Models with dark energy decaying into dark matter have been proposed in Cosmology to solve the coincidence problem. We study the effect of such coupling on the cosmic microwave background temperature anisotropies. The interaction changes the rate of evolution of the metric potentials, the growth rate of matter density perturbations and modifies the Integrated Sachs-Wolfe component of cosmic microwave background temperature anisotropies, enhancing the effect. Cross-correlation of galaxy catalogs with CMB maps provides a model independent test to constrain the interaction. We particularize our analysis for a specific interacting model and show that galaxy catalogs with median redshifts $z_m=0.1-0.9$ can rule out models with an interaction parameter strength of $c^2\simeq 0.1$ better than 99.95% confidence level. Values of $c^2\le 0.01$ are compatible with the data and may account for the possible discrepancy between the fraction of dark energy derived from WMAP 3yr data and the fraction obtained from the ISW effect. Measuring the fraction of dark energy by these two methods could provide evidence of an interaction.