It's both heat and voltage that do damage, but both combined make it worse. Voltage itself, regardless of temperature, will eventually result in electromigration failure (you can Google/Wikipedia for the technical description which eludes me, I just know it means the end of the line). Again, this is a matter of when, not if, and more voltage speeds the process up. This is regardless of temperatures, however, more heat adds to the effect speeding it up even more. Heat then also has it's own dangers as, well, temperatures. In the end, it's both, and the combination of them, at least that's what I've gathered from seeing many discussions on the two.
As far as I know, 1.4V was the most common rule to go by with the 45nm Core 2s. I think some others seem to say they've run their 45nm Core 2s a bit over 1.4V 24/7 for a year or two fine, but I cannot speak for that. As always, the limits aren't really set in stone (other than Intel's claimed limit which is 1.3625V). I'd personally try and stay below 1.4V myself, along with keeping temperatures down, but if you're getting temperatures that good at 1.45V, then who knows.
My FSB is 400MHz, with the multiplier at 10 (default). I could probably raise the FSB and drop the multiplier, but I like keeping the RAM/Northbridge voltages/temperatures down.