CC(=O)NCCc1c2cc(F)cc(C(CN(Cc3cc(C)on3)C(=O)NC3CC3)c3nnc(C)s3)c2[nH]c1
CC(=O)NCCc1c2cc(C#N)cc(C(CN(Cc3cc(C)on3)C(=O)NC3CC3)c3nnc(C)s3)c2[nH]c1
CC(=O)NCCc1c2cc(Cl)cc(C(CN(Cc3cc(C)on3)C(=O)NC3CC3)c3nnc(C)s3)c2[nH]c1
A few merges that would populate all four subsites (S1, S1', S2, S3). x0397 reaches from S1 to S1', and its N-methyl group points towards S2 (aromatic wheel) and S3. x0395 sits one C-C away from that N-methyl group; one half sits in S1', the other (non-aromatic) half in S2. Finally, replace that non-aromatic half with x0104, which reaches from S2 to S3. Try the nitrile and fluoro that poke into the H61/M165 hole for various hits - and toss in a chloro for good measure.
Why this might not work: H61, C145 and M165 move quite a lot for x0397, away from how they are for the aromatic hole. And probably a bitch to synthesise.