A gradient of positional information in an insect, Rhodnius.

Locke discovered a segmental gradient in Rhodnius which controls the polarity of the epidermal cell and gives positional information. The polarity is expressed by the orientation of folds in the adult epicuticle, which are aligned parallel to the contours in the gradient. It was later suggested that this gradient could be of a concentration of a diffusible substance. Because concentration gradients could be maintained in various ways we have simulated several models in the computer, and examined the results of rotating square pieces of model landscape through 90° and allowing diffusion. The gradient landscapes after different times and at equilibrium are plotted as contour maps and are compared with cuticle patterns from adult insects after rotation of square pieces of epidermis in larvae. One simple model, where the gradient depends only on the activities of a line of source cells at one end of the segment and a line of sink cells at the other, is eliminated by 2 observations: (1) the theoretical and experimental patterns are consistently different; and (2) when adults developing from operated larvae are made to form a supernumerary cuticle the first and second cuticles have almost identical patterns. This suggests that the gradient landscape has reached a steady state. In another model the cells are considered to act as homeostatic units in the gradient, and when moved to a new position they each attempt to maintain their original or ‘set’ concentration. Simulation of this model gives equilibrium patterns which are similar to the experimental results. It is suggested that the cells become ‘set’ at some stage in the cell cycle to the ambient concentration. This hypothesis predicts that after reaching initial equilibrium the pattern should change only if there are cell divisions. Adult insects are made to moult again under different conditions and it is found that pattern change is correlated with cell divisions. Locke also observed an asymmetry in the patterns after rotation of squares through 180°. Simulation showed that such asymmetry would result from each cell acting as a better homeostatic unit when moved one way in the gradient (for example when acting as a sink) than when moved the other (acting as a source). We do not claim that these comparisons eliminate all other classes of model, and present our conclusions in as general a form as possible.