First understand $\mathrm{F}$.
The first part
$\rho_{\mathrm{fuser}\leftarrow \mathrm{username}}(\pi_{\mathrm{username}}(\mathrm{FriendsWith}))$
contains everyone appearing in the first friendship column.
The second part
$\rho_{\mathrm{fuser}\leftarrow \mathrm{username2}}(\pi_{\mathrm{username2}}(\mathrm{FriendsWith}))$
contains everyone appearing in the second friendship column.
Their union therefore contains every user participating in at least one friendship.
Now consider
$\mathrm{Users}\times \mathrm{F}$.
Select matching usernames:
$\sigma_{\mathrm{username=fuser}}(\mathrm{Users}\times \mathrm{F})$.
These are precisely the $\mathrm{Users}$ tuples belonging to people who have at least one friend.
Project:
$\pi_{\mathrm{username,email}}\left(\sigma_{\mathrm{username=fuser}}(\mathrm{Users}\times \mathrm{F})\right)$.
So this relation represents:
users who have at least one friend.
But we need the opposite.
Start from all users:
$\pi_{\mathrm{username,email}}(\mathrm{Users})$
and subtract users who have friends:
$\pi_{\mathrm{username,email}}(\mathrm{Users})$ $-$ $\pi_{\mathrm{username,email}}\left(\sigma_{\mathrm{username=fuser}}(\mathrm{Users}\times \mathrm{F})\right)$.
Therefore, A is correct.