Organismal development requires the precise coordination of genetic programs to regulate cell fate and function. MEF2 transcription factors (TFs) play essential roles in this process but how these broadly expressed factors contribute to the generation of specific cell types during development is poorly understood. Here we show that despite being expressed in virtually all mammalian tissues, in the retina MEF2D binds to retina-specific enhancers and controls photoreceptor cell development. MEF2D achieves specificity by cooperating with a retina-specific factor CRX, which recruits MEF2D away from canonical MEF2 binding sites and redirects it to retina-specific enhancers that lack the consensus MEF2-binding sequence. Once bound to retina-specific enhancers, MEF2D and CRX co-activate the expression of photoreceptor-specific genes that are critical for retinal function. These findings demonstrate that broadly expressed TFs acquire specific functions through competitive recruitment to enhancers by tissue-specific TFs and through selective activation of these enhancers to regulate tissue-specific genes.

Despite significant heritability of autism spectrum disorders (ASDs), their extreme genetic heterogeneity has proven challenging for gene discovery. Studies of primarily simplex families have implicated de novo copy number changes and point mutations, but are not optimally designed to identify inherited risk alleles. We apply whole-exome sequencing (WES) to ASD families enriched for inherited causes due to consanguinity and find familial ASD associated with biallelic mutations in disease genes (AMT, PEX7, SYNE1, VPS13B, PAH, and POMGNT1). At least some of these genes show biallelic mutations in nonconsanguineous families as well. These mutations are often only partially disabling or present atypically, with patients lacking diagnostic features of the Mendelian disorders with which these genes are classically associated. Our study shows the utility of WES for identifying specific genetic conditions not clinically suspected and the importance of partial loss of gene function in ASDs.

To thrive in complex environments, animals and artificial agents must learn to act adaptively to maximize fitness and rewards. Such adaptive behavior can be learned through reinforcement learning1, a class of algorithms that has been successful at training artificial agents2-6 and at characterizing the firing of dopamine neurons in the midbrain7-9. In classical reinforcement learning, agents discount future rewards exponentially according to a single time scale, controlled by the discount factor. Here, we explore the presence of multiple timescales in biological reinforcement learning. We first show that reinforcement agents learning at a multitude of timescales possess distinct computational benefits. Next, we report that dopamine neurons in mice performing two behavioral tasks encode reward prediction error with a diversity of discount time constants. Our model explains the heterogeneity of temporal discounting in both cue-evoked transient responses and slower timescale fluctuations known as dopamine ramps. Crucially, the measured discount factor of individual neurons is correlated across the two tasks suggesting that it is a cell-specific property. Together, our results provide a new paradigm to understand functional heterogeneity in dopamine neurons, a mechanistic basis for the empirical observation that humans and animals use non-exponential discounts in many situations10-14, and open new avenues for the design of more efficient reinforcement learning algorithms.

Charged multivesicular body protein 1A (CHMP1A; also known as chromatin-modifying protein 1A) is a member of the ESCRT-III (endosomal sorting complex required for transport-III) complex but is also suggested to localize to the nuclear matrix and regulate chromatin structure. Here, we show that loss-of-function mutations in human CHMP1A cause reduced cerebellar size (pontocerebellar hypoplasia) and reduced cerebral cortical size (microcephaly). CHMP1A-mutant cells show impaired proliferation, with increased expression of INK4A, a negative regulator of stem cell proliferation. Chromatin immunoprecipitation suggests loss of the normal INK4A repression by BMI in these cells. Morpholino-based knockdown of zebrafish chmp1a resulted in brain defects resembling those seen after bmi1a and bmi1b knockdown, which were partially rescued by INK4A ortholog knockdown, further supporting links between CHMP1A and BMI1-mediated regulation of INK4A. Our results suggest that CHMP1A serves as a critical link between cytoplasmic signals and BMI1-mediated chromatin modifications that regulate proliferation of central nervous system progenitor cells.

Rapid phasic activity of midbrain dopamine neurons is thought to signal reward prediction errors (RPEs), resembling temporal difference errors used in machine learning. However, recent studies describing slowly increasing dopamine signals have instead proposed that they represent state values and arise independent from somatic spiking activity. Here we developed experimental paradigms using virtual reality that disambiguate RPEs from values. We examined dopamine circuit activity at various stages, including somatic spiking, calcium signals at somata and axons, and striatal dopamine concentrations. Our results demonstrate that ramping dopamine signals are consistent with RPEs rather than value, and this ramping is observed at all stages examined. Ramping dopamine signals can be driven by a dynamic stimulus that indicates a gradual approach to a reward. We provide a unified computational understanding of rapid phasic and slowly ramping dopamine signals: dopamine neurons perform a derivative-like computation over values on a moment-by-moment basis.

EphB receptor tyrosine kinases control multiple steps in nervous system development. However, it remains unclear whether EphBs regulate these different developmental processes directly or indirectly. In addition, given that EphBs signal through multiple mechanisms, it has been challenging to define which signaling functions of EphBs regulate particular developmental events. To address these issues, we engineered triple knock-in mice in which the kinase activity of three neuronally expressed EphBs can be rapidly, reversibly and specifically blocked. We found that the tyrosine kinase activity of EphBs was required for axon guidance in vivo. In contrast, EphB-mediated synaptogenesis occurred normally when the kinase activity of EphBs was inhibited, suggesting that EphBs mediate synapse development by an EphB tyrosine kinase-independent mechanism. Taken together, our data indicate that EphBs control axon guidance and synaptogenesis by distinct mechanisms and provide a new mouse model for dissecting EphB function in development and disease.