Ticks from diverse genera encode chemokine-inhibitory evasin proteins

To prolong residence on their hosts, ticks secrete many salivary factors that target host defense molecules. In particular, the tick Rhipicephalus sanguineus has been shown to produce three salivary glycoproteins named “evasins,” which bind to host chemokines, thereby inhibiting the recruitment of leukocytes to the location of the tick bite. Using sequence similarity searches, we have identified 257 new putative evasin sequences encoded by the genomes or salivary or visceral transcriptomes of numerous hard ticks, spanning the genera Rhipicephalus, Amblyomma, and Ixodes of the Ixodidae family. Nine representative sequences were successfully expressed in Escherichia coli, and eight of the nine candidates exhibited high-affinity binding to human chemokines. Sequence alignments enabled classification of the evasins into two subfamilies: C8 evasins share a conserved set of eight Cys residues (four disulfide bonds), whereas C6 evasins have only three of these disulfide bonds. Most of the identified sequences contain predicted secretion leader sequences, N-linked glycosylation sites, and a putative site of tyrosine sulfation. We conclude that chemokine-binding evasin proteins are widely expressed among tick species of the Ixodidae family, are likely to play important roles in subverting host defenses, and constitute a valuable pool of anti-inflammatory proteins for potential future therapeutic applications.

alus pulchellus (23) and Rhipicephalus appendiculatus (24)). Furthermore, saliva from several species of Ixodidae contains chemokine-binding activity suggested to be similar to that of evasin-3 (25). Thus, it is likely that evasins are ubiquitously expressed by a wide variety of tick species.
In addition to their functions in host defense, the interactions of chemokines with their receptors also contribute to excessive leukocyte recruitment in many inflammatory diseases (7,8). Thus, chemokine receptors and to a lesser extent chemokines themselves are popular targets for the development of antiinflammatory therapeutics (26,27). To this end, Proudfoot and co-workers (16, 28 -30) have demonstrated that R. sanguineus evasins are efficacious anti-inflammatory agents in animal models of psoriasis (16), lung injury (16,28), colitis (29), atherosclerosis (30), and ischemic stroke (30).
Motivated by the potential of evasins to inhibit chemokinemediated inflammation in other diseases, we have explored tick genomic and transcriptomic databases to identify homologues of evasin-1 and evasin-4. Here we describe the identification of more than 250 putative tick evasins, spanning three genera of hard ticks (Rhipicephalus, Amblyomma, and Ixodes), and the validation of representative members of this protein family as binders and inhibitors of human chemokines. We discuss the implications of our findings for evasin evolution and for future development of chemokine-targeted anti-inflammatory therapeutics.

Identification of evasin candidates from gene and transcript databases
To identify predicted tick proteins with high sequence identity to R. sanguineus evasin-1 and -4, we initially searched the UniProtKB protein sequence database as well as the tick genomic and transcriptomic data in VectorBase (Fig. 1, Phase 1). UniProtKB searches yielded 133 protein sequences with 9 -42% sequence identity to R. sanguineus evasin-1 and -4. These sequences were derived from one Rhipicephalus species (R. pulchellus) and five Amblyomma species (A. americanum, A. cajennense, A. maculatum, A. parvum, and A. triste).
VectorBase searches yielded 24 sequences from the species Ixodes ricinus and one putative evasin from Ixodes scapularis with relatively low (9 -20%) sequence identity to R. sanguineus evasin-1 and -4. To identify additional evasin candidates, we used the aligned sequences of the initial hits to generate a hidden Markov model (HMM) 4 with which we then interrogated four databases: the UniProtKB protein sequence database; the genomic, transcriptomic, and peptide data in VectorBase; data from the NCBI Transcriptome Shotgun Sequence Assembly Database; and salivary gland and viscera transcriptomes from the cattle tick Rhipicephalus microplus and the Australian paralysis tick Ixodes holocyclus (Fig. 1, Phase 2). In combination with the initial search results, these searches identified a total of 428 sequences, which were curated for conserved features (see below) to yield a final database of 257 putative evasins (accession numbers are listed in supplemental Table S1) with 9 -46% sequence identity to R. sanguineus evasin-1 and -4.

Conserved features of evasin sequences: C 8 and C 6 subfamilies
R. sanguineus evasin-1 and -4 contain several sequence features that are expected to be conserved in orthologous proteins. Eight conserved Cys residues form four disulfide bonds in the structure of evasin-1, most of which should be retained in other evasins. Because evasins are secreted proteins, N-terminal signal sequences are also anticipated to be present. In addition, R. sanguineus evasin-1 and -4 contain three conserved potential N-linked glycosylation sites (evasin-4 contains four additional sites) and have previously been shown to be glycosylated when expressed in both HEK293 cells and TN5 insect cells (15

Tick evasins from diverse genera
and -4 contain tyrosine residues adjacent to several acidic residues (sequences EDDEDY and EEEDDY, respectively), which we identified as likely to be post-translationally sulfated (12,31,32). This feature of evasin sequences has not been identified previously.
All of these sequence features are present in the majority of putative evasin sequences retrieved in the above searches as illustrated for 13 selected sequences in Fig. 2. Although the pattern of Cys residues was completely conserved in putative evasin sequences from Rhipicephalus and Amblyomma species,

Tick evasins from diverse genera
all Ixodes sequences were missing the fifth and/or eighth Cys, which are disulfide-bonded to each other in evasin-1. Interestingly, the Ixodes sequences retain the putative tyrosine sulfation site, but it is located after the first conserved Cys in these sequences in contrast to being before the first conserved Cys in the Rhipicephalus and Amblyomma sequences. Although the effects of these sequence differences on evasin structure and function remain to be determined, the sequence comparisons led us to propose that there are two distinct subfamilies of evasins homologous to R. sanguineus evasin-1 and -4, namely the C 8 evasins (found in Rhipicephalus and Amblyomma) and the C 6 evasins (found in Ixodes species). Both subfamilies are distinct from the evasin-3 type evasins.

Bacterial expression of evasins
Previous studies of recombinant evasin-1 and -4 have used mammalian or insect cell expression systems. To evaluate whether evasins expressed in a bacterial system would be folded and functional, we expressed N-terminally His 6 -tagged R. sanguineus evasin-1 and -4 in Escherichia coli. Evasin-4 was expressed in inclusion bodies but could be renatured under disulfide exchange conditions and purified chromatographically (Fig. 3A). Evasin-1 was expressed poorly and was not pursued further.
Binding of purified evasin-4 to each of several chemokines was detected using a competitive fluorescence anisotropybinding assay (Fig. 3B) (33). In this assay, a fluorescein-labeled peptide derived from the N-terminal region of a chemokine receptor is displaced from the chemokine, resulting in decreased fluorescence anisotropy with increased concentration of evasin. As shown by simulated displacement curves (supplemental Fig. S1), the shape of the curve is indicative of the equilibrium dissociation constant (K d ) between evasin and chemokine, although the concentration of evasin required for 50% displacement of the fluorescent peptide is substantially higher than the K d ; this difference is related to the concentrations of the chemokine and fluorescent peptide used in the assay and their affinity for each other. We observed that His 6 -tagged evasin-4 bound to each of six chemokines tested with K d values ranging from ϳ1 to ϳ20 nM ( Fig. 3B and Table 1). These results indicated that glycosylation and/or tyrosine sulfation is not essential for chemokine binding, consistent with previous results showing that deglycosylation of evasin-1 did not affect activity (15). We concluded that the E. coli system was appropriate for screening novel evasin candidates.
To determine whether the evasin candidates identified above are able to bind chemokines, we expressed 12 selected sequences in E. coli, two each from the species R. pulchellus, A. americanum, A. cajennense, A. triste, and I. ricinus and one each from A. maculatum and A. parvum; evasin candidates are identified by the names listed in Fig. 2 (e.g. ACA-01 and ACA-02 for the two candidates from A. cajennense). Initial expression trials indicated that the majority of candidates were sequestered in inclusion bodies, which were therefore harvested and denatured to allow the His 6 -tagged proteins to be partially purified. After renaturation by rapid dilution and subsequent immobilized metal affinity purification, nine of the 12 candidates (all except APA-01, ATR-01, and IRI-02) were isolated in sufficient quantities to screen for chemokine binding.

Evasin candidates bind to chemokines
The competitive fluorescence anisotropy-binding assay was used to assess the binding of the nine isolated evasin candidates to each of five CC chemokines previously expressed and purified in our laboratory (supplemental Fig. S2). With the exception of ACA-02 (from A. cajennense), all evasin candidates bound to at least one of the chemokines tested with K d values of Ͻ50 nM (Table 2). We conclude that these eight proteins are chemokine-binding evasins and, by extension, that the majority of evasin candidates identified in our sequence-based searches are also likely to be chemokine binders.
We selected one newly identified evasin from each genus (ACA-01, RPU-01, and IRI-01) and an additional putative evasin (IHO-01; Fig. 2) from the Australian paralysis tick I. holocyclus for detailed comparison with R. sanguineus evasin-4 in chemokine-binding and inhibition assays. These proteins were expressed, refolded, and purified by size exclusion chromatography (Fig. 4). Non-reducing SDS-PAGE (Fig. 4) indicated that Tick evasins from diverse genera the purified proteins did not contain intermolecular disulfide bonding, suggesting that they were predominantly correctly folded. These four evasins bound to five CC chemokines tested ( Fig. 4) with the K d values listed in Table 1). Notably, the two C 6 evasins (from Ixodes species) bind with lower affinity than the C 8 evasins (from Rhipicephalus and Amblyomma species) to the chemokines tested here. This may be an indication of differences in target chemokine selectivity, the requirement for post-translational modifications, or simply lower intrinsic affinity.

Evasin candidates inhibit chemokine activity
Chemokines activate their receptors to initiate a variety of G protein-dependent (e.g. inhibition of cAMP synthesis) and G protein-independent (e.g. ␤-arrestin recruitment) signaling pathways. To determine whether the newly identified evasins inhibit chemokine signaling via their cognate receptors, we determined the effects of evasins on the inhibition of cAMP production induced by treatment of CCR2-expressing HEK293 cells with the chemokines monocyte chemoattractant protein-1 (MCP-1/CCL2) and MCP-2/CCL8 or by treatment of CCR3-expressing HEK293 cells with the chemokines eotaxin-1/CCL11 and eotaxin-2/CCL24. As shown in Fig. 5 and Table 3, three of the novel evasins (RPU-01, ACA-01, and IHO-01) were able to inhibit the activity of one or more chemokine tested, although again the evasin from the Ixodes species exhibited relatively low potency. Notably, the selectivities for chemokine inhibition clearly differ among the evasins, suggesting that it may be possible to identify or engineer evasins for desired inhibitory selectivity against target chemokines.

Discussion
We have shown here that the genomes or transcriptomes of several species of hard ticks, spanning the three genera Rhipi-cephalus, Amblyomma, and Ixodes, encode numerous proteins with sequences similar to the previously characterized evasin-1 and -4 from R. sanguineus. Among 12 of these proteins screened, nine were expressed well in E. coli and were sufficiently renatured for binding measurements. Of these nine candidate evasins, eight bound to one or more chemokine(s) tested with affinities of ϳ50 nM or tighter. Moreover, three of the four fully purified evasins tested were found to inhibit chemokine signaling. These results strongly suggest that many of the proteins identified by our bioinformatics approach are indeed chemokine-inhibitory evasins.
During the final revisions of the current manuscript, a study by Singh et al. (34) describing the application of yeast surface display to identify several evasins from ticks in the genera Rhipicephalus and Amblyomma was published. Among the 10 evasins characterized in that study, two were the same as evasins we have characterized, namely ACA-01 (designated by Singh et al. (34) as P974_AMBCA) and RPU-01 (P467_RHIPU).
Considering that the current experiments were limited to a small subset of human chemokines, it seems likely that some of the evasins tested may be more potent inhibitors of different human chemokines or chemokines from other species, such as the natural hosts of the relevant tick species. In addition, it is possible that evasins may inhibit binding of chemokines to not only chemokine receptors but also glycosaminoglycans, which are important for establishment of chemokine gradients required for effective leukocyte trafficking in vivo (35). Of course, it also remains possible that evasins may bind to other host proteins in addition to chemokines. Future studies will be needed to investigate these hypotheses.
The new evasins identified here bound to a panel of related CC chemokines with K d values in the range ϳ5 nM to ϳ1 M. In comparison, R. sanguineus evasin-1 and -4 have been reported  to bind various chemokines with K d values ranging from ϳ30 pM to ϳ200 nM, although the reported values varied somewhat depending on the experimental methods used (13,16) and may also be dependent on expression systems used due to variable post-translational modifications. In addition, the new evasins exhibited distinct binding selectivities to the various chemokines. For example, ACA-01 binds to the chemokines eotaxin-1 and MCP-2 substantially more tightly than to eotaxin-2, eotaxin-3, or MCP-1, whereas RPU-01 has similar affinities for eotaxin-1, eotaxin-3, MCP-1, and MCP-2 and lower affinity for eotaxin-2 (supplemental Fig. S3 and Table 1). The existence of such affinity and selectivity differences is consistent with the sequence variation among the evasins and chemokines tested.
In particular, the evasin sequences differ substantially in the regions that interact with the chemokine in the structure of R. sanguineus evasin-1 bound to human CCL3 ( Fig. 2A) and in the few residues of evasin-4 previously found to contribute to binding of CCL5 (36,37). Similarly, there are numerous differences in the evasin-binding regions of the chemokines, such as the flexible N-terminal region. Future studies will be required to more fully elucidate the structural basis of the selectivity differences.
Despite the sequence variation in chemokine-binding residues, the evasins identified here share several highly conserved features, represented by the sequence motifs in Fig. 2. Like R. sanguineus evasin-1 and -4, most of these proteins (dubbed C 8 evasins) share a distinctive pattern of eight core Cys residues, which form four disulfide bonds in R. sanguineus evasin-1. Ixodes sequences (dubbed C 6 evasins) lack one of these disulfide bonds, suggesting that the missing disulfide may not be critical for folding and stability, but the C 6 evasins also bind more weakly to the chemokines tested here, so the additional disulfide may enhance binding affinity. Additional conserved features are glycine residues, glycosylation sites, and an apparent tyrosine sulfation site, which has not been identified previously.
Future studies will be required to determine whether the conserved post-translational modifications occur in vivo and influence the properties of evasins. Previous studies of the R. sanguineus evasin-1 and -4 in animal models (16,28,29) have utilized proteins produced in eukaryotic cell lines and therefore presumably decorated heterogeneously with various posttranslational modifications. We have shown here that homogeneous evasins produced in E. coli retain the ability to bind chemokines with high affinity, suggesting that the post-translational modifications are not essential for chemokine binding. However, it remains possible that these modifications influence such properties as binding affinity, selectivity, stability, or pharmacokinetics, thereby affecting the efficacy of evasins in disease models.

Tick evasins from diverse genera
Previously, the only tick evasins shown to bind and inhibit chemokines were the three R. sanguineus evasins. Our results indicate that evasins are likely to be broadly expressed by many hard tick species and therefore that chemokine inhibition has been an advantageous evolutionary strategy for ticks. As all hard ticks are obligatory blood feeders, this may reflect positive selection pressure to maintain immunomodulatory evasin proteins and to evade immune recognition and remain attached to hosts for extended periods to enable continued blood feeding.
Among the 257 putative evasins identified in our bioinformatics searches, there is substantial sequence variation within each tick species and between different species, both common characteristics of tick protein families (11). Within a single species, we identified numerous different evasin-like sequences. For example, we found 13 R. pulchellus sequences with pairwise sequence identities ranging from 14 to 50%. Considering the current and previous observations that chemokine selectivity can vary substantially among evasins (15,16,18), it seems likely that the expression of several evasins by the same species may be a strategy to more effectively suppress the activities of multiple host chemokines and thereby inhibit the recruitment of their target leukocytes.
Comparison of sequences from different species also indicated a high level of sequence divergence. The average Ϯ S.D. of cross-species pairwise identities was 20.1 Ϯ 6.7%, and pairwise identities were Ͻ50% for 99% of cross-species comparisons and Ͻ30% for 95% of cross-species comparisons. The maximum identity found to R. sanguineus evasin-1 or -4 was 46% (to sequence JAA60818.1 from R. pulchellus). These sequence variations suggest that there has been substantial evolutionary pressure for evasin sequences to diverge after tick speciation events. Because speciation of ticks would typically be associated with a new mammalian host, the evolutionary pressure may arise from the different arrays of chemokines expressed by the hosts for the different tick species.

Tick evasins from diverse genera
Many tick species are of medical or agricultural importance because they function as vectors for microbial pathogens. Within the genus Ixodes, I. ricinus and the related I. scapularis are of particular interest as vectors of the Lyme disease spirochete Borrelia burgdorferi (38 -40). Within the genus Rhipicephalus, R. microplus (previously known as Boophilus microplus) is considered the most economically important tick for the United States and Australian cattle industries as it can cause weight loss in host animals and transmit agents of several cattle diseases (41,42). Based on the prevalence of evasin sequences in the currently available databases, we anticipate that ongoing genome projects for I. scapularis and R. microplus are likely to reveal additional evasins as a mechanism of host immune evasion by these important tick species.
The numerous tick evasins identified in the current study represent a rich "library" of chemokine-binding proteins from which to identify proteins that selectively inhibit chemokines of interest in specific human or animal diseases. Notably, several of the newly identified evasins have high affinity for the chemokine eotaxin-1, a major chemoattractant for eosinophils and basophils, which play a key role in allergic responses and reactions to parasitic infestations. This is consistent with the role of evasins being to suppress host responses to ticks. However, it also suggests that evasins may be useful therapeutics in human diseases in which these types of leukocytes are involved as observed in a previous study of experimental colitis (29).

Conclusion
Our sequence similarity searches have identified more than 250 predicted tick protein sequences with features similar to R. sanguineus evasin-1 and -4 but substantial sequence variation in chemokine recognition residues. Among these sequences, we have identified two distinct subfamilies of evasins (C 8 and C 6 ). Among nine candidates screened for chemokine binding, eight exhibited nanomolar affinities for one or more chemokine tested. Our results suggest that most hard ticks express chemokine-binding evasins and that these evasins play important roles in evasion of host immune responses. The evasins identified here provide a valuable pool of potential inhibitors for development of chemokine-targeted anti-inflammatory therapies.

Bioinformatics and evasin identification
Sequences of the canonical R. sanguineus evasin-1 and evasion-4 were retrieved from UniProt (accession numbers P0C8E7 and P0C8E9). These were used as queries against Uni-ProtKB (restricted to Arthropoda; July 20, 2015 release) using the Basic Local Alignment Search Tool (BLAST) (43) with the BLASTp algorithm and against VectorBase (I. scapularis and I. ricinus transcriptome and genome sequences) (44) using tBLASTn with default parameters. Hits were ranked based on highest pairwise identity to the known evasin sequences using a pairwise identity matrix generated by Clustal Omega (45). The presence of the cysteines conserved in the canonical evasins was assessed. Putative evasin sequences were then analyzed using SignalP (46) to predict signal peptide cleavage sites. The putative evasin sequences with highest pairwise identity to the known evasins for each tick species were selected as candidates for cloning and expression. These candidate sequences were analyzed for the presence of putative sulfation sites using Sulfinator (47) and N-linked glycosylation sites by the motif N*(S/T).
Multiple sequence alignments of initial hits were generated using MUSCLE v3.8.31 (48). HMMs were constructed from these MUSCLE alignments using HMMER v3.1b1 (49). The HMMs were used with hmmsearch (default parameters) to identify more distantly related evasin sequences in UniProtKB (version January 11, 2016), NCBI Transcriptome Shotgun Assembly databases (downloaded February 10, 2016), Vector-Base (I. scapularis and I. ricinus transcriptome sets translated with TransDecoder (50) and six-frame translated genome sequences), and I. holocyclus and R. microplus transcriptome data sets provided by Manuel Rodriguez Valle. Sequences were retained if they contained conserved cysteine residues equivalent to the canonical evasins and a predicted signal sequence and were less than 200 residues long. Hits were compiled into a final non-redundant set of putative evasins.

Cloning, expression, and purification of putative evasins for screening
Plasmids containing the coding sequences of evasin candidates plus a C-terminal His 6 tag in the background expression vector pET28a were ordered from GenScript; codon usage was optimized for expression in E. coli. BL21(DE3) E. coli cells transformed with these plasmids were grown in 1 liter of LB medium supplemented with kanamycin sulfate (30 g/ml) until the optical density at 600 nm reached 0.6. Protein expression was initiated by addition of isopropyl ␤-D-1-thiogalactopyranoside (1 mM), and the culture was grown for either 4 h or overnight. Cells were harvested by centrifugation; resuspended in 15 ml of 20 mM Tris⅐HCl, 500 mM NaCl, 5 mM imidazole, 0.02% (w/v) NaN 3 , pH 8.0; sonicated; and centrifuged to separate the soluble from insoluble (inclusion body) fraction. Inclusion bodies were washed in 20 mM Tris⅐HCl, 500 mM NaCl, 5 mM imidazole, 0.02% (w/v) NaN 3 , 2 mM DTT, 0.5% (v/v) Triton X-100, pH 8.0, and denatured in 20 mM Tris⅐HCl, 6 M guanidine⅐HCl, 20 mM imidazole, 20 mM ␤-mercaptoethanol, pH 8.0, before being added to 15 ml of loose nickel-nitrilotriacetic acid resin (Qiagen) to purify recombinant protein by batch immobilized metal affinity chromatography. Denatured protein was then refolded via rapid dilution (flow rate, 0.1 ml/min) in 2 liters of 20 mM Tris⅐HCl, 400 mM NaCl, 2 mM reduced glutathione, 0.5 mM oxidized glutathione, 0.02% (w/v) NaN 3 , pH 8.0. This was followed by further purification by immobilized metal affinity chromatography (HisTrap HP 5-ml column, GE Healthcare).

Tick evasins from diverse genera
After initial screening, expression of selected evasins was repeated at 4-liter scale, including a final purification step of size exclusion chromatography (HiLoad 16/60 Superdex 75 prep grade column, GE Healthcare). The molecular weights of pure proteins were confirmed by mass spectrometry using an Orbitrap Fusion Tribrid mass spectrometer (Thermo Scientific) in intact protein mode.

Chemokine-binding assay
Competition-binding assays, measuring the displacement of a fluorescent receptor peptide from each of six CC chemokines (CCL2/MCP-1, CCL7/MCP-3, CCL8/MCP-2, CCL11/eotaxin-1, CCL24/eotaxin-2, and CCL26/eotaxin-3), were conducted for each candidate evasin to screen for chemokine-binding ability as described (33). Briefly, a solution of the candidate evasin was prepared in MOPS buffer (50 mM MOPS, pH 7.4), and a serial 2-fold dilution was conducted on a 96-well plate (final concentration range from 1 M to 15.6 nM). A mixed solution of the fluorescent peptide and chemokine (final concentrations of 10 nM and 100 nM, respectively) was added such that the final volume in each well was 200 l. Competition-binding assays were also performed using the purified evasins to measure the displacement of a fluorescent receptor peptide from each of five CC chemokines (MCP-1, MCP-2, eotaxin-1, eotaxin-2, and eotaxin-3) under similar conditions. Briefly, a solution of the purified evasin was prepared in MOPS buffer, and a serial 2-fold dilution (1.33-fold dilution for evasin-4) was conducted on a 384-well plate coated with a 0.001% (w/v) solution of poly-Llysine (final concentration range from 1 M to 15.6 nM or 300 nM to 9.5 nM for evasin-4). A mixed solution of the fluorescent peptide and chemokine (final concentrations of 10 and 100 nM, respectively) was added such that the final volume in each well was 20 l. In both cases, fluorescence anisotropy was measured 5 min after plating with excitation and emission wavelengths of 485 and 520 nm, respectively, using a BMG Labtech PHERAstar FS plate reader. Experiments were conducted in duplicate three times independently, and the mean anisotropy was fitted by non-linear regression analysis using GraphPad Prism v.6.0 software to the equation for a 1:1 competitive displacement curve, described previously (51), transformed to enable Ϫlog K d (pK d ) values ϮS.E. to be determined. Multiple t tests were conducted to assess significance of the differences in pK d .

Inhibition of chemokine signaling
Chemokine signaling via the receptor CCR2 or CCR3 was measured using a cAMP biosensor assay developed previously (52). The assay measures the ability of chemokines to inhibit forskolin-induced cAMP production in c-Myc-FLAG-CCR2 or c-Myc-FLAG-CCR3 FlpIn TREx HEK293 cells transiently transfected with the CAMYEL cAMP bioluminescence resonance energy transfer (BRET) biosensor (53). Cells were grown overnight in 10-cm dishes using DMEM ϩ 5% FBS before transfection using 6:1 (w/w) PEI:DNA. Tetracycline (10 g/ml) was added 24 h after transfection to induce expression of c-Myc-FLAG-CCR2 or c-Myc-FLAG-CCR3, and cells were seeded (25,000 cells/well) in a white, poly-D-lysine-coated 96-well plate (CulturPlates, PerkinElmer Life Sciences) and incubated overnight at 37°C in 5% CO 2 . The following day, cells were washed and equilibrated in Hanks' balanced salt solution for 30 min at 37°C. Cells were then incubated with the Rluc substrate coelenterazine h (final concentration, 5 M) for 5 min followed by a further 5-min incubation with a solution containing a fixed concentration of chemokine (10 nM MCP-1, 100 nM MCP-2, 100 nM eotaxin-1, or 100 nM eotaxin-2, corresponding to approximately the EC 80 , as shown in supplemental Fig. S4) with various concentrations of an evasin candidate that had been preincubated for ϳ15 min to allow binding of the evasin to the chemokine. A final 5-min incubation step was conducted after addition of forskolin (final concentration,10 M), which directly stimulates the production of cAMP via adenylyl cyclase. The Rluc and yellow fluorescent protein (YFP) emissions were then measured at 475 and 525 nm, respectively, using a BMG Labtech PHERAstar FS plate reader. Data are presented as a BRET ratio, calculated as the ratio of YFP to Rluc signals and expressed as the percentage of the forskolin-induced signal.