Method

Homology modeling

The homo sapiens active μ opioid receptor was built by taking the X-ray structure of musculus active μ opioid receptor (PDB code: 5C1M) as a template by using the homology modeling module of Discovery Studio 3.5 software(15 ). All water molecules in the X-ray structure were retained. Ramachandran plot was employed to evaluate the validity of the homology models in Figure S1 . Residues were numbered according to the generalized numbering scheme proposed by Ballesteros and Weinstein(19 ).

Molecular docking

The morphine, fentanyl and sufentanil were docked into the 3D structure of homo sapiens active μ opioid receptor. We used the Glide Docking module in Maestro 3.5 program to dock these compounds into the binding site of the active μ opioid receptor. These three systems were subjected to Monte Carlo Multiple Minimum conformational searches using the OPLS_2005 force field. The minimized conformations were selected for the next molecular dynamics simulations.

Molecular dynamic simulations

Eight systems were built for molecular dynamics simulations: (1) μ opioid receptor without ligand (Apo); (2) morphine-bound active MOR; (3) sufentanil-bound active MOR; and (4) fentanyl-bound active MOR; (5) mutant μ opioid receptor without ligand (Apo-W6.48L); (6) morphine-bound active mutant MOR (W6.48L); (7) sentinel-bound active mutant MOR (W6.48L); and (8) fentanyl-bound mutant MOR (W6.48L). MD simulations were performed using the Gromacs5.1.2 package. All systems were embedded in a hydrated POPC lipid bilayer. Water molecules were used to solvate the protein. After that, Na+ and Cl-ions were added in the water to neutralize the system using 0.15 mol/L NaCl. The steepest descent followed by conjugate gradient methods were used for the energy minimization for these systems. Then, we gradually heated the systems from 0 K to 310 K. The systems were subjected to equilibrate at constant pressure and temperature for 1 ns (310 K, 1 atm). Finally, the production MD simulations of eight systems were performed for 100 ns. All analyses of MD trajectories were performed using the tools implemented in the Gromacs5.1.2 package.

Plasmid construction and site-directed mutagenesis

To make wild-type and mutant (WT & mutant) MOR plasmids, full-length hOPRM1 cDNA was subcloned into the Tag-lite® pT8-SNAP vector (Cisbio, France). The Vazyme® Fast Mutagenesis Kit V2 (Vazyme, China)was used to introduce the D3.32A, W6.48L, H6.52L site mutation to the MOR recombinant plasmid, the nucleotide sequences of mutant MOR were confirmed by DNA sequencing and sequence alignment. Primers designed to amplify MOR plasmid containing mutant site are listed here:
D3.32A
F 5’-TCCATAGC TTACTATAACATGTTCACCAGC-3’
R 5’-ATAGTAAG CTATGGAGATCACTATCTTGCA-3’
W6.48L
F 5’-GTCTGCTT GACTCCCATTCACATTTACGTC-3’
R 5’-GGGAGTCA AGCAGACGATGAACACAGCCAC-3’
H6.52L
F 5’-CCCATTCT CATTTACGTCATCATTAAAGCC-3’
R 5’-GTAAATGA GAATGGGAGTCCAGCAGACGAT-3’

Cell lines

Chinese Hamster Ovary (CHO-K1) cells were purchased from ATCC and maintained in F12 medium containing 10% fetal bovine serum (FBS).
To obtain CHO stably expressing MOR (WT & mutant) cell lines, wild-type and D3.32A, W6.48L, H6.52L mutant MOR plasmids were respectively transfected into CHO-K1 cells using lipofectamine® LTX regent. The transfected cell mixtures were grown under the selection pressure of 200 μg/ml hygromycin for 10 days, then cells were seeded in 96-well plates at an approximate density of 1 cell per well to isolate clones. An appropriate clone was selected for further experimental studies. The MOR expression of CHO-MOR cell lines was confirmed by western-blot analysis (Figure S3A ). To verify the site mutation of each CHO-MOR cell line, we amplified the target MOR fragments from total genomic DNA (Figure S3B ). Then DNA sequencing analysis of the PCR products verified the correct site mutation of MOR in CHO-MOR cell lines. Membrane MOR expression of the CHO-MOR cell lines was measured by immunofluorescence analysis (Figure S3D ). The expression level of wild-type and D3.32A MOR on membranes was confirmed to be similar, and the W6.48L and H6.52L mutant MOR is approximately 63% of wild-type. CHO-MOR (WT & mutant) cells were maintained in F12 medium supplemented with 10% FBS and 40 μg/ml hygromycin.
To obtain CHO-stably expressing both MOR (WT & mutant) and β-arrestin2-EGFP cell lines, we transfected the β-arrestin2-EGFP plasmid into each CHO-MOR (WT & mutant) cell line. After growing under the selection pressure of 6 μg/ml puromycin for 10 days, cell mixtures were seeded in 96-well plates at an approximate density of 1 cell per well to isolate clones. An appropriate clone was selected for further experiments. The expression of β-arrestin2 was verified by western-blot analysis and the expression level of each cell line was confirmed to be similar (Figure S3C ). CHO-MOR (WT & mutant)-β-arrestin2-EGFP cells were maintained in F12 medium supplemented with 10% FBS and 40 μg/ml hygromycin and 2 μg/ml puromycin.
All cells were cultured according to the standard protocol in 37 ℃ incubator with 5% CO2.

Western-blot

Adherent CHO cells expressing MOR (WT & mutant) or β-arrestin2 were lysed in RIPA lysis buffer (50 mM pH 7.4 Tris, 150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitor cocktail and phosphatase inhibitor. The lysate was incubated at 37 ℃ for 10 min, then separated by SDS-PAGE of 8% polyacrylamide gels and transferred to PVDF membranes. Membranes were immunoblotted with anti-MOR antibody (Millipore, AB1580-Ⅰ) to verify the expression of MOR or immunoblotted with anti-β-arrestin 2 antibody (Cell signaling technology, 3857) to verify the expression of β-arrestin 2. To ensure equal loading into each lane, blots were also incubated with GAPDH antibody.

Immunofluorescence

Membrane MOR can be labelled with SNAP-Lumi4-Tb, and MOR-Tb level could be quantified by Envision Multilabel Reader at 615 nm. CHO-MOR (WT & mutant) cells were seeded on 35 mm dishes and grown to 80% confluence. Cells were incubated with SNAP-Lumi4-Tb solution (Cisbio, France) for 1 h. Then cells were digested and dispensed into 384-well plate at a density of 20000 cells per well, and the total test volume is 20 μl. Read the plate on Envision Multilabel Reader (Perkin Elmer) at 615 nm. The response represents the expression level of MOR.

Genomic DNA PCR

CHO-MOR (WT & mutant) cells were seeded on 35 mm dishes and grown to 80% confluence. Genomic DNA was extracted from cells using the Gentra pure gene cell kit (Qiagen, Germany). MOR DNA fragments were amplified from genomic DNA by PCR primers F: 5’-CTCGCCGTGAAAGAGTGGCT-3’, R: 5’-GGGCAACGGAGCAGTTTCT-3’. Then the PCR products were sequenced to verify the mutant site.

Radioligand saturation binding assay

Radioligand saturation binding assay was performed to determine the affinity of fentanyl in binding with MOR. Membrane proteins were extracted from CHO cells expressing MOR (WT & mutant), and 20 μg proteins were used for each reaction. In specific binding reaction, membranes were incubated with a series of concentrations of [3H]Fentanyl (ranging from 1.5625 nM to 50 nM) for 30 min at 37 ℃ in Tris-HCl buffer (50 mM Tris, pH 7.4). In nonspecific binding reaction, membranes were incubated with 5 μM naloxone and [3H]Fentanyl (ranging from 1.5625 nM to 50 nM) for 30 min at 37 ℃ in Tris-HCl buffer. The reaction mixtures were filtered over GF/C filters, and then the filters were washed three times by cold Tris-HCl buffer. Radioactivity was assayed by liquid scintillation counting overnight.

HTRF competitive binding assay

HTRF competitive binding assay was performed according to the Tag-lite® binding assay recommended protocol (Cisbio, France). This assay is based on the competition between the Tag-lite® fluorescent ligand and test compounds. CHO-MOR (WT & mutant) cells were labeled in batch with SNAP-Lumi4-Tb and suspended in Tag-lite® labeling buffer (TLB), then cells dispensed into 384 well small volume white plate (Thermo Scientific Nunc, USA) at a density of 5000 cells per well. To determine saturation binding constant Kd of the fluorescent ligand, cells were incubated with a series of concentrations (ranging from 0.1 nM to 100 nM) of the fluorescent ligand in TLB. Non-specific binding signal wells were incubated with 100 nM Naltrindole. For competition binding experiments, cells were incubated with varying concentrations (ranging from 10-4 M to 10-14 M) of test compounds in the presence of 8 nM fluorescent ligand. The total binding reaction volume is 20 μl, and the concentration mentioned above means final concentration. Read the plate on Envision Multilabel Reader (Perkin Elmer) at 665 nm and 615 nm after 3 h incubation. The HTFR ratio was calculated as the equation:
\begin{equation} \begin{matrix}HTRF\ Ratio=\frac{Signal\ 665\ nm}{Signal\ 615\ nm}\times 10^{4}\#Equation\ 1\\ \end{matrix}\nonumber \\ \end{equation}

cAMP assay

The cAMP assay was performed according to the cAMP-Gi kit protocol (Cisbio, France). CHO-MOR (WT & mutant) cells were harvested and suspended in stimulation buffer supplemented with 0.5 mM IBMX. Then cells were dispensed into 384-well small volume white plate (Thermo Scientific Nunc, USA) at a density of 5000 cells per well. Add a series of concentrations (ranging from 10-12 M to 10-6 M) of test compounds and 3 μM forskolin into wells (non-stimulated control wells add stimulation buffer instead of test compounds at this step), seal the plate and incubate at room temperature (RT) for 15 min. Add cAMP-Cryptate and anti-cAMP-d2 working solution into wells, seal the plate and incubate at RT for 1 h. The total binding reaction volume is 20 μl, and the concentration mentioned above means final concentration. Read the plate on Envision Multilabel Reader (Perkin Elmer) at 665 nm and 615 nm after 3 h incubation. The response was calculated as Equation 1 .

β-arrestin2 recruitment assay

CHO-MOR (WT & mutant)-βarrestin2 cells were seeded into 96-well, black-walled, clear-bottom assay plate (Costar, USA) at an appropriate density to ensure they will be 50%~70% confluent next day. For the assay, cells were serum-starved for 1 h, then treated with a series of concentrations (ranging from 10-12 M to 10-5 M) of test compounds for 5 min at 37 ℃. Cells were fixed and dyed using 4% paraformaldehyde (PFA) with Hoechst nuclear stain (1:10000) for 30 min. β-arrestin2 translocation images were captured with a 20× objective on an ArrayScan™ XTI High Content Analysis (HCA) Reader (Thermo Scientific, USA). The spots/nuclear ratio per well was quantified using Cellomics Spot® Detection BioApplication (Thermo Scientific, USA).

Data analysis and statistical procedures

GraphPad Prism (V7.0) software was used for curve fitting and data analysis. All data are shown as mean ± SEM of at least three independent experiments run in duplicate or triplicate.
For binding assays, concentration-response curves were fit to one-site binding models provided in Prism software to determine Kd, Bmax, and Ki.
For functional assays, data are shown by subtracting basal values and presented as percentage of DAMGO, then were fit to a non-linear regression (three-parameter) model to determine EC50 and Emax. As indicated, the difference between all the fitting parameters of wild-type MOR versus W6.48L mutant MOR was acquired via unpaired, two-tailed t-test.
We used the classic operational model analysis method to calculate ligand bias. The operational model (Black and Leff, 1983)(20 ) was applied to calculate bias as described by van der Westhuizen et al. (2014) (21 )and Laura M. Bohn et al. (2017)(22 ). The transduction ratio (τ/KA) was determined based on the equation:
\begin{equation} \begin{matrix}E=\frac{E_{\max}}{1+\left(\frac{1+\frac{A}{10^{\log\left(K_{A}\right)}}}{A\times 10^{\log\left(\frac{\tau}{K_{A}}\right)}}\right)^{n}}\#Equation\ 2\\ \end{matrix}\nonumber \\ \end{equation}
Where E is the effect of ligand, Emax is the maximal response, A is the molar concentration of the ligand, KAis the equilibrium dissociation constant, and the log(τ/KA) is the transduction coefficient(21 ,23 ). For each assay, the Emax is constrained to be a shared value, the basal is constrained to be the value of zero and n is constrained to the value of 1.
DAMGO was chosen as the reference ligand. To eliminate system deviation, the Δ log (τ/KA) of test ligand and SEM were calculated by Equation 3 and Equation 4 :
\begin{equation} \begin{matrix}\log\left(\frac{\tau}{K_{A}}\right)={\log\left(\frac{\tau}{K_{A}}\right)}_{\text{test}}-{\log\left(\frac{\tau}{K_{A}}\right)}_{\text{DAMGO}}\#\text{Equation\ }3\\ \end{matrix}\nonumber \\ \end{equation}\begin{equation} \begin{matrix}\text{SE}_{\left(\text{Δ\ log}\left(\frac{\tau}{K_{A}}\right)\right)}=\sqrt{\left(\text{SE}_{\left(\log\left(\frac{\tau}{K_{A}}\right)\right)_{\text{test}}}\right)^{2}+\left(\text{SE}_{\left(\log\left(\frac{\tau}{K_{A}}\right)\right)_{\text{DAMGO}}}\right)^{2}}\#Equation\ 4\\ \end{matrix}\nonumber \\ \end{equation}
The bias parameter ΔΔ log (τ/KA) and SEM were calculated by Equation 5 and Equation 6 :
\begin{equation} \begin{matrix}\ {\log\left(\frac{\tau}{K_{A}}\right)}_{\frac{\text{cAMP}}{\beta arr2}}={\operatorname{log}\left(\frac{\tau}{K_{A}}\right)}_{\text{cAMP}}-{\ \log\left(\frac{\tau}{K_{A}}\right)}_{\beta arr2}\#\text{Equation\ }5\\ \end{matrix}\nonumber \\ \end{equation}\begin{equation} \begin{matrix}\text{SE}_{\left(\text{ΔΔ\ log}\left(\frac{\tau}{K_{A}}\right)\right)}=\sqrt{\left(\text{SE}_{\left(\text{Δ\ log}\left(\frac{\tau}{K_{A}}\right)\right)_{test:cAMP}}\right)^{2}+\left(\text{SE}_{\left(\text{Δ\ log}\left(\frac{\tau}{K_{A}}\right)\right)_{test:\beta arr2}}\right)^{2}}\#\text{Equation\ }6\\ \end{matrix}\nonumber \\ \end{equation}

Materials

[3H]Fentanyl was purchased from Perkin Elmer (USA). DAMGO and Naltrindole were purchased from Tocris (UK). Forskolin and 3-isobutyl-1-methylxanthine (IBMX) were purchased from Sigma-Aldrich (USA). Morphine was purchased from QINHAI pharmaceutical factory (China). Fentanyl derivatives were synthesized by our institute. Cell culture reagents were purchased from Gibco (USA).