How does Boxed closure calls work just like closure calls?

Why does calling boxed closures work just like calling closures? Is it because of Deref implementation of Box ? Would it work on all smart pointers? As I tried some examples to better understand it, I get the following error.

fn main() {
    let mut x = String::new();
    let y = || {
        let t = &mut x;
    };
    let mut ww = Box::new(y);
    (&mut ww)();
}

This throws an error

error[E0525]: expected a closure that implements the `Fn` trait, but this closure only implements `FnMut`
  --> src/main.rs:22:13
   |
22 |     let y = || {
   |             ^^ this closure implements `FnMut`, not `Fn`
23 |         let t = &mut x;
   |                      - closure is `FnMut` because it mutates the variable `x` here
...
26 |     (&mut ww)();
   |     ----------- the requirement to implement `Fn` derives from here

Any documentation on how boxed closures work would be helpful as well.

from Recent Questions - Stack Overflow https://ift.tt/3IfvkoQ
https://ift.tt/eA8V8J