Abstract

In a typical G protein coupled receptor drug discovery campaign, an in vitro primary functional screening assay is often established in a recombinant system overexpressing the target of interest, which offers advantages with respect to overall throughput and robustness of compound testing. Subsequently, compounds are then progressed into more physiologically relevant but lower throughput ex vivo primary cell assays and finally in vivo studies. Here we describe a dynamic mass redistribution (DMR) assay that has been developed in a format suitable to support medium throughput drug screening in primary human neutrophils. Neutrophils are known to express both CXC chemokine receptor (CXCR) 1 and CXCR2 that are thought to play significant roles in various inflammatory disorders and cancer. Using multiple relevant chemokine ligands and a range of selective and nonselective small and large molecule antagonists that block CXCR1 and CXCR2 responses, we demonstrate distinct pharmacological profiles in neutrophil DMR from those observed in recombinant assays but predictive of activity in neutrophil chemotaxis and CD11b upregulation, a validated target engagement marker previously used in clinical studies of CXCR2 antagonists. The primary human neutrophil DMR cell system is highly reproducible, robust, and less prone to donor variability observed in CD11b and chemotaxis assays and thus provides a unique, more physiologically relevant, and higher throughput assay to support drug discovery and translation to early clinical trials.

SIGNIFICANCE STATEMENT Neutrophil dynamic mass redistribution assays provide a higher throughput screening assay to profile compounds in primary cells earlier in the screening cascade enabling a higher level of confidence in progressing the development of compounds toward the clinic. This is particularly important for chemokine receptors where redundancy contributes to a lack of correlation between recombinant screening assays and primary cells, with the coexpression of related receptors confounding results.

Introduction

The CXC chemokine receptor (CXCR) 2 is a G protein-coupled receptor (GPCR) that is widely expressed across immune cells (Cheng et al., 2019). Both CXCR2 and the closely related CXCR1 are Gαi-coupled GPCRs that predominantly signal through inhibitory G proteins to reduce intracellular cAMP. Like many chemokine receptors, CXCR1 and CXCR2 are activated by multiple circulating chemokines, with CXCR1 activated by CXCL6 and CXCL8, whereas CXCR2 is activated by CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, CXCL7, and CXCL8 (Ahuja and Murphy, 1996; Liu et al., 2016). This makes a CXCR2 screening cascade more complex than a single ligand-receptor system, with the ligand promiscuity of chemokine receptors often cited as a reason for the clinical failure of many chemokine-targeting therapies (Dyer, 2020).

As the most abundant leukocyte in circulation and a key regulator in disease processes, neutrophils are becoming an important pharmacological target (Németh et al., 2020). As such, detection of neutrophil activity in clinical studies has become increasingly significant, with CD11b upregulation now established as an important translational marker (Liston et al., 1998; Lazaar et al., 2011; Miller et al., 2015; Namour et al., 2016). CD11b is an integrin, which is upregulated in response to circulating chemoattractants, such as chemokines and complement proteins (Maas et al., 2018). CD11b binds to intracellular adhesion molecules 1 and 2 to mediate neutrophil adhesion and migration through the endothelium to sites of infection (Maas et al., 2018). CD11b upregulation has been used as a target engagement marker in many clinical trials, including for the CXCR2 small molecule antagonists danirixin and SB-656933, where the latter used CD11b data to inform clinical doses (Liston et al., 1998; Lazaar et al., 2011; Miller et al., 2015; Namour et al., 2016). Other common neutrophil assays include superoxide production and chemotaxis. While technological advancements have improved the quality of these assays over the years, they are often low throughput and labor intensive, prone to assay variability, and highly susceptible to interdonor variability, making them unsuitable for more intensive screening (Gomez-Lopez et al., 2011; Silvestre-Roig et al., 2019).

Dynamic mass redistribution (DMR) assays are a label-free method to measure ligand responses. They are based on two alternative detection methods, namely light refraction and electrical impedance. Light-based methods involve optical biosensors that measure the change in the wavelength of refracted light upon cellular activation, whereas impedance methods measure the change in the impedance of current flow between two electrodes (Scott and Peters, 2010; Fang, 2011). While it is still unknown precisely what these readouts measure, they are thought to provide a holistic view of cellular signaling, detecting more integrated, phenotypic downstream responses, such as changes in cell adhesion or shape (Scott and Peters, 2010; Grundmann and Kostenis, 2015). Although most studies have examined recombinant cells in these systems, the high sensitivity and ability to capture any form of signaling in a label-free set-up affords an ideal opportunity for primary cell studies (Grundmann and Kostenis, 2015; Hillger et al., 2017). Moreover, the ability to pick up multiple signaling pathways and modalities makes them potentially superior to standard functional assays, which measure a single readout of response.

The unique expression of receptor and effector proteins in different cell backgrounds can dramatically change compound activity when transferring from recombinant systems, where primary pharmacology screening usually occurs, to endogenously expressing systems (Eglen et al., 2008). This is particularly true for chemokine receptors, where redundancy is thought be a major contributor to drug failures (Dyer, 2020), especially for neutrophils, which coexpress CXCR1 and CXCR2 (Futosi et al., 2013; Lämmermann and Kastenmüller, 2019). Thus, our objective was to establish a DMR assay to measure chemokine activation in primary human neutrophils and to determine how these results translate from data generated using recombinant cell systems. We also aimed to examine more conventional neutrophil assays (namely, CD11b upregulation and chemotaxis) to validate the DMR assay for higher throughput compound screening in primary cells. There is precedent for measuring GPCR activation in neutrophil DMR assays (Schröder et al., 2011; Locker et al., 2015; Christensen et al., 2017; Frei et al., 2021); however, to our knowledge, this is the first time a more thorough examination and comparison of multiple functional readouts for neutrophil chemokine receptors has been undertaken. As CD11b upregulation is a common target engagement marker for neutrophil GPCR drugs, we were keen to establish how results translate from the DMR platform to CD11b and whether this assay potentially could be used as a screen to help predict activity in a clinical setting. Here we demonstrate that neutrophil DMR assays are robust and reproducible, generating both chemokine responses and antagonist profiles that are consistent across different neutrophil assay formats, supporting DMR as a high-throughput alternative to CD11b and chemotaxis assays in the screening cascade.

Materials and MethodsMaterials

All chemokines were from Genscript (Piscataway, NJ, USA). All other ligands were from Tocris (Bristol, UK) except for AZD-5069 (MedChemExpress, Monmouth Junction, NJ, USA), Navarixin (Toronto Research Chemicals, Ontario, Canada), Danirixin (Cayman Chemical, Ann Arbor, MI, USA), N-formylmethionyl-leucyl-phenylalanine (fMLF) (Sigma Aldrich, Poole, UK), and human C5a (PeproTech, London, UK). Compound 19 (Dwyer et al., 2006) was synthesized internally. All assay reagents were from Sigma-Aldrich unless otherwise stated. FITC-conjugated mouse IgG1 kappa anti-CD11b monoclonal antibody (clone ICRF44, catalog no. 301330), APC-conjugated mouse IgGM kappa anti-CD15 monoclonal antibody (clone HI98, catalog no. 301908), and PE-Cy7-conjugated mouse IgG1 kappa anti-CD16 monoclonal antibody (clone 3G8, catalog no. 302016) were from Biolegend (San Diego, CA, USA) while CXCR2 nanobodies 127D1 and 163E3 (Bradley et al., 2015) were a kind gift from Kymab (Cambridge, UK).

Cell Line Generation

CHO-K1 cells were transfected with a pSNAP vector (NEB, Hitchin, UK) containing the human CXCR1 or CXCR2 receptor using GeneJuice transfection reagent (Merck Millipore, Watford, UK) according to the manufacturer’s instructions. Cells were cultured in Dulbecco’s modified Eagle’s medium/Ham’s F-12 medium supplemented with 10% fetal bovine serum and 1 mg ml−1 geneticin for selection. Following 2 weeks of selection, cells were dilution cloned. Clonal CHO CXCR1 and CHO CXCR2 cells were maintained in Dulbecco’s modified Eagle’s medium/Ham’s F-12 medium supplemented with 10% FBS and 0.2 mg ml−1 geneticin to maintain selection pressure.

cAMP Assays

CHO CXCR1 and CHO CXCR2 cells were seeded overnight at 2000 and 1200 cells per well, respectively, in white 384-well plates. On experiment day, media was replaced with assay buffer (Hank’s balanced salt solution; Lonza, Basel, Switzerland) supplemented with 20mM HEPES and 0.1% bovine serum albumin, pH 7.4) in the absence or presence of antagonists, for 1 hour at 37°C. Chemokines, in the presence of 1 µM forskolin, were added and plates further incubated for 5 minutes. cAMP was detected using cAMP Gi kit (Cisbio, Codolet, France) according to the kit instructions. Plates were read on a PHERAstar FS microplate reader (BMG LabTech, Offenburg, Germany) using standard homogeneous time resolved fluorescence settings. Homogeneous time resolved fluorescence ratios were determined by dividing emissions at 665 nm by emissions at 620 nm and multiplying by 10,000.

Neutrophil Isolation–Ficoll-Dextran

Human whole blood from healthy volunteers (Cambridge Bioscience, Cambridge, UK) was mixed 1:1 with 2% dextran in PBS to sediment the red blood cells. After 30 minutes, the supernatant was removed and carefully layered onto 15 ml Ficoll-Paque (GE Healthcare, Pittsburgh, PA, USA) and centrifuged for 25 minutes [300g, room temperature (RT), slow acceleration and brake]. The supernatant was discarded and the pellet resuspended in PBS. Contaminating red blood cells were removed by osmotic shock; 20 ml ice-cold distilled water was added for 30 seconds before neutralization with 20 ml 2x PBS. Cells were pelleted by centrifugation (5 minutes, 300g) before being resuspended in PBS. Cells were counted with a Haematology Analyzer (Sysmex, Milton Keynes, UK).

Dynamic Mass Redistribution Assays

Ficoll-Dextran isolated cells were resuspended in assay buffer containing 0.4% DMSO and seeded 50,000 neutrophils per well onto uncoated 384-well Epic plates (Corning, NY, USA). Cells were left to settle to the bottom of the wells at RT for 2 hours. Plates were read using the Corning Epic BT System, which measures cellular responses as a change in the wavelength of refracted light. First, plates were read for 1 minute to establish baseline before pausing the read. Ligands were added slowly and to the top of the volume to avoid disturbing the cells, using a Bravo Automated Liquid Handling Platform (Agilent, Santa Clara, CA, USA). The plate was returned to the reader and the read resumed. For antagonist treatments, antagonists were added in the same manner with responses measured to determine the effect of compound alone. They were incubated for 1 hour at RT before the addition of chemokine as described earlier. For pertussis toxin (PTx) treatments, neutrophils were incubated with 3 μg ml−1 PTx for 1 hour at 37°C before being plated onto Epic plates as stated previously. For CHO CXCR1 or CXCR2 assays, cells were seeded 12,500 cells per well overnight onto uncoated 384-well Epic plates. Media was exchanged for assay buffer for 2 hours prior to conducting assay as described earlier.

Neutrophil Isolation–Magnetic Isolation

Neutrophils were isolated from human whole blood from healthy volunteers by immunomagnetic negative selection using the EasySep Direct Human Neutrophil Isolation kit (STEMCELL Technologies, Cambridge, UK) following manufacturer’s instructions. Briefly, EDTA was added to the blood to a final concentration of 1 mM before adding the isolation cocktail and RapidSpheres. After 5 minutes incubation at RT, the cells were diluted with 1 equal volume of PBS containing 1 mM EDTA and placed into the magnet for 10 minutes at RT to allow magnetic separation of labeled cells. Further enrichment was undertaken by performing two additional cycles of RapidSpheres addition and separation with the magnet. The final cell suspension was counted with a Haematology Analyzer with expected neutrophil purity > 90%. Cells were centrifuged (5 minutes, 300g) then resuspended in assay buffer.

Chemotaxis Assays

Magnetically isolated neutrophils were loaded with 5 µM Calcein AM (Invitrogen) for 30 minutes at 37°C before resuspension in assay buffer. Two hundred thousand cells were added to the upper chamber of a HTS 96-transwell chemotaxis plate with a 5 µm filter (Corning) and incubated for 30 minutes at 37°C in the presence or absence of antagonist. Following this incubation, chemokine was added to the lower well compartment and the filter plate assembly incubated at 37°C/5% CO2 for 45 minutes. The filter was removed and 100 µl of the lower compartment was transferred to a 96-well black bottom plate (Corning) and the fluorescence determined (excitation: 485 nm emission: 520 nm) using a PHERAstar microplate reader.

CD11b Assays

Fifty thousand magnetically isolated human neutrophils were preincubated with antagonist at 37°C for 30 minutes before the addition of chemokine for 30 minutes at 37°C. The assay was stopped by placing cells on ice and addition of an equal volume of ice-cold staining buffer (1.5 µg ml−1 anti-CD11b antibody + 0.4 µg ml−1 anti-CD15 antibody + 0.1 µg ml−1 anti-CD16 antibody in assay buffer). After staining for 30 minutes at 4°C, cells were washed twice with ice-cold assay buffer and fixed with 1% paraformaldehyde (20 minutes, 4°C). After fixation, cells were washed twice with ice-cold PBS and kept in 200 µl PBS at 4°C in the dark until ready to analyze. CD11b expression was determined using flow cytometry (FACS-Canto II; Becton, Dickinson and Company, Franklin Lakes, NJ, USA). Neutrophil CD11b expression was measured using the geometric mean of FITC fluorescence intensity on CD15-positive and CD16-positive granular cells with high forward and side-scatter properties; 2,000 positive events were counted per sample.

Data Analysis

DMR data were analyzed using Corning EpicAnalyzer software. Raw DMR traces were corrected by removal of background responses and the peak response between 3 and 7 minutes (neutrophil) or 5 and 12 minutes (CHO CXCR1 and CXCR2) extracted. All data were plotted in GraphPad Prism (San Diego, CA, USA) using four parameter logistic equation. Statistical analyses were also conducted with GraphPad Prism.

ResultsProfiling of Chemokines and Antagonists in CHO CXCR1 and CHO CXCR2 cAMP and DMR Assays

First, we profiled a range of chemokines and antagonists reported to be selective for CXCR1 and CXCR2 in CHO CXCR1 and CXCR2 cAMP and DMR assays to understand their potencies and selectivity profiles in a recombinant expression system. As a Gαi-coupled receptor, inhibition of cAMP is the primary signaling endpoint for CXCR1 and CXCR2, so we wanted to investigate how this compared with a DMR readout.

In the cAMP assay, all the chemokines examined were full agonists for the inhibition of forskolin-stimulated cAMP at the CXCR2 receptor, except for CXCL7 (Emax 88 ± 3%; Table 1; Fig. 1A). At CXCR1, all ligands acted as full agonists to inhibit forskolin-stimulated cAMP, except for CXCL1, which was a partial agonist (Emax 81 ± 6%) and CXCL2, which was inactive (Table 1; Fig. 1B). Between the two receptors, there were large potency differences for many of the chemokines tested, which is reflective of their reported variations in selectivity for CXCR2.

TABLE 1

Summary of chemokine responses in CHO CXCR2 and CXCR1 cAMP inhibition and DMR assays

Fig. 1.

Activity of chemokine ligands in CHO CXCR1 and CHO CXCR2 cAMP and DMR assays. Example chemokine concentration response curves for cAMP assays in CHO CXCR2 (A) and CXCR1 (B) assays. Representative kinetic DMR traces for CXCL8 at CXCR2 (C) and CXCR1 (D) with compound addition indicated by the arrow. Example chemokine concentration response curves for DMR assays for CXCR2 (E) and CXCR1 (F). There was a poor correlation in potencies between the two readouts for CXCR2 (G) but an excellent correlation for CXCR1 (H). Data are pooled (mean ± S.D.) from two to five independent experiments with data in correlation plots summarized from Table 1. For the kinetic traces, data are mean ± S.D. from duplicate wells.

In the DMR assay, chemokines induced concentration-dependent positive deflections in both CHO CXCR2 (Fig. 1C) and CHO CXCR1 (Fig. 1D) cells, which peaked around 5 minutes after compound addition. Peak responses were extracted for both receptors (Fig. 1, E and F), with CXCR2 results appearing broadly consistent with the results of the cAMP assay (Table 1). However, there was a trend for reduced agonist potencies in this assay relative to the inhibition of cAMP, and CXCL2 was also identified as a partial agonist. This is likely due to the increased signal amplification of the cAMP signal, leading to increased agonist potencies (Stott et al., 2016). However, the correlation in potencies between both assay formats for CXCR2 was poor (R2 = 0.27; Fig. 1G), though this is potentially due to the low spread in potency values across ligands at this target and the higher variability in the DMR responses. For CXCR1, the DMR readout revealed the partiality of most of the chemokines at this target, with only the two CXCL8 proteoforms showing full agonism (Fig. 1F; Table 1). Unlike CXCR2, there was an excellent correlation in potencies for CXCR1 between the two assay formats (R2 = 0.92; Fig. 1H) as well as a trend for increased potencies in DMR relative to cAMP, opposite to what was observed for CXCR2.

We then tested a range of small molecule antagonists, most of which have been previously described as CXCR2-selective, in the CHO CXCR1 and CXCR2 cAMP and DMR assays using both CXCL8 and CXCL1 as nonselective and selective chemokines, respectively (Fig. 2). Due to the weak activity of CXCL1 at CXCR1, we could not determine the activity of the antagonists for this receptor-ligand pair. All the small molecules tested were able to fully inhibit CXCR2 responses elicited by both CXCL8 and CXCL1, except for reparixin, which was inactive, and NVP CXCR2 20, which partially inhibited the CXCR2 responses (Fig. 2, A and B; Table 2). Reparixin is described as a high-affinity antagonist of both CXCR1 and CXCR2 (Bertini et al., 2004), so this inactivity was surprising. We cannot yet explain this contradictory observation; we tested multiple sources of the compound and quality control analysis by UV, and mass spectrometry (data not shown) did not highlight any issues with the compound stock. There were no significant differences in the potencies of the small molecules to inhibit CXCR2 signaling induced by CXCL1 or CXCL8 (P > 0.05; unpaired t test), but all ligands were significantly less potent at inhibiting CXCL8 signaling at CXCR1 when compared with CXCR2 (P < 0.05; unpaired t test; Fig. 2C; Table 2). We also examined two previously described CXCR2 specific nanobodies (Bradley et al., 2015). These nanobodies have been shown to either bind the CXCR2 N-terminus (clone 127D1), generating a potent, partial antagonist, or the extracellular loop (clone 163E3), generating a less potent but full antagonist (Bradley et al., 2015). In our hands, both nanobodies behaved as full antagonists, and, unlike the original report, 127D1 and 163E3 were equipotent at inhibiting both CXCL1 and CXCL8 responses (Table 2). These nanobodies have been shown not to bind CXCR1 (Bradley et al., 2015). When these antagonists were tested in the DMR assay, all ligands except reparixin and AZ 10397767 fully inhibited chemokine responses and the same rank order of pIC50 values was obtained (Fig. 2, D–F; Table 3), and there was a good correlation with the cAMP results (R2 ≥ 0.67; Fig. 2, G–I). However, this assay had a much smaller assay window and much greater variability than the cAMP format, which limited compound testing.

Fig. 2.

Activity of antagonists in CHO CXCR1 and CHO CXCR2 cAMP and DMR assays. Antagonists were preincubated with for 1 hour at RT prior to addition of EC80 of chemokine. Example antagonist concentration response curves at CXCR2 (A, B) and CXCR1 (C) in cAMP assays and CXCR2 (D, E) and CXCR1 (F) in DMR assays. pIC50s between assay formats correlated well (G–I). All data are pooled (mean ± S.D.) of two to four experiments with data in correlation plots summarized from Table 2.

TABLE 2

Summary of antagonist responses at CHO CXCR2 and CXCR1 receptors in cAMP inhibition assays against an EC80 challenge of CXCL1 and CXCL8

TABLE 3

Summary of antagonist responses at recombinant CHO CXCR2 and CXCR1 receptors in DMR assays against an EC80 challenge of CXCL1 and CXCL8

Chemokines Generate Robust and Reproducible DMR Responses That Are Specific to CXCR1 and CXCR2 in Isolated Neutrophils

We then looked to characterize the chemokine and antagonist responses in the neutrophil DMR assay. While there is some evidence for CXCL8-induced neutrophil DMR responses (Locker et al., 2015), this is the first time a more detailed assessment of chemokine responses in this assay, and how they compare with other neutrophil assays, has been undertaken. Upon addition of chemokine, neutrophils demonstrated a large and rapid concentration-dependent positive deflection which peaked within 4 minutes of treatment (Fig. 3, A–C), which was much larger in size than the CHO CXCR1 or CHO CXCR2 DMR assays (Fig. 1, C and D). All chemokines displayed very similar kinetic traces, despite the differing levels of selectivity for CXCR2 over CXCR1 (Table 1). Peak responses were extracted and data pooled (Fig. 3, D and E) with chemokine potencies and maximal responses described in Table 4. Despite the known issues of reproducibility in primary cell assays, ligand potencies were very consistent over multiple donors, although absolute response magnitude varied; for example, 30 nM CXCL8 responses ranged from ∼700 to 1200 Δpm between donors, but intra-assay variability was generally less than 10% (Supplemental Fig. 1) and much lower than that previously reported for formyl peptide receptor (FPR) 1 responses in the same assay (Christensen et al., 2017).

Fig. 3.

CXCR1/2 chemokines demonstrate large positive deflections in neutrophil DMR assays. Representative kinetic DMR traces of concentration response curves of CXCL1 (A) and CXCL8 (B) and representative 100 nM chemokine (C) with arrow indicating compound addition. Data are mean ± S.D. of triplicate wells. Data were baseline corrected by subtracting buffer wells and peak response between 3 and 7 minutes extracted and plotted (D, E). Data represent pooled (mean ± S.D.) of at least four independent experiments each from an individual donor, normalized to 30 nM CXCL8 response. Neutrophil DMR potencies were correlated with potencies from recombinant CHO-CXCR2 cAMP (F) and DMR (G) with CXCR1/2 dual chemokines (CXCL6 and CXCL8) labeled in red.

TABLE 4

Summary of chemokine responses in neutrophil DMR assays

We then compared the chemokine potencies in neutrophil DMR with those from CHO CXCR2 cAMP and observed a good correlation between the two assays (R2 = 0.63, P < 0.01; Fig. 3F). However, the correlation between CHO CXCR2 DMR and neutrophil DMR was much lower (R2 = 0.32, P = 0.09; Fig. 3G). Compared with the neutrophil DMR, the CHO CXCR2 DMR had a much lower signal (∼100 pM vs ∼1000 pM) and much greater variability. This could be reflective of different cellular processes within the different cell backgrounds upon CXCR2 activation, which could explain these results. When the neutrophil DMR potencies were compared with the CHO CXCR1 cAMP results, there was no correlation (R2 = 0.00; data not shown), likely due to the selectivity of most chemokines for CXCR2.

As DMR likely represents a holistic cellular response, we wanted to investigate whether the chemokine responses measured in neutrophils were dependent on Gαi activation using PTx to inactivate the Gαi protein. However, neutrophils are short-lived cells that usually only survive for less than 24 hours in the bloodstream (McCracken and Allen, 2014); this prohibited the usual overnight PTx treatment regimen. Instead, we used a modified protocol from Christensen et al. (Christensen et al., 2017) whereby we treated neutrophils for 1 hour with a high PTx concentration (3 μg ml−1) and included fMLF as a positive control to replicate that study. However, this treatment only inhibited responses by around 20% (Supplemental Fig. 2) unlike the original study, which could inhibit responses by around 70%. In both cases, this suggests that there is some involvement of Gαi in the neutrophil DMR responses, but it likely needs a longer treatment time to fully inactivate the Gαi proteins, which is challenging to do with such short-lived cells.

As neutrophils express several GPCRs, we wanted to confirm that the responses detected were specific to CXCR1 and CXCR2. The chemokines show differing levels of selectivity for CXCR1 and CXCR2; therefore, we examined antagonism of both CXCL1 and CXCL8 as CXCR2-selective and -nonselective chemokines, respectively. It is noteworthy that most of the small molecules studied, except for the inactive reparixin, induced a small, negative deflection following compound treatment alone (Fig. 4, A–C). This could be indicative of inverse agonism, which is generally characterized by an opposite response to the agonist. It has been demonstrated previously in DMR assays (Scott and Peters, 2010), and inverse agonism has been reported for both small molecule and nanobody CXCR2 antagonists (Bradley et al., 2009, 2015) but only in recombinant systems thus far. However, the scale of this response (∼40 Δpm) is dwarfed by the response elicited by the chemokines themselves. These negative deflections were not observed in the CHO CXCR1 or CXCR2 DMR assays, suggesting the presence of constitutively active receptors in neutrophils that were not detected recombinantly. However, this could be due to the relative size responses between cell types; the assay signal in the recombinant system may not have been large enough to detect the constitutive activity. Notably, the nanobody 127D1 displayed a significant response when tested alone at concentrations of 3 nM and greater, suggesting either agonism or nonspecific responses at these higher concentrations. This effect was not observed for 163E3, suggesting it was not a general nonspecific effect to nanobody treatment (Fig. 4, D and E). These concentrations were excluded from further analysis. These agonist-like responses were also not observed in the CHO CXCR2 cells and could be due to off-target effects specific to neutrophils. This further emphasizes the need for primary cell screening in the target cell type.

Fig. 4.

Kinetic traces of antagonist treatments. Representative kinetic DMR traces of small molecule antagonist (A–C) and nanobody treatments (D, E). Data represents mean ± S.D. of duplicate wells with arrow indicating compound addition.

All the small molecules examined inhibited both CXCL1- and CXCL8-induced DMR responses in a concentration-dependent manner, except for reparixin, which was inactive, in line with the recombinant data. When the CXCR2 nanobodies were examined, they showed significant inhibition of the CXCL1 response, but, unlike the recombinant assay data, 127D1 was 10-fold more potent than 163E3 and demonstrated partial antagonism. This is consistent with a previous report of the nanobody pharmacology (Bradley et al., 2015). However, when they were tested against CXCL8, they failed to significantly inhibit responses and potencies could not be derived due to incomplete curves (Fig. 5, A–F; Table 5). When the pIC50 values for CXCR1 inhibition were compared between the neutrophil DMR assay and the CHO CXCR2 cAMP or DMR assays, there was an excellent correlation (R2 = 0.73–0.74; Fig. 5, G and H). In contrast, the correlation for CXCL8 inhibition between neutrophil DMR and recombinant assays was lower for the DMR assay (R2 = 0.65) and failed to reach significance for the cAMP assay (R2 = 0.46; P = 0.06) (Fig. 5, I and J). CXCL1 displays one of the highest selectivities for CXCR2 (Table 1), so it is highly probable that at EC80 concentration, only CXCR2 receptors are being activated, whereas CXCL8 will activate both CXCR1 and CXCR2. The reduced correlation for CXCL8 is likely indicative of two receptors signaling in neutrophils, compared with the single receptor system measured in the CHO CXCR1 and CXCR2 cell lines.

Fig. 5.

Inhibition of CXCL1 and CXCL8 responses by CXCR1/2 antagonists. Antagonists were preincubated with neutrophils for 1 hour at RT prior to addition of EC80 of chemokine. Inhibition of CXCL1 (A) and CXCL8 (D) induced DMR responses by selected antagonists. Data are pooled (mean ± S.D.) of 4 to 15 independent experiments, each from an individual donor, performed in duplicate, normalized to EC80 concentration of chemokine. Representative DMR kinetic traces of EC80 chemokine addition following preincubation with increasing concentration of AZD-5069 (B, E) or 163E3 (C, F). Data points represent mean ± S.D. of duplicate wells. Correlation between pIC50 for neutrophil DMR and CHO CXCR2 CXCL1 cAMP (G), CXCL1 DMR (H), CXCL8 cAMP (I), or CXCL8 DMR (J).

TABLE 5

Summary of CXCL1 and CXCL8 inhibition by antagonists in neutrophil DMR assays

Neutrophil DMR Assays Are Sensitive to Different Antagonist Modalities

Next, we wanted to explore whether this assay was sensitive enough to pick up different antagonist modalities by performing Schild analysis using the CXCR2 nanobodies that have been described as noncompetitive (127D1) or competitive (163E3) (Bradley et al., 2015). We also compared these to AZD-5069, a slowly dissociating reversible small molecule antagonist (Nicholls et al., 2015) that may bind to an intracellular binding site previously identified for other CXCR2 small molecules (Bradley et al., 2009; Salchow et al., 2010; Liu et al., 2020). As shown in Fig. 6, all antagonists generated different profiles at CXCL1 that were broadly consistent with their reported pharmacological profiles. The N-terminal binder 127D1 acted as a negative allosteric modulator, where increasing concentrations of antagonist could no longer compete with the agonist due to co-operativity, leading to characteristic “stacking” of the agonist curves (Fig. 6A). However, the reported competitive binder 163E3 caused a decrease in the maximal response and did not show the competitive antagonist profile previously reported (Fig. 6B). The small molecule AZD-5069 also caused a decrease in the maximal response of CXCL1, appearing as a noncompetitive antagonist (Fig. 6C), which is likely due to its previously reported slow dissociation rate. This contrasts with the responses observed at CXCL8, where only AZD-5069 and 300 nM 163E3 showed any significant inhibition of the CXCL8 response (Fig. 6, D–F). Furthermore, unlike the CXCL1 inhibition, AZD-5069 failed to fully inhibit CXCL8 responses at 1 µM and much higher concentrations of compound are required to fully block the signaling of this chemokine. As none of these antagonists show competitive profiles, we could not determine a pA2 for any of the compounds.

Fig. 6.

Schild analysis AZD-5069 and CXCR2 nanobodies. Antagonists were preincubated with neutrophils for 1 hour at RT before addition of CXCL1 (A–C) or CXCL8 (D–F). Data are pooled (mean ± S.D.) of at least three independent experiments from individual donors performed in duplicate, normalized to vehicle treated 30 nM CXCL1 or 30 nM CXCL8 responses.

Neutrophil DMR Assays Yield Comparable Results to Traditional CD11b and Chemotaxis Assays

We have demonstrated that a DMR approach can be successfully applied to neutrophils to investigate chemokine pharmacology and provide robust and reproducible data. Given the holistic nature of the DMR signal, we aimed to confirm our observations in well-established assays. To support this, we assessed the upregulation of the CD11b integrin by flow cytometry, a standard read-out validated as a target engagement marker in neutrophils (Liston et al., 1998; Lazaar et al., 2011; Miller et al., 2015; Namour et al., 2016), and we used the low throughput, but more mechanistic, in vitro chemotaxis assay. All chemokines examined induced concentration-dependent upregulation of CD11b with results broadly complimenting those from the DMR assay (Fig. 7, A and B; Table 6); however, in contrast to the DMR assay, both the CD11b and chemotaxis assays show much greater donor to donor variability (Supplemental Fig. 3). This variability for the chemotaxis assays particularly meant we could test fewer antagonists in this assay. It also precluded the determination of an EC80 and prohibited standard IC50 determination in this assay format. A further advantage of the DMR assay is the lack of “bell-shaped” response that is characteristic of the chemotaxis assay.

Fig. 7.

Summary of CD11b and chemotaxis characterization data. CXCR1/2 chemokines induce upregulation of CD11b (A) with potencies correlating with potencies in the DMR assay (B). Antagonist activity (pIC50) was determined by challenge of an EC80 of CXCL1 (C) or CXCL8 (D) after preincubation of the antagonist for 30 minutes. CXCL1; Schild analysis of 127D1 in CD11b (E) and chemotaxis (F) assays and 163E3 in CD11b (G) and chemotaxis (H) assays. CXCL8; Schild analysis of 127D1 in CD11b (I) and chemotaxis (J) assays, and 163E3 in CD11b (K) and chemotaxis (L) assays. Schild analysis of AZD-5069 to inhibit CXCL1 (M) and CXCL8 (N) in CD11b assays. CD11b responses were normalized to 3 nM fMLF response and data are expressed as pooled (mean ± S.D) of three to nine independent donors. Chemotaxis responses were normalized to 10 nM chemokine response and data are expressed as pooled (mean ± S.D.) of four independent experiments from individual donors performed in quadruplicate. Data in the correlation plot is summarized from Table 6.

TABLE 6

Summary of chemokine responses in neutrophil CD11b upregulation assays

When challenged against an EC80 concentration of CXCL1, AZD-5069, 127D1, and 163E3 all inhibited responses with pIC50s of 8.19 ± 0.19, 9.55 ± 0.24, and 8.30 ± 0.26, respectively, which were consistent with their pIC50s in the DMR assay (Fig. 7C). However, AZD-5069 was the only antagonist to inhibit CXCL8 responses in the CD11b assay (Fig. 7D), and its potency was 69-fold lower at 6.53 ± 0.10. This is consistent with the DMR data and further supports that both CXCR1 and CXCR2 are involved in neutrophil chemokine responses. We also wanted to reproduce the Schild neutrophil DMR experiments to determine whether the noncompetitive profiles of the antagonists tested would also be mirrored in both CD11b and chemotaxis assays.

While the lower throughput of the chemotaxis prevented us from testing a wider range of antagonist concentrations, both assay formats agreed with the negative allosteric modulator profile of 127D1 and the reduction in maximal response with 163E3 when tested against CXCL1 responses (Fig. 7, E–H). In contrast, neither nanobody could block CXCL8 responses in either the CD11b or chemotaxis assays (Fig. 7, I–L). Furthermore, AZD-5069 demonstrated inhibitory profiles consistent with the DMR responses in CD11b, with noncompetitive inhibition of CXCL1 and increased concentrations required to fully inhibit CXCL8 (Fig. 7, M and N).

Neutrophil DMR Assays Can Be Applied to a Range of GPCRs

Finally, following the success of neutrophil DMR assays for detecting CXCR1 and CXCR2 responses, we investigated the application of the assay to other GPCRs expressed by neutrophils. As FPR1 responses in neutrophil DMR assays have previously been characterized (Christensen et al., 2017), the FPR1 agonist fMLF was chosen as a positive control. A range of ligands were then tested, including those of classic neutrophil chemoattractant receptors such as the formyl peptide receptors (FPR1, FPR2, and FPR3), leukotriene B4 receptors (BLT1 and BLT2), platelet activating factor (PAF) receptor, and complement component 5a receptor 1 (C5aR1), as well as other GPCRs that have been described as expressed on neutrophils, such as the free fatty acid receptor (FFAR) 2, and purinergic P2Y receptors (Le Poul et al., 2003; Futosi et al., 2013; Wang and Chen, 2018; Frei et al., 2021).

Responses to all ligands tested were detected and were shown to be reproducible and robust over multiple donors (Fig. 8; Table 7). All ligands exhibited a rapid initial peak within a few minutes of compound addition, similar to that observed with the CXCR1/CXCR2 chemokines, with some ligands, such as leukotriene B4, showing a later secondary peak (Fig. 8B), although for the purposes of analysis only the first peak was quantified. FPR responses were the largest in terms of wavelength change, with responses to both FPR1 and FPR2 selective ligands detected. As with the chemokine DMR studies, these ass