LFA-1 signals to promote actin polymerization and upstream migration in T cells Free

Migration of leukocytes from the vasculature into peripheral tissue is central to their role in fighting pathogens, promoting tissue repair and attacking solid tumors. This process, called transendothelial migration (TEM), is a key control point in the inflammatory response (Nourshargh and Alon, 2014). TEM is a multi-step process that begins with selectin-dependent cell rolling on the vasculature, followed by chemokine-induced cell arrest. At this point, integrins expressed on the leukocyte interact with their endothelial ligands, resulting in shear-resistant adhesion, cell spreading, and migration along the endothelium (Denucci et al., 2009; Ley et al., 2007; Nourshargh and Alon, 2014). Leukocyte migration along the vascular wall is a prerequisite for transmigration, and is thought to allow cells to search for ideal sites to cross the endothelial layer, such as three-way junctions (Gorina et al., 2014; Phillipson et al., 2006; Schenkel et al., 2004; Shaw et al., 2001; Sumagin and Sarelius, 2010). Intravital imaging of immune responses in vivo has revealed that leukocyte migration along the endothelial monolayer is not random and can be directed by shear flow forces (Bartholomäus et al., 2009; Phillipson et al., 2009). Interestingly, leukocytes have been observed to preferentially migrate against the direction of shear flow (Bartholomäus et al., 2009), an unexpected result given the extra energy expenditure needed to oppose head-on shear forces. Although it is not clear why leukocytes display this phenotype, in vitro studies have shown that T cells crawling against the direction of shear on inflamed endothelia are more likely to undergo transmigration (Anderson et al., 2019; Steiner et al., 2010), suggesting a link between these two mechanically demanding processes.

T cell adhesion and migration on the vascular wall are dependent on integrin interactions with their endothelial ligands. The two major integrin ligands expressed on inflamed endothelia are ICAM-1 and VCAM-1, which serve as binding partners for the T cell integrins LFA-1 (α_Lβ₂) and VLA-4 (α₄β₁), respectively. Although both of these integrin–ligand pairs can support adhesion, they seem to promote distinct migratory behaviors. The most striking example of this is observed under shear flow, in which T cells migrating on ICAM-1-coated surfaces preferentially migrate against the direction of shear flow (upstream migration), whereas T cells migrating on VCAM-1-coated surfaces migrate with the direction of shear flow (downstream migration) (Dominguez et al., 2015; Hornung et al., 2020; Kim and Hammer, 2019; Valignat et al., 2014, 2013). This integrin-dependent phenomenon has also been documented in B cells (Tedford et al., 2017), hematopoietic precursors (Buffone et al., 2018) and neutrophils (under some conditions; Buffone et al., 2019). These data point towards fundamentally different roles for ICAM-1 and VCAM-1 in coordinating leukocyte migration.

A biophysical model has been proposed to explain the ability of T cells to migrate upstream on ICAM-1. The model is based on the finding that LFA-1–ICAM-1 interactions occur toward the front of the cell, and induce the formation of a broadly spread leading edge and a raised, trailing uropod, whereas VLA-4–VCAM-1 interactions take place toward the back of the cell and fail to generate these shape changes (Hornung et al., 2020). The shape changes induced by LFA-1 engagement allow the uropod to act as a ‘wind vane’ that passively steers the T cell upstream (Valignat et al., 2014). Although this model relies on the idea that localized integrin adhesion results in the relevant cell shape changes, T cell spreading and migration have also been shown to involve integrin signaling. This is best documented for LFA-1; T cells that come into contact with surface-presented ICAM-1 immediately polarize and begin to migrate (Kelleher et al., 1990; Porter et al., 2002; Smith et al., 2003).

Recently, we characterized a signaling pathway downstream of LFA-1 that leads to cell spreading, actin polymerization, and migration (Roy et al., 2018). Central players in this pathway are the Crk adaptor proteins, which coordinate Src-dependent phosphorylation of the scaffolding ubiquitin ligase c-Cbl (also known as CBL), ultimately leading to phosphoinositide 3-kinase (PI3K) activity and actin responses. Importantly, disruption of this pathway by deleting Crk proteins perturbs LFA-1-dependent migration, showing that LFA-1 signaling is an important factor driving T cell migration on ICAM-1. Because the LFA-1–ICAM-1 interaction triggers strong migratory responses in T cells, we hypothesized that differential integrin signaling events account for the differences in behavior of T cells migrating on ICAM-1 versus VCAM-1. Therefore, we analyzed T cell migration and signaling in response to binding ICAM-1 or VCAM-1, with the goal of determining how these different integrin ligands control T cell migration.

We and others have previously shown that T cells preferentially migrate against the direction of shear on ICAM-1 and with the direction of shear on VCAM-1 (Anderson et al., 2019; Dominguez et al., 2015; Hornung et al., 2020; Kim and Hammer, 2019; Steiner et al., 2010; Valignat et al., 2014, 2013). To verify that our primary mouse T cells display this behavior, we activated T cells for 2 d and rested them for an additional 3 d in the presence of IL-2, a procedure that generates highly motile T cell blasts expressing both LFA-1 and VLA-4. These cells were allowed to adhere to surfaces coated with ICAM-1, VCAM-1, or both ligands mixed at a 1:1 ratio, and then subjected to shear flow of either 100 s⁻¹ or 800 s⁻¹, corresponding to roughly 1 dyn/cm² (0.1 Pa) and 8 dyn/cm² (0.8 Pa), respectively. Video microscopy was performed, and the migrating cells were tracked and analyzed as described in the Materials and Methods. Fig. 1A displays representative scattergrams of T cell migration tracks, with red tracks indicating upstream migration and blue tracks indicating downstream migration. These data show a clear tendency of T cells to migrate upstream on ICAM-1 and mixed surfaces, whereas migration on VCAM-1 was almost completely downstream. To quantify upstream versus downstream migration, we calculated the migration index (MI) for each cell. MI is determined by dividing the displacement in the x-direction (the axis parallel to the shear flow) by the total track length. An MI value of −1 indicates that cells migrate in a straight trajectory against the direction of flow, and an MI value of +1 indicates migration in a straight trajectory with the direction of flow. Quantification of at least three independent experiments confirmed that T cells tend to migrate against the direction of shear on ICAM-1 (Fig. 1B). T cells on mixed surfaces displayed a phenotype similar to those on ICAM-1 alone, showing that ICAM-1 ligation induces the dominant phenotype.

We reasoned that this upstream/downstream phenotype could result from large differences in LFA-1 or VLA-4 expression levels, or from the ability of the different integrins to promote firm adhesion. To test this, we surface labeled primary T cells for CD11a and CD18 (integrins αL and β2, respectively, which compose LFA-1), and CD49d and CD29 (integrins α4 and β1, respectively, which compose VLA-4). T cells expressed measurable levels of all four integrin subunits (Fig. 1C). As detailed further in Fig. S1, analysis of additional α- and β-integrin subunits revealed that LFA-1 and VLA-4 are likely the primary integrins that bind to ICAM-1 and VCAM-1, respectively (Fig. S1). To compare the ability of the two integrins to support adhesion, we utilized a standard plate-based adhesion assay to measure binding to surfaces coated with ICAM-1, VCAM-1, or both ligands in combination. Interestingly, VCAM-1 supported more robust adhesion than ICAM-1, and the mixed surfaces promoted adhesion that was slightly higher than VCAM-1 alone (Fig. 1D). Thus, both integrins act as functional adhesion molecules, and there is no correlation between the efficiency of adhesion and upstream/downstream migratory behavior. Taken together, these data show that primary mouse T cells migrate upstream on ICAM-1 and downstream on VCAM-1, and that this phenotype cannot be explained simply by differences in integrin expression or adhesive functions.

To better understand the differential nature of migration on ICAM-1 and VCAM-1, we analyzed the shape and behavior of migrating T cells. This was done in the absence of shear to eliminate confounding effects induced by the shear itself. We showed previously that engagement of LFA-1 on primary mouse T cells by surface-bound ICAM-1 leads to cell spreading, cytoskeletal remodeling, and persistent directional migration (Roy et al., 2018). To ask whether the VLA-4–VCAM-1 interaction induces similar responses, we conducted side-by-side analysis. Primary mouse T cells were allowed to interact with surfaces coated with ICAM-1, VCAM-1, or both ligands mixed at equal ratios. Poly-L-lysine was included as a non-specific control. Cells were then fixed, stained for F-actin and imaged using confocal microscopy. As shown in Fig. 2A, T cells responded to ICAM-1-coated surfaces by spreading and forming a broad, actin-rich leading edge. In contrast, T cells responding to VCAM-1 displayed a more elongated shape. Although a leading edge could usually be identified, it was narrower and lacked robust F-actin accumulation. Quantification of F-actin levels confirmed that spreading on ICAM-1 induced an increase in actin polymer, whereas F-actin levels in cells responding to VCAM-1 remained unchanged relative to the poly-L-lysine control (Fig. 2B). To ensure this was not due to decreased binding to VCAM-1, we coated plates with a 3.75× molar excess of VCAM-1 and observed no increase in spreading or leading edge formation (Fig. S2). T cells responding to mixed surfaces were indistinguishable from T cells responding to ICAM-1 alone (Fig. 2A,B), consistent with the finding that ICAM-1 induces the dominant phenotype. To evaluate how the different surfaces affect T cell migration, we tracked migrating cells using video microscopy. T cells migrating on ICAM-1 and mixed surfaces migrated significantly faster and more directionally than T cells on VCAM-1 (Fig. 2C,D). To more closely examine cell shape and actin dynamics during T cell migration, we utilized T cells expressing Lifeact–GFP. As observed in fixed cells labeled with phalloidin (Fig. 2A), T cells migrating on ICAM-1 or mixed surfaces adopted a spread morphology and displayed an actin-rich leading edge, with a broad contact area with the substrate and a raised uropod (Fig. 2E,F). In contrast, T cells migrating on VCAM-1 showed less contact with the stimulatory surface and seemed to adhere mostly through the center and uropod of the cell (Fig. 2E,F). This morphology is consistent with observations by others (Hornung et al., 2020; Valignat et al., 2014). Analysis of the movies revealed that whereas T cells migrating on ICAM-1-coated surfaces moved smoothly, showing sustained actin polymerization at the leading edge, T cells on VCAM-1 had periodic bursts of movement concomitant with brief actin flares at the front of the cell (Fig. 2F,G; Movies 1, 2). This differential behavior can be readily observed using a kymographic representation (Fig. 2G). Taken together, these data show that engagement of ICAM-1, but not VCAM-1, results in T cell spreading, actin polymerization, and fast migration. Additionally, these data suggest that ICAM-1 triggers specific signaling events that lead to these responses.

Recently, we identified the Crk family adaptor proteins as key signaling intermediates that promote actin polymerization and migration in T cells downstream of LFA-1 (Roy et al., 2018). The Crk family consists of three proteins, Crk I, Crk II, and CrkL, which are transcribed from two loci (Crk I and Crk II are splice variants encoded by the Crk locus) (Feller, 2001). In the absence of shear flow, T cells lacking all three Crk proteins (DKO T cells) migrate significantly slower than wild-type (WT) T cells on ICAM-1, and fail to spread and polymerize actin at the leading edge (Roy et al., 2018) (Fig. 3A,B). This phenotype is strikingly similar to that of WT T cells migrating on VCAM-1, suggesting that Crk protein signaling might be a determining factor in the cellular response to ICAM-1. To ask whether Crk signaling is also important for upstream migration, we imaged WT and DKO T cells migrating in the presence of shear flow (800 s⁻¹), as in Fig. 1. We found that DKO T cells were unable to migrate upstream on ICAM-1, with no apparent defect in downstream migration on VCAM-1 (Fig. 3C,D). This phenotype was not due differences in integrin levels on DKO T cells (Fig. S3). These data support a model in which Crk-dependent LFA-1 signaling drives T cell spreading, actin responses, and upstream migration.

We next sought to identify signaling differences downstream of LFA-1 and VLA-4 that could account for the differential cell behavior that we observed. In addition to activating canonical signaling pathways, such as the PI3K–AKT and ERK pathways, engagement of LFA-1 and VLA-4 induces the phosphorylation and activation of several signaling and scaffold proteins, including CasL (also known as NEDD9), cCbl (also known as CBL), FAK (PTK2), and Pyk2 (PTK2B) (Cheung and Ostergaard, 2016; Hunter and Shimizu, 1997; Meng and Lowell, 1998; Raab et al., 2017; Sato et al., 1995; Tabassam et al., 1999; van Seventer et al., 1998, 2001). Crk proteins are crucial for multiple signaling events downstream of LFA-1, including phosphorylation of CasL and cCbl, as well as activation of the PI3K pathway (Roy et al., 2018). Our observation that Crk-deficient T cells do not migrate upstream on ICAM-1 suggests that some of these signaling events may be necessary for upstream migration. If so, we reasoned that these events might occur in cells responding to ICAM-1, but not to VCAM-1. To our knowledge, a direct comparison of signaling events initiated by surface-presented ICAM-1 versus VCAM-1 has never been reported. To determine whether ICAM-1 and VCAM-1 differentially activate any of these downstream signaling pathways, we allowed T cells to interact with surfaces coated with either ICAM-1 or VCAM-1. The T cells were then lysed, and the activation state of the PI3K and ERK pathways were probed using phospho-specific antibodies. In addition, lysates were immunoprecipitated with anti-phosphotyrosine antibody, and probed with antibodies to CasL, cCbl, and Pyk2. ICAM-1 stimulated robust phosphorylation of AKT and ERK (ERK1 and ERK2, also known as MAPK3 and MAPK1), while stimulation with VCAM-1 induced only modest activation of these pathways (Fig. 4A,B). Similarly, robust phosphorylation of cCbl was only induced by ICAM-1 (Fig. 4A,B). Importantly, phosphorylation of CasL and Pyk2 occurred with similar efficiency after engagement of ICAM-1 and VCAM-1. These data demonstrate that both LFA-1 and VLA-4 are capable of signaling, but the repertoire of downstream signaling events is distinct.

Because of the clear differences in LFA-1 and VLA-4 signaling, we next wanted to directly test whether any of the LFA-1-specific signaling events were important for upstream migration. In particular, we focused on cCbl and PI3K, both of which show defective activation in Crk-deficient T cells. To target these proteins, we implemented a CRISPR knockout (KO) system in primary mouse T cells (Huang et al., 2019). T cell blasts cultured from Cas9-expressing mice were transduced with retroviral vectors containing non-targeting (NT), cCbl, or PI3Kδ gRNAs. T cells were then selected in puromycin for 3 d and used for experiments. Reductions in protein levels were confirmed by immunoblotting (Fig. 5A,B). To assess the role of these proteins in LFA-1-mediated responses under static conditions, we allowed T cells to interact with ICAM-1-coated surfaces and stained for F-actin. T cells expressing NT gRNA showed the canonical migratory phenotype, with a broad actin-rich leading edge (Fig. 5C). Deletion of cCbl had no effect on this phenotype, whereas loss of PI3Kδ moderately reduced both cell spreading and actin polymerization (Fig. 5C,D). To test responses under shear flow conditions, we tracked cells migrating on ICAM-1-coated surfaces with a shear rate of 800 s⁻¹ and calculated MI. Note that the procedures used for retroviral transduction and selection diminished the magnitude of upstream migration in all T cell populations, including the NT control. Nonetheless, T cells expressing either NT or PI3Kδ gRNAs remained capable of migrating upstream on ICAM-1, whereas T cells expressing the cCbl gRNA failed to migrate upstream (Fig. 5E). This pattern also held true when analyzing the percentage of cells migrating upstream (Fig. 5F). To best visualize this phenotype, we plotted the number of cells as a function of their MI for all the cells analyzed in Fig. 5F (Fig. 5G). When compared to control NT gRNA-treated T cells, a large proportion of T cells lacking cCbl had an MI between 0.1 and 0.4. This was also seen in representative scattergrams (Fig. S4). The finding that T cells expressing the PI3Kδ gRNA showed normal upstream motility was rather surprising, especially because these cells did exhibit diminished spreading and polarization under static conditions (Fig. 5C,D), and a modest decrease in track length under shear flow conditions (Fig. S4). We reasoned that this might be because deletion of PI3Kδ was incomplete. Alternatively, other isoforms of PI3K could play a role. To address these possibilities, we repeated this analysis using T cells treated with the pan-PI3K inhibitor LY294002. As shown in Fig. S5, this inhibitor resulted in overall decrease in directionality but, as with PI3Kδ deletion, treatment with LY294002 had no effect on upstream migration. This finding is consistent with previous work in human T cells (Kim and Hammer, 2019; Valignat et al., 2014). Taken together, these studies identify cCbl, but not PI3K, as a necessary component of the LFA-1 signaling pathway that drives upstream migration.

We and others have previously shown that T cells migrate upstream on ICAM-1 and downstream on VCAM-1 (Anderson et al., 2019; Dominguez et al., 2015; Hornung et al., 2020; Kim and Hammer, 2019; Steiner et al., 2010; Valignat et al., 2014, 2013). In this study, we investigated the role of integrin signaling events in driving upstream migration. We found that engagement of LFA-1 by surface-bound ICAM-1 triggered robust activation of many signaling events as well as cytoskeletal remodeling and cell shape changes, whereas engagement of VLA-4 by VCAM-1 failed to trigger some, but not all, of these events. Importantly, upstream migration on ICAM-1 could be reversed by deleting key proteins associated with LFA-1 signaling. Taken together, our data uncover fundamental signaling differences between LFA-1 and VLA-4 that directly affect T cell migration under shear flow.

Our first indication that LFA-1 signaling drives T cell migration came from our previous studies of the Crk adaptor proteins. We found that T cells lacking Crk proteins have defects in LFA-1 signaling cascades leading to activation of the PI3K pathway and phosphorylation of the ubiquitin ligase scaffold protein cCbl. These defects are associated with impaired T cell spreading, actin polymerization and migration in response to surface-bound ICAM-1 under static conditions (Roy et al., 2018). Here, we show that Crk-deficient T cells fail to migrate against the direction of shear flow on ICAM-1, suggesting a role for signaling in upstream migration. Importantly, although WT T cells responding to ICAM-1 adopt the canonical, flattened morphology with a low-profile leading edge and a raised uropod (Cheung and Ostergaard, 2016; Kelleher et al., 1990; Porter et al., 2002; Roy et al., 2018; Smith et al., 2003), DKO T cells fail to spread; they sit high on the coverslip, even at the front. Thus, even when interacting with ICAM-1, DKO T cells look strikingly similar to WT T cells migrating on VCAM-1.

The phenotype of DKO T cells, which fail to flatten and migrate upstream, fits well with the passive steering mechanism proposed by Valignat and co-workers, who argue that the direction of migration under shear is dictated by the shear forces acting on cells of different shape (Hornung et al., 2020; Valignat et al., 2014). Central to this model is the observation that ICAM-1 induces a spread morphology, with the exception of the aforementioned raised uropod. Shear forces over the spread cell make the uropod act as a ‘wind vane’, aligning the cell to face against the direction of shear. On the other hand, cells migrating on VCAM-1 adhere mostly at the rear, resulting in a raised front. In this instance, shear forces strike the front of the cell, pushing it downstream. In its simplest form, the passive steering mechanism does not explicitly require integrin signaling. Indeed, a signal-independent mechanism was proposed based on the lack of Ca²⁺ influx in cells migrating on either ICAM-1 or VCAM-1 (Hornung et al., 2020). However, although it is clear that Ca²⁺ is important for neutrophil migration (Dixit et al., 2012, 2011; Schaff et al., 2008), a requirement for Ca²⁺ signaling in integrin-dependent T cell migration is not well established. Additionally, the lack of Ca²⁺ signaling does not rule out a role for other signaling events. Numerous kinases and adaptor proteins are phosphorylated downstream of integrin engagement in T cells (Abram and Lowell, 2009), and at least some of these events depend on Crk protein expression. Thus, we conclude that Crk proteins function as signaling intermediates in the pathway linking LFA-1 engagement and upstream motility. Importantly, signaling and passive steering are not mutually exclusive. Indeed, we believe that the two are inextricably linked. We postulate a two-step mechanism in which Crk-dependent signaling downstream of LFA-1 drives cell shape changes, thereby allowing the cell to undergo passive steering.

To further explore the role of signaling in T cell migration under shear flow, we compared signaling events downstream of LFA-1 versus VLA-4 engagement. We found that engagement of LFA-1 by surface-bound ICAM-1 triggers robust signaling, including the activation of the PI3K and ERK pathways, as well as the phosphorylation and activation of scaffold and signaling proteins such as cCbl, CasL, and Pyk2. In contrast, engagement of VLA-4 by VCAM-1 does not trigger substantial activation of PI3K or ERK, nor phosphorylation of cCbl. We were particularly interested in the LFA-1-dependent activation of PI3K and cCbl, because these events are also blunted in Crk-deficient T cells (Roy et al., 2018). While knockout of PI3Kδ or pharmacological inhibition of PI3K had modest effects on track length and displacement, neither manipulation affected the ability of T cells to migrate upstream, consistent with earlier studies (Kim and Hammer, 2019; Valignat et al., 2014). In contrast, knockout of cCbl did perturb upstream migration. Curiously, however, cCbl KO T cells looked morphologically indistinguishable from WT T cells on ICAM-1. This represents an interesting exception to the idea that signaling works by inducing cell shape changes that allow passive steering, and indicates that cCbl promotes upstream migration through a distinct mechanism.

How does cCbl control T cell migration under shear flow? We showed previously that cCbl interacts with the p85 subunit of PI3K after LFA-1 ligation (Roy et al., 2018), but this does not seem to contribute to upstream migration, because inhibition of PI3K activity had no effect. In addition to p85, cCbl is known to interact with dozens of different proteins in a cell-type-dependent manner (Schmidt and Dikic, 2005). Thus, one or more other binding partners could affect T cell migration. In addition to its scaffold function, cCbl functions as a RING finger ubiquitin ligase. Indeed, most of the work studying cCbl in T cells has focused on the ubiquitin ligase activity of the protein, and it is generally accepted that cCbl acts as a negative regulator of T cell receptor signaling (Balagopalan et al., 2007; Murphy et al., 1998; Naramura et al., 1998; Rao et al., 2000; Thien et al., 2005, 1999). The role of cCbl-dependent ubiquitylation in T cell migration is largely unexplored. Since ubiquitylation can control protein degradation, trafficking or functioning, it is attractive to speculate that cCbl promotes upstream migration under shear flow conditions by ubiquitylating proteins needed for T cell migration. Future work aimed at determining the molecular mechanism of cCbl function will provide valuable new insights into T cell migration.

Taken together, our findings demonstrate a clear role for integrin-dependent signaling during T cell migration, and reveal that LFA-1 and VLA-4 deliver distinct signals that direct different migratory behaviors under both static and shear-flow conditions. Since many vessels express multiple integrin ligands, and ligand levels change with tissue type and inflammatory status, it will be important going forward to understand how T cells integrate integrin signaling pathways. Indeed, integrin crosstalk has already been shown to play a role during upstream migration (Buffone et al., 2019; Kim and Hammer, 2019). It will also be important to test how differential integrin signaling influences immune cell trafficking in vivo, especially during infiltration into sites of inflammation and solid tumors. A greater understanding of these events could guide the rational design of therapeutics to alter immune cell migration.

Anti-CD3 clone 2C-11 and the anti-CD28 clone PV1 were obtained from BioXCell. Anti-pTyr clone PY-20 was from Upsate (Millipore). Anti-HEF1 (CasL) clone 2G9 and anti-Pyk2 clone YE353 were obtained from Abcam. Anti-pERK (9101), anti-pAKT (4060), and anti-c-Cbl (2747) were from Cell Signaling Technology. Anti-AKT (559028) was from BD. Secondary antibodies conjugated to appropriate fluorophores and AlexaFluor-conjugated phalloidin were obtained from Thermo Fisher Scientific. Recombinant mouse ICAM-1-Fc and VCAM-1-Fc were purchased from R&D Systems. Antibodies for flow cytometry were from BioLegend. These included rat anti-CD4 APC (clone RM4-5), anti-CD49d FITC (clone R1-2), anti-CD49e PE-Cy7 (clone 5H10-27), anti-CD49f Pac-blue (clone GoH3), anti-CD11a PE (clone M17/4), anti-CD11b APC (clone M1/70), anti-CD29 488 (clone HMβ1-1), anti-CD18 PE (clone M18/2), and anti-ITGB7 PerCP/Cy5.5 (clone FIB27). All antibodies were used at a 1:100 dilution for flow cytometry. All antibodies were validated by the manufacturer.

All mice were housed in the Children's Hospital of Philadelphia animal facility, according to guidelines put forth by the Institutional Animal Care and Use Committee. C57BL/6 mice, originally obtained from The Jackson Laboratory, were used as a source of WT T cells. Mice expressing Lifeact–GFP have been described previously (Riedl et al., 2010). Additionally, mice lacking the Crk adaptor proteins in T cells (herein referred to as DKO) were generated by crossing CD4⁺ Cre mice with mice in which the two Crk loci have been floxed (Crk^fl/fl:CrkL^fl/fl mice) (Huang et al., 2015; Park and Curran, 2008). To generate DKO mice expressing Lifeact–GFP, Crk^fl/fl:CrkL^fl/fl mice were crossed with Lifeact–GFP mice, and the resulting Lifeact–GFP: Crk^fl/fl:CrkL^fl/fl mice were crossed with CD4⁺ Cre mice. Mice expressing Cas9 (C57BL/6J-congenic H11^Cas9; Jackson stock no. 028239) were used for CRISPR studies. All mice were 8–14 weeks of age when killed, and both males and females were used as a source of T cells.

Primary mouse CD4⁺ T cells were purified from lymph nodes and spleens by negative selection. Briefly, after removing red blood cells by incubation for 1 min in ACK lysis buffer (77 mM NH₄Cl, 5 mM KHCO₃ and 55 μM EDTA), cells were washed and incubated with anti-MHCII and anti-CD8 hybridoma supernatants (M5/114 and 2.43, respectively) for 20 min at 4°C. After washing, cells were mixed with anti-rat Ig magnetic beads (Qiagen BioMag), incubated for 15 min at 4°C, and subjected to three rounds of magnetic separation using a benchtop magnet. The resulting CD4⁺ T cells were then immediately activated on 24-well plates coated with anti-CD3 and anti-CD28 (2C11 and PV1, 1 μg/ml each) at 1×10⁶ cells per well. Activation was done in T cell complete medium, composed of DMEM (Gibco 11885-084) supplemented with 5% FBS, penicillin/streptomycin, non-essential amino acids, Glutamax, and 2 μl 2-mercaptoethanol. Unless otherwise indicated, all tissue culture reagents were from Gibco. After 48 h, cells were removed from activation and mixed at a 1:1 volume ratio with complete T cell medium containing recombinant IL-2 (obtained through the NIH AIDS Reagent Program, Division of AIDS, NIAID, NIH from Dr Maurice Gately, Hoffmann - La Roche Inc), to give a final IL-2 concentration of 20 units/ml. T cells were used at days 5–6 after activation.

Cells were first stained with Live/Dead Aqua (ThermoFisher) in PBS following the manufacturer's protocol. Staining was quenched using 1% bovine serum albumin (BSA) solution. For staining, antibodies were diluted 1:100 in FACS buffer (PBS, 5% FBS, 0.02% NaN₃, and 1 mM EDTA). Flow cytometry was performed using either a Cytoflex LX or CytoFlex S cytometer (Beckman Coulter) and data was analyzed using FlowJo software (FlowJo LLC). T cells were gated based on size, live cells, and expression of CD4⁺.

96-well plates (MaxiSorp, ThermoFisher) were coated with 2 μg/ml mouse ICAM-1, VCAM-1, or both ligands mixed (1 μg/ml each), in PBS overnight at 4°C. Plates were then washed three times with PBS, blocked with 1% BSA in PBS for 1 h at room temperature, and washed twice with PBS. Activated CD4⁺ T cells were used on day 5 after initial isolation. To prepare the cells, they were first labeled with Calcein AM (ThermoFisher) at a final concentration of 2.5 μM for 30 min at 37°C in serum-free DMEM. Cells were then washed and resuspended in T cell complete medium, and incubated at 37°C for 30 min. Cells were washed and resuspended in 2.5% BSA in PBS (with Ca²⁺ and Mg²⁺) and 1×10⁵ cells were added to each well on ice in triplicate. After a 20 min incubation, the plate was read on a Bio-Tek Synergy HT fluorescence plate reader to obtain the baseline measurements representing ‘maximum adhesion’ per well. The plate was then incubated at 37°C for 10 min, then washed and read again. The plate was washed and read a total of 2–4 times, until the signal from the unstimulated control was stable. To calculate the percentage adhesion, the fluorescence per well after washes was divided by the initial ‘maximum adhesion’ fluorescence reading per well. Background was subtracted using values obtained from empty wells.

Lab-Tek 8-chamber slides (ThermoFisher) were coated overnight at 4°C with 2 μg/ml ICAM-1, VCAM-1, or both ligands mixed (1 μg/ml each). For supplemental experiments using excess VCAM-1, VCAM-1 was coated at 10 μg/ml. Activated CD4⁺ T cells were resuspended in Leibovitz's L-15 medium (Gibco) supplemented with 2 mg/ml glucose and incubated at 37°C for 20 min. T cells were then added to the chamber slides for 20 min at 37°C, followed by fixation in 3.7% paraformaldehyde in PBS. Cells were then blocked and permeabilized in PSG (PBS, 0.01% saponin, 0.05% fish skin gelatin) for 20 min, followed by 45 min with fluorescent phalloidin (Molecular Probes) in PSG. Cells were washed in PSG, mounted, and imaged using a 63× PlanApo 1.4 NA objective on an Axiovert 200 M (Zeiss) with a spinning disk confocal system (Ultraview ERS6; PerkinElmer). Four z-planes spanning a total of 0.75 μm were collected at the cell–surface interface using an Orca Flash 4.0 camera (Hamamatsu). Image analysis of the rendered stacks was conducted using Volocity v6.3 software. Cells were identified using the ‘Find Objects’ command, using a low threshold on the actin channel, and total phalloidin staining was quantified per cell based on integrated pixel intensity.

For the visualization of F-actin in live cells, Lifeact–GFP mice served as the source of CD4⁺ T cells. Activated CD4⁺ T cells were added to Lab-Tek 8-chamber slides for 20 min, gently washed in warm L-15 medium, mounted, and imaged at 37°C using the spinning disk confocal system. For migration movies, 9 z-planes spanning a total of 2 μm were collected every 2 s. Projections of the full 2 μm were prepared in Fiji. For whole-cell reconstruction, 49 z-planes spanning a total of 12 μm were captured, and rendered using the orthogonal viewing tool in Fiji.

Surfaces were prepared by stamping 2 μg/ml of protein A/G (Biovision, San Francisco, CA) using PDMS stamps, onto UV ozone-treated, PDMS spin-coated glass slides. Surfaces were then blocked with 0.2% Pluronic F-127 (Sigma), washed, and subsequently incubated with 2 μg/ml of murine ICAM-1 Fc, VCAM-1 Fc, or a 1:1 mixture at 4°C overnight. Activated CD4⁺ T cells were resuspended in RPMI-1640 supplemented with 2 mg/ml D-glucose and 0.1% BSA and introduced into a flow chamber using a syringe. T cells were allowed to adhere for 30 min prior to the onset of shear (rate: 100 s⁻¹ or 800 s⁻¹). Cells were then imaged every minute for 30 min using a Nikon TE300 with a custom built environmental chamber at 37°C and 5% CO₂. For drug experiments, either DMSO or LY294002 (50 μM) were present for 50 min prior to the onset of shear, followed by 30 min of imaging under shear flow conditions. The movies were exported to ImageJ and cells were tracked and analyzed using the manual tracking plugin in ImageJ and a custom MATLAB script. Only cells that stayed in the field-of-view the whole movie were included in the analysis. Total track length, total displacement, and migration index were calculated. Migration Index (MI) describes the directionality of a cell by quantifying a ratio of a cell's axial displacement to total distance it has traveled. With flow from left to right of field-of-view, negative MI represents cells traveling upstream, whereas positive MI denotes cells traveling downstream.

Lab-Tek 8-chamber slides (ThermoFisher) were coated with 2 μg/ml ICAM-1, VCAM-1, or both ligands mixed, overnight at 4°C. Activated CD4⁺ T cells were washed and resuspended in L-15 medium containing 2 mg/ml D-glucose. T cells were then added to the chambers, incubated for 20 min, gently washed to remove all unbound cells, and imaged using a 10× phase contrast objective at 37°C on a Zeiss Axiovert 200M microscope equipped with an automated x–y stage and a Roper EMCCD camera. Time-lapse images were collected every 30 s for 10 min using SlideBook 6 software (Intelligent Imaging Innovations). Movies were exported into ImageJ, and cells were tracked using the manual tracking plugin to calculate speed. Directionality was calculated by taking the ratio of displacement and track length.

The recently described gRNA retroviral transfer vector MRIG (Huang et al., 2019) was modified to express the puromycin resistance gene in place of GFP. Specific gRNAs were cloned into this vector exactly as described by Huang et al. (2019). The gRNA sequences used in this study were as follows: non targeting (NT), 5′-GCGAGGTATTCGGCTCCGCG-3′; cCbl, 5′-TGTCCCTTCTAGCCGCCCAG-3′; and PI3Kδ, 5′-GGAGCGTGGGCGCATCACGG-3′. 293T cells (originally obtained from ATCC and screened periodically for mycoplasma) were transfected with MRIG-puro along with the retroviral packaging vector pCL-eco (Addgene 12371) at a 4:3 ratio using the calcium phosphate method (Kingston et al., 1999) and allowed to incubate overnight. The next morning, the medium was gently replaced, and cells were incubated for 24–30 h. Viral-containing supernatants were harvested and centrifuged at 1000 g for 10 min to remove cellular debris. Polybrene was added to clarified viral supernatants to a final concentration of 8 μg/ml, and supernatants were used immediately for T cell transductions.

CD4⁺ T cells harvested from Cas9-expressing mice were activated on 24-well plates coated with anti-CD3 and anti-CD28 (2C11 5 μg/ml and PV1 2 μg/ml) at 1×10⁶ cells per well in T cell complete medium. After 24 h, the conditioned medium was removed and reserved for later use, and gently replaced with viral supernatants (with polybrene), and incubated at 37°C for 10 min. The plate was then centrifuged at 1100 g for 2 h at 37°C. Directly following spinoculation, the plate was placed back in the incubator for 10 min. Viral supernatants were then removed and replaced with a mixture of fresh complete T cell medium mixed with conditioned medium at a 3:1 volume ratio, after which cells were cultured for an additional 24 h. T cells were then removed from activation, and mixed at a 1:1 volume ratio with complete T cell medium containing recombinant IL-2 to yield a final IL-2 concentration of 20 units/ml. After 4 h, puromycin was added to a final concentration of 4 μg/ml. Cells were used after 3 d of selection (day 5 after isolation).

60-mm tissue culture dishes (Corning, 430166) were coated with 2 μg/ml ICAM-1 or VCAM-1 overnight at 4°C. Activated CD4⁺ T cells were serum starved for 3 h in DMEM lacking all supplements, washed and resuspended in L15 medium supplemented with 2 mg/ml D-glucose, and incubated for 10 min at 37°C. 8.5×10⁶ T cells were then allowed to interact with surfaces coated with ICAM-1 or VCAM-1 for 20 min at 37°C. Cells stimulated on surface-bound ICAM-1 or VCAM-1 were lysed by aspirating the medium and adding 500 μl 1× ice cold lysis buffer (final composition: 1% Triton X-100, 150 mM NaCl, 50 mM Tris-HCl pH 7.5, 10 mM MgCl₂, 5 mM NaF, 1 mM sodium orthovanadate and Roche EDTA-free protease inhibitor cocktail). Unstimulated control cells were lysed in solution by adding an equal volume of 2× cold lysis buffer for a final volume of 500 μl. Lysates were incubated on ice with periodic vortexing for 15 min, followed by centrifugation for 10 min at 16,000 g at 4°C. A 50 μl aliquot of each whole-cell lysate was retained, mixed with 4× sample buffer containing DTT (50 mM final), and heated to 95°C for 10 min prior to separation by SDS–PAGE. The remainder of each lysate was used for immunoprecipitation.

For immunoprecipitation, 60 μl of Protein A agarose bead slurry (Repligen) were washed and resuspended in lysis buffer. Beads were then pre-bound to anti-phosphotyrosine (PY-20, 2.5 μg per condition) overnight at 4°C, with rotation. Antibody-charged beads were then washed three times with lysis buffer, mixed with 450 μl cell lysates, and rotated at 4°C overnight. Beads were then washed three times with lysis buffer, mixed with 30 μl of 2× sample buffer, boiled, and analyzed using SDS–PAGE.

Proteins were separated by SDS–PAGE using the Invitrogen Novex Mini-cell system with NuPAGE 4–12% BisTris gradient gels. Proteins were transferred to nitrocellulose membranes (0.45 μm, BioRad) and blocked using LI-COR blocking buffer mixed 1:1 with PBS for 1 h at room temperature. Primary antibodies were mixed in TBS containing 0.1% tween-20 (TBST) with 2% BSA and incubated with membranes overnight at 4°C on a shaker. Primary antibody dilutions were as follows: anti-HEF1 (1:350), anti-Pyk2 (1:1000), anti-pERK (1:1500), anti-pAKT (1:1000), anti-c-Cbl (1:1000), and anti-AKT (1:1000). Membranes were then washed three times for 10 min each in TBST, and incubated with fluorophore-conjugated secondary antibodies in TBST with 2% BSA for 1 h at room temperature. Membranes were washed three times for 10 min in TBST, and imaged using a LI-COR Odyssey imaging system. Quantification of bands was performed using ImageStudio software (LI-COR), and all measurements were within the linear range.

Statistics were calculated and graphs were prepared using GraphPad Prism 8. When only two groups were being compared, a t-test was used. When more than two groups were compared, a one-way ANOVA was performed using multiple comparisons with a Tukey correction. *P<0.05; **P<0.01; ***P<0.001

We thank the CHOP Flow Cytometry Core for access to flow cytometers. We also thank members of the Burkhardt and Hammer laboratories for discussions and critical reading of the manuscript.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.248328.reviewer-comments.pdf