Chemotaxis and swarming in differentiated HL-60 neutrophil-like cells

The human leukemia cell line (HL-60) is an alternative to primary neutrophils in research studies. However, because HL-60 cells proliferate in an incompletely differentiated state, they must undergo differentiation before they acquire the functional properties of neutrophils. Here we provide evidence of swarming and chemotaxis in differentiated HL-60 neutrophil-like cells (dHL-60) using precise microfluidic assays. We found that dimethyl sulfoxide differentiated HL-60 cells (DdHL-60) have a larger size, increased length, and lower ability to squeeze through narrow channels compared to primary neutrophils. They migrate through tapered microfluidic channels slower than primary neutrophils, but faster than HL-60s differentiated by other protocols, e.g., using all-trans retinoic acid. We found that dHL-60 can swarm toward zymosan particle clusters, though they display disorganized migratory patterns and produce swarms of smaller size compared to primary neutrophils.

The histograms of the critical CS show that a larger percentage of A, O, R patterns in DdHL-60 cells occur at cross-sections smaller than 10 µm 2 ( Fig. 3bi-iv), similar to what was observed in primary neutrophils 12 . The histograms of the critical CS show that A, O, R patterns in AdHL-60 cells are also less dispersed ( Fig. 3ci-iv). In summary, our data demonstrate that mechanical constriction in the tapered channel during chemotaxis interfere with the migratory response of dHL-60 to chemokine gradients more than with the migratory responses of human neutrophils. dHL-60 and primary neutrophils display distinct morphologies during chemotaxis through confined microfluidic channels. Our data shows that dHL-60 confined in microfluidic channels during chemotaxis are larger and longer than primary neutrophils. We observed that migrating dHL-60 display a broad leading edge toward the chemokine and a rounded tail end (Fig. 4a). When the cells advance through the tapered channels towards the smaller cross sections, their length gradually increases. The length of DdHL-60 cells, nDdHL-60 cells and primary neutrophils increases from ~ 38 to ~ 78 µm, ~ 30 to ~ 65 µm and from ~ 25 to ~ 65 µm respectively when migrating through the tapered micro-channel toward the chemokine (Fig. 4b).
Reduced directionality of dHL-60 during swarming compared to primary neutrophils. Next, we compared the abilities of DdHL-60, nDdHL-60, and human neutrophils to swarm toward zymosan particle clusters. Three distinct phases of swarming have been previously reported 9 . Swarming starts with random migration on the surface (scouting phase-5 min Fig. 5a). After the first neutrophil interacts with the cluster, the number of migrating neutrophils towards the zymosan cluster increases rapidly (growing phase-10-15 min Fig. 5a). Swarms reach their peak size 60-120 min later, after which the size remains stable (stabilization phase- Fig. 5a) (Supplementary video S2). We observed qualitatively similar aggregation dynamics in both DdHL-60 and nDdHL-60 cells. Aggregation starts with the cells moving randomly on the surface of the zymosan particle clusters (scouting phase-5 min Fig. 5a). Within minutes of interaction with the zymosan particle, an increasing number of dHL-60 cells migrate in the direction of the zymosan particle clusters (growing phase-10-30 min Fig. 5a). A period of fast growth is followed by slower growth. Swarms in nDdHL-60 cells reach their peak size after 60-120 min, after which the size of the aggregates decreases because cells leave the aggregate (Fig. 5a). During the stabilization phase, dHL-60 cells migrate in and out of the swarms (Fig. 5a) (Supplementary video S3 and S4).

Surface expression of LTB4 Receptor 1 and LTB4 secretion by dHL-60 cells. Having shown that
DdHL-60 cells exhibit swarm like behavior that is qualitatively similar to primary neutrophils but quantitatively distinct, we sought to determine whether the observed differences in DdHL-60 and primary cells was due to expression or secretion of LTB4 and its receptor. To test this, we checked for the expression of LTB4-R1 on viable DdHL-60, nDdHL-60, and primary neutrophils. We found that 73%, 96.2% and 98% of viable DdHL-60, nDdHL-60 and primary neutrophils express LTB4-R1, respectively (Fig. 7a,b). Next, we quantified the amount of LTB4 released during swarming in vitro using enzyme-linked immunosorbent assay (ELISA). For this, we incubated DdHL-60, nDdHL-60, or primary neutrophils with salmonella typhimurium for 4 h, then quantified the amount of LTB4 released into the supernatant. We found that the amount of LTB4 released by primary neutrophils was approximately twice the amount released by dHL-60 cells (Fig. 7c).
Blocking LTB-4 receptors alters DdHL-60 swarm-like behavior. To verify that the mechanisms involved in the aggregation of HL-60 cells are similar to swarming in primary neutrophils, we tested the role Here, we compared the aggregation of DdHL-60 cells in the presence of BLT1 and BLT2 receptors antagonists and a LTB4 synthesis inhibitor. We found that in the presence of BLT1 and BLT2 receptor antagonists LY255283, U75302 there was significant delay in the initiation and a reduction in the swarm size (Fig. 8a,b) (Supplementary video S6 and S7). Inhibition of the LTB4 pathway by the MK886 inhibitor alters mainly the final size, with little impact on the initiation, suggesting the release of LTB4 during the early stage of swarming is from pre-formed vesicles (Fig. 8a,b) (Supplementary video S6 and S8). The swarm area around zymosan clusters during the stabilization phase was ~ 2 times larger in control compared to drug treated cells (~ 25,000 μm 2 vs. 17,000 μm 2 vs. 15,000 μm 2 , for control vs. BLT1&2 vs. MK-886 treated cells) (Fig. 8b).

Discussion
Neutrophil swarming has been described as a crucial process of neutrophil tissue response needed to specifically regulate tissue protection and destruction during several inflammatory diseases 14 . Here we report that dHL-60 s are capable of swarming, qualitatively similar to primary neutrophils. The dHL-60 s display all three phases of swarming behavior (scouting, amplification, and stabilization phases) 8,9 . However, despite the qualitative www.nature.com/scientificreports/ similarities, significant differences also exist. The swarm size for similar targets is smaller and the trajectories of cells joining the swarms appear more disorganized for differentiated HL-60 cells compared to primary neutrophils. The differences in chemotaxis (migration fraction, migratory velocity, and directionality) between DdHL-60, nDdHL-60 and primary neutrophils may contribute to the differences observed in the swarming assay. A lower percentage of DdHL-60 cells migrate towards chemoattractants compared to primary neutrophils and which may be responsible for the observed smaller swarm size in DdHL-60. Furthermore, our data show that nDdHL-60 cells display improved percentage of migration comparable to primary neutrophils, however, with more arrest and retro taxis migratory patterns, which may be the cause of the smaller swarm size observed in nDdHL-60 cells. Moreover, our study also demonstrates that both DdHL-60 and ndDHL-60 cells migrate slower than primary neutrophils 9,10,16 .
Importantly, the third phase of swarming is characterized by the recruitment of distant neutrophils to an infection site, driven by expression of the high-affinity receptor for LTB4 (LTB4R1) on neutrophils 9 . Our data demonstrate that the observed swarm-like behavior in DdHL-60 and nDdHL-60 cells is LTB4 dependent, similar Figure 6. Quantification of primary neutrophil and HL-60 swarming. (a) Chemotactic index (CI) and speed over time, towards zymosan particle clusters in (i) primary neutrophils, (ii) DdHL-60, (iii) nDdHL-60. (b) The speed of migration increases with time towards zymosan particle clusters (scale bar: CI above 0.8 indicates the cells that are chemotaxing toward zymosan particle. speed (µm/min) = 4.5 means cell moving toward zymosan particle). (c) Primary neutrophils (blue curve) accumulation on targets is fast and form a bigger swarm and dHL-60 neutrophil (orange curve) aggregation proceeds fast and then continues slower over time. (d) Comparison of area of swarm formed between primary neutrophils (PN) and dHL-60 (HL-60). Comparison were performed by unpaired, two-tailed t-test; *P < 0.05). Error bars represent standard deviations for these measurements. Figure 5 shows representative image of experiment of N = 30 swarm spots. www.nature.com/scientificreports/  www.nature.com/scientificreports/ to the findings in primary neutrophils 8 . Though, our data demonstrates that both DdHL-60 and nDdHL-60 cells expresses comparable level of LTB4-R1 to primary neutrophils, but they both release lower amount of LTB4 compared to primary neutrophils during swarming. Since LTB4 release is crucial to the process of swarming in neutrophils, an observed impairment in the release of LTB4 may be the cause of smaller swarm size in DdHL-60 and nDdHL-60 cells.
The establishment of HL-60 cells as a viable model system for neutrophil swarming would have great potential value to the study of swarming. Previously, neutrophil swarming has been demonstrated in zebrafish larvae 15 and mouse tissues 16 . Human neutrophil swarming has also been observed in vitro 8,17 . Although dHL-60 cells can synthesize LTB4 and also possess LTB4 receptors 3 , to the best of our kowledge, swarming behavior in HL-60 cells has not been demonstrated before. Compared to these models, HL-60 cells would allow genetic manipulation not possible in primary neutrophils, thereby allowing direct interrogation of the molecular mechanisms at work in swarming without relying on chemical inhibitors with possible off-target effects. Furthermore, HL-60 cells are a long established and relatively easy to use cell line, which would benefit groups for which access to primary human neutrophils is too expensive or difficult.
In summary, our study provides the first evidence for swarming behavior in DdHL-60 cells. The DdHL-60 swarms are smaller than the primary neutrophils even though DdHL-60 s migrate at comparable speed. Differences in the expression of LTB4-R1 on both DdHL-60 and nDdHL-60 cells may explain the differences. Deficits in the mechanisms of intracellular communication or a combination of these factors may also be involved. Our study, therefore, suggests that DdHL-60 and nDdHL-60 could be useful models for the study of neutrophil chemotaxis and swarming, and further optimizations are needed.
Neutrophil isolation. Human blood samples from healthy donors (aged 18 years and older) were purchased from Research Blood Components, LLC. Human neutrophils were isolated within 1 h after the blood draw using a human neutrophil direct isolation kit (STEMcell Technologies, Vancouver, Canada) following the manufacturer's protocol. After isolation, neutrophils were stained with Hoechst 33,342 trihydrochloride dye (Life Technolo- www.nature.com/scientificreports/ gies). Stained neutrophils were then suspended in RPMI 1640 media containing 20% FBS (Thermo Fisher Scientific) at a concentration of 2 × 10 7 cells/ml in the chemotaxis device or 2.5 × 10 6 cells/mL in the swarming assay.
Gating strategy. Data were analyzed using FlowJo Software. The gating strategy was based on Forward/ Side Scatter (FSC/SSC) profile: (i) Cells of interest were obtained by gating on cell population based on size and granularity using (FSC vs SSC ) to exclude debris (Fig. 7ai). (ii) Single cells were identified by gating on cell of interest population by using forward scatter height (FSC-H) versus forward scatter area (FSC-A) density plot for double cells exclusion (Fig. 7aii). (iii) Viable cells were identified by gating on the single cell population negative for Zombie Aqua (Fig. 7aiii). Device fabrication. The microfluidic devices were fabricated as described by Wang and colleagues using standard soft lithography 12 . Briefly, two-layer master mold in negative photoresist (SU-8, Microchem, Newton, MA) were fabricated on a 4-inch silicon wafer. The first layer was 2 μm thin containing the patterns of the tapered migration channels. The second layer was 75 μm thick and consists of cell-loading channels (CLC) and chemokine chambers. A ratio of 10:1 PDMS base and curing agent were mixed, cast on the master mold, and degassed thoroughly (PDMS, Sylgard, 184, Elsworth Adhesives, Wilmington, MA). We transferred the wafer into an oven at 65•C to cure overnight. After curing, we peeled off the PDMS layer from the wafer and cut out individual devices using a scalpel. We punched the inlets and outlets of the devices using a 0.75 mm diameter biopsy puncher (Harris Uni-Core, Ted Pella) and irreversibly bonded them to a glass-bottom multi-well plate (MatTek Co., Ashland, MA). To prepare the glass slides for the swarming assay, plasma treated glass slides (Fisher brand Double Frosted Microscope Slides, Fisher Scientific, Waltham, MA, USA) were micro-patterned with a solution containing poly-L-lysin and FITC-ZETAG (1.6 mg/ml) using a Polypico micro-dispensing machine 17,18 . Zymosan particle clusters were used as targets for neutrophils swarms. A solution of 0.5 mg/mL zymosan particles in ultra-pure water (Gibco, life technologies, USA) was prepared and sonicated for 10 min before pipetting onto the glass slide and was allowed to adhere for 10mins on a hot plate. Excess zymosan particles was then washed thrice with PBS and stored at room temperature. Prior to the experiment, glass slides were placed in an open well chamber (Grace Bio-Labs). For HL-60 neutrophil-like cells, the chambers were coated with 50 μg/ml of fibronectin for 1 h at 37 °C to improve migration of the cells.
Microfluidic devices to study neutrophil chemotaxis. The microfluidic device used for this study was designed as described by 12 and it consists of an array of tapered channels, with a cross-sectional area of 20 μm 2 at the cell loading chamber end to 6 μm 2 at the chemo-attractant chamber end. The tapered channels are 500 μm in length and connect the cell-loading chamber (CLC) to several chemoattractant chambers. A chemoattractant gradient that increases toward the chemoattractant chamber is established along the tapered migration channels. This device enables us to compare the chemotaxis and migration between HL-60 cell line model of neutrophils and primary neutrophils. Using time-lapse imaging and automatic cell tracking on image J software, we were able to effectively compare the active migration patterns between a single DdHL-60 neutrophil-like cell and a single primary neutrophil with high spatial and temporal resolutions. www.nature.com/scientificreports/ Analysis of differentiated HL-60 neutrophil-like cells (dHL-60) migration. We used Track-mate module in Fiji ImageJ (ImageJ, NIH, version 2.0.0) to track and analyze cell trajectories automatically. We identified four migration behaviors, including persistent migration (P), arrest (A), oscillation (O), and retro-taxis (R). Persistent migration indicates neutrophils that migrated through the tapered channels without changing directions from the cell loading chamber to the chemoattractant chamber. Arrest describes neutrophils that migrated into the tapered channels but got trapped in the channels. Oscillation indicates neutrophils that change migration direction more than two times within the tapered channel. Retro-taxis describes neutrophils that migrated into the tapered channel but migrated back into the cell-loading chamber. Percentage of migration was calculated as thus: (N mc )/(N CLC )*100. Where N mc is total number of migrated cells in the tapered channel and N CLC is total number of cells in the cell loading chamber. Percentage of each migration pattern was calculated as thus: (N mp )/(N mc )*100. Where N mp is number of cells demonstrating a specific migration pattern and N mc is total number of migrated cells in the tapered channel.
Swarm size measurement. Changes in swarm size over time were estimated using track mate plugin in Image J. The cell-occupied area was measured from the DAPI channel for Hoechst labeled HL-60 neutrophil-like cells and primary neutrophil using filter and suitable threshold on image J 8 .
Chemotaxis and tracking analysis. To quantify the directional radial migration during swarming, we measured the distance of neutrophils to the zymosan spot and plotted the changes in this distance for individual tracks over time. The instantaneous chemotactic index (CI) at time t was then calculated as: CI(t) = − R′(t)/ X′(t) where: R′(t) = ∂/∂t ‖x − z‖‖ is the rate of change of the distance between the cell's position x and the zymosan particle cluster position z. Before chemotaxis, analysis the cell track positions x was determined using track mate plugin in Image J. Clusters of zymosan particles were segmented by Huang thresholding followed by selection of the largest connected component, and z was set to be the centroid of this object. Cellular migration speed (spline smoothed) is calculated as: X′(t) = ‖∂x/∂t‖.

Statistical analysis.
Statistical significance of the differences between multiple data groups were tested using two-way Analysis of Variance (ANOVA) in GraphPad Prism (GraphPad Software, version 8.3.0). Within ANOVA, significance between two sets of data was further analyzed using two-tailed t-tests. All the box plots consist of a median line, the mean value (the cross), and whiskers from minimum value to maximum value.