exposure to a new environment with a mild foot shock. The training caused two different methylation events. First, simply placing the animal in a new environment led, one hour later, to demethylation of histone H3 on amino acid lysine nine, dubbed, in short H3K9. The full contextual fear paradigm had the same effect, but in addition, the researchers found trimethylation of H3K4. This additional methylation event did not occur if the new environment and the foot shock were separated by two hours, indicating that it was not simply the result of stress from the foot shock. Interestingly, methylation of H3K9 is normally associated with transcriptional repression, whereas H3K4 methylation activates genes. Since the latter only emerged after the full contextual fear conditioning paradigm, the result implies “that the H3K4 mark may be an associative-learning-specific signal,” write the authors.
This set of experiments demonstrated a relationship between learning and methylation, but did not prove that the latter was a prerequisite for the former. To investigate that, Gupta and colleagues turned to an Mll methylase mutant mouse, since there is evidence this enzyme specifically modifies H3K4 (see Milne et al., 2002). The researchers found that Mll+/- heterozygotes were slow to catch on in the associative learning task. Twenty-four hours after training, methylase-deficient animals only froze one in five times when re-exposed to the fear context, whereas normal littermates froze four in five times. Curiously, the researchers did not directly measure H3K4 trimethylation in the Mll heterozygotes.
If both silencing and activation of genes through histone methylation is important for hippocampal-dependent learning and memory, then are specific sites of histone methylation especially important? In an attempt to address this question, the authors chose to examine methylation patterns around the promoters of BDNF and Zif268, two genes known to be involved in memory consolidation (see ARF related news story). When Gupta and colleagues isolated those regions by chromatin immunoprecipitation, they found increased H3K4 methylation at both the Zif268 and BDNF promoter 1 regions following contextual fear training but not context training alone. “Together, these results demonstrate active regulation of H3K4 trimethylation within specific gene promoters, and further demonstrate that H3K4 trimethylation is regulated in response to fear conditioning,” write the authors.
This research may open up a new avenue for studying memory formation, perhaps even a ”rotary” that connects several others as well. Histone acetylation and DNA methylation also play roles in learning and memory, and histone methylation can influence both. Methylation at H3K9, for example, competes with acetylation at the same site. In agreement with this, the authors found that the general histone deacetylase (HDAC) inhibitor, sodium butyrate, blocks H3K9 methylation induced by fear conditioning. Histone methylation can alter DNA methylation, which itself can alter gene expression. For example, work from Huda Zoghbi’s group at Baylor College of Medicine, Houston, Texas, showed that CREB can recruit the methyl-DNA binding protein MeCP2 to promoters to activate transcription (see Chahrour et al., 2008). Originally, the Rett syndrome gene MeCP2 was fingered as a transcriptional repressor. Interestingly, Gupta and colleagues present evidence that MeCP2 binding is also increased on the Zif268 promoter in response to histone methylation during contextual fear learning.
Histone acetylation is a hot area of memory research, given that HDAC inhibitors enhance learning and memory (see ARF related news story). There is also some evidence that histone methylation is changed in some cases of schizophrenia (see Akbarian and Huang, 2009). Whether histone methylation turns out to be relevant to memory disorders such as Alzheimer’s remains to be seen.—Tom Fagan.
Gupta W, Kim SY, Artis S, Molfese DL, Schumacher A, Sweatt JD, Paylor RE, Lubin FD. Histone methylation regulates memory formation. Journal of Neuroscience 2010, March 10; 30:3589-3599. Abstract