Defective migration and dysmorphology of neutrophil granulocytes in atypical chronic myeloid leukemia treated with ruxolitinib

Migration assays of neutrophil granulocytes, referred to as neutrophils, bear strong potential as valuable diagnostic and surveillance tools. As neutrophils are among the first cells to counter infections [1], altered neutrophil migration patterns were observed during acute and chronic conditions, such as sepsis [2, 3], asthma [4], chronic inflammatory bowel disease [5] or multiple sclerosis [6]. When recruited into inflamed tissue, neutrophils do rely on their ability to autonomously migrate towards the area of infection. Chemotactic stimuli attracting neutrophils are e.g. secreted pathogenic peptides containing formyl-methionine (fM) or CXCL cytokines, released by patrolling immune cells or endothelial cells in contact with the pathogens [7, 8]. Therefore, neutrophil migration is rapidly altered in response to ongoing inflammation. Additionally, neutrophil migration is altered in cancer. Reports showed altered neutrophil migration in human head and neck cancer as well as small lung cell carcinoma [9‐11]. These findings were corroborated in a number of murine cancer models revealing increased neutrophil motility [12]. However, despite the large body of evidence for the importance of motility for neutrophil function and its modification in disease states, there have been limited attempts to exploit this knowledge for the diagnosis of human diseases.

In this regard, we recently demonstrated that neutrophil migration strongly correlates with the revised international prognosis scoring system (IPSS-R) in myelodysplastic syndrome [13], representing a neoplasia affecting neutrophil functions such as degranulation and phagocytosis [14]. Here, neutrophils from severe MDS cases, with a high risk of blast transformation, displayed significantly lower migration speed than lower-risk MDS cases or neutrophils from healthy donors. Furthermore, the recovery of normal migration patterns during therapy correlated with a successful response to the treatment, pointing at the analysis of migration as a potential monitoring tool for therapy [13].

Atypical chronic myeloid leukemia (aCML) is a rare subtype of myelodysplastic / myeloproliferative neoplasms (MDS/MPN) characterized by poor prognosis and lack of standardized treatment algorithms [15, 16]. aCML mostly manifests in elderly patients with male predominance [17]. It is characterized by elevated white blood cell (WBC) counts, mainly due to increased granulocyte numbers, splenomegaly and severe dysgranulopoiesis with abnormal chromatin clumping [17, 18]. As all MDS/MPN subtypes, aCML lacks distinct genetic alterations facilitating the diagnosis [17]. However, aCML cases have been reported to be associated with mutations of spliceosome proteins, e.g. in U2AF1, of epigenetic modifiers, e.g. in ASXL1, TET2, EZH2, and of signaling molecules, e.g. in NRAS, KRAS, JAK2, CSF3R [19]. Especially the role of mutations in the CSF3R gene, coding for the G-CSF receptor, in aCML were controversially discussed as CSF3R mutations are defining mutations in the diagnosis of chronic neutrophilic leukemia (CNL) [20, 21]. However, studies also reported CSF3R mutations in aCML [22]. Hence, CSF3R mutations represent one of the overlapping features between CNL and aCML [19, 23]. Most CSF3R mutations found in human leukemias involve either truncations or membrane proximal mutations [22]. The CSF3R p.T618I mutation diagnosed in the case of the present study is a membrane proximal mutation causing ligand-independent activation of the down-stream signaling JAK/STAT pathway involving predominantly JAK1/2 in turn leading to unchecked neutrophil proliferation [23]. In this circumstance, the JAK1/2 inhibitor ruxolitinib has been reported as a potential effective CNL therapeutic option [24, 25]. In light of these promising reports, ruxolitinib treatment of CNL and aCML patients with CSF3R mutations is currently investigated within clinical trials (NCT02092324).

As the diagnosis of aCML is a complex endeavor and criteria for the monitoring of aCML therapy have not been established, we investigated the applicability of our novel migration assay in this disease setting.

Here, we analyzed neutrophil migration of an aCML patient in a longitudinal manner. Neutrophils of the present aCML case displayed severely reduced migration compared to healthy controls. Upon treatment with ruxolitinib, neutrophil migration remained at a low level, even though blood parameters and clinical presentation of the patient improved. Interestingly, before initiating ruxolitinib treatment, aCML neutrophils had a flattened morphology steadily normalizing upon treatment with ruxolitinib.

Our findings suggest that analyses of neutrophil migration and morphology might be a valuable diagnostic / monitoring tool for myeloid neoplasms in general. Hence, we conclude that neutrophil migration analyses may be suitable to monitor a spectrum of hematological diseases and should possibly be part of future diagnostic workup strategies and therapy monitoring.

Neutrophil migration is a promising novel functional parameter to identify states of disease in humans. To elucidate the applicability of our neutrophil migration assay [13] and to determine whether additional migration or morphological parameters in neutrophil migration assays are useful, we studied a single patient suffering from the rare neoplasia aCML.

We have shown that migration of aCML neutrophils remained diminished over the whole observation period. Furthermore, the expression of the low affinity IgG receptor FcγRIII, CD16, and carbohydrate antigen, CD15, which are normally strongly expressed by human mature neutrophils, remained below the levels of matched adult controls. CD16 is normally upregulated with increasing neutrophil maturation [35], yet this did coincide with the amount of immature neutrophils in the circulation of the patient. Furthermore, CD15 expression, which is also regulated during neutrophil maturation [36], was largely absent over the complete observation period and did also not correlate with peripheral immature neutrophil levels. Additionally, we observed a severe reduction of the key chemokine receptors fMLPR, CXCR1 and CXCR2 expression levels. This is remarkable as ruxolitinib therapy caused loss of the CSF3R (p.T618I) clone and was successful in reducing both WBC, peripheral neutrophil precursors and other disease manifestations, like night sweats, weight loss and fatigue. In MDS, however, successful treatment was indicated by a normalization of neutrophil migration. Thus, it is conceivable that the migration defect is the result of multiple mutations, not exclusively CSF3R (p.T618I). It is therefore important to consider that the aCML patient presented here harbored a large number of different mutations, which are typically observed in myeloid neoplasms. CCAAT/enhancer binding protein α (CEBPα) is a transcription factor important during granulocyte differentiation. Mutations in this protein generally cause a diminished activity by preventing DNA binding or downstream translation and reduced activity of CEBPα results in differentiation arrest of granulocytes and hyperproliferation of hematopoietic stem cells (HSCs) [37, 38]. The NRAS (G12D) mutation is one candidate that might directly influence neutrophil migration. Ras proteins are proto-oncogenes that are critical for the signal transduction from cell-surface receptors into inner machinery, thereby controlling cell proliferation, differentiation or cell death [39]. NRAS mutations are frequent in human myeloid leukemias and other cancers and the G12D mutation has been described to drive development of chronic MPN in mice [40]. Indeed, oncogenic NRAS is involved in heightened migration in melanoma cell lines and inhibition of NRAS by microRNAs was successful in reducing trans-well migration [41]. There is evidence that a delicate interplay between ERK and p38 MAPK is needed to regulate directed migration [42]. We have recently reported, that over-phosphorylation of the MAPK p38 upon high fMLP stimulation for 1 h correlated with reduced random migration speed in human neutrophils [13]. This might hint at different mechanisms to induce random versus directed migration or variations due to different analysis timepoint. The NRAS (G12D) mutation in human neutrophils might therefore cause defects in random migration by inducing the hyperphosphorylation of downstream targets.

Further studies are needed to determine whether random and directed migration are in fact differentially affected by NRAS (G12D) and whether there are specific mutations that affect neutrophil migration to distinguish MDS or aCML. Additionally, a study recently reported that ruxolitinib itself showed impairing effects on dendritic cell migration by inhibition of Rho-associated coiled-coil kinase (ROCK) [43]. While we cannot rule out that ruxolitinib therapy was partially responsible for the diminished neutrophil migration reported here, we found low migration speed and recruitment before onset of the therapy as well (Fig. 1e), hinting at a cell intrinsic mechanism underlying this phenomenon. Furthermore, especially the relative number of moving neutrophils upon fMLP, CXCL1 and PBS stimulated conditions increased with therapy start. Neutrophil speed however remained heavily impaired.

An interesting difference between the aCML and the MDS samples, was neutrophil morphology. In aCML the altered morphology normalized in concordance with the therapeutic success of ruxolitinib treatment. aCML neutrophils were significantly enlarged compared to healthy controls before therapy but whether this was the result of an enlarged cytoplasm or increased adherence remains unclear. However, neutrophil size did not correlate with the assessed peripheral blood parameters, like myeloblast and metamyelocyte counts (data not shown), ruling out the possibility that neutrophil progenitors in the assay were responsible for these changes. Additionally, neutrophil morphology normalized to healthy levels already 1 week after the initiation of ruxolitinib therapy. In fact, it has been reported that ruxolitinib causes microtubule instability in JAK (V617F) mutant HEL cells by inhibiting JAK2 and STAT3 activity [44] and might thereby change their morphology. Furthermore, it is conceivable that the CSF3R (T518I) mutant was responsible for the abnormal shape or adherence of neutrophil in our assay, which then normalized by clonal depletion as assessed by NGS (Fig. 4a).

Strikingly, we observed severely reduced expression levels of CD15 and CD16, as well as of the corresponding receptors to the chemokines used in this assay, fMLPR, CXCR1 and CXCR2. While we found no direct correlation between the levels of receptor expression and neutrophil migration behavior in healthy donors [13], it is conceivable that especially the severely impaired migration upon CXCL8 treatment might have been caused by the absence of its signaling receptor CXCR1 [45]. However, CXCL8 can also signal via CXCR2 [45], whose expression levels verged on normal over the course of disease but did not impact neutrophil migration when triggered with CXCL8. As CXCL8 is a key molecule required for the recruitment of neutrophils and their successful extravasation from the blood vessel system [46], this disrupted response to CXCL8 might explain, why the patient remained susceptible to bacterial and fungal infections throughout therapy.

With the case presented here, we provide evidence that the routine assessment of neutrophil migration and receptor expression provides a broader perspective on diseased neutrophils and impact of a treatment on the patient’s cells. We found compelling evidence that cell shape and degree of adherence were changed over the course of ruxolitinib treatment and coincided with the disappearance of specific clones. Furthermore, we noted an increase in neutrophil speed on day 67 to 12.5 μm/min upon fMLP stimulation (Fig. 4b), which corresponded to the bleeding of a gastric ulcer 6 days later, when the patient was admitted to hospital again. We therefore believe that assessment of neutrophil migration might also facilitate the surveillance of patients with a higher risk for infections. Interestingly, our data also demonstrate that aCML neutrophils were indeed still able to overcome their unresponsive state, but seemed to require additional, host-derived activations to migrate in an in vitro assay.