The velocity-addition formulae should indeed give the correct answer. Here's what happens if you work it out using vector Lorentz transformations.
We know that the four-velocity of the electron is $S'$ frame is $\tilde u = W(ce_t' + ue_x')$, where $W=(1-[u/c]^2)^{-1/2}$. Some notes on notation: $\tilde u$ denotes the four-vector while $u$ is just the three-velocity. I'm calling this Lorentz factor $W$ to avoid confusion with the Lorentz factor of the boost (which will just be $\gamma$). So once again, $u$ goes with $W$ the way $v$ goes with $\gamma$.
Now then, the boost between the $S,S'$ frames tells us that $e_t' = \gamma( e_t + v e_x/c)$ and $e_x' = \gamma(e_x + v e_t/c)$. We can be very mechanical here and just plug these expressions back into our original four-velocity. The result is
$$\tilde u = W(c[\gamma(e_t + ve_x/c)] + u[\gamma(e_x + ve_t/c)])=W\gamma([c+uv/c]e_t+\gamma[u+v]e_x)$$
Take $\tilde p = m_e \tilde u$, and you're done. The components fall out as
$$E/c = m_e W\gamma(c + uv/c), \quad p = m_e W\gamma(u + v)$$
In the limiting case that the boost velocity $v=0$, then $\gamma=1$ and we get $E/c=m_e W c \implies E = m_e W c^2$, which checks, and $p = m_e W u$, which also checks. Similarly, if $u=0$, then $W=1$, the correct limits hold, too. It's always a good idea to check limits; doing so helped me find a mistake in this calculation which I've since corrected!