The early "hormones" might simply have been the actual catalysts of the cellular function they signaled; they might have been both signal and agent, themselves.
The early receptors might have evolved, then, as helpers to those catalysts. As their effecacy improved via evolution, the effecacy of the hormone might have decreased, until the hormone itself had no purpose but to trigger the receptor.
When the function of each part depends on the prior existence of all the other parts, the Darwinian model seems to breaks down.
It doesn't, actually, because evolution works often by co-opting what is already present, not by creation ex nihilo. An analagous example is the construction of a stone arch - remove one piece and the whole thing fails. Yet, arches are constructed one piece at a time. How? Via scaffolds - simpler, less effective structures that are themselves able to be built piece by piece. You build the scaffold, it supports the arch; when the arch is completed the scaffold has no purpose and is removed.