The Gαq/11 Proteins Contribute to T Lymphocyte Migration by Promoting Turnover of Integrin LFA-1 through Recycling

The role of Gαi proteins coupled to chemokine receptors in directed migration of immune cells is well understood. In this study we show that the separate class of Gαq/11 proteins is required for the underlying ability of T cells to migrate both randomly and in a directed chemokine-dependent manner. Interfering with Gαq or Gα11 using dominant negative cDNA constructs or siRNA for Gαq causes accumulation of LFA-1 adhesions and stalled migration. Gαq/11 has an impact on LFA-1 expression at plasma membrane level and also on its internalization. Additionally Gαq co-localizes with LFA-1- and EEA1-expressing intracellular vesicles and partially with Rap1- but not Rab11-expressing vesicles. However the influence of Gαq is not confined to the vesicles that express it, as its reduction alters intracellular trafficking of other vesicles involved in recycling. In summary vesicle-associated Gαq/11 is required for the turnover of LFA-1 adhesion that is necessary for migration. These G proteins participate directly in the initial phase of recycling and this has an impact on later stages of the endo-exocytic pathway.


Introduction
Small chemoattractant peptides called chemokines direct T lymphocytes (T cells) to arrest on post-capillary venules at sites of infection or injury [1,2]. Chemokines bind to G protein-coupled receptors (GPCRs), initiating signalling that activates integrins such as lymphocyte function-associated antigen-1 (LFA-1, CD11a/CD18, aLb2) [3,4]. The chemokine GPCRs are coupled to heterotrimeric G proteins composed of a, b and c subunits and signal through active Gai-GTP and Gbc dimers leading to generation of intracellular effectors such as Ca 2+ and diacylglycerol [5,6]. One of the key downstream effectors of chemokine triggered signalling is the GTPase Rap1. It has several critical roles in LFA-1 activation that lead to arrest of circulating T cells onto vessels and their subsequent firm adhesion to and migration along the vessel walls and into tissue [4,7].
Other groups of G proteins such as the Gaq/11 family comprising Gaq, 11, 14 and 15/16, have also been implicated in immune cell functions such as migration but less is known about how they mediate their effects compared with Gai proteins [6,8]. Gaq and Ga11 are widely expressed and are the most homologous members of this family with many of their activities considered to be over-lapping. There are conflicting reports about the involvement of these Gaq/11 proteins in migration. A positive role was demonstrated by the failure both of Ga11-inhibited myeloid leukaemia cells to migrate to lymphoid tissues and of the LFA-1-mediated tissue invasion of a Ga11-inhibited T cell hybridoma [9,10]. Similarly neutrophils and dendritic cells from mice lacking Gaq (Gnaq 2/2 ) have deficient chemotactic responses, with the defect apparently not extending to Gnaq 2/2 T cells [11]. In contrast Gaq siRNA-mediated knockdown in the Jurkat T cell line enhanced migration in response to chemokine CXCL12 suggesting a repressive effect of the Gaq protein on motility [12].
In this study we have focussed on the role of Gaq and Ga11 in T cell migration mediated by the integrin LFA-1. Blocking Gaq/ 11 activity increased LFA-1-mediated adhesion and led to a reduction in the ability of T cells to migrate both randomly and towards chemokine. We show that this G protein family is required for the turnover of LFA-1 adhesions, has a specific role in their endocytosis and has an impact beyond its expression in the intracellular trafficking of LFA-1.

Gai2 is Needed for Directed but not Random Migration
To investigate involvement of different classes of heterotrimeric G proteins in T cell migration, we first asked whether the HSB2 T cell line was able to respond to a chemoattractant by testing its migration toward CXCL12 (SDF-1a) in a Transwell assay. Transfection of T cells with a dominant negative (DN) cDNA construct of the G protein, Gai2, which is involved in chemokinemediated chemoattraction [13], caused decreased migration towards CXCL12 (8764% decrease) (Fig. 1A). Placing CXCL12 in both upper and lower wells abrogated the directed movement of the T cells indicating that a chemotactic effect was being detected (data not shown). The transfection did not affect membrane expression of CXCR4, the CXCL12 receptor, compared with T cells transfected with vector control (data not shown).
Although unable to respond to the chemokine, T cells transfected with DN Gai2 cDNA had however the same capacity as cells expressing the vector control to migrate randomly on surfaces coated with the LFA-1 ligand, intercellular adhesion molecule-1 (ICAM-1) (Fig. 1B). There was no difference in either directionality or speed of migration between DN Gai2-treated versus vector control-treated T cells (DN Gai2, 12.861.1 mm/ min; control, 11.060.7 mm/min (mean6s.d.)).
To ask whether this distinction between directed and random migration applied more generally and to rule out autocrine chemokine stimulation, we treated T lymphoblasts with pertussis toxin (PTX) that inhibits Gai activity by catalyzing the ADPribosylation of Gai proteins. PTX had significant impact on directed migration towards chemokines CXCL12 and CXCL10 (IP-10) as expected (% decrease: CXCL12, 71.563.5%; CXCL10, 56.561.5%) (Fig. 1C). However, as with HSB2 T cells expressing DN Gai proteins, the ability of T lymphoblasts to migrate randomly on ICAM-1 with regard to either directionality or the speed of migration was unaffected by PTX (PTX, 9.060.6 mm/ min; control, 9.360.6 mm/min (mean6s.d.)) (Fig. 1D). These findings suggest that the Gai-containing heterotrimers are not utilized for random migration of T cells.

The Gaq/11 Subgroup is Needed for both Random and Directed T Cell Migration
We next investigated the effect on migration of the Gaq/11 subgroup of heterotrimeric G proteins whose function is insensitive to PTX [6]. DN cDNA constructs for the two major members of this group, Gaq and Ga11, had significant impact on LFA-1dependent migration when transfected into HSB2 T cells either singly or together. Both the extent of single cell tracking and overall speed were substantially reduced (DN Gaq, 1.560.5 mm/ min; DN Ga11, 3.460.9 mm/min; DN Gaq/11, 3.860.6 mm/ min; control, 7.960.7 mm/min) (mean6s.d.)( Fig. 2A).
To further confirm the role of these G proteins by a means alternative to cDNA transfection, we focussed on Gaq that gave the most robust siRNA knockdown. For Gaq siRNA-treated HSB2 T cells, Gaq protein was reduced by ,85% to a level of 16.561.8% of control siRNA-treated cells (mean6s.d. of n = 5 experiments) (Fig. 2B). For primary T cells the average level of Gaq siRNA knockdown was ,60% with Gaq expression reduced to 40.068.9% of control siRNA (mean6s.d. of n = 3 experiments). A significant decrease in single cell tracking and speed of random migration was observed with siRNA-treated primary T cells (Gaq siRNA, 4.460.4 mm/min, control siRNA, 10.760.5 mm/min) (Fig. 2C).
We next examined a possible role for Gaq in migration directed by chemokines. Gaq siRNA-treated primary T cells displayed a decrease of 4166% in chemotaxis to CXCL12 in a Transwell assay compared with control siRNA-treated T cells (Fig. 2D). An equivalent result was obtained when HSB2 T cells were transfected with DN Gaq/11 and tested for chemotaxis to CXCL12. The transfected cells were reduced to the background levels of migration ( Supplementary Fig. S1).
It was important to evaluate any role for Gaq/11 proteins in a shear flow assay that tests the ability of integrins to attach under conditions of mechanical stress as experienced in the circulation. The rolling and attaching behavior of Gaq siRNA-treated HSB2 T cells was assessed on chambers coated with ICAM-1 and Eselectin at a shear force of 1 dyne as previously described [14]. Both Gaq and control siRNA-treated T cells rolled normally on Eselectin and were able to attach both transiently and firmly to ICAM-1 ( Supplementary Fig. S2A). The ability of Gaq siRNA-treated cells to adhere normally was further confirmed in a static adhesion assay where DN Gaq/11-transfected T cells adhered to ICAM-1 comparably to the control T cells (Supplementary Fig. S2B).
Thus the Gaq/11 proteins have a role in the random migration of T cells and also when the cells are undergoing directed migration to a chemoattractant such as a chemokine. However the G proteins appear not to influence T cell adhesion to ICAM-1 under either static or mechanical shear conditions.

DN Gaq/11 Reduction Alters the Morphology of ICAM-1adhered T Cells
To gain further understanding of how signaling through Gaq/11 might regulate LFA-1-mediated migration, we investigated the morphology of T cells transfected with either a combination of DN Gaq/11 cDNAs or vector alone. The DN Gaq/11-treated T cells were polarized to the same extent as control T cells (DN Gaq/11, 83.560.7% versus control, 8661.5%) (Fig. 3A). However, DN Gaq/11-transfected T cells displayed increased total cell length compared with control T cells (DN Gaq/11, 20.563.5 mm versus control, 13.8563.5 mm (mean6s.d.)). Additionally the uropods of the majority of DN Gaq/11-transfected T cells were attached to the ICAM-1coated surface rather than elevated above it (DN Gaq/11, 81.562.5% versus control, 31.561.5%). Live cell images highlighted an abnormally attached rear of the DN Gaq/11transfected T cells providing further evidence that T cells lacked the ability to detach correctly (Fig. 3B, Supplementary Videos S1 and S2). A lack of effect on migration of T cells transfected with WT Gaq cDNA provided further evidence that the DN cDNAs were working as expected (data not shown).
These observations indicate that T cells transfected with DN Gaq/11 cDNAs are able to move the leading edge of the cell forward but have impaired ability to detach from the substrate ICAM-1.

Association of Gaq/11 with LFA-1 Endocytosis and Intracellular Vesicles
An association has been made previously between the failure of LFA-1 detachment and b2 subunit mutation leading to lack of LFA-1 endocytosis [15]. It was therefore of interest to test whether there was a connection between the ability of the Gaq/11-inhibited T cells to turnover LFA-1 adhesions and endocytosis or recycling of this integrin. To investigate LFA-1 internalization biochemically, intact HSB2 T cells were surface labelled with glutathione-cleavable biotin and allowed to migrate on ICAM-1 for 40 min followed by LFA-1 immunoprecipitation and blotting for biotinylated integrin. By removal of membrane LFA-1 using glutathione, total LFA-1 could be distinguished from internalized LFA-1 [16,17]. The T cells were also treated with primaquine (PQ), which is a lysosomotrophic amine that slows recycling by blocking membrane fusion of exocytic vesicles [17]. When Gaq siRNA-treated T cells were compared with control siRNA-treated cells there was a 3564% reduction in internalized LFA-1 based on total cell biotinylated LFA-1 levels (Fig. 4A).
Confocal microscopic images of T blasts revealed a pattern of Gaq-expressing intracellular vesicular structures with highest density of immunostaining in the juxta-nuclear region of the polarized cells and trailing edge with scattered distribution towards the front of the cell (Fig. 4B). Thus Gaq not only influences LFA-1 internalization, but is also associated with intracellular vesicles. These observations provide the first suggestion that the Gaq proteins might be involved in LFA-1 recycling.

Co-localization of Gaq with LFA-1 and Vesicle Markers
We used confocal microscopy to further define the association of Gaq with LFA-1 and other intracellular vesicle markers. Gaq colocalized with LFA-1 particularly prominently where endosomal vesicles are located in the juxta-nuclear region and trailing edge (Fig. 5). Pixel-by-pixel analysis yielded 95.461.8% overlap between Gaq and LFA-1 (n = 5 cells). Gaq staining also overlapped with EEA1, a key Rab5 effector protein [18] (Overlap analysis = 94.264.0%, n = 5 cells). However, although it was wellexpressed, there was essentially no overlap of Gaq with Rab11, a marker of a subset of late endosomal vesicles (Overlap analysis = 12.763.9%, n = 5 cells) and previously associated with LFA-1 recycling [16].
The GTPase Rap1 is required for the early events of LFA-1 adhesion [19] and also with its transport in intracellular vesicles [4,7,20,21]. We observed a partial overlap of Gaq with Rap1 (Overlap analysis = 67.7612.5%, n = 5 cells) suggesting that Rap1 has associations beyond those with Gaq.
In summary, confocal microscopy showed that there was selectivity in immunostaining of Gaq in that it overlapped with LFA-1 and with some, but not all, endosomal vesicle markers previously associated with this integrin.

Association of Gaq with LFA-1 and Vesicle Markers at Plasma Membrane Level
We next used TIRF microscopy to look in closer detail at the T cell membrane where LFA-1 was in contact with ICAM-1. At TIRF level, Gaq co-localized with LFA-1 in the main cell body where the integrin attached to ICAM-1 sparing the attached filopodia (Overlap analysis = 93.866.2%, n = 5 cells). Co-localization of Gaq staining with EEA1 was also observed (Overlap analysis = 81.6612.9%, n = 5 cells) (Fig. 6). Thus Gaq was expressed at membrane level with LFA-1 and EEA1. In contrast, Rab11 and Rap1 were both poorly visible at TIRF level. As both proteins were easily imaged at epifluorescence level, the implication is that they are expressed intracellularly but away from the plasma membrane.

Effect of Gaq Reduction on Vesicle Expression and Distribution
It was relevant to ask whether Gaq siRNA knockdown altered the expression or distribution pattern of intracellular vesicles. When Gaq siRNA-treated T cells were compared with control siRNA-treated cells, it was evident that the T cells with reduced Gaq displayed increased expression of LFA-1 and EEA1 that were both co-expressed with Gaq (Fig. 7A, Table 1). In addition, although Gaq only partially overlapped with Rap1, and not at all with Rab11, the staining associated with these vesicle markers was also increased in the Gaq siRNA-treated T cells. Therefore the reduction in Gaq expression had an impact not only on Gaqexpressing vesicles, but also on vesicles with which it did not codistribute.
Finally it might be expected that these effects of lack of Gaq on vesicular traffic would have an impact on membrane expression of LFA-1. Using laser scanning cytometry to examine LFA-1 expression per unit cell area, a comparison of T cells transfected with control versus DN Gaq/11 cDNAs revealed ,40% reduction in expression of LFA-1 on membranes of Gaq/11 compromised T cells consistent with intracellular accumulation of recycling vesicles (Fig. 7B).
The accumulation of several types of intracellular vesicles when the level of Gaq is diminished is consistent with there being a general slowdown in intracellular trafficking of LFA-1 that has a knock-on effect of causing reduction in LFA-1 at membrane level.

Discussion
In this study we find that the Gaq/11 class of G proteins has an essential function in the LFA-1-mediated migration of human T cells. These two G proteins are needed for a basic aspect of migration as their blockade affects both random migration and also when the T cells are undergoing directed chemokinemediated migration through the use of G protein, Gai2. The use of dominant negative cDNAs indicate an overlapping role for Gaq and Ga11 as has been previously suggested [6,8]. We provide evidence that Gaq/11 proteins participate in LFA-1 recycling and in particular that they regulate LFA-1 adhesion turnover, internalization and membrane expression. The effect on migration is in keeping with an earlier report indicating Ga11 involvement in the LFA-1-mediated migration and invasion of a T cell lymphoma [10]. Furthermore there is increasing evidence that recycling of integrin is generally essential for successful migration of leukocytes. For example, the presence of a5b1 [22,23] and LFA-1 [16] in recycling vesicles drives neutrophil migration.
An initial observation was that blockade of Gaq and Ga11 activity or Gaq siRNA knockdown caused excessive T cell adhesion to ICAM-1. Although the leading edge of the T cells displayed some forward motility, migration halted because of increased attaching, not at the front, but at the rear of the cell. Blocking Gaq activity however had no overall effect on LFA-1mediated adhesion of T cells under either static or shear flow conditions, suggesting that the pro-adhesive phase was not affected. The evidence thus pointed to an inability of Gaq/11compromised T cells to detach or turnover their LFA-1 adhesions.
A previous report showed that mutation of the b2 subunit in the endocytosis motif prevented not only endocytosis, but also LFA-1 turnover implying a connection between the two processes [15]. Similarly we found that Gaq affected de-adhesion and was also required for LFA-1 internalization. This was consistent with confocal images showing the presence of Gaq on intracellular vesicles and co-localization with LFA-1 and EEA1, a protein associated with the GTPase Rab5 that characterizes vesicles involved in the early endocytic stage of recycling [24].
The GTPase Rap1 plays an essential early role in lymphocyte arrest on the vasculature following chemokine stimulation [19] and this is consistent with its involvement in vesicle transport as well-demonstrated in other studies [20,21,25,26]. Gaq only partially co-distributed with Rap1 suggesting additional roles for the GTPase. Other studies report Rap1 to be associated with EEA1, Rab5 and Rab11 [21] or alternatively on vesicles with limited EEA1 and Rab7 overlap [26]. Thus, in terms of vesicle marker expression, the pattern of Rap1 expression is distinct from but overlapping with Gaq. It is of interest that Rap1 is also required for delivery of LFA-1 to the membrane [27] whereas we found no role for Gaq in this activity.
Gaq did not co-distribute with Rab11 that is displayed by a subset of late endosomal vesicles. This is of relevance because LFA-1 recycling has also been associated not only with Rap1, but with recycling vesicles expressing Rab11 and with trafficking of LFA-1 to the lamellipodia [16]. The data therefore support the idea that Gaq is associated with the initial endocytic phase of LFA-1 recycling and not with the later stages. Thus there must be heterogeneity in LFA-1-transporting vesicles with Gaq characterizing one set and Rap1 and Rab11, a partially overlapping and separate set respectively that are involved in the subsequent events of bringing LFA-1 to the membrane followed by exocytosis. Supporting evidence comes from a comparison between epifluorescence and TIRF observations that show both Rap1 and Rab11 to be most highly concentrated away from the level of close membrane contact of T cell LFA-1 with ICAM-1. The Rab11dependent release of LFA-1 into membrane ruffles at the leading edge of CHO cells is in keeping with their lack of substrate contact [16].
The influence of Gaq on the behavior of intracellular vesicles however goes beyond the vesicles that express it, as Gaq siRNA knockdown affects not only EEA1-expressing vesicles on which Gaq is co-expressed, but also Rap1-and Rab11-expressing vesicles. These vesicle markers normally display a juxta-nuclear distribution pattern with scattered representation towards the front of the polarized T cell. In Gaq siRNA-treated T cells, the juxtanuclear pattern is relaxed and this is accompanied by an increase in vesicles expressing not only LFA-1 and EEA1, but also Rap1 and Rab11. Such a ''log jam'' of vesicles and the reduction in LFA-1 internalization all point to a disturbance in the LFA-1 recycling network that follows on from reduction in Gaq with consequences beyond the vesicles that normally express it. Thus the findings indicate that the Gaq-expressing vesicles are linked in terms of function to other vesicles with later involvement in the endocytic/exocytic sequence of events involved in the intracellular movement of LFA-1 back to the membrane. An end result of the failure of normal recycling is the ,40% decrease in expression of membrane LFA-1. This is apparently not fully compensated for by a decrease in LFA-1 internalization in Gaq/11 blocked cells. It was also not reflected in diminished adhesion, but rather the reverse, consistent with disturbance in LFA-1 turnover as evident from impaired uropod retraction.
Some of these activities of Gaq/11 are similar to those of the Ga12/13 class of G proteins. T cells from Ga12/13-deficient mice also display increased LFA-1-mediated adhesion [28]. A speculation is that Ga12/13 may also be involved in recycling of integrin as Rab11-expressing vesicles that are Ga13-dependent are associated with an intracellular recycling compartment of T cells containing CXCR4/TCR heterodimers [29].
In summary we here define an early stage in LFA-1 recycling on T cells that is regulated by the Gaq/11 proteins. Although Gaq/ 11 has a direct role in endocytosis, there is evidence for more extended influence on an interconnected sequence of events involving other types of vesicles that traffic LFA-1 back to the T cell membrane. The presented evidence does not exclude the existence of LFA-1 recycling pathways that are completely independent of Gaq/11. A key issue for the future will be to determine the relationship between the different LFA-1-containing intracellular vesicles and the extent of their heterogeneity.

Cell Isolation, Culture and Transfection
Peripheral blood mononuclear cells were prepared from single donor leukocyte buffy coats (National Blood Service, Tooting, London, UK); T cells were expanded as previously described and used between days 10 and 14 [31]. Primary T cells were isolated using the MACS Pan T cell isolation kit II by negative depletion (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany). The human T lymphoblast CD3 2 T cell line, HSB2, isolated from an acute lymphoblastic leukemia source (ATCC number CCL-120.1, known as CCRF-HSB-2 or HSB2) was maintained in RPMI 1640/10% FCS [32].
HSB2 T cells (2610 7 cells) were washed in OptiMEM + GlutaMAX (Invitrogen, Paisley, UK) and electroporated with the following reagents all at 400 nM per reaction using a Gene Pulser with Capacitance Extender (Bio-Rad UK, Hemel Hempstead, UK) set at 960 mF and 300 mV: Gaq siRNAs (Experimentally Verified GNAQ5 and GNAQ6) or negative control siRNA (all Quiagen Ltd., Crawley, UK). Efficiency of individual siRNA knockdowns in T cells was evaluated by Western blotting. Primary T cells were transfected with the siRNAs also at 400 nM per reaction using Amaxa Human T cell Nucleofector kit (Lonza, Cologne, Germany).
Alternatively HSB2 T cells were transfected with full-length dominant negative human G protein alpha q, Q209L/D277N (Gaq, CloneID GNA0Q000X0), human G protein alpha 11, Q209L/D277N (Ga11, CloneID GNA11000X0), G-protein i2, G203T (Gai2, CloneID GNAI12000T0) (UMR cDNA Resource Center, University of Missouri) (cDNA constructs, pCDNA3.1 vector control, all @10 mg cDNA per reaction. The strategy for the DN Gaq/11 cDNA constructs involved creation of xanthine nucleotide binding mutants of Ga11 and by analogy Gaq mutants that act in a dominant negative fashion by binding to their appropriate receptors and blocking GTP-mediated activation [33]. Additionally the Gai2 DN mutant G203T has been described [34]. Flow cytometry following co-transfection of T cells with EGFP cDNA @ 0.1 mg cDNA per reaction revealed ,50% T cells were successfully transfected.
Transfected cells were maintained in RPMI 1640 with 10% FCS for up to 48 h for siRNA-, and 24 h or 48 h, for cDNAtransfected HSB2 cells. Transfected T cells were routinely .90% T cell viable.

LFA-1 Internalization Assay
This protocol was adapted from [35]. Glass coverslips (32 mm) were coated overnight with 3 mg/ml ICAM-1Fc, then blocked with 2.0% BSA. To biotinylate membrane proteins, washed T cells were re-suspended in 0.5 mg/ml EZ-link Sulpho-NHS-SS-Biotin (21331, Pierce, ThermoFisher Scientific, Loughborough, UK) at 25610 6 cells/ml and incubated on ice for 1 h. After washing, 4610 6 T cells in HBSS buffer were added to each ICAM-1-coated coverslip. Primaquine diphosphate (PQ) (160393, Sigma-Aldrich Ltd) at 300 mM was added and the cells incubated for 40 min at 37uC to allow internalization of receptors. To remove membrane bound biotin, fresh glutathione buffer (46 mM glutathione, 75 mM NaCl, 1 mM EDTA, 1% BSA, 75 mM NaOH) was added and the cells incubated on ice for 30 min. Controls for biotinylation of total LFA-1 were maintained in PBS.
To analyze biotinylated LFA-1, T cells were lysed with a standard buffer containing 0.2% NP40 buffer. Immunoprecipitation using anti-LFA-1 mAb 38 and subsequent blotting were performed as previously described [36]. Biotinylated LFA-1 was revealed by blot incubation with Streptavidin-HRP conjugate (RPN1231, GE Healthcare) in PBS/0.1% Tween 20 and ECL reagent (GE Healthcare). Samples were also probed withã atubulin mAb and anti-mouse IgG-HRP Ab (GE Healthcare) to check for equivalent sample loading.

Chemotaxis
For chemotaxis assays, T cells at 5610 6 cells/100 ml were allowed to migrate through 5 mm pore size Transwell insert wells coated with ICAM-1Fc as above (Corning, Acton, MA, USA). The lower wells contained either 600 ml RPMI 1640/0.1% BSA alone or medium plus 10 nM CXCL10 or CXCL12 (PeproTech EC Ltd, London, UK). After 90 min of incubation at 37uC and 5% CO 2 , inserts were discarded and the migrated T cells were counted by flow cytometry after recovery using ice-cold 5 mM EDTA/PBS. All samples were tested in triplicate.

Video Microscopy
35 mm glass-bottom microwell dishes (MatTek Corp., Ashland, MA, US) or m-slides VI (Ibidi GmbH, Martinsried, Germany) were coated overnight with 3 mg/ml ICAM-1Fc as above. The T cells (2610 6 cells/ml in HBSS with 20 mM HEPES (H.HBSS) were exposed to ICAM-1 for 10 min at 37uC and images captured using an Olympus MTV3 Inverted microscope using a 206 lens or Zeiss Axiovert 135TV Inverted microscope using a 636 lens plus AQM 2001 Kinetic Acquisition Manager software (Kinetic Imaging Ltd). The cells were tracked at 15 sec intervals with Motion Analysis software (Kinetic Imaging Ltd, Bromborough, UK) and data analyzed using a Mathematica notebook (Wolfram Research, Long Hanborough, UK) developed by D. Zicha (Cancer Research UK). Uropod attachment was quantified as previously reported [37]. Briefly individual live migrating T cells were observed using a visual assessment and scored by two observers. Analysis of the attachment status was accomplished by focus on both the T cell contact interface with ICAM-1 and the focal plane above this level.

Confocal and TIRF Microscopy
13 mm round glass coverslips or glass bottomed MatTek dishes were pre-coated with ICAM-1-Fc as above. Washed T cells (2610 5 cells/sample) were added to coated coverslips for 30 min. Adherent cells were fixed with fresh 3% paraformaldehyde in Pipes buffer (pH8) for 5 min at RT, washed and fixed again in 3% paraformaldehyde in Borax, (pH 11, Sigma-Aldrich Ltd) [14]. Cells were then permeabilized with 0.1% Triton-X-100 for 5 min at 4uC. Autofluorescence was quenched using fresh sodium tetraborate (1 mg/ml, pH8) for 15 min at RT. Coverslips were incubated with primary mAbs overnight at 4uC, followed by Alexa488-goat anti-mouse IgG or Alexa546-goat anti-rabbit IgG (Invitrogen, Paisley, UK) for 45 min. Images were acquired on a Zeiss Laser Scanning Microscope LSM 710 or 780 using Zen software and x63 DIC oil lens.
The polarity of the T cells was determined by examining their morphology. A migrating T cell displays a spreading lamellipodium at the leading edge and an elevated uropod at the trailing edge, although HSB2 cells attach less ideally that T lymphoblasts. We have confirmed the assignments in previous studies of both leading edge (F-actin cross-linking; a-actinin localization) and trailing edge (increased a-tubulin and ICAM-1/3 distribution) [14,36,38].
For quantification of the fluorescence using the microscope software, an arbitrary mean fluorescence value per designated cell area was determined. This was done by applying a filter between 500-1000 mHz to exclude both the background and saturated signals, leaving the positive signal of interest and generating the mean fluorescence intensity (MFI) 6 s.d.
The extent of co-localization of different markers was analyzed using Image J software and JACoB analysis. Co-localization was measured using the Manders coefficient to evaluate the overlap in fluorescence.

Detection of Cell Membrane LFA-1 Using Laser Scanning Cytometry
Epitope expression and the area of spread contour of T cells mounted on ICAM-1 coverslips were calculated based on their fluorescence using a Laser Scanning Cytometer (CompuCyte, Mass, USA) and WinCyte version 3 software (Compucyte). The technique has the advantage of allowing investigation of membrane expression of epitopes on lymphocytes that are adhered to and spread on the LFA-1 ligand ICAM-1. The approach was to allow cell attachment and migration followed first by cell fixing and labelling with anti-LFA-1 mAb 38 conjugated to AlexaFluor-488, then by permeabilization with 0.1% Triton X100 (5 min on ice) and labeling with Phalloidin-PE. The threshold level for measurement was set using the signal from Phalloidin-PE so that individual cells could be segmented and assigned an area value (microns squared). Within this area, the total integrated AF488 signal was measured by summation of the value of AF488 fluorescence for each pixel (10-bit scale per pixel). In Fig. 7C, we measured the change in AF488 measurements per average cell area of T cells treated with cDNA constructs of DN Gaq/11 compared with control cDNA. 6000 T cells were measured per coverslip.

Statistical Analysis
The migration and other assays are presented as mean6s.d. The unpaired Student's t test was performed using GraphPad Prism software version 5 for Macintosh computers. The following significant differences are as indicated: *, P,0.05; **, P,0.01 and ***, P,0.001.