Corticothalamic Axons Guide Retinal Axons into dLGN
James A. Shanks, Shinya Ito, Laura Schaevitz, Jena Yamada, Bin Chen, et al.
(see pages 5252–5263)
Retinal ganglion cell axons face many choices as they grow toward their targets. First they must grow toward the optic fissure and leave the retina. At the optic chiasm, they must choose to cross the midline or continue growing ipsilaterally. Ultimately, they must decide which target to innervate: the dorsal lateral geniculate nucleus (dLGN), superior colliculus, or one of many nuclei not involved in image formation. Axons make these choices by following attractive and repulsive cues that are secreted by, or expressed on, the surface of cells located near choice points.
Shanks et al. have discovered an unexpected source of guidance cues for retinal axons entering the dLGN: corticothalamic axons. Projections from cortical layer 6 reach the thalamus before retinal axons, but they appear unable to enter the dLGN until retinal afferents have completed their innervation (Seabrook, et al. 2013 33:10085). Visual corticothalamic axons (in fact, most of the cortex) are absent in transgenic mice lacking the RNA-binding protein Tra2ß, and Shanks et al. found that most retinal axons failed to enter the dLGN in these mice. Ganglion cell axons were absent in the dorsomedial portion of the dLGN and at least one class of retinal axons failed to enter the nucleus altogether. Similar defects were present in mice lacking only cortical layer 6.
Despite the pronounced deficit in dLGN innervation in Tra2ß-deficient mice, retinal innervation of the superior colliculus and olivary pretectal nucleus, which receive input from cortical layer 5, appeared normal. Moreover, visual responses of superior colliculus neurons were normal, and mice had normal pupillary light responses and circadian activity patterns. More surprisingly, Tra2ß-deficient mice showed normal fear conditioning to a flashing light stimulus, and they could learn to use a visual cue to find an escape platform in the Morris water maze.
These results suggest that corticothalamic and retinal ganglion cell axons promote each other's growth into the dLGN. Corticothalamic axons waiting outside the dLGN may secrete a factor that promotes ingrowth of retinal axons, which—previous studies suggest—make the dLGN hospitable for corticothalamic innervation. Identification of the guidance cue provided by corticothalamic axons will deepen our understanding of the complex mechanisms guiding retinal axon growth.
Superior Temporal Sulcus Encodes Dangerousness
Andrew C. Connolly, Long Sha, J. Swaroop Guntupalli, Nikolaas Oosterhof, Yaroslav O. Halchenko, et al.
(see pages 5373–5384)
Categorization of objects and concepts is a central element of human cognition. We may classify a given object according to numerous properties. For example, we can classify animals based on size and shape, fitness for consumption, or dangerousness. Understanding where these different categories are represented in the brain and how these different representations are related to each other anatomically is essential for elucidating the neural bases of cognition.
Connolly et al. approached this question by examining human brain responses to pictures of different animals using functional magnetic resonance imaging. They then used pattern classifiers and clustering analyses to identify anatomical regions that were differentially activated by different taxonomic groups (mammal, reptile, or bug) or levels of dangerousness (low or high). Previous work had indicated that taxonomy was represented in the lateral occipital complex, an area involved in object recognition. Connolly et al. expected that dangerousness would also be represented in this area, but with a different representational structure in multidimensional vector space. They found no evidence for this hypothesis, however. Instead, they found that dangerousness of stimuli could be decoded from activity patterns in four large functional clusters: in the early visual cortex and in the right anterior, middle, and posterior superior temporal sulcus (STS).
To examine the relative strength of encoding of dangerousness versus basic visual features and taxonomy in these clusters, the authors asked how well the relationships among neural activity patterns elicited by different stimuli matched classification models based on people's perceptual judgments of threat or taxonomy or a computational model of early visual processing. This representational similarity analysis suggested that activity in posterior STS was influenced by taxonomy and visual features, but that these influences were significantly weaker in anterior STS.
These results suggest that neural activity in the anterior STS plays a role in threat classification. Previous work has suggested that the STS is involved in perception of faces, gaze direction, and speech, leading to the hypothesis that this area plays a role in discerning other people's intentions. These new results support this hypothesis and suggest that the anterior STS is also involved in discerning animals' intentions, specifically if an animal is likely to pose a threat.
Footnotes
This Week in The Journal is written by Teresa Esch, Ph.D.