JAK2-V617F activates β1-integrin-mediated adhesion of granulocytes to vascular cell adhesion molecule 1

Chronic myeloproliferative neoplasia (CMN) represents a group of clonal disorders characterized by excessive proliferation of one or more of the myeloid, erythroid or megakaryocytic cell lineages. CMN includes several subentities such as polycythemia vera (PV), essential thrombocytosis (ET), primary myelofibrosis (PMF) and others. An activating somatic point mutation of the JAK2 gene (JAK2-V617F) was found in the majority (95%) of PV patients and in 50% of ET and PMF patients, respectively.^{1, 2, 3} The clinical course of CMN patients is characterized by increased risk of thrombosis, splenomegaly and an inflammatory response syndrome.^{4, 5} Clinical studies using JAK kinase inhibitors have shown considerable improvement of splenomegaly, constitutive symptoms and overall survival.^{4, 5, 6} JAK2-V617F activates proliferative and anti-apoptotic signaling pathways, thereby driving clonal proliferation of myeloid cells. In addition, JAK2-V617F was reported to activate Lu/BCAM-mediated erythrocyte adhesion through Rap1/Akt signaling in PV, a mechanism that may explain the increased risk of thrombosis in PV patients.⁷ However, in platelets of ET patients, impairment of the PI3 kinase/Rap1/integrin α_IIbβ₃ pathway was demonstrated and was unrelated to the mutation status of JAK2.⁸ In leukocytes of CMN patients, the effect of JAK2-V671F on integrin function and on adhesion is unknown.

Therefore, the aim of this study was to explore whether JAK2-V617F activates β1 integrin-mediated adhesion of granulocytes in CMN. Besides β2 and β3, the β1 integrin chain is expressed on human granulocytes, forming the major heterodimer VLA-4 (very late antigen-4) in combination with α4 integrin.^{9, 10} Integrins play essential roles in leukocyte activation by mediating rolling and firm adhesion to the endothelium, transmigration and trafficking into tissues.^{9, 10} In non-stimulated leukocytes, VLA-4 is expressed in a closed, inactive conformation. Upon external stimuli (for example, chemokines as SDF1α) VLA-4 rapidly undergoes a conformational change, thereby enhancing both the affinity and the avidity for its natural ligand, the VCAM1 molecule.

We first tested adhesion of granulocytes isolated from peripheral blood of JAK2-V617F-positive CMN patients on VCAM1-coated surface. Despite similar levels of β1 integrin expression as compared with healthy controls (Supplementary Figure 1a), primary JAK2-V617F-positive granulocytes showed an overall increase in adhesion to immobilized VCAM1 (Figure 1a). To test for an involvement of JAK2-V617F in activation of integrins, we assessed binding of soluble recombinant VCAM1/Fc (sVCAM1) to the granulocytes. Here the granulocytes of CMN patients demonstrated a significant increase in sVCAM1 binding as compared with healthy donors (Figure 1b, right). As several JAK2-mutated CMN samples showed values comparable to healthy donor controls, we assessed for a potential influence of mutational burden on the phenotype. Here sVCAM1 binding closely correlated with the JAK2-V617F allelic ratio, which is highly variable depending on stage and clinical CMN subtype (Figure 1b, left).¹¹ To further study JAK2-V617F-mediated β1 integrin activation in more detail, 32D myeloid progenitor cells ectopically expressing Epo-R plus either JAK2-WT or JAK2-V617F were generated.^{1, 2, 3, 12} As for the patient granulocytes, 32D JAK2-V617F cells displayed a strong increase in static adhesion to immobilized VCAM1 (Figure 1c). The enhanced adhesion was reversed by inhibition of JAK2 kinase activity (Figure 1c) and not due to altered expression levels of the β1 integrin (CD29) (Supplementary Figure 1b). The adhesion assay employed here was performed using (human-)Fc-tagged and Fc-free VCAM1 in parallel. No differences could be observed, indicating that the human-Fc-tag does not result in unspecific binding on murine cells (data not shown). In the sVCAM1 binding assay, JAK2-V617F led to a sixfold increase in soluble ligand binding as compared with JAK2-WT cells (Supplementary Figure 1c). Pharmacological inhibition of JAK2-V617F downregulated sVCAM1 binding in a time-dependent fashion (Supplementary Figure 1d, and data not shown), without affecting β1 integrin surface expression and cell viability (data not shown).

Figure 1

Peripheral blood was obtained from healthy volunteers and JAK2-V617F-positive patients (PV, ET and PMF) who were untreated with JAK inhibitors after informed consent and upon approval by the local ethics committee (protocol no MD115108). Mononuclear cells were removed by Ficoll-Paque density gradient centrifugation, followed by lysis of erythrocytes with BD FACS Lysing solution (BD Biosciences, Franklin Lake, NJ, USA). (a) Static adhesion assay of primary granulocytes from healthy donors (n=5) and JAK2-V617F-positive patients (n=5) on immobilized VCAM1 (R&D Systems, McKinley, MN, USA, ADP5-050) was performed as described before for ICAM1.¹⁵ (b) sVCAM1 binding assay using soluble VCAM1/Fc (R&D Systems, 862-VC) in primary granulocytes from healthy donors (n=10) and JAK2-V617F-positive patients (n=10) as described previously for ICAM1¹⁵ (right). Correlation of sVCAM1 binding of granulocytes isolated from JAK2-V617F-positive patients with JAK2-V617F allelic burden of peripheral blood cells (left). (c) Static adhesion of 32D JAK2-WT (WT) and JAK2-V617F (V617F) cells on immobilized VCAM1 (R&D Systems, 862-VC) in the absence and presence of JAK inhibitor I treatment (shown are results obtained upon subsequent washing steps II, III and IV) as described before for ICAM1.¹⁵ Cells were treated either with DMSO (−) or with JAK inhibitor I (200 nM) (+) for 16 h. Three independent experiments were performed. *P⩽0.05; **P⩽0.01; ***P⩽0.001 (unpaired, two-tailed Student’s t-test).

Next, we investigated a potential change to the open, high-affinity conformation of β1 integrin chains by using the high-affinity conformation-specific antibody 9EG7. Supplementary Figure 2a shows a significant increase in binding of 9EG7 in 32D JAK2-V617F cells, indicating a change from the bent to the open conformation of the β1 integrin chain.

Considering the potential of JAK2-V617F to induce production of chemokines/cytokines, which in turn may cause increased ligand binding of integrins, we co-cultured 32D JAK2-WT and JAK2-V617F cells. The presence of the mutant had no apparent effect on sVCAM1 binding in JAK2-WT cells, indicating a cell intrinsic effect of JAK2-V617F on integrin activation (Supplementary Figure 2b).

As the small GTPase Rap1 has been reported to play a role in β1 integrin-mediated adhesion,¹³ we employed pull-down experiments of activated Rap1. In 32D JAK2-V617F cells, a strong increase in Rap1 activation was observed (Figure 2a, left), which was suppressed following pharmacological inhibition of JAK2 kinase (Figure 2a, right). Prominent Rap1 activation was also observed in primary JAK2-V617F-positive granulocytes (Figure 2b). Employing the Rap1 inhibitor FTS-A,¹⁴ we observed dose-dependent reduction in adhesion without apparent influence on cell death (Figure 2c, and data not shown). Although shRNA-targeting of Rap1 was only moderately effective in our experiments, the reduction in adhesive capacity correlated with the efficacy of RNAi-mediated inactivation of Rap1 (Supplementary Figures 2c and d). Thus, in granulocytes and 32D myeloid progenitors, Rap1 is activated by JAK2-V617F as described for erythrocytes of PV patients⁷ and may play an important role in JAK2-V617F-activated β1 integrin adhesion.

Figure 2

(a) Rap1 activation in 32D cells expressing either JAK2-WT or JAK2-V617F detected by pull-down of GTP-bound Rap1 as indicated by the manufacturer (Thermo Scientific, Waltham, MA, USA) (right). Inhibition of Rap1 activation by JAK inhibitor I treatment (0.5 μM; 4 h) in 32D JAK2-V617F cells; (−) indicates DMSO control (left). Lower panels show quantitative analysis of Rap1 activation. (b) Rap1 activation in primary human granulocytes isolated from JAK2-V617F-positive patients. Immunoblots show two independent cases of increased Rap1 activation in primary granulocytes isolated from healthy donors and JAK2-V617F-positive patients, respectively. (c) Effect of Rap1 inhibitor FTS-A treatment (16 h) on static adhesion of 32D JAK2-V617F cells on immobilized VCAM1; (−) indicates DMSO control. Three independent experiments were performed. *P⩽0.05; **P⩽0.01; ***P⩽0.001 (unpaired, two-tailed Student’s t-test).

Together, our findings indicate a novel role for JAK2-V617F in activation of β1 integrins and enhanced adhesion of granulocytes and 32D myeloid progenitors to VCAM1-coated surfaces. As VCAM1 is abundantly expressed on endothelial cells, this newly identified characteristic may play a critical role in abnormal interaction of granulocytes with the endothelium in JAK2-V617F-positive CMN.