Supporting Information Structure-based identification of novel ligands ...
Supporting Information
Structure-based identification of novel ligands targeting multiple sites within a chemokine- G protein–coupled receptor interface †
Emmanuel W Smith1, †Amanda M Nevins2, Zhen Qiao3, Yan Liu3, Anthony E Getschman2, Sai L
Vankayala4, M Trent Kemp4, Francis C Peterson2, Rongshi Li3, Brian F Volkman*2, Yu Chen*1 Table of contents Page S2. Comparative binding energy analysis method Page S3. Comparative binding energy analysis results Page S5. Figure S1. Comparative binding energy analysis graph Page S6. Figure S2. Novel small molecule ligand discovery using rigid docking Page S7. Figure S3. Novel small molecule ligand discovery using ensemble docking Page S8. Figure S4. Novel small molecule ligands docking poses Page S9. Scheme S1. Page S10. Supporting Information References
S1
Comparative binding energy analysis method Prime MM-GBSA calculations in the Schrodinger suite (release 2015-4, Schrödinger, LLC, New York, NY, 2015) were used to compute the binding free energy for the CXCR4 N-terminal peptide when bound to CXCL12,1, 2 by using the NMR structure of constitutively monomeric CXCL12 (L55C/I58C) bound to the non-sulfated p40 CXCR4 peptide (PDB ID: 2N55), and then by using the monomer from the NMR structure of dimeric CXCL12 bound to the sulfated p38 CXCR4 peptide (PDB ID: 2K05).3 Both NMR complex structures were processed and minimized using the protein preparation wizard in the Schrodinger suite.1, 4, 5 Missing hydrogens were added to the structures, bond orders were corrected, disulfide bonds in close proximity were created, and the protonation states were assigned using PROPKA at a pH value of 7.6 The hydrogen-bonding network in the complexes was then optimized by reorienting amide groups of asparagine and glutamine residues, the imidazole ring of histidine residues, and hydroxyl and thiol groups. Restraint minimization was performed to relax strained bonds, angles and steric clashes, until the root mean square deviation (RMSD) reached a maximum cutoff of 0.30 Å. After the protein minimization phase, CXCL12 and the CXCR4 peptide in each case were separated into different entries to treat CXCL12 as the receptor, and the CXCR4 peptide as a ligand for probing the binding free energy. The VSGB 2.0 implicit solvation model4 of the Prime MM-GBSA module1, 2 was used during the MM-GBSA calculations and polar hydrogens were minimized in order to retain the experimental conformations. The binding free energy was decomposed into per-residue binding free energy contribution on the peptide using the Prime energy visualizer module in Schrodinger’s suite.2 MM-GBSA calculation: Ereceptor(minimized))
∆Gbind
=
Ecomplex(minimized)
−
(Eligand(minimized)
+
S2
Comparative binding energy analysis results. To gain additional insight into the properties of the I4/I6 binding site compared to other hotspots, and analyze its recognition epitopes on the CXCR4 peptide, we used Schrodinger Prime MM-GBSA1,
2
to calculate the energetic
contribution of each residue side chain on the receptor N-terminus to the PPI using the constitutively monomeric NMR CXCL12 (L55C/I58C) structure complexed to the non-sulfated N-terminal CXCR4 peptide (PDB ID: 2N55) (Fig. S1). Before the energetic contribution calculations, the experimentally determined structure was subjected to energy minimization as a complex and as individual proteins. The contribution of each residue side chain of the CXCR4 peptide to the binding energy was calculated considering interactions in both the complex state as well those in the free peptide. Based on these results, I4 is the residue contributing most to the binding of the non-sulfated N-terminus of CXCR4 to CXCL12 (Fig. S1), consistent with the observation that I4 makes the most contacts in the NMR complex structure. However, I6 seems to contribute little to the binding energy despite its favorable contacts with CXCL12 in the complex state. This is potentially due to the energetic cost associated with the loss of favorable interactions in the free peptide state. Surprisingly, the Y12 and Y21 residues of the CXCR4 peptide do not appear to contribute favorably to the binding interaction (Fig. S1). As shown by the FTMap analysis, and by the more favorable interactions between sY12/sY21 and these regions of the NMR complex of CXCL12 with the sulfated CXCR4 peptide (PDB ID: 2K05),3 the (s)Y12/(s)Y21 binding pockets may themselves assume configurations more suitable for ligand binding in the dimeric state. Therefore we also calculated the energetic contribution to the PPI by residues of the sulfated CXCR4 N-terminal peptide using one monomer of the dimeric NMR CXCL12 structure complexed to the sulfated N-terminal CXCR4 peptide (PDB ID: 2K05)
S3
and investigated whether the sulfation of Y12 and Y21 improved contributions to binding (Fig. S1). Based on the calculations, both sY12 and sY21 contribute more favorably to binding than their non-sulfated counterparts in the CXCR4 peptide in complex with the constitutively monomeric CXCL12 (L55C/I58C) (Fig. S1). However, while the two calculations agree qualitatively with previous experimental results demonstrating the importance of O-sulfation, they also underestimate the significance of the sulfotyrosine residues’ contributions to binding, and in particular for sY21. Close examination revealed structural differences between the protein conformations in the starting complex structure and the model post energy minimization, particularly for the CXCR4 peptide. Therefore it is possible that without the experimental restraints used for the modeling of the NMR structure the calculations were unable to accurately capture all of the interactions between CXCL12 and the CXCR4 peptide.
S4
Figure S1. Comparative binding energy analysis graph. Bar graph of residue-by-residue side chain binding energy contribution for both non-sulfated N-terminal CXCR4 peptide from monomeric CXCL12 complex (blue graph) (PDB ID: 2N55) and sulfated N-terminal CXCR4 peptide from dimeric CXCL12 complex (orange graph) (PDB ID: 2K05).
S5
0.6
1800
0.4
L 6 6
Y 6 1
I 5 1
K 5 6
N 4 6
L 3 6
R 4 1
O
T 3 1
0.2 0.0
L 2 6
O
0.8
A 2 1
O
S 1 6
O
1.0
S 6
OH
1.2
C 1 1
O
K 1
Compound 1 (ZINC C04181455) δ Chemical Shift (ppm)
A
PDB ID: 2N55 (conf 1)
Compound 2 (ZINC C40310216) O
OH
O
δ Chemical Shift (ppm)
B 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf 1)
AminoAcid Residue
Compound 3 (ZINC C16480049)
O
O
O
δ Chemical Shift (ppm)
C 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf 1)
Amino Acid Residue
Figure S2. Novel small molecule ligand discovery using rigid docking. 2D 1H-15N HMQC spectroscopy identified three compounds that bound to WT-CXCL12 (1, 2, and 3) from the compounds chosen via rigid docking virtual screening. Chemical shifts for each compound were assigned and mapped to monomeric CXCL12 (PDB ID: 2N55). A docking prediction for each compound is included showing overlap between perturbed residues and the binding pocket.
S6
A δ Chemical Shift (ppm)
Compound 4 (ZINC C44978491) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
I51
K56
N46
L36
R41
T31
L26
A21
S16
S6
C11
K1
0.0
PDB ID: 2N55 (conf-13)
Amino Acid Residue
B δ Chemical Shift (ppm)
Compound 5 (ZINC C69492022) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf-13)
Amino Acid Residue
C CH3
O
O HN
P HO
H3C
OH
OH CH3
δ Chemical Shift (ppm)
Compound 6 (ZINC C12998741) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
K1
S6
0.0 F
PDB ID: 2N55 (conf-13)
Amino Acid Residue
D
δ Chemical Shift (ppm)
Compound 7 (C15782120) 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
I51
K56
N46
R41
L36
T31
L26
A21
S16
S6
C11
K1
0.0
PDB ID: 2N55 (conf-5)
Amino Acid Residue
E δ Chemical Shift (ppm)
Compound 8 (ZINC C07362052) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf-16)
Amino Acid Residue
Figure S3. Novel small molecule ligand discovery using ensemble docking. 2D 1H-15N HMQC spectroscopy identified five compounds that bound WT-CXCL12 (4-8) from the compounds chosen via ensemble docking virtual screening. Chemical shifts for each compound were assigned and mapped to monomeric CXCL12 (PDB ID: 2N55). A docking prediction for each compound is included showing overlap between perturbed residues and the binding pocket.
S7
A W57 V23 R20
Y61
Compound 1
H25 K27
PDB ID: 2N55 (conf 1)
B
E60 A21
Compound 6
K43
K64
L66
V23
A65
N67
K24 H25 PDBID: 2N55 (conf 13)
C K24 I28 Compound 9
A40 C11
V39
PDB ID: 2K05 (conf 1, mon A)
Figure S4. Novel small molecule ligands docking poses. A. Docking pose of 1 in constitutive monomeric CXCL12 (PDB ID: 2N55, conformation 1) including mapping and labeling of significantly perturbed residues. B. Docking pose of 6 in constitutive monomeric CXCL12 (PDB ID: 2N55, conformation 13) including mapping and labeling of significantly perturbed residues. C. Docking pose of 9 in monomer A of constitutive dimeric CXCL12 (PDB ID: 2K05, conformation 1) including mapping and labeling of significantly perturbed residues.
S8
O
S N H
O N H
O
NaOH NH 2
HN S
10
O NH
Br
EtO
N
K 2CO3 / DMF 11
OEt
S N
N N
LiOH THF/H2O
O OEt
O 12
O
OH
S N
N N
O OH
O 9
O
Scheme S1. 2-((4-(carboxymethoxy)-6-phenyl-1,3,5-triazin-2-yl)thio)acetic acid (9) was prepared according to the reported procedure (Basyoumi, M. N.; El-Khamry, A.-M. Bull Chem Soc Jpn 1979, 50(1/2), 3728).
S9
Supporting Information References (1)Greenidge, P. A.; Kramer, C.; Mozziconacci, J. C.; Wolf, R. M. MM/GBSA binding energy prediction on the PDBbind data set: successes, failures, and directions for further improvement. J. Chem. Inf. Model. 2013, 53, 201-209. (2)Prime, version 4.2; Schrödinger, LLC: New York, 2015. (3)Veldkamp, C. T.; Seibert, C.; Peterson, F. C.; De la Cruz, N. B.; Haugner, J. C., 3rd; Basnet, H.; Sakmar, T. P.; Volkman, B. F. Structural basis of CXCR4 sulfotyrosine recognition by the chemokine SDF-1/CXCL12. Sci. Signal. 2008, 1, ra4. (4)Li, J.; Abel, R.; Zhu, K.; Cao, Y.; Zhao, S.; Friesner, R. A. The VSGB 2.0 model: a next generation energy model for high resolution protein structure modeling. Proteins 2011, 79, 27942812. (5)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, 221-234. (6)Li, H.; Robertson, A. D.; Jensen, J. H. Very fast empirical prediction and rationalization of protein pKa values. Proteins 2005, 61, 704-721.
S10
Structure-based identification of novel ligands targeting multiple sites within a chemokine- G protein–coupled receptor interface †
Emmanuel W Smith1, †Amanda M Nevins2, Zhen Qiao3, Yan Liu3, Anthony E Getschman2, Sai L
Vankayala4, M Trent Kemp4, Francis C Peterson2, Rongshi Li3, Brian F Volkman*2, Yu Chen*1 Table of contents Page S2. Comparative binding energy analysis method Page S3. Comparative binding energy analysis results Page S5. Figure S1. Comparative binding energy analysis graph Page S6. Figure S2. Novel small molecule ligand discovery using rigid docking Page S7. Figure S3. Novel small molecule ligand discovery using ensemble docking Page S8. Figure S4. Novel small molecule ligands docking poses Page S9. Scheme S1. Page S10. Supporting Information References
S1
Comparative binding energy analysis method Prime MM-GBSA calculations in the Schrodinger suite (release 2015-4, Schrödinger, LLC, New York, NY, 2015) were used to compute the binding free energy for the CXCR4 N-terminal peptide when bound to CXCL12,1, 2 by using the NMR structure of constitutively monomeric CXCL12 (L55C/I58C) bound to the non-sulfated p40 CXCR4 peptide (PDB ID: 2N55), and then by using the monomer from the NMR structure of dimeric CXCL12 bound to the sulfated p38 CXCR4 peptide (PDB ID: 2K05).3 Both NMR complex structures were processed and minimized using the protein preparation wizard in the Schrodinger suite.1, 4, 5 Missing hydrogens were added to the structures, bond orders were corrected, disulfide bonds in close proximity were created, and the protonation states were assigned using PROPKA at a pH value of 7.6 The hydrogen-bonding network in the complexes was then optimized by reorienting amide groups of asparagine and glutamine residues, the imidazole ring of histidine residues, and hydroxyl and thiol groups. Restraint minimization was performed to relax strained bonds, angles and steric clashes, until the root mean square deviation (RMSD) reached a maximum cutoff of 0.30 Å. After the protein minimization phase, CXCL12 and the CXCR4 peptide in each case were separated into different entries to treat CXCL12 as the receptor, and the CXCR4 peptide as a ligand for probing the binding free energy. The VSGB 2.0 implicit solvation model4 of the Prime MM-GBSA module1, 2 was used during the MM-GBSA calculations and polar hydrogens were minimized in order to retain the experimental conformations. The binding free energy was decomposed into per-residue binding free energy contribution on the peptide using the Prime energy visualizer module in Schrodinger’s suite.2 MM-GBSA calculation: Ereceptor(minimized))
∆Gbind
=
Ecomplex(minimized)
−
(Eligand(minimized)
+
S2
Comparative binding energy analysis results. To gain additional insight into the properties of the I4/I6 binding site compared to other hotspots, and analyze its recognition epitopes on the CXCR4 peptide, we used Schrodinger Prime MM-GBSA1,
2
to calculate the energetic
contribution of each residue side chain on the receptor N-terminus to the PPI using the constitutively monomeric NMR CXCL12 (L55C/I58C) structure complexed to the non-sulfated N-terminal CXCR4 peptide (PDB ID: 2N55) (Fig. S1). Before the energetic contribution calculations, the experimentally determined structure was subjected to energy minimization as a complex and as individual proteins. The contribution of each residue side chain of the CXCR4 peptide to the binding energy was calculated considering interactions in both the complex state as well those in the free peptide. Based on these results, I4 is the residue contributing most to the binding of the non-sulfated N-terminus of CXCR4 to CXCL12 (Fig. S1), consistent with the observation that I4 makes the most contacts in the NMR complex structure. However, I6 seems to contribute little to the binding energy despite its favorable contacts with CXCL12 in the complex state. This is potentially due to the energetic cost associated with the loss of favorable interactions in the free peptide state. Surprisingly, the Y12 and Y21 residues of the CXCR4 peptide do not appear to contribute favorably to the binding interaction (Fig. S1). As shown by the FTMap analysis, and by the more favorable interactions between sY12/sY21 and these regions of the NMR complex of CXCL12 with the sulfated CXCR4 peptide (PDB ID: 2K05),3 the (s)Y12/(s)Y21 binding pockets may themselves assume configurations more suitable for ligand binding in the dimeric state. Therefore we also calculated the energetic contribution to the PPI by residues of the sulfated CXCR4 N-terminal peptide using one monomer of the dimeric NMR CXCL12 structure complexed to the sulfated N-terminal CXCR4 peptide (PDB ID: 2K05)
S3
and investigated whether the sulfation of Y12 and Y21 improved contributions to binding (Fig. S1). Based on the calculations, both sY12 and sY21 contribute more favorably to binding than their non-sulfated counterparts in the CXCR4 peptide in complex with the constitutively monomeric CXCL12 (L55C/I58C) (Fig. S1). However, while the two calculations agree qualitatively with previous experimental results demonstrating the importance of O-sulfation, they also underestimate the significance of the sulfotyrosine residues’ contributions to binding, and in particular for sY21. Close examination revealed structural differences between the protein conformations in the starting complex structure and the model post energy minimization, particularly for the CXCR4 peptide. Therefore it is possible that without the experimental restraints used for the modeling of the NMR structure the calculations were unable to accurately capture all of the interactions between CXCL12 and the CXCR4 peptide.
S4
Figure S1. Comparative binding energy analysis graph. Bar graph of residue-by-residue side chain binding energy contribution for both non-sulfated N-terminal CXCR4 peptide from monomeric CXCL12 complex (blue graph) (PDB ID: 2N55) and sulfated N-terminal CXCR4 peptide from dimeric CXCL12 complex (orange graph) (PDB ID: 2K05).
S5
0.6
1800
0.4
L 6 6
Y 6 1
I 5 1
K 5 6
N 4 6
L 3 6
R 4 1
O
T 3 1
0.2 0.0
L 2 6
O
0.8
A 2 1
O
S 1 6
O
1.0
S 6
OH
1.2
C 1 1
O
K 1
Compound 1 (ZINC C04181455) δ Chemical Shift (ppm)
A
PDB ID: 2N55 (conf 1)
Compound 2 (ZINC C40310216) O
OH
O
δ Chemical Shift (ppm)
B 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf 1)
AminoAcid Residue
Compound 3 (ZINC C16480049)
O
O
O
δ Chemical Shift (ppm)
C 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf 1)
Amino Acid Residue
Figure S2. Novel small molecule ligand discovery using rigid docking. 2D 1H-15N HMQC spectroscopy identified three compounds that bound to WT-CXCL12 (1, 2, and 3) from the compounds chosen via rigid docking virtual screening. Chemical shifts for each compound were assigned and mapped to monomeric CXCL12 (PDB ID: 2N55). A docking prediction for each compound is included showing overlap between perturbed residues and the binding pocket.
S6
A δ Chemical Shift (ppm)
Compound 4 (ZINC C44978491) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
I51
K56
N46
L36
R41
T31
L26
A21
S16
S6
C11
K1
0.0
PDB ID: 2N55 (conf-13)
Amino Acid Residue
B δ Chemical Shift (ppm)
Compound 5 (ZINC C69492022) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf-13)
Amino Acid Residue
C CH3
O
O HN
P HO
H3C
OH
OH CH3
δ Chemical Shift (ppm)
Compound 6 (ZINC C12998741) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
K1
S6
0.0 F
PDB ID: 2N55 (conf-13)
Amino Acid Residue
D
δ Chemical Shift (ppm)
Compound 7 (C15782120) 1.2 1.0 0.8 0.6
1800
0.4 0.2 L66
Y61
I51
K56
N46
R41
L36
T31
L26
A21
S16
S6
C11
K1
0.0
PDB ID: 2N55 (conf-5)
Amino Acid Residue
E δ Chemical Shift (ppm)
Compound 8 (ZINC C07362052) 1.2 1.0 0.8
1800
0.6 0.4 0.2 L66
Y61
K56
I51
N46
R41
L36
T31
L26
A21
S16
C11
S6
K1
0.0
PDB ID: 2N55 (conf-16)
Amino Acid Residue
Figure S3. Novel small molecule ligand discovery using ensemble docking. 2D 1H-15N HMQC spectroscopy identified five compounds that bound WT-CXCL12 (4-8) from the compounds chosen via ensemble docking virtual screening. Chemical shifts for each compound were assigned and mapped to monomeric CXCL12 (PDB ID: 2N55). A docking prediction for each compound is included showing overlap between perturbed residues and the binding pocket.
S7
A W57 V23 R20
Y61
Compound 1
H25 K27
PDB ID: 2N55 (conf 1)
B
E60 A21
Compound 6
K43
K64
L66
V23
A65
N67
K24 H25 PDBID: 2N55 (conf 13)
C K24 I28 Compound 9
A40 C11
V39
PDB ID: 2K05 (conf 1, mon A)
Figure S4. Novel small molecule ligands docking poses. A. Docking pose of 1 in constitutive monomeric CXCL12 (PDB ID: 2N55, conformation 1) including mapping and labeling of significantly perturbed residues. B. Docking pose of 6 in constitutive monomeric CXCL12 (PDB ID: 2N55, conformation 13) including mapping and labeling of significantly perturbed residues. C. Docking pose of 9 in monomer A of constitutive dimeric CXCL12 (PDB ID: 2K05, conformation 1) including mapping and labeling of significantly perturbed residues.
S8
O
S N H
O N H
O
NaOH NH 2
HN S
10
O NH
Br
EtO
N
K 2CO3 / DMF 11
OEt
S N
N N
LiOH THF/H2O
O OEt
O 12
O
OH
S N
N N
O OH
O 9
O
Scheme S1. 2-((4-(carboxymethoxy)-6-phenyl-1,3,5-triazin-2-yl)thio)acetic acid (9) was prepared according to the reported procedure (Basyoumi, M. N.; El-Khamry, A.-M. Bull Chem Soc Jpn 1979, 50(1/2), 3728).
S9
Supporting Information References (1)Greenidge, P. A.; Kramer, C.; Mozziconacci, J. C.; Wolf, R. M. MM/GBSA binding energy prediction on the PDBbind data set: successes, failures, and directions for further improvement. J. Chem. Inf. Model. 2013, 53, 201-209. (2)Prime, version 4.2; Schrödinger, LLC: New York, 2015. (3)Veldkamp, C. T.; Seibert, C.; Peterson, F. C.; De la Cruz, N. B.; Haugner, J. C., 3rd; Basnet, H.; Sakmar, T. P.; Volkman, B. F. Structural basis of CXCR4 sulfotyrosine recognition by the chemokine SDF-1/CXCL12. Sci. Signal. 2008, 1, ra4. (4)Li, J.; Abel, R.; Zhu, K.; Cao, Y.; Zhao, S.; Friesner, R. A. The VSGB 2.0 model: a next generation energy model for high resolution protein structure modeling. Proteins 2011, 79, 27942812. (5)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, 221-234. (6)Li, H.; Robertson, A. D.; Jensen, J. H. Very fast empirical prediction and rationalization of protein pKa values. Proteins 2005, 61, 704-721.
S10
