Helix 8 in chemotactic receptors of the complement system

Host response to infection involves the activation of the complement system leading to the production of anaphylatoxins C3a and C5a. Complement factor C5a exerts its effect through the activation of C5aR1, chemotactic receptor 1, and triggers the G protein-coupled signaling cascade. Orthosteric and allosteric antagonists of C5aR1 are a novel strategy for anti-inflammatory therapies. Here, we discuss recent crystal structures of inactive C5aR1 in terms of an inverted orientation of helix H8, unobserved in other GPCR structures. An analysis of mutual interactions of subunits in the C5aR1—G protein complex has provided new insights into the activation mechanism of this distinct receptor. By comparing two C5aR receptors C5aR1 and C5aR2 we explained differences between their signaling pathways on the molecular level. By means of molecular dynamics we explained why C5aR2 cannot transduce signal through the G protein pathway but instead recruits beta-arrestin. A comparison of microsecond MD trajectories started from active and inactive C5aR1 receptor conformations has provided insights into details of local and global changes in the transmembrane domain induced by interactions with the Gα subunit and explained the impact of inverted H8 on the C5aR1 activation.


Figure A. Global changes of C5aR1 induced by Gi subunits. (I) -A superposition of inactive
C5aR1 (blue-to-red, the C5aR1 crystal structure -5O9H) and active C5aR1 (grey, the homology model based on 6OMM, the cryo-EM structure of FPR2 with Gi). Both conformations of C5aR1, active and inactive, were subjected to microsecond MD. The structures extracted from the last frames of MD simulations starting from inactive and active C5aR1 were superposed in (II). A global movement of TM6 in inactive C5aR1 towards the active conformation induced by interactions with the Ga subunit was observed. The Ga subunits (orange) also overlapped regardless of the starting model of C5aR1 used for building the simulation system (inactive vs. active).
(I) (II) Figure B. A homology model vs. a microsecond MD-refined model of active C5aR1 -the location of G protein subunits. (I) -An RMSD plot representing the differences in the positions of Ca in the Ga subunit. Here, the TM cores in all of the frames were superposed to show only the variability of the Ga position with respect to the starting homology model. (II) -The same RMSD plot for Gb. Here, the Ga subunits of each frame were superposed to observe only the variability in the Gb position with respect to Ga. (III) -An RMSD plot for Gg. Here, the Gb subunits of each frame were superposed to observe the variability in the Gg position with respect to Gb. Interfaces Ga -Gb and Gb -Gg were of the least conformational variability. (IV) A superposition of the homology model of C5aR1 based on FPR2 with Gi (dark blue) and the C5aR1 conformation extracted from the last microsecond MD simulation frame (blue-to-red and orange). The N-terminus of Ga was slightly rotated inwards (a green arrow) in comparison to the FPR2 template, but its C-terminus interacting with the receptor remained in the place. Contact maps generated for a starting homology model of inactive C5aR1 with Ga (I, III, V) and for a microsecond MD-refined model of C5aR1 with Ga (II, IV, VI). (I, II) -contact maps for the receptor only. The most visible changes in the receptor conformation were marked with circles (magenta). They included: breaking the TM3-TM6 lock and TM6-TM7 interactions during activation and changes in the TM7-ICL4-H8 region. The most visible was the loss of TM6-TM2 interactions through the TM core of the receptor. (III, IV) -contact maps for the receptor and the Ga subunit. Interacting sequence regions were marked with the same circles as in Fig. 2. In comparison to Fig. 2 Figure E. Amino acid composition of ICL4 loops in C5aR receptors. In both cases, C5aR1 (I) and C5aR2 (II), polar amino acids are the most populated except for one Leu residue, but ICL4 (green) in C5aR2 is shorter than in C5aR1. Flexible ICL4 loops in both receptors did not form any regular secondary structure during MD simulations.

Figure F. Loss of crucial interactions between C5aR2 and Gi subunits. (I, II) -RMSD plots
showing a loss of interactions between C5aR2 and the Ga and Gb subunits. In both cases, the distances exceeded 8 Å, the maximal distance between the centers of mass of two residues to describe them as still being in contact. (I) -The RMSD plot for the Ca -Ca distance in the C5aR2 complex corresponding to the 325-349 distance in the C5aR1 complex (see Fig. 3D). This shows the loss of interactions between C5aR2 and Ga. (II) -The RMSD plot for the distance in the C5aR2 complex corresponding to the 331-57 distance in the C5aR1 complex (see Fig. 4D). This shows the loss of interactions between C5aR2 and Gb. (III, IV) -A comparison of the last simulation frames of C5aR1 (grey and dark blue) and C5aR2 (blue-tored and orange) showed completely different relative positions of H8 and the C-terminus of Ga in these two receptors despite their similar starting conformation based on active FPR2 with Gi. It was caused by shorter ICL3 and ICL4 loops in C5aR2 in comparison to C5aR1. (III) -a side view, (IV) -an intracellular view. Similar loss of contacts between the receptor and Ga can be observed in contact maps (V-VIII). (V, VI) -contact maps generated for the simulation started from active C5aR2 based on FPR2, (VII-VIII) -contact maps generated for the simulation started from inactive C5aR2 based on the crystal structure of C5aR1. Left panels (V and VIII) correspond to the beginning of simulations while right panels correspond to last frames of simulations. Regardless the starting conformation of C5aR2 used for the simulation, loss of several interactions was observed confirming the decreased stability of such simulation system. For example, intra-receptor interactions (ICL2 -C-terminus incl. H8, marked with right black circles) were lost. Also, C-terminal regions of both, the receptor and Ga (dark blue circles) did not interact by the end of simulations so close like in case of C5aR1 (see Fig. 2). Glu325 -Lys349 contact on the C5aR1 -Ga interface (see Fig. 3) was lost in the case of C5aR2. Moreover, there were discrepancies between simulations started from active C5aR2 and inactive one in this region (compare VI and VIII). Finally, regions involved in three-body interactions in C5aR1 (see Fig. 4, yellow circles), in C5aR2 were involved in far fewer contacts (yellow circles) suggesting that such interactions were not present at all.
(VII) (VIII) Figure G. A comparison of loop modeling algorithms in Rosetta. KIC (I), CCD (II), 50 topscoring loop models were selected from 5000 generated models. Here, an inactive structure of C5aR1 (PDB id: 5O9H) was used for loop modeling. CCD allowed for a more extensive conformational search also including TM helices but was slightly biased towards regular secondary structures.
(I) (II) Figure H. Beta-arrestin binding site -C5aR receptors vs. b1AR. (I) b1AR in complex with arrestin-2 (PDB id: 6TKO). Polar contacts were marked with dashed yellow lines. (II) a superposition of b1AR (green), C5aR1 (orange), and C5aR2 (grey). Here, a crystal structure of inactive C5aR1 and a homology model of inactive C5aR2 based on it were used. For both C5aR receptors, alanine (ICL1 -blue, ICL2 -green) substitutions of active site residues were observed in comparison to b1AR. Only for C5aR2 Arg/Leu (TM3 -marine green) and Gln/Gly (TM2 -light blue) substitutions were observed. H8 (red) was in the same place in all three receptors (II). Arrestin binding sites in details: b1AR (III), C5aR1 (based on b1AR) (IV), and C5aR2 (based on b1AR) (V). In (II-V) receptors were shown in blue-to-red color schemes.

Figure I. Residues important for C5aR1 function.
Here (I -II), the residues described in Figures  3 and 4, involved in the G protein coupling and signal transduction, were shown in orange, with the 'DRF' motif shown in red. The residues described in Table 1 as involved in the receptor activation were shown in yellow, the residues involved in the b-arrestin binding were shown in green, and the residues involved in the formation of H8 were shown in magenta. The latter were shown in detail in (II) -the side view. Both, (I) and (II), represent the refined C5aR1 structure extracted from the last frame of microsecond simulations, started from active C5aR1 based on FPR2.