Detergent selection as a determinant of lysate melting behavior and drug-target hit-calling in PISA

Abstract

Thermal proteome profiling (TPP) and proteome integral solubility alteration (PISA) assays measure drug-target interactions by monitoring protein thermal stability across the proteome. While detergents are routinely used in lysate-based thermal profiling, the field lacks consensus on whether detergents should be present during the melting step or only added afterward as an extraction buffer, and whether detergent identity matters for this choice. Here, we evaluate how commonly used detergents and the timing of their use in thermal stability workflows affect proteome-wide thermal stability and PISA hit calling in TF-1 lysates. We find that NP-40 and DDM produce highly correlated melting profiles when used exclusively as post-melt extraction buffers, but diverge substantially when present during the melting step. DDM in particular prevents the thermally-induced loss in solubility of large classes of proteins, such as cell surface proteins, and these effects propagate directly into PISA hit calling. Performing the PISA melt in DDM versus NP-40 results in the gain and loss of distinct drug-target interactions for both the PAK4 inhibitor PF-3758309 and the PLK1 inhibitor volasertib. Notably, DDM enables detection of a volasertib-TMEM97 interaction that was previously not detected in NP-40. However, we also find that the stabilization effects of DDM mask the identification of some known PISA hits for these drugs. We further introduce a four-parameter logistic model of protein melting to aid in modeling of these findings and a linear regression framework for PISA hit calling that outperforms pairwise t-tests in low-replicate settings. Together, these results establish detergent selection as a tunable experimental variable in thermal profiling and suggest that some drug-target engagements previously attributed exclusively to intact-cell context may be recoverable in lysates with appropriate buffer conditions.