A functional radioreceptor assay of alpha-V-beta-3 (αvβ3) inhibitors in plasma: Application as an ex vivo pharmacodynamic model

J. Biochem. Biophys. Methods 65 (2005) 107 – 120 www.elsevier.com/locate/jbbm

A functional radioreceptor assay of alpha-V-beta-3 (avh3) inhibitors in plasma: Application as an ex vivo pharmacodynamic model Margery A. Chaikin a, Ju´an Jose´ Marugan a, Gerald W. De Vries b, Peter Baciu b, Jeffrey Edelman b, Ming Ni b, Bruce E. Tomczuk a, Wenxi Pan a, Zihong Guo a, Beth Anaclerio a, Kristi Leonard a, Stephen H. Eisennagel a, Christopher J. Molloy a, Carl L. Manthey a,* a

Johnson & Johnson Pharmaceutical Research and Development, 665 Stockton Drive, Exton, PA 19341, United States b Allergan Inc. 2525 DuPont Drive, Irvine, CA 92612-1599, United States Received 20 April 2005; accepted 26 October 2005

Abstract Development of avh3-integrin inhibitors has been hampered by a lack of pharmacodynamic endpoints to identify doses that inhibit avh3 in vivo. To address this need, we developed an avh3 radioreceptor assay (RRA) that could be performed in 100% plasma. The RRA was based on 125I-echistatin binding to plateimmobilized avh3. Small molecule avh3 inhibitors efficiently competed echistatin binding to avh3 when the assay was carried out in buffer. However, when carried out in 100% plasma, the RRA revealed a 45 to N 3000-fold loss in compound potencies. The losses in potency reflected, in part, the high plasma protein binding by the compounds examined. The RRA was adapted as an ex vivo pharmacodynamic model. Echistatin binding was measured in the presence of plasma harvested at timed intervals from rats dosed with select compounds. Using this pharmacodynamic model, compound and dose selection was optimized for further testing in models of corneal angiogenesis. Moderate anti-angiogenic activity was achieved when rats were dosed sufficient to achieve sustained (N 50%) plasma inhibition through the trough interval. Thus, the RRA provided a simple technique to rank order compound potency in plasma, and could find general

Abbreviations: AUC, area under the curve; avh3, alpha-V-beta-3; bFGF, basic fibroblast growth factor; RGD, arginine-glycine-aspartic acid; RRA, radioreceptor assay. * Corresponding author. Tel.: +1 610 458 5264x6559; fax: +1 610 458 8259. E-mail address: [email protected] (C.L. Manthey). 0165-022X/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jbbm.2005.10.004

108

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

use as an ex vivo pharmacodynamic assay to select compounds and doses for preclinical and clinical proofof-principle studies. D 2005 Elsevier B.V. All rights reserved. Keywords: Integrin; Echistatin; Pharmacodynamic; Angiogenesis; Plasma; Binding

1. Introduction Integrins are heterodimeric transmembrane proteins consisting of one a- and one h-subunit. As a class, integrins mediate cellular binding to matrix proteins and to other cells. Expression of the avh3 integrin is low in most healthy tissues, but expression is markedly increased on new capillary sprouts during angiogenesis in some tissues [1–3]. Antibodies that block avh3 function were shown to inhibit angiogenesis induced by tumors grown in chick chorioallantoic membranes [4]. Later studies demonstrated that inhibitory antibodies and peptides could block retinal neovascularization [2,5] and tumor angiogenesis and growth [1,6] in animal models. These data motivated many laboratories to seek non-peptidic inhibitors of avh3 that could be dosed by the oral route for the treatment of cancer and other diseases where angiogenesis contributes to pathogenesis (reviewed in [7]). In addition to vascular expression, avh3 integrin is expressed on osteoclasts and is required for optimal bone resorption [3]. For this reason, avh3 inhibitors have been sought additionally for the treatment of osteoporosis [8–11]. The function of avh3 is mediated, in part, by interaction with arginine-glycine-aspartic acid (RGD) exposed on the surface of several extracellular matrix proteins including vitronectin [12]. Small synthetic cyclic pentameric peptides such as cRGDfV compete the binding of avh3 to vitronectin [13]. Also known to block avh3-vitronectin interactions are several peptides found in snake venom such as echistatin [14]. These too contain an RGD sequence that is critical for binding to avh3 [15]. Ku et al. [16] provided an early example of the use of computational chemistry to design nonpeptide bRGD-mimeticQ integrin inhibitors. Subsequently, numerous research teams designed many additional novel compounds based on the RGD-mimetic concept (reviewed in [7]). Nonetheless, relative to the large investment of effort, there have been only a few reports on nonpeptide compounds with in vivo anti-angiogenic activity [17–19], and, to the best of our knowledge, no non-peptidic compounds have entered clinical trials for cancer. High plasma protein binding is a feature of several known non-peptide RGD-mimetics, and such binding may be a general feature that attenuates the in vivo activity of this class of agents [9,20]. Even so, more success has met efforts to develop avh3 inhibitors for the prevention of bone resorption as illustrated by the recent report that the avh3 antagonist, L-000845704, increased bone mineral density in postmenopausal women [11]. To interpret positive or negative pharmacology results, it is important to establish that the target of interest has indeed been blocked. Pharmacodynamic assays are used for this purpose and seek to measure directly the in situ activity of the target protein or some acute physiological response that can be attributed to target inhibition. For example, inhibition of enzyme targets may reduce in vivo conversion of substrate to product. Or, when the target is a receptor, measurement of an acute physiologic response to ligand may be measured. Development of pharmacodynamic assays for avh3 inhibitors presented a challenge; avh3 is not an enzyme, and although it has binding partners, (e.g., vitronectin) these bligandsQ are not known to induce acute physiologic responses. Thus, interpretation of the in vivo pharmacology of avh3 inhibitors has been hampered by a lack of data on target inhibition in vivo.

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

109

Pathogenic angiogenesis has been associated with vascular permeability [21], and, during angiogenesis, avh3 likely functions in an environment accessible to plasma proteins. To assess the impact of plasma proteins on compound activity we developed a radioreceptor assay (RRA) that measured the binding of avh3 to 125I-echistatin in 100% plasma. The RRA proved useful to rank order compound potency in plasma and as a quick test of species-specific plasma effects. Further, the RRA was adapted to serve as an ex vivo pharmacodynamic model wherein compound activity could be estimated in plasma harvested from dosed animals, and we have correlated this activity with moderate efficacy in two models of angiogenesis. 2. Materials and methods 2.1. Reagents Compounds 1–4 were synthesized as described [22,23]. The following materials were purchased: citrated animal plasmas (Pel-Freez Biologicals), 125I-echistatin ~2000 Ci/mMol (Amersham Pharmacia Biotech), unlabeled echistatin (Peninsula Laboratories), placental avh3 (Chemicon), cyclo(Arg-Gly-Asp-DPhe-Val) (Peptide Institute Inc.), Fraction V bovine serum albumin (Sigma Chemicals) and Isoplate HB 96 well plates (Perkin Elmer). 2.2. RRA protocol avh3 (50 ng/well), in assay buffer (20 mM Tris pH 7.5 containing 150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2 and 1 mM MnCl2), was used to coat Wallac Isoplate HB 96 well plates overnight at 4 8C. Coated plates were then blocked 2 h at room temperature in assay buffer containing 1% BSA. To the blocked plates, 125I-echistatin was added at a final concentration of 0.25 nM with or without compound in assay buffer or in 100% platelet poor plasma in a total reaction volume of 100 Al. Test conditions were evaluated in triplicate wells. For saturation and competition binding assays, the reactions were incubated for 2 h at room temperature. For on-rate experiments, reactions were incubated for the times noted. Following incubation, the plates were washed twice with assay buffer. Microscintillant was added (100 Al/well) and bound 125I-echistatin was counted using a Wallac 1450 Microbeta Plus Liquid Scintillation Counter. For all assays, non-specific binding was determined by addition of z 200 molar excess unlabeled echistatin. Data analysis was performed using GraphPad PrismR or Microsoft Excel. 2.3. Ex vivo pharmacodynamic studies Groups of three male Sprague–Dawley rats were dosed subcutaneously with 50 mg/kg of compound 1 or with 50 or 120 mg/kg of compound 2 in 40% 2-hydroxypropyl-h-cyclodextrin. Dose volume was 4 ml/ kg. Blood was harvested from each rat just prior to dosing (T = 0) and at timed intervals after dosing as indicated. Blood was drawn through the tail vein into syringes coated with sodium citrate, and platelet-poor plasma was prepared from the fresh blood by centrifugation (10,000 g  5 min). Undiluted plasma was adjusted to contain 0.25 nM 125I-echistatin, and 100 Al of plasma were added to microtiter plate wells containing immobilized avh3. Binding of 125I-echistatin was determined in duplicate by RRA as described above. 2.4. Plasma protein binding Rat plasma or PBS was adjusted to contain 10 AM of test compound. Plasma samples were submitted to centrifugation (2 min  10,000 g) on an Amicon Centrifree Micropartition unit with a 30 K MW cutoff to separate bound (retained) and free (filtered) compound. Fifty microliters of filtrate were extracted with acetonitrile and analyzed on an API mass spectrometer. Extracted samples were analyzed using LC-MS/

110

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

MS. The LC-MS/MS system consisted of an API-2000 triple quadrupole mass spectrometer using the TurboIonspray source, a Shimadzu LC-10ADvp pumping system, a CTC/Leap HTS PAL autosampler, and a C8 LC column (Princeton Chromatography, 5u, 50  3.0 mm). The chromatography system consisted of an isocratic mobile phase of 70% methanol, 30% water, and 0.1% formic acid at a flow rate of 0.4 ml/min. Data was collected and MRM analysis was performed using Analyst 1.1 from Applied Biosystems. Percent bound was calculated as [1 (AUCfiltered plasma / AUCunfiltered buffer)]  100. 2.5. Rat basic FGF (bFGF) corneal angiogenesis model Three-month-old Sprague–Dawley rats, weighing approximately 250 g, were purchased from Charles River Laboratories (Wilmington, MA). The rats were anesthetized with intramuscular injection of ketamine (50 mg/kg) and xylazine (5 mg/kg) for surgery and eye examination. Corneal neovascularization was induced by an intra-stroma implantation of an Elvax pellet (DuPont) containing 300 ng bFGF. One eye of each rat was used. Topical 10% Cyclogel was instilled for mydriasis during the eye examination in order to see neovascularization clearly on the cornea without the background of normal iris vasculature. A central corneal incision about 0.1 mm deep and 2–3 mm wide was made at the apex of the rat cornea. Radial midstromal tunnels toward the corneoscleral limbus were formed with a cyclodialysis spatula. Implants were inserted into these preformed tunnels and positioned at the 12 o’clock or 6 o’clock meridians 1 mm from the limbus. Topical antibiotic ointment (bacitractin/neomycin/polymyxin) was applied once after surgery. The rats received the angiogenic challenge on day 0. Compounds formulated in 40% 2hydroxypropyl-h-cyclodextrin were administered s.c. twice daily for 9 days, from day 0 to day 8. Compound treatment groups were compared directly to control groups receiving vehicle alone. The growth of corneal neovascularization and efficacy of test compounds were monitored by photography. The area of corneal neovascularization was measured and analyzed using ImageNet 2000 (Topcon). Data was analyzed by Student’s t-test. 2.6. Rat cautery-induced corneal angiogenesis model A silver/potassium nitrate cautery applicator was applied for 2 s to the center of each cornea of anesthetized female Sprague–Dawley rats (5/group). Rats were dosed subcutaneously twice daily with either vehicle alone (40% 2-hydroxypropyl-h-cyclodextrin) or 120 mg/kg compound 2 in vehicle on day 0 through the morning of day 4. On day 4, rats were euthanized and the vasculature perfused with FITCdextran. Eyes were enucleated, and the isolated corneas were flat-mounted for analysis. Corneal neovascularization was measured using fluorescence microscopy and ImagePro software.

3. Results We reported previously [22,23] the identification of indol-1-yl propionic acid inhibitors of avh3-integrin. Compounds 1–4 (Table 1) were reported to exhibit IC50 values of 0.25–0.7 nM in an ELISA that measured avh3 binding to biotinylated-vitronectin, but were inactive at 10 AM against the closely related platelet integrin, aIIbh3 [23]. avh3 inhibitors were sought as antiangiogenic agents. Because pathogenic angiogenesis has been associated with vascular permeability [21], avh3 may function in an environment accessible to plasma proteins. For this reason we sought to determine compound activity in whole plasma. The ELISA-based avh3 assay did not function in the presence of plasma. This was due, presumably, to the low sensitivity of the colorimetric assay, and to occupancy of avh3 by proteins endogenous to plasma. We therefore sought a radioligand to exploit the sensitivity of radioreceptor assays. Echistatin is a snake venom protein that binds avh3 with sub-nanomolar affinity [15]. Because binding was known to be RGD-dependent, and because 125I-echistatin was commercially available, echistatin was selected as a surrogate ligand for assay development.

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

111

Table 1 Potencies of non-peptide inhibitors and cRGDfV in buffer and plasma: correlation with % PPB Compound

Structure

Assay buffer

Plasma

Ratioa

% PPBb

IC50 (nM) Compound 1

H N

N

O

0.31c (0.08)

126 (11)

406

98.4

0.62 (0.31)

1960 (410)

3160

99.8

2.68 (0.19)

7560 (1610)

2820

99.8

5.78 (1.59)

11,000 (2800)

1900

99.4

2.10 (0.71)

95.2 (20.9)

45

31.6

N COOH N

Compound 2

H N

N

O N COOH

O O

Compound 3

H N

N

O N N

Compound 4

H N

N

COOH

O N COOH F

cRGDfV a b c

Ratios were calculated as the plasma IC50 values divided by the buffer IC50 values. Percent plasma protein binding values were determined as described in Materials and methods. IC50 values were the mean and SEM (parentheses) of 3 to 8 determinations.

3.1. Development of the radioreceptor assay (RRA) The binding of 125I-echistatin to plate immobilized avh3 was saturable (Fig. 1A). Twenty-five nanograms (0.1 pmol) of the immobilized purified integrin were incubated with increasing (0.004–1 nM) concentrations of 125I-echistatin. A one-site binding model ( Y = B max * X / (K d + X) was used to calculate an apparent K d of 0.29 nM (95% Cl = 0.27–0.30 nM). Saturation was achieved with approximately 0.01 pmol bound indicating that a significant portion of the immobilized avh3 failed to recognize the labeled peptide, possibly due to immobilization conditions or to the specific activity of the protein preparation. 125 I-echistatin binding to avh3 could be readily measured in plasma (Fig. 1B), due to the high affinity of echistatin to avh3 and the excellent signal-to-noise ratio afforded by the radioisotope. The principle effect of plasma on the echistatin–integrin interaction was to decrease the association rate (k on) from 80 min in buffer to 530 min in plasma. Binding of 125I-echistatin to avh3 was specific in either buffer or in 100% rat plasma (Fig. 1C). Increasing concentrations of unlabeled echistatin were used to compete 125I-echistatin for 2 h at room temperature. In either assay format (buffer or plasma), 125I-echistatin binding was reduced N 99% by 100 nM unlabeled echistatin. The calculated ED50s for echistatin were 0.3 and 1.9 nM in buffer and plasma, respectively, consistent with the approximate 6.6-fold reduction in the association rate in plasma. At 2 h, the signal to noise of specific binding in plasma was approximately 50 : 1 providing an RRA with a good dynamic range.

112

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

A 40000

30000

0.3

B/F

CPM

0.4

20000

0.2 0.1

10000

0.0 0.0

2.5

5.0

7.5

10.0

Bound

0 0.00

0.25 125

0.50

0.75

1.00

I-Echistatin (nM)

B

Net CPM

30000

20000 Buffer Plasma 10000

0 0

1000

2000

3000

Minutes

C 100

% Control

Buffer 75

Plasma

50 25 0 -2

-1

0

1

2

3

Echistatin (log nM) Fig. 1. Binding of 125I-echistatin to immobilized avh3 was saturable and was specific in buffer and in plasma. (A) The binding of 125I-echistatin (0.004–1.0 nM) to plate immobilized avh3 (25 ng/well) was measured as described in Materials and methods. Specific binding (squares) was calculated from total binding by subtracting non-specific binding (inverted triangles). Nonspecific binding was determined experimentally in the presence of no less than a 200 molar excess of unlabelled echistatin. The binding reaction proceeded for 2 h at room temperature. The inset displays the Scatchard plot of saturation binding data. (B) Plates containing immobilized avh3 (50 ng/well) were incubated with 0.25 nM 125Iechistatin in buffer or plasma. At times noted, wells were washed and counted. On-rates were determined using a one-site exponential association and were calculated as association half-lives. (C) Plates coated with 50 ng/well avh3 were incubated with 125I-echistatin and increasing concentrations of unlabelled echistatin in buffer or plasma. Following a 2h incubation at room temperature, the plates were washed and counted.

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

113

3.2. Small molecule avb3 inhibitors compete echistatin binding Compounds 1–4 from our lead discovery effort were tested for competition of 125I-echistatin binding to avh3 in the RRA (Fig. 2A and Table 1). Initial experiments were in assay buffer.

A: Buffer 100

% Total Signal

Compound 1 Compound 2

75

cRGDfV 50

25

0 -2

-1

0

1

2

3

log (nM)

B: Rat plasma

% Total Signal

100

75

50

Compound 1 25

Compound 2 cRGDfV

0 -1

0

1

2

3

4

log (nM)

C: Species plasma

% Total Signal

100 75 50

Canine Murine Rabbit Human

25 0 -1

0

1

2

3

4

compound 1 (log nM) Fig. 2. The cyclic RGD-peptide, cRGDfV, and small molecule RGD-mimetic compounds competed echistatin binding to avh3 in assay buffer or plasma. Two-fold dilutions of inhibitors were prepared in assay buffer containing 0.25 nM 125Iechistatin (panel A) or in 100% platelet poor rat plasma containing 0.25 nM 125I-echistatin (panel B). The prepared dilutions were transferred to plates coated with 50 ng/well avh3 and the binding of 125I-echistatin to immobilized avh3 was measured as described in Materials and methods. (Panel C) The potency of compound 1 was determined as described above in the presence of plasmas from the species indicated.

114

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

Compounds 1 and 2 competed echistatin binding with IC50 values below 1 nM and were more potent than the well-studied cyclic RGD peptide, cRGDfV [4]. These data corroborate other reports that nonpeptide RDG mimetics compete echistatin binding to avh3 [24]. 3.3. RRA revealed variable and dramatic loss of compound potency in plasma Next, we tested the ability of our leads to compete 125I-echistatin binding in 100% rat plasma. The potencies of compound 1–4 in plasma were reduced 400 to N 3000-fold when compared to potencies in buffer (Fig. 2B and Table 1). In contrast, the impact of plasma on the potency of cRGDfV was only ~45-fold. High plasma protein binding could potentially account for loss of activity in plasma. The percent of each compound determined to be plasma protein bound are reported in Table 1. Consistent with the large loss in potencies, compounds 1–4 were highly bound. Further, loss in potency appeared to correlate with the percentage of compound remaining free in plasma. For example, 1.6% of compound 1 and only 0.2% of compounds 2 and 3 were free in plasma. The 8-fold difference in unbound fraction was consistent with the ~8-fold greater plasma-induce loss in activity of compounds 2 and 3 relative to compound 1. Compared to compounds 1, 2, and 3, compound 4 was intermediate in both plasma protein binding and in plasma-induced fold loss of potency. The peptide, cRGDfV was the least bound and least effected by plasma. The identity of the plasma protein(s) binding compounds 1–4 was not identified in this study. However, albumin may play a significant role since compound 1 was determined to be 98% bound in a solution containing a physiological concentration of albumin (40 mg/ml). Knowledge of plasma protein binding and plasma potency can be important in understanding species-specific efficacy, toxicology, and pharmacokinetics. We found the RRA to be a rapid and versatile means of assessing compound activity in plasma harvested from different species. Citrated plasma from dog, mouse, rabbit, and human were substituted for rat plasma in the RRA (Fig. 2C). The favorable activity of compound 1 in canine plasma reflected moderate dog plasma protein binding (93%) relative to higher plasma protein binding (z 99%) measured in murine, rabbit, and human plasma. Compound affinity can be influenced strongly by the activation state of integrin heterodimers [25]. Divalent cations are essential for integrin function, and manganese in particular can induce a high affinity integrin conformation [25,26]. We postulated that loss of compound potency in citrated plasma might reflect the negligible concentration of manganese in plasma and the Table 2 The impact of anticoagulants and divalent cations on the potency of compound 1 in human plasma Compound 1 IC50 (nM) a

Experiment 1

Experiment 2b

Na citrate 160 Na citrate + TIc 395

Heparin Hirudin 115 250 Na citrate + TIc + 1 mM CaCl2, MgCl2, and MnCl2 429

a The IC50 value for compound 1 inhibition of echistatin–avh3 interaction was determined in the presence of purchased human citrated or heparinized-plasma or plasma prepared from a normal volunteer from blood drawn in the presence of hirudin. b The IC50 value for compound 1 was determined in citrated human plasma adjusted to contain thrombin inhibitor (TI) with or without divalent cations as indicated. The thrombin inhibitor was used to prevent clotting of plasma upon addition of divalent cations. c Thrombin inhibitor [28].

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

115

divalent cation chelating capacity of citrate. However, we found that similar to citrated plasma, plasma prepared using either heparin or hirudin also reduced the potency of compound 1 (Table 2). This finding ruled out divalent cation chelation as a mechanism to explain the loss of potency in plasma. In a second experiment (Table 2), citrated plasma supplemented to contain 1 mM MnCl2 also reduced compound 1 activity. This ruled out low plasma manganese as a mechanism to explain the loss of potency in plasma. 3.4. Use of the RRA as a pharmacodynamic tool enabling interpretation of in vivo pharmacology To select the best compound for proof-of-principle pharmacology studies, we evaluated our leads using an ex vivo pharmacodynamic model adapted from the RRA. In this model, rats were dosed with compound, and plasma was harvested subsequently at timed intervals. The plasmas were adjusted to contain 125I-echistatin, and binding of 125I-echistatin to immobilized avh3 was measured. In initial studies, the pharmacodynamic activity of compound 1 was compared to compound 2 at 50 mg/kg (Fig. 3A). Shortly after rats were dosed with either compound, plasma blocked echistatin binding to avh3 by N95%. Less inhibitory activity was observed in plasmas harvested at 4 and at 6 h, and by 8 h, echistatin binding was inhibited only 12% and 43% by

CPM (% Control)

A

100

Compound 1 (50 mpk)

80

Compound 2 (50 mpk)

*

60

*

40 20 0 0

1

2

4

6

8

Hours after dose

B 2000 Compound 2 (120 mpk)

CPM

1600

Rat#1

1200

Rat#2 Rat#3

800 400 0 0

2

4

8

12

Hours after dose Fig. 3. Pharmacodynamics of compound 1 and 2 in rats determined ex vivo. In panel A, groups of three male Sprague– Dawley rats were dosed s.c. with 50 mg/kg of compound 1 (grey bars) or compound 2 (black bars) and citrated blood was harvested from each rat just prior to dosing (T = 0) and at 1, 2, 4, 6, and 8 h post-dosing. In panel B, three rats were dosed s.c. with 120 mg/kg compound 2, and citrated blood was harvested from each rat just prior to dosing (T = 0) and at 1, 2, 4, 8 and 12 h post-dosing. Platelet-poor plasma was prepared from the fresh blood. The undiluted plasma was adjusted to contain 0.25 nM 125I-echistatin and binding of echistatin to plate immobilized avh3 was determined as described in Materials and methods. Values in panel A represent the mean values and SEM of three rats. Asterisk indicates greater inhibition by compound 2 compared to compound 1 at 6 and 8 h ( p b 0.05, Student’s t-test).

116

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

plasmas harvested from animals dosed with compound 1 and compound 2, respectively. Inhibition by compound 2 was significantly better than compound 1 at 6 and 8 h. Thus, even though compound 1 was N 10-fold more potent than compound 2 in plasma (Table 1), compound 2 demonstrated superior pharmacodynamic activity. This could be explained by the improved pharmacokinetics of compound 2 in rats. Whereas compound 1 exhibited the high rate of clearance of 65 ml/min/kg, compound 2 had a clearance of only 5 ml/min/kg (data not shown). Preliminary investigation indicated that twice daily administration of 50 mg/kg of compound 1 was ineffective in suppressing corneal angiogenesis in rats. Due to superior pharmacodynamic activity, compound 2 was selected for subsequent studies. One hundred and twenty milligrams per kilogram was the highest dose that could be delivered in a soluble formulation, and this dose was chosen to obtain maximal exposure. In the first study, corneal neovascularization was induced by intra-stromal implantation of an Elvax pellet containing bFGF. Ten rats were treated twice daily with vehicle or with compound 2 and the area of corneal neovascularization was quantified on day 9. The neovascularization area in vehicle-treated rats was 2.492 F 0.260 mm2 versus 1.715 F 0.111 mm2 in compound 2-treated rats, representing a 31% inhibition of angiogenesis ( p = 0.01, Student’s t-test). Representative corneas from vehicle and compoundtreated rats are shown in Fig. 4. In a second, more aggressive model, angiogenesis was induced following silver/potassium nitrate cautery. Five rats were treated twice daily with vehicle or with compound 2 and the area of corneal neovascularization was quantified on day 4. The neovascularization area in the vehicle-treated rats was 16.6 F 0.6 mm2 versus 13.3 F 1.7 mm2 in compound 2-treated rats, representing a 20% inhibition of angiogenesis ( p b 0.01, Student’s ttest). Thus, although the suppression was moderate, the impact of compound 2 on corneal angiogenesis was reproduced in two models. The remaining neovascularization that was not blocked in rats dosed with compound 2 may be avh3-independent. Alternatively, the dose may not have achieved adequate inhibition of avh3. To assess the latter possibility, we examined the pharmacodynamic activity of compound 2 dosed at 120 mg/kg (Fig. 3B). Plasma harvested 2 and 4 h after compound 2 administration

Fig. 4. Compound 2 inhibited bFGF-induced corneal angiogenesis in rats. Corneal neovascularization was induced by an intra-stroma implantation of an Elvax pellet (DuPont) containing 300 ng bFGF in one eye of each of twenty rats. Ten rats were treated twice daily (s.c.) with vehicle and ten rats were treated twice daily (s.c.) with 120 mg/kg compound 2. Angiogenesis was quantified on day 9 as described in Materials and methods. Rats within each group were ranked based on the area of neovascularizaion. Shown are the implanted eyes of the 4th, 5th, and 6th ranked rats of the vehicle control (panel A) compound 2 (panel B)-treated groups.

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

117

inhibited echistatin binding to avh3 by at N95%, and inhibition remained strong at 8 h. By 12 h, inhibition was incomplete, but still N 50%. With the aid of LC-MS, compound 2 was determined to be present at 5.3 F 0.8 AM in plasma 12 h after dosing. Thus, consistent with the PD assay, trough levels of compound 2 were still higher than the experimental plasma IC50. Overall, these data suggest to us that corneal angiogenesis can proceed, at least in part, in the context of substantial inhibition of vascular avh3. 4. Discussion Herein we reported the development of a simple, rapid RRA that can be used to measure the bioactivity avh3 inhibitors in 100% plasma. Further, we have shown how the RRA can be adapted as an ex vivo pharmacodynamic assay, and have provided one example where activity in plasma was correlated with moderate efficacy in two models of angiogenesis. It is proposed that this RRA could find general use as an ex vivo pharmacodynamic assay to select compounds and doses for preclinical proof-of-principle studies and for clinical development. A critical observation made in the course of these studies was the considerable (400 to N 3000fold) loss in activity when our synthetic avh3 inhibitors were assayed in plasma rather than in buffer. Compounds 1–4 were highly plasma protein bound (N 98%), and the losses of potency may reflect, in part, low levels of free compound in plasma. Loss of plasma activity appeared to correlate with plasma protein binding. For example, plasma reduced the potencies of cRGDfV, compound 1, and compound 2 by 45-, 406-, and 3161-fold, respectively. Thus, compound 1 lost ~1 logarithmic unit more potency than cRGDfV, and compound 2 lost ~1 logarithmic unit more potency than compound 1. This correlated well with a corresponding progressive logarithmic reduction in the percentage of free compound in plasma (i.e., 68.4%, 1.6%, and 0.2%, respectively). Compounds 3 and 4 also experienced a N 1000-fold loss of potency in plasma commensurate with the high plasma protein binding exhibited by these compounds. The current report was limited to a series of indol-1-yl propionic acids. However, we have tested several diverse chemical series of avh3 inhibitors (from the literature and our own laboratory), and have consistently observed a 100 to N10,000-fold loss of potency in plasma (unpublished results). Indeed, these compounds invariably exhibited high plasma protein binding, a finding consistent with reports of high plasma protein binding by other small molecule RGD-mimetic inhibitors [9,20]. Thus, high plasma protein binding and plasmainduced loss of potency may be commonly encountered impediments to the development of novel RGD-mimetic compounds. Current LC-MS-MS methods for measuring plasma protein binding are laborious, time consuming, and expensive. In contrast, the RRA described herein is a simple and rapid assay that is useful for rank ordering compound activity in plasma. Importantly, the RRA allowed good compound discrimination even when plasma protein binding was N 98%. Moreover, the RRA can be used to assay the potency of compounds in plasmas harvested from diverse species. Such data may contribute to understanding differences in efficacy, toxicology, and pharmacokinetics across species. Although the relative loss of compound potencies in plasma appeared to correlate with plasma protein binding, the absolute loss of potencies were consistently 6 to 9-fold in excess of what could be explained by plasma protein binding alone. For example, plasma protein binding of compound 1 (98.4%) would account for a 62.5-fold loss in potency, which is 6.5-fold less than the observed 406-fold loss observed experimentally. One explanation for this discrepancy, might be slow diffusion rates (and consequently slower on-rates) of compounds in plasma versus buffer as was observed for echistatin.

118

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

Numerous pharmaceutical research groups have sought non-peptide avh3 inhibitors as antiangiogenic agents, especially for the treatment of cancer. However, as noted by Tucker [27], the number of reports on anti-angiogenic activities of small molecule avh3 inhibitors has not been commensurate with the global effort to develop such inhibitors, and, to the best of our knowledge, no such compounds have entered clinical trials as anticancer agents. It is difficult to know the reason for this apparent lack of progress, particularly in light of the promising preclinical efficacy of RGD-peptides [2,5,6] and neutralizing antibodies [1], both of which are undergoing phase II clinical trials in cancer. One possibility is that antibodies may act by mechanisms distinct from small molecule RGD-mimetic inhibitors. However, it is more difficult to argue that RGD-peptides are mechanistically distinct from the RGD-mimetic inhibitors. Another possibility, suggested by our results, is that the potency of some inhibitors may be severely reduced in plasma. In any event, the field has been impeded by the absence of pharmacodynamic assays to establish that avh3 has been blocked successfully in dosed animals. Without such assays, it is possible that pharmacological studies have been performed with compound doses inadequate to achieve avh3 blockade. We propose that the best application of the RRA described herein is in the selection of compounds and doses for proof-of-principle pharmacology studies. Herein, the RRA was adapted as an ex vivo pharmacodynamic model to determine the inhibitory activity of compounds in the plasmas of dosed animals. Using this approach, compound 2 was shown to exhibit superior pharmacodynamic activity when compared to compound 1. Further, it was established that a dose of 50 mg/kg of compound 2 resulted in incomplete inhibition as early as 4 h following administration. Consequently, we selected compound 2 for our initial proof-ofconcept studies, and utilized a dose of 120 mg/kg, the highest dose deliverable in a soluble formulation. This regimen suppressed bFGF-induced and cautery-induced angiogenesis in rats by 31% and 20%, respectively. The RRA revealed that plasma harvested from rats dosed with 120 mg/kg of compound 2 blocked avh3 ex vivo by N95% through at least 4 h and N 50% through 12 h. Overall, the data suggest that although compound 2 has anti-angiogenic activity, a significant amount of corneal angiogenesis can proceed in the context of substantial (albeit incomplete) inhibition of avh3. The data do not exclude the possibility that complete inhibition may result in greater efficacy. However, as a practical matter, it appears that further inhibition would be difficult to achieve with compound 2 due to dosing limitations, and a compound with improved pharmacodynamic activity would be needed to test this hypothesis. In addition to its role in angiogenesis, avh3 mediates osteoclast attachment to bone. Very recently, Murphy et al. [11] reported significant anti-resorptive activity of the small molecule avh3 inhibitor, L-000845704, in women with postmenopausal osteoporosis. It is perhaps notable that this highly active compound was characterized by moderate plasma protein binding [10]. The promising data in osteoporosis may engender renewed interest in avh3 as a therapeutic target. Towards this end, the 125I-echistatin-avh3 RRA described herein provides a simple, rapid and versatile bioanalytical technique to rank order compound potency in plasma, and to estimate pharmacodynamic activity ex vivo following dosing of animals. The RRA may be useful for compound optimization and for dose selection for future studies designed to explore the therapeutic utility of avh3 inhibitors. 5. Simplified description Herein we described an RRA based on 125I-echistatin binding to plate-immobilized avh3. The RRA can be used to rank order the potencies of small molecule inhibitors of avh3, and is

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

119

an improvement over existing ELISA assays because it can be performed in 100% plasma. When performed in plasma, the RRA is a simple and rapid method to determine the relative impact of plasma protein binding on compound activity. The development of avh3 inhibitors has been hampered by a lack of pharmacodynamic assays. The RRA can be adapted for use as an ex vivo pharmacodynamic model by measuring 125I-echistatin binding to immobilized avh3 in the presence of plasmas harvested from animals dosed with compounds. Using this pharmacodynamic model, compound and dose selection can be optimized. Thus, the RRA provides a simple technique to rank order the potency of avh3 inhibitors in plasma, and could find general use as an ex vivo pharmacodynamic assay to select optimized compounds and doses for preclinical and clinical proof-of-principle studies. References [1] Brooks P, Stromblad S, Klemke R, Visscher D, Sarkar F, Cheresh D. Anti-integrin avh3 blocks human breast cancer growth and angiogenesis in human skin. J Clin Invest 1995;96:1815 – 22. [2] Friedlander M, Theesfeld C, Sugita M, Fruttiger M, Thomas M, Chang S, Cheresh D. Involvement of integrins avh3 and avh5 in ocular neovascular diseases. Proc Natl Acad Sci U S A 1996;93:9764 – 9. [3] Horton MA. The alpha v beta 3 integrin bvitronectin receptorQ. Int J Biochem Cell Biol 1997;29:721 – 5. [4] Brooks P, Montgomery A, Rosenfeld M, Reisfeld R, Hu T, Klier G, et al. Integrin avh3 antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels. Cell 1994;79:1157 – 64. [5] Luna J, Tobe T, Mousa S, Reilly T, Campochiaro P. Antagonists of integrin avh3 inhibit retinal neovascularization in a murine model. Lab Invest 1996;75:563 – 73. [6] Lode H, Moehler T, Ziang R, Jonczyk A, Gillies S, Cheresh D, Reisfeld R. Synergy between an antiangiogenic integrin av antagonist and an antibody–cytokine fusion protein eradicates spontaneous tumor metastases. Proc Natl Acad Sci 1999;96:1591 – 6. [7] Henry C, Moitessier N, Chapleur Y. Vitronectin receptor alpha(V)beta(3) integrin antagonists: chemical and structural requirements for activity and selectivity. Mini Rev Med Chem 2002;2:531 – 42. [8] Lark M, Stroup G, Dodds R, Kapadia R, Hoffman S, Hwang S, et al. Antagonism of the osteoclast vitronectin receptor with an orally active nonpeptide inhibitor prevents cancellous bone loss in the ovariectomized rat. J Bone Miner Res 2001;16:319 – 27. [9] Coleman PJ, Brashear KM, Askew BC, Hutchinson JH, McVean CA, Duong LT, et al. Nonpeptide avh3 antagonists: Part 11. discovery and preclinical evaluation of potent avh3 antagonists for the prevention and treatment of osteoporosis. J Med Chem 2004;47:4829 – 37. [10] Breslin MJ, Duggan MD, Halczenko W, Harman GD, Duong LT, Fenandez-Metzler C, et al. Nonpeptide avh3 antagonists. Part 10: in vitro and in vivo evaluation of a potent 7-methyl substituted tetrahydro-[1,8]naphthyridine derivative. Bioorg Med Chem Lett 2004;14:4515 – 8. [11] Murphy M.G., Cerchio K., Stoch S.A., Gottesdiener K., Wu M., Recker R.for the L-000845704 study group. Effect of L-000845704, an avh3 antagonist, on markers of bone turnover and bone mineral density in postmenopausal osteoporotic women. J Clin Endocrinol Metab 2005;90:2022 – 8. [12] Cherny RC, Honan MA, Thiagarajan P. Site-directed mutagenesis of the Arginine-Glycine-Aspartic Acid in vitronectin abolishes cell adhesion. J Biol Chem 1993;268:9725 – 9. [13] Pfaff M, Tangemann K, Muller B, Gurrath M, Muller G, Kessler H, et al. Selective recognition of cyclic RGD peptides of NMR defined conformation by alpha Iib beta 3, alpha V beta 3, and alpha 5 beta 1 integrins. J Biol Chem 1994;269:20233 – 8. [14] McLane MA, Marcinkiewicz C, Vijay-Kumar S, Wierzbicka-Patynowski I, Niewiarowski S. Viper venom disintegrins and related molecules. Proc Soc Exp Biol Med 1998;219:109 – 19. [15] Marcinkiewicz C, Vijay-Kumar S, McLane MA, Niewiarowski S. Significance of RGD loop and c-terminal domain of echistatin for recognition of aIIh3 and avh3 integrins and expression of ligand-induced binding site. Blood 1997;90:1565 – 75. [16] Ku TW, Ali FE, Barton LS, Bean JW, Bondinell WE, Burgess JL, et al. Direct design of a potent non-peptide fibrinogen receptor antagonist based on the structure and conformation of a highly constrained cyclic RGD peptide. J Am Chem Soc 1994;115:8861 – 2. [17] Reinmuth N, Liu W, Ahmad SA, Fan F, Stoeltzing O, Parikh AA, et al. avh3 integrin antagonist S247 decreases colon cancer metastasis and angiogenesis and improves survival in mice. Cancer Res 2003;63:2079 – 87.

120

M.A. Chaikin et al. / J. Biochem. Biophys. Methods 65 (2005) 107–120

[18] Ker JS, Wexler RS, Mousa SA, Robinson CS, Wexler EJ, Mohamed S, et al. Novel small molecule alpha v integrin antagonists: comparative anti-cancer efficacy with known angiogenesis inhibitors. Anticancer Res 1999;19:959 – 68. [19] Kumar CC, Malkowski M, Yin Z, Tanghetti E, Yaromko B, Nechuta T, et al. Inhibition of angiogenesis and tumor growth by SCH221153, a dual alpha(v)beta3 and alpha(v)beta5 integrin receptor antagonist. Cancer Res 2001;61:2232 – 8. [20] Ward KW, Azzarano LM, Bondinell WE, Cousins RD, Huffman WF, Jakas DR, et al. Preclinical pharmacokinetics and interspecies scaling of a novel vitronectin receptor antagonist. Drug Metab Dispos 1999;27:1232 – 41. [21] Nage JA, Brown LF, Senger DR, Lanir N, Van de Water L, Dvorak AM, et al. Pathogenesis of tumor stroma generation: a critical role for leaky blood vessels and fibrin deposition. Biochim Biophys Acta 1989;948:305 – 26. [22] Marugan JJ, Manthey C, Anaclerio B, Lafrance L, Lu T, Markotan T, et al. Design, synthesis, and biological evaluation of novel potent and selective alpha(v)beta(3)/alpha(v)beta(5) integrin dual inhibitors with improved bioavailability. Selection of the molecular core. J Med Chem 2005;48:926 – 34. [23] Leonard K., Pan W.B., Anaclerio J., Gushue Z., Guo M., Chaikin R., et al. Non-peptidic avh3 antagonists containing indol-1-yl propionic acids. Bioorg Med Chem Lett 2005;15:2679 – 84. [24] Thibault G, Tardif P, LaPalme G. Comparative specificity of platelet aIIh3 integrin antagonists. J Pharm Exp Ther 2001;296:690 – 6. [25] Chen LL, Whitty A, Lobb RR, Adams SP, Pepinsky RB. Multiple activation states of integrin alpha4beta1 detected through their different affinities for a small molecule ligand. J Biol Chem 1999;274:13167 – 75. [26] Mehta RJ, Diefenbach B, Brown A, Cullin E, Jonczyk A, Gussow D, et al. Transmembrane-truncated avh3 integrin retains high affinity for ligand binding: evidence for an binside-outQ suppressor? Biochem J 1998;330:861 – 9. [27] Tucker GC. av integrin inhibitors and cancer therapy. Curr Opin Investig Drugs 2003;4:722 – 31. [28] Lu T, Tomczuk B, Bone R, Murphy L, Salemme FR, Soll RM. Non-peptidic phenyl-based thrombin inhibitors: exploring structural requirements of the S1 specificity pocket with amidines. Bioorg Med Chem Lett 2000;10:83 – 5.