Scan out from Black’s king in all directions and you find not one but two pieces lined up with it: the knight on e5 and the rook on b5. Both should be considered possible targets for pins. White can pin the knight along the file with a vertical attacker—his rook: Re1. But the knight has protection from the rook on b5, and by the time White can get another attacker to the scene (i.e., his f-pawn) Black will be able to move his king and release the knight from its paralysis.
So now compare White’s other possible pin: Ba4. This time it works. The rook has no guard, so it is immediately vulnerable; if Black moves his king to release it from the pin, the rook gets taken anyway with BxR.
Just for fun, think back to the original idea of Re1. It would work if Black’s rook were on b8—though not if Black’s rook were on b6, for then Black would reply to Re1 with Re6. In that case the knight on e5 still would be subject to a relative pin (if it moved, the rook behind it would get taken); but if White tried to take advantage with f2-f4, attacking the pinned knight, Black would have the discovered attack Nf3+. No matter what White did, he at least would lose his rook for a knight.