Confinement by Liquid‐Liquid Interface Replicates In Vivo Neutrophil Deformations and Elicits Bleb‐Based Migration

2.1. Exclusive Liquid Repellency (ELR) Enables Liquid Microchannels of Heights Less than Single Cells

In this section, the surface engineering principles enabling construction of liquid channels of heights less than a single cell are described. Recently, groups have engineered liquid‐walled microfluidic channels by either printing aqueous media onto a chemically homogeneous solid substrate under oil overlay,^{[ 41} , ⁴² , ^{43 ]} or by depositing media onto superhydrophilic regions of a chemically patterned (i.e., heterogeneous) substrate.^{[ 44 ]} In either case, interfacial forces within the three phase (solid‐aqueous‐oil) system resist aqueous fluid expansion and thus pin the media to user‐controlled geometries. In such strategies, however, obtaining channel heights on single cell length scales (< 10 µm leukocyte diameter) is challenging, as the forces pinning the aqueous phase to the desired geometry do not reach the theoretical maximum possible within a three‐phase liquid‐liquid‐solid system. Here, we leverage our recent discovery of exclusive liquid repellency (ELR)^{[ 45} , ⁴⁶ , ^{47 ]} to control interfacial geometry with the theoretical maximum of aqueous‐confining forces, enabling engineering of conduit heights comparable to that of single cells (≈1‐10 µm).

Engineering such channels is rooted in foundational principles of interfacial physics. The degree to which a solid surface repels a liquid within a three‐phase solid‐liquid‐liquid system is reflected by the Young's contact angle (𝜃) that a liquid droplet (Liquid 1) makes on the surface (Figure  1A, top). Young's contact angle is determined by the balance between all interfacial energies (𝛾) present within the system. Meticulous engineering of this energy balance results in a surface exhibiting exclusive liquid repellency (ELR), that is, complete repulsion of a liquid, resulting in a contact angle of 180°.^{[ 45} , ^{46 ]} Here, we differentially pattern a glass substrate to exhibit regions of ELR to aqueous media (Liquid 1) in the presence of silicon oil (Liquid 2) (i.e., Aqueous Repellent Surface (ARS)) (Figure 1A, middle), or regions of ELR to silicon oil (Liquid 1) in the presence of media (Liquid 2) (i.e., Oil Repellent Surface (ORS)) (Figure 1A, bottom). Controlling surface patterning of ARS and ORS regions to specific geometries (Figure S2, Supporting Information) results in a double‐ELR system that provides the theoretical maximum of interfacial forces to pin both media and oil onto their preferred surface. Importantly, this enables the construction of channels bound by a liquid‐liquid interface with heights less than single cell diameter (< 10 µm) (Figure 1B). Media is deposited onto the pre‐patterned surface by “sweeping” a hanging droplet of media across the substrate with a pipette to construct an array of > 50 channels on a standard microscope slide within seconds (Figure 1C). Here, we implement a simple pattern that results in two hemispheric droplets connected by a single, linear, liquid channel (Figure 1D,E). The flanking droplets act as inlet and outlet ports to seed cells, chemokines, etc. by direct pipetting within an open system. Notably, however, the method allows for creation of an endless number of configurations (Figure S3, Supporting Information).

In the body, leukocytes migrate upon a number of different adhesive substrates. Of these, collagen is the most abundant ECM protein in the body and is present within a variety of tissue contexts in vivo. That said, high viscosity and rapid polymerization rate limits incorporation of fibrous collagen into traditional closed microchannels on sub‐100 µm length scales. Existing methods often require external pumping of collagen into closed channels,^{[ 48 ]} decreasing throughput and increasing adoption barrier. Here, channels are constructed in open fluid, allowing for direct physical access so that performing the sweep technique with neutralized collagen yields microchannels on single cell length scales coated by fibrillar collagen (Figure 1F,G; Figure S4, Supporting Information).

2.2. The Liquid‐Liquid Interface Confines, yet Deforms in Response to, Immobile Cells

In microchannel systems composed of rigid walls, the degree of confinement, and thus cell shape is determined entirely by the channel geometry. However, during interstitial migration in vivo, leukocyte shape is determined by an equilibrium state between cell‐generated and resistive pressures provided by the surrounding cellular environment. Thus, we sought to investigate whether the liquid‐liquid interface 1) possesses sufficient rigidity to confine cells and 2) is “soft” enough to deform in response to single cell‐scale forces (i.e., single cell stiffness). Primary human neutrophils were incorporated into the hanging droplet during the sweep and thus placed directly into channels of varying width (200 to 30 µm) (Figure  2A). In the absence of cells, channel height is determined by an equilibrium state reflecting the pressure exerted by the aqueous phase itself; channel height (red line) increases with channel width (Figure 2C; Figure S5A, Supporting Information). Cells placed into progressively smaller channels exhibited decreased cell height (blue line), implicating sufficient force applied by the interface to mechanically confine single cells (Figure 2B,C; Figure S5B–F, Supporting Information). This demonstrates the ability to alter cell confinement by varying channel geometry. Notably, when channel height is significantly less than cell diameter (30 and 100 µm width channels), then cell height exceeds that of the channel alone (in the absence of cells). This implies that cells exert passive forces to deform the interface when initial channel height is sufficiently small. Indeed, confocal imaging along the length of a 30 µm width channel reveals that cells deform the interface from an initial equilibrium height (h_0,equilib ) to an increased equilibrium height (h_cell,equilib ) (Figure 2D). Thus, the interface is not a rigid barrier such as plastic or PDMS elastomer but is deformable by the presence of individual cells.

In contrast to rigid substrates with fixed stiffness dependent on material properties alone, the pressure exerted by a liquid‐liquid interface is dependent both on material components of the interface (i.e., oil and media) and interfacial curvature. Thus, we next sought to quantify the confining pressures exerted on cells by the interface. According to the Law of Laplace, the pressure exerted by any curved liquid‐liquid interface is a function of the interfacial tension (material dependent constant; γ_OM ≈ 41.8 mN m⁻¹ for water and silicone oil^{[ 46 ]}) and the radii of interfacial curvature across all curved planes (side yz and cross‐sectional xz in this case). Within this double ELR system that completely immobilizes the aqueous phase to the predefined channel width, deformations in channel height alter interfacial curvature (𝜅) and thus the reciprocal, radius of curvature (R = 1/𝜅) (Figure 2E). The result is an equilibrium confining pressure that reflects the pressures exerted by cells themselves. Critically, interfacial curvature (and thus pressure) is controllable by altering channel width (Figure 2F). While the law of LaPlace describes the pressure at the aqueous‐oil interface, which takes on the curvature of the apical cell surface, here we use this model as an approximation of interfacial pressure confining cells. Interfacial pressures in the presence (blue) or absence (red) of cells decrease with increased channel width (Figure 2G). This demonstrates the ability to control confining pressures by altering channel geometry.

This mathematical framework would predict that the stiffest part of the cell, the nucleus, exerts the greatest pressure to deform the interface. A number of studies have demonstrated that the cell nucleus can support a surface tension to resist deformation.^{[ 49} , ⁵⁰ , ^{51 ]} However, the low lamin A/C content of the neutrophil nucleus decreases its stiffness and contributes to a unique lobulated morphology compared to other cell types.^{[ 52} , ⁵³ , ^{54 ]} This imparts a high degree of nuclear flexibility, enabling the neutrophil to navigate efficiently through confined spaces.^{[ 52 ]} Therefore, it is unclear whether the malleable neutrophil nucleus generates a pressure distinct from cortical tension to generate greater deformation forces on a soft confining interface. To examine this, we analyzed a subset of cells whereby lobes of the nucleus were positioned at distinct locations along the length of a cell. We find that cell regions containing nuclear lobes generate greater interfacial deformation (Figure S6C, Supporting Information). Furthermore, treatment with Trichostatin A to decrease nuclear stiffness^{[ 55} , ⁵⁶ , ^{57 ]} increases cell area (Figure S6B, Supporting Information) and decreases cell height (Figure 2H). Such nuclear softening results in a flat cell morphology independent of nuclear lobe positioning (Figure S6C, Supporting Information). Taken together, these data indicate that the neutrophil nucleus influences the degree of cell confinement. In contrast to rigid‐walled channels where cell confinement is purely a function of predefined geometrical constraints, these data support a model whereby the channels adopt a confining pressure reflecting an equilibrium defined by the physical properties of cells themselves (Figure 2I).

2.3. Neutrophil Motility is Dependent on Required Interaction with the Interface

Given that the interface exerts confining pressures comparable to cells themselves, we sought to examine how interaction with the deformable barrier regulates cell motility. Creation of channels followed by sequential addition of neutrophils to the inlet droplet opposite from chemoattractant in the outlet droplet establishes a chemotaxis assay under confinement (Figure  3A; Figure S7 and Movie S1, Supporting Information). In this assay, cells occupy a layer of culture media between the collagen layer and oil‐media interface, thus chemotaxis would require active deformation of the interface to allow space for migration (Figure 3B). The initial cells to enter the channel, denoted as “pioneer” cells, migrate slowly, followed by “trailing” cells that migrate rapidly (Figure 3C,D). Confocal imaging reveals differential interfacial heights before and after pioneer cell passage (Figure 3E,F), establishing that pioneer cells actively deform the interface to allow space for migration. Notably, the interface is elastic and relaxes after cell passage (Figure S8A,B, Supporting Information), however, in most cases, the sheer number of trailing cells maintains an interfacial height and limits interface reversion to an original state. Therefore, while both populations are confined by the interface in the transverse (z) direction (Figure S8A, Supporting Information), pioneer cells must also interact with the interface in the axial (y) direction at the cell front. It is reasonable that this mechanical resistance is responsible for decreased pioneer cell speed. However, it is also possible that pioneer cell migration alters the chemical gradient in a manner to increase motility of trailing cells. To investigate this, we added a combination of fMLP and FITC dye of similar molecular weight (FITC 389.382 g mol⁻¹ versus fMLP 437.56 g mol⁻¹) to visualize an approximate chemokine gradient during the course of migration (Figure S8C, Supporting Information). We found that pioneer cell passage did not significantly alter the chemical gradient (Figure S8D, Supporting Information). To further decouple positioning within the chemical gradient from required interaction with the interface, we analyzed a subset of trailing cells that migrate fast enough to catch up to and thus become pioneer cells. As these cells move through the stable global gradient, they significantly decrease in speed once they reach the interface (Figure 3G). Taken together, these data indicate a purely mechanical mechanism whereby the interaction with (and deformation of) the interface restricts cell motility. Pioneer cells actively deform the channel interface, leaving decreased mechanical resistance for trailing cells which in turn migrate faster. From a technical standpoint, these results demonstrate the ability to decouple chemical cues such as position within a stable gradient, from mechanical cues of interaction with the interfacial barrier to isolate a role of physical cues on cell motility. Given the ability to tune pressures exerted by the interface, we next studied how alterations in the physical properties of the surrounding environment regulate cell motility.

2.4. Liquid‐Liquid Interface Generates a Tunable Gradient of Confinement that Regulates Motility

In contrast to immobile cells, migratory cells typically exhibit morphological polarity, largely dependent on nuclear positioning as they extend cytoplasmic protrusions at the leading edge to deform the surrounding environment. Rigid microchannels impose fixed environmental stiffness that is not responsive to cell‐generated forces, so that all cell components (nucleus and cytoplasm) typically assume the entirety of available space within the channel. Here, we find that pioneer neutrophils exhibit rear‐ward nuclear positioning and front‐ward cytoplasmic protrusions (Figure  4A; Figure S9A and Movie S2, Supporting Information). These regions deform the interface to different degrees; the cell rear induces maximal interfacial deformation, characterized by increased interfacial curvature (Figure 4B) and thus maximal confining pressures, compared to frontward protrusions (Figure 4C). Thus, in contrast to immobile cells directly placed into channels, which exhibit symmetric deformations in the xz and yz planes, pioneer cells interacting with the interface generate spatially variant radii of curvature across both planes. The result is a gradient of confining pressures across the length of the cell. This gradient is tunable by altering channel width (Figure S9D, Supporting Information). Increased width generates a more shallow gradient across cell length (yz plane), with less significant differences in confining pressure between the forward protrusion and nucleus‐containing rear (Figure 4D). Therefore, tuning interface curvature (xz plane) controls cell shape (yz plane) during pioneer cell migration. Bulk analysis confirms that increased channel width results in decreased confining pressures over both the cell rear (Figure S9E, Supporting Information) and frontward protrusions (Figure S9F, Supporting Information). Thus, akin to interstitial migration through elastic tissues in vivo, the greater the force generated by the cell, the higher resistive pressure exerted by the surrounding elastic environment.

To explore how altering confining pressure gradients regulates motility, we engineered channels of varying width (30 versus 100 µm) and length (300 versus 500 µm, respectively) that exhibit different confining pressures yet similar chemical gradients (Figure S10, Supporting Information). While the gradient within 30 µm width channels was slightly steeper than that in 100 µm channels (typically increasing motility), cells migrated significantly slower, suggesting that increased confining pressures negatively impacts motility (Figure 4E). Ultimately, these data indicate that the liquid‐liquid interface is sensitive to differential pressures generated across a single migratory cell, and that cell morphology and motility are tunable by controlling channel geometry. These in vitro data would predict the extension of thin protrusions to initially deform surrounding cells in vivo, followed by the cell rear eliciting maximal surrounding cell deformations during interstitial migration through elastic cellular tissues. To examine this, we next established an in vivo model to capture deformations of surrounding cells during neutrophil interstitial migration within the larval zebrafish.

2.5. Tuning Interface Pressures Recapitulates Surrounding Cell Deformations During Interstitial Migration In Vivo

Migratory leukocytes sense membrane deformations to regulate motility and effector function.^{[ 58} , ⁵⁹ , ^{60 ]} Membrane shape in vivo is determined by a balance between outward cell generated forces and resistive pressures arising from surrounding tissues. Thus, the magnitude of surrounding cell deformations reflects the relative force relationship between migrating leukocytes and surrounding cells. We next explored whether tuning the resistive pressures applied by the interface could recapitulate relevant leukocyte morphology and surrounding cell deformations during interstitial migration in vivo.

Limitations in optical access of most in vivo models have hindered the ability to capture leukocyte physical interactions with surrounding cells during migration. Here, we leverage the genetic tractability and optical transparency of the larval zebrafish to capture deformations of neutrophils and surrounding keratinocytes during interstitial migration within the zebrafish epidermis (Figure  5A). We performed high resolution imaging of spontaneous neutrophil motility within the epidermis of a transgenic larval zebrafish model expressing BFP in the neutrophil cytoplasm, mCherry in the nucleus and acGFP in surrounding basal keratinocytes (Tg(LyzC:TagBFP/LyzC:H2B‐mCherry) x Tg(Krtt1c19e:acGFP)). Such imaging reveals that neutrophils interact with and deform surrounding basal keratinocytes during interstitial migration (Movie S3, Supporting Information). Neutrophils first extend a cytoplasmic protrusion to exert outward forces on surrounding keratinocytes, followed by a period of maximal keratinocyte deformation during passage of the cell body (Figure 5B). As predicted within liquid channels, the nucleus is rear‐ward oriented, and passage of the cell rear induces maximal deformations of surrounding cells (Figure 5C). Tuning the interfacial shape by controlling channel width (30 µm versus 100 µm) reproduces neutrophil protrusion width (Figure 5D) and epithelial cell deformations during passage of the cell rear (Figure 5E) observed in vivo. These data indicate that tuning of baseline channel pressures reproduces relevant transient force balances during interstitial migration in vivo, validating the relevance of this system to model physical cell‐cell interactions during interstitial leukocyte migration.

2.6. Neutrophils Transition to a Bleb‐Based Mechanism of Force Generation to Deform an Interface Exerting Cell‐Scale Confining Pressures

We next sought to explore the mechanism of force generation that cells employ at the leading edge to deform the liquid interface and generate space for migration. On 2D substrates, leukocytes largely rely on polymerization of branched actin networks stabilized by actin‐related protein 2/3 (Arp2/3) to extend protrusions at the leading edge. Recent studies have demonstrated that under rigid confinement, neutrophils can employ a bleb‐based motility mechanism independent of actin polymerization at the leading edge,^{[ 18 ]} however, it is not known how cells generate forces to deform a surrounding elastic environment exerting cell scale confining forces. Blebbing is driven by Rho‐kinase dependent contraction at the rear driving increased intracellular hydraulic pressure. This drives detachment and rapid projection of the cell membrane through the F‐actin cortex at the cell front (Figure  6A). Blebs form on significantly faster timescales (< 2 µm s⁻¹ and < 10 s lifespans) than pseudopodia.^{[ 12 ]} We investigated the hypothesis that interaction with the interface provoked a bleb‐based mechanism of migration to generate outward deforming forces.

We first examined migration of primary neutrophils under inhibition of the Arp‐2/3 complex using the small molecule CK666. While migration was inhibited within an established 2D migration system (Figure S11A,B, Supporting Information), it was unaffected under confinement within liquid channels (Figure 6B). This suggests motiliy independent of Arp‐2/3 mediated actin polymerization at the leading edge. Intriguingly, rapid imaging (1 s intervals) revealed sheet‐like pseudopodial protrusions in trailing cells and formation of rapid, rounded protrusions that mimic the shape and kinetics of blebs in pioneer cells (Figure 6C; Figure S11C and Movie S4, Supporting Information). These bleb‐like structures are disproportionately formed by pioneer cells interacting with the interface compared to trailing cells (Figure S11D, Supporting Information). Inhibition of contractility under treatment with Rho‐Kinase inhibitor Y‐27632 hinders the ability of pioneer cells to migrate through the interfacial barrier (Figure 6D). Under such treatment, pioneer cell speed is sufficiently slow, such that trailing cell motility is limited by that of pioneers and thus were excluded from analysis. Attempts at confirming blebbing in primary neutrophils by visualization of F‐actin was limited by photobleaching of fluorescent actin probes and resistance to transfection. Thus, we imaged migration of neutrophil‐like PLB‐985 cells under stable expression of LifeAct‐mRuby. Pioneer neutrophil‐like cells exhibited prominent blebbing at the leading edge compared to trailing cells that favored pseudopodial‐based protrusions (Figure 6E,F; Figure S11E and Movies S5‐S6, Supporting Information). Furthermore, trailing cells that eventually reach the interface to become pioneers rapidly transition from pseudopodial to bleb‐based protrusions (Figure 6G; Movie S7, Supporting Information). Taken together, these data indicate a mechanism of bleb‐based force generation at the leading edge to deform an interface exerting cell‐scale confining pressures.

Construction of Liquid‐Liquid Microchannels

A chambered coverglass (no. 1.5 borosilicate glass, 0.13 to 0.17 mm thick; Thermo Fisher Scientific, 155 360) was treated first with O₂ plasma (Diener Electronic Femto, Plasma Surface Technology) at 60 W for 3 min and then moved to an oven for grafting of PDMS‐Silane (1,3‐dichlorotetramethylsiloxane; Gelest, SID3372.0; ≈10 µL per device) onto the glass surface by chemical vapor deposition (CVD) at 50 °C for 30 min. The grafted slide was masked by a PDMS stamp (stamp construction described in Li et al.^{[ 47 ]}) and treated with O₂ plasma at 60 W for 3 min to pattern ORS regions. After surface patterning, the PDMS stamp was removed and stored in a clean space for reuse. The chambered coverglass slides were overlaid with oil (silicone oil, 5 cSt; Sigma‐Aldrich, 317 667), so that regions not exposed to O2 plasma exhibit ELR to aqueous media (i.e., ARS surface). Liquid‐liquid channels were created by sweeping a hanging droplet of culture media across the patterned surface with a wide‐orifice pipette. Aqueous fluid from the hanging droplet spontaneously deposits onto ORS regions only (Figure 1D,E). To seed channels with fibrous collagen, a mixture of one part rat tail collagen I (Corning, 10 mg mL⁻¹, on ice), one part neutrophil culture media (Roswell Park Memorial Institute (RPMI) 1640 Medium, Thermo Fisher Scientific,11 875 093) + 2% fetal bovine serum (FBS; Thermo Fisher Scientific, 10 437 010) and 1% Penicillin/Streptomycin)) and one‐part 2X HEPES buffer was swept across the patterned glass substrate. The result is a final concentration of 3.33 mg mL⁻¹ rat tail collagen at pH 7.2. For experiments where cells were placed directly into the channels (Figure 2), primary neutrophils (see isolation methods below) were added to the culture media component of this mixture at a concentration of 80 000 µL⁻¹ (final 26 667 µL⁻¹ following the 1:3 dilution with collagen and 2X HEPES). In any case, the collagen (± cells) mixture was incubated at 37 °C, 21% O2, 5% CO2, and 95% RH for 15 min to allow for polymerization before transfer to an on‐stage incubator (Bold Line, Okolab) for imaging.

Whole Blood Collection and Neutrophil Isolation

All blood samples were drawn according to Institutional Review Boards (IRB) approved protocols per the Declaration of Helsinki at the University of Wisconsin‐Madison in the Beebe Lab (IRB# 2020‐1623) and Huttenlocher Lab (IRB# 2017‐003). Informed consent was obtained from all subjects in the study. Whole blood was collected with standard ethylenediaminetetraacetic acid (EDTA) tubes (BD, #36 643, EDTA [K2] STERILE, 1 054 688) and then stored at RT (≈ 22 °C) or 37 °C in stationary storage before isolation.

Neutrophils were isolated from whole blood using magnetic bead‐based negative selection per protocol using either the EasySep Direct Human Neutrophil Isolation Kit (STEMCELL, 19 666) or the MACSxpress Neutrophil Isolation Kit (Miltenyi Biotech, 130‐104‐434) purified with the MACSxpress Erythrocyte Depletion Kit (Miltenyi Biotech, 130‐098‐196). After isolation, the neutrophil pellet was directly resuspended in RPMI 1640 culture media with 2% FBS and 1% Penicillin/Streptomycin (as above). Following resuspension, liquid cultures were stored at 37° in a standard CO₂ incubator (Thermo Scientific, HERACELL 240i) for no more than 1 h before seeding into the device and transfer to an onstage incubator for imaging.

Quantification of Cell and Channel Heights Following Passive Sweep

Following isolation, primary neutrophils were labeled immediately after resuspension by incubation at 37 °C and 5% CO₂ for 15 min with the cytoplasmic dye Calcein‐AM green (1 µm; Invitrogen, C3100MP) and the nuclear dye Hoescht 33 342 (10 µg mL⁻¹; Invitrogen, H1399) in RPMI 1640 media with 2% FBS and 1% Penicillin/Streptomycin. Cells in media (concentration 80 000 µL⁻¹) were combined with collagen (10 mg mL⁻¹, on ice) and 2X HEPES at 1:1:1 ratio (as above) and mixed by pipette. The resultant mixture was immediately swept across the patterned surface as described above to place cells directly into channels. Regarding experiments to soften the cell nucleus (Figure 2H; Figure S6B‐C, Supporting Information), primary neutrophils were incubated at 37 °C with 10 µM Trichostatin A (TSA) for 1 h before seeding into channels. Confocal imaging was performed on a Nikon AR1 spinning disk microscope courtesy of the UW Optical Core to gather z‐stacks with step size of 200 nm. Cell and channel heights (Figure 2C,H) were measured in FIJI/ImageJ from cross‐sectional orthogonal views at regions where a cell was or was not present (both within the same channel). Interfacial curvature (Figure 2G) was calculated by transferring images (both xz and yz planes) to Microsoft PowerPoint and manually drawing a circle on top of the image to match the curvature of the interface (in both the xz and yz planes). The radius of this circle yields the radius of curvature (inverse of curvature) in that plane. Interfacial pressures were calculated by the LaPlace equation; P_int = γ(1/R_xz + 1/R_yz ). A single representative cell at the center of the channel was chosen for analysis per unique channel replicate, with a total of three channel replicates for each independent human donor (n = 3).

Quantification of Cell Morphology and Motility During Chemotaxis

To study cell chemotaxis under confinement by the liquid‐liquid interface (Figure 3), channels were constructed by sweeping with neutralized collagen and incubated at 37° to allow polymerization, as described above. Immediately after sweep, excess volume was manually removed by pipette to ensure equal volumes in both inlet and outlet droplets. Following isolation from whole blood, neutrophils were re‐suspended in RPMI 1640 media containing 2% FBS and 1% Penicillin/Streptomycin at a concentration of 40 000 cells per µL. In experiments where interfacial height or curvature was measured from volumetric confocal stacks (Figures 3 and 4), media was visualized by addition of fluorescein (FITC) sodium salt (40 µM; Sigma Aldrich, F6377). Vybrant DiD (Invitrogen, V22887) was used to label the cell membrane; 5 µL dye was added to 1 mL of cell suspension in above media immediately following isolation from whole blood (per the Invitrogen protocol) and incubated at 37 °C and 5% CO₂ for 15 min prior to washing and cell resuspension to the desired concentration of 40k cells per µL for seeding into the liquid inlet droplets.

Following resuspension, 1.5 µL of cell suspension was added to the inlet droplet by pipette. Immediately after, 1.5 µL of 100 nM N‐Formylmethionyl‐leucyl‐phenylalanine (fMLP; Millipore Sigma F3506) in the same media was added to the outlet droplet. After such addition, the cell suspension was mixed by pipetting to ensure homogeneity and cell positioning near the channel entrance. Addition of equal amounts of volume to both the inlet and outlet droplets is essential to restrict generation of convective flow throughout the channel, so that chemokine gradients are driven by diffusion alone. Cell motility was visualized on a Nikon epi‐fluorescent microscope (Nikon Eclipse TE3000 with a Nikon Plan Fluor 20x/0.50 objective, motorized stage from Ludl Electronic Products and Prime BSI Express camera from Teledyne Photometrics or a Nikon TI Eclipse inverted microscope with Nikon Plan Fluor 20x/0.50 objective and Hamamatsu camera). Quantification of cell speed was performed in ImageJ/FIJI using the manual tracking plug‐in of time lapse movies taken at time intervals of 30 s. Volumetric imaging of cells interacting with the interface was performed by labeling the media with FITC dye and imaging with the aforementioned Nikon AR1 spinning disk confocal microscope with step size of 200 nm. Through such volumetric imaging, interfacial heights were measured by the height of the FITC labeled media layer. Similarly, measurements of interfacial pressures during pioneer cell migration (Figure 4) were made from measuring interfacial curvature (defined by the FITC labeled media) from cross‐sectional (xz plane) and side (yz plane) volumetric images as described above. Briefly, images were transferred to Microsoft PowerPoint and interfacial radii of curvatures at the cell front and rear (in both the xz and yz planes) measured by manually drawing a circle to match the curvature of the interface (see Figure S9, Supporting Information for visual depiction).

Zebrafish Maintenance and Handling

Animal care and use was approved by the Institutional Animal Care and Use Committee of University of Wisconsin and strictly followed guidelines set by the federal Health Research Extension Act and the Public Health Service Policy on the Humane Care and Use of Laboratory Animal, administered by the National Institutes of Health Office of Laboratory Animal Welfare. All protocols using zebrafish in this study were approved by the University of Wisconsin‐Madison Research Animals Resource Center (protocol M005405‐A02). To generate the larval zebrafish model used to visualize neutrophil morphology and keratinocyte deformations, previously published neutrophil‐cytoplasm labeled Tg(LyzC:TagBFP) fish were crossed with nuclear labeled Tg(LyzC:H2B‐mCherry) fish to generate a stable line (Tg(LyzC:TagBFP/LyzC:H2B‐mCherry). These fish were bred with a stable line expressing cytoplasmic acGFP in basal keratinocytes Tg(Krtt1c19e:acGFP)) (Gift of Dr. Alvaro Sagasti) to generate larvae used in each experiment. All fish were maintained on the AB background. Following breeding, fertilized embryos were transferred to E3 medium (5 mM NaCl, 0.17 mM KCl, 0.44 mM CaCl₂, 0.33 mM MgSO₄, 0.025 mM NaOH, and 0.0003% Methylene Blue) and maintained at 29 °C. Larval zebrafish were anesthetized using 0.2 mg mL⁻¹ tricaine (ethyl 3‐aminobenzoate; Sigma‐Aldrich) before any experimentation or live imaging.

Tracking In Vivo Keratinocyte Deformations

Dechorionated, tricaine‐anesthetized three days‐post fertilization (dpf) larvae were mounted in 2% low‐melting point agarose (Sigma‐Aldrich) on a 35 mm glass bottom dish (#1.5H Glass, Cellvis, CA, USA). Time‐lapse imaging of spontaneous neutrophil migration through basal keratinocytes in the yolk sac and neck region was performed on a spinning‐disk confocal (CSU‐X; Yokogawa) on a Zeiss Observer Z.1 inverted microscope and an electron‐multiplying charge‐coupled device Evolve 512 camera (Photometrics), with an EC Plan‐Neofluar 40×/NA 0.75 air objective (1‐2 µm optical sections, 2355 × 512 resolution) using ZenPro 2012 software (Zeiss). Images were captured at 10 s intervals with z‐step intervals of 3 µm. Quantification of deformations was performed on resulting maximal intensity projections; background subtraction was manually performed in FIJI/ImageJ in images displayed in the main text (Figure 5).

Quantification of Bleb Formation in Primary Neutrophils and Neutrophil‐Like PLB‐985 Cells

Both primary neutrophils and PLB‐985 neutrophil‐like cells were used to determine the migration mode utilized by neutrophils during interaction with the confining interface (Figure 6). Immediately following isolation from whole blood, primary neutrophils were incubated with 10 µM CK666 (Millipore Sigma, SML0006) or LY‐27632 (Millipore Sigma, Y0503) to inhibit CK‐666 or Rho‐Kinase at 37 °C and 5% CO₂ for 30 min. Bleb‐like structures in primary neutrophils were quantified from brightfield time lapse images taken at 1 s intervals.

PLB‐985 cells expressing LifeAct‐mRuby were generated by lentiviral transfection. HEK293T cells were grown to 70% confluency in a 10‐cm tissue culture dish for each lentiviral target and transfected with pLV‐EF1a‐IRES‐Neo (Addgene #85 139) containing LifeAct‐mRuby, VSV‐G, and CMV 8.9.1. The 72‐h viral supernatant was collected and concentrated using a lentivirus concentrator (Lenti‐X; Takara Bio Inc.) following the manufacturer's instructions. PLB‐985 cells (1 × 10⁶ total number) were infected with viral supernatant for 3 days in the presence of 15 µg mL⁻¹ polybrene. Stable cell lines were generated with 1 µg mL⁻¹ neomycin selection. Resulting PLB‐985 cells expressing LifeAct‐mRuby were differentiated into neutrophil‐like cells by treatment with 1.25% DMSO (Sigma‐Aldrich, D2650) for 6 days at 37 °C and 5% CO₂ in RPMI 1640 with 2% FBS and 1% Penicillin/Streptomycin. Cell motility was assayed in an equivalent assay as that of primary cells, with imaging performed on the previously described Zeiss Observer Z.1 inverted spinning disc confocal microscope, at 0.5‐1 s intervals. Background subtraction was manually performed in FIJI/ImageJ in images displayed in the main text (Figure 6).

Statistical Analysis

Data were analyzed (Prism 9.0; GraphPad Software). Statistical significance was assessed using Student's t tests when comparing two conditions/groups, and one‐way analysis of variance (ANOVA) corrected using the Tukey's test when comparing multiple groups. All n and p‐values are noted in the figure captions.