It depends. If A activates the complex XYZ by binding to it, you have something like XYZ + A -> XYZ_A. If the function of A is catalytic, you might need something like XYZ + A -> XYZ_A -> XYZ*_A -> XYZ* + A (where the star denotes the active form). Depending on the speed of the intermediary steps and your level of detail, you might be able to leave out the steps in between.

Also, you might want to add the formation of the complex to your model, something like X + Y -> XY and XY + Z -> XYZ (or similar, dependent on how the complex forms)

I'm not sure about Reactome but BioPax is just a kind of XML, so you can represent it as a table for further processing or (if you are familiar with programming) parse it to directly exctract the information you need.

Thanks for your answers. In extension of my question I want to know how I can write the simple relationship A + B = AB ; C ----> AB ?

Can I write C ---> A ; C ----> B and ignore the complex formation, although there is a restriction that A and B work together in a complex form.

Again, it depends. Does C really activate both A and B (provided they are in the complex)? Or does it activate the complex as such? In the latter case, nothing wrong with C----> AB (if you make clear that you mean activation by the arrow), and you don't have to specify how the complex AB is formed. Depending your modelling environment, you could also use a conditional statement such as

C---> A IF A is bound to B

There are several ways of representing any given biochemical system - it really depends on what you want to do with your model.