Background
CP-690,550, a selective inhibitor of the JAK family of protein tyrosine kinases, is being developed as an immunosuppressive and anti-inflammatory agent for the treatment and prevention of acute allograft rejection, RA, psoriasis and other immune mediated diseases [
1‐
6].
In clinical trials, CP-690,550 administration resulted in a dose-related decrease in PBNCs in active RA patients [
7,
8] within 2 weeks of treatment, but not in psoriasis patients [
9], renal allograft patients [
10] or normal volunteers [
11] for up to 14 and 28 days of treatment, respectively. In the RA trial [
7,
8], as observed in other RA studies [
12], patients were found to have baseline PBNCs which were at or above the upper limit of the reference range for normal human subjects. Following treatment with CP-690,550 for 2 weeks, PBNCs in these patients were found to decrease to within the normal reference range, and showed a strong dose-related correlation with the anti-inflammatory activity of the compound [
13].
Multiple inflammatory cytokine receptors signal through pathways involving JAK1 and JAK3, and their inhibition with CP-690,550 likely leads to anti-inflammatory and immunosuppressive activity. Conversely, JAK2 is required for signaling through several growth factor receptors and is important for myeloid and erythroid hematopoiesis [
14‐
16].
The aim of the current study was to characterize the potency and selectivity of CP-690,550 for the JAK family members and to determine if PBNC reductions in the context of arthritis are related to the anti-inflammatory efficacy of CP-690,550 (through JAK 1 and JAK3 inhibition), or due to inhibition of hematopoiesis through inhibition of JAK2 at efficacious exposures. The
in vitro potency and selectivity of CP-690,550 were determined using recombinant human kinases and whole blood cytokine induced STAT phosphorylation assays. In studies published previously, CP-690,550 was shown to inhibit arthritis development and bone destruction in the rat AIA model [
17]. In the current study, rat AIA was used to characterize the
in vivo effects of inflammation and CP-690,550 treatment on PBNCs and bone marrow myeloid progenitors, and
in vitro bone marrow progenitor cell differentiation assays were used to determine the direct effects of CP-690,550 on granulopoiesis. Additionally, pharmacokinetic and pharmacodynamic modeling was used to determine the drug concentration-effect relationship between JAK kinase inhibition and neutrophil reductions in the non-clinical and clinical studies.
Results of this study suggest that the CP-690,550 mediated reduction in PBNCs in the rat AIA model, and likely in human RA patients, is due to the anti-inflammatory action of the compound by the suppression of cytokines and chemotactic factors which elevate neutrophil counts in the peripheral compartment.
Methods
Enzyme Potency and Selectivity Assays
Recombinant kinase domains of JAK2 and JAK3 were purchased from Invitrogen (Madison, WI), and recombinant GST-fusions of JAK1 (residues 852-1142) and TYK2 (residues 870-1187, containing a C1187 S modification) kinase domains were expressed and purified at Pfizer Laboratories. MgATP was obtained from Sigma Chemical Company (St. Louis, MO). JAKtide (FITC-KGGEEEEYFELVKK) and IRS-1 (5-FAM-KKSRGDYMTMQIG) peptides were purchased from American Peptide Company (Sunnyvale, CA). FITC-conjugated antibodies against human CD8, mouse CD8, mouse CD11b and rat CD3; PE-conjugated antibodies against human CD3, human CD14, mouse CD3 and rat CD4; and AlexaFluor®647 (Ax647) conjugated pSTAT1, pSTAT3, pSTAT5 and pSTAT6 monoclonal antibodies were from BD Biosciences (San Jose, CA). The PE-conjugated antibody against mouse F4/80 was from eBioscience (San Diego, CA). Recombinant human, mouse and rat cytokines were from R&D Systems (Minneapolis, MN).
In Vitro Kinase Assays
A peptide mobility shift assay was used to quantify the phosphorylation of JAKtide (for JAK2 and JAK3) or IRS-1 (for JAK1 and TYK2) peptide substrates. Endpoint reactions were carried out at the apparent K
m (MgATP) for each enzyme (4 μM for JAK2 and JAK3, 40 μM for JAK1 and 7 μM for TYK2) in the presence of CP-690,550 and 1 μM substrate. Samples were analyzed by LabChip 3000 from Caliper Life Sciences (Hopkinton, MA). Data was analyzed using HTS Well Analyzer Software from Caliper Life Sciences to determine the amount of product formed, and was expressed as percent of control activity based on uninhibited and no enzyme controls. Dose-response data were fit using 4 parameter logistic fit software to determine IC
50. To determine the inhibition constant (K
i) for each enzyme, initial velocities were measured at varying concentrations of CP-690,550 and MgATP. The data from each experiment were fit to competitive, noncompetitive and uncompetitive inhibition models, to determine nonlinear fit and Lineweaver-Burk transformations [
18].
In Vitro Whole Blood STAT Phosphorylation Assays
Heparinized normal human, DBA/1 mouse or Lewis rat whole blood was pre-incubated with CP-690,550 for 1 hour and stained with lineage-specific antibodies. Specifically, species-specific antibodies to CD3 and CD8 were used to identify human, mouse and rat total T cells or the CD8 subset, while anti-CD14 was used to label human monocytes, anti-CD11b and anti-F4/80 were used to label mouse monocytes, and anti-CD3 and anti-CD4 were used to identify rat monocytes (CD3-, CD4+). Blood was stimulated with or without cytokine (IL-2, IL-4, IL-6, IL-7, IL-15, IL-21 and IFN-γ at 100 ng/mL, GM-CSF at 20 ng/mL or IFN-α at 1000 units/mL) at 37°C for 8-20 minutes, and activation was stopped by the addition of Lyse/Fix Buffer (BD Biosciences) following the manufacturer's protocol. Cells were washed, permeabilized in ice-cold Perm Buffer III (BD Biosciences) for 20 minutes, and stained with Ax647-conjugated phospho STAT-specific monoclonal antibodies. Flow cytometric analysis (FACS) was performed on a FACSCalibur equipped with an HTS plate loader (BD Biosciences) running Cellquest software. Twenty thousand cellular events were collected per sample, and listmode data was analyzed using FlowJo software (Treestar, Ashland, OR). Ax647 geometric mean channel fluorescence (MCF) was determined from either the CD3+ T cell population, the CD3+/CD8+ T cell subset, or from CD14+ human monocytes, CD11b+/F4/80+ mouse monocytes or CD3-/CD4+ rat monocytes. Percent of control STAT phosphorylation was determined at each concentration of CP-690,550 by comparison of Ax647 MCF with those from unstimulated and cytokine stimulated controls. IC50 values were determined using four parameter logistic fit software.
Colony forming cell (CFC) assays were performed by StemCell Technologies, Inc. as previously described by Pereira [
19]. CP-690,550 was prepared in DMSO and added to culture dishes at a final concentration of 0.1% DMSO. Human bone marrow hematopoietic cells (Lonza, Walkersville, MD), were added to the appropriate media formulations to obtain final plating concentrations. CFC assays contained 30% fetal bovine serum/1% bovine serum albumin. Cultures were plated and incubated at 37°C, 5% CO
2, and colonies were evaluated and scored microscopically. A dose-response graph was generated plotting the log of the compound concentration versus the percentage of control colony growth using Microcal Origin software. IC
50 values were calculated using the Boltzman equation. CP-690,550 was evaluated in two separate experiments and each IC
50 value in Table
1 is a representation of 6 separate dose-response curves. Cytotoxicity was evaluated in Huh 7 cells using the MTT (3-(4,5-dimethylthiazol-2-yl)-2 diphenyltetrazolium bromide; Sigma) reduction assay.
Table 1
CP-690,550 In Vitro Potency, Selectivity and Inhibition of Myeloid Progenitor Cell
Recombinant human kinase
| |
IC
50
(nM)
| |
K
i
(nM)
|
JAK1 | | 3.2 ± 1.4 | | 0.68 ± 0.12 |
JAK2 | | 4.1 ± 0.4 | | 0.99 ± 0.04 |
JAK3 | | 1.6 ± 0.2 | | 0.24 ± 0.02 |
TYK2 | | 34.0 ± 6.0 | | 4.39 ± 0.27 |
CP-690,550 Inhibition of Cytokine Signaling in Human Whole Blood
A
|
Cytokine
|
JAK
|
pSTAT
|
CD3
+
T cells
IC
50
(nM)
|
Monocytes
IC
50
(nM)
|
IL-2 | 1/3 | 5 | 28 ± 5 | NA |
IL-4 | 1/3 | 6 | 50 ± 5 | NA |
IL-7 | 1/3 | 5 | 38 ± 9 | NA |
IL-6 | 1/2 | 1 | 54 ± 7 | NA |
IL-6 | 1/2 | 3 | 367 ± 49 | 406 ± 68 |
IFN-α | 1/TYK2 | 1 | 44 ± 4 | 148 ± 41 |
IFN-γ | 1/2 | 1 | NA | 178 ± 38 |
Species Comparison of CP-690,550 Inhibition of Cytokine Signaling in Whole Blood
A
|
Cytokine
|
JAK
|
pSTAT
|
Cells
|
Human
IC
50
(nM)
|
Mouse
IC
50
(nM)
|
Rat
IC
50
(nM)
|
IL-15 | 1/3 | 5 | CD8+ T cells | 56 ± 6 | 42 ± 12 | NA |
IL-21 | 1/3 | 3 | CD3+ T cells | 25 ± 6 | NA | 187 ± 48 |
IL-6 | 1/2 | 1 | CD3+ T cells | 54 ± 7 | 185 ± 46 | 62 ± 14 |
GM-CSF | 2 | 5 | Monocytes | 1377 ± 185 | 4379 ± 655 | 877 ± 171 |
CP-690,550 Inhibition of Human and Mouse Myeloid Progenitor Cell Differentiation
B
|
Total Human Myeloid Colonies
IC
50
(nM)
|
Human
CFU-G
C
Colonies
IC
50
(nM)
|
Total Mouse Myeloid Colonies
IC
50
(nM)
|
Mouse
CFU-G Colonies
IC
50
(nM)
|
870 ± 120 | 930 ± 30 | 1210 ± 490 | 1128 ± 524 |
Rat Adjuvant-Induced Arthritis
Female Lewis rats, 160-180 grams, were obtained from Harlan Laboratories (Indianapolis, IN). Food and water were available ad libitum. The Pfizer Institutional Animal Care and Use Committee reviewed and approved the animal use in these studies. The animal care and use program is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care. Adjuvant was prepared with Mycobacteria butyricum (Lot #264010, Difco Laboratories, Detroit, MI) in a 15 mg/mL suspension with squalene oil (Sigma) with a PT3100 homogenizer (Kinematica, Bohemia, NY). On day 0, rats were anesthetized with CO2/O2 and three 50 μL injections were administered subcutaneously at the base of the tail. Rats were dosed orally with either vehicle (0.5% methylcellulose/0.025% Tween-20) alone or CP-690,550 suspended in vehicle. Dosing was initiated on day 14 and continued twice daily through day 21 post-immunization. The volume of both the right and left hind paw of each rat was measured by plethysmometer (Model 7140, Ugo Basile, Collegeville, PA) and the combined volume of both paws was calculated. Blood was collected in EDTA microtainer tubes (Becton Dickinson, Franklin Lakes, NJ) under CO2/O2 anesthesia. Absolute total neutrophil counts in each sample were measured from whole blood using a Cell-Dyne 3700 instrument (Abbott Laboratories, Abbott Park, IL). Plasma CP-690,550 concentrations were measured using LC/MS/MS techniques. Population plasma levels for each dose were assessed based on one-compartment pharmacokinetics and mean elimination rate estimates. Daily (24 hr) area-under-the-curve (AUC) exposures were subsequently calculated based on the trapezoidal rule. Pharmacodynamic values were assessed on day 21 by calculating the mean paw volume change for each treatment group compared to the mean vehicle control. These values were normalized with the mean vehicle control paw volume change to yield a fractional pharmacodynamic response. The mean fractional pharmacodynamic response values were plotted against their mean pharmacokinetic AUC values and modeled with an Emax non-linear regression model with a maximal pharmacodynamic response constraint of 1 (GraphPad Prism® v. 5.01; GraphPad Software Inc., La Jolla, CA).
Rat Bone Marrow Analysis by Flow Cytometry
Rat bone marrow suspensions were collected, prepared and analyzed as described by Criswell, et al [
20]. Total nucleated cell count (TNCC) was determined using a hematology analyzer (Advia 2120, Siemens, Terrytown, NY). FACS was performed on a Beckman Coulter FC500 equipped with a 15-mW argon-ion laser set at 488 nm and calibrated prior to use. Fluorescent signals were obtained through bandpass filters at 525 nm for the green fluorescence of DCF analysis and at 575 nm for the red fluorescence of PE-tagged lymphocyte analysis. A minimum of 35,000 nucleated cells for DCF analysis and 70,000 nucleated cells for lymphocyte analysis were acquired on each bone marrow sample with listmode date files storing values for forward scatter (FS), FL1, and log integral red fluorescence (FL2). CXP software was used for data analysis of the following bone marrow subpopulations: maturing and proliferating myeloid, maturing and proliferating erythroid, megakaryocytes and lymphocytes. Summary statistics (mean and standard deviation) were calculated for each treatment group for percentage and absolute values, as well as, myeloid:erythroid (M:E) ratios. Student's t-tests were performed to assess statistical differences.
Plasma Cytokine Immunoassays
Cytokines were detected in plasma using Luminex-based rat IL-17 LINCOplex kits (Millipore/LINCO, St. Charles, MO) and Luminex 200 instrumentation (Luminex Corporation, Austin, TX) following the manufacturer's instructions, or Meso Scale Discovery Ultra -Sensitive rat IL-6 kits and Sector Imager 6000 instrumentation (Meso Scale Discovery, Gaithersburg, MD) following the manufacturer's protocol.
Discussion
In a recent analysis of kinase inhibitor selectivity within the human kinome, CP-690,550 was found to be highly selective for the JAK family kinases, and to have similar binding affinities for both JAK2 and JAK3 [
22]. While binding to the kinase domain of JAK1 was not examined in that study, it has been suggested that therapeutic efficacy of CP-690,550 in some indications may be due to cross-over inhibition of other JAKs [
23]. As we have demonstrated in the present study, CP-690,550 is a potent inhibitor of all JAKs in
in vitro kinase assays where truncated kinase domain constructs were used. The K
i and IC
50 were measured for each enzyme, and the relative JAK enzyme rank order potency for CP-690,550 is comparable using either measure of inhibitory potency.
Inhibition of JAK-dependent STAT phosphorylation in whole blood was also examined as a measure of JAK potency and selectivity. The comparable potency observed in human T cells stimulated with several γc cytokines suggests that CP-690,550 is capable of inhibiting multiple STAT pathways equally (IC
50 25-56 nM) within the same cell population. Similar potency was also observed against IL-6 and IFN-α driven STAT1 phosphorylation in T cells, whereas IL-6 activation of STAT3 was less sensitive. Although the gp130 component of the IL-6 receptor complex can associate with JAK1, JAK2 or TYK2, JAK1 plays an essential role in cell signaling, since JAK1-deficient cells fail to respond to IL-6 or other gp130 cytokines [
24]. Thus, while CP-690,550 can inhibit cytokine activation pathways associated with JAK1, subtle differences in JAK or STAT utilization by specific receptors may influence inhibitor potency and selectivity. Further examination of the mechanism behind differential JAK inhibitor sensitivity of IL-6 signaling pathways in T cells will be intriguing. In monocytes, CP-690,550 had reduced activity compared to T cells, yet both type I and type II interferon signaling was inhibited with IC
50 values below 200 nM, while GM-CSF signaling through a receptor that only utilizes JAK2 was significantly less affected. It is unclear mechanistically how CP-690,550 might spare JAK2 in cellular assays, although one explanation for reduced cellular potency could be that a cytoplasmic activator of JAK2, such as SH2-Bβ, can partially counteract the effects of kinase inhibition [
25]. Other possible explanations could be JAK-specific differences in inhibitor potency between isolated kinase domains and full-length enzymes or even against JAK hetero/homodimers. The selectivity profile of CP-690,550 in mouse and rat whole blood was similar to that observed in human, suggesting that in contrast to its activity on isolated JAK kinases, CP-690,550 may have selectivity for JAK1 and JAK3 over JAK2 in whole blood or in cellular assays where JAKs are in their native conformation. If JAK2 sensitivity to CP-690,550 is indeed reduced in cells, our findings suggest that inhibition of the enzyme may not be required to block signaling through receptors for IL-6 and IFN-γ, as inhibition of JAK1 could be sufficient.
The two most likely mechanistic hypotheses which could explain observed clinical reductions in PBNC in RA subjects by CP-690,550 are: 1) suppression of bone marrow myeloid progenitor cell differentiation via a loss of selectivity and inhibition of JAK2; and 2) suppression of chemotactic and hematopoietic growth factors as an extension of the anti-inflammatory activity of CP-690,550.
Several lines of evidence suggest that the first hypothesis is not valid at fully efficacious doses of CP-690,550. First, reductions in PBNC were not observed non-clinically (in rodents (data not shown) and non-human primates [
21] for up to 6 and 9 months of treatment, respectively; although a slight reduction in red blood cell parameters was observed in both species at dose levels where inhibition of JAK2 would be expected (data not shown)) or clinically in healthy volunteers, psoriasis and allograft patients (for up to 14 and 28 days of treatment, respectively) [
9,
10,
21]. Secondly, RA patients in general, and most patients enrolled in the CP-690,550 RA clinical trial, were found to have elevated baseline PBNC, and treatment with CP-690,550 resulted in reductions in PBNC to within the normal human reference range [
7,
8]. Thirdly,
in vitro human selectivity data suggests that clinically efficacious doses of CP-690,550 of up to 30 mg would inhibit the JAK1 and JAK3 enzymes, but not JAK2 (Tables
1 and
2).
It has previously been demonstrated that JAK2 is important in the signal transduction cascade for several hematopoietic growth factor receptors, including granulocyte-colony stimulating factor (G-CSF) and granulocyte macrophage-colony stimulating factor (GM-CSF) [
3,
14‐
16]. It is this established role of JAK2 in regulating granulocyte hematopoiesis that suggests the involvement of JAK2 inhibition in the PBNC reductions in RA patients.
Milici et al previously showed that CP-690,550 treatment in the rat AIA model inhibited arthritis development and bone destruction [
17]. In the present study, we used the rat AIA model and an
in vitro model of human and rodent (mouse) myeloid progenitor cell differentiation to characterize the effects on PBNC observed in RA patients. Additionally, we used pharmacokinetic and pharmacodynamic (PK/PD) modeling to determine the exposure-effect relationships and correlations between JAK inhibition, efficacy and PBNC reductions in both the non-clinical models and in RA patient populations.
Results from this study demonstrate a strong PK/PD relationship between inhibition of JAK1 and JAK3, efficacy, and the inhibition of inflammatory cytokines and neutrophilia in the rat AIA model with CP-690,550 (Tables
1 and
2, Figures
1 and
2). Increases in arthritis, inflammatory cytokines and PBNC in AIA rats correspond to increasing disease severity and progression. CP-690,550 dose-dependently inhibits all three parameters to levels observed in normal animals. Additionally, inhibition of arthritis occurs at a lower dose than the reductions in PBNC (Figure
2F), further suggesting that the reductions in neutrophils are secondary to inhibition of inflammation. Neither generalized myelosuppression nor neutropenia were observed in the rat AIA model, or in normal animals [
21], even at exaggerated clinically efficacious exposures of CP-690,550.
Neutrophilia observed in the rat AIA model was also determined to be, at least in part, due to an increase in granulopoiesis in the bone marrow, and this effect was inhibited at a fully efficacious dose of CP-690,550 (Figure
3). These effects correlated with inhibition of both IL-17 and IL-6 which, in addition to TNF-α, IL-8 and GM-CSF, are pivotal inflammatory cytokines linked to the inflammation, neutrophilia and neutrophil chemotaxis that promote the progression of arthritic disease [
26‐
30].
In
in vitro hematopoiesis assays, CP-690,550 had no effect on human myeloid progenitor cell differentiation at concentrations which fully inhibit cytokine signaling pathways through JAK1 and JAK3 but do not appear to inhibit JAK2 in whole blood (Table
1), and are fully efficacious in rodent arthritis models and in human RA patients (Table
2, Figure
2) [
17]. At exaggerated
in vitro concentrations which exceed selectivity against JAK1 and JAK3 and do inhibit cytokine signaling through JAK2, inhibition of myeloid progenitor cell differentiation was observed (Table
1). Similar results were observed in the mouse, although the IC
50 values were reduced relative to human (Table
1). These results are consistent with previous reports demonstrating a role of JAK2 in G-CSF and GM-CSF signaling and myeloid progenitor cell differentiation [
14‐
16,
26]. Although we did not directly evaluate inhibition of JAK2 or
in vitro granulopoiesis in the rat, inhibition of arthritis, cytokines and neutrophilia in the AIA model were observed at plasma exposures that were well below the human JAK2 IC
50 and effects on GM-CSF stimulated STAT phosphorylation in the rat whole blood assay. Furthermore, we did not observe any inhibitory effects on erythropoiesis in the rat (normal or AIA) over the course of CP-690,550 treatment, which would have been expected if JAK2 was inhibited [
14,
16]. However due to the extended half-life of RBCs such analysis is limited in this model. These findings were consistent with those of Manshouri et al, who demonstrated minimal inhibition of human ex-vivo expanded erythroid progenitors at CP-690,550 concentrations up to 1 μM [
31]. Similarly, in a recent publication by Lin et al., dose-dependent inhibition of erythropoietin-driven reticulocyte formation by CP-690,550 in mice was only observed at doses above that required to fully inhibit IL-2 signaling [
32].
Collectively, results from this investigation suggest that the reductions in PBNC in human RA patients may be an indirect consequence of the anti-inflammatory activity of CP-690,550 and/or the inhibition of JAK1 and JAK3 activity, but not JAK2. However, it is recognized that this compound may have additional pharmacological effects on either granulopoiesis and/or PBNC at exaggerated clinical dose levels where loss of selectivity and cross-inhibition of JAK2 may occur.
Macrophages and lymphocytes have a well established role in the onset and progression of arthritis, but the role of neutrophils has been less clear [
33]. However, several reports looking at the role of neutrophils in both RA patients and in non-clinical models of inflammatory arthritis indicate that these cells are likely to be involved in both the onset and progression of arthritic disease, particularly in the process of joint degradation [
26,
28,
33‐
42]. Neutrophils infiltrating into an arthritic joint can release proteolytic enzymes and reactive oxygen species which can increase inflammation and accelerate the destruction of the bone and cartilage [
43‐
45]. Therefore, it is plausible based upon the present study that the inhibition of neutrophilia by CP-690,550, as observed in both human RA patients and in the rat AIA model, is a desirable and beneficial pharmacological effect of CP-690,550.
Competing interests
All authors were full time employees of Pfizer Inc at the time this work was performed. This study was sponsored by Pfizer Inc.
Authors' contributions
DMM participated in the design and coordination of the studies and writing the manuscript. MME, CLF, JDW, CJG JLH, MJS, and MLB carried out the in vivo and in vitro studies. MIJ carried out some of the in vitro assays and helped write the manuscript. XL helped coordinate the in vitro assays and editing of the manuscript. MED performed the PK analysis/modeling and writing of the manuscript. SKR helped design and coordinate the rat AIA bone marrow assays. NK contributed intellectually to the design the studies and editing of the manuscript. DLM participated in the design and coordination of the studies and helped to draft and edit the manuscript. All authors read and approved the final manuscript.