Fig 2: (A–D) The areas occupied by the dendritic arbors were also significantly reduced (A), again drawn from samples of cells distributed across similar retinal eccentricities (B) from retinas of comparable areas (C). Soma sizes were also determined, revealing a slight but non-significant reduction in somal area (D). (E,F) Fields containing DiI-labeled dendritic arbors (red) were also labeled for PNA-positive cone pedicles (blue), in order to quantify the number of pedicles contacted (yellow circles) in Sox5-CKO (F) and littermate control (E) retinas. Calibration bar = 10 µm. (G) The number of cone pedicles contacted by those dendritic arbors was significantly reduced in the Sox5-CKO retina. (H) Average cone densities were unchanged in the Sox5-CKO retina. (I) The number of cone pedicles contacted increased with the areal size of the dendritic arbor, in both groups. Means and standard errors are shown within the violin plots in (A,D,G), while means and individual retinas are indicated in (C,H). n = the number of cells (A,D,G) or retinas (C,H) sampled.

Fig 3: (A–D) Seven examples of DiI-labeled axon terminals (A,B) and dendritic arbors (C,D) from Sox5-CKO (B,D) and littermate control (A,C) retinas, in Z-stack projections from retinal wholemounts. Note the smaller areal domains of both axonal and dendritic territories in the Sox5-CKO retinas, enclosed here by convex polygons (red). Calibration bar = 10 µm.

Fig 4: (A–D) The number of DiI-labeled dendritic endings (white dots) at individual PNA-labeled cone pedicles (blue) (A,B) showed considerable variability across the dendritic fields (C,D), in both littermate control (A,C) and in Sox5-CKO retinas (B,D). Calibration bar = 1 µm in panels (A,B), and = 10 µm in panels (C,D). (E,F) The number of terminals per pedicle declined gradually with distance from the labeled Type 7 soma, in both conditions. (G,H) As expected, the total number of dendritic terminals contacting all of the pedicles by a cell was significantly reduced in the Sox5-CKO retina (G), but the average number at each pedicle per cell was significantly increased (H). The average number of dendritic terminals at each pedicle per cell is considerably more variable in the Sox5-CKO retinas, due primarily to the presence of hyper-innervated pedicles (H). (I) Pooling the entire population of sampled pedicles, the frequency distribution of the number of pedicles with a given number of terminals is shifted to the right in the Sox5-CKO retinas.

Fig 5: Summary diagram depicting the reduced size of axonal and dendritic arbors in the Sox5-CKO retina, indicative of a role for Sox5 in promoting process outgrowth in the plane of the IPL and OPL, respectively. Sox5 also contributes specificity to the targeting of those axonal arbors within S4 of the IPL, made apparent by the ectopic extension into S5 in its absence. The hyper-innervation of pedicles by dendritic branches is interpreted to arise secondarily, from the lack of shared pedicle contacts with neighboring Type 7 dendritic arbors, inferred from the reduction in dendritic field area by a population that normally tiles the retina.