The hard part is to observe that there exist two different Sylow-3 subgroups. Indirect assumption: there is only one (which is a normal subgroup, of course). Denote it by $P_9$.
If $P_9$ is normal, then any 5-Sylow $P_5$ normalizes it: so $P_5$ acts on $P_9$ by conjugation. This action must be trivial: there are two groups of order $9$, and none of the two has an automorphism of order $5$. (Check this!)
But then elements of $P_5$ and $P_9$ commute, so $P_9$ is contained in the normalizer of $P_5$. But this is nonsense: the normalizer of $P_5$ is $P_5$ itself, as its index is 36 (by Sylow's theorem, this index is the number of $5$-Sylows).
The $36$ Sylow-5 subgroups provide $144$ elements of order $5$: all such subgroups contain $4$ nontrivial elements (of order $5$).
Everything else must be squeezed into the remianing $36$ elements.
Assume indirectly that all Sylow-3 subgroups intersect trivially. All of them have $8$ nontrivial elements. so that would provide us with at least $8n_3\leq 36$ elements whose order is a power of $3$. Thus, $n_3\leq 4$.
Combining this with the observation $n_3\geq 2$ and the Sylow theorems, $n_3=4$ is the only possibility.
Thus there are $32$ nontrivial elements contained in $3$-Sylows, i.e., of order which is a power of $3$.
There are only 4 elemets remaining: they must be the elements of a 2-Sylow. Hence, there is only one 2-Sylow, which must be a normal subgroup.
But then we get the same contradiction as above: there are only two $4$-element grups, and none of them has an automorphism of order 5. So this Sylow-2 subgroup would normalize any Sylow-5 subgroup, which is impossible.