Research ArticleImmunology

Superresolution imaging of the cytoplasmic phosphatase PTPN22 links integrin-mediated T cell adhesion with autoimmunity

See allHide authors and affiliations

Science Signaling  04 Oct 2016:
Vol. 9, Issue 448, pp. ra99
DOI: 10.1126/scisignal.aaf2195

Release the phosphatase!

T cells need to move through the circulation, attach to endothelial cells, transmigrate into tissues, and stably interact with target cells. The phosphatase PTPN22 targets phosphorylated tyrosines in Src and Syk family kinases, many of which are phosphorylated and activated in migrating T cells in response to the binding of the integrin LFA-1 to its ligand ICAM-1. Burn et al. used superresolution microscopy to show that PTPN22 formed clusters in nonmigrating T cells, which were dispersed in T cells that migrated on surfaces coated with ICAM-1. Freed from these complexes, PTPN22 interacted with its targets near the front of the migrating T cell, which inhibited LFA-1 signaling. In contrast, clusters containing the PTPN22 R620W mutant, a variant that is associated with autoimmune diseases, failed to disaggregate in migrating T cells, and thus, LFA-1 clustering and signaling were not inhibited. Together, these data suggest how a mutation associated with autoimmunity dysregulates T cell adhesion and migration.


Integrins are heterodimeric transmembrane proteins that play a fundamental role in the migration of leukocytes to sites of infection or injury. We found that protein tyrosine phosphatase nonreceptor type 22 (PTPN22) inhibits signaling by the integrin lymphocyte function-associated antigen–1 (LFA-1) in effector T cells. PTPN22 colocalized with its substrates at the leading edge of cells migrating on surfaces coated with the LFA-1 ligand intercellular adhesion molecule–1 (ICAM-1). Knockout or knockdown of PTPN22 or expression of the autoimmune disease–associated PTPN22-R620W variant resulted in the enhanced phosphorylation of signaling molecules downstream of integrins. Superresolution imaging revealed that PTPN22-R620 (wild-type PTPN22) was present as large clusters in unstimulated T cells and that these disaggregated upon stimulation of LFA-1, enabling increased association of PTPN22 with its binding partners at the leading edge. The failure of PTPN22-R620W molecules to be retained at the leading edge led to increased LFA-1 clustering and integrin-mediated cell adhesion. Our data define a previously uncharacterized mechanism for fine-tuning integrin signaling in T cells, as well as a paradigm of autoimmunity in humans in which disease susceptibility is underpinned by inherited phosphatase mutations that perturb integrin function.


Integrins function as adhesion receptors that control cell-cell and cell-matrix interactions, thereby regulating the migration of cells into tissues. Lymphocyte function-associated antigen–1 (LFA-1; also known as CD11a/CD18 or αLβ2 integrin) is the major integrin used by T cells. In addition to mediating T cell adhesion and migration, LFA-1 transduces environmental cues that affect a wide range of cellular functions, including cell differentiation, proliferation, cytokine production, cytotoxicity, and cell survival (16). Optimal function requires changes in the conformation and clustering of LFA-1 in ways that promote cell adherence, which are achieved through two distinct, yet overlapping, signaling pathways (7). Inside-out signaling is initiated by antigen or chemokine receptors and results in conformational changes in LFA-1 that increase its affinity for its ligands, such as intercellular adhesion molecule–1 (ICAM-1), and the binding of cytoplasmic signaling modules to integrin tails (8, 9). Outside-in signaling is initiated after LFA-1 engages ICAM-1, and it leads to the Src- and Syk-mediated phosphorylation of tyrosines, the plasma membrane translocation of Rap1 (Ras-related protein 1), and the binding of talin and kindlin-3 to the cytoplasmic tail of the β2 subunit of LFA-1 (3, 10). Although counterregulation of LFA-1–dependent protein tyrosine phosphorylation is required for repeated cycles of adhesion, de-adhesion, protrusion, and contraction of T cells (11, 12), the protein tyrosine phosphatases that support this function are not well understood.

Protein tyrosine phosphatase nonreceptor type 22 (PTPN22; also known as Lyp in humans and PEP in mice) is a protein tyrosine phosphatase that dephosphorylates Src and Syk family kinases (13). Interest in PTPN22 has grown considerably since initial reports of strong associations between a missense single-nucleotide polymorphism in the PTPN22 gene (1858C>T, encoding the R620W variant) and a growing number of autoimmune diseases, including type I diabetes, rheumatoid arthritis, and systemic lupus erythematosus (14). Functional studies have focused on how the phosphatase regulates antigen receptor signaling in lymphocytes, but precisely how the R620W variant confers susceptibility to autoimmune disease is unknown. The fact that the R620W mutation targets the P1 polyproline domain of PTPN22 has raised the possibility that impaired interactions between the mutant phosphatase and the Src homology 3 (SH3) domain of C-terminal Src kinase (Csk), a negative regulator of Src family kinases, could disrupt mechanisms that have evolved to attenuate antigen receptor signaling in lymphocytes (14). Despite these insights, some studies have suggested that the R620W variant is a loss-of-function mutation, whereas others pointed to a gain of function (15, 16). Here, we investigated whether PTPN22 regulated LFA-1–mediated signaling because the kinases Lck and ZAP70, which are substrates of PTPN22, are also phosphorylated after engagement of LFA-1 (17, 18). Our experiments were also motivated by the possibility that perturbations in integrin signaling might play a role in provoking immune-mediated inflammatory diseases in individuals carrying the PTPN22-R620W variant.


PTPN22 colocalizes with its substrates at the leading edge of migrating T cells

To examine whether PTPN22 regulated outside-in integrin signaling, we plated activated primary human T cells onto ICAM-1–coated glass and then fixed, stained, and imaged the cells by confocal microscopy. PTPN22 mainly polarized to the lamellipodium at the leading edge of migrating T cells (on ICAM-1), with some staining in the uropod, whereas PTPN22 was not polarized in nonmigrating T cells on glass coated with poly-l-lysine (PLL) (Fig. 1A). Specificity of staining was confirmed by gene knockdown, by analysis of patterns of staining obtained with monoclonal and polyclonal antibodies specific for PTPN22, and by localization studies with T cells expressing green fluorescent protein (GFP)–tagged PTPN22 (PTPN22-GFP) (fig. S1, A to C). Confocal microscopy and fluorescence intensity plots revealed the phosphorylation of activating tyrosine residues on Lck (pY394), ZAP70 (pY493), and Vav (pY174), which were colocalized with PTPN22 at the leading edge of migrating cells (Fig. 1A and fig. S1D) but not nonmigrating cells (fig. S1E). Analysis and quantification of total internal reflection fluorescence (TIRF) microscopy images confirmed the plasma membrane–proximal associations between PTPN22 and its phosphorylated substrates at the leading edge of migrating T cells, when compared to nonmigrating T cells on PLL (fig. S2, A and B).

Fig. 1 PTPN22 is an inhibitor of LFA-1 signaling, colocalizing with its phosphorylated substrates at the leading edge of migrating T cells.

(A) Primary human T cell blasts were layered onto glass slides coated with PLL or ICAM-1 for 20 min before being stained with mouse anti-PTPN22 antibody (green) and the indicated phosphospecific antibodies (red), and then imaged by confocal microscopy. The direction of migration is indicated by large white arrows. Scale bar, 10 μm. Data represent the analysis of 30 to 40 cells from four independent experiments. (B) Human T cells layered onto PLL- or ICAM-1–coated plates for the indicated times were lysed, subjected to immunoprecipitation (IP) with mouse anti-PTPN22 antibody or control immunoglobulin G (IgG), and then analyzed by Western blotting with antibodies against the indicated proteins. Western blots are representative of five independent experiments. (C) T cells were transfected with scrambled (Scram) or PTPN22-specific (PTPN22) siRNAs and were then cultured for a further 48 hours before being plated onto PLL- or ICAM-1–coated plates. After 20 min, the adherent cells were harvested, lysed, and analyzed by Western blotting with antibodies against the indicated proteins. Western blots are representative of three experiments. (D) T cell blasts were transfected with scrambled or PTPN22-specific siRNAs and cultured for 24 hours before being plated onto ICAM-1–coated plates. The migration of single cells was tracked by time-lapse microscopy. Dot plots show pooled data from three experiments with the mean speeds ± SD of 100 to 150 T cells transfected with the indicated siRNAs. ****P < 0.0001. (E) T cells were mock-transfected (no DNA) or were transfected with the indicated GFP expression vectors. After 24 hours, the velocity of the cells on ICAM-1 was quantified as described for (D). Data are means ± SD derived from three pooled experiments analyzing a total of 50 to 90 cells. ****P < 0.0001; ns, not significant. (F) T cells isolated from the lymph nodes of Ptpn22+/+ or Ptpn22−/− littermate mice were allowed to migrate for 20 min on PLL- or ICAM-1–coated plates. (Top) Lysates of adherent cells were analyzed by Western blotting with antibodies against the indicated proteins. Western blots are representative of three experiments. (Bottom) Fold changes in the abundance of pERK1/2 in the indicated cells plated on ICAM-1 relative to the abundance of pERK1/2 in cells plated on PLL were determined. Data are means ± SD of three experiments. *P = 0.042. (G) Human T cell blasts were generated from genotyped donors expressing PTPN22-R620 (RR), PTPN22-R620W (RW), or PTPN22-W620 (WW) and allowed to migrate on PLL- or ICAM-1–coated plates for 20 min. (Left) Cells were lysed and analyzed by Western blotting with antibodies against the indicated proteins. (Bottom) Fold changes in the abundance of pERK1/2 in the indicated migrating cells relative to the abundance of pERK1/2 in nonmigrating cells were determined. Data are means ± SD of eight experiments. Pairwise comparisons (two-tailed t test): RR versus RW, *P = 0.0337; RW versus WW, *P = 0.0144; RR versus WW, **P = 0.0093.

Integrin signaling stimulates an association between PTPN22 and its phosphorylated substrates

Biochemical analysis of whole-cell lysates confirmed the inducible phosphorylation of integrin signaling intermediates in migrating T cells in response to different integrin ligands (fig. S3, A and B), and the specificity of this response was confirmed in experiments with soluble anti–ICAM-1 antibodies (fig. S3C). Lysates of nonmigrating and migrating cells were subjected to immunoprecipitation with an anti-PTPN22 antibody and were then analyzed by Western blotting with an anti-phosphotyrosine antibody. Whereas a single phosphoprotein band was detected in the nonmigrating T cells, many more phosphoprotein bands were detected in the migrating T cells, suggesting the inducible association of PTPN22 with phosphotyrosine substrates (fig. S3D). Three of these bands resolved to molecular masses corresponding to phosphorylated Lck (56 kDa), ZAP70 (70 kDa), and Vav (118 kDa). The specificity of these interactions and the associations between these substrates and PTPN22 over time were confirmed by Western blotting analysis of anti-PTPN22 immunoprecipitates with phosphospecific antibodies (Fig. 1B). Furthermore, PTPN22 associated only with phosphorylated protein substrates, as suggested by Western blotting analysis of total Lck and ZAP70 abundances (Fig. 1B), which indicates that the phosphorylation of its substrates in migrating T cells was required for their interaction with PTPN22.

PTPN22 is an inhibitor of integrin signaling, and the R620W variant is a loss-of-function mutant

Targeting PTPN22 with small interfering RNA (siRNA) reduced the abundance of PTPN22 protein in migrating primary human T cells by ~50 to 60%, without affecting the abundance of LFA-1 (fig. S3E), but increased the ICAM-1–dependent phosphorylation of Lck (pY394), ZAP70 (pY493), and Vav (pY174) when compared to that in migrating cells treated with control siRNA (Fig. 1C). Unlike for pY493-ZAP70, there was a less pronounced increase in the phosphorylation of the regulatory Tyr319 site of ZAP70, which is not a target of PTPN22 (13). PTPN22 knockdown was also associated with increased T cell motility (Fig. 1D), whereas overexpression of the GFP-tagged PTPN22-R620 variant (wild type), but not a catalytically inactive C227A variant or GFP alone, in PTPN22-sufficient T cells reduced cell motility (Fig. 1E). The phosphatase activity of PTPN22 was therefore required for the regulation of integrin-mediated cell motility. Integrin-dependent phosphorylation of the classical mitogen-activated protein kinases (MAPKs) extracellular signal–regulated kinase 1 (ERK1) and ERK2 (collectively referred to as ERK1/2) was also observed after engagement of LFA-1 (Fig. 1C), consistent with its localization in nascent adhesion complexes and its contribution to the “motor phase” of lamellipodium protrusion (1921). ERK1/2 phosphorylation was blocked by the Lck inhibitor PP2 (fig. S3F), suggesting that the integrin-dependent activation of Lck was upstream of ERK phosphorylation, as previously reported (18). Consistent with these data, ERK1/2 phosphorylation was enhanced in migrating Ptpn22−/− murine T cells (Fig. 1F) and in human T cells carrying either one or two copies of the PTPN22 genetic variant encoding PTPN22-W620 (Fig. 1G). These data suggest that PTPN22 is an inhibitor of LFA-1 signaling and that the R620W variant is a loss-of-function mutant.

PTPN22 and Csk localize in plasma membrane–proximal clusters that decluster during LFA-1–stimulated migration

Diffraction-limited microscopy showed substantial localization of PTPN22 at the leading edge of migrating T cells after engagement of LFA-1 (Fig. 1A). To obtain quantitative imaging data of this phenomenon at the plasma membrane, we used TIRF–direct stochastic optical reconstruction microscopy (dSTORM) superresolution microscopy and quantitative cluster analysis. These techniques quantify protein (localization) and cluster number, as well as cluster size and the density of PTPN22 protein localized at the plasma membrane of stationary cells or at the leading edge of migrating T cells. The visualization of single molecules can be represented by point maps and cluster heat maps. Using this approach, we found that PTPN22 was highly clustered at the plasma membrane of nonmigrating T cells (plated on PLL), whereas migrating T cells contained smaller, less dense clusters (Fig. 2A), which was confirmed by quantitative cluster analysis (Fig. 2, B and C); localizations per cluster (Fig. 2C) reflect the number of PTPN22 molecules per cluster. Quantification by Ripley’s K function, a measure of the extent of clustering, also revealed a substantial reduction in clustering [peak of L(r)-r curve] and the presence of smaller clusters (position of the peak on the x axis) at the leading edge (Fig. 2D), further confirming that PTPN22 was substantially less clustered in migrating T cells.

Fig. 2 PTPN22 exists in large clusters that disperse upon engagement of LFA-1.

(A) Primary T cell blasts were generated from the peripheral blood of PTPN22-R620 homozygous donors and layered onto PLL- or ICAM-1–coated plates for 20 min before they were fixed, permeabilized, and stained with mouse anti-PTPN22 antibodies. Images were acquired with a Nikon N-STORM microscope, and molecule distributions were analyzed with cluster analysis algorithms. For each condition, N-STORM images are representative of PTPN22 molecule distributions at the whole-cell level and in 4-μm2 region maps (boxed), which were selected from the leading edge of the migrating T cell. Pointillist (Gaussian-fitted) and pseudocolored heat maps are representative of cluster data acquired after the processing of N-STORM image regions with the cluster analysis algorithm. Scale bars, 5 μm for the PLL condition; 10 μm for the ICAM-1 condition. Data are representative of three experiments analyzing 40 to 50 cells per experiment. (B and C) Cluster analysis of the images represented in (A) was used to define (B) the diameter of the PTPN22-R620–containing clusters and (C) the number of localizations of PTPN22-R620 per cluster for >900 clusters in nonmigrating cells (PLL) and migrating cells (ICAM-1). Data are pooled from three independent experiments analyzing 50 cells per experiment. ****P < 0.0001. (D) Ripley’s K function curves (mean ± SEM) were generated to quantify the degree of clustering of PTPN22-R620 in nonmigrating T cells (PLL, solid line) and migrating T cells (ICAM-1, dashed line). Data are representative of three experiments, analyzing between 30 and 50 cells per experiment.

Csk also exhibited this declustering phenomenon upon stimulation of LFA-1 (15% of localizations were in clusters of migrating cells versus 30% in nonmigrating cells), although the heat maps revealed more heterogeneity in Csk clusters compared to those of PTPN22 (fig. S4, A to C). Concurrent with the LFA-1–dependent declustering of PTPN22 and Csk, we detected increased association between the phosphatase and the kinase in cell lysates immunoprecipitated with antibodies against either PTPN22 or Csk (fig. S4, D and E), whereas the adaptor protein associated with glycolipid-enriched microdomains (PAG, also known as Csk-binding protein) (22) had less Csk and PTPN22 associated with it in ICAM-1–stimulated cells (fig. S4, F and G). Consistent with these findings, tyrosine phosphorylation of PAG, which promotes the association between PAG and Csk and the retention of Csk in plasma membrane microdomains (23), was also reduced upon engagement of LFA-1 (fig. S4H). These data indicate that the declustering of both PTPN22 and Csk coincides with the dissociation of Csk from dephosphorylated PAG and the increased association of PTPN22 with Csk.

Nanoscale organization of PTPN22-R620 and PTPN22-W620 clusters in migrating T cells

Closer examination of PTPN22-R620 localizations at the nanoscale level revealed that the transition from the nonmigrating to the migrating state was associated with three cluster characteristics. First, there was a modest reduction in the total number of clusters (Fig. 3A, closed symbols). Second, PTPN22-R620–containing clusters became smaller, which is based on a marked reduction in the percentage of PTPN22 molecules that occurred within clusters (Fig. 3B, closed symbols), which is consistent with the cluster heat maps and the changes in the diameter and density of clusters upon stimulation with ICAM-1 (Fig. 2, A to C). Third, there was a marked increase in the number of PTPN22-R620 localizations at the leading edge (Fig. 3C, closed symbols). Thus, in response to the stimulation of LFA-1 by ICAM-1, PTPN22 molecules were dispersed from clusters and accumulated at the plasma membrane in closer proximity to LFA-1 signaling intermediates.

Fig. 3 Retention of PTPN22-W620 at the plasma membrane is impaired.

(A to E) Primary T cell blasts were generated from the peripheral blood of PTPN22-R620 and PTPN22-W620 homozygous donors and layered onto PLL- or ICAM-1–coated plates for 20 min before they were fixed, permeabilized, and stained with mouse anti-PTPN22 antibodies. N-STORM images were acquired as described for Fig. 2, and cluster analysis was used to define (A) the number of PTPN22-R620 and PTPN22-W620 clusters per region (n = 126 regions), (B) the percentage of the indicated PTPN22 variant localizations in the clusters (n = 126 clusters), and (C) the number of the indicated PTPN22 variant localizations per region (n = 126 regions). Bars represent means ± SD. *P < 0.02; ****P < 0.0001; ns, not significant. (D) Pointillist (Gaussian-fitted) and pseudocolored heat maps are representative of cluster data acquired after the processing of N-STORM image regions, as described for Fig. 2A. Scale bars, 5 μm for the PLL condition; 10 μm for the ICAM-1 condition. Data show the analysis of 50 cells per experiment and are representative of three experiments. (E) Ripley’s K function curves (mean ± SEM) were constructed to quantify the degree of clustering of PTPN22-R620 (dashed line) and PTPN22-W620 (solid line) in T cells migrating on ICAM-1. Data are representative of three independent experiments. (F) T cells expressing PTPN22-R620 (RR) or PTPN22-W620 (WW) derived from homozygous donors were plated on PLL- or ICAM-1–coated surfaces for 20 min. (Left) The cells were then lysed, subjected to immunoprecipitation with anti-Csk antibody, and analyzed by Western blotting with antibodies against the indicated proteins. Western blots are representative of three experiments. (Right) Quantification of the ratio of the abundance of PTPN22 to that of Csk, relative to that for cells plated on PLL. Data represent means of three independent experiments. Tukey’s ordinary one-way analysis of variance (ANOVA) multiple comparisons test, ***P < 0.0005; **P < 0.005; ns, not significant.

The clustering characteristics of the disease-associated W620 mutant PTPN22 were similar to those of the common R620 variant in nonmigrating cells (Fig. 3D, top; compare open versus closed symbols for cells plated on PLL in Fig. 3, A to C). In contrast, migrating T cells expressing the PTPN22-W620 mutant had increased numbers and density of clusters (Fig. 3, A and B, open symbols), which were evident in the region point maps and heat maps (Fig. 3D, bottom). The amplitude of Ripley’s K function also suggested that PTPN22-W620 was more clustered than PTPN22-R620 (Fig. 3E). Unexpectedly, the total number of molecules of PTPN22-W620 at the plasma membrane was substantially reduced compared to that of PTPN22-R620 after stimulation of LFA-1 (Fig. 3C, open versus closed symbols for ICAM-1). This finding was consistent with the point maps, which showed reduced numbers of localizations outside of the clusters (Fig. 3D, bottom, and fig. S5A), which could not be explained by differences in protein abundances at the whole-cell level (fig. S5, B and C).

Using simulated data to model the consequences of varying the density of the nonclustered background of molecules that surround clusters, we found that both the linearity and the gradient of Ripley’s K function depended on the number of nonclustered localizations (fig. S5, D and E). These simulations, together with the observed Ripley’s K function data (Fig. 3E), indicated that differences between the clustering of PTPN22-R620 and PTPN22-W620 were not a result of changes in the clustering behavior of W620 but rather were a consequence of the lack of localization of PTPN22-W620 outside clusters, a finding that was consistent with the pointillist maps (fig. S5A). We surmised that declustered PTPN22-W620 molecules were not retained at the plasma membrane in migrating T cells to the same extent as were declustered PTPN22-R620 molecules. To explore the mechanism behind this difference in plasma membrane localization, we tested whether the P1 domain mutation in PTPN22-W620 compromised its binding to SH3 domain–containing protein partners, such as Csk, in response to integrin stimulation. Homozygous donor-derived PTPN22-R620– or PTPN22-W620–expressing T cells that migrated on ICAM-1 were lysed, subjected to immunoprecipitation with anti-Csk antibody, and analyzed by Western blotting with an anti-PTPN22 antibody. The results demonstrated that the association between Csk and PTPN22-W620 was reduced in migrating T cells compared to that between Csk and PTPN22-R620 (Fig. 3F). This suggests that upon stimulation of LFA-1, the retention of PTPN22 at the plasma membrane depends on the formation of a complex with Csk or other SH3 domain–containing proteins.

PTPN22 associates with the LFA-1 signaling complex and inhibits LFA-1 clustering

The LFA-1–dependent adhesion and migration of T cells are regulated by conformational changes in the extracellular domain of the α and β chains of the integrin, as well as the physical clustering of heterodimers at the plasma membrane (2426). Furthermore, Lck and ZAP70 associate with the cytoplasmic tail of the β chain of ligand-bound LFA-1 (17, 27). Initial evidence for an association between LFA-1 and its regulator, PTPN22, was derived from confocal images showing that LFA-1 and PTPN22 were colocalized at the leading edge (Fig. 4A). By TIRF microscopy, the colocalization of LFA-1 and PTPN22 was also observed at the interface with ICAM-1 at the leading edge (Fig. 4B), returning a Manders’ colocalization coefficient (MCC) of 0.43 ± 0.14. This association was also supported by coimmunoprecipitation experiments, in which LFA-1–PTPN22 interactions were observed to increase as a function of LFA-1 engagement (Fig. 4C). To address the mechanism for this association, we examined immunoprecipitates from the human leukemic Jurkat cell line and its Lck-deficient derivative JCaM1.6 (which both have comparable amounts of ZAP70 and PTPN22), and we found that PTPN22 did not appear to form complexes with LFA-1 in the absence of Lck (Fig. 4D and fig. S6). These data suggest that PTPN22 associates with the LFA-1 signaling complex in an Lck-dependent manner, where it interacts with substrates to regulate integrin signaling.

Fig. 4 PTPN22 colocalizes with LFA-1 at the leading edge of migrating T cells.

(A and B) T cells migrating on an ICAM-1–coated surface were stained with antibodies specific for PTPN22 (green) and LFA-1 (red) and imaged by (A) confocal microscopy or (B) TIRF microscopy. The direction of migration is shown by white arrows. (B) (Left) The boxed region is shown under higher magnification, and the colocalization of PTPN22 and LFA-1 is indicated by small white arrows. (Right) Bright-field images with intensity scales for PTPN22 and LFA-1 in stained cells. Data are representative of regions selected from the leading edges of 30 to 40 cells from four independent experiments. (C) (Left) T cells that were plated on PLL- or ICAM-1–coated surfaces were lysed, subjected to immunoprecipitation with anti–LFA-1 antibody or control IgG, and analyzed by Western blotting with antibodies against the indicated proteins. Western blots are representative of three experiments. (Right) Quantification of the ratio of the abundance of PTPN22 to that of LFA-1 from T cells plated onto ICAM-1 relative to that for cells plated on PLL. Data are means derived from three independent experiments. (D) Jurkat cells (JK) and their Lck-deficient derivatives (JCaM1.6) were plated onto PLL-coated surfaces for 20 min before being lysed. (Left) Cell lysates were analyzed by Western blotting with antibodies against the indicated proteins. (Right) Cell lysates were subjected to immunoprecipitation with mouse anti-PTPN22 antibody or control IgG and analyzed by Western blotting with antibodies specific for the indicated proteins. Western blots are representative of two experiments (for the reciprocal experiment, see fig. S6). ms, mouse antibody; gt, goat antibody.

To examine the relationship between the regulation of LFA-1 signaling by PTPN22 and the function of LFA-1, we first investigated how the loss of PTPN22 function might affect LFA-1 clustering at the cell surface. Superresolution TIRF-dSTORM maps showed increased clustering of LFA-1 at the leading edge of migrating T cells expressing the PTPN22-W620 variant compared to that in cells expressing PTPN22-R620 (Fig. 5A), which was corroborated by Ripley’s K function (Fig. 5B). Although there were no differences in the total number of LFA-1 localizations or in the number of clusters between cells expressing the two different PTPN22 variants (Fig. 5, C and D), the percentage of LFA-1 molecules participating in clusters were increased in T cells expressing the PTPN22-W620 variant (Fig. 5E). Thus, the loss of PTPN22 function was associated with increased LFA-1 clustering at the leading edge of migrating T cells.

Fig. 5 Expression of the PTPN22-W620 mutant enhances LFA-1 clustering at the leading edge of migrating T cells.

(A) T cells from homozygous donors expressing PTPN22-R620 or PTPN22-W620 were layered onto PLL- or ICAM-1–coated plates for 20 min before being fixed and stained with anti–LFA-1 antibody. Images were acquired, and the molecular distributions were analyzed as described in Fig. 2. For each PTPN22 variant, representative images are shown for their cell surface LFA-1 molecule distributions. Single regions were used to generate pointillist and pseudocolored cluster heat maps. Scale bar, 5 μm. Data show representative images from an analysis of 70 to 80 cells per genotype and are representative of three experiments. (B) Ripley’s K function curves (mean ± SEM) were constructed to quantify the degree of clustering of surface LFA-1 in migrating T cells that expressed PTPN22-R620 (dashed line) or PTPN22-W620 (solid line). Data are representative of three independent experiments. (C to E) Cluster analysis of N-STORM images derived from PTPN22-R620– and PTPN22-W620–expressing T cells migrating on ICAM-1 was used to quantify (C) the number of LFA-1 localizations per region (n = 83 regions), (D) the number of LFA-1 clusters (n = 83 regions), and (E) the percentage of LFA-1 localizations in clusters (n = 83 clusters). Data are means ± SD. ****P < 0.0001; ns, not significant.

Integrin-dependent adhesion is increased in T cells expressing the loss-of-function PTPN22-W620 variant

In light of these findings, we evaluated the functional consequences of enhanced LFA-1 clustering by directly comparing the adhesion of T cells expressing either PTPN22-R620 or PTPN22-W620 under conditions of shear flow. T cells expressing the disease-associated PTPN22-W620 variant were more adherent to ICAM-1 than cells expressing the R620 variant (Fig. 6A). Ptpn22−/− murine T cells phenocopied the disease-associated variant, being more adherent than wild-type T cells under shear flow (Fig. 6B) and more adherent in de-attachment assays (Fig. 6C). Thus, a loss-of-function PTPN22 mutant that enhanced LFA-1 signaling increased LFA-1 clustering and cell adhesion. Together, our results suggest that in migrating T cells, PTPN22 disperses from large, plasma membrane–proximal clusters into smaller clusters that are capable of interacting with the LFA-1 signaling complex, where inhibition of signaling leads to reduced integrin clustering and adhesion. If the localization of PTPN22 at the plasma membrane is compromised, regulation of integrin signals is uncoupled, and both integrin clustering and cell adhesion are enhanced.

Fig. 6 T cells expressing loss-of-function PTPN22 mutants or those deficient in PTPN22 are more adherent under shear flow.

(A) T cells expressing PTPN22-R620 or PTPN22-W620 derived from homozygous donors were flowed over glass slides coated with ICAM-1 (5 μg/ml) at a shear flow rate of 0.5 dyne/cm2 and imaged by time-lapse, wide-field microscopy. Cells adhering to ICAM-1–coated glass were counted every 1 min for a total of 8 min. Pooled data for cells of each genotype were derived from 12 independent experiments. *P = 0.049. (B) T cells were generated from the lymph nodes of Ptpn22+/+ and Ptpn22−/− littermate mice, and their adherence under shear flow was quantified as described in (A). A representative experiment (left) and pooled data (right) show the mean number of adherent cells ± SD after 10 min for cells of each genotype, based on six independent experiments. **P = 0.002. (C) T cells from the indicated mice were prepared as described in (B) and allowed to adhere to glass slides coated with ICAM-1 (5 μg/ml) for 10 min. Shear force was applied at the indicated flow rates, and time-lapse, wide-field movies were acquired. The numbers of adherent T cells were counted every 1 min for 10 min. Phase decay analysis was used to generate half-life values for cell attachment over time. Data represent 11 independent experiments. *P = 0.0472; **P = 0.005; ***P = 0.0008; ns, not significant.


Protein tyrosine phosphatases are now established as key regulators of integrin signaling (11, 12). The protein tyrosine phosphatase–PEST (PTP-PEST) family of phosphatases, which consists of PTPN12 and PTPN22, can be added to a growing list of inhibitors of integrin function, which includes DOK1 (RhoH, docking protein 1), calpain, and the ubiquitin ligase SHARPIN (SHANK-associated regulator of G protein signaling homology domain–interacting protein) (2831). Single-molecule localization microscopy enabled us to image PTPN22 at the nanoscale level and to document that the plasma membrane–proximal declustering of PTPN22 is linked both temporally and spatially to its inhibitory function. This is in contradistinction to the opposing clustering behavior reported previously for kinase-associated signaling modules (32, 33), which enables digital signaling and increases signal transduction fidelity (34). Whether clustering in the steady state is unique to PTPN22 or is a common mechanism for sequestering phosphatases from their substrates will require further study.

The inhibitory functions of PTPN22 were confirmed by gene targeting in mouse and human T cells. We also found that a catalytically active PTPN22 was required to inhibit integrin-dependent cell motility. Experiments with PTPN22-W620–expressing T cells from homozygous donors indicated that the disease-associated variant was a loss-of-function mutant, at least in the context of LFA-1 signaling. The basis for this functional difference is underpinned in part by the impaired binding of the mutated P1 polyproline domain of PTPN22-W620 with SH3 domain–containing proteins, notably Csk. The spatiotemporal dynamics of PTPN22-Csk interactions and the effect of disrupting these associations on PTPN22 function are complex and may be signal-specific. For example, Vang et al. (35) demonstrated that dissociation of PTPN22 from Csk is a prerequisite for targeting the phosphatase to plasma membrane lipid raft domains, where it attenuates T cell receptor (TCR) signaling. PTPN22-W620 partitions into rafts more efficiently than does PTPN22-R620, whereas the forced dissociation of PTPN22 from Csk with a recombinant Csk-SH3 domain also reduces TCR signaling. These data support the gain-of-function hypothesis with respect to the PTPN22-W620 variant and TCR signaling, and suggest that the effect of the disease-associated mutant is context-dependent. Whereas uncoupling of the association between Csk and PTPN22-W620 seems a consistent feature relevant to TCR- and integrin-dependent signaling (36, 37), we cannot rule out the possibility that disrupting the interactions between PTPN22 and other SH3 domain–containing proteins could contribute to the signaling phenotypes reported. A comparative biochemical analysis of the PTPN22-R620 and PTPN22-W620 interactomes, using technologies that were reported for PAG (38), would provide a systematic and unbiased approach to address this issue.

We present a model depicting how PTPN22 inhibits LFA-1 signaling and how the PTPN22-W620 mutant enhances LFA-1 signaling and integrin-dependent adhesion (Fig. 7). According to this model, PTPN22 exists in large clusters in the steady state, where it is sequestered from its substrates. Active signals stimulate declustering, which is an event that we are now studying in the context of TCR and LFA-1 stimulation and one that targets clustered pools of both PTPN22 and Csk. Precisely how the clusters disaggregate is not known, but the process serves to deliver monomers of PTPN22 to the plasma membrane zone, enabling interactions with its binding partners. Lck is constitutively associated with the β subunit of LFA-1 (17) and is required for the recruitment of PTPN22 to the LFA-1 signaling complex. In PTPN22−/− cells, the LFA-1–stimulated phosphorylation of these intermediates goes unchecked, manifesting as the enhanced phosphorylation of signaling intermediates and augmented integrin-dependent signaling, cell motility, and adhesion. The outcome of expressing the PTPN22-W620 is the same, except that in this case, the total cellular amounts of the phosphatase are equivalent to those in PTPN22-R620–expressing T cells, but the W620 variant fails to bind to Csk (or possibly other SH3 domain–containing proteins). Instead, PTPN22-W620 is distributed throughout the rest of the cell rather than being retained near the LFA-1 signaling complex. Failure to attenuate LFA-1 signals is associated with much larger, denser clusters of LFA-1 at the cell surface, equipping the cell with domains of increased adhesiveness.

Fig. 7 Mechanistic model for the regulation of integrin signaling by PTPN22.

The transition of LFA-1 from a low- or intermediate-affinity state (middle) to a high-affinity state (left) is characterized by the phosphorylation of Lck, ZAP70, and Vav, which are associated with the cytoplasmic tail of the β2 subunit of LFA-1. Subsequently, spatiotemporal regulation of LFA-1 signals is mediated by the dispersal of PTPN22 and Csk from clusters, the disassociation of PTPN22 and Csk from PAG at the plasma membrane, and the increased association of PTPN22 with Csk through the P1 domain of PTPN22 and the SH3 domain of Csk. PTPN22-Csk complexes target their phosphorylated substrates in the LFA-1 signaling complex, which leads to the attenuation of LFA-1 signaling. Although the declustering of the loss-of-function PTPN22-W620 mutant is preserved, the binding of PTPN22-W620 to Csk is impaired (right). The mutant phosphatase is not retained at the plasma membrane, and in the absence of membrane-proximal binding partners, such as Csk, PTPN22-W620 diffuses away from the plasma membrane. As a consequence, LFA-1 signal intensity is augmented and sustained, further promoting LFA-1 clustering at the cell surface and increasing integrin-dependent adhesion (right). The spatiotemporal organization of PTPN22-R620 (PTP), PTPN22-W620 (PTP-W), Csk, talin (Tal), Lck, Vav1 (Vav), ZAP70 (Zap), and kindlin is illustrated. P denotes phosphorylation on tyrosine residues.

How might this increased cell adhesiveness translate to altered cell function in vivo? We suspect that the consequences of increased adhesion under shear flow conditions could perturb multiple phases of the homing of cells to lymph nodes and tissues, including the adhesion of cells on vascular endothelium coupled to transmigration into tissues, trafficking across high endothelial venules within lymphoid organs, and interactions between cells or with the surrounding extracellular matrix. Integrins have other functions in addition to mediating adhesion and migration because they promote interactions between T cells and antigen-presenting cells in ways that underpin the earliest steps in T cell activation and differentiation (9, 39), as well as downstream effector responses, such as cytokine production and cytotoxic functions (46). We propose that aberrations in integrin function be included in the repertoire of mechanisms underpinning a predisposition to autoimmune disease in individuals carrying loss-of-function PTPN22 mutations (40). Loss of immune tolerance, however, will depend on the balance of function between effector and regulatory T (Treg) cells because our work has previously demonstrated that the increased adhesiveness of Ptpn22−/− Treg cells is associated with their greater potency (41).


Antibodies and integrin ligands

Mouse monoclonal antibody and affinity-purified goat polyclonal antibody raised against human PTPN22 were purchased from R&D Systems. Antibodies specific for Vav1 (C-14), ZAP70 (1E7.2), Lck (3A5), and Csk (C20) were from Santa Cruz Biotechnology. Rabbit polyclonal antibodies against pSrc (Tyr416), pZAP70 (Tyr319/Syk-Tyr352), pZAP70 (Tyr493/Syk-Tyr526), p44/42 MAPK (ERK1/2) (Thr202/Tyr204) (197G2), α- and β-tubulin, and β-actin were from Cell Signaling Technology; antibodies specific for pVav1 (Tyr174) (EP5107), PAG (ab14989), and CD11a were from Abcam. The monoclonal anti–LFA-1 antibody mAb38 was a gift from N. Hogg (Francis Crick Institute, London, U.K.). Horseradish peroxidase (HRP)–conjugated sheep anti-mouse IgG (Amersham), goat anti-rabbit Ig–HRP (Dako), mouse monoclonal light chain–specific anti-goat IgG (Jackson Laboratory), goat anti-mouse IgG, goat anti-rabbit IgG, donkey anti-mouse IgG, donkey anti-rabbit IgG, donkey anti-goat IgG, and the IgG Zenon antibody labeling kit were from Life Technologies. Recombinant human ICAM-1/CD54 Fc chimera, recombinant murine ICAM-1/CD54 Fc chimera, recombinant VCAM-1, and fibronectin were obtained from R&D Systems.


Ptpn22−/− mice were generated and genotyped as described previously (41). The line was rederived into the Biological Services Unit at King’s College London and bred on a C57BL/6 background for 10 generations under specific pathogen–free conditions, in compliance with the Home Office regulations and local ethically approved guidelines. Sex- and age-matched Ptpn22+/+ and Ptpn22−/− littermates were used in experiments.

Media, cell culture, and transfection

Human T cells were cultured in complete medium [Iscove’s modified Dulbecco’s medium (IMDM), 10% fetal bovine serum (FBS), penicillin, and streptomycin], whereas mouse T cells were cultured in Glutamax-RPMI, 10% FBS, 50 μM β-mercaptoethanol (β-ME), 100 μM sodium pyruvate, 20 mM Hepes, penicillin, and streptomycin. Glutamax-RPMI, 50 μM β-ME, 100 μM sodium pyruvate, 25 mM Hepes, penicillin, and streptomycin were used as cell migration medium. Human peripheral blood mononuclear cells (PBMCs) were isolated from whole blood with Lymphoprep (STEMCELL Technologies) and were stimulated with phytohemagglutinin (1 μg/ml, Thermo Fisher Scientific) in IMDM medium supplemented with 10% fetal calf serum, penicillin, and streptomycin (PAA) for the first 48 hours and with interleukin-2 (IL-2; 20 ng/ml, aldesleukin, Novartis) for up to 10 days. T cells were purified with a Pan T Cell Isolation kit (catalog no. 130-091-156, Miltenyi Biotec) to a purity of >97%, as determined by flow cytometry analysis. Donors belonging to TwinsUK ( and selected on the basis of their rs2476601 genotype (PTPN22 C1858T, corresponding to PTPN22-R620W protein) also provided PBMCs after informed consent. Transfection of primary human T cell blasts was performed with the Amaxa Nucleofector and Human T Cell Nucleofection Kit, program T-020 (Lonza). Plasmid DNA (2 μg) and siRNA pools (Invitrogen, Thermo Fisher Scientific) were used to transfect 1 × 107 T cells. Mouse T cells were generated from splenic and lymph node cell suspensions, adjusted to a density of 3 × 106 cells/ml, and cultured in complete medium. Cells were stimulated with concanavalin A (1 μg/ml, Sigma) for 48 hours, subjected to a Ficoll gradient, and resuspended in complete medium supplemented with IL-2 (20 ng/ml) at a density of 2 × 106 cells/ml. All adhesion and migration assays were performed with mouse T cells after they had been cultured for 4 to 5 days in IL-2. The human leukemic T cell line, Jurkat, and its Lck-deficient derivative JCaM1.6 were also used in experiments (42).

Generation of PTPN22-GFP constructs

A pEF5HA plasmid encoding PTPN22 (a gift from N. Bottini, La Jolla Institute for Allergy and Immunology, La Jolla, CA) was sequenced and used as a plasmid backbone for site-directed mutagenesis to generate a panel of PTPN22 mutants, and a 3′ GFP fragment was introduced by subcloning. All constructs were verified by sequencing. Targeted mutations were introduced with the following specific primer pairs: R620W CCACTTCCTGTATGGACACCTGAATCATTTA (forward) and TAAATGATTCAGGTGTCCATACAGGAAGTGG (reverse); C227A, TGTTCCCATATGCATTCACGCCAGTGCTGGCTGTGGAAGGACTGG (forward) and CCAGTCCTTCCACAGCCAGCACTGGCGTGAATGCATATGGGAACA (reverse).

T cell stimulation and signaling with integrin ligands

Glass coverslips (32 mm, VWR International) were coated in six-well sterile plates with the integrin ligands ICAM-1–Fc (3 μg/ml, unless indicated otherwise), VCAM-1–Fc (3 μg/ml), or fibronectin (10 μg/ml) or with PLL (Sigma) overnight at 4°C, washed three times with phosphate-buffered saline (PBS), and blocked for 1 hour at room temperature with 2% bovine serum albumin (BSA) in PBS. T cell blasts were rested in migration medium for 30 min and then added to coverslips (at 3 × 106 cells per coverslip) and incubated for 20 min at 37°C. Unbound cells were then aspirated, and 1 ml of lysis buffer was added to the coverslips (1 × 6-well plate, 2 × 107 cells per well) before being lysed on ice for 20 min. Lysates were cleared by centrifugation and diluted in sample buffer for Western blotting or immunoprecipitation experiments.

Western blotting and immunoprecipitations

Cells were lysed directly in 2× SDS–polyacrylamide gel electrophoresis (SDS-PAGE) sample buffer or in lysis buffer containing 1% Triton X-100 (Sigma) with phosphatase and protease inhibitors (Roche). Proteins were separated by SDS-PAGE and transferred to Immobilon-P polyvinylidene difluoride membranes by standard Western blotting techniques. After incubation with primary and secondary antibodies, reactive bands on the blots were visualized by SuperSignal chemiluminescent reaction (Pierce Biotechnology) in a ChemiDoc Station (Bio-Rad). For immunoprecipitations, 1 to 2 μg of control or specific antibody were added to cell lysates overnight at 4°C, which was followed by the addition of 20 μl of magnetic beads (Millipore) and incubation for 1 hour at 4°C. The beads were washed three times in lysis buffer before being eluted with 20 μl of boiling 2× SDS-PAGE sample buffer. Cell lysates and immunoprecipitates were used immediately or were stored at −80°C until needed for analysis.

Immunofluorescence staining and microscopy

Coverslips, glass-bottomed dishes (MatTek), or eight-well glass-bottomed microscopy chambers (ibidi) were coated overnight at 4°C with human recombinant ICAM-1–Fc (3 μg/ml) or 0.01% PLL, washed three times in Hanks’ balanced salt solution, and blocked in 5% BSA for 1 hour. T cells were resuspended at 2.5 × 105 cells/ml in migration medium that had been equilibrated overnight at 37°C in 5% CO2 and added to glass coated with ICAM-1–Fc. After 20 min of migration, the cells were pH shift–fixed [that is, they were treated for 5 min with 3% paraformaldehyde (PFA)–80 mM Pipes dipotassium salt (pH 6.8), supplemented with 2 mM Mg2+ and 5 mM EGTA, which was followed by treatment for 10 min with 3% PFA and 100 mM Borax (pH 11)] and then permeabilized with 0.1% Triton X-100 for 5 min at 4°C. Autofluorescence was quenched by treating the samples with NaBH4 (1 mg/ml) for 15 min. The chambers were blocked with 10% goat serum for 1 hour and then incubated with primary antibody overnight and an appropriate secondary antibody for 20 min at room temperature. Confocal microscopy was performed with a Zeiss LSM 700 Axio Imager M2 system at ×63 magnification (Plan-Apochromat 63×/1.40 Oil M27, Zeiss; zoom, 4; 512 × 512 pixel scan). TIRF images were obtained with a Zeiss observer Z1 (inverted) microscope equipped with a TIRF slider [Plan-Apochromat 100×/1.40 Oil DIC (UV) VIS-IR, Zeiss]. Images were collected, processed, and analyzed with SlideBook 5.5 (3i) or ImageJ software. Colocalization of signaling proteins was determined according to the method of Dunn et al. (43). Briefly, TIRF images were processed with the automatic method of local background subtraction by median filtering and finally by small-value subtraction with SlideBook6 software. Four regions (1.6 μm × 1.6 μm) were chosen in the front lamella of the cell. Subsequently, MCC was calculated with the inbuilt SlideBook6 function. Results from five regions were averaged per cell in four different experiments and analyzed with Prism 6.0 software (GraphPad).

Direct stochastic optical reconstruction microscopy

dSTORM imaging was performed on a Nikon N-STORM microscope with a 100× 1.49 numerical aperture oil immersion TIRF objective. Cells were imaged under TIRF illumination with a 647-nm laser with photoactivation at 405 nm in oxygen-scavenging buffer [including glucose oxidase (50 μg/ml), HRP (25 μg/ml), and 75 mM cysteamine in base buffer (pH 8.0)]. Fluorescence was collected on an Andor iXon EM-CCD camera. Acquisition time was between 5 and 15 min, with an integration time of 10 ms. Molecular coordinates were calculated with Nikon NIS N-STORM software using a photon threshold of 3000 per point spread function.

Cluster analysis

The output from the NIS N-STORM software was in the form of pointillist x-y coordinates of the localized fluorophores. Data were divided into nonoverlapping 2 μm × 2 μm square regions, avoiding cell boundaries. Ripley’s K function can be used to quantify the level and size scale of molecular clustering, and the use of this function has previously been demonstrated in experiments with T cells (4446). The K function was calculated for each region with the Excel plug-in SpPack (47) with the following equation:K(r)=An2i=1nj=1nδijwhere δij = 1 if the distance between molecules i and j is less than r; otherwise, δij = 0. Therefore, δij represents the drawing of concentric circles of radius r around each point, i, and counting how many other points, j, are encircled. As defined, the K(r) value scales linearly with the circle area for increasing r and is therefore converted into the L function with the following equation:L(r)=K(r)/π

This function is represented by a plot of L(r)-r versus r. In the case of a completely spatially random (CSR) distribution of molecules, L(r)-r = 0 for all r. If L(r)-r is positive, this represents clustering on a particular spatial scale, r. A negative L(r)-r value represents a more regular distribution than that of CSR (negative clustering). Edge effects were corrected by means of a toroidal wrap. Confidence intervals (95%) were calculated by simulating 100 CSR distributions with the same total molecular density as that of the experimental data. To test the effect of reducing the density of monomers (that is, the CSR background) on the Ripley’s K function curves, we analyzed simulated data. A 3 μm × 3 μm region was simulated with a single Gaussian profile cluster at its center (SD = 100 nm, 100 points), which was then overlaid with a CSR background. We then varied the density of the CSR overlay and analyzed the resulting Ripley’s K function curves. In the presence of zero background, the curve decays linearly to negative infinity at increasing r; however, in the presence of increasing background, the curve asymptotically approaches zero (as is the case for a pure CSR distribution). To generate the pseudocolored cluster maps, the degree of clustering of each molecule was calculated with Getis’ variant of Ripley’s K function (48). This is simply the L(r) value omitting averaging over all molecules in the region (j). Therefore, for each molecule, i, this value is given by the following equation:L(r)i=Anj=1nδij/π

Molecules at the edge of the region of interest had their L(r) value corrected with a buffer region of width r. To generate the cluster maps, the L(r) values were interpolated onto a 5-nm resolution grid with MATLAB software, and the L(r) color surface was pseudocolored. To extract cluster statistics, the map was thresholded at a value of L(r) = 200, with areas above this value considered to be clusters. Clusters were separated with an 8-connectivity rule that enabled us to extract the number of clusters, cluster sizes, number of molecules per cluster, and other parameters.

Flow cytometry

T cell blasts were stained in cold flow cytometry (fluorescence-activated cell sorting) buffer (0.5% BSA and 0.01% sodium azide in PBS) with antibodies and live/dead discrimination, washed, fixed with 2% PFA, and analyzed with a FACSCalibur flow cytometer. Data analysis was performed with FlowJo software (Tree Star Inc.).

Time-lapse microscopy

To monitor cell attachment under shear flow, glass-bottomed flow chambers (μ-Slide VI0.4, ibidi) were coated with ICAM-1–Fc (5 μg/ml), and T cell blasts (1 × 106 cells/ml) were flowed over the glass at 0.5 dyne/cm2 for 8 to 10 min in migration medium. Wide-field, time-lapse movies were acquired in 10 areas per slide. Cell numbers were counted over time, and the data were presented as means ± SD. To monitor cell detachment under shear flow, T cells isolated from Ptpn22+/+ and Ptpn22−/− mice (5 × 106 cells/ml in migration medium) were allowed to adhere to glass-bottomed flow chambers (μ-Slide VI0.1, ibidi) coated with ICAM-1–Fc (5 μg/ml) for 10 min, and then migration medium (incubator-equilibrated overnight) was applied at shear flow rates from 5 to 30 dynes/cm2. Wide-field, time-lapse images were acquired for 10 min, and cell counts were accrued every 1 min. Data were plotted exponentially reflecting loss of cell numbers over time, and the phase decay half-life was calculated with Prism 6.0 software (GraphPad). Values represent the rate at which the T cells detached from the glass over time; lower values indicate increased detachment.

Statistical analysis

All statistical analyses were performed with Prism 6.0 software (GraphPad). Distributions of data points and their variance were determined, and parametric or nonparametric tests were applied, as appropriate. Comparisons between two groups were evaluated with a Mann-Whitney U test; unpaired Student’s t tests were used for normally distributed data. Comparison of three or more independent conditions was determined with the rank-based nonparametric Kruskal-Wallis H test or with Tukey’s ordinary one-way ANOVA multiple comparisons test for normally distributed data. For comparison of cell motility, cell adhesion, and clustering data between groups, unpaired two-tailed t tests were used. Student’s t tests were also applied to compare densitometric measurements between independent Western blotting experiments. Differences were considered to be statistically significantly different when P < 0.05.


Fig. S1. PTPN22 and its phosphorylated substrates localize to the leading edge of migrating T cells.

Fig. S2. TIRF microscopy confirms the colocalization of PTPN22 with its phosphorylated substrates.

Fig. S3. Engagement of integrin leads to the phosphorylation of PTPN22 substrates in migrating T cells.

Fig. S4. Csk declusters upon engagement of LFA-1 and associates with PTPN22 in migrating T cells.

Fig. S5. Analysis of cluster characteristics and the abundances of PTPN22-R620 and PTPN22-W620 in migrating T cells.

Fig. S6. PTPN22 associates with LFA-1 in an Lck-dependent manner.


Acknowledgments: We thank N. Hogg for providing anti–LFA-1 antibodies, N. Bottini for supplying PTPN22 cDNA constructs, N. Hogg and S. Fagerholm for helpful comments on the manuscript, and T. Prevost for advice on all statistical analysis. Funding: This research was supported by Arthritis Research UK grants 16409 (to R.Z.), 19652 (to A.P.C., L.M.S., and R.Z.), 20525 (to G.H.C., A.P.C., L.M.S., and R.Z.), and 20848 (to V.L.M.); a Royal Society collaborative travel grant (to L.M.S. and A.P.C.); Wellcome Trust project grant 086054 (to R.J.B., A.P.C., and R.Z.); Wellcome Trust Investigator Award 096669 (to R.Z.); a Strategic Award from the Wellcome Trust for the Centre for Immunity, Infection and Evolution (095831 to R.Z.); Swedish Medical Council awards K2010-80P-21592-01-4 and K2010-80X-215917-01-4; Stiftelsen Olle Engquist Byggmästare, I&A Lundberg forskningsstiftelse, Gyllenstiernska Krapperups-stiftelsen, Gustav V 80 jubileum fond, Nanna Svartz, and Crafoord awards (to L.M.S.); Anna-Greta Crafoords postdoctoral fellowship (to K.P.); a European Research Council starter grant (337187 to D.M.O.); and a Marie Curie Career Integration Grant (334303 to D.M.O.). This work was also supported by infrastructure funded by the National Institute for Health Research Biomedical Research Centre at Guy’s and St. Thomas’ NHS Foundation Trust and King’s College London (reference: guysbrc-2012-17) and by the Nikon Imaging Centre at King’s College London. Author contributions: G.L.B., G.H.C., K.P., and M.S. performed most of the experiments and analyzed the data; M.S., J.G., S.M., M.Y., G.A., and N.P. contributed to the imaging experiments and their analysis, including simulations; V.L.M. contributed to the analysis of cell adhesion under flow; C.S.-B., H.P., F.C., and R.J.B. contributed to the execution and analysis of mouse experiments; T.J.V. identified and provided PTPN22-genotyped donors; G.L.B., G.H.C., R.Z., D.M.O., L.M.S., and A.P.C. designed the experiments and analyzed the data; and G.L.B., G.H.C., L.M.S., and A.P.C. wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Signaling

Navigate This Article