The effect of divalent metal cations on the αv integrin binding site is ligand and integrin specific

Biomedicine & Pharmacotherapy 110 (2019) 362–370

Contents lists available at ScienceDirect

Biomedicine & Pharmacotherapy journal homepage: www.elsevier.com/locate/biopha

The effect of divalent metal cations on the αv integrin binding site is ligand and integrin specific

T

Eleanor R. Hall, Robert J. Slack

⁎

Fibrosis Discovery Performance Unit, Respiratory TAU, GlaxoSmithKline, Gunnels Wood Road, Stevenage, Hertfordshire, UK

ARTICLE INFO

ABSTRACT

Keywords: αv integrins Divalent metal cations Radioligand binding Affinity Integrin selectivity

The binding of orthosteric ligands to integrins requires the presence of divalent metal cations bound to metal ionbinding sites located in the I domains of the integrin α and β subunits. In this study the influence of the type and concentration of divalent metal cation present was investigated on a single arginyl-glycinyl-aspartic acid (RGD) ligand across the αv integrin sub-family and single αv integrin (αvβ6) with different ligands. These relationships were determined using radioligand binding studies completed with [3H] ligands and purified αv integrin protein preparations. The binding of [3H]compound 1 to the RGD site on individual αv integrins demonstrated a unique profile in relation to the type and concentration of divalent metal cation present. The use of physiological concentrations of Mg2+ and Ca2+ in simulated lung fluid altered the αv integrin selectivity profile of [3H] compound 1 in terms of affinity and the level of receptor occupancy. In addition, different RGD ligands for the αvβ6 integrin behaved differently under the same divalent metal cation conditions. In conclusion, this study demonstrates the need to determine the individual relationship between RGD ligands and the integrins they may engage in vivo, especially when determining selectivity profiles for potential RGD-mimetic small molecule therapeutics, with organ and disease state also considered.

1. Introduction One of the key cell adhesion and signalling proteins in mammals are the integrins [1]. These are heterodimeric, transmembrane glycoprotein receptors that are made up of an α and β-subunit (18 and 8 variants respectively that can make up to 24 heterodimers [2]), bound in a noncovalent complex that forms the ligand binding site. Integrins have the ability to signal in both directions across the plasma membrane by either binding extracellular ligands or interacting with the cytoskeleton via their intracellular domains [3]. The crystal structure of the heterodimer comprising of one α- and one β-subunit have been solved for a number of integrins [4,5]. Both subunits are characterised by a large extracellular ectodomain (made up of several different domains), a single transmembrane helix and a short intracellular cytoplasmic domain. Integrins can exist in both activated (upright conformation) and inactivated (bent conformation) states [6] where they demonstrate a high and low affinity for ligands respectively. Activation can occur via many mechanisms that include extracellular ligand binding (referred to as ‘outside-in’ signalling),

intracellular β-integrin tail activation (referred to as ‘inside-out’ signalling) and divalent metal cation occupancy of the ligand-binding pocket. The bi-directional signalling displayed by integrins makes them an essential receptor family to enable human cells to respond to changes in the extracellular environment but also able to influence the extracellular environment. Information from the extracellular environment is communicated into cells via ligand binding to integrins resulting in changes in cell polarity, cytoskeletal structure, gene expression and cell survival and proliferation [7]. In addition, ligand binding shifts the integrin affinity state from low (inactive) to high (active). In the opposite direction, intracellular activators such as talin can bind to the cytoplasmic tail of the β-subunit evoking a conformational change that shifts the integrin to a high affinity state more readily able to bind extracellular ligands enabling cell migration, extra-cellular matrix assembly and remodelling [8]. In addition to these activation mechanisms, divalent metal cations (Ca2+, Mg2+ and Mn2+) are not only a pre-requisite of binding of integrin ligands [9] but can also influence the activation state [6]. Multiple binding sites for divalent metal cations within the βI domain of the integrin β-subunit have been

Abbreviations: MIDAS, metal ion-dependent adhesion site; ADIMIDAS, adjacent to metal ion-dependent adhesion site; LIMBS, ligand-induced metal ion-binding site; RGD, arginyl-glycinyl-aspartic acid; tLAP, truncated latency associated peptide; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate; DMSO, dimethylsulfoxide; NSB, non-specific binding; LS, liquid scintillation; ANOVA, one-way analysis of variance; Bmax, total number of binding sites ⁎ Corresponding author at: Fibrosis DPU, Respiratory TAU, GlaxoSmithKline, Gunnels Wood Road, Stevenage, Hertfordshire, SG1 2NY, UK. E-mail address: [email protected] (R.J. Slack). https://doi.org/10.1016/j.biopha.2018.11.130 Received 22 October 2018; Received in revised form 26 November 2018; Accepted 27 November 2018 0753-3322/ © 2018 GlaxoSmithKline Plc. Published by Elsevier Masson SAS. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/).

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

identified including the metal ion-dependent adhesion site (MIDAS), adjacent to metal ion-dependent adhesion site (ADMIDAS) and ligandinduced metal ion-binding site (LIMBS) [9]. MIDAS physiologically binds Mg2+ that is required for ligand binding whilst LIMBS functions as a positive regulatory site [10] and ADMIDAS as a negative regulatory site [11]. Mn2+ for the majority of integrins increases ligand binding affinity whilst Ca2+ can increase or decrease depending on the concentration and which of the LIMBS or ADMIDAS it engages [9]. However, little is still known on the physiological role of the integrin ion binding sites and is important to consider this variable when setting up systems to probe ligand-integrin interactions and endeavouring to standardise divalent metal cation type and concentration where possible. This becomes even more critical for ligands proposed to be developed as therapeutics when the key endogenous divalent metal cations and their abilities to alter ligand binding may not be known. It is therefore prudent to ensure ligand affinity is measured in a full spectrum of divalent metal cations and concentrations to provide best and worst-case binding profiles, with plasma/tissue levels of these divalent metal cations used as starting points. Depending on the target organ and the divalent metal cation concentrations present in the fluid the target and drug are exposed to, the binding profile is likely to differ but can be begun to be investigated using simulated body fluids [12]. In this study the effect of divalent metal cation type and concentration has been investigated using tool radioligands, with a focus on the αv sub-group of the arginyl-glycinyl-aspartic acid (RGD) family of integrins. The aim of this is to investigate the similarities and differences of the effect of divalent metal cation between integrins with the same ligand and between ligands at the same integrin, to give insight into the impact in a physiological in vivo setting. As part of this study the selective αvβ6 small molecule compound 1 [13] was further characterised against the αv integrins to determine its use as a tool for characterising the binding of unlabelled ligands to this receptor family.

peptide-1 GRRGDLATIHG (tLAP1), tLAP2 (YTSGDQKTIKS), and tLAP3 (HGRGDLGALKK) and the αvβ6 selective peptide NAVPNLRGDLQVLAQKVART (A20FMDV2 derived from the foot-and-mouth disease virus [19]) were synthesized by Cambridge Research Biochemicals (Cleveland, UK). A20FMDV2 and compound 1 were radiolabeled with [3H] by Quotient Bioresearch (Radiochemicals) Ltd. (Cardiff, UK) and had specific activities of 20 Ci/mmol ([3H]A20FMDV2), 11 Ci/mmol and 196 Ci/mmol (low and high specific activities for [3H]compound 1) (Fig. 1). Compound 2 was radiolabeled with [3H] by RC TRITEC Ltd. (Teufen, Switzerland) and had a specific activity of 16.1 Ci/mmol. All other chemicals and reagents were purchased from Sigma-Aldrich Co. Ltd. (Gillingham, UK) unless otherwise stated. 2.2. Recombinant soluble integrin proteins Purified soluble protein preparations (recombinantly derived from Chinese Hamster Ovary cells) for the human αvβ1, αvβ3, αvβ5, αvβ6 and αvβ8 integrin proteins profiled were all purchased from R&D Systems Inc. (Minneapolis, MN, USA). Protein preparations were reconstituted in sterile PBS to a stock concentration of 100 μg/ml (except for αvβ1 and αvβ5 (50 μg/ml)) and aliquots stored at −80 °C until use. 2.3. Radioligand binding studies 2.3.1. General protocols for radioligand binding assays All radioligand binding experiments were performed in 96-deep well plates (Greiner Bio-One, Firckenhausen, Germany) at 37 °C in binding buffer (25 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 100 mM NaCl, 2 mM MgCl2 (unless otherwise stated) and 1 mM 3[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS) at pH 7.4 (NaOH))) or simulated lung fluid (SLF) (0.5 mM MgCl2, 103 mM NaCl, 4 mM KCl, 0.9 mM Na2HPO4, 0.4 mM Na2SO4, 2.5 mM CaCl2.2H2O, 7 mM C2H3NaO3, 31 mM NaHCO3 and 0.3 mM C6H6O7.2H2O.3Na, 1 mM CHAPS at pH 7.4) [12]. Experiments were completed in a total volume of 0.5 ml consisting of 50 μl/well of either unlabeled compound at varying concentrations or vehicle (1% dimethylsulfoxide (DMSO)), 50 μl of [3H]compound 1 (11 Ci/mmol for αvβ3/αvβ5 and 196 Ci/mmol version for αvβ1/αvβ6/αvβ8) or [3H] compound 2 and 0.4 ml/well of purified integrin (1 nM of αvβ1 and αvβ8, 3 nM αvβ3 and αvβ5, and 25 pM αvβ6). Non-specific binding (NSB) was determined with 10 μM SC-68448. Specific binding was

2. Materials and methods 2.1. Materials Compound 1 [13], compound 2 [14,15], SC-68448 [16], SB-267268 [17] and cilengitide [18] (Fig. 1) were synthesized by the Fibrosis Discovery Performance Unit Medicinal Chemistry group at GSK Medicines Research Centre (Stevenage, UK). Truncated latency associated

Fig. 1. The chemical structures of the small molecule RGD-mimetics used in this study.

363

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

Fig. 2. Saturation binding of [3H]compound 1 to the human αv integrins in different buffer conditions. Specific binding of [3H]compound 1 was measured by incubation of increasing concentrations of radioligand with soluble human αvβ1 (A), αvβ3 (B), αvβ5 (C), αvβ6 (D) and αvβ8 (E) integrin protein in the presence of either vehicle (1% DMSO) or 10 μM SC-68448. Plates were then filtered after a 6 h or 24 h (αvβ6) incubation and the amount of radioligand bound measured by liquid scintillation spectroscopy. Saturation binding was completed in binding buffer (2 mM Mg2+) or simulated lung fluid (2.5 mM Ca2+ and 0.5 mM Mg2+). Specific binding was measured by subtracting the non-specific binding (10 μM SC-68448) from the total radioligand binding in the presence of vehicle (1% DMSO). Specific saturation binding data were fitted to a one affinity site model with Hill slope (see 2.4 Data Analysis section). Data shown are the mean ± SD of duplicate points and are representative of four individual experiments with comparable results. DPM, disintegrations per minute; SLF, simulated lung fluid.

measured by subtracting the NSB from the total radioligand binding in the presence of vehicle. Plates were incubated with gentle agitation for the time periods indicated and binding terminated by rapid vacuum filtration through a 48-well Brandel harvester (Brandel Inc. Gaithersburg, MD, USA) onto GF/B filter papers pre-soaked in 0.3% v/v poly-

ethylenimine. Samples were washed rapidly three times with ice cold dH2O and filters transferred into liquid scintillation (LS) vials containing 4 ml LS fluid (Ultima-Flo™ M, PerkinElmer LAS UK Ltd., Beaconsfield, UK). The amount of radioligand bound to integrin protein was measured by LS spectroscopy using a TriCarb 2900 TR LS counter (PerkinElmer LAS UK Ltd., Beaconsfield, UK). By the same method the concentration of total radioligand added to each well was calculated for data analysis. All radioligand binding assays were completed with a final DMSO concentration of 1%. 2.3.2. Radioligand characterization Association, dissociation and saturation binding studies were performed with [3H]compound 1 to determine radioligand binding kinetics at the αv integrins using human recombinant soluble protein. For saturation binding, αv integrin protein was incubated with increasing concentrations of radioligand for 6 h or 24 h (αvβ6) prior to filtration. For association binding, αv integrin protein were incubated with varying concentrations of [3H]compound 1 for varying times up to 4 h prior to filtration. For dissociation binding, αv integrin protein were pre-incubated for 30 min with a fixed concentration of radioligand (∼KD) and dissociation initiated by addition of 10 μM SC-68448 before incubation for varying times up to 72 h prior to filtration. To demonstrate the divalent cation dependency of αv integrin binding the specific binding window of [3H]compound 1 and [3H]compound 2 at a saturating concentration (∼10 x KD) was measured in the absence and presence of a range of concentrations of Ca2+, Mg2+ and Mn2+ following a 6 h incubation. 2.3.3. Determination of RGD ligand affinity For determining the affinity of RGD ligands at the αv integrins, competition binding studies were completed as previously described [15]. Briefly, integrin protein was incubated with a fixed concentration of [3H]compound 1 and increasing concentrations of unlabeled test ligand for 6 h or 24 h (αvβ6) prior to filtration. 2.4. Data analysis Analysis of in vitro radioligand binding experiments was completed using Prism 5.0 (GraphPad Software, San Diego, CA, USA). Association, dissociation, saturation and competition binding data were fitted to models previously described [13]. To allow the results between binding experiments to be combined, where appropriate, disintegrations per minute values were normalized to give the amount of radioligand bound per amount of purified protein (pmol/mg). All statistical analyses were completed using Prism 5.0 (GraphPad Software, San Diego, CA, USA) and differences of P < 0.05 considered to be statistically significant. Statistical significance between two data sets was tested using a Student’s unpaired t-test. One-way analysis of variance (ANOVA) was used for comparison of more than two datasets and, where significance was observed, an appropriate post-test completed. 364

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

Table 1 Saturation binding parameters for [3H]compound 1 with soluble human αv integrin protein in binding buffer and simulated lung fluid. Specific saturation binding data were fitted to a one affinity site model with Hill slope. Data shown are mean values ± SEM for four individual determinations. pKD, negative log10 of KD. Binding Buffer Integrin

pKD

αvβ1 αvβ3 αvβ5 αvβ6 αvβ8

8.65 7.49 7.61 10.6 8.59

± ± ± ± ±

Simulated Lung Fluid

0.04 0.29 0.06 0.02 0.11

Bmax (pmol/mg)

pKD

990 ± 168 3,636 ± 997 2,155 ± 244 3,481 ± 66 822 ± 77

8.50 7.01 7.85 10.4 9.24

Bmax (pmol/mg) ± ± ± ± ±

0.21 0.22 0.02 0.07 0.04

293 ± 136 3,222 ± 736 2,134 ± 434 3,521 ± 39 1,282 ± 316

Fig. 3. Association binding of [3H]compound 1 to the human αv integrins. In association studies specific binding was measured by incubation of [3H]compound 1 with soluble human αvβ1 (A), αvβ3 (B), αvβ5 (C), αvβ6 (D) and αvβ8 (E) integrin protein in the presence of either vehicle (1% DMSO) or 10 μM SC-68448. Association binding data were generated at the radioligand concentrations indicated (∼KD and 4 x KD) and globally fitted to the association kinetic model (see 2.4 Data Analysis section). Data shown are the mean ± SD of duplicate points and are representative of at least four individual experiments with comparable results. DPM, disintegrations per minute. Table 2 Kinetic binding parameters for [3H]compound 1 with soluble human αv integrin protein. Specific association binding data were globally fitted to the association kinetic model to determine kon values. Dissociation binding data were fitted to a one or two-phase (αvβ6) dissociation model to generate koff values (fast and slow for αvβ6). koff values were subsequently used to calculate dissociation half-life (t1/2) values using the equation t1/2 = 0.693/ koff. Data shown are mean values ± SEM for at least four individual determinations. Integrin αvβ1 αvβ3 αvβ5 αvβ6 αvβ8

kon (M−1. min−1) 5.33 1.55 1.06 3.56 1.41

± ± ± ± ±

koff 8

0.96 × 10 0.26 × 108 0.09 × 108 0.22 × 108 0.13 × 108

0.70 1.18 0.40 1.22 0.09

t1/2 ± ± ± ± ±

−1

0.05 min 0.19 min−1 0.03 min−1 0.28 h−1 (fast); 0.04 ± 0.01 h−1 (slow) 0.01 min−1

365

1.00 0.63 1.77 0.70 8.72

± ± ± ± ±

0.07 min 0.09 min 0.17 min 0.20 h (fast); 19.9 ± 4.42 h (slow) 1.67 min

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

using binding buffer (2 mM Mg2+) or SLF (0.5 mM Mg2+ and 2.5 mM Ca2+). Specific binding data from all saturation experiments were best fitted to a one affinity site model (Fig. 2). This analysis resulted in the saturation binding parameters shown in Table 1. Comparable affinity of [3H]compound 1 with the αv integrins was observed with that determined historically in binding buffer (in the presence of 2 mM Mg2+) with a high selectivity for αvβ6 displayed [13]. In SLF the binding of [3H]compound 1 to αvβ5 and αvβ6 was comparable to that in binding buffer suggesting the increased Ca2+ and decreased Mg2+ concentrations had limited effect. A reduced affinity and total number of binding sites (Bmax) over the concentrations of [3H]compound 1 tested was observed for both αvβ1 and αvβ3 in SLF compared with binding buffer, with the largest effect observed with αvβ1 where the Bmax was reduced by over 3-fold. An increase in affinity was observed at αvβ8 with [3H] compound 1 in SLF in addition to an increase in the Bmax. Overall the high selectivity of [3H]compound 1 for αvβ6 over the other αv integrins was improved in SLF compared with binding buffer. 3.2. Association and dissociation binding of [3H]compound 1 with and from the human αv integrins The association of [3H]compound 1 with the αv integrins was observed to follow a single phase (Fig. 3) with kon values comparable between integrins (Fig. 3; Table 2). [3H]compound 1 dissociation from the αv integrins followed a single-phase dissociation profile for all except αvβ6 where a two-phase dissociation model was best fitted (Fig. 4). Dissociation t1/2 values followed the order of affinity determined for [3H]compound 1 (low affinity = fast dissociation and high affinity = slow dissociation) with a fast dissociation observed from αvβ1, αvβ3 and αvβ5, a moderately slower dissociation from αvβ8, in comparison to a very slow dissociation from αvβ6, made up of a fast and slow phase (Fig. 4; Table 2). 3.3. Human αv integrin binding divalent cation dependency The regulation of ligand binding to integrins has been shown to be dependent on divalent cations due to the presence of allosteric cation binding sites within their protein structure [2,13,20,21]. The sensitivity of [3H]compound 1 binding to the αv integrins was measured by titrating divalent metal cations Mg2+, Mn2+ and Ca2+ against a fixed saturating concentration of [3H]compound 1 (Fig. 5). Mn2+ was shown to be the most potent activator of αvβ1, reaching maximum binding at ∼1.5 μM, followed by Mg2+, that demonstrated a comparable Bmax. Ca2+ potentiated [3H]compound 1 αvβ1 binding at lower concentrations, however at concentrations of > 150 μM binding was then inhibited producing a bell-shaped response (Fig. 5A). For αvβ3, Mn2+ was shown to be the most potent activator of [3H]compound 1 binding with Bmax achieved at ∼25 μM. Mg2+ was the next most potent activator of αvβ3, however it did not reach the same Bmax as Mn2+. In addition, Ca2+ could also only support partial binding with some evidence of reduced binding at the top concentration tested (10 mM) (Fig. 5B). Mn2+ was shown to be the most potent activator of αvβ5, reaching maximum binding at ∼25 μM, followed by Ca2+ that demonstrated a reduced Bmax in comparison. The least potent was Mg2+, however this demonstrated a comparable Bmax to that observed with Mn2+ (Fig. 5C). For αvβ8, Mn2+ was shown to be the most potent activator of [3H]compound 1 binding with Bmax achieved at ∼25 μM, followed by Ca2+ and then Mg2+. All divalent cations were able to support maximal binding of [3H]compound 1 to αvβ8 (Fig. 5E). Binding of [3H]compound 1 to αvβ6 required the presence of divalent cations with Ca2+, Mg2+ and Mn2+ able to support the binding of this radioligand (Fig. 5D). However, the concentration required to enable binding of [3H]compound 1 with αvβ6 was different between divalent cations with Mn2+ able to support binding at much lower concentrations compared with Mg2+ and Ca2+. In addition, Ca2+ was unable to produce maximal binding, up to a concentration of 10 mM,

Fig. 4. Dissociation binding of [3H]compound 1 from the human αv integrins. The [3H]compound 1 dissociation profile from αvβ3 and αvβ5 (A), αvβ1 and αvβ8 (B) and αvβ6 (C) are shown. For dissociation studies soluble human integrin protein were pre-incubated for 1 h with a fixed concentration of radioligand before dissociation was initiated by addition of 10 μM SC-68448. Plates were then filtered after the time indicated and the amount of radioligand bound was measured by liquid scintillation spectroscopy. Dissociation binding data were fitted to a one or two-phase (extra-sum-of-square F test, P < 0.05 for αvβ6) dissociation model to generate koff values (fast and slow for αvβ6). Data shown are the mean ± SD of duplicate points and are representative of at least four individual experiments with comparable results. DPM, disintegrations per minute.

Unless otherwise indicated data shown are either mean ± SD or, where three or more data points/individual experiments have been completed, mean ± SEM. 3. Results 3.1. Radioligand saturation binding of [3H]compound 1 with the human αv integrins [3H]compound 1 saturation binding studies were carried out to determine binding affinity against the αv integrins. This was completed in the presence of different concentrations of divalent metal cations 366

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

Fig. 5. Saturation binding of [3H]compound 1 to the human αv integrins in the presence of different types and concentrations of divalent cations. Specific binding was measured by incubation of [3H]compound 1 with soluble human αvβ1 (A), αvβ3 (B), αvβ5 (C), αvβ6 (D) and αvβ8 (E) integrin protein in the presence of either vehicle (1% DMSO) or 10 μM SC-68448 at different concentrations of Ca2+, Mg2+ or Mn2+ with experiments carried out using 10 x KD concentrations of radioligand. Plates were then filtered after a 6 h incubation and the amount of radioligand bound measured by liquid scintillation spectroscopy. Specific binding was measured by subtracting the non-specific binding (10 μM SC-68448) from the total radioligand binding in the presence of vehicle (1% DMSO) and DPM values converted to pmol/ mg. Data shown are the mean ± SEM of four individual experiments carried out in quadruplicate.

compared with Mn2+ and Mg2+. Binding of [3H]compound 2 and [3H]A20FMDV2 to αvβ6 also required the presence of divalent cations with Ca2+, Mg2+ and Mn2+ able to support the binding of [3H]compound 2 and [3H]A20FMDV2 (Fig. 6A and B). The propensity to support the binding of [3H]A20FMDV2 is comparable between all the cations tested and all enabled maximal binding of [3H]A20FMDV2 to αvβ6. However, Mn2+ could support this at much lower concentrations compared with Mg2+ and Ca2+ with this trend also observed with [3H] compound 2. However, with [3H]compound 2 Ca2+ was unable to produce maximal binding, up to a concentration of 10 mM, compared with Mn2+ and Mg2+. The divalent cation binding profile of [3H] compound 1 was more comparable to that observed with [3H]A20FMDV2 (Figs. 5D and 6B). However, the partial ability to support binding in the presence of Ca2+ was more comparable to [3H] compound 2, although the level to which binding was enabled was greater for [3H]compound 1 (Figs. 5D and 6A).

against [3H]compound 1 following a 6 h or 24 h (αvβ6) incubation period (Fig. 7). The pKI values determined for a range of unlabelled integrin ligands are summarised in Table 3. The pan-αv integrin RGDmimetic SC-68448, the cyclic peptide cilengitide and tLAP1 have been used historically to control competition binding studies against this receptor class [14] and the affinity estimates in this study were in good agreement with previous data. The lack of affinity demonstrated for tLAP2 against all the αv integrins, due to the lack of the RGD sequence, also acted as a negative control. A more accurate determination of the affinity of tLAP3 has been completed in this study and it displayed a comparable high selectivity for αvβ6 with that of tLAP1, but with a higher affinity for all the αv integrins it was shown to bind. The high affinity and selectivity of SB-267268 for the αvβ3/5 integrins has been confirmed [17] apart from αvβ1 where a moderate affinity was also demonstrated for this RGD-mimetic small molecule (Fig. 7). 4. Discussion

3.4. Competition binding studies with [3H]compound 1 at the human αv integrins

Divalent metal cations have been shown to influence the ligand binding site on integrins via the MIDAS, LIMBS and ADIMIDAS sites and inducing conformational changes that can result in the transition from an inactive to an active form and vice-versa [6,9–11]. One area that has

To determine the affinity of unlabelled integrin ligands at the αv integrins, competition displacement binding curves were measured

367

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

of a range of concentrations of Mn2+, Mg2+ and Ca2+. The traditional approach to measuring the affinity of ligands for integrins is to use systems containing 2 mM Mg2+ as this is its human plasma concentration and most integrins are at resting states under these physiological conditions [23]. However, saturation binding studies investigating [3H]compound 1 in binding buffer compared with SLF demonstrated the importance of using physiologically relevant divalent metal cation concentrations and the differing effects when comparing affinity and maximal receptor binding with the αv integrins. For this molecule an overall improved selectivity profile was determined and in the case of αvβ1 a marked reduction in Bmax was observed suggesting an even greater selectivity in terms of the receptor occupancy that would be observed in vivo. There is also evidence that concentrations of divalent metal cations change during different physiological processes like wound healing [24]. It is likely that under disease states there may also be fluctuations in divalent metal cations that make the assessment of drug binding at a range of conditions essential to provide an understanding of a best and worst-case scenario. When the focus shifted to investigate different ligands at the same integrin, αvβ6 was chosen as a case study. The profile observed for the small molecule ligand [3H]compound 1 was comparable to that observed with the peptide ligand [3H]A20FMDV2, in terms of the potency of the divalent metal cations effect. However, the partial ability to support binding in the presence of Ca2+ was more comparable to another small molecule [3H]compound 2, although the level to which binding was enabled was greater for [3H]compound 1. This suggests that different αvβ6 ligands can demonstrate different binding profiles when in the presence of the same type and concentration of divalent metal cation. This is not only the case between distinct types of ligand i.e. peptide versus small molecule, but also within i.e. small molecule versus small molecule. This highlights that each structurally dissimilar αvβ6 ligand, and likely other ligands designed to bind to the other αv integrins, would need to be assessed on an individual basis and that this may be important when investigating more physiologically relevant divalent metal cation conditions that would depend on the location of the target e.g. the disease organ of interest in the case of a therapeutic. One limitation of this study, as not measured directly, is the impact of divalent metal cations on the kinetics of the αv RGD binding site interaction. However, from the affinity changes already observed some of these effects could be predicted. Although some changes in association rates will occur, generally in this target class increased ligand affinity is correlated with slower dissociation and therefore more straightforward to predict [14,25]. Therefore, under physiological lung conditions for compound 1 you would anticipate a comparable dissociation rate for αvβ1, αvβ5 and αvβ6, but a potential increase in the rate for αvβ3 and decrease in the rate for αvβ8 to that measured in binding buffer. Therefore, in addition to affinity changes and when put in terms of drug residency times, these shifts in dissociation rates would have the potential to alter the selectivity profile of an RGD-mimetic ligand by also changing the drugs duration of αv integrin engagement. Although these changes are likely minimal for compound 1 as it already demonstrates a high selectivity for αvβ6, for other chemically distinct

Fig. 6. Saturation binding of [3H]compound 2 and [3H]A20FMDV2 to the human αvβ6 integrin in the presence of different types and concentrations of divalent cations. Specific binding was measured by incubation of [3H]compound 2 (A) or [3H]A20FMDV2 (B) with soluble human αvβ6 integrin protein in the presence of either vehicle (1% DMSO) or 10 μM SC-68448 at different concentrations of Ca2+, Mg2+ or Mn2+ with experiments carried out using 10 x KD concentrations of radioligand. Plates were then filtered after a 6 h incubation and the amount of radioligand bound to αvβ6 measured by liquid scintillation spectroscopy. Specific binding was measured by subtracting the nonspecific binding (10 μM SC-68448) from the total radioligand binding in the presence of vehicle (1% DMSO) and DPM values converted to pmol/mg. Data shown are the mean ± SEM of four individual experiments carried out in quadruplicate.

not been fully explored is the effect of cation type and concentration on the αv integrin sub-family RGD binding site and how this compares for the same ligand at different integrins or different ligands at the same integrin. To gather further information regarding the relationship between ligands, divalent metal cations and concentrations thereof, in this study saturating concentrations of radioligands were tested at increasing concentrations of either Ca2+, Mg2+ or Mn2+. This was completed by evaluating a radiolabelled version of compound 1, a selective αvβ6 small molecule currently in development for inhaled delivery for idiopathic pulmonary fibrosis [22], that was profiled under a range of conditions. As such, in the context of delivering a drug into the lung to target a protein on the baso-lateral surface of the airway epithelial layer, SLF was investigated in addition to investigating the effect

Table 3 The pKI values at the human αv integrins from competition binding between [3H]compound 1 and a range of integrin ligands. Competition binding data were fitted using non-linear regression analysis (four-parameter logistic equation with variable slope [26]) to generate IC50 values. IC50 values generated were converted to KI values using the Cheng-Prusoff equation [27]. Data shown are mean values ± SEM for four individual determinations. pKI, negative log10 of KI. Integrin Ligand

αvβ1

SC-68448 tLAP1 tLAP2 tLAP3 SB267268 Cilengitide

7.64 ± 5.47 ± < 5.23 6.01 ± 7.87 ± 6.64 ±

αvβ3 0.06 0.04 0.11 0.18 0.10

8.93 ± 5.67 ± < 5.40 6.56 ± 9.58 ± 8.16 ±

0.05 0.04 0.02 0.03 0.01

368

αvβ5

αvβ6

8.04 ± 0.06 < 5.37 < 5.37 < 5.37 9.59 ± 0.03 7.68 ± 0.07

8.56 ± 8.34 ± < 5.48 8.83 ± 5.63 ± < 5.48

αvβ8 0.05 0.16 0.12 0.04

7.08 ± 0.14 < 5.22 < 5.22 5.98 ± 0.07 < 5.22 < 5.22

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack

Fig. 7. Competition displacement binding curves for integrin ligands against [3H]compound 1 and the human αv integrins. Full competition binding curves were generated by incubating unlabelled integrin ligand at a range of concentrations with soluble human αv integrin protein and [3H]compound 1. Plates were then filtered after a 6 h or 24 h (αvβ6) incubation and the amount of radioligand bound was measured by liquid scintillation spectroscopy. Total and non-specific binding values were measured in the presence of vehicle (1% DMSO) and 10 μM SC-68448 respectively and were used to calculate the % inhibition of radioligand bound to each αv integrin. Competition binding displacement curve data are the mean ± SEM of four individual experiments carried out in singlicate.

RGD-mimetics, as exhibited for compound 2, there is the potential for different divalent metal cation profiles and therefore selectivity profiles in relation to affinity and receptor kinetics. In addition, even with relatively low affinity for the αv integrins beyond αvβ6, compound 1, has proven a useful tool to investigate the effects of divalent metal cations on this receptor family and to profile ligand binding. In the latter case, a close correlation with affinities measured with other tool radioligands has been demonstrated.

References [1] R.O. Hynes, Integrins: a family of cell surface receptors, Cell 48 (1987) 549–554. [2] R.O. Hynes, Integrins: bidirectional, allosteric signaling machines, Cell 110 (2002) 673–687. [3] R.J. Faull, M.K. Ginsberg, Inside-out signaling through integrins, J. Am. Soc. Nephrol. 7 (1996) 1091–1097. [4] T.L. Lau, C. Kim, M.H. Ginsberg, T.S. Ulmer, The structure of the integrin αIIbβ3 transmembrane complex explains integrin transmembrane signalling, EMBO J. 28 (2009) 1351–1361. [5] X. Dong, L.Z. Mi, J. Zhu, W. Wang, P. Hu, B.H. Luo, T.A. Springer, αvβ3 integrin crystal structures and their functional implications, Biochemistry 51 (2012) 8814–8828. [6] S. Tiwari, J.A. Askari, M.J. Humphries, N.J. Bulleid, Divalent cations regulate the folding and activation status of integrins during their intracellular trafficking, J. Cell. Sci. 124 (2011) 1672–1680. [7] S.J. Shattil, C. Kim, M.H. Ginsberg, The final steps of integrin activation: the end game, Nat. Rev. Mol. Cell Biol. 11 (2010) 288–300. [8] D.A. Calderwood, I.D. Campbell, D.R. Critchley, Talins and kindlins: partners in integrinmediated adhesion, Nat. Rev. Mol. Cell Biol. 14 (2013) 503–517. [9] B.H. Luo, C.V. Carman, T.A. Springer, Structural basis of integrin regulation and signaling, Annu. Rev. Immunol. 25 (2007) 619–647. [10] J. Chen, A. Salas, T.A. Springer, Bistable regulation of integrin adhesiveness by a bipolar metal ion cluster, Nat. Struct. Biol. 10 (2003) 995–1001. [11] A.P. Mould, S.J. Barton, J.A. Askari, S.E. Craig, M.J. Humphries, Role of ADMIDAS cationbinding site in ligand recognition by integrin alpha 5 beta 1, J. Biol. Chem. 278 (2003) 51622–51629. [12] M.R.C. Marques, R. Loebenberg, M. Almukainzi, Simulated biological fluids with possible application in dissolution testing, Dissolut. Technol. 18 (2011) 15–29. [13] E.R. Hall, L.I. Bibby, R.J. Slack, Characterisation of a novel, high affinity and selective αvβ6 integrin RGD-mimetic radioligand, Biochem. Pharmacol. 117 (2016) 88–96. [14] J.E. Rowedder, S.B. Ludbrook, R.J. Slack, Determining the true selectivity profile of αv integrin ligands using radioligand binding: applying an old solution to a new problem, SLAS Discov. 22 (2017) 962–973. [15] A.L. Wilkinson, J.W. Barrett, R.J. Slack, Pharmacological characterization of a tool αvβ1 integrin small molecule RGD-mimetic inhibitor, Eur. J. Pharmacol. 842 (2019) 239–247. [16] C.P. Carron, D.M. Meyer, J.A. Pegg, V.W. Engleman, M.A. Nickols, S.L. Settle, W.F. Westlin, P.G. Ruminski, G.A. Nickols, A peptidomimetic antagonist of the integrin αvβ3 inhibits leydig cell tumor growth and the development of hypercalcemia of malignancy, Cancer Res. 58 (1998) 1930–1935. [17] W.H. Miller, R.M. Keenan, R.N. Willette, M.W. Lark, Identification and in vivo efficacy of small-molecule antagonists of integrin αvβ3 (the vitronectin receptor), Drug Discov. Today 5 (2000) 397–408. [18] S.L. Goodman, G. Hölzemann, G.A.G. Sulyok, H. Kessler, Nanomolar small molecule inhibitors for αvβ6, αvβ5, and αvβ3 integrins, J. Med. Chem. 45 (2002) 1045–1051. [19] D. Logan, R. Abu-Ghazaleh, W. Blakemore, S. Curry, T. Jackson, A. King, S. Lea, R. Lewis, J. Newman, N. Parry, D. Rowlands, D. Stuart, E. Fry, Structure of a major immunogenic site on foot-and-mouth disease virus, Nature 362 (1993) 566–568. [20] E.F. Plow, T.A. Haas, L. Zhang, J. Loftus, J.W. Smith, Ligand binding to integrins, J. Biol. Chem. 275 (2000) 21785–21788. [21] R.J. Slack, M. Hafeji, R. Rogers, S.B. Ludbrook, J.F. Marshall, D. Flint, S. Pyne, J.C. Denyer, Pharmacological characterization of the αvβ6 integrin binding and

5. Conclusion The effect of divalent metal cations on the binding of ligands to the RGD site on αv integrins varies between integrins with the same ligand and between ligands with the same integrin. This highlights the requirement to characterise the individual relationship between RGDmimetic inhibitors, the integrins they will engage in vivo and the predicted physiological divalent metal cations at target site and in disease state. In addition, this should also be taken into consideration when setting up screening assays to identify and optimise drug candidates against this target class. Statement of conflicts of interest All authors are currently or have been employees of GlaxoSmithKline but no conflict of interest beyond this is declared. Declaration of interest None. Acknowledgements The authors would like to acknowledge the Fibrosis DPU Medicinal Chemistry team at GlaxoSmithKline for the synthesis of compound 1, compound 2, SC-68448, SB-267268 and cilengitide. The authors would also like to acknowledge Prof. John Marshall and Cancer Research Technology for their work in developing A20FMDV2. This work was fully funded by GlaxoSmithKline. 369

Biomedicine & Pharmacotherapy 110 (2019) 362–370

E.R. Hall, R.J. Slack internalization kinetics of the foot-and-mouth disease virus derived peptide A20FMDV2, Pharmacology 97 (2016) 114–125. [22] P.A. Procopiou, N.A. Anderson, J. Barrett, T.N. Barrett, M.H.J. Crawford, B.J. Fallon, A.P. Hancock, J. Le, S. Lemma, R.P. Marshall, J. Morrell, J.M. Pritchard, J.E. Rowedder, P. Saklatvala, R.J. Slack, S.L. Sollis, C.J. Suckling, L.R. Thorp, G. Vitulli, S.J.F. Macdonald, Discovery of (S)-3-(3-(3,5-dimethyl-1H-pyrazol-1-yl)phenyl)-4-((R)-3-(2-(5,6,7,8-tetrahydro-1,8-naphthyridin-2-yl)ethyl)pyrrolidin-1-yl)butanoic acid, a nonpeptidic αvβ6 integrin inhibitor for the inhaled treatment of idiopathic pulmonary fibrosis, J. Med. Chem. 61 (2018) 8417–8443. [23] T. Vorup-Jensen, T.T. Waldron, N. Astrof, M. Shimaoka, T.A. Springer, The connection between metal ion affinity and ligand affinity in integrin I domains, Biochim. Biophys. Acta 1774 (2017) 1148–1155.

[24] J.J. Grzesiak, M.D. Pierschbacher, Shifts in the concentrations of magnesium and calcium in early porcine and rat wound fluids activate the cell migratory response, J. Clin. Invest. 95 (1995) 227–233. [25] X. Dong, B. Zhaoa, F. Linb, C. Lua, B.N. Roger, T.A. Springer, High integrin αvβ6 affinity reached by hybrid domain deletion slows ligand-binding on-rate, Proc. Natl. Acad. Sci. U. S. A. 115 (2018) E1429–E1436. [26] A.V. Hill, The mode of action of nicotine and curari, determined by the form of the contraction curve and the method of temperature coefficients, J. Physiol. 39 (1909) 361–373. [27] Y. Cheng, W.H. Prusoff, Relationship between the inhibition constant (KI) and the concentration of inhibitor which causes 50 per cent inhibition (I50) of an enzymatic reaction, Biochem. Pharmacol. 22 (1973) 3099–3108.

370