CCL20 neutralization by a monoclonal antibody in healthy subjects selectively inhibits recruitment of CCR6⁺ cells in an experimental suction blister

Abstract

Aims

GSK3050002, a humanized IgG1κ antibody with high binding affinity to human CCL20, was administered in a first‐in‐human study to evaluate safety, pharmacokinetics (PK) and pharmacodynamics (PD). An experimental skin suction blister model was employed to assess target engagement and the ability of the compound to inhibit recruitment of inflammatory CCR6 expressing cells.

Methods

This study was a randomized, double‐blind (sponsor open), placebo‐controlled, single‐centre, single ascending intravenous dose escalation trial in 48 healthy male volunteers.

Results

GSK3050002 (0.1–20 mg kg⁻¹) was well tolerated and no safety concerns were identified. The PK was linear over the dose range, with a half‐life of approximately 2 weeks. Complex of GSK3050002/CCL20 increased in serum and blister fluid with increasing doses of GSK3050002. There were dose‐dependent decreases in CCR6⁺ cell recruitment to skin blisters with maximal effects at doses of 5 mg kg⁻¹ and higher, doses at which GSK3050002/CCL20 complex in serum and blister fluid also appeared to reach maximum levels.

Conclusions

These results indicate a relationship between PK, target engagement and PD, suggesting a selective inhibition of recruitment of CCR6⁺ cells by GSK3050002 and support further development of GSK3050002 in autoimmune and inflammatory diseases.

What is Already Known about this Subject

• CCL20 and its receptor CCR6 are found up‐regulated in tissues of patients with chronic inflammatory conditions.

• CCR6 is the main chemokine receptor of Th17 cells, which are implicated in many chronic inflammatory conditions.

What this Study Adds

• Levels of GSK3050002/CCL20 complex appeared to increase in a dose‐dependent manner and reach maximum levels at doses of 5 mg kg⁻¹ and higher both in serum and interstitial fluid, suggesting that a saturable target engagement has been achieved at the high doses.

• GSK3050002 inhibited recruitment of CCR6⁺ cells selectively into an experimental suction skin blister.

Tables of Links

TARGETS
GPCRs 2	CCR6
CCR4	CXCR3

LIGANDS
CCL1	CCL22
CCL2	CXCL10
CCL20	CXCL8

These Tables list key protein targets and ligands in this article that are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 1, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 2.

Introduction

Chemokines and chemokine receptors are involved in leukocyte recruitment and play an important role in the pathogenesis of inflammatory diseases. With over 50 chemokines and 20 chemokine receptors, they form a complex biological system with most chemokines being classified as inflammatory, although several have been attributed mainly homeostatic functions or even a dual function 3. Many chemokines act as ligands for multiple receptors and vice versa (redundancy) and affinity of chemokines for receptors may be altered by binding to glycosaminoglycans and by homo‐ or heterodimerization 4, 5. Also, certain chemokine receptors may act as decoy receptors, or interceptors, by binding chemokines, but not resulting in signalling 6. Such a complex system with redundancy and altered functionality depending on the local environment is a likely contributor to the failure of many clinical trials evaluating therapeutic targeting of chemokine biology, and this has to be taken into account for the development of more effective inhibitors of chemokines and chemokine receptors.

The C‐C motif chemokine ligand 20 (CCL20), or macrophage inflammatory protein‐3 alpha, has low homology to other chemokines and is the only chemokine known to interact with CC chemokine receptor 6 (CCR6) 7, 8, 9. CCL20 is normally expressed constitutively at low levels, but can be induced in response to pro‐inflammatory cytokines and through activation of Toll‐like receptors 7. Tissues that show constitutive mRNA expression of CCL20 include liver, lung and appendix. Lower levels are detected in the lymph nodes, intestine and skin 7. CCR6 is expressed on a variety of immune cells including memory and regulatory T‐cells 8, 9, naïve and activated B cells 10, dendritic cells 11 and Th17 T‐cells 12. Both CCL20 and CCR6 are found to be upregulated in diseased tissue of patients with chronic inflammatory conditions such as psoriasis, arthritic conditions and inflammatory bowel disease 13, 14, 15. Inhibition of CCL20 in preclinical models has been shown to inhibit disease activity in models of skin inflammation, arthritis and colitis 16, 17, 18, 19 and CCR6 knockout mice are more resistant to autoimmune and inflammatory‐mediated diseases, including psoriasis and arthritis 16, 20. Thus, an inhibitor of the CCL20–CCR6 pathway may be effective in the treatment of diseases characterized by trafficking of immune cells to sites of inflammation.

GSK3050002 (also known as E6071) was developed by Morphotek, KAN Research Institute, Inc. and Eisai Co., Ltd., as a humanized IgG1κ antibody with high binding affinity (48 pM) to human CCL20. Here we report the results of a first‐in‐human study, in which this neutralizing antibody was evaluated in a randomized, placebo‐controlled, single dose escalation study in male healthy volunteers to assess safety, tolerability, pharmacokinetics, immunogenicity and biological activity. An experimental skin suction blister model 21 was employed in this study to assess pharmacokinetics, target engagement and the ability of the antibody to selectively inhibit recruitment of inflammatory CCR6 expressing cells.

Materials and methods

Study design and participants

As no preclinical reproductive toxicity testing has been completed, the study was restricted to healthy male volunteers (aged 18–65 years) based on medical history, physical examination, electrocardiograms (ECG) and laboratory testing. Other key eligibility criteria included body mass index between 18 and 29 kg m⁻², alanine amino transferase, alkaline phosphatase and bilirubin ≤ 1.5 × upper limit of normal, and QTcF < 450 msec based on a single ECG. Subjects with current or chronic history of liver disease or known hepatic or biliary abnormalities (with the exception of Gilbert's syndrome or asymptomatic gallstones) as well as heavy smokers were excluded. Subjects could not have used prescription or non‐prescription drugs or have been immunized with live‐attenuated vaccines within 4 weeks prior to dosing until 19 weeks after dosing.

This study was designed as a randomized, double‐blind (sponsor open), placebo‐controlled, single ascending dose escalation trial (ClinTrials.gov identifier: NCT01984047). The subjects and investigators were blinded to treatment allocation, but unblinded review of safety data by the central sponsor study team was allowed at dose escalation committee meetings where necessary. Subjects were enrolled in six sequential cohorts of eight subjects each to receive single ascending intravenous (IV) doses, infused over 2 h. Within each cohort, subjects were randomized to receive either GSK3050002 or placebo in a 3:1 ratio. Safety and pharmacokinetic data (including adverse events, vital signs, ECGs, clinical chemistry and haematology) from 14 days post‐dose from at least three of the six subjects who had received GSK3050002 must have been available in order for a dose escalation decision to be taken.

Subjects remained in‐house for approximately 3 days after dosing, and returned for 8 outpatient visits, scheduled over the following 81 days after discharge from the unit, and a subsequent follow‐up visit 7–14 days after the last study visit.

The determination of the planned starting dose for this study and the dose range tested were according to the principles of the European Agency for the Evaluation of Medicinal Products (EMEA) and Food and Drug Administration (FDA) guidelines 22, 23. The initial dose, determined from the animal studies used in the non‐clinical assessments of both the pharmacology and toxicology of GSK3050002, provided reasonable assurance that there were no undue or unforeseen risks for the first administration of GSK3050002 to humans at the dose levels proposed in this study.

The study was conducted at a single study centre (GlaxoSmithKline Clinical Unit Cambridge, Addenbrooke's Centre for Clinical Investigation, Addenbrooke's Hospital, Cambridge) in the United Kingdom according to the ethical principles of ‘good clinical practice’ (GCP) and the Declaration of Helsinki after obtaining written informed consent from each patient. The protocol and its amendments were approved by the local Ethics Committee (NHS Health Research Authority NRES Committee South Central – Berkshire B).

Pharmacokinetic assessments

Serum PK concentrations

In each of the cohorts, blood samples (2.5 ml) for PK analysis of serum GSK3050002 were collected on Days 1, 2, 3, 7, 14, 21, 28, 42, 56 and 84. On Day 1, samples were taken at pre‐dose, at the end of IV infusion (approximately 2 h), at 4 h and 12 h post‐dose. On Day 2, samples were taken at 24 h and 36 h post‐dose.

Initially, serum samples from treatment groups 0.1 and 0.5 mg kg⁻¹ were analysed for GSK3050002 using a validated analytical method based on sample dilution, followed by colorimetric immunoassay analysis. Samples were diluted fivefold with assay buffer and GSK3050002 was captured with biotinylated h‐MIP‐3α (h‐CCL20) antigen and detected using a horseradish peroxidase labelled mouse anti‐human IgG₁ antibody. The lower limit of quantification (LLQ) for GSK3050002 was 0.1 μg ml⁻¹ using a 50 μL aliquot of fivefold diluted human serum with a higher limit of quantification (HLQ) of 2.5 μg ml⁻¹.

After initial analysis, matrix (serum) interference with the immunoassay was observed in one subject in the 0.1 mg kg⁻¹ treatment group and two subjects in the 0.5 mg kg⁻¹ treatment group. Subsequently, an additional analytical method based on sample dilution, immunocapture and trypsin digest, followed by ultra high pressure liquid chromatography tandem mass spectrometry (UHPLC–MS/MS) analysis was validated. Samples were diluted fivefold with assay buffer and GSK305002 was extracted and immunocaptured with biotinylated h‐MIP‐3α (h‐CCL20) antigen followed by tryptic digest. The tryptically derived representative peptide ‘LLIYGATNLADGVPSR’ was selected from the complementary determining region of GSK3050002 and was quantified by UHPLC–MS/MS using a TurboIonSpray™ interface and multiple reaction monitoring. The LLQ in this second method for GSK3050002 was 0.2 μg ml⁻¹ using a 50 μL aliquot of fivefold diluted human serum with a HLQ of 20 μg ml⁻¹. A cross‐validation was performed between these two methods using anonymized, pooled samples from subjects dosed with GSK3050002 that had no observed interference and demonstrated these two methods were equivalent. The three subjects with matrix interference from the 0.1 and 0.5 mg kg⁻¹ treatment groups as well as samples from subjects from treatment groups 1, 5, 10 and 20 mg kg⁻¹ GSK3050002 were analysed and reported using the UHPLC–MS/MS‐based assay.

Blister PK concentrations

GSK3050002 was quantified in human blister fluid, obtained on Baseline, Day 3 and Day 15, by an exploratory method based on the validated human serum immunocapture UHPLC–MS/MS method. The analytical method format, sensitivity, assay range and tryptically derived representative peptide fragment were the same as those used for the human plasma method. Human serum standard calibrators and quality controls were used as a surrogate matrix to analyse the human blister fluid samples. Human serum and human blister fluid were crossed over and shown to be equivalent.

PK and PKPD analyses

Actual blood sampling times were used to determine individual pharmacokinetic (PK) parameters. The GSK3050002 concentration–time results were used to assess the following PK parameters: maximum serum concentration (C _max), area under the curve (AUC), clearance (CL), steady‐state volume of distribution (V _ss), and terminal half‐life (t _1/2). The GSK3050002 concentration–time data were subjected to noncompartmental PK analysis using WinNonlin Phoenix version 6.4.

In addition, concentration–time data were analysed using a population PK method. Finally, a PKPD model was developed using observed drug and drug–CCL20 complex data in blood to characterize the in vivo binding affinity to CCL20.The software NONMEM (version7.2 ICON Solutions) and PsN (Perl Speaks NONMEM, version 3.2.4) were used for the development of the nonlinear mixed effects PK/PD models.

Statistical analyses

No formal hypothesis testing or comparisons were made between treatment groups. The sample size was based on feasibility. Assessments of the dose–response relationship of PK endpoints from the noncompartmental PK analysis and of preliminary dose proportionality were conducted. An estimation approach was used to assess the effect of single doses of GSK3050002 relative to placebo for the PD endpoints.

Safety assessments

Safety assessments were conducted at Baseline and throughout the study until the follow‐up visit and included the following: adverse events (AEs), clinical laboratory evaluations (chemistry, haematology and urinalysis), vital signs and 12‐lead ECG results. These data were descriptively analysed.

Immunogenicity

The immunogenicity assays involved consecutive screening, confirmation and titration. Samples were collected from all study subjects, including those receiving placebo, at the following times: predose, Days 14, 28 and 84. The samples were analysed for the presence of anti‐GSK3050002 antibodies using a validated electro‐chemiluminescent (ECL) assay on the Meso Scale Discovery (MSD) platform. Briefly, samples were mixed to generate a homogeneous mixture containing 250 ng ml⁻¹ biotinylated GSK3050002, ruthenylated GSK3050002 and 5% serum sample in assay diluent (1% casein in PBS). Samples were incubated for 1 h on a streptavidin MSD plate that was blocked overnight with casein in phosphate buffered saline (PBS). After blocking and sample incubation, the plates were washed with 0.1% Tween‐20 in PBS. Following incubation, MSD read buffer (2×) was added to the plates and analysed using a MSD Sector Imager 6000. The resulting intensity of luminescence is directly proportional to the quantity of detected antidrug antibodies in the sample. Normal human serum and anti‐GSK3050002 antibody spiked in normal human serum were used as negative and positive controls, respectively. The screening assay cut point was generated by analysing an adequate number of normal human samples to provide a statistically valid assessment of both biological and assay variability. Using a risk‐based approach, the screening cut point was determined at upper 95% confidence limit. A confirmatory assay was used to determine whether a screening positive response was specific for the drug. The confirmation cut point was determined based on the percentage of signal inhibition due to addition of free drug using the screening assay format. If samples were exposed to an excess concentration of free drug with a resulting signal inhibition percentage greater than or equal to the confirmatory assay cut point, then the antidrug antibody response was said to be specific for the drug. The confirmed antidrug antibody positive samples were further run in titration assay to determine relative titres. In this assay, the screening cut point was used as the titration cut point and the assay titre was the dilution factor interpolated at the cut point multiplied by the assay minimum required dilution. All samples (pre‐dose and post‐doses) were initially tested by the above antidrug screening assay and those samples with potential anti‐GSK3050002 antibodies were confirmed by immunocompetition using excess drug. Only samples that tested positive in the confirmation step were reported as positive and these samples were then tested at multiple dilutions to obtain the antibody titre. Subjects in whom anti‐GSK3050002 antibodies were still detectable at Day 84, but for whom pre‐existing (negative at Day −1) anti‐GSK3050002 antibodies were not detected, were further monitored at or around 6 months post‐dose.

Experimentally‐induced suction blister assessments

Suction blisters were induced as described by Akbar et al. 21. Briefly, negative pressure was applied via a suction chamber on naive skin of the volar (flexor) surfaces of the left or right forearm for approximately 2–4 h using a clinical suction pump at Day −1 and on Days 2 and 14 post‐dose. The procedure was performed at warm room temperature (~22°C) until a single unilocular blister was formed. The blister was protected overnight with a rigid adhesive dressing. Blister fluid was harvested approximately 21 (± 3) h after skin blister induction, in order to measure target engagement and PD effects.

Free and complex CCL20 in blood and blister fluid

Free CCL20 levels as reported for the clinical study were measured in the serum and blister fluid by an enzyme linked immunoassay (ELISA; LLQ 78 pg ml⁻¹). Plates were coated with anti human CCL20 (R&D Systems) and CCL20 in serum samples detected using subsequent incubations with biotinylated goat anti‐human CCL20, streptavidin‐HRP and 3,3,5,5,′tetramethylbenzidine. Absorbance was measured at 450 nm. For selected samples CCL20 was also measured by MSD and following prior protein A/G depletion. MSD plates were coated overnight with GSK3050002 at 5 μg ml⁻¹, washed with 0.1% Tween‐20 in PBS and blocked with diluent 2 (MSD) for 30 min at room temperature. Subsequently, plates were washed and samples (25 μl, undiluted) incubated for 1 h at room temperature. After washing, 25 μl ruthenylated detection antibody (mouse anti‐CCL20) was incubated for 1 h at room temperature followed by washing and addition of 150 μl read buffer (2×; MSD). Luminescence was detected using a MSD Sector Imager 6000. For protein A/G depletion, samples were loaded onto Handee Spin cups with Protein A/G beads (Thermo‐Fisher), mixed and incubated for 1 h at room temperature on an orbital shaker. The cups were then centrifuged at 300 × g and the flowthrough collected for MSD analysis as above. In blister fluid, also CCL2, CCL3, CCL4, CCL11, CCL13, CCL17, CCL22, CCL26, CXCL10, IL‐6, TNF‐α, IL‐1β and IFN‐γ were measured using a validated MSD kit (V‐plex) according to manufacturer's protocol.

Serum and blister fluid levels of GSK3050002‐CCL20 complex were measured using an ECL‐based ligand binding immunoassay on the MSD platform (LLQ 156.25 pg ml⁻¹). Total CCL20 was captured on streptavidin MSD plates using biotinylated goat anti‐CCL20 polyclonal antibody and GSK3050002/CCL20 detected using ruthenylated mouse anti‐idiotype antibody.

Immune cell phenotyping and quantification

Immune cell phenotyping was performed on whole blood and blister fluid using antibody panels as described in Table S1. Selected immune cells in blister fluid were gated as follows: T‐cells (CD3⁺), Th1 (CD3⁺CD4⁺CXCR3⁺CCR6⁻), Th2 (CD3⁺CD4⁺CXCR3⁻CCR6⁻), Th17 (CD3⁺CD4⁺CXCR3⁻CCR6⁺) and Th1/17 (CD3⁺CD4⁺CXCR3⁺CCR6⁺) T‐cells, while monocytes and granulocytes were gated based on forward and side scatter. Flow cytometry was performed using a BD FACS Canto II.

In vitro chemotaxis

Chemotaxis was assessed using Transwell inserts (Costar). CD4⁺ memory T‐cells were purified from peripheral blood mononuclear cells (PBMC) using magnetic beads (Miltenyi) and 1 × 10⁶ cells were allowed to migrate towards the indicated chemokines (all R&D Systems) for 2 h at 37°C. The inserts were then removed, migrated cells collected and cells analysed by flow cytometry using Trucount tubes (BD Biosciences) for quantification of absolute cell numbers. Migration is presented as an index with migration towards tissue culture medium set at one. Chemokinesis controls were included for most experiments by including the same concentration of chemokine in both the upper and lower well. Chemotaxis towards CCL20 was also performed in the presence of increasing concentrations of GSK3050002. Migration of PBMC was also tested in response to cultured HT‐29 cells that were stimulated overnight with 10 ng ml⁻¹ tumour necrosis factor‐α (TNF‐α). Migrated cells were identified by flow cytometry using CD3, CD4, CD25, CD127, CCR4, CCR6, CCR8 and CXCR3. For migration in response to HT‐29 cells, a panel consisting of CD3, CD4, CD14, CD19, CD56, CD66b and CCR6 was used. Flow cytometry was performed using a BD Canto II.

Results

Demographics and disposition

A total of 48 male subjects were dosed and completed the study as planned. The subjects were well balanced with respect to demographic parameters across the six cohorts. The mean (SD) demographics were 38.9 (8.70) years of age, 80.5 (9.85) kg body weight, 25.3 (2.19) kg m⁻² body mass index.

Safety

There were no clinically important changes in vital signs, ECGs, or clinical laboratory parameters in the study. Additionally, there were no deaths, serious AEs or withdrawals due to AEs. Administration of GSK3050002 at all doses was well tolerated. The AE profile for GSK3050002, across all dose groups, was similar to that for the placebo group and there was no apparent dose relationship for the types or frequencies of reported AEs in the GSK3050002 treatment arms (Table 1). The majority of AEs were reported by single subjects with nasopharyngitis and headache the most frequently reported AEs across all treatment groups during the study. The majority of AEs were mild to moderate in intensity. There were no AEs deemed related to the study drug during the blinded review of AEs by the investigator in the GSK3050002 groups.

Table 1.

Summary of adverse events reported by >2 subjects across all dose groups

	GSK3050002
Preferred term	0.1 mg kg −1 (n = 6)	0.5 mg kg −1 (n = 6)	1 mg kg −1 (n = 6)	5 mg kg −1 (n = 6)	10 mg kg −1 (n = 6)	20 mg kg −1 (n = 6)	Placebo (n = 12)
ANY EVENT	3 (50%)	4 (67%)	2 (33%)	4 (67%)	6 (100%)	6 (100%)	11 (92%)
Nasopharyngitis	1 (17%)	0	0	0	2 (33%)	2 (33%)	6 (50%)
Headache	1 (17%)	2 (33%)	0	1 (17%)	0	3 (50%)	3 (25%)
Gastroenteritis	0	0	1 (17%)	1 (17%)	0	1 (17%)	0
Abdominal discomfort	0	0	0	0	2 (33%)	0	0
Toothache	1 (17%)	0	0	0	0	1 (17%)	1 (8%)
Seasonal allergy	0	1 (17%)	0	0	0	0	2 (17%)
Contusion	0	0	0	1 (17%)	0	1 (17%)	0
Decreased appetite	0	1 (17%)	0	0	1 (17%)	0	0
Arthralgia	0	0	0	0	0	0	2 (17%)
Back pain	0	0	0	0	0	1 (17%)	1 (8%)
Cough	0	0	0	1 (17%)	0	2 (33%)	0
Oropharyngeal pain	0	0	0	0	2 (33%)	0	1 (8%)

Pharmacokinetics

The pharmacokinetics of serum GSK3050002 were linear over the dose range of 0.1–20 mg kg⁻¹, with a mean terminal half‐life (t _1/2) of approximately 2 weeks (Figure 1A and Table 2). The typical distribution volume ranged from 7.7 to 9.8 l (for a 70‐kg body weight) and the systemic clearance ranged from 15.3 to 23 ml h⁻¹ (for a 70‐kg body weight) as derived from AUC and t _1/2 values in Table 2 (excluding data from 0.1 mg kg⁻¹). Dose proportionality was assessed using a power model analysis in which the estimated mean slope for C _max and AUC (with 90% confidence interval [CI]) were 1.025 [0.996, 1.054] and 1.079 [1.038, 1.120], respectively. In the suction blister fluids, the average dermal interstitial concentration of GSK3050002 was approximately 20% of the concentration in the serum when measured on Days 3 and 15 (Figure 1B).

Figure 1.

Linear PK distribution of GSK3050002 in serum and blister fluid. (A) GSK3050002 in serum, (B) GSK3050002 in blister fluid. Data is presented as mean + SD (serum: n = 12 for placebo, n = 6 for GSK3050002; blister fluid: n = 9–11 for placebo, n = 3–6 for GSK3050002 with the exception of 0.1 mg kg⁻¹; n = 1–3)

Table 2.

Summary of selected non‐compartmental PK parameters for plasma GSK3050002 concentration

Treatment group	n	AUC 0–∞ (μg.h ml −1 ) a	C max (μg ml −1 ) a	t 1/2 (h) a
0.1 mg kg−1	6	298 b(35.97)	2.24 (42.59)	98.6 b (39.77)
0.5 mg kg−1	6	1520 (30.83)	11.3 (8.49)	294 (63.48)
1 mg kg−1	6	4390 (15.11)	23.4 (14.11)	334 (11.84)
5 mg kg−1	6	19 300 (20.86)	117 (11.74)	314 (20.70)
10 mg kg−1	6	43 200 (29.20)	257 (10.92)	369 (19.71)
20 mg kg−1	6	91 600 (19.35)	503 (10.96)	349 (13.28)

^aGeometric mean (%CVb)

^b n = 5

Results from the population PK analyses were consistent with the noncompartmental PK analysis. A two‐compartmental model was retained for the final model to describe the plasma PK data. The typical central volume was 3.63 l (95% CI: 3.44–3.83), the typical peripheral volume was 3.19 l (2.89–3.52) and the typical systemic clearance was 19.8 ml h⁻¹ (95% CI: 18.3–21.4). Inter‐compartmental clearance was 15.6 ml h⁻¹ (95% CI: 13.5–18.0).

Immunogenicity

The reporting of incidence of anti‐GSK3050002 antibodies was based on recent white paper recommendations 24. A total of seven of the 36 subjects developed antidrug antibodies post‐dose, of which four had been treated with GSK3050002 and three with placebo (Table 3). In addition, pre‐existing antibodies were detected for three subjects (one in the GSK3050002 group and two in the placebo group). No direct correlations could be established between the presence of anti‐GSK3050002 antibodies with specific AEs or the PK profiles.

Table 3.

Summary of individual anti‐GSK3050002 titres for positive subjects

Treatment group	Day				6 Month follow‐up a
	−1	14	28	84
Subjects with pre‐existing anti‐GSK3050002 antibodies at baseline
0.5 mg kg−1	+ (320)	–	–	–	Not doneb
Placebo	+ (320)	+ (1280)	+ (320)	+ (640)	Not done
Placebo	+ (320)	+ (320)	+ (320)	+ (320)	Not done
Subjects with post‐dose anti‐GSK3050002 antibodies
5 mg kg−1	–	–	–	+ (320)	Negative
10 mg kg−1	–	–	+ (80)	+ (80)	Negative
Placebo	–	–	–	+ (80)	Negative
Placebo	–	–	+ (40)	+ (40)	Negative
1 mg kg−1	–	–	+ (320)	–	Not done
20 mg kg−1	–	+ (40)	–	–	Not done
Placebo	–	–	+ (80)	–	Not done

^aSubjects with post‐dose anti‐GSK3050002 antibodies on Day 84 were further monitored at approximately 6 months post‐dose.

^bSubjects with anti‐GSK3050002 antibodies at Baseline or with a transient response that did not persist till Day 84, were not considered for 6 month follow‐up sample.

Of the four subjects dosed with GSK3050002 that had detectable antibody levels post‐dose, three had transient anti‐GSK3050002 antibody responses that were detected only at one sampling time point during the course of the treatment. None of the subjects that showed a persistent anti‐GSK3050002 antibody response until Day 84 remained positive when followed up at 6 months post‐dose. Antibody titres in all the treatment‐induced subjects ranged from 1:40 to 1:320 and peaked at approximately 2–4 weeks post‐dosing. One treated subject had a pre‐dose transient titre response that was 16‐fold above the minimum required dilution (i.e. 320), but did not show any boosting effect post‐treatment. Three other subjects had transient titre responses (2‐ and 16‐fold above the minimum required dilution) and one subject a persistent titre response (fourfold above minimum required dilution) that later confirmed negative. All the subjects that showed treatment‐induced antidrug antibody had values that were just above the cut point with transient low titres. Of the placebo subjects, two with pre‐existing antibodies had consistent antibody titres throughout the study. The other three placebo subjects showed transient titres that were only twofold above the minimum required dilution and were completely negative by the next sampling point. In general, for both the placebo and the treated group, the titration values were approximately twice the low positive control and were transient in nature. Furthermore, the titration dilutions displayed a very gradual decrease and the samples had a less than twofold decrease in the assay response with each dilution as opposed to an expected rapid decrease with each dilution (signal decreasing twofold for each dilution). This could be indicative of assay interference by factors other than the antidrug antibody. Taken together, the data suggest that low concentrations of antidrug antibody were detected in four subjects that received GSK3050002 and there was no apparent effect of the presence of anti‐GSK3050002 antibodies on the PK of GSK3050002.

Target engagement

Serum concentrations of drug–CCL20 complexes increased with increasing doses of GSK3050002 and appeared to reach maximum levels at doses ≥5 mg kg⁻¹ (Figure 2A). For the placebo group, the serum drug–CCL20 levels for all subjects remained at below the level of quantification (BLQ) and did not change over time. In skin blister fluid, drug–CCL20 complex levels also increased in a dose‐dependent manner and complexed CCL20 levels in blister fluid were similar at doses ≥5 mg kg⁻¹ on Days 3 and 15, suggesting saturation at these doses (Figure 2B). Complex concentrations in blister fluid were approximately 20% of those measured in serum at the corresponding time points. This blood:tissue ratio of 5:1 for drug–CCL20 complex accumulation reflects the same distribution of GSK3050002 to the interstitial dermal compartment observed in PK assays.

Figure 2.

Dose dependent increase over time of complexed and free CCL20 in serum and blister. (A) GSK3050002/CCL20 complex in serum, (B) GSK3050002/CCL20 complex in blister fluid, (C) free CCL20 by ELISA in serum and (D) free CCL20 by ELISA in blister fluid. (E) Free CCL20 comparison of ELISA, MSD and protein A/G depletion of samples of a single subject that received 20 mg kg⁻¹ GSK3050002. (F) Comparison of predicted vs. actual GSK3050002/CCL20 concentration data. Data is presented as mean + SEM in A–D (serum: n = 12 for placebo, n = 6 for GSK3050002; blister fluid: n = 9–11 for placebo, n = 3–6 for GSK3050002). Data from D of 0.1 mg kg⁻¹ cohort is not included as this was not measured with the validated assay. Data in F shows actual data as data points with error bars representing SEM and the lines represent predicted data with 95% confidence intervals as shaded areas

Surprisingly, concentrations of free CCL20 in serum also appeared to increase in a dose‐dependent manner and this was mirrored in the skin blister fluid, with free CCL20 levels still being elevated at Day 15 in all dose cohorts compared to placebo (Figure 2C–D). Levels of free CCL20 in blister fluid were similar at all time points in the placebo group. As these levels of free CCL20 detected in both serum and blister fluid after GSK3050002 were unexpected, free CCL20 in serum was also evaluated using an electro‐chemiluminescent‐based detection method (MSD), for which samples did not require dilution. Interestingly, when samples were analysed by MSD, levels of free CCL20 appeared reduced compared to the results using ELISA. Furthermore, when samples were applied to a protein A/G column prior to free CCL20 detection, to deplete immunoglobulins, including free as well as CCL20‐complexed GSK3050002, a further reduction in free CCL20 was observed (Figure 2E), suggesting sample processing may result in disassociation of CCL20–GSK3050002 complex.

A PKPD model using only PK and drug–CCL20 complex data was developed to assess the in vivo affinity of GSK3050002 to CCL20 target. Free CCL20 data in serum were not used in the model due to unexpected increased levels compared with baseline. Free target was assumed to be synthesized at a zero‐order rate (k _syn) and degraded at a first‐order rate (k _deg). For simplicity, we assumed target binding only in the central compartment between antibody concentration and CCL20 levels. A quasi‐equilibrium solution was considered among drug, free target and drug–CCL20 complex 25. Estimated in vivo affinity was approximately 350 pM with a rapid (approximately 10–20 min) CCL20 target turnover estimated based on drug–CCL20 complex data (Figure 2F).

Effects on CCR6‐expressing cells

The experimental skin suction blister provided an opportunity to evaluate the PD effects of GSK3050002 on CCR6⁺ cells. Analysis of immune cells that are recruited into the blister following its formation showed that CCR6⁺ CD4⁺ T‐cells, but not CD8⁺ T‐cells, were enriched in blister fluid when compared to blood (Figure 3A). Furthermore, CCL20 expression levels in blister fluid at baseline were significantly higher than in blood (Figure 3B).

Figure 3.

Characterization of the suction blister as a model to evaluate recruitment of CCR6 expressing cells. (A) Representative flow cytometry data showing enrichment of (i) CCR6⁺CD4⁺ T‐cells, but not (ii) CD8⁺ T‐cells, in the blister compared to blood. (B) Free CCL20 in serum and blister at pre‐dose (baseline). Data in A is from pre‐dose (baseline) samples from the same subject. Data in B, n = 48 of serum (43 at below level of quantification; BLQ) and n = 38 for blister (12 at BLQ), line indicates mean

Administration of GSK3050002 resulted in a dose‐dependent reduction in the percentage of CCR6⁺ T‐cells that were recruited into the skin blisters (Figure 4A). Compared to the percentage of CCR6⁺ T‐cells in blisters induced prior to dosing, a dose‐dependent decrease in percentage of CCR6⁺ T‐cell was observed at both Day 3 and Day 15, which appeared maximal at dose of 5 mg kg⁻¹ and higher (Figure 4B), which is in line with the observed saturation of GSK3050002–CCL20 complex at these doses in both blood and blister fluid (Figure 2A–B). The inhibition of CCR6⁺ T‐cells recruited into the blister correlated with the formation of GSK3050002/CCL20 complex (Figure 4C) and was selective for CCR6⁺ T‐cells, as no inhibition of total CD3⁺ T‐cells (CCR6⁺ and CCR6⁻ combined; Figure 4D), CCR6⁺ monocytes (13 ± 7.6% of monocytes expressing CCR6 at baseline) or CCR6⁺ granulocytes (17 ± 12% of granulocytes expressing CCR6 at baseline) was observed (Figure S1A and C). Neither was inhibition of total monocytes or granulocytes observed (Figure S1B and D). Inhibition of recruitment of CD8⁺ T‐cells was investigated and similar to all CCR6⁺ T‐cells, although more variable, a GSK3050002 dose‐dependent reduction in recruitment of CCR6⁺CD8⁺ T‐cells was observed, but not for total CD8⁺ T‐cells (Figure 4E and F). Investigations into specialized CD4⁺ T helper subsets were then undertaken. Although not verified by cytokine or transcription factor expression, CD4⁺ T‐cells were identified as follows: CXCR3⁻CCR6⁻ as Th2, CXCR3⁺CCR6⁻ as Th1, CXCR3⁻CCR6⁺ as Th17 and CXCR3⁺CCR6⁺ as Th1/17 T‐cells. Inhibition of cell recruitment by GSK3050002 was found to be prominent for CCR6⁺ Th17 cells only. The number of CCR6⁺ Th17 cells recruited into the blisters decreased with increasing doses of GSK3050002 (Figure 4G), while CCR6⁻ T‐cell (Th1 and Th2) numbers slightly increased (Figure 4I and J). The T‐cell population that expressed both CCR6 and CXCR3, designated as Th1/17, neither increased nor decreased in the blister fluid following GSK3050002 dosing, which most likely reflects their heterogeneous composition (Figure 4H). Other CCR6⁺ immune cells could not be analysed in more detail as the volume of blister fluid that was retrieved was limited and only allowed for analysis of one detailed flow panel, which focused on the CD4⁺ T helper subsets. In blood, no differences in immune cell populations following GSK3050002 dosing were observed (data not shown).

Figure 4.

GSK3050002 mediates selective inhibition of CCR6⁺ cells in a suction blister model. (A) Representative histograms showing dose‐dependent inhibition of CCR6⁺ T‐cells. (B) CCR6⁺ T‐cells, (C) CD3⁺CCR6⁺ T‐cell correlation with GSK3050002/CCL20 complex in blister fluid, (D) total CD3⁺ T‐cells, (E) CD8⁺CCR6⁺ T‐cells, (F) total CD3⁺CD8⁺ T‐cells, (G) Th17 T‐cells, (H) Th1/17 T‐cells, (I) Th1 T‐cells and (J) Th2 T‐cells. All data in A is from Day 3, data in B, D–J of indicated cell types is represented as change from baseline, data from Day 3 blister is shown in grey and data from Day 15 blister in black squares. Data in B, D–I is represented as mean ± SEM. In C, circles are data from Day 3, squares from Day 15, open symbols are placebo and GSK3050002 doses are represented as a greyscale with black the highest (20 mg kg⁻¹) dose

In vitro, CCR6⁺ T‐cells, such as Th17 and Th1/17 cells (identified using differential expression of CXCR3 and CCR6 as in the clinical study), migrated towards CCL20, a response that could be inhibited by GSK3050002, whereas CCR6⁻ T‐cells, such as Th1 and Th2 T‐cells, did not migrate towards CCL20 (Figure S2A–B). Th1 and Th2 cells migrated towards CXCL10 and CCL22, respectively (Figure S2C–D). Th17 cells also migrated in response to CCL22 (Figure S2D) and 86% of Th17 cells expressed its receptor CCR4 (Figure S2G). Interestingly, CCL22 was expressed at high levels in the skin blisters (Figure 5 and Table S2), but despite the ability of Th17 cells to respond to CCL22 in vitro, dosing of GSK3050002 inhibited the recruitment of Th17 cells into the blisters. This is supportive of a non‐redundant role for CCL20 in the migration of Th17 cells into the skin blister. In addition to chemokines, the expression of cytokines was also evaluated in the skin blister. No effect of GSK3050002 was observed on expression of IFN‐γ, IL‐1β, IL‐6 or TNF‐α when compared to baseline or placebo (Table S2).

Figure 5.

Suction blisters contain a mixture of pro‐inflammatory chemokines. Data from indicated chemokines were scaled by median absolute deviation after subtraction of the median and shown as a heat map. Left column indicates chemokines measured, right column depicts corresponding chemokine receptors

Regulatory T (Treg) cells, identified as CD4⁺CD25^hiCD127^lo T‐cells, were observed to migrate towards CCL20, which could be inhibited by GSK3050002 (Figure S2A–B). In vitro analysis showed that 71% of Treg cells expressed CCR6, while 94% of Treg cells expressed CCR4 (Figure S2G). As expected, Treg cells migrated potently in vitro towards CCL22 and their response was two‐ to threefold higher than that observed for Th17 cells (Figure S2D). The in vitro data suggested that Treg cells should also be able to migrate in response to CCL1, CXCL10 and CCL22 (Figure S2B–D) and both CXCL10 and CCL22 were present at high levels in the blisters (Figure 5 and Table S2). Unfortunately, the cell numbers in blister fluid was limited and Treg cells could not be analysed.

The comparison of the response to multiple chemokines was hampered by the fact that these in vitro assays depend on the use of recombinant protein rather than native, endogenous proteins. For instance, the migratory response towards CXCL10, although specific for CXCR3⁺ Th1 and Th1/17 cells, seemed modest compared to other chemokines, in particular CCL22, at the same concentration (Figure S2C and D). Therefore, in vitro chemotaxis using a mixture of pro‐inflammatory chemokines produced by TNF‐α stimulated HT‐29 cells 26 was assessed and this also showed that GSK3050002 inhibited migration of CCR6⁺ cells despite the presence of other chemokines (Figures S2F and S3). Both in vitro and in vivo data are supportive of GSK3050002 selectively inhibiting recruitment of CCR6⁺ immune cells in a pro‐inflammatory environment.

Discussion

Administration of GSK3050002 to healthy male volunteers was well‐tolerated. There were no drug‐related AEs, SAEs or deaths during the study. In addition, there were no clinically significant changes in laboratory values, vital signs or ECGs.

The pharmacokinetics of GSK3050002 were typical for an IgG1 monoclonal antibody with high binding affinity to a soluble target 27 without any apparent effect of target‐mediated drug disposition. The pharmacokinetics were linear over the dose range of 0.1–20 mg kg⁻¹, with a mean terminal half‐life of GSK3050002 of approximately 2 weeks. GSK3050002 was also detected in skin blister fluid, but at concentrations of approximately 20% of that in the serum.

A small number of subjects (four out of 36) treated with GSK3050002 developed transient antidrug antibodies post dose, of which none were still positive at a 6‐month follow‐up visit. As the titration assay showed a very gradual decrease in signal with dilution, as opposed to the expected twofold decrease in signal, it is likely that the titres observed may not correspond to actual high concentrations of antidrug antibody. No direct correlations could be established between specific AEs, pharmacokinetics and presence of anti‐GSK3050002 antibodies.

Antidrug antibodies were also detected post‐dose in three subjects in the placebo group. It is unclear why antidrug antibodies were detected, but it is not uncommon to detect transient antidrug antibody responses in placebo groups 28, 29. A possibility could be that endogenous, low‐affinity antibodies, reactive to the drug, could generate a non‐specific antidrug antibody response in the placebo group 30 or that they are the result of false positives in the assay. Antidrug antibodies were also detected at baseline (Day −1) in three subjects (one in the GSK3050002 group and two in the placebo group) indicating the presence of pre‐existing antidrug antibodies. These pre‐existing antidrug antibodies may be components of the natural antibody population or components of the adaptive immune response to environmental antigens or homologous biotherapeutics 31.

High accumulation of GSK3050002–CCL20 complex was observed in both serum and skin blister fluid. Peak levels of complex appeared at doses ≥5 mg kg⁻¹ in both serum and skin blister fluid. An unexpected increase in free CCL20 was also observed in both serum and blister fluid. Limited data on kinetics of chemokine neutralization are available from literature. Following inhibition of CCL2 with high affinity monoclonal antibodies, a similar accumulation of drug–CCL2 complex (>2000‐fold vs. baseline) was observed and interestingly, free CCL2 levels were also found to have increased following dosing 32, 33. Inhibition of IL‐8 (CXCL8) using a monoclonal antibody showed a dose‐dependent inhibition of free IL‐8 in washing fluid, which was present at high levels at baseline, but unfortunately no free IL‐8 levels were measured in blood 34.

It is worth noting that the measured baseline levels of CCL20 unlikely reflect the real synthesis and degradation rates (Baseline = k_syn/k_deg); in fact, the accumulation of drug–target complex can be derived from the ratio of complex and free target degradations (fold increase ~ k_deg/k_complex). Assuming a degradation rate similar to the free drug elimination, accumulation of complexes above 1000‐fold compared with baseline levels can be explained by extremely fast target turnover. An alternative hypothesis is to assume that levels of baseline in blood do not represent the overall target expression, in this case a pool of chemokines sequestered in a matrix and not in equilibrium with the systemic concentrations needs to be assumed.

Subsequent analysis of free CCL20 revealed that using an electro‐chemiluminescent‐based detection method, which did not require predilution of the samples, free CCL20 levels were reduced and when all immunoglobulin, including GSK3050002, was depleted prior to measuring free CCL20, the free CCL20 levels were even further reduced. This would suggest that ex vivo manipulation of serum samples, such as dilution, results in disassociation of CCL20–GSK3050002 complex, and consequently measurement of free CCL20 levels that are higher than in vivo 35. It is unlikely that free CCL20 would increase following GSK3050002 dosing if concentrations at baseline are in equilibrium with CCL20 concentrations in other tissue compartments. While the observed increase of free CCL20 can be at least partially explained by dilution factors or ex vivo handling as shown by the ELISA vs. MSD results, trafficking and subsequent disassociation of GSK3050002–CCL20 complex from a pool compartment into the circulation, can also affect the free CCL20 measurements in blood as the ratio of free and complexed CCL20 is an equilibrium related to the in vivo affinity of GSK3050002. Comparison of PK and PD data, suggest that GSK3050002 should be in excess of the observed free CCL20. In blister fluid, GSK3050002 was present at levels of 30–180 nM on Day 15 for doses of 5 mg kg⁻¹ and higher, while GSK3050002–CCL20 complex was present at 2–3.35 nM. This would suggest that there is a molar excess of at least 15‐fold of free GSK3050002 vs. complex of GSK3050002/CCL20 in the skin blisters. Such excess should, in theory, be able to neutralize the detected free CCL20 levels. Unfortunately, it could not be experimentally verified whether any free CCL20 was indeed present in serum. Attempts were made to evaluate the biological activity of free CCL20, but such assays failed to reach the required sensitivity (data not shown). Furthermore, a dose‐dependent reduction in recruitment of CCR6⁺ T‐cells, but not CCR6⁻ T‐cells, to skin blister fluid was observed in subjects dosed with GSK3050002, with maximal effects observed at doses ≥5 mg kg⁻¹. This effect is consistent with the predicted pharmacodynamic effect of neutralization of CCL20 in the skin and would be inconsistent with a local increase in free active CCL20.

Simulations of target engagement after repeat dosing, based on CCL20 turnover estimated from GSK3050002–CCL20 complex data, were also performed to guide dose selection for chronic treatment in a patient population. Assuming the same GSK3050002 binding affinity to CCL20 target, doses of 5 mg kg⁻¹ (with a minimum of once every 2 weeks dosing frequency) and above are predicted to achieve at least an average 90% inhibition of free CCL20 over the dosing interval. These predictions are consistent with the maximal reduction in the percentage of CCR6⁺ cells in the total T‐cell pool observed in blister fluid on Day 15 post‐dose and in line with the observed saturation of GSK3050002–CCL20 complex and predicted free CCL20 inhibition at these doses. As GSK3050002 appeared to selectively inhibit recruitment of CCR6⁺ T‐cells, in particular Th17 cells, in the blisters, GSK3050002 may have potential therapeutic benefit in many chronic inflammatory conditions such as psoriasis, rheumatoid arthritis or inflammatory bowel disease, which are associated with Th17 cells 36. It should be noted, however, that Th17 cells in this study were identified as CD4⁺CXCR3⁻CCR6⁺ T‐cells and IL‐17 secretion or transcription factor expression was not investigated.

Further studies are required to evaluate whether GSK3050002 may inhibit other CCR6⁺ pathogenic cells and whether GSK3050002 may unwantedly inhibit non‐pathogenic cells such as Treg cells. Unfortunately, insufficient numbers of blister fluid cells were obtained in the clinical study to reliably evaluate recruitment of Treg cells in the blister. In vitro results show that Treg cells were capable of migrating towards CCL1, CXCL10, CCL20 and CCL22. In particular, Treg cells migrated in vitro towards CCL22, with almost all Treg cells expressing CCR4, and their response was about two‐ to threefold higher than that of Th17 cells. Similarly, in vivo, other chemokines, including CCL22, may mediate recruitment of Treg cells. This may be different from Th17 cells, as their recruitment was inhibited by GSK3050002 despite the ability of Th17 cells in vitro to respond to CCL22 and high CCL22 levels in the blisters. No effect of GSK3050002 was observed on monocyte recruitment in vivo, despite the ability GSK3050002 to inhibit in vitro monocyte migration. It is likely that other chemokines in the blister may play a more dominant role in monocyte recruitment compared to the supernatant used in vitro. A likely candidate for this could be CCL2, which was expressed at much higher levels in the blister compared to the in vitro system.

Conclusions

The use of the suction skin blister model has enabled the assessment of target engagement and pharmacodynamic effects in a target tissue at an early stage of clinical development. The observation of a dose‐dependent pharmacodynamic effect in the skin blister, consistent with the predicted mechanism of action of neutralization of bioactive CCL20, is encouraging. Maximal effects on CCR6⁺ T‐cell migration to the skin, as measured in blister fluid, were observed in dose cohorts where the accumulation of drug–CCL20 complex had reached maximum levels, supporting a local pharmacological neutralization of CCL20 activity. These results point to a relationship between pharmacokinetics, target engagement and pharmacodynamic effect and support further development of GSK3050002 in autoimmune and inflammatory diseases.

Competing Interests

Funding for this study was provided by GlaxoSmithKline and Morphotek, Inc. (NCT01984047). All listed authors were employees of GlaxoSmithKline (GSK) during the conduct of the study, hold GSK stock or stock options, and meet the criteria for authorship set forth by the International Committee for Medical Journal Editors.

The authors are grateful to the staff at GSK Clinical Unit in Cambridge for conducting the clinical study, to the subjects who participated in this study and to Jennifer Wright and Rajendra Singh for technical assistance.

Contributors

All listed authors met the criteria for authorship set forth by the International Committee for Medical Journal Editors. G.B., S.Z., K.H., A.C., S.B., D.R., L.F., V.N., D.F., M.H. and R.T. contributed to study design, review of analysed data and the manuscript. A.W. and J.O. contributed to data acquisition and analysis. Each author contributed important intellectual content during manuscript drafting. Editorial support (development of the first draft, assembling tables and figures, collating author comments, and referencing) was provided by GD Scientific & Medical Writing, LLC) and was funded by GSK.

Supporting information

Table S1 Immunophenotyping flow cytometry panels used in the study

Table S2 Chemokine and cytokine expression in suction blisters

Figure S1 GSK3050002 does not inhibit monocytes or granulocytes in a suction blister model. (A) CCR6⁺ monocytes, (B) total monocytes, (C) CCR6⁺ granulocytes and (D) total granulocytes. All data is represented as change from baseline, data from Day 3 blister is shown in grey and data from Day 15 blister in black squares. Data is represented as mean ± SEM

Figure S2 Migratory response of T‐cell subsets in vitro. Migration of Th17, Th1/17, Treg, Th1 and Th2 T‐cells towards (A) CCL20, (C) CXCL10, (D) CCL22 and (E) CCL1. (B) Inhibition of migration towards (B) CCL20 or (F) a chemokine mix produced following TNF‐α stimulation of HT‐29 cells by GSK3050002. (G) Chemokine receptor expression of T‐cell subsets, n/a = not applicable as gating based on these receptors. Data in A–F is shown as mean + SD and migration index of 1 equals migration observed to tissue culture medium. White bars represent chemokinesis control. In (B) lightest grey bar represents isotype control. Concentration of chemokine in (A) are 0.3, 30, 300 ng ml⁻¹ CCL20, (C) 3, 10, 30, 100 ng ml⁻¹ CXCL10, (D) 1, 3, 10, 30, 100 ng ml⁻¹ CCL22 and (E) 1, 3, 10, 30 ng ml⁻¹ CCL1. GSK3050002 was used in (B) at 0, 0.1, 10, 100 μg ml⁻¹ with 30 ng ml⁻¹ CCL20 and in (F) at 0, 0.156, 0.625, 2.5, 6.67 μg ml⁻¹. Cell types in (A–F) were identified as per clinical protocol. Cells in (G) were identified as follows: PBMC (FSC⁺SSC⁺CCR6⁺), B cells (CD19⁺CCR6⁺), monocytes (CD14⁺CCR6⁺), T‐cells (CD3⁺CCR6⁺), CCR6⁻ (FSC⁺SSC⁺CCR6⁻). Data in (G) is represented as mean ± SD

Figure S3 Chemokines produced by HT‐29 cells in response to TNF‐α stimulation. Grey cells represent missing values. Data of selected chemokines are plotted as separate graphs as mean + SD, n = 5,ULOQ; upper limit of quantification

Bouma, G. , Zamuner, S. , Hicks, K. , Want, A. , Oliveira, J. , Choudhury, A. , Brett, S. , Robertson, D. , Felton, L. , Norris, V. , Fernando, D. , Herdman, M. , and Tarzi, R. (2017) CCL20 neutralization by a monoclonal antibody in healthy subjects selectively inhibits recruitment of CCR6⁺ cells in an experimental suction blister. Br J Clin Pharmacol, 83: 1976–1990. doi: 10.1111/bcp.13286.