Assuming the use of big o notation refers to the limiting behavior of $x$, as $x$ tends to $\infty$, then the following argument will work, if you are describing some other limiting behavior the requirements on x can be adjusted accordingly.
If we have for some $n_0$, that there exists a positive constant $M$, such that for all $x>n_0$ $$|f(x)|\leq M|g(x)|$$
And we also have for some other $a_0$, that there exists a positive constant $C$ such that for all $x>a_0$, $$|g(x)|\leq C|h(x)|$$
Then we may pick the maxima of either $n_0$ or $a_0$ so that we can say both statements hold for all $x>\text{max}{(a_0,n_0)}$, thus both inequalitys holds for the same range of x, and we are permitted to manipulate one with the other. From here we can deduce that
$$|f(x)|\leq MC|h(x)|$$
For all $x$ such that, $x>\text{max}{(a_0,n_0)}$, which is the exact definition of $$f(x)=O(h(x))$$
Where the constant for which the absolute value of $f(x)$ is less then it times $h(x)$ is equal to $MC$, and the statement holds for all $x>\text{max}{(a_0,n_0)}$