i would suggest to create a node for each venue, and then to create properties latitude and longitude for this type of node. then, you must also index these properties.
to solve the problem with venues around a given point, well, maybe creating some country supernodes could help - e.g. europe, middle europe, czech republic, Prague. but this requeires the db of such country names. i would rather create virtual country parts:
imagine the world globe as a 2d map. than make a grid on it to get a matrix of NxM squares - the best would be divide it by using some global pattern like each +1 of latitude and longitude is a virtual country part.
after you got this virtual country parts than simply connect all your venue nodes to its supernode which have properties lat's+long's within this virtual country part.
than, when searching for all nodes within you radius, you must find first the country part( or more parts when crossbordered) and then manualy check all nodes within this part(s) whether or not they fit into your radius.
at least you don't need to check all nodes but only a subset. i'm not aware of such an index solution which could handle the radius problem, only manual check is the only way.
do you need an example using neo4jphp? if so, i'll find some time to make one.
EDIT: i found that some solution already exist like https://github.com/neo4j/spatial so maybe my answer isn't appropriate and you should wait for an domain expert answer.