I have been baffled with this question approx. for a dozen of years for now. There is a default "lock & key" mechanism of action assumed for the interaction between a signalling molecule and its receptor, but while applying a bit of combinatorics and biophysics to it I find the following inconsistencies.
Suppose a long-lasting neuropeptide is circulating in a CSF or blood macrocirculation. Or suppose a norepinephrine or other neurohormones are released by hypothalamic and pituitary cells into the macrocirculation for sympathetic or parasympathetic signalling. Accroding to the logic of "lock & key" mechanism, in order for a "key" signalling molecule to activate its "lock" receptor an active site of this molecule - the molecule partition which directly reacts with the receptor as it is generally not the whole molecule which reacts with the receptor - should come in direct contact with the relevant active site of its in-membrane receptor. So by such combinatorial logic, if there are thousands of receptors on outer membranes of cells in CSF or vessel endotelium along the way of which this "key"-molecule is travelling, this molecule should try to "lock into" every such potential "lock" receptor in order to "check" if this is the right "lock". In other words, while travelling across long distances such "key" signalling molecules should not just float in CSF or blood vessels as is generally depicted, but slide across all outer cell membranes in CSF or across vascular endotelium in blood or lymph circulation in order to "check" every potential "lock" receptor for finding their relevant receptors for eventual activation. It is definitely not the case, but it the result of applying combinatorial "lock&key" logic.
Somehow it seems (for myself, at least) that it is just assumed that a "key" molecule released at distances which are millions> of times larger than the molecule size should just "lock" for granted into its appropiate receptors throughout the organism. But suppose a "key" molecule travels in CSF or interstitial fluid in the vicinity of its appropriate membrane "lock" receptor, but with the molecule's reaction (active) site pointed away from its receptor or at an angle. Should the "key" molecule not then react with its "lock" receptor?? Is there a minimum effective reaction distance established between a "key" molecule and its "lock" receptor in order to effectively activate the receptor?
The latter argument holds true also for small distance 20-40 nm synapses with molecules released in it whose size would be minimum 1000 times smaller than the synapse size. Should a neurotransmitter molecule not react with its receptor in a synapse if the molecule's active site would be pointed 180 degrees away from the receptor's active site?