When you want to prove something about cardinalities you actually say "I don't really care who is in the set, I just care about its size".
This means that you don't really care that $(1,2,3)\notin\mathbb N^2\times\mathbb N$, because once you proved that $\mathbb N^2\times\mathbb N$ is countable you have a natural bijection: $$\left(\left(x,y\right),z\right)\mapsto\left(x,y,z\right)$$
This function is clearly bijective, and while you cannot write "clearly" in a homework assignment - it is not hard to prove that for yourself as well. Every $(n+1)$-tuple is just an $n$-tuple with an additional element.
Alternatively, you can use the fact that $\mathbb N\times\mathbb N$ is countable as well and just "skip" the above part, by doing it implicitly as follows:
So now the induction step would be to take $f_n\colon\mathbb N^n\to\mathbb N$ which is a bijection, define $g\colon\mathbb N^{n+1}\to\mathbb N\times\mathbb N$ as follows:
$$g((a_1,\ldots,a_{n+1})) = (f_n(a_1,\ldots,a_n),a_{n+1})$$
This function is injective since two different tuples will either have a different coordinate $k\le n$ in which case the $f_n$ part would be different by injectivity of $f_n$; or different coordinate at the $n+1$ place, in which case the right coordinate of the image will be different.
It is also surjective due to a similar argument. And then $f_{n+1}$ is the composition of $g$ with a fixed bijection from $\mathbb N\times\mathbb N$ to $\mathbb N$.