The best way to do this is in SEM which allows the role of moderating variables to be specifically modeled. But that is much more complex than regression and SPSS won't do it. In regression you want to have ideally a theory of which variables will moderate (perhaps by consulting the literature on your topic). Alternately you can look for the impact on the variables influenced by the moderating variables when you remove the moderators. There are signs to look for, such as dramatic changes (for instance a variable's sign reversing when the moderator is removed).
It is, in OLS, as much an artform as a science.