Suppose that's not true. We will have there exists $s>0$, such that with positive probability, $W_t$ won't hit zero after time $t=s$, i.e., $$\mathbb{P}^0(W_t\neq 0 \text{ for any }t\ge s)>0.$$
Note that using simple Markov property, the left hand side equals
$$\mathbb{E}^{0}(\mathbb{P}^{W_s}(W_t\neq 0 \text{ for any }t\ge 0)).$$
Now for any $x$, (for simplicity we assume $x>0$)
$$\mathbb{P}^{x}(W_t\neq 0 \text{ for any }t\ge 0)=\lim_{M\uparrow \infty}\mathbb{P}^{x}(\text{hit }M\text{ before hit }0)=\lim_{M\uparrow \infty}\frac{x}{M}=0.$$
Hence $$\mathbb{P}^0(W_t\neq 0 \text{ for any }t\ge s)=\mathbb{E}^{0}(\mathbb{P}^{W_s}(W_t\neq 0 \text{ for any }t\ge 0))=0,$$
which leads to a contradiction.