Validation of in vivo pharmacodynamic activity of a novel PDGF receptor tyrosine kinase inhibitor using immunohistochemistry and quantitative image analysis Free

With the advent of agents directed against specific molecular targets in drug discovery, it has become imperative to show a compound's cellular impact on the intended biomolecule in vivo. The objective of the present study was to determine if we could develop an assay to validate the in vivo effects of a compound. Hence, we investigated the in vivo pharmacodynamic activity of JNJ-10198409, a relatively selective inhibitor of platelet-derived growth factor receptor tyrosine kinase (PDGF-RTK), in tumor tissues after administering the compound orally in a nude mouse xenograft model of human LoVo colon cancer. We developed a novel assay to quantify the in vivo anti-PDGF-RTK activity of the inhibitor in tumor tissue by determining the phosphorylation status of phospholipase Cγ1 (PLCγ1), a key downstream cellular molecule in the PDGF-RTK signaling cascade. We used two antibodies, one specific for the total (phosphorylated and unphosphorylated forms) PLCγ1 (pan-PLCγ1) and the other, specific for phosphorylated form of PLCγ1 (ph-PLCγ1) to immunohistochemically detect their expression in tumor tissues. Computer-assisted image analysis was then used to directly compare the ratio of ph-PLCγ1 to pan-PLCγ1 immunolabeling intensities in serial sections (5 μm) of tumors obtained from vehicle- and JNJ-10198409-treated tumor-bearing mice. Our data showed statistically significant, dose-dependent differences in the ph-PLC/pan-PLC ratio among the four treatment groups (vehicle, 25, 50, and 100 mg/kg b.i.d.). These results confirmed this compound's ability to suppress PDGF-RTK downstream signaling in tumor tissues in vivo. In addition to this specific application of this in vivo validation approach to those targets that use PLCγ as a downstream signaling partner, these methods may also benefit other drug discovery targets.

The successful discovery of new agents with therapeutic potential requires demonstration of in vivo penetration and action in target tissues by an investigational compound, which is especially important for agents that suppress intracellular growth signal cascades. In models of therapeutic action in oncology, the efficacy of a compound is traditionally evaluated using measurements such as the reduction of tumor volume or tumor weight. However, there remains a need to validate the mechanism of action of a given compound by measuring the compound's activity at its intended molecular target in the tumor tissue in vivo.

JNJ-10198409 is a dual-mechanism, antiangiogenic, and tumor cell antiproliferative agent, and is a potent ATP competitive inhibitor of platelet-derived growth factor receptor tyrosine kinase (PDGF-RTK) with an IC₅₀ of 2 nmol/L (1). The compound was active in the in vitro rat aortic ring assay of angiogenesis (IC₅₀ = 15 nmol/L), as well as in the in vivo Matrigel mouse model of growth factor–stimulated angiogenesis (1). Furthermore, it had shown dose-dependent inhibitory activity against tumor growth in vivo in several human tumor xenograft models, including LoVo human colon tumors (1). In these studies, we showed that JNJ-10198409 inhibited the activation of PDGF-RTK in ex vivo tumor tissues by examination of the phosphorylation status of a downstream signaling partner, phospholipase Cγ1 (PLCγ1). PLCγ1 was involved in intracellular signaling, and in response to extracellular stimuli, such as PDGF, became activated via phosphorylation, which subsequently hydrolyzed PIP2 to generate two secondary messengers: inositol trisphosphate and diacyglycerol (2).

Therefore, inhibition of PDGF-RTK will in turn inhibit downstream PLCγ1 phosphorylation (ph-PLCγ1). Based upon this foundation, we designed a process to evaluate the in vivo pharmacodynamic effects of JNJ-10198409 by assessing the status of ph-PLCγ1 to pan-PLCγ1 (pan = phosphorylated and unphosphorylated forms) in serially sectioned (5 μm) tumors from vehicle- and JNJ-10198409–treated mice using immunohistochemistry and computer-assisted image analysis. The tumor measurement data showed a statistically significant, antitumor activity. A significant dose-dependent reduction in the ratio of ph-PLCγ1 to pan-PLCγ1 in tumor tissues from mice treated with compound was shown which validated the in vivo pharmacodynamic activity of the compound.

JNJ-10198409 [(6,7-dimethoxy-2,4-dihydro-indeno [1,2-c]pyrazol-3-yl)-(3-fluro-phenyl)-amine] was synthesized using a process that has been previously described (3). For the in vivo tumor xenograph studies, the compound was suspended in 0.5% methyl cellulose (0.5%) and given orally by gavage at the indicated doses at a dosing volume of 10 mL/kg.

The LoVo human colon carcinoma cell line was obtained from the American Type Culture Collection (Manassas, VA). Cell culture reagents were obtained from Life Technologies, Inc. (Grand Island, NY) and the cells were maintained at 37°C plus 5% CO₂ as exponentially growing monolayers in the recommended growth medium.

Female athymic nude mice (nu/nu) were obtained from Charles River Laboratories (Wilmington, MA) and were maintained according to NIH standards. Mice (9–10 weeks of age) were group-housed under clean-room conditions in sterile microisolator cages on a 12-hour light/dark cycle at 21°C to 22°C and 40% to 50% humidity. Mice were fed an irradiated standard rodent diet ad libitum. All animals were housed in the Johnson & Johnson Pharmaceutical Research and Development Laboratory Animal Medicine facility, which is fully accredited by the American Association for Assessment and Accreditation of Laboratory Animal Care. All procedures involving animals were conducted in compliance with the NIH Guide for the Care and Use of Laboratory Animals and protocols were approved by the Johnson & Johnson Pharmaceutical Research and Development Internal Animal Use and Care Committee.

PDGFR-β kinase activity was measured in streptavidin-coated 96-well Flash plates (Perkin-Elmer Life and Analytical Sciences, Boston, MA) using purified recombinant human PDGFR-β kinase domain protein (source described below) and PLCγ peptide (Biotin) KHKKLAEGSAYEEV-amide (synthesized at Johnson & Johnson Pharmaceutical Research and Development, La Jolla, CA) as the substrate for phosphate incorporation. Inhibition of PDGFR-β kinase activity was measured by quantifying the amount of [³³P] incorporated from [³³P]-γ labeled ATP of various concentrations of test compound compared with DMSO into the immobilized PLCγ substrate in the presence controls. Briefly, PDGFR-β enzyme was diluted in reaction buffer consisting of 50 mmol/L Tris-Cl (pH 8), 10 mmol/L MgCl₂, 0.1 mmol/L Na₃VO_4, 1 mmol/L DTT, 1% DMSO, 0.25 μmol/L peptide substrate, 0.8 μCi per well [³³P]-γ-ATP (2,000–3,000 Ci/mmol), and 5 μmol/L ATP in the presence or absence of test compound. Following incubation at 30°C for 1 hour, the reaction was terminated by washing with PBS-EDTA (100 mmol/L) and the plates counted in a scintillation counter. Each test compound was evaluated at eight serially diluted concentrations in single log units over the range of 0.001 to 10 μmol/L. The percentage of inhibition from DMSO control was calculated at each drug concentration and the IC₅₀ values were calculated using nonlinear regression analysis (GraphPad Prism, GraphPad Software, Inc., San Diego, CA).

The oral antitumor efficacy of JNJ-10198409 was evaluated in vivo using the human tumor nude mouse xenograft growth delay model. Female athymic nude mice were inoculated s.c. in the inguinal region of the thigh with 1 × 10⁶ tumor cells in 0.1 mL volume (day 0). For growth delay studies, mice were randomly assigned to treatment groups and were dosed orally by gavage with vehicle (control group) or experimental compound in vehicle beginning 3 days after tumor cell inoculation. Compound was prepared in 0.5% methylcellulose and treatments were given orally b.i.d. (twice daily) for 31 consecutive days. Tumor growth was measured twice weekly (Mondays and Fridays) using electronic calipers for the duration of the study. Tumor area (mm²) was calculated using the formula (l × w): where l, length; w, width of the tumor. Mice were examined at each daily dose for clinical signs of adverse, drug-related side effects. At study termination, mice were euthanized and solid tumors were surgically excised intact and weighed. Tumors were fixed in 10% formalin for immunohistochemistry. Differences between treated and control mice were analyzed statistically by ANOVA and Dunnett's t test, with a P value of ≤0.05 (two-tailed) considered statistically significant. GraphPad Prism (version 3) was used for all statistical analysis and for graphic presentation.

Formalin-fixed tumors were trimmed and processed for paraffin embedding according to conventional methods. Five-micron sections were serially cut, mounted onto SuperFrost Plus+ (Fisher Scientific, Pittsburgh, PA) microscopic slides and dried overnight. Tissue sections on microscopic slides were dewaxed and rehydrated and processed for immunohistochemistry (4, 5). Slides were heated in Target buffer (Dako, Carpinteria, CA), cooled, placed in PBS [(pH 7.4) Zymed Labs, South San Francisco, CA], and treated with 3.0% H₂O₂ for 10 minutes at room temperature. All incubations (30 minutes) and washes were done at room temperature. Normal blocking serum (Vector Labs, Burlingame, CA) was placed on all slides for 10 minutes. After a brief rinse in PBS, sections were incubated with the primary antibodies to the phosphorylated and unphosphorylated forms of PLCγ1 (pan-PLCγ1) (1:200, Santa Cruz, CA) and to the phosphorylated form of PLCγ1 (ph-PLCγ1) (1:200, Santa Cruz). Slides were then washed in PBS and treated with biotinylated secondary antibodies (Vector Labs). After washing in PBS, the avidin-biotin-horseradish peroxidase complex reagent (Vector Labs) was added. Slides were washed and subsequently treated twice with 3,3′-diaminobenzidine (Biomeda, Foster City, CA) for 5 minutes, rinsed in distilled water, and counterstained with hematoxylin. Additional negative control slides included the replacement of the primary antibody with non–immune serum (Vector Labs) or the antibody dilution buffer (Zymed Labs).

Image analysis values (Olympus BX-50, Sony 3CCD DKC-5000 camera, Image Pro, v. 4.1.0.9, Media Cybernetics, Phase III Imaging, Glen Mills, PA) were set to measure 3,3′-diaminobenzidine-positive immunoreactivity in all serially sectioned tissues (6). Specifically, the immunolabeled areas were identified through the software (point and click on immunolabeled areas) to generate the image analysis values, which were obtained as a function of area density using the identical threshold range (0–197) for each channel (red, green, and blue). These values were calibrated by comparing the negative control slides (value = 0) to the immunolabeled slides and were maintained as a standard throughout the analyses. As an additional control, all of the captured tissue images for these analyses were stored with identical photographic settings (brightness = 55, contrast = 55). Control for the immunolabeling was based on the condition that all of the slides were treated at the same time with the same reagents using the same 3,3′-diaminobenzidine development time (twice for 5 minutes). Analyses were done under a very low magnification using a 2× objective (30× final magnification; totaling an average 18.66 mm² scanning area per tissue) to include as much tissue area as possible in order to avoid the need to select “random” fields for analysis. Data were calculated as a percentage of the amount of immunolabeling in the total tissue area (total area of positive pixels/total number of pixels) independent of specific cellular labeling patterns and subjected to Student's t test statistical analysis using SigmaStat (SPSS Inc., San Rafael, CA) and plotted using SigmaPlot (SPSS Inc.). A P value <0.05 was considered a significant difference.

JNJ-10198409 (Fig. 1) is a novel investigational anticancer agent with a dual mechanism of action that has both antiangiogenic activity through blockade of the PDGF-RTK pathway and a direct tumor cell antiproliferative activity. The compound is a potent ATP competitive inhibitor of PDGF-RTK and is active in the in vitro rat aortic ring angiogenesis assay as well as in the in vivo Matrigel mouse model of growth factor–stimulated angiogenesis (1, 3).

As shown in Fig. 2, JNJ-10198409 is a potent PDGFR inhibitor with an IC₅₀ of 2 nmol/L in the PDGF-RTK kinase assay (see Materials and Methods). Although the compound is very active against PDGFR, it is much less potent against other closely related receptor tyrosine kinases, such as vascular endothelial growth factor and fibroblast growth factor (data not shown). In a broad kinase screening against more than 100 kinases, the compound has shown good selectivity with only significant activity against c-Abl, and moderate activity against c-Src and several c-Src-related kinases (data not shown). This data indicates that JNJ-10198409 is a potent PDGFR inhibitor with high degree of selectivity.

To validate the presence and investigate the changes in expression of PDGF receptor-β and the phosphorylated and total amounts of PLCγ1, we use Western blot analysis on LoVo cell lysates under various conditions (Fig. 3). These data showed that with vehicle and increasing doses of JNJ-10198409 (0.001, 0.1, and 1.0 μmol/L) for 24 hours, we did not detect any changes of PDGF receptor-β expression (Fig. 3A). Interestingly, we detected a dose-dependent reduction of phosphorylated PLCγ1 expression (Fig. 3B) in spite of consistent expression patterns of total PLCγ1 in the cell lysates (Fig. 3C). These data suggested that our compound could inhibit the phosphorylation of PLCγ1 in vitro.

The compound had also shown a dose-dependent oral antitumor activity against several human tumor xenograft models, including LoVo human colon tumors, from which the tumor tissues were collected for the current study (Fig. 4). Subsequently, it became imperative to develop a method to evaluate the in vivo activity of the compound against its putative antiangiogenic target, PDGF-RTK and the resulting inhibition of downstream signaling events. As shown in Fig. 4, orally administered JNJ-10198409 produced a statistically significant, dose-dependent reduction in tumor growth. After 31 days of twice daily p.o. dosing, mean final tumor area (mm²) was inhibited by 15%, 64%, and 91% at doses of 25, 50 and 100 mg/kg, p.o. b.i.d., respectively.

As the study terminated, tumors were excised intact, weighted, and fixed in 10% formalin for subsequent immunohistochemistry analyses. Positive PLCγ1 immunolabeling was presented as brown staining (Fig. 5). We did not observe any detectable immunolabeling in the negative controls (data not presented). Representative adjacent images were presented in Fig. 5 and showed positive pan-PLCγ1 (Fig. 5A and C) and ph-PLCγ1 (Fig. 5B and D) immunolabeling (arrows) in serially sectioned xenograft tumors in an animal from the vehicle-treated group (Fig. 5A and B) and from the high dose JNJ-10198409-treated (100 mg/kg) group (Fig. 5C and D). Grossly, less amounts of ph-PLCγ1 immunolabeling (Fig. 5D) were detected as compared with the pan-PLCγ1 immunolabeling (Fig. 5C) in the adjacent, serially sectioned area of the same tumor and compared with the typical labeling intensities of the vehicle-treated animals. However, a robust means to quantify the relative amounts of ph-PLCγ1 and pan-PLCγ1 immunolabeling intensity beyond the soft, subjective (weak, moderate, or strong) criteria was required. Therefore, we employed computer-assisted image analysis to quantify the relative amount of immunolabeling in tissue sections.

The resultant data showed a dose-dependent, statistically significant decrease of the ratio or percentage of ph/pan-PLCγ1 (Fig. 6). About 60% (63.14 ± 9.36% SE; n = 14) of the total PLCγ1 was phosphorylated in the vehicle-treated animals, which may represent basal levels. A minimal alteration in staining was observed (P < 0.05) in the ratio of ph/pan-PLCγ1 in the low dose (25 mg/kg) treatment group (58.83 ± 11.12% SE; n = 8). However, we observed significant decreases in the percentages of ph-PLCγ1 immunolabeling in the tumors obtained from the moderate (48.00 ± 6.08% SE; 50 mg/kg) and high (30.85 ± 5.34%; 100 mg/kg) compound dose groups (P < 0.01 and P < 0.005, respectively).

Critical in the discovery of new agents with multiple pharmacologic activities is the development of an in vivo validation assay to determine if a compound can reach the desired target tissue to affect a relevant cellular process, which is the objective of this study. One such agent is JNJ-10198409, a potent inhibitor of PDGF-receptor tyrosine kinase activity (Fig. 2) with the ability to suppress tumor growth in nude mouse human tumor xenograft models, as well as inhibit tumor cell antiproliferative activity that is independent of its anti-PDGF-RTK effect (1). This compound shows good selectivity against PDGFR with only significant activity against c-Abl in a broad kinase screen against >100 kinases, and is highly active against PDGFR in in vitro enzymatic kinase assays. To better understand the antitumoral action of the agent, we sought pharmacodynamic evidence of pharmacologic activity in a signaling cascade relevant to the PDGFR kinase activity of the agent. We first showed that LoVo colon cells express the PDGF receptor and then set out to determine if the compound suppressed downstream signaling events of PDGFR activation. Additional in vivo antitumor growth efficacy of the compound was observed in four other tumor xenograft models, including H460 lung, MDA-MB-231 breast, A375 melanoma, and PC3 prostate.¹

Because our initial in vitro experiments showed that this PDGF-RTK inhibitor reduced downstream phosphorylation of PLCγ1 in the LoVo tumor cell line treated with the compound, we wanted to determine the phosphorylation status of PLCγ1 (ph-PLCγ1) in tumor tissue from animals treated with the compound. PLCγ1 is a direct tyrosine kinase substrate (7, 8). The activation of PLCγ1 following the stimulation of RTK catalyzes the hydrolysis of phosphatidylinositol 4,5-bisphosphate to produce the second messenger molecules inositol trisphosphate and diacyglycerol. Although inositol trisphosphate triggers the mobilization of intracellular Ca²⁺, diacyglycerol acts as an endogenous activator of protein kinase C (2). The role of PLCγ1 as a promoter of mitogenic response was previously not well-established (8). However, recent studies indicate that PLCγ1 may be necessary to promote the induction of immediate early genes, including c-Fos, by growth factors such as PDGF (9). Targeted disruption of PLCγ1 gene in mice results in embryonic lethality, further supporting the role of PLCγ1 in mitogenesis and cellular proliferation (10). Several studies have indicated that PLCγ1 may be essential for growth factor induction of DNA synthesis (11–13). Furthermore, the deficiency of PLCγ1 gene can impair vasculogenesis in mice, indicative and consistent with potential promitogenic and proliferative functions of PLCγ1 (9). Therefore, the phosphorylation of PLCγ1 by PDGF-RTK activation may mediate the proliferative signaling and promote cell growth as well, although the role of mitogen-activated protein kinases in mitogenesis and proliferation in response to growth factor and mitogen stimulation is much better understood and established. The present study shows that JNJ-10198409 inhibits the molecular signal transmitting from PDGFR phosphorylation. This is more convincing than simply illustrating the inhibition of PDGFR phosphorylation by the compound because that may not sufficiently show the functional activity of the agent in vivo.

Our study showed that JNJ-10198409 inhibited the targeted PDGF-initiated cascade in the tumor following oral administration. However, because of the multiple mechanisms of action of this compound, the observed anti-PDGF activity cannot be directly correlated with reduction in tumor size observed with the tumor measurement data. Nevertheless, by demonstrating the blocking of PDGF-RTK activity in the tumor, more specifically the PLCγ1 phosphorylation with the PDGF-RTK inhibitor, we can show that the compound is affecting a relevant molecular target, the PDGF receptor tyrosine kinase. Using immunohistochemistry and computer-assisted image analysis, we showed a dose-dependent reduction of ph-PLCγ1 in the targeted tumors, which was the goal of this study. These data confirmed the pharmacologic potential of our compound in vivo. In addition, tumor cells in a cross-section of the tumor tissue represent various stages of the cell cycle and this immunohistochemistry approach measures the entire section. Although employing immunohistochemistry to validate or characterize physiologic processes in vivo is not unique, the application of computer-assisted image analysis to semiquantitate immunoreactivity is novel. For example, in a related study, an inhibitor of c-Met was used to block c-Met phosphorylation (14). In this study, immunoblot analysis was used to validate efficacy because representative micrographs of ph-c-Met and total c-Met, which showed a decrease in ph-c-Met as compared with vehicle (14). In another recent study, anti-Flt-3 and antiphosphotyrosine antibodies and immunoblot analysis were employed to evaluate the intended therapeutic response of an investigational agent in a phase 1 study (15). Because no phospho-specific antibody was used in that study, a highly selective antiphosphotyrosine antibody was used to evaluate phosphorylated Flt-3 versus total Flt-3 protein (15). In another pharmacodynamic study evaluating an investigational kinase inhibitor, reduction in phosphorylated PDGF receptor and phosphorylated extracellular signal-regulated kinase was shown by immunohistochemical methods but without image analyses (16).

In summary, the present study showed that the in vivo phosphorylation status of PLCγ1 was altered as reflected by the decreased ratio of ph-PLCγ1/pan-PLCγ1 in tumor tissues from mice that were treated with the compound orally and that the reduction of this signal was associated with increased dose. Even though the phosphorylation of PLCγ1 is a key downstream cellular molecule in the PDGF-RTK signaling cascade, it is certainly possible that our compound may be affecting the phosphorylation status of PLCγ1 independent of PDGF-RTK signaling. Whatever the case, the data obtained from this study showed that our compound negatively affected PLCγ1 phosphorylation in vivo in a dose-dependent manner. It is expected that the applications of other chemo-treatments will not yield similar findings unless they also target the PDGF-RTK/PLCγ1 pathway. Therefore, this approach represents a pharmacodynamic methodology with relatively quick readout and specificity for developing a targeted anti-PDGF-RTK inhibitor. This immunohistochemistry-based method using a pair of ph-PLCγ1 and pan-anti-PLCγ1 antibodies was reliable, reproducible, and relatively inexpensive for validating and quantifying the in vivo pharmacodynamic action of the compound against the intended molecular target. Our approach can be applied to validate and quantify the in vivo activities of investigational agents against diverse molecular targets given the increasing availability of phosphospecific antibodies. Furthermore, it will have additional applications for agents that act against any upstream signaling partners of PLCγ including other receptor tyrosine kinases such as vascular endothelial growth factors or fibroblast growth factors.

We thank Debra Polkovitch, Danielle Lawrence, Meghan Towers, and Brenda Hertzog for their excellent histologic technical expertise, and Stacy Skrzat for carrying out the in vivo study from which the tumor tissues were harvested.