Cys170 is the palmitoylation site of OPRM1
We used wild-type HEK293 cells (HEK) and HEKOPRM1 cells (HEK cells heterologously expressing OPRM1 with HA spliced at the amino terminus) to validate the palmitoylation assay [13
]. HA-tagged receptors were precipitated with HA antibody and protein G agarose. The following procedures were used to determine receptor palmitoylation: 1) Free sulfhydryl groups in precipitated receptors were blocked with N-ethylmaleimide (NEM). 2) Palmitoylated cysteines were hydrolyzed with hydroxylamine. 3) Biotin was conjugated to the de-palmitoylated cysteines in the immunoprecipitated receptors with btn-BMCC (1-biotinamido-4- [4'- (maleimidomethyl) cyclohexane carboxamido] butane). The amount of biotin linked to the receptors was determined by immunoblotting.
We observed intensive biotin labeling in the HEKOPRM1 cells but not in the HEK cells, which suggested that the palmitoylation we detected was specific to OPRM1 (Figure , Lanes 1-2). Since NEM was used to block free sulfhydryl groups, the immunoreactivity of biotin increased when the NEM step was omitted (Figure , Lane 4). In addition, the immunoreactivity of biotin decreased when the hydroxylamine step was omitted (Figure , Lanes 5) and no biotin was detected when btn-BMCC was not used (Figure , Lanes 6), which confirmed that the assay was suitable for detecting the palmitoylation of OPRM1. To further confirm that the assay could be used to detect palmitoylation specifically, we used 2-bromopalmitate (2-BP), a palmitoylation inhibitor, to block all palmitoylation. A low level of receptor palmitoylation was observed when HEKOPRM1 cells were pretreated with 2-BP for 12 h (Figure , Lane 3), which also indicated that the palmitoylation we detected was on OPRM1. Since palmitoylation is important for cell function, prolonged treatment with or high concentrations of 2-BP will affect cell viability. In the current study, the treatment time and concentration of 2-BP were determined empirically so as to inhibit receptor palmitoylation while having minimal effect on cell viability. Thus, the 2-BP treatment paradigm we used did not completely block receptor palmitoylation.
Figure 1 Cys170 is the palmitoylation site of OPRM1. (a) Palmitoylation assays were performed in HEK and HEKOPRM1 cells. The amounts of palmitoylated receptor were normalized against that in HEKOPRM1 (Lane 2). 50 μM 2-BP was used to treat HEKOPRM1 for (more ...)
Mutation of the two conserved cysteine residues (C8.53 and C8.58) in the C terminus of OPRM1 does not affect palmitoylation [3
]. Thus, we predicted the only other intracellular cysteine, C3.55(170), to be the putative palmitoylation site. To confirm this hypothesis, each of these three cysteines was mutated to alanine individually, and the mutants were stably expressed in HEK cells to obtain the following: HEKC170A, HEKC346A, and HEKC351A. Although C170A is difficult to stably express in CHO cells [3
], we were able to successfully express a high level of C170A in the cell membrane of HEK cells, possibly because of differences between cell lines or our use of poly-L-lysine during cell culture. As shown in Figure , Lanes 2 and 5-6, we detected similar amounts of palmitoylated receptor in HEKOPRM1, HEKC346A, and HEKC351A cells. Furthermore, the amount of palmitoylated receptor in HEKC170A cells was similar to that in 2-BP-pretreated HEKOPRM1 cells (Figure ). These results suggest that C3.55(170) but not C8.53(346) or C8.58(351) is indeed the palmitoylation site of OPRM1.
Our subsequent [3H]-diprenorphine saturation binding assay using isolated cell membranes indicated that there was no difference in the amounts of receptors in the cell membranes of the HEKOPRM1, HEKC170A, HEKC346A, and HEKC351A cells (membrane receptors were expressed in HEK, HEKOPRM1, HEKC170A, HEKC346A, and HEKC351A cells at 0.06 ± 0.05, 6.87 ± 1.14, 7.36 ± 1.10, 5.12 ± 1.67, and 6.80 ± 1.61 pmol/mg protein respectively [Figure ]). Our FACS analysis using an antibody against the HA-tag further confirmed that there was no difference in the amounts of membrane receptor between the four cell lines (Figure ). Since the HA-tag was located in the receptor N-terminus, and the cell membrane was not disrupted during the analysis, the results obtained from the FACS assay should represent the actual amounts of membrane receptor. Lastly, immunoblotting also indicated that the overall receptor expression levels were similar between the four cell lines (Figure ). In light of these results, it is reasonable to conclude that C3.55(170) is the major palmitoylation site of OPRM1. For the sake of consistency, we will now use "OPRM1" to refer to the wild-type OPRM1, "C170A" to refer to the palmitoylation-deficient mutant, and "receptor" to indicate both the wild type and mutants.
Of note, agonist treatment did not affect receptor palmitoylation when morphine and fentanyl were used to challenge the HEKOPRM1 cells (Figure ). Since the studies described here focused on how receptor palmitoylation influences receptor signaling, the effects of agonists on receptor palmitoylation or other subsequent observations are not discussed in depth.
Receptor palmitoylation stabilizes morphine-induced signaling and receptor-Gαi2 coupling
To determine the influence of palmitoylation on receptor signaling, we monitored morphine-induced adenylyl cyclase inhibition and ERK phosphorylation. Morphine-induced adenylyl cyclase inhibition is defined by the ability of morphine to inhibit the forskolin-induced increase in the intracellular cAMP level. We analyzed morphine-induced ERK phosphorylation by calculating the percentage increase of phosphorylated ERK when compared to basal level.
As summarized in Table , no difference in the affinities for ligands (morphine, naloxone, and Cys2-Tyr3-Orn5-Pen7-amide [CTOP]) was detected between OPRM1 and C170A. For example, the KI of CTOP was 11 ± 1.4 nM in HEKOPRM1 cells, and was 9.5 ± 2.1 nM in HEKC170A cells (Table ). In addition, the expression levels of OPRM1 and C170A in the cell membrane were similar (Figure ). However, morphine induced less signaling in HEKC170A than in HEKOPRM1 cells. The ability of morphine to induce adenylyl cyclase inhibition in HEKC170A was approximately 75% of that in HEKOPRM1, when maximum inhibition was analyzed (Table ). The ability to induce ERK phosphorylation in HEKC170A was approximately 69% of that in HEKOPRM1 (Table and Figure ). Morphine also induced receptor signaling in 2-BP-treated HEKOPRM1 cells and in HEKC170A cells (Table and Figure ). These results suggest that palmitoylation blockage impairs receptor signaling induced by morphine.
Palmitoylation does not affect the binding of agonists.
Palmitoylation impairs morphine-induced receptor signaling.
Both of the signaling events that we monitored are mediated via Gαi2 [12
]. In addition, up-regulation and down-regulation of Gαi2 in HEKOPRM1 cells significantly affects adenylyl cyclase inhibition and ERK phosphorylation induced by morphine. Thus, we thought it likely that the impaired signaling described above was indicative of a decrease in receptor-Gαi2 coupling. We next investigated the effect of receptor palmitoylation on Gαi2 coupling.
When we analyzed the immunoreactivity of OPRM1 and Gαi2 on the cell membrane (Figure ), the colocalization between receptor and Gαi2 in HEKOPRM1 cells was more obvious than in HEKC170A cells. Similar observations were noted in our co-immunoprecipitation experiments (Figure and ). When we used a Gαi2 antibody to perform co-immunoprecipitation, the amount of OPRM1 co-immunoprecipitated with Gαi2 was greater than that of C170A. When HA antibody was used to immunoprecipitate the receptor, more Gαi2 was co-immunoprecipitated with OPRM1 than with C170A. These results indicate that the interaction between Gαi2 and C170A is weaker than that between Gαi2 and OPRM1.
Figure 2 Palmitoylation contributes to Gαi2 coupling. (a) The colocalization between HA-tagged receptor and Gαi2 was determined in HEKOPRM1 and HEKC170A cells. Images were analyzed as described in Methods. (b) Anti-HA antibody was used to precipitate (more ...)
We further investigated the interaction between receptor and Gαi2 with FRET analysis. The normalized net FRET between CFPOPRM1 and YFPGαi2 was much higher than that between CFP and YFP, suggesting that OPRM1 and Gαi2 were in close proximity of each other (≤ 10 nm). We performed FRET analysis on the cell membrane to exclude any possible influence from the intracellular expression of fluorescence constructs. The normalized net FRET between CFPOPRM1 and YFPGαi2 was higher than that between CFPC170A and YFPGαi2 (Figure ). Because 1) we kept the expression of the fluorescence constructs, like CFPOPRMA1 and YFPGαi2 to similar levels by titrating the amounts of plasmids used for transfection, 2) we used immunoblotting to monitor expression during our studies, and 3) we determined overall fluorescence intensities prior to our FRET and colocalization studies, the FRET difference supports the conclusion that blockage of receptor palmitoylation in the C170A mutant impairs Gαi2 coupling.
The YFP/CFP tagged receptors had similar functions when either FLAG- or HA-tagged. Morphine-induced adenylyl cyclase inhibition in the cells caused transient expression of these receptor constructs with similar KIs: 9.8 ± 1.1 nM (HA-tagged OPRM1), 10.7 ± 1.4 nM (FLAG-tagged OPRM1), 8.9 ± 1.2 nM (CFPOPRM1), and 9.5 ± 0.8 nM (YFPOPRM1). Thus, the FRET experiments should be indicative of the functional characteristics of the receptors.
Receptor palmitoylation facilitates homodimerization and subsequent Gαi2 coupling
We investigated the possible contribution of OPRM1 palmitoylation to homodimerization by performing FRET analysis between CFPOPRM1/CFPC170A and YFPOPRM1/YFPC170A. As indicated in Figure , the normalized net FRET between CFPOPRM1 and YFPOPRM1 was 0.49 ± 0.03, whereas it was 0.07 ± 0.02 between CFPC170A and YFPC170A in the cell membrane. In addition, when the HEK cells were co-transfected with CFPOPRM1 and YFPC170A or with CFPC170A and YFPOPRM1, the normalized net FRETs were 0.30 ± 0.05 and 0.27 ± 0.05, respectively. The colocalization and co-immunoprecipitation assays between HAOPRM1/HAC170A and FLAGOPRM1/FLAGC170A confirmed the results of the FRET assay (Figure and ). In summary, our colocalization, co-immunoprecipitation and FRET studies suggest that the amount of the OPRM1-OPRM1 homodimer is greater than the amount of the OPRM1-C170A dimer, and the amount of the OPRM1-C170A dimer is greater than that of the C170A-C170A homodimer, when similar levels of receptors are expressed. It is reasonable, therefore, to suggest that the ability of C170A to form a homodimer is lower than that of OPRM1.
Figure 3 Palmitoylation stabilizes homodimerization. (a) FRET analysis was performed after transfecting combinations of CFPOPRM1/CFPC170A and YFPOPRM1/YFPC170A into HEK cells. (b-c) FLAGOPRM1/FLAGC170A and HAOPRM1/HAC170A were transfected into HEK cells. The colocalization (more ...)
Because the amounts of homodimer decreased sequentially from Lane 1 to Lane 4 in Figure , we used FRET analysis to determine if the decrease affected receptor-Gαi2 coupling (Figure ). We transiently transfected YFPGαi2 with either CFPOPRM1 or CFPC170A into HEKOPRM1 and HEKC170A cells. We considered two caveats in these experiments and took the following steps to ensure the success of the studies: 1) we determined that HEKOPRM1 and HEKC170A cells expressed similar amounts of membrane receptors; and 2) we tightly controlled transient transfection of CFPOPRM1 and CFPC170A in order to reach similar expression levels.
Figure 4 Palmitoylation stabilizes Gαi2 coupling. (a) The FRET between CFPOPRM1 and YFPGαi2 and the FRET between CFPC170A and YFPGαi2 were determined in HEKOPRM1 and HEKC170A cells. (b-c) Colocalization between FLAGOPRM1 and Gαi2 (more ...)
According to our hypothesis, if receptor palmitoylation affects Gαi2 coupling, a similar sequential decrease in receptor-Gαi2 coupling should be observed between OPRM1 homodimer, OPRM1-C170A dimer and C170A homodimer. As indicated in Figure , we observed that the normalized net FRET between CFPOPRM1 and YFPGαi2 was greater than that between CFPC170A and YFPGαi2 in both HEKOPRM1 and HEKC170A cells. The normalized net FRET between CFPOPRM1 and YFPGαi2, as well as between CFPC170A and YFPGαi2, was greater in HEKOPRM1 than in HEKC170A. These results suggest a positive correlation between the receptor palmitoylation and Gαi2 coupling.
This correlation can be explained by two potential mechanisms. One possibility is that the homodimer's affinity for Gαi2 is much higher than the monomer's affinity for Gαi2; this mechanism is supported by a previous report [9
]. A second possibility is that the C170A monomer's affinity for Gαi2 is much lower than the OPRM1 monomer's affinity for Gαi2. If the second mechanism was the dominant one, the FRET between transiently transfected CFPC170A and YFPGαi2 would be smaller in HEKOPRM1 cells than in HEKC170A cells, because OPRM1's higher affinity for YFPGαi2 would result in a higher competition for Gαi2 in HEKOPRM1 than in HEKC170A. However, our FRET analysis produced the opposite result: the FRET between CFPC170A and YFPGαi2 was higher in HEKOPRM1 cells than in HEKC170A cells (Figure ). These observations suggest that the reduced receptor dimerization in the absence of palmitoylation leads to decreased Gαi2 coupling. Our additional colocalization and co-immunoprecipitation studies further supported this hypothesis (Figure and ). In total, these results indicate a correlation between receptor homodimerization and Gαi2 coupling.
Receptor palmitoylation facilitates cholesterol association in the receptor signaling complex
In order to determine the detailed mechanisms underlying these phenomena, we utilized the observed interaction between cholesterol and palmitoyl group in the crystal structure of β2
]. Before we could determine the existence of a similar cholesterol-palmitoyl interaction in the OPRM1 complex, however, we first needed to quantify the existing cholesterol in the receptor complex. Because direct detection of cholesterol within the homodimer requires purification of the receptor to homogeneity, and there is no guarantee that the cholesterol-receptor association will stay intact during purification, we instead examined the amount of cholesterol incorporated into the receptor signaling complex using a new method, described below.
To determine cholesterol association with the receptor complex, we used HA-antibody to precipitate the HA-tagged receptor. In this method, if cholesterol does associate with the receptor complex specifically, greater amounts of cholesterol should be precipitated by the HA antibody when compared to immunoprecipitation with no antibody. To avoid possible influence from the usage of antibody, FLAG antibody was used as control antibody, as no protein was FLAG-tagged in current paradigm. Cholesterol association with the receptor signaling complex was indicated by the additional amount of cholesterol precipitated by the HA antibody compared with that precipitated by a control antibody. Extensive washing with lysis buffer containing Triton X-100 and digitonin ensured the removal of cholesterol that was nonspecifically associated with the receptor signaling complex.
Using this procedure, HA antibody precipitated more cholesterol in HEKOPRM1 cells than did FLAG antibody or PBS. Since the receptor was HA-tagged at the N-terminus, any detected cholesterol in the precipitated receptors could be identified as cholesterol associated with receptor signaling complex. Further, in control experiments using HEK cells, the two antibodies and PBS precipitated similar amounts of cholesterol (Figure ). These results confirm that this assay detects cholesterol associated with the receptor signaling complex.
Figure 5 Palmitoylation facilitates cholesterol association. (a) Cholesterol associated with receptor complex was determined in HEK, HEKOPRM1 and HEKC170A cells. The amount of cholesterol precipitated with PBS in HEK cells was used for normalization. (b-c) HEKOPRM1 (more ...)
We also noted decreased cholesterol association in HEKC170A cells (Figure ). The amount of cholesterol precipitated with the HA antibody was similar to that precipitated with the FLAG antibody, suggesting that the mutation on C3.55(170) contributes to the cholesterol association. Although the assay could not distinguish between cholesterol that associates with the receptor directly and cholesterol that associates with other proteins within the signaling complex, receptor palmitoylation appears to regulate the amount of cholesterol that associates with the complex.
Cholesterol association facilitates homodimerization and Gαi2 coupling
Because 1) a cholesterol-palmitoyl interaction has been suggested in the β2-AR crystal structure, and 2) we demonstrated that receptor palmitoylation facilitates receptor dimerization and G protein coupling of OPRM1 (Figure to Figure ), we sought to discover whether cholesterol has the same functions in the latter receptor. To determine the contribution of cholesterol association to receptor signaling, we treated the cells with simvastatin, an HMG-CoA reductase inhibitor. We assayed receptor dimerization and G protein coupling with FRET, colocalization and immunoprecipitation.
The cellular cholesterol content decreased on the cell membrane of HEKOPRM1 cells after treatment of the cells with 0.5 μM simvastatin for 12 h. We were able to prevent the decreases in cholesterol content by including 20 ng/ml cholesterol during the simvastatin treatment (Figure ). As expected, simvastatin treatment also induced a decrease in cholesterol level on the membrane of HEKC170A cells (Figure ).
We also assessed how cholesterol depletion influences its association with the receptor signaling complex. Simvastatin treatment decreased the association of cholesterol with the receptor complex, but this could be prevented by including 20 ng/ml cholesterol in the culture medium (Figure ). Further, simvastatin not only decreased the amount of cholesterol precipitated in the "PBS" group, it also impaired the ability of the HA antibody to precipitate more cholesterol than FLAG antibody. Since cholesterol association was not detected in the HEKC170A cells, simvastatin treatment had no effect in these cells (Figure ).
Since simvastatin treatment decreased the cellular cholesterol content, we used the FRET assay to determine whether cholesterol content affects receptor dimerization and G protein coupling. The normalized net FRET between CFPOPRM1 and YFPOPRM1 in simvastatin-treated HEK cells was decreased compared to untreated cells and could be reversed by inclusion of cholesterol during the simvastatin treatment (Figure ). A similar simvastatin-mediated decrease was observed with CFPOPRM1 and YFPGαi2 and could also be reversed by the inclusion of cholesterol during the simvastatin treatment (Figure ). However, the cholesterol depletion induced by simvastatin did not affect the homodimerization (Figure ) or G protein coupling of C170A (Figure ). Therefore, the presence of cholesterol within the receptor signaling complex is critical for receptor homodimerization and Gαi2 coupling.
Figure 6 Reducing cellular cholesterol affects homodimerization and G protein coupling. (a) HEK cells were transfected with CFPOPRM1 and YFPOPRM1 or transfected with CFPC170A and YFPC170A for 24 h. These cells were than treated with PBS (Control), 0.5 μM (more ...)
We further illustrated the relationship between receptor palmitoylation, cholesterol association, and receptor dimerization by incubating cells with the palmitoylation inhibitor 2-BP. We observed a decrease in cholesterol associated with the OPRM1 signaling complex after 2-BP treatment (Figure ). Because of the inhibitory effect of palmitoylation blockage on Gαi2 membrane targeting [14
], the influence of 2-BP on Gαi2 coupling was not investigated. We also saw a reduction in the normalized net FRET between CFPOPRM1 and YFPOPRM1 after 2-BP treatment (Figure ). In addition, 2-BP treatment did not affect cholesterol association with C170A or the homodimerization of C170A, since palmitoylation blockage in C170A already impaired these two functions to basal levels (Figure ).
Figure 7 Palmitoylation inhibitor impairs homodimerization and cholesterol association. (a) HEKOPRM1 or HEKC170A cells were treated with 50 μM 2-BP or vehicle for 12 h, and the cholesterol associated with receptor complex was measured. (b) HEK cells were (more ...)
Computational modeling suggests that palmitoyl-cholesterol interaction stabilizes the OPRM1 homodimer
We undertook modeling studies to confirm that a specific cholesterol interaction with palmitoylated C3.55(170) may enhance the interactions at the homodimer interface of OPRM1. The OPRM1 model we developed for the modeling studies reported here is a homology model that uses the β2
-AR crystal structure as a template [15
]. As mentioned in Methods
, OPRM1 has two TMHs that differ in the position of helix deforming residues from the template β2
-AR (TMH2: P2.58 OPRM1 vs. P2.59 β2
-AR; TMH4: P4.59 OPRM1 vs. P4.60 β2
-AR). Our Conformational Memories (CM) calculations revealed that the location of P2.58 in OPRM1 causes the pitch of TMH2 to change after the proline such that residue 2.60 faces into the binding pocket. This same residue position in the β2
-AR resides in the TMH2/3 interface. These results are consistent with the conformation of TMH2 in the CXCR4 crystal structure (CXCR4 also has a Pro at 2.58) [16
]. The TMH4 region from 4.53 to 4.58 is SSAIGLP in OPRM1. Our CM calculations showed that the presence of the G2.56 so close to P2.58 causes a wider turn in TMH4 than is seen in β2
-AR. The net result is that TMH4 leans more towards TMH5. One result of this change is the lipid exposure of residue 4.59, a key residue in the TMH4 dimer interface (see below). These two key helix changes, along with the resulting changes in helix packing, distinguish the OPRM1 binding pocket (and lipid face) from that of β2
Our detailed modeling procedures are described in Methods. Figure illustrates the position of cholesterol relative to the palmitoyl and the TMH bundle. Due to the extreme tilt of TMH3 in the TMH bundle, the intracellular end of TMH3 (orange) is between the intracellular ends of TMH4 (yellow) and TMH5 (cyan). This position of TMH3 allows the cholesterol to pack between the C3.55(170) palmitoyl and TMH4. Figure provides an extracellular view of the final energy-minimized OPRM1 homodimer. In the resultant dimer, cholesterol is packed against the TMH4 interface and TMH3. The palmitoyl at C3.55(170) is packed against the cholesterol with TMH5, blocking cholesterol from leaving the interface. Table provides a summary of the resultant interaction energies for the palmitoylated OPRM1 homodimer/cholesterol complex. It is clear here that the major energetic contributions to the interaction energies between the protomers are van der Waals (VDW) energies. The homodimer interface residues with VDW contributions are N4.41, I4.44, C4.48, I4.51, and I4.56, with a total energy of -14.76 kcal/mol. The cholesterol associated with protomer A interacts with protomer B residues R4.40, N4.41, K4.43, I4.44, and V4.47, contributing an additional -2.44 kcal/mol, and the cholesterol associated with protomer B contributes an additional -2.39 kcal/mol. Thus, the total cholesterol interactions (-4.83 kcal/mol) contribute 24.7% to the total interaction energy at the homodimer interface (-19.59 kcal/mol), suggesting that the interaction between cholesterol and palmitoyl facilitates OPRM1 homodimerization
Figure 8 Computational modeling of the OPRM1 homodimer interface. (a) This figure illustrates the position of cholesterol relative to the palmitoyl and the OPRM1 TMH bundle. The view is from lipid looking toward TMH4 (yellow). The OPRM1 model is displayed in molecular (more ...)
Homodimer interface interaction energies.