Oh, that's too cute! +1 :) :)
In general, I'd suggest using another transcendental constant, one that is less likely to show up in the solution of a differential equation so that the back substitution E -> α doesn't clobber an exponential function. (Examples: EulerGamma, Catalan, Zeta[7],..., but not Pi obviously though it, too, will work here.)
