Supporting Information

Supporting Information

Virtual screening for ligand discovery at the σ

receptor

Daniel A. Greenfield

_{, Hayden R. Schmidt}

_{, Meredith A. Skiba}

_{, Michael D. Mandler}

_{, Jacob}

Anderson

_{, Piotr Sliz}

_{*, and Andrew C. Kruse}

_*

1

_{Harvard Medical School, Department of Biological Chemistry and Molecular Pharmacology, Boston, MA 02115,}

USA.

_{Harvard University, Department of Chemistry and Chemical Biology, Cambridge, MA 02138, USA.}

_Boston

Children’s Hospital, Boston, MA 02115, USA.

_{These authors contributed equally.}

_{Present address: University of}

Lausanne, Center for Integrative Genomics, Lausanne CH-1015, Switzerland.

*Correspondence should be

addressed to: [email protected] or [email protected].

Experimental procedures Computational Methods

Prior to docking, the σ1 receptor structure (PDB ID: 5HK1) was put

through the Protein Preparation Protocol34_{. This added hydrogens,}

removed co-crystallized water molecules, enumerated bond orders, optimized the receptor’s hydrogen bond network, and ran a restrained minimization to alleviate backbone clashes within the receptor backbone. Additionally, Asp126 was protonated in our docking screens on the basis of its apparent hydrogen bond with Glu172 in the crystal structure of the receptor35_.

The eMolecules ligand library was prepared using the Ligand Preparation Protocol34 _{to generate three-dimensional geometries,}

assign proper bond orders, and generate accessible tautomer and ionization states for the ligands prior to virtual screening36_.

The grid generation protocol in Glide Docking defines the target area (binding site) for each ligand to be tested within during the docking program. This was centered between the carboxylates of Glu172 and Asp126 and was kept constant as the docking precision was increased. Docking began with high throughput virtual screening (HTVS), in which all compounds in the eMolecules library were tested. All ligands with Glide scores more favorable than the score of the re-docked co-crystallized ligand PD144418 were put through standard precision screening (SP). All compounds scoring better than 1 standard deviation from the mean were tested in extra precision (XP), and the top 10% of XP ligands were put through flexible XP docking, where the receptor backbone was given freedom to shift in order to fit ligands to the orthosteric site.

To perform enrichment analysis similar to previous approaches37_{, a set of 15 known σ}

1 receptor binders were collected

from the PDSP Ki Database38 _{and put through the ligand}

preparation protocol. These ligands were then tested in varying precision levels after being merged with a set of 100,000 eMolecules ligands of similar molecular weight to the known binders. The average molecular weight of the known binders taken from the PDSP Ki Database is 351.07 g/mol, and the sample of ligands that were selected for enrichment with known binders had molecular weights between 300 and 400 g/mol in an attempt to force Glide to focus on differences in chemical composition rather than physicochemical descriptors. Tanimoto similarity coefficients were calculated using ECFP4 molecular fingerprints with the Schrödinger Canvas tool.

Schrodinger Drug Discovery Platform installation was supported by SBGrid Consortium and all data was generated and processed on SBGrid servers.39

Recombinant receptor expression and preparation of Sf9 cell membranes

Membranes were prepared from Sf9 insect cells infected with baculovirus encoding either recombinant σ1 or σ2 receptor,

using a protocol adapted from that of Vilner et al., as previously described35, 40, 41_{. Briefly, either the σ}

1 receptor or σ2 receptor,

which was recently identified as TMEM9741_{, were expressed in Sf9}

cells using the pFastBac baculovirus expression system. Cells were lysed by osmotic shock and membrane fractions were separated by repeated centrifugation and washing steps, followed by dounce homogenization. Total protein content was quantified using the BioRad DC protein assay, and membranes were stored in 100 μL aliquots at a concentration of 3-10 mg/mL at -80 °C until use.

Screening docking hits for σ1 and σ2 receptor binding

Membrane radioligand binding assays were performed as described with slight modifications35, 41, 42_{. All compounds were diluted to 10}

μM in 50 mM Tris pH 8.0. Membrane samples were also diluted and homogenized in 50 mM Tris pH 8.0 using a needle and syringe. A total of 2.5 μg of protein was added to each well of a 96-well block in a final volume of 100 μL, which also contained 1 μM of the compound and 10 nM of the appropriate radioligand (3_H

(+)-pentazocine for the σ1 receptor, and 3H DTG for the σ2 receptor).

All reactions were performed in triplicate. Haloperidol was used as a positive control, and buffer alone served as a negative control. Samples were incubated at room temperature with shaking for 1.5 h. The reaction was then terminated and samples were then applied to glass fiber filters (Merck Millipore) using a Brandel cell harvester. Filters were soaked with 0.3% (w/v) polyethylimine prior to use. After harvesting, filters were soaked overnight in scintillation vials containing 5 mL of Cytoscint scintillation fluid and measured on a Beckman Coulter LS 6500 scintillation counter.

Saturation binding in Sf9 cell membranes.

3_{H (+)-pentazocine and} 3_{H DTG saturation binding to Sf9}

membranes respectively expressing σ1 or σ2 receptor were

determined using a saturation binding assay similar to that described by Chu and Ruoho42_{,which we have described}

previously35, 41_{. The membranes (2.5 μg and 0.5 μg total protein per}

reaction for σ1 and σ2 expressing membranes, respectively) were

incubated in a 100 μL reaction buffered with 50 mM Tris pH 8.0, containing 0-300 nM 3_{H (+)-pentazocine for σ}

1 binding, or 0-300

nM 3_{H DTG for σ}

2 receptor binding. In the σ2 receptor binding

assay, concentrations of 100 and 300 nM DTG were achieved by isotopic dilution to minimize 3_{H DTG expenditure. For both}

receptors, nonspecific binding was assayed by parallel reactions containing 2 μM haloperidol. Reactions were incubated at 37 °C for 1.5 h and then terminated, filtered, and measured as for the single-point binding described previously. All reactions were performed in triplicate in a 96-well block. KD values were

calculated using non-linear regression tools in Graphpad Prism.

Competition binding assays in Sf9 membranes.

3_{H (+)-pentazocine or} 3_{H DTG competition curves testing the}

binding the docked compounds to the σ1 and σ2 receptors

respectively, were performed similarly to the protocol described by Chu and Ruoho42_{,with slight modifications we have used}

previously35, 41_{. Briefly, Sf9 insect membranes overexpressing}

either σ1 receptor (2.5 μg of total protein per reaction) or σ2 receptor

(0.5 -- 2.5 μg total protein per reaction) were incubated in a 100 μL reaction buffered with 50 mM Tris pH 8.0, with the appropriate radioligand at a 10 nM concentration and eight competing cold compound concentrations ranging from 10 pM – 100 μM. Reactions were incubated at 37 °C for 1.5 h to reach equilibrium, and then terminated by filtration as described previously. Ki values were computed by directly fitting the data and using the experimentally determined probe KD to calculate a Ki value, using

the Graphpad Prism software. This process implicitly uses a Cheng-Prusoff correction, so no secondary correction was applied.

LC/MS analysis of compound purity

Compounds were purchased from the following vendors: 1 - ChemDiv; 2, 3, 4, 5, 6 - ChemBridge; 7, 8, 9, 10 – Enamine Screening Compounds; 11 – Life Chemicals; 12 – Vitas M Labs; 13 – Otava; PD144418 – R+D Systems; haloperidol – Tocris. Purity of purchased compounds was analyzed using an Agilent 1100 series HPLC equipped with a Phenomenex Luna 3u C18 column (250 x 4.6 mm). Samples were separated with 0-100% gradient of acetonitrile in water over 40 minutes with detection at 254 nm. High resolution electrospray mass spectra (HRMS) were obtained on an Agilent QTOF 6530.

1 HRMS (ESI+) m/z calcd for C18H18F3N3S [M + H] 366.1246,

found 366.1243, 99% pure; 2 HRMS (ESI+) m/z calcd for C21H31N3O [M + H] 342.2540, found 342.2545, 94% pure; 3

HRMS (ESI+) m/z calcd for C18H22N4OS [M + H] 343.1587, found

343.1587, 90% pure; 4 HRMS (ESI+) m/z calcd for C16H22N3O2F3

[M + H] 346.1737, found 346.1734, 97% pure; 5 HRMS (ESI+) m/z calcd for C18H15N2OF3 [M + H] 333.1209, found 333.1210,

99% pure; 6 HRMS (ESI+) m/z calcd for C19H22N6O [M + H]

351.1928, found 351.1918, 92% pure; 7 HRMS (ESI+) m/z calcd for C18H19N3OClF [M + H] 348.1273, found 348.1279, 95% pure;

8 HRMS (ESI+) m/z calcd for C17H13N3BrF3 [M + H] 396.0318,

found 396.0322, 97% pure; 9 HRMS (ESI+) m/z calcd for C18H15N3Cl2F2 [M + H] 382.0684, found 382.0681, 88% pure; 10

HRMS (ESI+) m/z calcd for C21H23N4OF [M + H] 367.1929, found

367.1943, 86% pure; 11 HRMS (ESI+) m/z calcd for C22H23N3O

[M + H] 346.1914, found 346.1912, 98% pure; 12 HRMS (ESI+) m/z calcd for C19H19N2OF3 [M + H] 349.1522, found 349.1538,

93% pure; 13 HRMS (ESI+) m/z calcd for C20H19N2OF3 [M + H]

AUTHOR INFORMATION

Corresponding Authors

* Piotr Sliz, Department of Biological Chemistry and Molecular Pharmacology, Blavatnik Institute, Harvard Medical School, Boston, MA 02115; E-mail: [email protected]

* Andrew Kruse, Department of Biological Chemistry and Molecular Pharmacology, Blavatnik Institute, Harvard Medical School, Boston, MA 02115, [email protected]

Present Addresses

†University of Lausanne, Center for Integrative Genomics, Lausanne CH-1015, Switzerland

Author Contributions

D.A.G. designed and carried out computational docking and analysis of results, and together with H.R.S. performed radioligand binding assays and associated data analysis. M.D.M., M.A.S., and H.R.S. analyzed compound purity. J.A. carried out

computational docking of additional compounds. P.S. Co-designed the computational pipeline and co-supervised the docking experiments. A.C.K. supervised the work and together with D.A.G. and H.R.S. wrote the manuscript. ‡These authors contributed equally.

Funding Sources

This work was supported by NIH grant R01GM119185 (A.C.K.), and grant number FG-2017-9226 from the Alfred P. Sloan Foundation (A.C.K.). H.R.S was supported by an NSF Graduate Research Fellowship DGE1745303. D.A.G was supported by BCMP Scholars Internship Program and Molecular Biophysics Training Grant NIGMS T32 GM008313.

ACKNOWLEDGMENT

We would like to thank Dr. Matthew Henke and the Harvard Medical School Analytical Chemistry core for assistance with LC/MS.

ABBREVIATIONS

NMDA, N-methyl-D-aspartate; ECFP, Extended Connectivity Fingerprints; HTVS, High Throughput Virtual Screening; SP, Standard Precision Virtual Screening; XP, Extra Precision Virtual Screening; 4-IBP, N-(benzylpiperidin-4yl)-4-iodobenzamide; PD144418, 1,2,3,6-tetrahydro-5-[3-(4-methylphenyl)-5-isoxazolyl]-1-propylpyridine; NE-100, N, N-dipropyl-2-[4-methoxy-3-(2-phenylethoxy)phenyl]-ehthylamine monohydrochloride; DTG, 1,3-Di-(2-tolyl)guanidine

Supplementary figures

Figure S1. Chemical structures of the ligands tested in s receptor binding assays, and PD144418, the ligand from the s receptor crystal structure used in docking.

Figure S2. Docked poses of the thirteen compounds chosen for experimental characterization. The s1 receptor is

Table S1. Measured Ki affinity values for each compound for both the s1 receptor. Bold values are those used in main

text Table 1.

σ

receptor values (M)

Compound

Rep. 1

Rep. 2

mean

SEM

SD

95% CI

1

2.25E-08 6.16E-08 4.21E-08 1.96E-08 2.76E-08 3.83E-08

2

ND

ND

3

1.27E-07 9.10E-08 1.09E-07 1.80E-08 2.55E-08 3.53E-08

4

1.80E-08 3.07E-08 2.44E-08 6.35E-09 8.98E-09 1.24E-08

5

1.91E-07 1.17E-07 1.54E-07 3.70E-08 5.23E-08 7.25E-08

6

ND

ND

7

2.204E-07 2.53E-07 2.37E-07 1.63E-08 2.31E-08 3.19E-08

8

ND

ND

9

1.64E-07 1.77E-07 1.71E-07 6.50E-09 9.19E-09 1.27E-08

10

ND

ND

11

ND

ND

12

9.15E-07 4.77E-07 6.96E-07 2.19E-07 3.10E-07 4.29E-07

Table S2. Measured Ki affinity values for each compound for both the s2 receptor. Bold values are those used in main

text Table 1.

σ

receptor values (M)

Compound Rep. 1

Rep. 2

Rep. 3

mean

SEM

SD

95% CI

1

4.66E-07 3.89E-07 9.49E-07 6.01E-07 1.75E-07 3.04E-07 4.21E-07

2

ND

ND

ND

3

6.43E-07 2.68E-06 2.37E-06 1.90E-06 5.49E-07 1.10E-06 1.52E-06

4

7.55E-08 1.53E-08

4.54E-08 3.01E-08 4.26E-08 5.90E-08

5

ND

ND

ND

6

ND

ND

ND

7

3.15E-07 2.50E-07 1.24E-07 2.30E-07 5.61E-08 9.71E-08 1.35E-07

8

ND

ND

ND

9

9.85E-07 1.69E-06 3.87E-06 2.43E-06 8.68E-07 1.50E-06 2.08E-06

10

ND

ND

ND

11

ND

ND

ND

12

ND

ND

ND

Supplementary references

34. Sastry, G. M.; Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W., Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput Aided Mol Des 2013, 27 (3), 221-34.

35. Schmidt, H. R.; Zheng, S.; Gurpinar, E.; Koehl, A.; Manglik, A.; Kruse, A. C., Crystal structure of the human sigma1 receptor. Nature 2016, 532 (7600), 527-30.

36. Halgren, T. A. M., R. B.; Friesner, R. A.; Beard, H. S.; Frye, L. L.; Pollard, W. T.; Banks, J. L., , Glide: A New Approach for Rapid, Accurate Docking and Scoring. 2. Enrichment Factors in Database Screening J. Med. Chem. 2004, 47, 1750–1759. 37. Huang, N.; Shoichet, B. K.; Irwin, J. J., Benchmarking sets for molecular docking. J Med Chem 2006, 49 (23), 6789-801.

38. Roth B.L., L. E., Patel S., Kroeze W.K., The Multiplicity of Serotonin Receptors: Uselessly Diverse Molecules or an Embarrassment of Riches? Neuroscientist 2000, 6 (4), 252-262.

39. Morin, A.; Eisenbraun, B.; Key, J.; Sanschagrin, P. C.; Timony, M. A.; Ottaviano, M.; Sliz, P., Collaboration gets the most out of software. Elife 2013, 2, e01456.

40. Vilner, B. J.; John, C. S.; Bowen, W. D., Sigma-1 and sigma-2 receptors are expressed in a wide variety of human and rodent tumor cell lines. Cancer Res 1995, 55 (2), 408-13.

41. Alon, A.; Schmidt, H. R.; Wood, M. D.; Sahn, J. J.; Martin, S. F.; Kruse, A. C., Identification of the gene that codes for the sigma2 receptor. Proc Natl Acad Sci U S A 2017, 114 (27), 7160-7165.