Integrin crosstalk allows CD4+ T lymphocytes to continue migrating in the upstream direction after flow

Abstract

In order to perform critical immune functions at sites of inflammation, circulatory T lymphocytes must be able to arrest, adhere, migrate and transmigrate on the endothelial surface. This progression of steps is coordinated by cellular adhesion molecules (CAMs), chemokines, and selectins presented on the endothelium. Two important interactions are between Lymphocyte Function-associated Antigen-1 (LFA-1) and Intracellular Adhesion Molecule-1 (ICAM-1) and also between Very Late Antigen-4 (VLA-4) and Vascular Cell Adhesion Molecule-1 (VCAM-1). Recent studies have shown that T lymphocytes and other cell types can migrate upstream (against the direction) of flow through the binding of LFA-1 to ICAM-1. Since upstream migration of T cells depends on a specific adhesive pathway, we hypothesized that mechanotransduction is critical to migration, and that signals might allow T-cells to remember their direction of migration after the flow is terminated. Cells on ICAM-1 surfaces migrate against the shear flow, but the upstream migration reverts to random migration after the flow is stopped. Cells on VCAM-1 migrate with the direction of flow. However, on surfaces that combine ICAM-1 and VCAM-1, cells crawl upstream at a shear rate of 800 s⁻¹ and continue migrating in the upstream direction for at least 30 minutes after the flow is terminated—we call this ‘migrational memory’. Post-flow upstream migration on VCAM-1/ICAM-1 surfaces is reversed upon the inhibition of PI3K, but conserved with cdc42 and Arp2/3 inhibitors. Using an antibody against VLA-4, we can block migrational memory on VCAM-1/ICAM-1 surfaces. Using a soluble ligand for VLA-4 (sVCAM-1), we can promote migrational memory on ICAM-1 surfaces. These results indicate that, while upstream migration under flow requires LFA-1 binding to immobilized ICAM-1, signaling from VLA-4 and PI3K activity is required for the migrational memory of CD4+ T cells. These results indicate that crosstalk between integrins potentiates the signal of upstream migration.

Insight box

Immune cells circulate in the blood vessels to home to sites of inflammation. During homing, immune cells migrate under shear stress on endothelium expressing adhesion molecules before exiting blood vessels. T cells can migrate against the direction of flow if a specific receptor is engaged. Here, we have assessed the mechanisms behind upstream migration by determining the conditions that govern persistence in directional motion after the flow is stopped. CD4+ T cells migrate upstream and can continue upstream migration only if two receptors are simultaneously engaged. We have discovered a key intracellular signaling molecule that enables communication between the two receptors. This work elucidates the underlying mechanisms of upstream migration and provides insight into the mechanochemical control of T cell migration.

INTRODUCTION

Cell migration is a crucial for T lymphocytes to home and reach sites of inflammation [1]. T lymphocytes need to stop, roll and adhere on the endothelial layer to reach the sites of inflammation [2, 3]. This process, known as the leukocyte adhesion cascade, is regulated by multiple adhesive interactions and intracellular signaling cascades. In response to inflammatory stimuli, vascular endothelial cells express chemokines and cellular adhesion molecules (CAMs) on their surface to recruit T lymphocytes to the sites of inflammation [4, 5]. To facilitate firm adhesion and migration along the endothelium, T lymphocytes express the integrins Lymphocyte Function-associated Antigen-1 (LFA-1, α_Lβ₂) and Very Late Antigen-4 (VLA-4, α₄β₁) to bind to Intracellular Adhesion Molecule-1 (ICAM-1) or Vascular Cell Adhesion Molecule-1 (VCAM-1), respectively, on the endothelial surface. After firm adhesion, T lymphocytes become migratory, first migrating on the apical surface of endothelial cells, and then transmigrating. Despite the obvious importance of hydrodynamic interactions in the vasculature, T lymphocyte migration is often studied in the absence of external shear flow. Not only do leukocytes have to withstand shear forces within blood vessel while rolling and adhering to surfaces and move, but migration involves spreading and cell reorganization under hydrodynamic forces to generate directional motion.

The complex signaling cascade activated downstream of integrin binding has been well studied in T lymphocytes in the absence of shear flow. Upon ligand binding, integrins recruit adaptor proteins to the cytoplasmic tail of integrins, connecting the integrins to cytoskeleton for force transmission and movement [6–13]. Scaffolding molecules propagate the extracellular signal to PI3K and Rho-family small GTPases for migration by subsequent recruitment and activation. Phosphatidylinositol 3-kinase (PI3K) on the plasma membrane promotes further recruitment and activation of GTPases to the plasma membrane [14, 15]. Rho-family small GTPases are often involved in establishing cell polarity and fostering actin polymerization [16]. Cdc42 and Rac localize at the leading edge of cells promoting actin branching via N-WASP and Arp2/3, and WAVE in the case of Rac [15, 17–19]. Rho, on the other hand, regulates myosin-dependent contractility at the rear end of cells [20, 21]. However, it is unknown how these signaling molecules affect the directionality of migration in response to shear flow.

Interestingly, it has been reported that T cells can crawl upstream against the direction of flow, depending on the chemistry of the surface. Valignat and coworkers reported that primary T lymphocytes orient themselves against the direction of flow and crawl upstream on surfaces coated with ICAM-1 and the chemokine SDF-1α [22]. Upstream migration has been observed in vivo; effector T cells exhibited intravascular upstream migration in a rat model of neurological autoimmune lesions [23]. A study in our own group showed that only a small amount of LFA-1-ICAM-1 interactions were necessary for the upstream migration of T lymphocytes; on surfaces coated with both VCAM-1 and ICAM-1, T lymphocytes migrated upstream even when ICAM-1 represented less than 10% of the substrate ligand, reinforcing upstream migration dictated by the LFA-1 and ICAM-1 interactions [24]. Upstream migration on ICAM-1 is not limited to T lymphocytes but has also been demonstrated by hematopoietic stem and progenitor cells (HSPCs), marginal zone B-cells, and neutrophils when Mac-1 was blocked to isolate LFA-1/ICAM-1 interactions [25–28].

We have started to search for clues regarding the mechanisms behind upstream migration. Previously, we showed that after subjecting cells to repeated cycles in which the flow was maintained for 10 minutes and then stopped for 10 minutes, cells on ICAM-1 surfaces moved randomly when the flow was stopped, while cells on surfaces presenting a mixture of VCAM-1/ICAM-1 persisted in crawling against the direction in which the flow had been, even after the flow was stopped [24]. In other words, the cells displayed directional persistence, which we call ‘migrational memory’, after the flow was terminated on surfaces in which both VCAM-1 and ICAM-1 are mixed.

Here, we aim to expand this investigation and address how CD4+ T lymphocytes maintain their directionality. Previously, we found that migrational memory was maintained for 10 minutes after the flow was terminated on surfaces that were functionalized with both VCAM-1 and ICAM-1 [24]. We ask now, how much longer do cells migrate upstream after flow, and what role does intracellular signaling and integrin crosstalk play in persistence? We indeed find that although the engagement of LFA-1 with ICAM-1 is absolutely required for upstream migration, the independent activation of VLA-4 can stimulate a persistent memory of the direction of upstream migration on ICAM-1 surfaces. Furthermore, we show that the activity of PI3K is needed for persistent memory; inhibitors of PI3K induce CD4+ T cells to revert to random migration when the flow is terminated. Surprisingly, based on studies using appropriate inhibitors, GTPases and Arp2/3 have no effect in remembering the direction of upstream migration. Therefore, we propose that directional memory requires the crosstalk between LFA-1 and VLA-4, and signaling through PI3K.

MATERIALS AND METHODS

Cell culture and reagents

Human Primary CD4+ and CD8+ T lymphocytes were acquired from the Human Immunology Core at the University of Pennsylvania (P30-CA016520). Cells were activated with PHA (MP Biomedicals, Santa Ana, CA) in RPMI-1640 with 10% FBS for 3 days, followed by subsequent activation with IL-2 (Corning, Corning, NY) in RPMI-1640 with 10% FBS and penicillin and streptomycin.

For shear flow experiments, cells were kept in RPMI-1640 supplemented with 2-mg/ml D-glucose and 0.1% BSA. CD4+ T cells were introduced into a flow chamber using a syringe, adhered for 30 minutes to surfaces prior to shear flow, and imaged. CD4+ T lymphocytes were treated with inhibitors 30 minutes prior to seeding, unless noted otherwise. The inhibitors were present in media at all times throughout the experiment. Pharmacological inhibitors used for experiments were the following: Wortmannin (Sigma, St. Louis, MO) LY294002 (CST, Beverly, MA), CT-04 (Cytoskeleton, Denver, CO), NSC23766 (Millipore, Temecula, CA), ML-141 (Millipore), and CK666 (Sigma). For modulating VLA-4 ligand binding, cells were incubated with the following for 15 minutes: soluble VCAM-1 (R&D Systems, Minneapolis, MN) and blocking antibodies, isotype control, anti-α₄ (9F10) or anti-β₁ (P5D2) (Biolegend, San Diego, CA).

Substrate preparation

Substrates were prepared as explained previously [29–31]. Briefly, 25 mm × 75 mm × 1 mm glass slides (Thermo-Fisher, Hampton, NH) are spin-coated with degassed Poly (dimethylsiloxane) (PDMS) cross-linked in 10:1 ratio with its curing agent (Dow Corning, Midland, MI). Spin-coated PDMS slides are then allowed to cross-link at 65°C for 1 hour. PDMS slides were then treated with UV for 8 minutes prior to printing.

Preparation of chimeric areas using microcontact printing

A 10:1 PDMS base:curing agent mix was used to make PDMS stamps for microcontact printing. Degassed PDMS solution was cured over a flat silicon wafer for an hour at 65°C. Then, cured PDMS was cut into cubes with 1 cm² flat stamping surface area. Trimmed stamps were sonicated in 200 proof ethanol for 10 minutes and rinsed with diH₂O. Surfaces were prepared by microcontact printing 2 μg/ml of protein A/G (Biovision, San Francisco, CA) onto UV ozone-treated PDMS spin-coated glass slides at room temperature. Surfaces were then blocked with 0.2% (w/v) Pluronic F-127 (Sigma). Subsequently, 10 μg/ml of ICAM-1 Fc chimera (R&D Systems), 10 μg/ml of VCAM-1 Fc chimera (R&D Systems), or 10 μg/ml consisting 1:1 concentration of VCAM-1 and ICAM-1 Fc chimera adhesion molecules were adsorbed at 4°C overnight to create ICAM-1, VCAM-1 or VCAM-1/ICAM-1 mixed surfaces, respectively.

Cell tracking and data acquisition

Images were acquired every minute for an hour using a Nikon TE300 with custom environmental control chamber at 37°C and 5% CO₂. Cells were exposed to first 30 minutes of shear flow, followed by 30 minutes of no flow (denoted as ‘post-flow’). Two shear rates, 100 and 800 s⁻¹, were used. While static conditions last 30 minutes, flow and post-flow groups were taken for an hour. Images were analyzed using Manual Tracking (https://imagej.nih.gov/ij/plugins/track/track.html) in ImageJ (https://imagej.nih.gov/ij/, NIH, Bethesda, MD) and MATLAB (The MathWorks, Natick, MA). Centroids of cells were tracked using Image plug-in Manual Tracking. Cells only stayed in the field-of-view for an entire hour were included in the analysis. With (x, y) coordinates of cells, Migration Index (MI), speed, and persistence time were calculated using a custom MATLAB script. MI describes the directionality of cells 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, while positive MI denotes cells traveling downstream. Persistence time was obtained by calculating mean squared displacement (MSD) of each cell and fitting to Dunn Equation [32].

RESULTS

CD4+ T lymphocytes retain directional persistence post flow at a high shear rate using both VLA-4 and LFA-1

It has been previously shown that T lymphocytes migrate against the direction of flow on surfaces with immobilized ICAM-1 [22, 24, 28]. Dominguez and coworkers had tested cells on various ratios of ICAM-1 and VCAM-1 at a shear rate of 800 s⁻¹, and discovered that cells migrated against the direction of flow whenever there was any ICAM-1 on the surface. Upstream migration with the combinations of ICAM-1 and VCAM-1 did not depend on the concentration of VCAM-1 present. Furthermore, when the flow at a shear rate of 800 s⁻¹ was stopped, only cells on surfaces in which VCAM-1/ICAM-1 had been mixed maintained upstream direction [24]. Here, we expand on this study and further investigate the factors that control T cell memory of directional motion in the absence of flow. Based on our previous results [24], we hypothesized that VLA-4 engagement is important for the persistence of motion when the flow is terminated. On three different surfaces, ICAM-1, a mixture of VCAM-1/ICAM-1, and VCAM-1, CD4+ T lymphocytes were exposed to shear flow for 30 minutes, then tracked for another 30 minutes after the flow was terminated. Figure 1 shows cells traces from a representative experiment.

Figure 1.

CD4+ T lymphocytes on ICAM-1, VCAM-1/ICAM-1, or VCAM-1 surfaces migrate according to the direction of flow, but only cells on surfaces with a mixture of VCAM-1/ICAM-1 maintain upstream directionality after the flow is terminated, and only after having been exposed to flow at a shear rate of 800 s⁻¹. (A) CD4+ T lymphocytes on ICAM-1 under flow at a shear rate of 800 s⁻¹ at time points 0, 15 and 30 minutes, followed by 30 minutes of no flow at time points 0, 15 and 30 minutes. (B) CD4+ T lymphocytes on VCAM-1/ICAM-1 surfaces under flow at a shear rate of 800 s⁻¹ at time points 0, 15 and 30 minutes, followed by 30 minutes of no flow at time points 0, 15 and 30 minutes. (C) CD4+ T lymphocytes on VCAM-1 mixed surface under flow at a shear rate of 800 s⁻¹ at time points 0, 15 and 30 minutes, followed by 30 minutes of no flow at time points 0, 15 and 30 minutes. Points indicate the end of the cell track. Scale bar = 150 μm.

CD4+ T lymphocytes on ICAM-1 surfaces migrated upstream at shear rates of 100 and 800 s⁻¹ (Fig. 2A-a,c) during flow. Such upstream migration, represented by a negative Migration Index (MI) (Fig. 2D and Supplementary Fig. 1C), is more prominent at a shear rate of 800 s⁻¹, which is consistent with previously published studies [24]. However, on ICAM-1 surfaces, after the flow is turned off, CD4+ T lymphocytes show no memory of their directional motion and migrate randomly (Fig. 2A-b,d, D and G). No significant difference in speed and persistence times was detected (Supplementary Fig. 1 A–C) after the flow was terminated. While LFA-1 engaging ICAM-1 is required for upstream migration, the binding is not sufficient to maintain directional persistence once the flow is stopped.

Figure 2.

CD4+ T lymphocytes on ICAM-1 or VCAM-1/ICAM-1 mixed surfaces display upstream migration, and downstream migration on VCAM-1, but only cells on VCAM-1/ICAM-1 mixed surface display migrational memory after having been exposed to flow at a shear rate of 800 s⁻¹. (A) Scattergrams of cells on ICAM-1 at low (100 s⁻¹, a and b) and high (800 s⁻¹, c and d) shear rates. (B) Scattergrams of cells on VCAM-1/ICAM-1 at low (100 s⁻¹, a and b) and high (800 s⁻¹, c and d) shear rates. (C) Scattergrams of cells on VCAM-1 at low (100 s⁻¹, a and b) and high (800 s⁻¹, c and d) shear rates. Migration Index of cells on (D) ICAM-1, (E) VCAM-1/ICAM-1 mixed surface, and (F) VCAM-1. Migration Index of cells on (G) ICAM-1, (H) VCAM-1/ICAM-1 mixed surface, and (I) VCAM-1, plotted over time after the flow is turned off. Asterisk: P < 0.05 compared to MI = 0. Red tracks indicate cells moving against the direction of flow, and gray tracks are cells moving with the direction of flow. The direction of flow is from left to right.

On surfaces in which VCAM-1/ICAM-1 are mixed, CD4+ T lymphocytes migrated against shear flow at shear rates of 100 and 800 s⁻¹ (Fig. 2B-a,c). Similar to CD4+ T lymphocytes on ICAM-1 surfaces, cells on VCAM-1/ICAM-1 mixed surfaces showed a shear rate dependence of upstream migration (Fig. 2E), crawling more avidly upstream at higher shear rates. While upstream migration requires from LFA-1-ICAM-1 interactions, VLA-4-VCAM-1 binding does not interfere with upstream migration. After the flow is turned off, CD4+ T lymphocytes on VCAM-1/ICAM-1 mixed surfaces maintained the upstream orientation (Fig. 2B-b,d and E). This upstream migration can persist for as long as 30 minutes, indicated by negative MI over this time (Fig. 2H). To observe persistent migration in the upstream direction, a shear rate of 800 s⁻¹ was needed. At 100 s⁻¹ shear rate, cells exhibited no collective memory of upstream migration, even on surfaces in which VCAM-1 and ICAM-1 were mixed. Thus, we conclude that while the direction of migration during flow is directed by LFA-1, the persistence of memory of direction post flow is dependent on VLA-4-VCAM-1 interactions and requires a high level of stimulus (higher shear rates).

We additionally confirmed that CD4+ T cells on VCAM-1 surfaces exhibit a strong downstream migration under flow at shear rates of 100 and 800 s⁻¹ (Fig. 2C), as demonstrated by a positive MI (Fig. 2F). This observation was consistent with previous results [24]. Cells on VCAM-1 migrated with the direction of flow and continue its direction of motion post flow (Fig. 2I). Since cells on VCAM-1 never migrate upstream, they do not have any upstream migration to remember.

CD8+ T cells also migrated upstream under flow on surfaces containing ICAM-1. However, unlike CD4+ T cells, CD8+ T cells exhibited random migration once the flow was turned off on all surfaces (Supplementary Fig. 2). These results highlight that while upstream migration on ICAM-1 is consistent with CD4+ and CD8+ subsets, only CD4+ T cells remember the upstream migration on VCAM-1/ICAM-1 mixed surfaces. To concentrate on studying how cells remember the upstream migration post flow, we focused solely on CD4+ T lymphocytes on surfaces containing ICAM-1 or a mixture of VCAM-1/ICAM-1 at a shear rate of 800 s⁻¹ for the remainder of the study.

PI3K is required for persistent upstream migration post flow in CD4+ T lymphocytes

Integrin-ligand binding initiates a wide range of intracellular signaling pathways which determine cell polarity and involving the recruitment of actin or microtubules to power movement [33, 34]. Using our result that CD4+ T cells persist in the upstream direction after a shear rate of 800 s⁻¹ on surfaces in which VCAM-1 and ICAM-1 are mixed, we investigated the role of PI3K in maintaining upstream migration using a PI3K inhibitor, wortmannin. On VCAM-1/ICAM-1 mixed surfaces, inhibiting PI3K did not block upstream migration during flow at a shear rate of 800 s⁻¹ (Fig. 3A-a–e and B). After the flow is removed, CD4+ T lymphocytes on VCAM-1/ICAM-1 surfaces with wortmannin lost their memory of upstream migration and showed no preference in direction (Fig. 3A-f–j and B). We measured the MIs continuously during the entire period during and after flow. During the 30 minutes of flow, the MI stayed negative, indicating PI3K inhibition did not block upstream migration (Supplementary Fig. 3D). After the flow is turned off, the MI rapidly approached zero, unlike the MI of the DMSO control group which remained negative, signifying PI3K inhibition led to random migration after the flow was terminated (Fig. 3D). At different concentrations of Wortmannin, CD4+ T cells on VCAM-1/ICAM-1 all preserved upstream migration under flow but lost the directionality post flow (Supplementary Fig. 3). Thus, on VCAM-1/ICAM-1 surfaces, without proper PI3K activity, CD4+ T lymphocytes lose migrational memory. Treating CD4+ T cells with Wortmannin on ICAM-1 did not interfere with upstream migration under flow (Supplementary Fig. 4). Wortmannin also did not affect the memory effect of CD4+ T cells on ICAM-1, as cells on ICAM-1 never remembered the upstream migration to begin with. This indicates that reduction in PI3K activity does not affect upstream migration when flow is present. Inhibiting with another PI3K inhibitor, LY294002, exhibited consistent results on both types of surfaces as the one with wortmannin (Supplementary Fig. 5). We conclude that PI3K is not imperative for upstream migration under flow, but necessary for migrational memory.

Figure 3.

CD4+ T lymphocytes on VCAM-1/ICAM-1 mixed surfaces lose migrational memory upon inhibition of PI3K with Wortmannin. (A) Scattergrams of cells with a range of dose of Wortmannin during flow with 800 s-1 shear rate (a–e) and post-flow (f–j). Red tracks indicate cells moving against the direction of flow, and gray tracks are cells moving with the direction of flow. The direction of flow is from left to right. (B) Migration Index of cells with DMSO or 50 nM of Wortmannin on VCAM-1/ICAM-1 mixed surface after 30 minutes of each flow condition. (C) Speed of cells with DMSO or 50 nM of Wortmannin on the mixed surface after 30 minutes of each flow condition. (D) Migration Index of cells with DMSO or 50 nM of Wortmannin plotted over the course of 30 minutes of post flow. NS: Not significant compared to DMSO control of corresponding flow condition.

Upstream migration both during and post flow is maintained independent of GTPases

We next hypothesized that cdc42, which is known to establish polarity in cell migration [15, 17–19], would be required to preserve the upstream direction after the flow is removed. We inhibited cdc42 with ML-141 to test whether cdc42 affects post-flow polarity and the persistence of directional migration. Inhibiting cdc42 with ML-141 had no significant effect on CD4+ T lymphocytes migrating upstream on either ICAM-1 or VCAM-1/ICAM-1 surfaces (Fig. 4). In the presence of ML-141, CD4+ T lymphocytes on ICAM-1 showed upstream migration under flow (Fig. 4A and B), but no preferential direction of migration after the flow was removed (Fig. 4C). In the presence of ML-141, CD4+ T cells on VCAM-1/ICAM-1 mixed surfaces preserved their upstream orientation during and post flow (Fig. 4D–F). This result suggests that CD4+ T cells with reduced cdc42 activity were still able to migrate upstream. Cdc42 inhibition also had no effect in migration after the flow has turned off. Inhibiting Rho or Rac also had no effect on upstream migration under flow or post-flow migratory behavior on both surfaces (Supplementary Fig. 6). Together, we conclude that inhibiting Rho, Rac, or cdc42 does not affect the ability to migrate upstream or to remember the upstream migration post flow.

Figure 4.

Inhibiting Cdc42 with 10 μM of ML-141 had no effect in migration during or after flow of CD4+ T lymphocytes. (A) Migration Index of cells on ICAM-1 after 30 minutes. (B and C) Migration Index plotted over the course of 30 minutes of each flow condition. (D) Migration Index of cells on VCAM-1/ICAM-1 after 30 minutes. (E and F) Migration Index plotted over the course of 30 minutes of each flow condition. Asterisk: P < 0.05 compared to MI = 0. NS: Not significant compared to DMSO control of corresponding flow condition.

Actin branching at lamellipodia also has no role in upstream migration or directional persistence in CD4+ T lymphocytes

At the leading edge of a moving cell, actin networks are constantly being restructured to foster cellular movement. Actin filaments are branched into mesh-like networks at lamellipodia, guided by Arp2/3 complexes. Since the actin network is the main driving force of lamellipodial extension in a moving cell, we hypothesized that hindering activity of Arp2/3 and reorganization of actin filaments would obliterate upstream migration and post-flow migrational memory. To investigate the effect of actin branching in migration under shear flow, CD4+ T lymphocytes were treated with different concentrations of CK666 on ICAM-1 or ICAM-1/VCAM-1 mixed surfaces. CD4+ T lymphocytes treated with CK666 on ICAM-1 surfaces robustly migrated against the direction of flow (Fig. 5A and B). Once the flow was removed, cells no longer retained the direction established during shear flow and displayed no significant difference in post-flow migration (Fig. 5C). We noticed a slight decrease in speed upon Arp2/3 inhibition (Supplementary Fig. 7B), but no significant discrepancies in persistence time (Supplementary Fig. 7C). CD4+ T cells on ICAM-1 with different concentrations of CK666 all preserved upstream migration under flow and no longer remember the upstream direction post flow (Supplementary Fig. 7A, G and H). CD4+ T lymphocytes treated with CK666 on VCAM-1/ICAM-1 mixed surfaces also showed upstream migration at a shear rate of 800 s⁻¹ (Fig. 5D and E). After the flow has been turned off, cells still maintained preferential upstream orientation that was established during flow (Fig. 5F). Cells treated with CK666 showed decreased speed, and persistent time was observed during and post flow (Supplementary Fig. 7E and F). CD4+ T cells on VCAM-1/ICAM-1 in other concentrations of CK666 all migrated upstream during and post flow (Supplementary Fig. 7D, I, and J), indicating that Arp2/3 inhibition of CD4+ T cells on VCAM-1/ICAM-1 surfaces had no effect on post-flow migrational memory. Sequestering actin polymerization with Latrunculin A (LatA) and Cytochalasin D (CytoD) completely abolished migration on both ICAM-1 and VCAM-1/ICAM-1 surfaces during and post flow (Supplementary Fig. 8). In conclusion, while actin filaments are required for migration, branch formation with Arp2/3 did not affect migratory patterns during or post flow regardless of which type of integrin was activated.

Figure 5.

CD4+ T lymphocytes maintained upstream migration and displayed migrational memory with 25 μM of Arp2/3 inhibitor, CK666. (A) Migration Index of cells on ICAM-1 after 30 minutes. (B and C) Migration Index of cells on ICAM-1 plotted over the course of 30 minutes of each flow condition. (D) Migration Index of cells on VCAM-1/ICAM-1 after 30 minutes. (E and F) Migration Index of cells on VCAM-1/ICAM-1 plotted over the course of 30 minutes of each flow condition. Asterisk: P < 0.05 compared to MI = 0. NS: Not significant compared to DMSO control of corresponding flow condition.

Altering ligand binding of VLA-4 alone can modulate the directional persistence post flow in CD4+ T lymphocytes

While LFA-1 engaging immobilized ICAM-1 drives upstream migration, we performed additional experiments to verify the role VLA-4 plays in fostering migrational memory. We hypothesized that if a signal is propagated through VLA-4, inhibition of VLA-4 with blocking antibodies would eliminate migrational memory in CD4+ T lymphocytes, whereas independent activation using a soluble ligand (sVCAM-1) would preserve migrational memory.

When CD4+ T cells on VCAM-1/ICAM-1 surfaces were treated with VLA-4 blocking antibodies, forcing cells to only engage ICAM-1 with LFA-1, cells migrated upstream under flow, but no longer retained the directional memory after the flow was turned off (Supplementary Fig. 9D–H). Blocking VLA-4 had no effect on migrational speed (Supplementary Fig. 9F). This strengthens our hypothesis in that VLA-4 binding to VCAM-1 is required for CD4+ T cells to preserve their directional migration after flow.

CD4+ T lymphocytes on ICAM-1 surfaces treated with soluble VCAM-1 (sVCAM-1) migrated upstream under flow and continued the upstream direction post flow (Fig. 6 and Supplementary Fig. 9A–C). No immobilized VCAM-1 was present in these experiments. Activating VLA-4 with soluble VCAM-1 had no effect on speed (Supplementary Fig. 9A). However, soluble VCAM-1 stimulated cells on ICAM-1 surfaces to preserve directional memory after the flow is turned off (Supplementary Fig. 9C). This suggests that simultaneous signaling from both LFA-1 and VLA-4 is required for the directional memory of upstream migration. However, it is the signal, not the adhesive interaction, of VLA-4/VCAM-1 that is required. This result further highlights that, while upstream migration is driven by LFA-1 and surface-bound ICAM-1 on the surface, post-flow upstream migration can be potentiated by VLA-4 binding to either immobilized or soluble VCAM-1.

Figure 6.

CD4+ T lymphocytes on ICAM-1 surfaces migrate upstream during and post-flow in the presence of soluble VCAM-1. (A) Scattergrams of cells on ICAM-1 during (a and b) and post (c and d) flow at a shear rate of 800 s⁻¹. Red tracks indicate cells moving against the direction of flow, and gray tracks are cells moving with the direction of flow. The direction of flow is from left to right. (B) Migration Index of cells on ICAM-1 plotted over the course of 30 minutes of each flow condition. Asterisk: P < 0.05 compared to MI = 0. NS: Not significant compared to isotype control of corresponding flow condition.

DISCUSSION

Intravascular crawling under shear flow is necessary for lymphocytes in order to effectively migrate in blood vessel and perform immunological functions. Direction of migration varies depending on types and concentrations of adhesion molecules cells engage and the shear rates cells experience. Numerous studies have shown different types of cells orienting against the direction of flow in vitro and in vivo. For example, upstream migration under flow has been observed with marginal zone B cells, HSPCs, effector T cells on ICAM-1 and in a rat model, and neutrophils when Mac-1 is blocked [22–28]. However, the intracellular mechanism of upstream migration for leukocytes remains poorly understood. Previously, we investigated how LFA-1 directs upstream migration of T lymphocytes. Cells on combinations of VCAM-1 and ICAM-1 elicited upstream migration under flow. The study further showed that cells on VCAM-1/ICAM-1 mixed surfaces migrated upstream during flow at a shear rate of 800 s⁻¹, and that cells maintained the direction of motion when flow was stopped for 10 minutes [24]. This study expands this work to understand the mechanisms of how cells persist to migrate upstream.

After the flow is removed, CD4+ T cells on ICAM-1 lost their directional signal and migrated at random. Cells on VCAM-1 migrated downstream under flow and continued in this direction when flow was terminated. On the other hand, on surfaces in which VCAM-1 and ICAM-1 are mixed, cells migrate against the direction of flow a shear rate of 800 s⁻¹ maintained their directionality even after the flow had been stopped. Thus, it appears that while LFA-1-ICAM-1 interactions are needed for upstream migration, the engagement of VLA-4 with VCAM-1 is needed for memory. Preservation of upstream migration post flow was not observed with CD8+ T lymphocytes, regardless of types of adhesion molecules or shear rates.

Clearly, some signaling pathway is at the root of memory persistence. Numerous studies have elucidated intracellular signal transduction pathways connecting lymphocyte adhesion to movement. Specifically, PI3K and Rho family of small GTPases are involved in cellular migration and cytoskeletal reorganization. PI3K phosphorylates PIP2 to PIP3, which is implicated in cell polarity and migration, and is often found in the lamellipodia of migrating cells [15, 35, 36]. Roy et al. proposes a model that the engagement of LFA-1 to immobilized ICAM-1 activates Src family kinases, which induces the binding of Crk/CasL to c-Cbl. This complex then activates PI3K catalytic function and promotes the production of PIP3 [35]. However, the kinetics and signaling of PI3K of leukocytes under flow are not well known, because most studies on PI3K have been focused with cells in the absence of shear flow. Here, our data show that inhibiting PI3K has no significant effect on upstream migration under shear flow. During flow, CD4+ T lymphocytes with hindered PI3K activity are still able to orient against the direction of flow on both ICAM-1 and VCAM-1/ICAM-1 mixed surfaces. However, without PI3K, CD4+ T lymphocytes no longer exhibited migrational memory on VCAM-1/ICAM-1 surfaces. Our findings suggest that while VLA-4-VCAM-1 interaction promotes post-flow directional persistence, the ability for VLA-4 to maintain the direction of upstream migration in concert with LFA-1 is PI3K dependent.

Downstream of PI3K, Rho-family GTPases play critical roles in cell migration reorganizing cytoskeleton and plasma membrane. Rho-family GTPases are known to be involved in reorganizing plasma membrane and cytoskeletal structures for efficient migration [17, 37]. Cdc42 and Rho activities in migration essential for chemotaxis [18, 38]. Rho and Rac inhibit each other by localizing at polarizing ends to further establish cell polarity and guide amoeboid motion. Here, we next investigated if GTPases affect migration under shear flow and post-flow directionality. However, inhibiting Rho, Rac, and cdc42 all had no significant effects in upstream migration or post-flow migrational memory. One possibility is that the absence of chemokines in our system also may explain the lack of significant effects upon inhibition of GTPases.

At the leading edge of a moving cell, actin filaments are organized in a mesh-like network at lamellipodia. Arp2/3 initiates actin nucleation at an existing actin filament, in conjunction with actin-polymerizing and depolymerizing factors. While Arp2/3 is highlighted in T cell activation at immunological synapse or migration under static conditions, its role in T cell migration in response to shear flow is poorly understood [39–42]. At a shear rate of 800 s⁻¹, CK666 reduced speed and persistence times but had no effect on the directionality of CD4+ T lymphocytes during flow. CK666 had no effect in post-flow migrational memory as well; cells on VCAM-1/ICAM-1 mixed surfaces with CK666 still maintained persistent directionality post flow. Surprisingly, perturbing Arp2/3 activity with CK666 does not affect upstream migration under flow via LFA-1-ICAM-1 interactions. Upstream migration post flow is regulated independently of Arp2/3. Perturbing actin polymerization with latrunculin A and cytochalasin D completely eliminated migration as both inhibitors disturb actin polymerization and eliminate actin filament formation. As actin filaments are the primary component of lamellipodial cytoskeleton and force generation for movement, this result was expected. These CD4+ T lymphocytes were only able to loosely interact with the surface and roll across the surface, rather than actively migrate. On VCAM-1/ICAM-1 mixed surfaces, a subpopulation of cells would roll and form tether-like membrane extensions while rolling, similar to tether structures seen with neutrophils during rolling [43–45]. These results suggest that while actin polymerization is required for any migratory phenotype, forming Arp2/3-guided branches may not be necessary for CD4+ T lymphocytes to migrate against the direction of flow and maintain the directionality after the flow is removed.

The mechanism of upstream migration remains rather mysterious. Valignat et al. have suggested that a uropod provides a passive self-steering structure in T cells during flow, similar to a wind vane [46]. While T cells are using their uropods to sense the direction of flow, determining the direction of migration (perhaps the placement of the uropod) ultimately depends which integrins are engaged. This suggests that in order to orient themselves against the direction of flow with the presence of ICAM-1, not only the physical formation of the uropod structure but also the reorientation of cytoskeletal network and cellular polarization, which are often driven by intracellular signaling pathways, are imperative for cells to reorient themselves and persistently migrate in the opposite direction of flow. This study implies that the simultaneous activation of LFA-1 and VLA-4 integrins could also propagate actin reorganization to support the upstream migration of T cells.

While LFA-1 alone is sufficient to promote upstream migration, we have shown that VLA-4 and LFA-1 together is essential for directed motion of CD4+ T lymphocytes post flow. VLA-4 alone does not support upstream migration. Although there are several reports of β₁- and β₂-subunits cross-talking intracellularly, the details of the crosstalk have not been fully elucidated [34, 47–49]. Specifically, several groups have noted that the activation of one integrin phosphorylates the other and thus affects the adhesion and migration on surfaces with the affected integrin’s cognate ligands [49–52]. Grönholm et al. and Porter and Hogg have reported LFA-1 activity affecting VLA-4 adhesion and migration on VCAM-1 [49, 51]. Chan et al. have investigated how VLA-4 augments LFA-1-mediated adhesion to ICAM-1 [52]. Our results also support integrin crosstalk by highlighting the difference in migration post flow between VCAM-1/ICAM-1 and ICAM-1 surfaces. Specifically, we speculate that integrin crosstalk provides some signal that is required for the continued upstream persistence of cell motion. CD4+ T lymphocytes on VCAM-1/ICAM-1 showed no sign of directional memory when blocked VLA-4 from binding to VCAM-1, which further supports that cells require both LFA-1 and VLA-4 to remember the direction of flow. However, cells on ICAM-1 with soluble VCAM-1 retained migrational memory, suggesting that both soluble or surface-bound form of VCAM-1 can trigger VLA-4 activity which then initiates the crosstalk of LFA-1 and VLA-4 and promotes the directional memory. Importantly, the result that suggests the occupancy of VLA-4, not its adhesive activity, is critical. While uropods provide physical structure to detect the direction of flow and LFA-1 to promote upstream migration, VLA-4-VCAM-1 binding may strengthen LFA-1 adhesion to ICAM-1, which, in turn, will promote robust upstream migration in CD4+ T cells. This somehow promotes the maintenance of polarization or stabilizes the uropods for longer to maintain upstream migration after flow longer than cells would have with LFA-1 alone.

We have also shown that migrational memory is dependent upon PI3K. We speculate that PI3K promotes crosstalk between LFA-1 and VLA-4. As phosphorylation of integrins subunits is required for the recruitment of integrin-binding proteins such as α-actinin for cytoskeletal reorganization [48], simultaneous activity from VLA-4 could provide a long-lived support of LFA-1 activation through PI3K, thus maintaining upstream migration during and after flow. In addition to integrin-associated adaptor proteins, this mechanism could also involve Crk and c-Cbl, which will be investigated in future work [35]. However, the intracellular localization and the time scale of signaling proteins to reorganize actin cytoskeleton in response to shear flow still remain unanswered.

Combining these insights, it appears that both the morphological structure of the cell and intracellular signaling cascades that are initiated by integrin binding to adhesion molecules are required for persistent upstream migration. While the upstream direction is determined by LFA-1, the persistence of the upstream direction is maintained by simultaneous VLA-4 activation and consequent signaling generated from the crosstalk of the two integrins. Because upstream migration in response to flow has been implicated to be a relevant physiological process in vitro and in vivo [23–26, 28, 46], this study provides a new platform to uncover how integrin crosstalk influences migration during flow and how cells migrate after the absence of shear flow.

Funding

This work was supported by National Institutes of Health [GM123019].

Conflict of interest statement

None declared.