Prevents the immediate e5 break.

If 3...c5, then you might try d5. Li - Wang 1-0 continued: 4.d5 e6 5.Nc3 exd5 6.Qxd5 Qxd5 7.Nxd5 Bd6 8.Nd2 Nc6 9.Nxc4 Bb8 10.g3 Nge7 11.Nxe7 Kxe7 12.Bg2 Nd4 13.Ne3 Re8 14.b3 Be5 15.Rb1 Rb8 16.f4 Bd6 17.Kf2 Kf8 18.Bb2 b6 19.Rhd1 Ba6 20.Rd2 Re7 21.Nd5 Rxe2+ 22.Rxe2 Nxe2 23.Bf1 Bb7 24.Bxe2 Bxd5 25.Rd1 Bxb3 26.axb3 Rd8 27.Be5 Ke7 28.Bxg7 Rg8 29.Bc3 a5 30.Bc4 Rc8 31.f5 1-0

Qc2 is another try. Also e5. Bg5 is fine though.

?! There are some beautiful continuations after Bxd5. Na4 also takes care of that monstrous bishop.

?! Stylish but inaccurate. It does two things, but that weak square is going to be nasty. Better was to prepare b5 by opposing the rook and queen.

? An anti-positional move. It's hard to say what this does. It's a centralized knight, but it's better to live with it than reduce to a hopeless game.

?! Qd2 would break the pin.

? Loses to the immediate g5. Better to lose a pawn on d4 than the game.

!? Probably the best practical chance? The plan is to give up the d4 pawn for activity or head to an inferior rook endgame where you can hope for an error.

? This appears to be a miscalculation to the sixth rank. h4, which you did consider, prevents the king from moving. Black is kind of at a loss for how to improve.