1 d4 d5 2 c4 c6 3 Nc3 e5

A sometimes-expert at the local chess club consistently plays the Slav Defense.  The lone master at the club has shown us why one does not play 1 d4 d5 2 c4 c6 3 Nc3:  3...e5.  But Aronian just played 3 Nc3 against Vallejo Pons, so I assume there's a refutation of 3...e5.  Does anyone know it?

There is no problem with 3.Nc3 for white in the Slav Defense.  It may lead to sharper play than usual if black accepts the (temporary) pawn sacrifice (3...dxc4). Super GMs play that move (3.Nc3) fairly regularly, too -- it's not just Aronian and not just yesterday.  Often, the line just transposes to the more typical lines after 3.Nf3.

What is the big deal with 3...e5, according to your local master?  That black is allowed to play it, compared to the lines after 3.Nf3 -- okay.  But what is black's gain, according to this person?

 1 d4 d5 2 c4 c6 3 Nc3 e5 4 dxe5 d4, and Black gets a powerful attack.  I tried 4 e4 and came out OK only because Black didn't play ...Qa5 at the appropriate point.  I'll have to get out the game and a chess set to analyze it further, but I know he's shown us how the attack can develop.  It isn't in my copy of MCO-13, but then, that book is somewhat old. 

I noticed in the opening database that the most popular and promising move for white is not 4.dxe5?! (which would be suspicious anyway: taking a sacrificed pawn in a line that you're unfamiliar with), but 4.e3.  Look it up in Game Explorer and see some sample games for white; there was one by Karpov even.