Membrane cholesterol access into a G-protein-coupled receptor

Cholesterol is a key component of cell membranes with a proven modulatory role on the function and ligand-binding properties of G-protein-coupled receptors (GPCRs). Crystal structures of prototypical GPCRs such as the adenosine A2A receptor (A2AR) have confirmed that cholesterol finds stable binding sites at the receptor surface suggesting an allosteric role of this lipid. Here we combine experimental and computational approaches to show that cholesterol can spontaneously enter the A2AR-binding pocket from the membrane milieu using the same portal gate previously suggested for opsin ligands. We confirm the presence of cholesterol inside the receptor by chemical modification of the A2AR interior in a biotinylation assay. Overall, we show that cholesterol's impact on A2AR-binding affinity goes beyond pure allosteric modulation and unveils a new interaction mode between cholesterol and the A2AR that could potentially apply to other GPCRs.


Supplementary note 3 -Biotinylation experiments
1. Details about the methodology: originally, the Substituted-Cysteine Accessibility Method (SCAM) was developed to elucidate water-accessible residues in membrane-spanning proteins like channels 14 , transporters 15 or binding-site crevices 16 . In the absence of crystallography data, this method requires a systematic mutation of every protein residue into a cysteine followed by an assessment of ligand binding properties. Throughout a decade, the SCAM method has been employed by Javitch et al. to explore water-accessible residues in one G protein coupled receptor, namely the dopamine D2 receptor (D2R) [16][17][18] . As mentioned in this work, an ideal starting point would be to create a cysteine-free pseudo-wild-type background that is insensitive to the reagents and has normal expression and function. However, intense efforts failed to achieve such construct for the D2R 19 . It is likely that the lack of cysteine residues in cysteine-free pseudo wild type GPCRs impacts protein expression, folding and thus the formation of a functional receptor.
In the present study, we adapted the SCAM method to explore the reactivity of cysteine residues under different conditions in the A2AR interior. The basis of the employed methodology relies on two key aspects: a. Methanethiosulfonate (MTS) reagents react with sulfhydryl groups of accessible cysteine residues 19 . Thus, only cysteines at the water-accessible surface of the receptor, namely the extracellular side or interior of the membrane-spanning segment, will be reactive to MTS reagents (i.e. cysteine residues at the lipid-accessible surface will not be reactive). b. MTS reagents are specific for free sulfhydryl groups, and thus residues engaged in a disulfide bridge are not reactive even if they are accessible to water 19,20 .
In contrast to the work from Javitch et al., in the present study we know the exact location of water-accessible cysteines in the A2AR thanks to the availability of high resolution crystallography data (PDB:3EML). This knowledge is crucial for correct interpretation of biotinylation and binding experiments as outlined below.
2. Ability of cysteine residues to react with MTSEA-B in the A2AR. In order to assess the ability of the A2AR to react with the biotinylation reagent MTSEA-B, the receptor was scanned for cysteine residues, their location and engagement in disulfide bridges. As shown by high resolution crystallography data (PBD: 3EML) ( Supplementary Fig. 14 Supplementary Fig. 15, chemical modification of these cysteine residues by MTSEA-B would clearly overlap with the orthosteric binding site of the A2AR. In summary, a detailed structural characterization of the A2AR suggests that all cysteines outside the receptor are non-reactive as they are engaged in disulfide bridges and that only cysteines inside the receptor are reactive towards the biotinylation reagent MTSEA-B. Hence, regardless of which cysteines become biotinylated in the receptor interior, cholesterol needs to enter the receptor interior to exert its action. It is worth noting that based on the experimental set-up used in this paper, we cannot pinpoint the exact cysteine residue reacting with the biotinylation reagent. However, provided that MTSEA-B should access the receptor from the extracellular side, we can speculate that the first cysteine residue (i.e. C3.30) found on the MTSEA-B entrance pathway into receptor should mostly react with the biotinylation reagent.

Supplementary note 4 -Exploring exit pathways for cholesterol from A2AR ligand binding site
Methyl-β-cyclodextrin (MβCD), the cholesterol-depleting agent we use in our experiments, is not likely to remove cholesterol from the interior of the A2AR pocket. Thus, it is reasonable to think that cholesterol first has to leave spontaneously the interior of the A2AR towards the membrane bulk before being removed by MβCD. In this context, we explored the energetic cost of extracting cholesterol from the A2AR binding crevice into the aqueous phase. As we discuss in the main text, the extraction of cholesterol to the water phase is very costly, namely 120 kJ·mol -1 (~50 KBT). Due to its hydrophobic moiety, cholesterol is highly reluctant to become exposed to the water phase but rather establishes a strong interaction with the extracellular loop 2 of the protein (see Supplementary Fig. 16, region labeled as C). A visual inspection of our simulations shows that shortly after pulling forces are applied upwards (i.e. ξ within 1-2 nm), cholesterol changes its initial orientation with respect to the membrane plane from a rather fluctuating parallel orientation to a more stable perpendicular position. Interactions between cholesterol hydroxyl group and water molecules present at the interior of the protein somehow shield cholesterol hydrophobic ring and play an important role in the position and stability of cholesterol inside and the pocket.
As highlighted in Supplementary Fig. 16, the energy landscape of cholesterol exiting the A2AR towards the water phase can be divided in three main regions by labels A, B and C. While cholesterol exiting from A2AR through the water phase is unlikely, our PMF calculations show a thermally accessible region which might be a key for a direct exit pathway into the membrane bulk. This region is located between ~20 kJ·mol -1 (~8 KBT) (label A in Supplementary Fig. 16) and ~37 kJ·mol -1 (~15 KBT) (label B in Supplementary Fig. 16). A visual inspection of the simulations shows that the lowest energy barrier of this region (i.e. label A) corresponds to cholesterol rigid ring progressing through the space limited by TM2 and TM7. Once cholesterol overcomes this barrier, the molecule is ready to head towards TM1-TM2 and/or TM1-TM7 exit to the membrane bulk. But in order to proceed to any of these exit gates, cholesterol also tail needs to cross the TM2 -TM7 corridor which involves an extra energy cost (i.e. label B). Label C mainly corresponds to the cholesterol molecule in the water phase, mostly interacting with the extracellular loop 2 of the A2AR.
In addition, as shown in label A of Supplementary Fig. 16, the position of cholesterol oxygen along the z axis in this region is very similar to that of membrane cholesterol bulk (reference density profile, blue line) which would eventually allow a comfortable insertion of cholesterol into the membrane upon exiting the receptor via TM1-TM2 and/or TM1-TM7. Interestingly, previous work 21 on the β2-adrenergic receptor using random acceleration molecular dynamics have highlighted that TM1-TM2 and TM1-TM7 are potential ligand entry/exit pathways. It is important to note that during our simulations the head of two phospholipids from the upper leaflet of the membrane (i.e. extracellular) significantly occupy both TM1-TM2 and TM1-TM7 grooves, as previously described in other computational and experimental studies.
This suggests that exit via TM1-TM2 and TM1-TM7 grooves may be additionally modulated by the diffusion of these phospholipids, an event way beyond our simulated time scale

Supplementary Tables
Supplementary Table 1  Abbreviations: 1,2-dipalmitoyl-sn-glycero-3phosphocholine (DPPC), 1,2-distearoyl-sn-glycero-3phosphocholine (DSPC), 1,2-dioleyl-sn-glycero-3phosphocholine (DOPC), 1-stearoyl-2docosahexaenoyl-sn-glycero-3-phosphocholine (SDPC) and sphingomyelin (SM), cholesterol (CHOL), and water (W), respectively.  Cholesterol was extracted from the ligand binding site of the A2AR receptor to the extracellular water phase using umbrella sampling simulations. While the black curve is the average free energy difference, standard deviation values obtained from bootstrapping are shown as black error bars. The reaction coordinate (ξ) corresponds to the distance along the z axis between the centre of mass of the A2AR backbone and the oxygen atom of cholesterol. Maroon, blue and orange lines show reference mass density profiles (kg m -3 , right y axis) of the whole membrane (all lipid molecules), membrane cholesterol bulk (cholesterol oxygen atoms), and cholesterol at the binding site (cholesterol oxygen atom, data taken from one unbiased simulation), respectively.