Conceptually it can be viewed as the special coprime case of the general fact that PIDs are gcd domains, whose ideals are generated by any gcd of their generators, since
$$\begin{align}
\overbrace{(a_1,\,\ldots,\,a_n)}^{\large (a_1)\,+\cdots +\, (a_n)} &\,=\, D,\ \ \ \text{wich means, by def'n of ideal sum, that}\\[.2em]
\iff\ \ \ C\supseteq (a_1),..,(a_n) &\iff \, C\,\supseteq\, D,\ \ \ \text{for all ideals $\,C\subseteq A$}\\[.2em]
\iff\ \ \ (c) \!\supseteq \!(a_1),..,(a_n) &\iff (c)\supseteq (d)\ \ \ \text{by $A$ is a PID} \\[.2em]
\iff\ \ \ \ \ c\ \ \mid\ \ a_1,\ldots,a_n\,\ &\iff\ \ c\ \ \mid\ \ d,\,\ \ \ \text{by contains = divides in a PID}\\[.2em]
\iff\ \ \ \ \ \ \ \gcd(a_1,\cdots, a_n)\! &\ \ \sim\ d,\ \ \ \text{where $\,\sim\,$ means associate}
\end{align}\qquad$$
where the final $\!\iff\!$ uses the gcd universal property, and the first is the universal property of the ideal sum $\,C\supseteq A_1+A_2\!\!\iff\! C\supseteq A_1,A_2,\, $ where $(a_1,\ldots,a_n) = (a_1)+\cdots (a_n)\,$ by definition.
As is often the case, the proof is straightforward - practically writing itself - when we employ the basic universal properties that define the objects.
For motivation see here for the case when the PID $=\Bbb Z$.