Environmental changes can elicit alterations in the form, behavior and/or physiology of all species, and this developmental response to environment is known as phenotypic plasticity. Despite its ubiquity, the molecular basis for phenotypic plasticity is not fully understood. The pea aphid, Acyrthosiphon pisum, serves as a model for an extreme form of phenotypic plasticity, known as polyphenism. Changes in photoperiod stimulate a switch in female aphid reproductive mode from asexual to sexual reproduction over the course of one generation without changes in genotype. This reproductive polyphenism results in female aphids with ovaries of one of two types: sexual ovaries (producing haploid oocytes via meiosis), or asexual ovaries (producing identical diploid aphid clones via parthenogenesis). To better understand how aphid ovaries could produce different outputs, we surveyed the transcriptomes of sexual and asexual ovaries using RNA-seq. Among genes that exhibited greater than two-fold differences in gene expression between sexual and asexual ovaries, we identified several aubergine paralogs, which encode for germline-specific members of the Argonaute small RNA-binding protein family. The A. pisum genome contains eight aubergine paralogs and at least two piwi paralogs. We are currently comparing the expression patterns of these aphid aubergine paralogs between asexual and sexual aphid ovaries. Aubergine proteins in other species are thought to help suppress the activity of transposable elements, which are found in high quantities throughout the A. pisum genome. Together, these experiments will help elucidate a potential relationship between aubergine paralogs and aphid reproductive plasticity.

The signal and resolution during in vivo imaging of the mouse brain is limited by sample-induced optical aberrations. We find that, although the optical aberrations can vary across the sample and increase in magnitude with depth, they remain stable for hours. As a result, two-photon adaptive optics can recover diffraction-limited performance to depths of 450 μm and improve imaging quality over fields of view of hundreds of microns. Adaptive optical correction yielded fivefold signal enhancement for small neuronal structures and a threefold increase in axial resolution. The corrections allowed us to detect smaller neuronal structures at greater contrast and also improve the signal-to-noise ratio during functional Ca(2+) imaging in single neurons.

Early stages of visual processing are thought to decorrelate, or whiten, the incoming temporally varying signals. Because the typical correlation time of natural stimuli, as well as the extent of temporal receptive fields of lateral geniculate nucleus (LGN) neurons, is much greater than neuronal time constants, such decorrelation must be done in stages combining contributions of multiple neurons. We propose to model temporal decorrelation in the visual pathway with the lattice filter, a signal processing device for stage-wise decorrelation of temporal signals. The stage-wise architecture of the lattice filter maps naturally onto the visual pathway (photoreceptors -> bipolar cells -> retinal ganglion cells -> LGN) and its filter weights can be learned using Hebbian rules in a stage-wise sequential manner. Moreover, predictions of neural activity from the lattice filter model are consistent with physiological measurements in LGN neurons and fruit fly second-order visual neurons. Therefore, the lattice filter model is a useful abstraction that may help unravel visual system function.

Early stages of sensory systems face the challenge of compressing information from numerous receptors onto a much smaller number of projection neurons, a so called communication bottleneck. To make more efficient use of limited bandwidth, compression may be achieved using predictive coding, whereby predictable, or redundant, components of the stimulus are removed. In the case of the retina, Srinivasan et al. (1982) suggested that feedforward inhibitory connections subtracting a linear prediction generated from nearby receptors implement such compression, resulting in biphasic center-surround receptive fields. However, feedback inhibitory circuits are common in early sensory circuits and furthermore their dynamics may be nonlinear. Can such circuits implement predictive coding as well? Here, solving the transient dynamics of nonlinear reciprocal feedback circuits through analogy to a signal-processing algorithm called linearized Bregman iteration we show that nonlinear predictive coding can be implemented in an inhibitory feedback circuit. In response to a step stimulus, interneuron activity in time constructs progressively less sparse but more accurate representations of the stimulus, a temporally evolving prediction. This analysis provides a powerful theoretical framework to interpret and understand the dynamics of early sensory processing in a variety of physiological experiments and yields novel predictions regarding the relation between activity and stimulus statistics.

A decline in ocular lens transparency known as cataract afflicts 90% of individuals by the age 70. Chronic deterioration of lens tissue occurs as a pathophysiological consequence of defective water and nutrient circulation through channel and transporter proteins. A key component is the aquaporin-0 (AQP0) water channel whose permeability is tightly regulated in healthy lenses. Using a variety of cellular and biochemical approaches we have discovered that products of the A-kinase anchoring protein 2 gene (AKAP2/AKAP-KL) form a stable complex with AQP0 to sequester protein kinase A (PKA) with the channel. This permits PKA phosphorylation of serine 235 within a calmodulin (CaM)-binding domain of AQP0. The additional negative charge introduced by phosphoserine 235 perturbs electrostatic interactions between AQP0 and CaM to favour water influx through the channel. In isolated mouse lenses, displacement of PKA from the AKAP2-AQP0 channel complex promotes cortical cataracts as characterized by severe opacities and cellular damage. Thus, anchored PKA modulation of AQP0 is a homeostatic mechanism that must be physically intact to preserve lens transparency.

Aberrations and random scattering severely limit optical imaging in deep tissue. Adaptive optics can in principle drastically extend the penetration depth and improve the image quality. However, for random scattering media a large number of spatial modes need to be measured and controlled to restore a diffraction limited focus. Here, we present a parallel wavefront optimization method using backscattered light as a feedback. Spatial confinement of the feedback signal is realized with a confocal pinhole and coherence gating. We show in simulations and experiments that this approach enables focusing deep into tissue over up to six mean scattering path lengths. Experimentally the technique was tested on tissue phantoms and fixed brain slices.

Electronic and biological systems both perform complex information processing, but they use very different techniques. Though electronics has the advantage in raw speed, biological systems have the edge in many other areas. They can be produced, and indeed self-reproduce, without expensive and finicky factories. They are tolerant of manufacturing defects, and learn and adapt for better performance. In many cases they can self-repair damage. These advantages suggest that biological systems might be useful in a wide variety of tasks involving information processing. So far, all attempts to use the nervous system of a living organism for information processing have involved selective breeding of existing organisms. This approach, largely independent of the details of internal operation, is used since we do not yet understand how neural systems work, nor exactly how they are constructed. However, as our knowledge increases, the day will come when we can envision useful nervous systems and design them based upon what we want them to do, as opposed to variations on what has been already built. We will then need tools, corresponding to our Electronic Design Automation tools, to help with the design. This paper is concerned with what such tools might look like.

Somatic sexual dimorphisms outside of the nervous system in Drosophila melanogaster are largely controlled by the male- and female-specific Doublesex transcription factors (DSX(M) and DSX(F), respectively). The DSX proteins must act at the right times and places in development to regulate the diverse array of genes that sculpt male and female characteristics across a variety of tissues. To explore how cellular and developmental contexts integrate with doublesex (dsx) gene function, we focused on the sexually dimorphic number of gustatory sense organs (GSOs) in the foreleg. We show that DSX(M) and DSX(F) promote and repress GSO formation, respectively, and that their relative contribution to this dimorphism varies along the proximodistal axis of the foreleg. Our results suggest that the DSX proteins impact specification of the gustatory sensory organ precursors (SOPs). DSX(F) then acts later in the foreleg to regulate gustatory receptor neuron axon guidance. These results suggest that the foreleg provides a unique opportunity for examining the context-dependent functions of DSX.

We present an approach for the automatic reconstruction of neurons from 3D stacks of electron microscopy sections. The core of our system is a set of possible assignments, each of which proposes with some cost a link between neuron regions in consecutive sections. These can model the continuation, branching, and end of neurons. The costs are trainable on positive assignment samples. An optimal and consistent set of assignments is found for the whole volume at once by solving an integer linear program. This set of assignments determines both the segmentation into neuron regions and the correspondence between such regions in neighboring slices. For each picked assignment, a confidence value helps to prioritize decisions to be reviewed by a human expert. We evaluate the performance of our method on an annotated volume of neural tissue and compare to the current state of the art [26]. Our method is superior in accuracy and can be trained using a small number of samples. The observed inference times are linear with about 2 milliseconds per neuron and section.