Yes, absolutely. This is how I'm running two different setups. A 45609 in both locations one is wired to line, neutral, and load (lights), the other just to line and neutral. Tied together with the 3-way program and they work beautifully. Usually less than a 1/2 second of delay when I hit the "fake" switch before the main switch responds.
Now, having said this, there is one potential major pit fall: distance from hub. I mentioned this briefly in an earlier post, but to elaborate a little: GE (as well as Intermatic and Evolve) switches do not use the patent that Lutron has and therefore can NOT report to the hub that the switch has been pressed.
SmartThings uses a little trick to get around this partially. Basically the switch just sends a generic node command and SmartThings figures this may be because the switch was switched, so does an immediate poll of the switch, determines it's current state and updates it as on or off. Works great for many operations, but it fails in three ways:
First, when a dimmer is turned off this does work. (The poll command is too fast and reports as still on as the dim fads to off).
Second, when an aux switch in a wired three-way setup is press, it never sends the node command
Third, if the switch does not have direct communication to the hub, the command isn't received. The mesh network doesn't pass on this node command to the hub.
So... long story short: If you're using a dimmer on either side of the wireless three-way things will operate a little weird and won't always work the way you expect them too.
If either of the switches is too far from the hub that it can't communicate directly with it, then things will operate a little weird and won't always work the way you expect them too.