Basically, for efficiency. for example the immutable classes can share storage for identical objects. Harmless-looking statements in an object oriented language can often create and destroy temporary objects, and the overheads are much smaller if the contents of these are known to be immutable.
You should only use mutable strings if you know the contents of the string may change after you have created it.
A StringBuilder class just pre-allocates a big array to hold the characters of a string you are yet to create. It prevents the need to call the allocator over and over again if concatenating a bunch of (immutable) shorter strings.
When there is a LOT of string concatenation going on at runtime, it is often more efficient (significantly more) to use StringBuilder.