Proxy Re-Encryption provides the ability to transform a ciphertext encrypted under some key to the one under another key. This seems interesting. But, something seems strange.
Suppose Alice has a key pair $(sk_a, pk_a)$ and Bob also has a key pair $(sk_b, pk_b)$. If Alice and Bob generate a proxy re-encryption key to transform a ciphertext under $pk_a$ to the one under $pk_b$ and vice versa, they can send messages without revealing their secret keys.
At this point, what about the following scenario? Bob securely transmits his public key $pk_b$ to Alice. Then, Alice encrypts her message $m$ using $pk_b$. There is no need to transform the ciphertext. Also, Alice can send Bob her public key $pk_a$ when she wants to receive the message from Bob.
What is the advantage of proxy re-encryption? I think there is no potential difference between a standard public key encryption scheme and proxy re-encryption.