Can it also be used as a one-time encryption scheme?
You cannot build asymmetric encryption from just hash functions. There is an impossibility proof for that. So for asymmetric encryption the answer is no. See Merkle Puzzles are Optimal - Boaz Barak, Mohammad Mahmoody-Ghidary for the proof.
For symmetric encryption, you can simply use the hash function directly to build encryption, no need to bother with Winternitz.
But you could use quantum-key-exchange together with hash based signatures to achieve something similar to asymmetric encryption.
Concerning many-time signatures
It's straight forward to build stateful many-time signatures from one-time signatures. Just create as many one time keys you need, compute a hash tree over their public keys and public its root. The hash tree allows you to efficiently prove what the nths public key is and proceed with the one-time signature form there.
But if you don't want to keep track of which one-time signatures you already used, you need a stateless many-time signature. It is possible to build stateless hash signatures, but it's a bit trickier. SPHINCS is a recent example.