In silico interaction analysis of Cannabinoid Receptor Interacting Protein 1b (CRIP1b) – CB1 Cannabinoid Receptor

Abstract

Cannabinoid Receptor Interacting Protein isoform 1b (CRIP1b) is known to interact with the CB₁ receptor. Alternative splicing of the CNRIP1 gene produces CRIP1a and CRIP1b with a difference in the third exon only. Exons 1 and 2 encode for a functional domain in both proteins. CRIP1a is involved in regulating CB₁ receptor internalization, but the function of CRIP1b is not very well characterized. Since there are significant identities in functional domains of these proteins, CRIP1b is a potential target for drug discovery. We report here predicted structure of CRIP1b followed by its interaction analysis with CB₁ receptor by in-silico methods. A number of complementary computational techniques, including, homology modeling, ab-initio and protein threading, were applied to generate three-dimensional molecular models for CRIP1b. The computed model of CRIP1b was refined, followed by docking with C terminus of CB₁ receptor to generate a model for the CRIP1b- CB₁ receptor interaction. The structure of CRIP1b obtained by homology modelling using RHO_GDI-2 as template is a sandwich fold structure having beta sheets connected by loops, similar to predicted CRIP1a structure. The best scoring refined model of CRIP1b in complex with the CB₁ receptor C terminus peptide showed favourable polar interactions. The overall binding pocket of CRIP1b was found to be overlapping to that of CRIP1a. The Arg82 and Cys126 of CRIP1b are involved in the majority of hydrogen bond interactions with the CB₁ receptor and are possible key residues required for interactions between the CB₁ receptor and CRIP1b.

Graphical abstract

The CRIP1b (cyan) in complex with the CB₁ receptor C-terminal peptide (blue) showing the hydrogen bond interactions between CRIP1b and C-terminal tail of the CB₁ receptor (S⁴⁶⁴TDTSAEAL⁴⁷²). Here, the computational models of CRIP1b and CB₁ receptor C-terminal peptide are used to depict the relation between CRIP1b and CB₁ receptor.

1. Background

Cannabinoid receptors are a family of G protein-coupled receptors that are involved in a wide variety of physiological and homeostatic processes [1–5]. The two cannabinoid receptors, the CB₁ receptor and the CB₂ receptor, have been identified including multiple splice variants of CB₁in humans [1, 2]. The CB₁ receptor is predominantly expressed in brain where it seems to play an important role in regulating analgesia, appetite, obesity, learning and memory, synaptic plasticity, substance abuse disorders, and neuro-protection [1, 3–5].TheCB₂ receptor, in contrast is found primarily in peripheral tissue, including tissue associated with immune system like spleen and tonsils, and is involved in cannabinoid mediated immune response[1,2]. Cannabinoid receptors bind and are activated by the lipid derived endocannabinoid ligands, anandamide and 2-arachidonoylglycerol[3,4]. CB₁ receptor signaling and regulation have assumed increased biomedical significance because of their involvement in variety of physiological disorders and diseases [3–5].

The CB₁ receptor signals mainly through the activation of G proteins of Gi/o family, and regulates Gi/o-mediated inhibition of adenylyl cyclase, inhibition of Ca2+ channels, interaction with receptor tyrosine kinases and non-receptor Src kinases to activate the Extracellular Signal-Regulated Kinase (ERK1/2), and other cellular signaling via interactions with Gi/o proteins[1,2]. The CB₁ receptor also interacts with associated proteins that serve as regulators of cellular signaling (FAN, β-arrestins) or receptor trafficking (β-arrestins, SGIP1, GASP) [6–8]. In addition, the CB₁receptoralso interacts with CRIP1a and CRIP1b, the cannabinoid receptor interacting proteins (CRIPs) that function to regulate cellular signaling and CB₁ receptor trafficking [9–16].

CRIP1a and CRIP1b are cytosolic proteins that were first discovered when the Lewis laboratory determined that deletion of the CB₁ receptor C-terminal released a tonic inhibition of N-type Ca²⁺ channels in neurons, concluding that the C-terminal performed an auto-inhibition function [5,6]. A yeast two-hybrid screen identified CRIP1a as a key associated protein [11]. The cannabinoid receptor interacting protein isoforms 1a and 1b (CRIP1a and CRIP1b) are generated by alternative splicing of CNRIP1 gene located on chromosome number 2 of the human genome [11]. CRIP1a is a 164-amino acid protein, whereas CRIP1b is a 128-amino acid protein, and both interact with the CB₁ cannabinoid receptor. Although CRIP1a is conserved among mammals and fish, the presence of CRIP1b is limited to primates [11]. CRIP1 proteins have been has been identified immunologically in neurons in the superior cervical ganglion [11], striatum [17], glutamatergic CA3 pyramidal neurons in the hippocampus [18, 19] and retina [20].

The function of CRIP1a has been clearly established in CB₁ receptor selection of Gi/o protein preference and signaling, and competition for β-arrestin to attenuate CB₁ receptor internalization[14], whereas the function of CRIP1b is not known [11]. Exogenous CRIP1a reversed CB₁ receptor-mediated tonic inhibition of Ca²⁺ currents, whereas CRIP1b could not[11]. CRIP1a-knockdown clones in the model neuroblastoma N18TG2 cells exhibited enhanced ERK1/2 phosphorylation efficacy in response to CP55940 and displayed a leftward shift in CP55940-mediated inhibition of forskolin-stimulated cAMP accumulation [12,13]. CB₁ receptor-mediated G_i3 and G_o activation by CP55940 was attenuated by CRIP1a over-expression, but robustly enhanced in CRIP1a-knockdown clones. Conversely, CP55940-mediated G_i1 and G_i2 activation was significant enhanced in cells over-expressing CRIP1a, but not inCRIP1a-knockdown clones [14]. These studies suggest that endogenous levels of CRIP1a modulate CB₁ receptor-mediated signal transduction by facilitating a G_i/o subtype preference for G_i1 and G_i2, accompanied by an overall suppression of G protein-mediated signaling in neuronal cells [12,13].

CRIP1a plays a role in the control of agonist-driven CB₁R cell surface regulation by competing with β-arrestins and thereby attenuating agonist-mediated internalization[14,15]. CRIP1a over-expression attenuated the agonist-induced redistribution of endogenous β-arrestin to punctae aggregates and subsequent loss of cell surface CB₁receptors. Conversely, CRIP1a knock-down augmented agonist-mediated β-arrestin redistribution to punctae[14]. Co-immunoprecipitation studies indicated thatCRIP1a competes with β-arrestin for binding to the CB₁ receptor, which attenuates the action of β-arrestin to mediate CB₁R internalization[14].Peptides mimicking the CB₁ receptor C-terminus could pull-down CRIP1a from cell extracts and recombinant CRIP1a. Co-immunoprecipitation of CB₁ receptor protein complexes demonstrated that distal C-terminal peptides competed for the CB₁ receptor association with CRIP1a [14], demonstrating the importance of this domain of the CB₁ receptor for CRIP binding to the CB₁ receptor. A similar distal C-terminal sequence found in the mGlu8a receptor has also been found to interact with CRIP1a, and may sub serve the same functions as for regulation of the CB₁ receptor [16].

Efforts to establish a cellular function for CRIP1b have not yielded definitive conclusions. Both CRIP1a and CRIP1b proteins interact with the last nine amino acids (S⁴⁶⁴TDTSAEAL⁴⁷²) of the distal CB₁-C terminus tail of the human CB₁ cannabinoid receptor (Figure 1). The CB₁ C-terminus binding is dependent upon a domain (amino acids 34 to 100) common to both CRIP1a and CRIP1b, as determined by studies of deletion mutations [11]. CRIP1a and CRIP1b possess distinctly divergent C-terminal domains, such that CRIP1b has the potential to exhibit functional differences from CRIP1a. CRIP1 proteins share no sequence homology to other proteins, so they cannot be categorized into a particular functional family based upon primary structure. Although recent modeling studies of CRIP1a by Ahmad et al., [21] revealed a structure containing beta sheets connected by loops, there is no report on the structure of CRIP1b so far.

Figure 1.

Schematic outline of the general workflow for CRIP1b modeling and its interaction analysis withCB₁ receptor.

The present study deals with structure prediction of CRIP1b followed by analysis of its interaction with aCB₁ receptor peptide by in-silico methods. A number of complementary computational techniques, including, homology modeling, ab-initio and protein threading, were applied to generate three dimensional molecular models for CRIP1b. The computed model of CRIP1b was refined, followed by docking with the C-terminus of CB₁ receptor to generate a model for the CRIP1b- CB₁ receptor interaction. The steps followed are described in a flow chart given in Figure 1.The aim of this study was to finding the key residues involved in CRIP1b- CB₁ receptor interaction in order to provide clues to CRIP1b function.

2. Methods

2.1 CB₁ receptor C terminus peptide modeling

The terminal 9-mer sequence from the distal CB₁ receptor C-terminus (S⁴⁶⁴TDTSAEAL⁴⁷²) was modeled using PEP-Fold [22]. This method, based on structural alphabet SA letters to describe the conformations of four consecutive residues, couples the predicted series of SA letters to a greedy algorithm and a coarse-grained force field. The most-native clusters of all peptides are ranked in the top 10 generated models [23].

2.2 CRIP1b modeling and Validation

The secondary structure for CRIP1b was predicted by various algorithms like HMMSTR, SSPRO4, SSPAL, JNET, SPARROW, SABLE, NNSSP, NETSURFP, SSP, PSS FINDER, SPINE, and PSIPRED[24–34]. A consensus of secondary structure results was generated by using GeneSilico Metaserver [35].

The suitable template for the CRIP1b model was selected by a homology search against Protein Data Bank using SwissModel server, NCBI-Blastp, PDBBLAST, CSBLAST, COMA, Hhblits, HHSearch, FFAS and Phyre [36–45]. The three dimensional model was built using MODELLER in Accelrys Discovery Studio 2.5 [46]. Loop refinement protocol was incorporated to increase the accuracy of the model [47]. The energy minimization was performed by Conjugate gradient minimization protocol using CHARMM force field in Discovery studio [48–50].

Apart from homology modeling, ab-initio and threading methods using Robetta, Muster and iTasser, were used to calculate a model for CRIP1b [51–55]. The Robetta algorithm first attempts homology modeling by searching for confident homologs from BLAST, PSI-BLAST, FFAS03 or 3D-Jury. If a homolog is not available, then the de novo Rosetta fragment insertion method is used for prediction. Muster and iTasser following a threading algorithm, based on fold recognition. The iTasser (Iterative Threading ASSEmbly Refinement) follows a hierarchical approach, starting with LOMETS to identify structural templates from the PDB followed by full-length atomic model construction using iterative template fragment assembly simulations [53–55]. MUSTER (MUlti-Sources ThreadER) combines sequence profile alignment with multiple structural information data to generate sequence-template alignments for model prediction [52].

The quality of the models generated by different methods was analyzed using Discrete Optimized Protein Energy, or DOPE score. DOPE is an atomic distance-dependent statistical potential from a sample of native structures, and is extracted from a non-redundant set of crystallographic structures [56]. The model geometry was validated by using Ramachandran plot [57] in Discovery Studio. The PyMol was employed to aid visualizations, analysis and image preparation of all the obtained models [58].

2.3 Protein-peptide docking

The HADDOCK (High Ambiguity Driven DOCKing) server was used for docking CRIP1b- CB₁ receptor nine-mer peptide [59–61]. As the site where C-terminus of CB₁ receptor binds exactly on CRIP1b is not known, the 9-mer peptide was docked with all possible binding sites with no restriction for unbiased evaluation. The HADDOCK performs docking in three stages, starting with randomization of orientations and rigid-body minimization, and then semi-flexible simulated annealing in torsion angle space is performed, followed by refinement in Cartesian space with explicit solvent. During the final step, the complex is immersed in a suitable solvent environment, using TIP3P water model [62]. Then, the models are subjected to molecular dynamics simulation at 300K [63]. Initially position restraints are applied on the non-interface heavy atoms, which are relaxed in later stages to facilitate side chain optimization [59–61]. Docking results obtained as ligand poses on different binding site clusters based on positional interface ligand RMSD were individually inspected, after which high scoring models were passed into the refinement step.

2.4 Refinement of docked complexes

The structures of CRIP1b in complex with the CB₁-receptor C-terminus 9-mer peptide (S⁴⁶⁴TDTSAEAL⁴⁷²) obtained from docking results were subjected to FireDock (Fast Interaction REfinement in molecular DOCKing) for rescoring and refinement [64]. The rearrangement of the interface side-chains followed by adjustment of the relative orientation of the molecules, are incorporated in the refinement process. The refined structures are ranked according to their binding energy, which is represented by global energy. The atomic contact energy (ACE), van der Waals forces (Attractive and Repulsive VdW) and hydrogen bonds (HB) contribute to the calculation of global energy, leading to final ranking [64,65]. The best scoring complex was further analyzed with PyMOL, Discovery Studio and Protein Interactions Calculator (PIC)[66]. PIC computes various interactions such as disulphide bonds, interactions between hydrophobic residues, ionic interactions, hydrogen bonds, aromatic-aromatic interactions, aromatic–sulphur interactions and cation–π interactions within a protein or between proteins in a complex [66].

3. Results and Discussion

3.1 Model of 9-mer peptide(S⁴⁶⁴TDTSAEAL⁴⁷²) of CB₁ receptor C terminus tail

A de novo approach using PEP-Fold server was used to calculate the structure of the 9-mer peptide (S⁴⁶⁴TDTSAEAL⁴⁷²) (Figure 2). The PEP-Fold ranks generated models according to their sOPEP, the coarse grained energy. The best-ranked model was selected for docking studies. The energy minimization of the selected peptide was performed by Conjugate gradient protocol under CHARMM force field.

Figure 2.

The de novo model of peptide from distal tail of CB₁ receptor C-terminus

3.2 The CRIP1b models

The human CRIP1b sequence was retrieved from NCBI with accession NP_001104571. The CRIP1b sequence was subjected to secondary structure calculation. The consensus result of various algorithms showed beta sheets as the major secondary structure component (Figure 3).

Figure 3.

The secondary structure of CRIP1b predicted by various protocols, the consensus of results was taken into account to estimate secondary structural content in CRIP1b. The beta sheets are represented by ‘E’ and alpha helices by ‘H’.

The Blastp results (Supporting file: Table S1) did not show any suitable sequence homolog for the purposes of modeling structures. Other tools were utilized for more sensitive data mining in search of CRIP1b structural homologs. The results produced by PDBBLAST, CSBLAST, COMA, Hhblits, HHSearch, FFAS and Phyre were low scoring (Supporting file: Table S1 A-G). The SwissModel suggested RHO GDP-DISSOCIATION INHIBITOR 2 (1DS6_B)[63] as a suitable template for modeling with 27.5% identity with CRIP1b (Supporting file: Table S2).This template was also recently suggested for CRIP1a homolog by Ahmed and colleagues [21]. Because our goal was to seek commonality or distinct differences from CRIP1a, we chose to use this template. The RHO-GDI2 structure-CRIP1b sequence alignment (figure 4) was used as input for computing CRIP1b models, using Modeller (Discover studio), followed by loop refinement and energy minimization (Figure 1A). Apart from homology modeling, other approaches like ab-initio and threading were also employed to obtain CRIP1b models. For comparison, we also calculated CRIP1a model, using the same template RHO-GDI 2.

Figure 4. Sequence alignment of Rho-GDI 2/1DS6_B (template) and CRIP1b sequence.

The ‘1DS6_b(PDB)’ represents secondary structure of Rho-GDI2 in PDB, and consensus secondary structure represents predicted secondary structure of CRIP1b.

3.3 Selection of a Valid Model

Because multiple CRIP1b models were generated by different approaches, it was important to choose a valid model with the best quality for further studies. For this purpose DOPE scores of all the models were calculated to facilitate their ranking. One of the CRIP1b models generated by homology modeling (with RHO_GDI 2 as template) showed the lowest DOPE value (Figure 5). The ab initio models generated by Robetta were closest to, but not better than, the models generated by Discovery studio in terms of their high DOPE score, (Table 1). The best ranking CRIP1b structure was validated with Ramachandran plot having all the residues in favorable region(Supporting file: Figure S1). The secondary structure data was also in agreement with the CRIP1b model, showing a sandwich fold model with beta sheets.

Figure 5.

The CRIP1b model (A) and its alignment with the template Rho-GDI 2, cyan colored (B)

Table 1. The DOPE score of CRIP1b models.

The models are arranged according to their DOPE score.

CRIP1b Model	DOPE Score
DS_model4	−13222.31836
DS_model1	−12861.45606
Robetta_model1	−12259.22266
Robetta_model5	−12102.96289
Robetta_model2	−12075.74316
Robetta_model3	−11994.18164
Robetta_model4	−11134.72266
DS_model3	−10764.875
DS_model2	−10442.48145
DS_model5	−10083.26953
Muster_model2	−7846.150391
Muster_model9	−7585.559082
Itasser_model1	−7320.54834
Muster_model4	−7285.194336
Itasser_model2	−7151.129883
Muster_model3	−7069.747559
Itasser_model4	−6522.483887
Muster_model7	−6313.573242
Muster_model8	−6313.573242
Muster_model5	−5812.94873
Muster_model6	−5812.632324
Muster_model1	−5803.634277
MultiCom_model	−5381.104004
Itasser_model5	−5299.068359
Muster_model10	−4933.700684
Itasser_model3	−1822.536377

Because CRIP1a and CRIP1b sequences are identical within the initial 110 residues, comparison between their models was performed. Although the model of CRIP1a is already reported by Ahmed et al.[21], the PDB coordinates of model are not available, so we derived the CRIP1a model in the same manner as CRIP1b to draw comparisons between the two. The residues lying in the C-terminal region (amino acids 155–164) were not covered by the template; hence they were modeled by an ab initio approach using Robetta and incorporated to form a full length model of CRIP1a. The homology model (with RHO-GDI 2 as template) showed the best DOPE scores; secondary structure data and Ramachandran Plot validated it. CRIP1a also showed a sandwich fold model with beta sheets (Supporting file: Figure S2A). The CRIP1a and CRIP1b models were aligned (Supporting file: Figure S2B), the majority of their secondary structure is not only similar to each other, but also agrees with the CRIP1a model reported by Ahmed et al. [21].The best ranked models of CRIP1b as well as CRIP1a were chosen for further interaction studies with the CB₁ receptor C-terminus 9-mer peptide.

Molecular docking of CRIP1b with the CB₁ receptor C-terminus peptide

To study interactions between CRIP1b and CB1 receptor C-terminus distal tail, the CRIP1b was docked with the 9-mer peptide (S464TDTSAEAL472) model generated earlier. Here, we have used HADDOCK 2.2, which implements solvated docking approach along with conventional Haddock docking protocol [67]. Each molecule and its associated solvation shell are considered as rigid body, due to which an encounter complex is formed with a water layer between the two protein chains (in our case CRIP1b and CB1R peptide). The non-interfacial waters are removed and during subsequent energy minimization stage the remaining waters and protein chains are treated are separate rigid bodies. Afterwards a biased Monte Carlo procedure is applied to remove water molecules, based on their probabilities for water-mediated contacts for specific amino acids. This procedure is repeated until only 25% of the initial interfacial water molecules remain. During the final steps, water molecules are removed if their interaction energy is unfavorable. The interaction energy (sum of van der Waals and electrostatic water–protein energies) should be less than 0.0 kcal/mol.

As the site where the C-terminus of CB₁ binds exactly on CRIP1b is not known, the 9-mer peptide was docked with all possible sites on CRIP1b. The Haddock analysis returned 36 poses, grouped in nine clusters based on positional interface ligand RMSD. Twelve poses from the top three clusters were scattered at different positions on the CRIP1b surface (Figure 6). The binding pockets of these three clusters revealed the residues involved (Figure6). The CRIP1a model was also docked with CB₁ receptor C-terminus peptide (Supporting file: Figure S3).

Figure 6. Docked poses of CB1 receptor C terminus peptide on CRIP1b model.

Docking studies yielded various docked complexes with peptide on distinct positions. All the poses were refined and scored. On the left side of figure, the colored spheres show the positions of peptide on CRIP1b in different poses. While on the right side, binding pockets of top scoring poses from three different clusters are written with differently colored residue fonts. Also, the chemical properties of residues are represented by oval background of different colors i.e., negative charged by pink, positive charged by violet, hydrophobic by green, polar by light blue and glycine by cream.

Refined poses of CRIP1b in complex with the CB₁receptor C-terminus distal tail

All of the docked poses of CRIP1b were subjected to refinement, followed by binding energy (referred to as global energy in Kcal/mol) calculation (Table 2). The best scoring refined pose of CRIP1b in complex with CB₁ receptor C-terminus peptide was taken into account for interaction analyses. The best scoring pose for CRIP1a in complex with CB₁ receptor C-terminus peptide was retrieved under same set of parameters (Supporting file: Table S3). The interactions for both complexes were calculated by PIC.

Table 2. FireDock scores.

The docked poses of CRIP1b in Complex with the CB₁ 9-mer peptide were refined and ranked by their binding energy (Global energy), which were contributed by attractive and repulsive Van der Waals (VdW) forces, atomic contact energy(ACE) and hydrogen bonds(HB).

Rank	Solution Number	Global Energy	Attractive VdW	Repulsive VdW	ACE	HB
1	11	−38.77	−17.43	3.72	−8.89	0.00
2	27	−38.19	−14.59	1.42	−10.07	0.00
3	12	−35.05	−17.69	4.42	−4.54	−1.16
4	34	−33.46	−20.68	6.30	−6.97	−1.47
5	19	−31.99	−19.96	8.32	−8.86	−1.61
6	26	−31.76	−20.19	3.95	−5.47	−2.04
7	8	−31.36	−18.81	7.90	−5.35	−3.84
8	2	−30.78	−19.25	8.78	−10.22	−3.06
9	35	−28.31	−17.35	5.69	−9.30	−0.50
10	10	−28.17	−18.47	4.22	−4.06	−2.22
11	7	−27.43	−14.61	1.10	−3.12	−2.64
12	17	−26.42	−18.67	5.56	−4.24	−1.82
13	29	−23.96	−12.25	4.14	−6.86	−1.10
14	6	−23.81	−19.43	4.94	−7.83	−2.44
15	23	−23.73	−20.79	6.42	−0.73	−4.38
16	30	−22.75	−15.15	3.61	−5.71	−1.67
17	36	−22.71	−14.68	7.75	−6.77	−1.41
18	1	−22.47	−12.56	6.79	−7.48	−1.82
19	37	−22.47	−12.56	6.79	−7.48	−1.82
20	21	−22.36	−16.79	6.70	−6.95	−1.72
21	15	−20.78	−13.24	3.74	−3.67	−0.36
22	24	−20.38	−12.79	2.14	−2.77	0.00
23	31	−18.24	−17.65	4.80	−2.24	−1.98
24	4	−16.62	−20.56	7.33	−3.08	−1.02
25	20	−16.41	−17.54	6.70	−5.84	0.00
26	9	−15.62	−14.81	0.66	−0.40	−1.69
27	32	−15.45	−11.91	0.75	−4.81	0.00
28	28	−14.71	−13.76	4.69	−3.96	−1.16
29	16	−14.07	−17.12	12.77	−2.34	−1.52
30	33	−12.81	−13.58	4.74	−2.54	−1.58

The last nine amino acids of the C-terminus of the CB₁ receptor (Figure 2) are mostly polar in nature. The polar interactions are expected to be the major interaction between CRIP1b and CB₁ receptor. Our model suggests that CRIP1b forms hydrophobic and hydrogen bond interactions with the CB₁ receptor C-terminus peptide (Table 3). The hydrogen bonds involve Glu77, Arg82, Tyr85, Tyr124, Cys126 from CRIP1b and Ser464, Thr465, Ser468, Ala469, Glu470, Leu472 from the CB₁ receptor peptide (Figure 7). Similarly interactions for CRIP1a with CB₁receptor peptide were calculated and hydrophobic and hydrogen bond interactions were found (Supporting file: Table S4). The residues, namely Ser63, Asp81, Val83, Tyr 126, Tyr128, from CRIP1a were involved in the majority of interactions (Supporting file: Figure S4). The binding pockets of CRIP1b and CRIP1a lies in similar region between residues 61 and 128, mainly encompassing residues at position 82, 85 and 126. If we compare the binding pockets within a 4Å radius of the CB₁ receptor peptide in CRIP1a reported by Ahmed and colleagues [21] with our data, a large number of common residues are observed. The residues at position 64, 68, 82, 85, 124 and 126 lie in binding pockets of CRIP1b as well as CRIP1a. This is in agreement with the model of CRIP1a obtained by Ahmed et al.[21].

Table 3. Interactions between CRIP1b and the CB₁ receptor C-terminus 9-merpeptide.

The Chain A represents CRIP1b while chain B represents CB₁-R peptide. The distances are represented as Dd-a (between Donor and Acceptor) and Dh-a(between Hydrogen and Acceptor) while angles are represented as A(d-H-N)(between Donor-H-N) and A(a-O=C)(between Acceptor-O=C). The MO stands for Multiple Occupancy. The undefined values are indicated as “-“.

A. Binding pocket residues of CRIP1b within 4 Angstroms of CB1 R -peptide
Position	64	68	77	79	82	85	86	124	126
Residue	ILE	LEU	GLU	ASP	ARG	TYR	TYR	TYR	CYS

B. Hydrophobic Interactions within 5 Angstroms
Position	Residue	Chain	Position	Residue	Chain
64	ILE	A	472	LEU	B
68	LEU	A	469	ALA	B
85	TYR	A	469	ALA	B
89	TYR	A	472	LEU	B
124	TYR	A	469	ALA	B

C. Ionic Interactions within 6 Angstroms
Position	Residue	Chain	Position	Residue	Chain
82	ARG	A	466	ASP	B

D. Protein-Protein Main Chain-Side Chain Hydrogen Bonds
Donor				Acceptor				Parameters
POS	CHAIN	RES	ATOM	POS	CHAIN	RES	ATOM	MO	Dd-a	Dh-a	A(d-N)	A(a-O=C)
77	A	Glu	OE1	465	B	Thr	O	1	3.35	2.70	119.29	172.19
77	A	Glu	OE1	465	B	Thr	O	2	3.35	3.43	76.62	172.19
82	A	Arg	NH1	465	B	Thr	O	1	2.50	3.13	45.15	133.79
82	A	Arg	NH1	465	B	Thr	O	2	2.50	2.33	87.44	133.79
82	A	Arg	NH2	465	B	Thr	O	1	2.82	3.58	37.22	122.79
82	A	Arg	NH2	465	B	Thr	O	2	2.82	2.65	88.58	122.79
85	A	Tyr	OH	468	B	Ser	O	–	3.05	—	—	87.81
124	A	Tyr	OH	469	B	Ala	O	–	2.38	—	—	135.51
126	A	Cys	SG	469	B	Ala	O	–	2.98	—	—	147.60
126	A	Cys	SG	470	B	Glu	O	–	3.92	—	—	101.32
126	A	Cys	SG	472	B	Leu	O	–	2.30	—	—	125.72
126	A	Cys	SG	472	B	Leu	OXT	–	3.51	—	—	64.40
464	B	Ser	N	124	A	Tyr	OH	–	3.16	—	179.42	—
469	B	Ala	N	85	A	Tyr	OH	–	3.27	3.97	39.03	—

Figure 7. Interaction between CRIP1b (wheat colored) and the CB₁-receptor 9-mer peptide (magenta stick model).

The CRIP1b residues involved in the binding domain are represented by wheat colored sticks; those involved in hydrogen bonds are connected to the CB₁ receptor peptide by blue dashed lines.

Conclusion

There are previous experimental data indicating CRIP1-CB₁ receptor interactions [11]; however, due to the absence of crystal/solution structures of the complex or CRIP1b protein, these interactions have never been studied at the atomic level. For CRIP1 proteins, the majority of studies have been performed on CRIP1a and the data on CRIP1b are very limited. Our study is the first attempt to study structure and interactions of CRIP1b with the CB₁ receptor. To predict and ensure accurate three-dimensional models of CRIP1b structure; a number of different modeling techniques (Ab initio, fold recognition and homology modeling using RHO-GDI 2 as a template) were incorporated. RhoGDI 2 is known to regulate the GDP/GTP exchange reaction of Rho proteins by inhibiting the dissociation of GDP and the subsequent binding of GTP to them [68,69]. The CB₁ cannabinoid receptors promote focal adhesion kinase (FAK) activation, and data suggest that CB₁-stimulated FAK phosphorylation and activation can be mediated by G_12/13 regulation of Rac1 and RhoA [70]. Although we have data indicating a role for CRIP1a in the Gi/o family regulation, we have not explored the interactions between CRIP proteins, the G12/13 proteins, and the Rho family proteins. Rho-GDI 2 is characterized by a sandwich fold beta structure, and exhibits the best identity with CRIP1b. The template (Rho-GDI 2) used for homology modeling by us is same as the one used by Ahmad et al., for CRIP1a studies (obtained by threading), but we obtained it from SwissModel template search [38]. The sequence alignment of Rho-GDI2 and CRIP1b encompassed all 128 residues of CRIP1b. Apart from homology models, other models obtained from different methods were also individually evaluated and ranked accordingly. The best scoring CRIP1b model (with Rho-GDI2 as a template) having a beta sandwich fold validated by Ramachandran Plot was selected for interaction studies.

The 9-mer peptide (⁴⁶⁴STDTSAEAL⁴⁷²) from the CB₁ receptor distal C-terminus, was modeled using PEPFOLD, and docked as a ligand to the best-scored model of CRIP1b. The HADDOCK protocol generates multiple poses based on energy and scoring functions. These poses were further refined and the best scoring pose in terms of its interaction energy was chosen for interaction analysis. The best scoring refined model of CRIP1b in complex with the CB₁ receptor C-terminus peptide showed favorable polar interactions. Overall, the binding pocket of CRIP1b was found to be similar to that of CRIP1a (Figure 8). The Arg82 and Cys126 of CRIP1b, involved in the majority of hydrogen bond interactions with the 9-mer peptide, are possible key residues required for interactions between the CB₁ receptor and CRIP1b. Though we didn’t see bonds with Lys130 as described by Ahmed et al., but Lys130 exists in the binding pocket of CRIP1a in our results(Supporting file: Table S4A). The Presence of Lys130 within the 4 angstroms radius of CB₁R peptide indicates the high possibility of contacts between Lys130 of CRIP1a and CB₁R peptide. Apart from Lys130, the other residues (Asn61, Ser63, Asp81, Arg82, Tyr126 and Asn127) of CRIP1a overlap with binding pocket defined by Ahmad et al., within the 4 angstroms radius of CB₁R peptide. The CRIP1a binding pocket defined by us not only overlaps with data generated by Ahmad et al., but also with CRIP1b binding sites residues lying in region 64 to 126.

Figure 8. Binding pockets of CRIP1a (green) and CRIP1b (cyan) for CB₁ receptor peptide (magenta).

The above portion of image represents positions of CB₁ receptor peptide on CRIP1a and 1b, while lower portion shows their respective binding pockets.

Our computational studies indicate that CRIP1b shares similar CB₁ receptor binding pocket residues with CRIP1a, but there are functional differences between the two proteins. As it is known that CRIP1a (164 residues) and CRIP1b (128 residues) amino acid sequences are identical for the initial 110 residues, it can be postulated that the remaining 54 residues of CRIP1a may contribute to differences in their function. CRIP1a significantly attenuates tonic inhibition of voltage-gated Ca²⁺ channels by the CB₁receptor but CRIP1b does not. CRIP1a but not CRIP1b has a predicted palmitoyalation site that may contribute to its localization at the plasma membranes [11]. Also, CRIP1a but not CRIP1b contains a PDZ class I ligand in its C-terminal tail, which could indicate a potential for interactions with proteins containing PDZ domains [11]. There is no such ligand in the CRIP1b C-terminal tail, which could account for the functional differences between CRIP1a and CRIP1b, in spite of having similar CB₁R interaction binding pocket.

As there were previously no experimental data available for CRIP1b structure, information obtained from our modeling studies may be used to guide design of future site directed mutagenesis experiments. Understanding structure and interactions of CRIP1 proteins will lead to understanding underlying mechanisms of regulation of the CB₁ receptor.

Highlights.

• The three dimensional molecular model of CRIP1b is derived.

• The CRIP1b is a sandwich fold structure.

• Molecular docking studies establish Arg82 and Cys126 as the key residues of CRIP1b, involved in interaction with CB₁ receptor

• The binding pocket of CRIP1b has a overlap to that of CRIP1a.

ABBREVIATIONS

CRIP

Cannabinoid receptor interacting protein

FAN

Factor Associated with Neutral sphingomyelinase

GASP

GPCR-associated sorting protein

TIP3P

transferable intermolecular potential with 3 points

THC

Tetra-hydro cannabinol

DOPE

Discrete optimized protein energy