The tension is high anytime the queens are faced off against each other, as they are here. Each side looks at how the opposite queen is guarded, asking whether the guard might be undermined or whether there is a way to exploit its required attendance by the queen's side. In this case Black’s queen is guarded by the knight on e5. White can’t take the knight, but that isn't the point; the point is that the knight can’t afford to leave its square—and thus that White can play his only check (1. Nc6+) with impunity. Black replies Kb7. Other king moves are possible, but this one seems most attractive as it puts some heat on White’s knight. Now ask: why doesn’t Black instead play RxN (with either rook)? The answer is that when White played Nc6 he unmasked the threat of BxN, which would destroy the guard of Black’s queen. If Black tries RxN, White thus has BxN+—and since it’s a check (a crucial point), Black would have to spend time moving his king. White then has QxQ. Replying to 1. Nc6 with Kb7 means that if White plays BxN it won’t give check and so won't be nearly as dangerous.
All right; think of the position resulting from 1. Nc6+, Kb7 as a little adjustment of the board White can force at no cost. What would then be possible? Since your knight is in the picture you should be thinking about forks and see 2. Ne7, attacking both Black rooks. This sounds good, but now wait. 2. Ne7 is not a check, so it gives Black time to make any move he likes. The danger is not that he will play QxQ, for then White replies Nc1xQ and the fork still hangs over Black’s head. No, the problem is that Black will reply to Ne7 with RxNc1+, a crushing riposte; for now Black is the one who has destroyed the only guard of White’s queen and has done it with check. After White plays RxR, Black has QxQ. So 2. Ne7 doesn’t look so good for White after all. Yet White can rehabilitate the idea by pausing to play QxQ on his second move—then playing 3. Ne7. Now if Black plays RxNc1 it doesn’t matter. White recaptures and wins the exchange.
Black still has an idea. Remember that after 1. Nc6+, Kb7; 2. QxQ, NxQ; 3. Ne7 (the fork), Black now has a knight on g4. He, too, thus has a knight fork to offer of White’s two rooks: Nf2. Thus it can go 3. Ne7, Nf2—but then White plays 4. NxRg6, and after 4. …NxR, 5. RxN, White has won material. (Black's rook was loose; White's rook wasn't.) Okay, but now suppose it goes 3. Ne7, Re6; 4. NxRc8, Nf2. This way Black is hoping to take one of White’s rooks and then capture the knight on c8. Ah, but this doesn’t work either, for now White has 5. NxBb6 and he still wins a piece. So Black is better off losing the exchange as described at the end of the previous paragraph. You need to see these variations, though, as they might have ruined everything. The point to remember is that when either or both sides are moving their knights around, you have to keep asking about the next capture or fork the pieces can give from their new squares. And when you're operating without checks, as White largely is here, it makes life complicated. You have to keep asking what counterthreats the enemy could make elsewhere.
This position is full of other useful lessons large and small. Start with the small: looking back now, you can see that the knight on e5 was overworked. It guarded Black’s queen and also the c6 square, which White was able to use as a springboard for a double attack with his knight. Second, the layout of Black’s rooks here is worth a stare; they are nicely arranged for a knight fork.
But the most important lesson here involves move order. It all works for White so long as he liquidates the queens at precisely the right moment, neither too soon nor too late. He doesn’t do it on the first move because keeping the Black queen on the board also keeps the Black knight where it is. He doesn’t wait until the third move because then Black has time to turn the tables. He does it on the second move, after the Black queen’s presence has fulfilled the useful (to White) purpose of freezing the e5 knight long enough for White to play Nc6+. A final way to state the key point: notice that the safety of each queen from the other is the paramount consideration at every turn. If either party is able to capture the guard of the enemy queen and do it with check, he will win the queen and the game. In the end neither side loses a queen (they merely are traded), but the pressures created for each by the need to avoid that result dictates much of the other tactical operations. It is the reason why Black’s first reply is not RxNc6, and it is the reason why White’s second move must not be Ne7.