The first Tudor domain of PHF20L1 (PHF20L1 Tudor1) recognizes both histone methylation and non-histone methylation to play versatile roles, e.g., PHF20L1 Tudor1 binds to the oncogenic target DNA (cytosine-5) methyltransferase 1 (DNMT1) to prevent it from degradation. However, the crystal structure of the PHF20 Tudor domain, a homolog of PHF20L1, reveals a closed aromatic cage of the Tudor domain. It is thus highly desirable to interrogate the ligand-recognition mechanism of PHF20L1 Tudor1, which will in turn validate the potential druggability of this target. Here, we solved the crystal structure of the free form PHF20L1 Tudor1, which adopts the closed conformation similar to PHF20. NMR relaxation dispersion and molecular dynamics (MD) simulations suggest a pre-existing low-population conformation with a remarkable rearrangement of aromatic cage residues. Such structural rearrangement is further revealed by the crystal structures of PHF20L1 Tudor1 in complex with the lysine 142 methylated (K142 me1 ) DNMT1, and a small molecule cosolvent 2-(N-morpholino)ethanesulfonic acid (MES), respectively. This result thus ignites interest in the discovery of small molecule inhibitors against PHF20L1 Tudor1. The hit identified from NMR fragment-based screening protrudes into the same open form aromatic cage of PHF20L1 Tudor1, and blocks the interaction between PHF20L1 Tudor1 and methylated DNMT1. Further free form crystal structures of key mutants reveal one open form and one closed form aromatic cage, which is energetically trapped observed in the NMR relaxation dispersion and MD simulations. The binding of DNMT1 with PHF20L1 Tudor1 mutants was also recapitulated in cancer cells. The mutagenesis thus alters the structure, dynamics and eventually the function of PHF20L1 Tudor1. Our results demonstrate that PHF20L1 Tudor1 utilizes the same conformational selection mechanism to recognize ligands, regardless of whether it is a natural substrate or a small molecule identified from fragment-based screening. Albeit at a low population, the pre-existing ligand-binding conformation shall shift the paradigm in the druggability assessment of a dynamic protein, even though it may lack a small molecule binding pocket in its free form structure. The inhibition of PHF20L1 paves an alternative way to target DNMT1 degradation.