Partial ligand-receptor engagement yields functional bias at the human complement receptor, C5aR1

The human complement component, C5a, binds two different seven-transmembrane receptors termed C5aR1 and C5aR2. C5aR1 is a prototypical G-protein–coupled receptor that couples to the Gαi subfamily of heterotrimeric G-proteins and β-arrestins (βarrs) following C5a stimulation. Peptide fragments derived from the C terminus of C5a can still interact with the receptor, albeit with lower affinity, and can act as agonists or antagonists. However, whether such fragments might display ligand bias at C5aR1 remains unexplored. Here, we compare C5a and a modified C-terminal fragment of C5a, C5apep, in terms of G-protein coupling, βarr recruitment, endocytosis, and extracellular signal-regulated kinase 1/2 mitogen-activated protein kinase activation at the human C5aR1. We discover that C5apep acts as a full agonist for Gαi coupling as measured by cAMP response and extracellular signal-regulated kinase 1/2 phosphorylation, but it displays partial agonism for βarr recruitment and receptor endocytosis. Interestingly, C5apep exhibits full-agonist efficacy with respect to inhibiting lipopolysaccharide-induced interleukin-6 secretion in human macrophages, but its ability to induce human neutrophil migration is substantially lower compared with C5a, although both these responses are sensitive to pertussis toxin treatment. Taken together, our data reveal that compared with C5a, C5apep exerts partial efficacy for βarr recruitment, receptor trafficking, and neutrophil migration. Our findings therefore uncover functional bias at C5aR1 and also provide a framework that can potentially be extended to chemokine receptors, which also typically interact with chemokines through a biphasic mechanism.

The complement peptide C5a, a potent chemotactic agent and an anaphylatoxin, is one of the most critical activation products of the human complement system (1). C5a is a 74amino acid-long peptide that is generated upon the enzymatic cleavage of complement component C5 by C5 convertases. Abnormal levels of C5a and subsequent signaling triggered by it are crucial in a range of inflammatory disorders including sepsis, vasculitis, and trauma (1,2). C5a exerts its effects via two seven-transmembrane receptors, namely the C5aR1 and C5aR2 (also known as C5L2) (3). Of these, C5aR1 is a prototypical G-protein-coupled receptor (GPCR) 4 that is expressed in macrophages, neutrophils, and endothelial cells (3). Upon binding of C5a, C5aR1 couples to the G␣ i subtype of heterotrimeric G proteins, resulting in inhibition of cAMP levels and mobilization of intracellular Ca 2ϩ (3). Subsequently, C5a also triggers the phosphorylation of C5aR1 followed by recruitment of ␤-arrestins (␤arrs) and receptor internalization (3).
Structurally, C5a harbors four different helices and connecting loops, and it is stabilized by the formation of three disulfide bonds (4). C5a interacts with C5aR1 through two distinct interfaces: one involves the core of C5a with the N terminus of the receptor, whereas the other involves the C terminus of C5a with the extracellular side of the transmembrane helices of C5aR1 ( Fig. 1A) (5). It has been proposed that the structural determinants for high-affinity binding are provided by the first set of interaction, whereas the second set of interaction is responsible for driving functional responses through the receptor (5). Peptides derived from the C terminus of C5a can bind to C5aR1, albeit with much lower affinity compared with C5a, and they can also trigger functional responses (6,7). Whether such peptides may induce differential coupling of G protein versus ␤arrs and exhibit biased functional responses remains completely unexplored.
Here, we focus on a modified hexapeptide, referred to as C5a pep hereafter (Fig. 1B), which displays the highest binding affinity to C5aR1 among various C5a fragments (7), and char- cro ARTICLE acterize it vis-à-vis C5a with respect to G␣ i and ␤arr coupling, functional outcomes, and cellular responses. In particular, we measure the ability of C5a and C5a pep to inhibit forskolin-induced cAMP as a measure of G␣ i -coupling, inducing ␤arr recruitment and trafficking, receptor endocytosis, ERK1/2 MAP kinase activation, IL-6 release, and neutrophil migration. We identify a significant bias in G␣ i versus ␤arr coupling, endocytosis versus ERK1/2 activation, and IL-6 release versus neutrophil migration between the two ligands. These findings establish a framework for investigating ligand-induced functional bias at C5aR1 and pave the way for subsequent characterization of physiological outcomes arising from such ligands.

C5a pep is a full agonist for G␣ i coupling
Although C5a pep exhibits the highest binding affinity for C5aR1 among the peptides derived from and modified based on the C terminus of C5a, its binding affinity for C5aR1 is still significantly lower than C5a (ϳ70 nM for C5a pep and ϳ1 nM for C5a) (7). We therefore first measured the ability of C5a pep to trigger G␣ i coupling to C5aR1 in HEK-293 cells using the GloSensor assay (8). The cells were stimulated with forskolin to generate cAMP followed by incubation with various doses of C5a and C5a pep . We observed that both C5a and C5a pep inhibited cAMP level to a similar extent at saturating concentrations ( Fig. 1C). As expected, based on their binding affinities for the receptor, C5a pep was ϳ100-fold less potent in cAMP inhibition compared with C5a (IC 50 ϭ ϳ0.26 nM for C5a and IC 50 ϭ ϳ16 nM for C5a pep ). We also measured the efficacy of C5a pep in C5aR1 expressing CHO cells using the LANCE cAMP assay (9) and observed a pattern of efficacy and potency very similar to that in HEK-293 cells (Fig. 1D).

C5a pep is a partial agonist for ␤arr recruitment
Upon C5a stimulation, C5aR1 undergoes phosphorylation and recruits ␤arrs, which is important for receptor desensitization and internalization, similar to other prototypical GPCRs (10,11). Thus, we next measured the ability of C5a pep to induce ␤arr coupling using a standard co-immunoprecipitation assay. There are two isoforms of ␤arrs known as ␤arr1 and 2 that exhibit a significant functional divergence despite a high level of sequence and structural similarity (12). We expressed either ␤arr1 or ␤arr2 with FLAG-tagged C5aR1 in HEK-293 cells and then measured their interaction upon ligand stimulation. As presented in Fig. 2 (A and B), we observed a robust recruitment of both isoforms of ␤arrs upon stimulation of cells with C5a. Figure 1. A modified C terminus C5a peptide, C5a pep , is a full agonist for G␣ i coupling. A, schematic representation of C5a binding to C5aR1. There are two sites of interaction, one involving the N terminus of C5aR1 and the other involving the extracellular loops. B, the primary sequence and modification of C5a pep , which is derived based on the C terminus of C5a. C, C5a pep behaves as a full-agonist in GloSensor-based cAMP assay. HEK-293 cells expressing C5aR1 were transfected with F22 plasmids. 24 h post-transfection, the cells were stimulated with indicated concentrations of C5a and C5a pep followed by recording of bioluminescence readout. D, G␣ i coupling induced by C5a and C5a pep measured in CHO cells using LANCE cAMP assay. CHO cells stably expressing C5aR1 were stimulated with the indicated concentrations of respective ligands. The data presented in C and D represent the means Ϯ S.E. of three to five independent experiments, and the EC 50 values of C5a and C5a pep were analyzed using unpaired t test. ***, p Ͻ 0.001.

Ligand bias at the human complement receptor
Interestingly, the levels of ␤arr recruitment induced by C5a pep were significantly lower compared with C5a, even at saturating ligand concentrations. As a control, we stimulated cells with W54011, an antagonist of C5aR1 (13), and as expected, it did not elicit any significant levels of ␤arr recruitment. To probe whether there may be a temporal difference in C5aR1-␤arr interaction pattern for C5a versus C5a pep , we further carried out a time-course experiment for ␤arr recruitment. However, C5a pep -induced ␤arr recruitment was significantly lower than C5a (Fig. 2, C and D). Taken together with the cAMP data presented above, these findings suggest that C5a pep is a fullagonist for G␣ i -coupling but a partial agonist for ␤arr recruitment.

C5a pep triggers slower endosomal trafficking of ␤arrs
␤arrs are normally distributed in the cytoplasm and upon agonist stimulation; they traffic to the membrane and interact with receptors (14). Subsequently, upon prolonged agonist stimulation, ␤arrs either dissociate from the receptor (class A pattern of ␤arr recruitment) or co-internalize with activated receptors in endosomal vesicles (class B pattern of ␤arr recruitment) (14). To probe whether C5a pep might differ from C5a with respect to ␤arr trafficking patterns, we co-expressed C5aR1 with either ␤arr1-YFP or ␤arr2-YFP and visualized the trafficking of ␤arrs using confocal microscopy. We observed that C5a pep was capable of promoting surface translocation of ␤arrs at similar levels as C5a during the early phase of agonist stimulation (Fig. 3, second column). However, we observed that C5a pep was significantly slower in promoting the appearance of ␤arrs in endosomal punctae and vesicles compared with C5a ( Fig. 3, third column), although ultimately it did induce robust endosomal localization of ␤arrs (Fig. 3, fourth column).

C5a pep exhibits a bias between ERK1/2 MAP kinase activation and receptor endocytosis
To probe whether C5a pep may also exhibit differential efficacy compared with C5a in functional assays, we next measured the ability of C5a pep to induce receptor endocytosis and ERK1/2 MAP kinase activation in HEK-293 cells. In agreement with our data on ␤arr recruitment and trafficking, we observed lower levels of receptor endocytosis induced by C5a pep compared with C5a (Fig. 4A). This observation hints that ␤arrs may play a crucial role in endocytosis of C5aR1, and therefore, weaker ␤arr recruitment by C5a pep translates into lower endocytosis. Interestingly however, C5a pep was as efficacious as C5a in stimulating phosphorylation of ERK1/2 MAP kinase in HEK-293 cells, at least at the time points that were tested in this experiment (Fig. 4, B and C). Thus, correlation of maximal levels of endocytosis triggered by C5a and C5a pep with the ERK1/2 phosphorylation reveals a bias of C5a pep in these two functional responses. Full efficacy of C5a pep for ERK1/2 phosphorylation, similar to that in cAMP assay, suggests that this response may be primarily driven by G␣ i . To probe this possibility, we measured ERK1/2 phosphorylation after pretreatment of cells with , 10 M) followed by cross-linking using DSP. Subsequently, FLAG-C5aR1 was immunoprecipitated using anti-FLAG antibody agarose, and co-elution of ␤arrs was visualized using Western blotting. B, densitometry-based quantification of data presented in A (averages Ϯ S.E.; n ϭ 2) normalized with respect to signal for C5a-␤arr1 condition (treated as 100%) and analyzed using two-way ANOVA. ***, p Ͻ 0.001. C, a time-course co-immunoprecipitation experiment to measure the interaction of C5aR1 with ␤arr2. The experiment was performed following the protocol as indicated except that cells were stimulated for different time points. D, densitometry-based quantification of the data presented in C (average Ϯ S.E.; n ϭ 2) normalized with respect to signal for C5a at 30 min condition (treated as 100%) and analyzed using two-way ANOVA. *, p Ͻ 0.05.

Ligand bias at the human complement receptor
pertussis toxin (PTX) and observed a robust inhibition of ERK1/2 phosphorylation for both C5a and C5a pep (Fig. 5

, A-D).
Previous studies on several G␣ i -coupled receptors have documented an interplay of ␤arrs and G␣ i in agonist-induced ERK1/2 phosphorylation. For example, nicotinic acid induced ERK1/2 phosphorylation downstream of GPR109A can be inhibited by either PTX pretreatment or ␤arr knockdown (15). Similarly, carvedilol-induced ERK1/2 phosphorylation downstream of ␤1-adrenergic receptor requires a contribution from both G␣ i and ␤arrs (16,17). This interesting aspect of GPCR signaling is being actively explored and discussed in the literature (18 -22). Although our data shows nearly complete inhibition of C5a/C5a pep -induced ERK1/2 phosphorylation for C5aR1, evaluating the contribution of ␤arrs in this process would be an interesting avenue for future studies.

C5a pep exhibits ␤arr isoform bias at a chimeric C5aR1
The interaction of ␤arrs with GPCRs is a biphasic process, which involves the receptor tail (i.e. phosphorylated C terminus) and the receptor core (cytoplasmic surface of the transmembrane bundle) (11,23). Based on the stability of their interaction with ␤arrs, GPCRs are categorized as class A and B, which represent transient and stable interactions, respectively (14). Receptors having clusters of phosphorylatable residues in their C terminus (such as the vasopressin receptor, V2R) typically interact stably with ␤arrs. Because C5aR1 does not harbor

Ligand bias at the human complement receptor
such clusters, we generated a chimeric C5aR1, where we grafted the C terminus of the V2R on C5aR1 and refer to this construct as C5a-V2R (Fig. 6A). We first measured the cAMP response of C5a-V2R upon stimulation with C5a and C5a pep and observed a similar efficacy and potency profile as for the WT C5aR1 (Fig.  6B). Similarly, we also observed comparable levels of ERK1/2 phosphorylation induced by both C5a and C5a pep downstream of C5a-V2R (Fig. 6, C and D). We next measured the interaction of ␤arrs with C5a-V2R and expectedly observed an enhanced ␤arr recruitment for C5a-V2R compared with C5aR1 (Fig. 7, A and B). Interestingly, however, unlike C5aR1, we observed an equal recruitment of ␤arr1 by C5a pep and C5a with C5a-V2R (Fig. 7, C and D). On the other hand, similar to C5aR1, ␤arr2 recruitment induced by C5a pep was still significantly weaker than C5a at this chimeric receptor (Fig. 7, A and B). These findings reveal a ␤arr isoform recruitment bias at the chimeric receptor, a pattern that has significant implications if applicable to other GPCRs as well. For example, several high-throughput screening assays of ␤arr recruitment, such as the Tango assay, utilize chimeric GPCRs with V2R tail (24). This may result in miscalculation of ligand bias, and therefore, the patterns observed using chimeric receptors in primary screening should be reconfirmed using WT receptors for calculating bias profile.

C5a pep is a partial agonist for ␤arr recruitment at C5aR2
C5a interacts with two distinct seven-transmembrane receptors, C5aR1 and C5aR2. Of these, C5aR2 does not exhibit any detectable G-protein coupling as measured in functional assays, although it robustly recruits ␤arrs upon agonist-stimulation (9,25,26). In line with previous reports in the literature, comparison of the primary sequences of C5aR1 and C5aR2 identifies mutations in highly conserved DRY and NPXXY motifs (Fig. 8A). We observed that C5a pep acts as a partial agonist for ␤arr2 recruitment at C5aR2, similar to C5aR1 (Fig. 8B). Interestingly, however, unlike C5aR1, we did not observe agonist-induced phosphorylation of ERK1/2 downstream of C5aR2 in HEK-293 cells (Fig. 8, C and D). Although the mutations in the DRY and NPXXY motifs may be the primary determinants for the lack of functional G-protein coupling, it is also plausible C5aR2 adopts an active conformation upon agonist binding that is distinct from that of C5aR1. Such a conformation may preclude functional G-protein coupling but allow receptor phosphorylation and ␤arr interaction. However, future studies designed to investigate agonist-dependent conformational changes in C5aR1 and C5aR2 are required to probe such a possibility.

C5a pep elicits biased cellular responses
C5aR1 is endogenously expressed at high levels in macrophages and neutrophils, where it modulates multiple inflammatory responses (3). Stimulation of C5aR1 in human macrophages reduces LPS-induced release of IL-6 (2), whereas neutrophil C5aR1 activation induces rapid chemotaxis (3). To assess whether C5a pep might exhibit a bias at the level of these cellular responses, we utilized primary human monocyte-derived macrophages (HMDMs) and human polymorphonuclear neutrophils (hPMNs) to measure IL-6 release and migration, respectively. Measuring G i -mediated cAMP responses in these primary cells is technically challenging, and therefore, we measured Ca 2ϩ mobilization in HMDMs upon stimulation by C5a and C5a pep . We observed a full-agonist profile of C5a pep similar to that observed in cAMP assays in HEK-293 and CHO cells, and the potency of these two agonists for Ca 2ϩ mobilization was also similar to that observed in HEK-293 and CHO cells (Fig. 9A). Quantification and comparison of C5a and C5a pep for

Ligand bias at the human complement receptor
inducing ␤arr recruitment in HMDMs and hPMNs at endogenous level of the receptor are technically very challenging and require development of novel sensors and assays going forward. Interestingly, however, C5a pep stimulation resulted in only a submaximal phosphorylation of ERK1/2 in HMDMs compared with C5a (Fig. 9B). This is in contrast with our observation in HEK-293 cells in which C5a pep stimulated ERK1/2 phosphorylation at levels similar to C5a (Fig. 4, B and C). We also found that ERK1/2 phosphorylation in HMDMs was sensitive to PTX pretreatment as in HEK-293 cells (Fig. 9C). This is particularly interesting considering that C5a pep behaves as full agonist in Ca 2ϩ mobilization experiments in HMDMs. There is growing evidence for context-specific effector coupling and functional responses downstream of several GPCRs (27,28). These emerging findings have refined our current understanding of biased signaling by providing substantial evidence of additional levels of complexities in GPCR signaling. Our data with C5a pep , especially in the context of ERK1/2 activation in HMDMs, further add to this important paradigm of GPCR signaling.
We next compared the ability of C5a and C5a pep to inhibit LPS-induced IL-6 release in HMDMs and to stimulate chemotaxis in hPMNs. We observed that the inhibition of LPSinduced IL-6 release in HMDMs was comparable for both C5a and C5a pep (Fig. 10A) and was sensitive to PTX pretreatment (Fig. 10B). Notably, however, we observed that C5a pep displayed a significantly blunted response in neutrophil chemotaxis compared with C5a even at saturating doses (Fig. 10C), although similar to IL-6 release, chemotaxis was also sensitive to PTX. These observations therefore uncover that C5a pep exhibits bias at the level of cellular responses when compared with C5a in primary cells expressing endogenous levels of C5aR1. It is inter-esting to note here that both LPS-induced IL-6 release and hPMN chemotaxis are sensitive to PTX pretreatment, suggesting that these processes are driven primarily by G␣ i coupling to C5aR1. However, because ␤arr knockdown or knockout studies in HMDMs and hPMNs are technically challenging and typically have suboptimal efficiency, we cannot rule out either a direct contribution of ␤arrs or an interplay of ␤arrs and G␣ i in these processes, and it remains to be explored in future studies (16).
It is important to note that chemokines, like complement C5a, also interact with their cognate GPCRs through a biphasic mechanism. Thus, it is tempting to speculate that fragments derived from the C terminus of chemokines may also exhibit biased signaling through their receptors. Our findings also suggest that a more comprehensive analysis of C5a fragments might yield additional ␤arr-biased ligands at C5aR1. Although crystal structures of C5aR1 bound to small molecule antagonists have been determined recently (29,30), a C5a-bound structure is still not available. Future high-resolution data of C5a-bound C5aR1 may provide structural insights into differential engagement of C5a pep compared with C5a and how these differential interactions in the ligand-binding pocket yield transducer-coupling bias.
In summary, we discover C5a pep as a biased C5aR1 agonist at the levels of G␣ i versus ␤arr coupling, functional outcomes, and cellular responses. Going forward, an interesting avenue might be to evaluate the physiological responses elicited by C5a pep in vivo. Given that C5a attenuates LPS-mediated cytokine production from macrophages (31) (Fig. 10), a biased ligand such as C5a pep that retains this beneficial activity, although diminishing the more pro-inflammatory activities of neutrophil migra- Figure 5. C5a/C5a pep -induced ERK1/2 phosphorylation is sensitive to PTX treatment. A, HEK-293 cells stably expressing C5aR1 were incubated with 100 ng/ml PTX for 12-16 h followed by serum starvation and ligand stimulation. Subsequently, ERK1/2 phosphorylation was measured by Western blotting. B, densitometry-based quantification of data presented in A from three independent experiments normalized with respect to maximal response (treated as 100%) and analyzed using one-way ANOVA. ***, p Ͻ 0.001. C, effect of PTX treatment on C5a pep -induced ERK1/2 phosphorylation measured as in A above. D, densitometry-based quantification of data presented in C, normalized, and analyzed as in B.

Ligand bias at the human complement receptor
tion, may be a novel therapeutic approach for treating inflammatory disorders. Furthermore, as we have demonstrated the recruitment of both isoforms of ␤arr (i.e. ␤arr1 and 2) to C5aR1, it would be interesting to evaluate the shared and distinct roles of these ␤arr isoforms in regulation of C5aR1. It is also notable that the second receptor activated by C5a (C5aR2) does not exhibit any detectable G-protein coupling but displays robust ␤arr recruitment (25). Although we did not observe agonistinduced ERK1/2 phosphorylation downstream of C5aR2 upon C5a stimulation, it is plausible that it may activate other signaling pathways, and a comprehensive study is required to probe this possibility in future.

General reagents, constructs, and cell culture
Most of the reagents were purchased from Sigma unless mentioned otherwise. The coding region of human C5aR1 was cloned in pcDNA3.1 vector with the N-terminal signal sequence and a FLAG tag. Coding regions of bovine ␤arr1 and ␤arr2 were cloned in pCMV vector. C5a pep was synthesized from Genscript. Recombinant human C5a was either purchased from Sino Biological or purified following a previously published protocol (4). Ultrapure lipopolysaccharide from Escherichia coli K12 strain was purchased from Invivogen. BSA was purchased from Sigma. For cell culture, trypsin-EDTA, Hanks' balanced salt solution (HBSS), HEPES, Dulbecco's modified Eagle's medium (DMEM), phenol-red free DMEM, Ham's F-12 medium, Iscove's modified Dulbecco's medium (IMDM), and penicillin/streptomycin were purchased from Thermo Fisher Scientific. Dulbecco's PBS (DPBS) was purchased from Lonza.
To generate HMDM, human buffy coat blood from anonymous healthy donors was obtained through the Australian Red Cross Blood Service. Human CD14ϩ monocytes were isolated from blood using Lymphoprep density centrifugation (STEMCELL) followed by CD14ϩ MACS magnetic bead separation (Miltenyi Biotec). The isolated monocytes were differentiated for 7 days in IMDM supplemented with 10% FBS, 100 units/ml penicillin, 100 g/ml streptomycin, and 15 ng/ml recombinant human macrophage colony stimulating factor (Peprotech) on 10-mm square dishes (Sterilin). Nonadherent cells were removed by washing with DPBS, and the adherent differentiated HMDMs were harvested by gentle scraping.
hPMNs were obtained from venous whole blood (20 ml) collected from healthy volunteers under informed consent. The samples were collected using venepuncture into BD K2EDTA Vacutainer blood collection tubes and processed within 5 h. For neutrophil isolation, the anticoagulated blood was firstly layered over a Lymphoprep (STEMCELL) density gradient and centrifuged (800 ϫ g for 30 min at 22°C), followed by residual erythrocytes removal using hypotonic lysis. Isolated PMNs were counted and resuspended in a HBSS-based migration buffer (containing calcium and magnesium, supplemented with 20 mM HEPES and 0.5% BSA).

Preparation of C5aR1, C5a-V2R, and C5aR2 expressing stable HEK-293 cell line
50 -60% confluent HEK-293 cells were transfected with 7 g of FLAG-tagged C5aR1/C5a-V2R/C5aR2 DNA complexed with 21 g of polyethylenimine (PEI). Next day, stable selection was started with optimal dose of G418 along with untransfected cells kept as negative control. After completion of stable selection, clonal population was prepared by the limited dilution method. The highest expressing clones were propagated further and kept under G418 selection throughout the course of experiments. Surface expression of C5aR1, C5a-V2R, and C5aR2 was measured using a previously described whole-cell surface ELISA protocol (32).

ERK1/2 phosphorylation assay
Agonist-induced ERK1/2 phosphorylation was measured primarily using a previously described Western blotting-based protocol (33). C5aR1-, C5a-V2R-, or C5aR2-expressing stable cell lines were seeded into 6-well plate at a density of 1 million cells/well. The cells were serum-starved for 12 h followed by stimulation with 1 M of C5a and 10 M of C5a pep , respectively, at selected time points. After the completion of time course, the Figure 7. C5a pep induces differential recruitment of ␤arr1 and 2 at C5aV 2 R. A, HEK-293 cells expressing FLAG-C5a-V2R and ␤arr2 were stimulated with saturating concentrations of different ligands followed by cross-linking using DSP. Subsequently, FLAG-C5a-V2R was immunoprecipitated using anti-FLAG antibody agarose, and co-elution of ␤arr2 was visualized using Western blotting. B, densitometry-based quantification of data presented in A (averageϮS.E.; n ϭ 2) normalized with respect to signal for C5a-␤arr2 30 min condition (treated as 100%) and analyzed using two-way ANOVA. **, p Ͻ 0.01; *, p Ͻ 0.05. C, interaction of C5aR1 or C5a-V2 2 R with ␤arr1 measured essentially in similar fashion as described in A above. D, densitometry-based quantification of data presented in C (averages Ϯ S.E.; n ϭ 3) normalized with respect to signal for C5a-C5aR1 condition (treated as 100%) and analyzed using two-way ANOVA. **, p Ͻ 0.01.

Ligand bias at the human complement receptor
Ligand bias at the human complement receptor medium was aspirated, and the cells were lysed in 100 l of 2ϫ SDS dye/well. The cells were heated at 95°C for 15 min followed by centrifugation at 15,000 rpm for 10 min. 10 l of lysate was loaded per well and separated on SDS-PAGE followed by Western blotting. The blots were blocked in 5% BSA (in TBST) for 1 h and incubated overnight with rabbit phospho-ERK (catalog no. 9101/CST) primary antibody at 1:5000 dilution. The blots were washed thrice with TBST for 10 min each and incubated with anti-rabbit HRP-coupled secondary antibody (1:10,000, catalog no. A00098/Genscript) for 1 h. The blots were washed again with TBST for three times and developed with Promega ECL solution on chemidoc (Bio-Rad). The blots were stripped with low pH stripping buffer and then reprobed for total ERK using rabbit total ERK (catalog no. 9102/CST) primary antibody at 1:5000 dilution. To measure the effect of PTX on ERK activation, HEK-293 cells stably expressing C5aR1 were seeded in 6-well plate. The cells were treated with 100 ng/ml PTX for 12-16 h followed by serum starvation. The cells were stimulated with C5a and C5a pep for the indicated time points. The samples were prepared and processed as mentioned previously.
The ligand-induced phospho-ERK1/2 signaling in HMDMs was assessed using the AlphaLISA Surefire Ultra p-ERK1/2 (Thr 202 /Tyr 204 ) kit (PerkinElmer) following the manufacturer's protocol. Briefly, HMDMs were seeded (50,000/well) in tissue culture-treated 96-well plates (Corning) for 24 h and serumstarved overnight. All ligand dilutions were prepared in serumfree medium containing 0.1% BSA. For stimulation, the cells were incubated with respective ligands for 10 min at room temperature and then immediately lysed using AlphaLISA lysis buffer on a microplate shaker (450 rpm, 10 min). For the detection of phospho-ERK1/2 content, cell lysate (5 l/well) was transferred to a 384-well ProxiPlate (PerkinElmer) and added to the donor and acceptor reaction mix (2.5 l/well, respectively) with 2-h incubation at room temperature in the dark. On a Tecan Spark 20M, following laser irradiation of donor beads at 680 nm, the chemiluminescence of acceptor beads at 615 nm was recorded. To measure the effect of PTX, culture medium was changed into serum-free IMDM containing 200 ng/ml PTX or vehicle only for overnight incubation. Afterward, ERK1/2 phosphorylation was measured as described above.

Receptor internalization assay
50 -60% confluent HEK-293 cells were transfected with 3.5 g of FLAG-tagged C5aR1 DNA by polyethylenimine method of transfection at DNA:PEI ratio of 1:3. 24 h post-transfection, 0.15 million cells/well were seeded in 24-well plate (precoated with 0.01% poly-D-lysine). After 24 h, the cells were serumstarved for 6 h followed by stimulation with 1 M of C5a and 10 M C5a pep , respectively, for selected time points. After stimu-lation, the cells were washed once with ice-cold 1ϫ TBS. The cells were then fixed with 4% (w/v) paraformaldehyde on ice for 20 min and washed thrice with TBS to remove paraformaldehyde. Blocking was done with 1% BSA prepared in 1ϫ TBS for 1.5 h. This was followed by incubation of cells with HRP-conjugated anti-FLAG M2 antibody (Sigma) at a dilution of 1:2000 prepared in 1% BSA ϩ 1ϫ TBS for 1.5 h. Afterward, the cells were washed thrice with 1% BSA ϩ 1ϫ TBS. Surface expression was measured by incubating cells with 200 l of 3,3Ј,5,5Ј-tetramethylbenzidine (Genscript) per well, and reaction was stopped by transferring 100 l of developed colored solution to a 96-well plate already containing 100 l of 1 M H 2 SO 4. Absorbance was read at 450 nm in a multiplate reader (Victor X4). For normalization, cell density was measured using janus green. Briefly, 3,3Ј,5,5Ј-tetramethylbenzidine was removed, and the cells were washed twice with 1ϫ TBS followed by incubation with 0.2% (w/v) janus green for 15 min. Destaining was done with three washes of 1 ml of distilled water. Stain was eluted by adding 800 l of 0.5 N HCl per well. 200 l of this solution was transferred to a 96-well plate, and absorbance was read at 595 nm. Data normalization was done by dividing A 450 by A 595 values.

GloSensor assay for cAMP measurement
50 -60% confluent HEK-293 cells were co-transfected with 3.5 g each of C5aR1 or C5a-V2R and 22F (Promega) plasmids. 24 h post-transfection, the cells were trypsinized and harvested by centrifugation at 1500 rpm for 10 min. The medium was aspirated, and cells were resuspended in luciferin sodium solution (0.5 mg/ml) (Gold Biotech) prepared in 1ϫ HBSS (Gibco) and 20 mM HEPES, pH 7.4. The cells were then seeded in a 96-well plate at a density of 0.4 million cells/well and kept at 37°C for 1.5 h in the CO 2 incubator followed by incubation at room temperature for 30 min. Basal reading was read on luminescence mode of multiplate reader (Victor X4), and cycles were adjusted until basal values were stabilized. The cells were then incubated with 1 M forskolin, and readings were recorded until maximum luminescence values were obtained. This was followed by stimulation of cells with specified concentrations of C5a and C5a pep , and values were recorded for 1 h. The data were normalized with respect to minimal stimulation dose of ligand after basal correction.

Cross-linking and co-immunoprecipitation
50 -60% confluent HEK-293 cells were co-transfected with C5aR1 or C5a-V2R and ␤arr1/␤arr2 plasmids by PEI (as mentioned earlier). 48 h post-transfection, the cells were serumstarved for 6 h and stimulated with respective doses of C5a/ C5a pep , harvested, and proceeded for cross-linking experiment. The cells were lysed by Dounce homogenization in 20 mM Figure 8. Sequence alignment of C5aR2 with C5aR1, ␤arr2 recruitment, and ERK1/2 phosphorylation. A, the sequences of human C5aR1 and C5aR2 were retrieved from Uniprot and aligned on M-coffee server with default parameters. Alignment reliability was assessed by core/TCS and generated alignment was visualized using Espript 3. Specific mutations in the DRY and NPXXY motif are highlighted. B, HEK-293 cells expressing C5aR2-Venus and ␤arr2-Rluc8 constructs were first incubated with luciferase-substrate for 2 h. Subsequently, the cells were stimulated with respective ligands, and BRET signals were monitored using a dose-response curve. The data represent averages Ϯ S.E. of three independent experiments, and the EC 50 values are compared using unpaired t test. ***, p Ͻ 0.001. C, HEK-293 cells expressing C5aR1 or C5aR2 were stimulated with C5a (100 nM) for indicated time points followed by detection of ERK1/2 phosphorylation using Western blotting. D, densitometry-based quantification of data presented in C normalized with C5a response for C5aR1 (treated as 100%) and analyzed using two-way ANOVA. ***, p Ͻ 0.001.

Sequence alignment of C5aR1 and C5aR2 sequences
Reviewed sequences of human C5aR1 and C5aR2 (also referred to as C5L2) were retrieved from Uniprot. Sequences were aligned on M-coffee server with default parameters. Alignment reliability was assessed by core/TCS, and generated alignment was visualized using Espript 3.

BRET assay for measuring the interaction of ␤arr2 with C5aR2
C5a-mediated ␤arr2 recruitment to C5aR2 was measured using bioluminescent resonance energy transfer (BRET) as previously described (9). Briefly, HEK-293 cells were transiently transfected with C5aR2-Venus and ␤arr2-Rluc8 constructs using XTG9 (Roche). At 24 h post-transfection, the cells were gently detached using 0.05% trypsin-EDTA and seeded (100,000/well) onto white 96-well TC plates (Corning) in phenol-red free DMEM containing 5% FBS. On the following day, the cells were firstly incubated with the substrate EnduRen (30 M; Promega) for 2 h (37°C, 5% CO 2 ). On a Tecan Spark 20M microplate reader (37°C), the BRET light emissions (460 -485 and 520 -545 nm) were continuously monitored for 25 reads with respective ligands added after the first 5 reads. The ligandinduced BRET ratio was calculated by subtracting the Venus (520 -545 nm) over Rluc8 (460 -485 nm) emission ratio of the vehicle-treated wells from that of the ligand-treated wells.

Confocal microscopy
For visualization of ligand-induced ␤arr recruitment, HEK-293 cells were co-transfected with C5aR1 and ␤arr1-YFP or ␤arr2-YFP plasmids in 1:1 ratio (total 7 g) by PEI. 24 h posttransfection, 1 million cells were seeded in 35-mm glass bottom dish precoated with 0.01% poly-D-lysine. After 24 h, the cells were serum-starved for 6 h and stimulated with respective doses of C5a (1 M) and C5a pep (10 M). For live-cell imaging, images were acquired using Carl Zeiss LSM780NLO confocal microscope for specified time intervals, and image processing was done in ZEN lite (Zen-blue/ZEN-black) software from Zeiss. Confocal microscopy experiments were performed on Figure 9. C5a pep exhibits bias at the level of Ca 2؉ mobilization and ERK1/2 phosphorylation in HMDMs. A, C5a pep is a full agonist in Ca 2ϩ mobilization assay in HMDM cells. HMDMs were first loaded with Fluo-4 calcium indicator followed by the addition of respective ligands and subsequent measurement of fluorescence intensity. The data were normalized with maximal response obtained for C5a and represent means Ϯ S.E. of triplicate experiments performed from three independent donors, and the EC 50 values of C5a and C5a pep were analyzed using unpaired t test. ***, p Ͻ 0.001. B, HMDMs were stimulated with respective ligands for 10 min at room temperature before being lysed. The phospho-ERK1/2 content in the lysate was detected using AlphaLISA Surefire Ultra p-ERK1/2 kit. The data were normalized with maximal response obtained for C5a and represent the means Ϯ S.E. of triplicate experiments performed from five independent donors, and the EC 50 values of C5a and C5a pep were analyzed using unpaired t test. ****, p Ͻ 0.0001. C, C5a/C5a pep -induced ERK1/2 phosphorylation in HMDMs is sensitive to PTX. HMDMs seeded in tissue culture-treated 96-well plates were preincubated with 200 ng/ml PTX or vehicle in serum-free IMDM overnight. Subsequently, the cells were stimulated with C5a/C5a pep , and ERK1/2 phosphorylation was measured using AlphaLISA Surefire Ultra p-ERK1/2 kit. The data are normalized with maximal response obtained for C5a and represent the means Ϯ S.E. of triplicate experiments performed from three independent donors and analyzed using paired two-way ANOVA. ***, p Ͻ 0.001; ****, p Ͻ 0.0001. CTL, control.

Ligand bias at the human complement receptor
two independent set of cells (i.e. independent transfections), and multiple cells were imaged and analyzed.

Intracellular calcium mobilization assays
Ligand-induced intracellular calcium mobilization was assessed using Fluo-4 NW Calcium Assay kit (Thermo Fisher Scientific) following the manufacturer's instructions. Briefly, HMDMs were seeded (50,000/well) in black clear-bottom 96-well TC plates (Corning) for 24 h before the assay. The cells were first stained with the Fluo-4 dye in assay buffer (1ϫ HBSS, 20 mM HEPES) for 45 min (37°C, 5% CO 2 ). Respective ligands were prepared in assay buffer containing 0.5% BSA. On a Flexstation 3 platform, the fluorescence (excitation/emission, 494/ 516 nm) was continually monitored for a total of 100 s with ligand addition performed at 16 s.

Chemotaxis assays
Ligand-induced hPMN migration was assessed using 6.5-mm Transwell polycarbonate membrane inserts with 3.0-m pore (Corning) to create a modified Boyden chamber (31). Freshly isolated hPMNs were seeded onto inserts (500,000/well) for 20 min (37°C, 5% CO 2 ) in a HBSS-based migration buffer as described in previous section. To initiate cell migration, respective ligands prepared in migration buffer were added to the receiver wells in duplicates. After 60-min migration (37°C, 5% CO 2 ), the inserts were gently washed once with DPBS, and the residual cells on the upper side of the membrane were removed using a cotton swab. Migrated cells were detached by adding 500 l/well Accumax solution (Invitrogen) to the receiver wells (10 min, room temperature) and then counted using a Bio-Rad TC20 TM automated cell counter. For PTX treatment experiments, freshly isolated hPMNs were incubated with 3 g/ml pertussis toxin (Tocris) or vehicle only and seeded onto inserts (500,000/well) (37°C, 5% CO 2 ) in migration buffer. Upon 4-h incubation with PTX, cell migration was measured as described above.

Measurement of cytokines release using ELISA
The immunomodulatory effect of respective C5aR1 ligands on LPS-induced cytokine release was assessed in primary human macrophages as previously described (34). HMDMs were seeded in 96-well TC plates (100,000/well) for 24 h before Figure 10. C5a pep exhibits bias between LPS-induced IL-6 release and neutrophil migration. A, C5a pep behaves as a full-agonist for lowering of LPSinduced IL-6 release in HMDM cells. HMDMs were stimulated using 10 ng/ml LPS in the absence or presence of C5a or C5a pep . Subsequently, the levels of IL-6 present in the supernatant after 24 h of stimulation were quantified using ELISA, background-corrected with the values obtained for serum-free medium/BSA, and normalized with maximal response for LPS (i.e. treated as 100%). The data represent normalized means Ϯ S.E. of triplicate measurements conducted in cells from four independent donors, analyzed using two-way ANOVA. *, p Ͻ 0.05; ****, p Ͻ 0.0001. B, LPS-induced IL-6 response is sensitive to PTX. HMDMs seeded in 96-well plates were incubated overnight with 200 ng/ml PTX or vehicle followed by stimulation with respective ligands. Subsequently, the IL-6 content in the supernatant was quantified using ELISA kits. The data are presented, normalized, and analyzed as in A above. C, C5a pep elicits only partial-agonist response in migration of human PMNs. Freshly isolated hPMNs seeded into Transwell inserts were stimulated with respective ligands added to the receiver wells and then allowed to migrate for 1 h. The number of migrated cells was recorded and normalized to the maximal C5a-induced migration. The data represent the means Ϯ S.E. of triplicate measurements conducted in cells from three independent donors. D, C5a pep -induced neutrophil migration is sensitive to PTX. Freshly isolated hPMNs were preincubated with 3 g/ml PTX or vehicle for 4 h, and cell migration was initiated by adding respective ligands. Ligand-induced migration was measured as in C, and the data are normalized with control (i.e. no stimulation) and analyzed suing two-way ANOVA. *, p Ͻ 0.05. CTL, control.
Ligand bias at the human complement receptor treatment. All ligands were prepared in serum-free IMDM containing 0.1% BSA. For stimulation, the cells were co-treated with LPS and respective C5aR1 ligands for 24 h (37°C, 5% C0 2 ). The supernatant was collected and stored at Ϫ20°C till use. IL-6 levels in the supernatant were quantified using respective human ELISA kits (BD OptEIA) as per the manufacturer's protocol.

Data collection, processing, and analysis
All experiments were conducted in triplicate and repeated on at least three separate days (for cell lines) or using cells from at least three donors (for HMDMs) unless otherwise specified. The data were analyzed using GraphPad software (Prism 8.0) and expressed as means Ϯ S.E. The data from each repeat were normalized accordingly before being combined. For all doseresponse assays, logarithmic concentration-response curves were plotted using combined data and analyzed to determine the respective potency values.