Signaling and functional competency of neutrophils derived from bone‐marrow cells expressing the ER‐HOXB8 oncoprotein

Neutrophils play a central role in immunity and inflammation via their intrinsic ability to migrate into inflamed tissue, to phagocytose pathogens, and to kill bacterial and fungi by releasing large quantities of superoxide anions and lytic enzymes. The molecular pathways controlling neutrophil microbicidal functions are still unclear, because neutrophils have a short half‐life and are resistant to genetic manipulation. Neutrophil‐like cells (NLC) can be generated from myeloid progenitors conditionally immortalized with the ER‐HoxB8 oncoprotein, but whether these cells can replace neutrophils in high‐throughput functional assays is unclear. Here, we assess the ability of NLC derived from ER‐HoxB8 progenitors to produce ROS and to perform chemotaxis and phagocytosis. We compare the Ca2+ responses and effector functions of NLC to primary murine neutrophils and document the molecular basis of their functional differences by mRNA profiling. Pro‐inflammatory cytokines enhanced the expression by NLC of neutrophil surface markers and transcription factors. Ca2+ elevations evoked in NLC by agonists, adhesion receptors, and store depletion resembled the physiological responses recorded in primary neutrophils, but NLC expressed reduced amounts of Ca2+ signaling proteins and of chemotactic receptors. Unlike their myeloid progenitors, NLC produced H2O2 when adhered to fibronectin, migrated toward chemotactic peptides, phagocytosed opsonized particles, and generated intracellular ROS. NLC phagocytosed as efficiently as primary neutrophils but produced 50 times less ROS and migrated less efficiently toward chemoattractant. Our data indicate that NLC can replace neutrophils to study Ca2+ signaling and phagocytosis, but that their incomplete granulocytic differentiation limits their use for chemotaxis and ROS production assays.

enhanced by cytokines, chemokines, or microbial products. 5,6 During infection, the circulating cells are recruited from the blood stream to inflamed tissue following chemotactic gradients, a multistep process that is finely regulated. [7][8][9] Elevations in cytosolic Ca 2+ concentration regulate neutrophil adhesion, extravasation, migration, and phagocytosis, as well as several of their antimicrobial functions. [10][11][12] Signal transduction from integrin, chemotactic, and phagocytic surface receptors involve Ca 2+ elevations mediated by phospholipase C that are sustained by receptor-operated or by store-operated Ca 2+ entry (SOCE) channels.
TRPC, TRPM, and TRPV Ca 2+ channels have been implicated in the control of neutrophil functions [13][14][15][16] and the role of the 3 ORAI1-3 SOCE channel isoforms and of their 2 regulatory proteins STIM1 and STIM2 has been clarified recently. [16][17][18] Novel functions of neutrophils in immunity and inflammation are continuously unravelled, 19 but several fundamental questions regarding the biology of these versatile cells remain to be clarified. Progress is slow because neutrophils are terminally differentiated cells that cannot be expanded in vitro, have a short life-cycle preventing long-term assays, and are largely resistant to transfection or transduction methods of genetic manipulation. Consequently, neutrophil functions are usually explored in animal models or in cell lines using a combination of genetic and pharmacological approaches. The variability of approaches used to study these complex cells greatly complicate the interpretation and comparison of results emanating from different studies. 20 Current research practices favor the use of cell lines to comply with the 3R guidelines aiming to reduce, replace, and refine animal experimentation. 21 The available cell lines HL-60 or PLB-985, derived from myeloma, accumulate further mutations in long-term cultures and require chemicals to induce their differentiation into neutrophillike cells. In this study we explore the potential of murine myeloid progenitors conditionally immortalized with the homeobox oncoprotein HoxB8 driven by the estrogen receptor (ER-HoxB8), originally designed by Wang et al. 22 These cells hold several advantages: (i) progenitor cells can be isolated from transgenic mouse strains and easily immortalized by the introduction of ER-HoxB8; (ii) The cells are from a non-cancerous background and recapitulate the physiological neutrophil differentiation program; (iii) the progenitors can be efficiently transduced using lentiviral vectors to provide a genetically tractable neutrophil model.
The HoxB8 system was recently used to establish the role of the non-muscle myosin Myh9 in neutrophil trafficking in vitro 23 and to monitor neutrophil trafficking during inflammation in vivo. 24,25 These studies validate the use of this cellular model to study neutrophil trafficking in vitro and in vivo and highlight its advantages for rescue experiments or adoptive transfer of cells bearing mutations resulting in lethal phenotypes in mice. However, whether sufficient numbers of neutrophil-like cells can be generated for high-throughput approaches and whether ER-HoxB8-derived neutrophils faithfully recapitulate the neutrophil Ca 2+ signaling toolkit is unknown. In this study, we compare neutrophil-like cells (NLC) immortalized from ER-HoxB8 progenitors to primary murine neutrophils. We evaluate the efficiency of chemotaxis, ROS production, and phagocytosis in these neutrophilic cells and their potential in high-throughput screens to evaluate regulators of store-operated Ca 2+ entry channels in neutrophils.

Mice, cells, and cell lines
Primary murine neutrophils were isolated from bone marrow col- Thomas Kaufmann (University of Bern, Switzerland).

Myeloid cell line culture and differentiation
ER-HoxB8 progenitor cells (ER-HoxB8) were originally isolated from healthy mice (C57/BL6) and the line was kindly provided by Prof. Cells were harvested at day 6 and used the same day.

May-Grünwald-Giesma staining
Cells for staining were isolated or cultured as described above, washed once, and seeded on polylysin-coated (0.1%) cover slips (CS, 12 mm) and allowed to adhere. Cells were air-dried and fixed with methanol (100%), washed once with PBS, and immediately stained.

Three hundred eighty-four-well plate format
The dose-dependency of SOCE inhibition by 2-APB and GSK-7975A was performed with Calcium6 dye (#R8194, Molecular devices) on a Hamamatsu FDSSµCell Imager (Hamamatsu Photonics K.K., Japan). Cells were loaded as described above, washed, and re-suspended in 1.5 mM Ca 2+ containing buffer and seeded on endothelial-coated microplates at 20,000 cells/well. Baseline Ca 2+ levels were recorded for 2 min, inhibitors added for 9 min before Tg addition, and SOCE recorded for 9 min. To calculate IC 50 values, the fluorescence responses (F/F0) obtained in the absence of inhibitors were set to 100% (maximal response) and the response obtained at a maximal dose of inhibitor were set to 0% (minimal response).  Primer sequences used for qRT-PCRs are listed in Table 1.

Cytokines promote the differentiation of ER-HoxB8 myeloid progenitors into neutrophil-like cells
To assess the maturation state of neutrophil-like cells (NLC) derived from ER-HoxB8-immortalized murine hematopoietic cells, we mea-  the percentage of Gr-1 positive cells to ∼90%, compared to 5% in progenitors and 99% in primary murine neutrophils (Fig 1A and E).  Summary of expression data obtained by qRT-PCR. Samples were collected each day over a time course of 6 days during the in vitro differentiation of immortalized myeloid progenitors Hoxb8-SCF following no treatment (Hoxb8), the adapted protocol using additional cytokines used for the presented study (NLCs (+Cytokines) and the published differentiation protocol by Wang et al. (NLCs (standard protocol, SP)), Primary Neutrophils (day 1 = day of isolation from murine bone marrow and sample preparation). Differentiation protocols and primer sequences for qRT-PCR can be found in the materials and methods section of the manuscript.

Calcium signals evoked in ER-HoxB8-derived neutrophil-like cells
The expression of the fMLF receptor was low in progenitors and increased in NLC and neutrophils ( Fig. 3F; Supplementary Fig. S3 (Fig. 3H). These data validate the use of NLC as neutrophil surrogates for high-throughput drug screening assays and indicate that the SOCE channels expressed in NLC are inhibited by 2-APB and relatively resistant to GSK-7975A.

ER-HoxB8-derived neutrophil-like cells recapitulate some, but not all, neutrophil effector functions
We next assessed whether NLC express a phagocytic NADPH oxidase (NOX2) able to produce the large amounts of ROS required for efficient neutrophils extravasation and pathogen clearance. 36,37 Adhesion of neutrophils to the matrix glycoprotein fibronectin activates NOX2-dependent ROS production, 38,39 an effect that is suppressed at high fibronectin concentrations due to the engagement of V 3 -integrins. 40,41 When adhered to plates coated with low concentrations of fibronectin (0.3 µg/ml), NLC, but not progenitors, produced significant amounts of extracellular H 2 O 2 reported by AmplexUltra Red fluorescence (Fig 4A and B). Adhesion-mediated H 2 O 2 production was abrogated by the NADPH-oxidase blocker DPI and reduced by extracellular Ca 2+ chelation and by the Ca 2+ channel inhibitor 2-APB ( Fig. 4A and B). µg/ml (Fig. 4C, white bars). A similar dose-dependent H 2 O 2 production was measured in neutrophils while ROS were undetectable in progenitors at all fibronectin concentrations (Fig. 4C, black and gray bars).
Importantly, the rates of H 2 O 2 production were 55-fold lower in NLC compared to neutrophils (2.8 × 10 3 vs. 1.59 × 10 5 at 1 µg/ml, Fig. 4C, white and gray bars, note different axis scales). Adhesion thereby activates NOX2-dependent ROS production in NLC, an effect partly dependent on the entry of Ca 2+ across membrane channels, but the rates of ROS production are 1.5 order of magnitude lower than those of neutrophils.
To clarify the molecular basis of the adhesion-mediated ROS responses, we measured the expression of genes coding for the adhesion receptor Mac1 (Itgam/Itgb2, M 2 integrin CD11b/CD18) and for the cytosolic and membrane NOX2 subunits p47 phox (Ncf1) and gp91 phox (Cybb). Cybb expression increased by 50-fold in NLC compared to progenitors while Itgam, Itgb2, and Ncf1 expression was conserved, all transcripts being 4-20 less abundant in NLC than in neutrophils ( Fig. 4D; Supplementary Fig. S3). To establish the presence of a functional NOX2, we assessed gp91 phox protein expression by western blot. A prominent ∼60 kD band was detected by the anti-gp91 phox antibody in WT neutrophils that was absent in neutrophils from gp91 phox knock-out mice (Fig. S2D). This band was undetectable in ER-HOXB8derived progenitors and barely detectable in NLC. Quantification of the gels indicated that NLC express ∼20 times less gp91 phox than primary neutrophils. The ability of NLC to produce ROS thus correlates with an increased expression of the rate-limiting membrane flavocytochrome gp91 phox , but this expression remains at low levels, likely limiting the NOX2 activity of NLC.
We further examined the ability of NLC to migrate toward chemoat- We next assessed the capacity of NLC to phagocytose opsonized fluorescent yeast particles by flow cytometry. Zymosan particles, added at a 1:5 cell to target ratio for 30 min, were taken up 3-4 times more efficiently by NLC and neutrophils than by progenitors (14 ± 0.6 vs. 35 ± 2, P = 0.0001 and 46 ± 6%, P < 0.01 for progenitors, NLC, and neutrophils, respectively, Fig 4E, # ). Ca 2+ removal decreased particle uptake (to 9.7 ± 0.4, 17 ± 1, and 23 ± 1%, Fig. 4F, *) with NLC and neutrophils retaining a significantly higher phagocytic index (P < 0.001 and P < 0.0001 for NLC and neutrophils vs. progenitors, respectively, Fig. 4F). An EDTA-containing buffer was used to favor the detachment of non-internalized beads and particle internalization was verified microscopically. The increased phagocytic ability of NLC correlated with an increased production of intracellular ROS, Data are from at least n = 2 independent experiments for primary neutrophils and from at least n = 3 for progenitors and NLC, * P < 0.05; ** P < 0.01; ***P < 0.001; and **** P < 0.0001, unpaired t-test with Welch's correction, with * , # , and $ differentiating the compared conditions for statistical analysis as indicated in the main text

DISCUSSION
In this study, we characterize the molecular and cellular basis of the  24 We confirm and extend these findings by comparing the effector functions of HoxB8-derived neutrophils and of primary neutrophils using high-throughput, quantitative assays and a range of activating conditions.
In our initial validation experiments, we modified the original culture protocol by adding pro-inflammatory cytokines, 40  From a Ca 2+ signaling standpoint, we show here that the responses evoked in neutrophils by chemotactic agents, adhesion receptors, and store depletion were recapitulated in NLC. In contrast, progenitors failed to respond to agonists and exhibited significantly larger SOCE rates than both NLC and neutrophils (Fig. 3). The high number of cells generated enabled us to test a whole range of concentrations of stimuli ( Fig. 3C; Supplementary Fig. S2A and B) and of inhibitors ( Fig. 3G  were undetectable in ER-HoxB8 progenitors, indicating that NLC had acquired a granulocytic phenotype. Although NLC qualitatively recapitulated these 3 effector functions, we observed important quantitative differences with neutrophils. NLC produced substantially less ROS than bone marrow-derived murine neutrophils and migrated less efficiently toward chemoattractants. In contrast, NLC took up zymosan particle as efficiently as neutrophils. NLC can therefore replace neutrophils to study phagocytosis but only provide qualitative information to study chemotaxis and ROS production. We could relate the signaling and functional competency of NLC to the transcription of proteins involved in chemotactic sensing, ROS production, and phagocytosis. NLC chemotaxis correlated with an increased expression of the chemotactic receptor Fpr1, their H 2 O 2 production with increased expression of the rate-limiting NOX2 subunit Cybb, and their ability to perform phagocytosis with a decreased expression of the inhibitory immunoglobulin receptor FcgR2b. The expression levels of Fpr1 and Cybb, however, were still considerably lower in NLC than in neutrophils, as were the levels of the gp91 phox protein component, consistent with the higher respiratory burst and chemotactic capacity of primary cells. This was also the case for the granule marker proteins Lactoferrin and MMP9, whose transcripts were much higher in neutrophils than in NLC (Fig. 4) and whose protein levels were 3-fold lower, prompting us to forego degranulation experiments. Our expression profiling data therefore indicate that genes conferring chemotactic, phagocytic, and oxidative capacity to granulocytes are expressed in NLC, but at lower levels than in neutrophils ( Supplementary Fig. S3), consistent with the quantitative functional differences observed between NLC and primary cells. On the other hand, our observation that NLC phagocytose zymosan particles as efficiently as neutrophils and that this ability correlates with a decreased expression of the low-affinity, inhibitory receptor rather than with an increased expression of the activating immunoglobulin receptors make this cell line a good model to study the impact of Fcy isoform expression on phagocytosis.
Complete reconstitution of neutrophil effector functions with the ER-HoxB8 system might require culture conditions mimicking processes occurring in the bone marrow niche. 48 Indeed, our observation that adhesion to fibronectin or to endothelial cells modulate NLC Ca 2+ and ROS responses show that these cells quickly adapt to their microenvironment. Despite its current limitations, the ER-HoxB8 system has potential as an alternative model to animal experimentation, since cell lines could be derived from knockout mouse models or genetically engineered to provide sufficient material for large-scale experiments. Using CRISPR-Cas9 approaches, genetically modified HoxB8 cell lines were generated to visualize Ca 2+ signals during dendritic cells migration 49 or to control their differentiation into osteoclasts. 50 In conclusion, we show that NLCs derived from murine bonemarrow cells conditionally immortalized with the ER-HoxB8 oncoprotein recapitulate three major effector functions of neutrophils in vitro.
We validate the use of NLC in high-throughput fluorescence screens and document the molecular basis of their signaling, chemotactic, phagocytic, and oxidative capacity. Our data indicate that NLC can replace neutrophils to study phagocytosis, chemotaxis, and ROS production, although their incomplete differentiation imparts important molecular and functional differences with primary neutrophils.

AUTHORSHIP
S.S. designed the study, performed experiments, analyzed data, and wrote the manuscript. C.C. and C.F. performed experiments, handled mice, and performed primary cell isolation. N.D. designed the study, supervised experiments, and wrote the manuscript.