The submitted manuscript “Schweinfurthins and their analogues are highly selective cellular probes for oxysterol-binding protein” reports interesting data for this natural product (SW) and a valuable panel of 58 analogs. This work builds from well-established previous work on this family of proteins and analogs, including characterization of inhibitor classes (“oxybipins” and “ORPphilins”) that target oxysterol binding protein (OSBP) and potentially other family members. The phenotypic results of this inhibition are disruption of Golgi integrity resulting in anti-cancer and anti-viral effects. Of note, the cytotoxicity profiles of SW and analogs have previously been published.

The current work presents synthesis of 13 SW analogs; these data are provided as supplemental information; additional citations are provided in an excel sheet that would hopefully also be included as supplemental information. Overall, the data reporting screening of the SW analog library is of interest to the field, although with issues concerning replication and a lack of error estimations that complicate comparisons. The manuscript makes compelling correlations of IC50 (or are they Kd?) values with NCI GI50 measurements for specific classes of SW analogs. Docking studies could be better supported with relative scores and variations in poses of individual compounds given. These do appear to rationalize most of the differences noted in binding. The manuscript seeks to demonstrate selectivity of compounds for OSBP over its close homolog ORP4; in. My opinion, the experimental design as presented is flawed and, as such, results do not support this conclusion.

Overall there is significant work reported in this manuscript and it will be of value to the field. However, there are limitations that are not addressed, and improvement in data presentation and descriptions should be made to strengthen the manuscript.

Details supporting these comments:

In total, 59 “SW analogs” (including SW A and G) were screened at a single-dose (10 µM) for activity against the ligand binding domains of recombinant OSBP, ORP1, ORP2, Aster-A, -B, -C, and STARD1, 3, 4, and 5. Strikingly, at this dose only OSBP appears to be reactive (e.g., “60-100% inhibition”), and with variation correlated to specific SW modifications. The remaining presentation seeks to capture these correlations to identify the structure-activity relationship of SW analogs for the OSBP recombinant protein.

The nomenclature used to present the data in the table of Figure 1 is somewhat problematic, where “inhibition” and “affinity” are used interchangeably. I understand that this arises from the combination of methods, where a competition assay is used for most, resulting in “inhibition” [of binding to a probe], while the STARD members are screened by a melting point assay, resulting in a metric associated with “affinity”. The screens use appropriate assay controls to ensure signals are arising from the expected source; this would be strengthened, however, if control compounds were used. The reported values do not contain error estimations (replicate details also appear to be missing); these are critical to distinguish potential differences in values (e.g., is 78 nM different from 106 nM, or are these values within error estimates?).

Given the expansive collection of known OSBP and other “inhibitors”, there would be significant power in benchmarking the current library to these known entities. These data would have more utility if the reported IC50 values were converted to apparent Kd values; the methods indicate that this was done (“Curves were fitted to the normalized data via nonlinear regression to allow the determination of kd values”), yet the tables and discussion refer to IC50 values. At multiple points in the narrative IC50 values are referred to as “affinity”, which isn’t strictly the case (“relative competitive binding” between two molecules may be more appropriate?). Two probes are listed for the competition assay; please define which probe is used in which assays.

An undefined abbreviation “SBP” (sterol binding pocket?) appears (pg 8) in the description of docked poses for compounds and elsewhere.

Fig 2 could be improved by labeling R positions of ligands, especially as these are referred to in the description of putative interactions. Fig 2 does not convincingly demonstrate selectivity imposed by Thr 491, as the main chain atom is forming bonds. If this information is important to demonstrate selectivity versus ORP4, a similar view for OPR4 should be provided (or simulations with position 491 of OSBP mutated to the equivalent in ORP4). Somewhere the bond distance range (or specific value for those indicated) for assigning these bonds needs to be defined.

The descriptions of the compound modifications and rationalization based on binding poses is clear, and the concluded SAR is supported by the presented data (sans error measurements, however).

Pg 13 refers to R490/R440; however, this should probably be R490/R449? (residue 440 is a Phe in Fig 2 and elsewhere in the text).

The use of intrinsic fluorescence of selected compounds is a clever approach and yields both direct binding (although no error measurements or indication of replicates are provided) and visualization data. The high content imaging provides interesting insights, however, it is very difficult to verify ER localization based on these images. The authors are appropriately cautious in the interpretation, simply suggesting that the compounds may be co-localizing with OSBP. It is unclear if this concentration (10 µM) and incubation time (30 min) would be sufficient to elicit a phenotype; however, such data would give more support to the purported activity. (Fig 3A legend has a small typo “absorbance”)

Pg 18 states “corresponding published log GI50 (half-maximal growth inhibitory concentration) values of all 59 SWs screened against NCI-60 cancer cell lines”, but no citation is given – can you please add this? It is a bit confusing because if it is published the data and accompanying SI for this manuscript may be redundant / not needed?

Pg 20 describes results of experiments assessing OSBP and ORP4 stability – can the authors please clarify if ORP4S is representative of all ORP4 in these cells? Note, Fig S1 is fluorescence data, while figure S2 are quantified Western blots (please note the legend for fig S2 also incorrectly refers to panel 1D of fig S1, and the labels used for panels and in the legend in Fig S3 are really unclear also).

The Western blot approach is unclear – the methods describe how the ITDRF samples were prepared, but do not include details for the Western blotting procedure – presumably it is the same as listed for “Immunoblotting analysis”? It is unclear because this section begins “For GPP130 degradation....” and does not the revert to generalized methods for all sample types. Details of quantification methods are not included. GPP130 measurements appear to be at different dosages between the two presented experiments, making the number of replicates unclear (potentially one measurement per concentration). The higher dose experiment (30 – 100 nM) includes limited dose dependence for compounds 1, 6, and 46, while compound 7 shows a dose responsive loss. The other panel depicts responses to 3-30 nM compound with dose response for compounds 1, 6, 8, and 46. Overall, these results do not seem to replicate well, as comparison of the y axes of response signal does not correlate.

The ITDRF experiments have problematic interpretations, and no loading control is included such that quantification is questionable. It appears that the blots are probed with an anti-ORP4S antibody (?); if so, it is recommended that results reporting and discussion limit comments to ORP4S (rather than “ORP4”). While the stabilizing effect of compounds 1, 7, and 46 are demonstrated with OSBP when lysate is heated to 53C, no impact was noted for ORP4S. However, no temperature gradients are presented and it is possible that ORP4S is stabilized by compounds but has a lower melting point, such that at 53C it precipitates regardless of compound presence. As such, claims of preferential stabilization cannot be supported. In light of this, the title should not make claims for selectivity for OSBP.

Methods for a Sterol Transfer Assay and Micro-plate Based Cholesterol Transfer Assay are given, but I do not see any results provided nor are these discussed? Perhaps these methods are intended for a different manuscript?

The conclusion states “The interactions with D453 and K577 at the bottom of the binding site are crucial for OSBP binding, while the R490/R440 interaction cluster is critical for a high affinity.” However, no mutations were made to these residues, and no validation of docking is provided; the statement should indicate that predictions suggest these interactions are important rather than being stated as fact. The same care in stating conclusions should be used when discussing linker impacts on orientation.

“stabilization of OSBP but not ORP4, confirming the excellent selectivity of SWs towards OSBP in comparison to other STPs.” “The detailed mechanistic investigation reported here supports the notion that OSBP is the primary target of the SWs, showing remarkable selectivity even over the closely related ORP4.” These concluding statements overstate the observations, which rest solely on the ITDRF experiments that do not adequately demonstrate a lack of stabilization of ORP4S, but simply show that its melting temperature is likely lower than that of OSBP. These experiments are further confounded by a lack of loading controls for normalization of technical errors. Therefore, the conclusion surrounding selectivity is not supported by data.

The authors are commended on their attention to previous suggestions, and previous concerns have been addressed.