It is not $F(A)$ that is isomorphic to $\mathbb{Z}$, but rather $K(A)$. In general, $F(A)$ does not have a group structure, and the Grothendieck group $K(A)$ is the most free way of imposing a group structure on $F(A)$ such that exact sequences behave as specified.
An element of $C$ is a finite formal sum $\sum_{k \geq 0} n_k [V_k]$ where $V_k$ is a $k$-dimensional vector space over the field $A$ and $n_k \in \mathbb{Z}$. If you consider that element in $C/D = K(A)$, then since you can always find an exact sequence
$$
0 \rightarrow V_{k-1} \rightarrow V_k \rightarrow V_1 \rightarrow 0
$$
(where the subscript denotes the dimension of the vector space), it is not hard to show by induction that $[V_k] = k [V_1]$ in $K(A)$. Thus every element in $K(A)$ has the form $\sum_{k \geq 0} n_k k [V_1]$ and the map sending this element to $\sum_{k \geq 0} n_k k \in \mathbb{Z}$ is easily seen to be an isomorphism (exercise).
Hint for the PID case: Use the exact sequence
$$
0 \rightarrow A \xrightarrow{p} A \rightarrow A / (p) \rightarrow 0
$$
to show that $[A / (p)] = 0$ in $K(A)$. What does every finitely generated module over a PID look like? Use this, the above observation and the argument in the case when $A$ is a field to prove that $K(A) \cong \mathbb{Z}$.