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 . 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 . 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 Ki and IC50 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 (IC50 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 . 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 IC50 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 . 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  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 . 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 , 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) . 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 IC50 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 IC50 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 . 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 .
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 . 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.