While it's quite easy to give a derivation of
$$\forall x ~ \bigg(A(x) \implies B(x)\bigg) \implies \bigg(\forall x~ A(x) \implies \forall x~ B(x)\bigg)$$
in a system that contains the rule of Universal Instantiation, how do you give one in a Hilbert style system like the one in Derek Goldrei's Predicate and Propositional Calculus: A Model of Argument. His Axiom 4,
$$\forall x (A \implies A[t/x]) \tag{Axiom 4}$$
where $t$ is free for $A$, seems to work as a kind of UI, but I'm not that sure. What I'm trying to say is: Does Axiom 4 warrant the inference:
$$\forall x~ (A(x) \implies B(x))$$ $$\downarrow$$ $$\text{(Apply Axiom 4)}$$ $$\downarrow$$ $$(A(x) \implies B(x))$$
...?
1)$∀x (A(x) \to B(x))$ --- premise [a]
2) $∀x A(x)$ --- premise [b]
3) $A(x) \to B(x)$ --- from 1) by Ax.4 and modus ponens
4) $A(x)$ --- from 2) by Ax.4 and modus ponens
5) $B(x)$ --- from 3) and 4) by mp
6) $∀x B(x)$ --- from 5) by Gen rule : $x$ is not free in [a] nor in [b]