In the paper: "Universal One-Way Hash Functions and their Cryptographic Applications". The following is proven: "If 1-1 one-way functions exist, then the one-way based signature scheme described above is secure." For a signature scheme defined in the paper, where by secure they mean "existentially unforgeable under an adaptive chosen plaintext attack", if I understand correctly this is EU-CMA.
Now SPHINCS constructs a post-quantum EU-CMA, PQ-EU-CMA (SPHINCS definition) scheme from the stronger assumptions of among others collision-resistant hashes. If I understand correctly CRHFs are black-box separated from OWFs, in "Finding collisions on a one-way street: Can secure hash functions be based on general assumptions?". Is there any work proving PQ-EU-CMA from OWFs only?
Based on the suggestions by lamontap and some more thought, is there a study of PQ-EU-CMA from some post-quantum generalization of OWFs? Such as those defined in One-Way Functions Imply Secure Computation in a Quantum World