This Week in The Journal

The Acetyltransferase GCN5 Limits Diencephalon Size

Jonathan J. Wilde, Julie A. Siegenthaler, Sharon Y.R. Dent, and Lee A. Niswander

Early in development, gradients of morphogens activate different transcriptional programs along the neural plate, dividing primordial neural tissue into domains that become the telencephalon, diencephalon, midbrain, and hindbrain. Where and when morphogens and their downstream effectors are expressed must be precisely controlled to ensure that each domain forms and grows properly. Excessive or insufficient growth can prevent neural tube closure and cause neurological deficits.

Loss of GCN5 acetyltransferase activity (right) causes mispatterning and expansion of the diencephalon, as indicated by expression of two diencephalic markers (green and magenta). See Wilde et al. for details.

The specification and growth of the diencephalon depends on localized expression of the morphogen sonic hedgehog (Shh): broader-than-normal expression of Shh leads to excessive production of diencephalic tissue. Wilde et al. have identified a molecular pathway that limits Shh expression to prevent such diencephalic expansion. They first found that the diencephalon was selectively expanded in mice in which the acetyltransferase GCN5 lacked enzymatic activity. In addition, Shh expression was expanded, and signaling by WNT, which promotes Shh expression, was increased in mutant mice. Because retinoic acid represses WNT signaling, and because GCN5 associates with retinoic acid receptor α (RARα), the authors asked whether GCN5 regulates expression of genes controlled by retinoic acid. Indeed, GCN5 mutation decreased expression of such genes.

Previous work indicated that one target of GCN5 acetylation is TACC1, a protein that interacts with RARα and is expressed in the developing forebrain. Wilde et al. found that in the absence of retinoic acid, wild-type GCN5, RARα, and TACC1 formed a complex localized to retinoic acid response elements in chromatin. Addition of retinoic acid caused GCN5-dependent acetylation of TACC1, which led to dissociation of TACC1 from the chromatin-bound complex.

These results suggest that growth of diencephalic tissue is restricted by retinoic acid, which binds to RARα at retinoic acid response elements in DNA. This binding triggers acetylation of TACC1 by GCN5, which causes TACC1 to dissociate from DNA. This in turn allows transcription of the associated genes. These genes likely repress WNT signaling and Shh expression, thus defining the limits of diencephalic specification. Future work will need to determine how retinoic acid production is regulated to prevent excessive restriction of diencephalon growth.

Orbitofrontal Cortex Encodes Updated Reward Identity

James D. Howard and Thorsten Kahnt

(see pages 2627–2638)

Many factors influence behavioral choices. What you order from a menu, for example, depends not only on what foods you like, but also on how recently you have eaten those foods and how hungry you are. All these factors must be incorporated into a neural representation of the expected value of each option before a choice is made. Evidence suggests that such expected values are encoded by activity in the orbitofrontal cortex. How values are updated to reflect current states, and how such updating influences decisions, is poorly understood, however. Therefore, Howard and Kahnt used functional magnetic resonance imaging to investigate changes in neural representations of food-related odors before and after one odor was devalued.

Hungry participants first rated the appeal of several odors, and two appetizing odors were selected for use in a subsequent choice task, which was performed in a scanner. In each trial of the task, subjects chose whether to sniff a high or low concentration of one of the appealing odors. They then consumed a meal consisting of food associated with one of these odors, thus devaluing that odor. Finally, subjects performed the choice task again. Pattern-based analysis was then used to compare neural representations of expected rewards before and after the meal.

Before the meal, subjects chose to sniff the higher concentration of both odors; after the meal, subjects chose the lower concentration of the devalued odor, but continued to choose the high concentration of the non-devalued odor. Pattern analysis indicated that reward identity was encoded in medial and lateral aspects of the posterior orbitofrontal cortex (pOFC). While representations of both rewards changed in the medial pOFC after the meal, only the representation of the devalued reward changed in the lateral pOFC. Additional analyses indicated that functional coupling between the lateral pOFC and ventromedial prefrontal cortex (vmPFC)—an area that encodes identity-independent reward value—decreased selectively for the devalued reward after the meal. The larger this decrease, the more subjects' preference for the high concentration of the devalued odor decreased.

These results support the hypothesis that the lateral pOFC represents the predicted values of specific rewards, updates these values as conditions change, and communicates these values to vmPFC to guide decisions. How outcome representations change at the neural ensemble level remains unknown and should be investigated in future studies.