Functional nanoscale organization of signaling moleculesdownstream of the T cell antigen receptor

Eilon Sherman1, Valarie Barr1, Suliana Manley2, George Patterson2, LakshmiBalagopalan1, Itoro Akpan1, Carole K. Regan1, Robert K. Merrill1, Connie L. Sommers1,Jennifer Lippincott-Schwartz2, and Lawrence E. Samelson1,*

1Laboratory of Cellular and Molecular Biology, CCR, NCI, NIH, Bethesda, MD, USA 208922Cell Biology and Metabolism Program, NICHD, NIH, Bethesda, MD, USA 20892

AbstractReceptor-regulated cellular signaling often is mediated by formation of transient, heterogeneousprotein complexes of undefined structure. We used single and two-color photoactivated-localization microscopy (PALM) to study complexes downstream of the T cell antigen receptor(TCR) in single molecule detail at the plasma membrane of intact T cells. The kinase ZAP-70distributed completely with the TCRζ chain and both partially mixed with the adapter LAT inactivated cells thus showing localized activation of LAT by TCR-coupled ZAP-70. In resting andactivated cells LAT primarily resided in nanoscale clusters as small as dimers whose formationdepended on protein-protein and protein-lipid interactions. Surprisingly, the adapter SLP-76localized to the periphery of LAT clusters. This nanoscale structure depended on polymerizedactin and its disruption affected TCR-dependent cell function. These results extend ourunderstanding of the mechanism of T cell activation and the formation and organization of TCR-mediated signaling complexes, findings also relevant to other receptor systems.

IntroductionSignaling complexes are heterogeneous, dynamic multi-molecular structures inducedfollowing receptor binding and which mediate signal transduction. They are nucleated byactivated receptors and membrane-bound scaffold proteins and are often found inmicroclusters (Cebecauer et al., 2010; Harding and Hancock, 2008; Schlessinger, 2000).Using diffraction-limited light microscopy, studies of T cell activation have demonstratedthat microclusters form at the plasma membrane (PM) of T cells upon engagement of theantigen specific receptor (TCR) (Campi et al., 2005; Yokosuka et al., 2005). Thesemicroclusters contain molecular complexes comprised of the TCR, signaling enzymes suchas ZAP-70 and phospholipase C-γ1 (PLC-γ1) and several adapter proteins including LAT,SLP-76, Grb2 and Gads (Bunnell et al., 2002). Similar structures have been seen in B cellsand mast cells following engagement of the immunoreceptors in these cells (Harwood andBatista, 2010b; Wilson et al., 2001).

© Published by Elsevier Inc.*Correspondence should be addressed to: Dr. Lawrence E. Samelson, Laboratory of Cellular and Molecular Biology, CCR, NCI, NIH,37 Convent Dr., Bethesda, MD 20892-4256, 301-496-9683, samelson@helix.nih.gov.Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to ourcustomers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review ofthe resulting proof before it is published in its final citable form. Please note that during the production process errors may bediscovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

NIH Public AccessAuthor ManuscriptImmunity. Author manuscript; available in PMC 2012 November 23.

Published in final edited form as:Immunity. 2011 November 23; 35(5): 705–720. doi:10.1016/j.immuni.2011.10.004.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Although signaling complexes and microclusters have been studied extensively by variousbiochemical and imaging techniques, the spatial organization of individual molecules withinsignaling complexes or of complexes within microclusters has not been addressed. Muchremains to be learned about complexes and microclusters including, their full sizedistribution, their potential heterogeneity, their arrangement in the PM and the molecularrequirements for their formation. Imaging the spatial relationship of molecules at the singlemolecule level might resolve these questions. Toward this goal, we studied TCR interactionswith signaling molecules and signaling complexes downstream of the TCR. We found ananoscale organization to these interactions and structures with unexpected propertiesrelevant for understanding T cell activation.

ResultsMost LAT molecules are not within microclusters

We began by studying the organization of LAT at the PM by diffraction-limited microscopy.Cells stably expressing LAT-Dronpa [(Ando et al., 2004); Figure S1A,B] were dropped andimaged on antibody coated coverslips (Bunnell et al., 2003) that either stimulate the TCR(αCD3) or avoid TCR stimulation (αCD45). Using confocal microscopy, LAT moleculesshowed pronounced signaling microclusters in both Jurkat T cells (Figure 1A, αCD3) andhuman peripheral blood T cells (PBTs) upon activation (Figure S1C, αCD3+αCD28). Oncontrol, non-activating surfaces, there were fewer cells with LAT clusters as detected byintensity thresholding (Figure 1A, αCD45, Figure S1D,F, and see SI for further details). Theextent of spreading under both conditions was similar (Figure S1E,G), but there was a farlower extent of LAT or phosphorylated proteins in clusters under non-stimulating conditions(Figure S1H). Importantly, we note that LAT microclusters accounted for only a small partof total LAT molecules at the cell surface as measured by total fluorescence using eitherconfocal (~24% and ~8% for stimulating and non-stimulating conditions; see Figure S1I) ortotal internal reflection (TIRF) microscopy (~20% and 12% for stimulating and non-stimulating conditions; see Figure S1J). These observations emphasize the need tocharacterize the role and organization of all of the LAT molecules at the PM including thosethat are not in microclusters (Lillemeier et al., 2010; Lillemeier et al., 2006; Wilson et al.,2001).

LAT resides in pre-existing nanoclusters with a shift towards larger clusters uponstimulation

To better characterize PM LAT clustering with single molecule resolution, we usedphotoactivated-localization microscopy (PALM) imaging to observe individual LATmolecules, tagged with photoactivatable fluorescent proteins [LAT-Dronpa (Ando et al.,2004) or LAT-PAmCherry (Subach et al., 2009)]. We chose stimulation conditions that bothmaximized cell spreading and the presence of LAT in the narrow optical section detected byPALM in TIRF mode (Figure S2A,B). PALM images from Jurkat T cells (Figure 1B,H) andnormal human peripheral blood T cells (PBTs) (Figure 1E) were analyzed under activatingand non-activating conditions. We measured the probability density of locating individualLAT-PAmCherry molecules with a precision of ~20nm (Figure S2C). We next analyzed thescale of LAT clustering using pair-correlation functions (PCFs or g(r); Figure 1C,F,I; FigureS1L; see also Analyses section in SI). In addition, using a cluster analysis algorithm and acareful choice of the distance threshold for clustering (see Analyses section in SI and FigureS1M,N), we were able to identify individual LAT clusters in the PM and describe their sizedistribution in detail (Figure 1D,G,J). We found that LAT was always significantly pre-clustered under non-stimulating conditions (i.e., αCD45-coated coverslips) as well asactivating conditions on the PM of both Jurkat (Figure 1B-D,H-J) and PBTs (Figure 1E-G).Clustering can be observed in univariate PCF data (Figure 1C,F,I) when the value of g(r)

Sherman et al. Page 2

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

(bold lines) exceeds the 95% confidence interval due to calculated random moleculardistribution (dashed lines; refer to the Heterogeneous Poisson null model in the Analysessection in SI). PCFs also revealed that molecules in defined clusters were separated by aslittle as 20 nm. The majority of LAT molecules resided in very small nanoclusterscontaining only two, three or a few detectable molecules (Figure 1D,G,J). Though smallnanoclusters predominated, there was a wide size distribution as indicated by the cluster sizestatistics (Figure S1L).

We were struck by the presence of LAT clustering as seen by confocal (Figure S1D) andPALM imaging (Figure 1B,E,H) in cells spread on the coverslip in response to non-activating αCD45 antibodies. Therefore, we tested coverslips coated with antibodies againstother surface molecules including CD28, CD43 or CD18, which also lead to cell adherenceand spreading without activating the TCR (Figure 1K; confocal imaging results not shown).Regardless of which control antibodies we used or their effect on cell or microclusterappearance, we found that LAT was distributed in nanoclusters (Figure 1L). We concludethat in the basal state LAT is organized in nanoclusters at the PM in Jurkat and PBT cells.

Upon stimulation with αCD3, we observed an increase in the extent of LAT clustering(Figure 1C,F,I) and a modest shift towards larger clusters (Figure 1D,G,J) in the sizedistribution curves, without a change in the overall shape of the distributions (Figure1D,G,J). These changes were more pronounced in Jurkat T cells than in PBT cells.Differences between cell types and cell-to-cell variability in the pattern of cellular spreadingand the extent of protein expression contributed to the differences in the amount ofclustering observed in our analyses (Figure 1C,D,F,G,I,J). A similar size distribution wasdetermined for LAT clusters at the PM of both fixed and live cells (Figure 1H-J and FigureS1K) imaged under stimulating and non-stimulating conditions. The live cell PALM resultconfirms that the nanocluster distribution we observed was not an artifact of fixation.

Two factors might contribute to the dominance of small nanoclusters in the size-distributionand PCFs derived from unstimulated and stimulated cells. First, most of the LAT moleculesat the PM do not reside in microclusters (Figure S1I,J). Second, large aggregates such asmicroclusters are broken down by our analyses into their smallest detectable constituents,such that a single microcluster viewed by diffraction-limited microscopy is revealed to be acollection of much smaller nanoclusters.

LAT clusters are recruited to early contact areas within lamellaeTo further understand the significance of the irregular distribution of LAT on the coverslipand LAT organization in basal nanoclusters, we conducted two-color PALM imaging ofcells expressing the fluorescent conjugates LAT-Dronpa and a PM marker, TAC-PAmCherry (Subach et al., 2009) (TAC is CD25, the IL-2 receptor alpha subunit; seeMaterials and Methods section in SI and Figure S2D,E regarding two-color PALMimaging). In this experiment LAT clusters were found at the sites of contact where PMlamellae containing the randomly distributed TAC-PAmCherry first touched the stimulatingsurface (Figure 2A). As the cells spread over time, more of the PM made contact with thesurface and only LAT clusters were retained with the irregular lamellar pattern at sites ofinitial cellular contact (Figure 2B; see also Figure S2A). An irregular distribution of PM atthe coverslip was also observed using non-stimulating conditions (Figure 2C). We concludethat these results explain the first-order heterogeneity of LAT distribution, which we defineas the irregular distribution of micron-sized LAT clusters at the PM. That is, LATmicroclusters first appear and persist in areas of membrane contact with the activatingsurface, resulting in micron-sized features that reflect these initial contacts. Importantly thisexperiment also shows that second-order molecular heterogeneity, the nanoscaleorganization of LAT microclusters, is not a feature of all membrane proteins. The PCF of

Sherman et al. Page 3

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

TAC showed that it is randomly distributed (the TAC PCF curve falls within the 95%confidence interval of the heterogeneous Poisson null model) and not clustered within thePM (Figure 2D; for data of unstimulated cells, see Figure S2F,G).

In the next section interactions between the TCR, ZAP-70 and LAT are extensively studiedusing two-color PALM and bivariate analyses. The distribution of these moleculesindividually can also be studied as above with univariate statistics (such as PCFs). Thedistribution of the TCR has been studied previously biochemically and by PALM(Lillemeier et al., 2010; Schamel et al., 2005). In T cells interacting with αCD45 on thecoverslip we found that the TCRζ chain was found in nanoclusters with a distributionfavoring small nanoclusters much like LAT (Figure S2H). Since the mechanism ofactivation in our studies is antibody-induced clustering of the TCR, we can not comment onthe distribution of the TCR following physiologic activation.

Two-color PALM captures the efficient interaction of TCRζ and ZAP-70 in activated T cellsWe used two-color PALM to study the organization of signaling molecules downstream ofthe TCR. We began by imaging TCRζ-Dronpa and ZAP-70-PAmCherry. Upon TCRactivation, the protein tyrosine kinase ZAP-70 is recruited to the phosphorylated zeta chainsof the TCR (TCRζ), becomes activated and then phosphorylates multiple tyrosines on LATand other protein substrates (Wang et al., 2010). Phosphotyrosines on LAT then serve asdocking sites for multiple adapters and enzymes (Balagopalan et al., 2010). As expectedfrom previous imaging and biochemical studies (Bunnell et al., 2002; Wang et al., 2010), wefound that ZAP-70 co-localized with TCRζ upon activation (Figure 2E) and was far lessefficiently recruited to TCRζ without TCR activation (Figure 2F). We used bivariatecorrelation statistics to test whether the two molecules, TCRζ and ZAP-70, mixed randomly.If they did, the bivariate curve (g12(r), bold lines in Figure 2G) would fall between the twocurves that define the 95% confidence interval of a random mixing model (dotted lines inFigure 2G), which serves as a null hypothesis (see Analyses section in SI). Displacement ofthe curve above the 95% confidence interval would indicate significant attraction betweenthe two molecules beyond the random mixing model, while displacement below the 95%confidence interval would suggest that repulsion or segregation govern the relativepositioning of the two molecules.

For the case of TCRζ and ZAP-70, the bivariate analysis demonstrated complete mixing ofthe two molecules following TCR activation (Figure 2G, red lines). This result shows thatZAP-70 is recruited to TCR clusters in a manner independent of TCR cluster size. Incontrast, the cells that engaged the non-activating surface revealed little interaction betweenTCRζ and ZAP-70 consistent with the little or absent TCRζ phosphorylation found inunactivated T cells (Figure 2G, green lines). We also note systematic differences in theabsolute height of the bivariate correlation curves and their 95% confidence intervalsbetween activated and non-activated cells. To quantify the significance of these differences,we plotted the normalized level of bivariate correlation for multiple individual cells at scalesbelow 60nm (essentially, the y-axis intercepts of the bold lines in comparison to theintercepts of the dotted lines shown in Figure 2G). The results indicate that TCRζ andZAP-70 consistently mix in a more homogenous way (approximately 100%) underactivating conditions than under non-activating conditions (far less than 50%, Figure 2H).

TCRζ and LAT exist in overlapping pools that segregate more upon T cell activation andserve as hot spots for LAT activation

The TCR and LAT have been suggested to exist in separate domains that concatenate uponcell activation without mixing (Lillemeier et al., 2010). We used 2-color PALM to imageLAT-Dronpa and TCRζ-PAmCherry expressed in Jurkat T cells on αCD3-coated or on

Sherman et al. Page 4

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

control αCD45-coated coverslips. Two-color PALM images showed nanoscale domains ofboth LAT and TCRζ that significantly overlapped under both activating and non-activatingconditions (Figure 3A,B).

In activated samples (Figure 3C) the bivariate analysis revealed homogeneous mixing of thetwo molecules at separations over ~450nm, but a strong deviation from this model belowthis length scale as indicated by the location of the curve below the zone defined for 95%confidence level of the homogeneous mixing (random labeling) model. However, undernon-activating conditions the bivariate correlation between the molecules was much closerto homogeneous mixing at all length-scales (Figure 3D). Comparing the differences betweenthe two conditions for multiple cells, we found that TCRζ and LAT tend to mix in a morehomogenous way under non-activating conditions over all size ranges and then segregatemore from each other upon activation (Figure 3H).

In light of the partial overlap of molecular distribution between TCRζ and LAT followingactivation, we were interested in the localization of ZAP-70 relative to LAT to provideinformation on the means by which LAT becomes phosphorylated. In contrast to itscomplete mixing with TCRζ, ZAP-70 showed only a partial overlap in distribution withLAT molecules upon T cell activation (Figure 3E,G,H) or under non-activating conditions(Figure 3F,G,H). This distribution was similar to the partial mixing of LAT with TCRζ(Figure 3A,C). Since we failed to detect efficient mixing of ZAP-70 with LAT withactivation, we propose that ZAP-70 phosphorylates LAT when it is bound to TCRζ or invery close proximity to this TCR subunit. At any point in time there is a limited number ofsuch “hot spots” where activated TCRs interact with and phosphorylate LAT.

Protein-protein and protein-lipid interactions are both required for intact LAT nanoclusterformation

The mechanisms responsible for the formation of LAT-nucleated signaling complexes havebeen studied with controversial results. Early studies identified LAT as a molecule residingin lipid rafts (Wilson et al., 2004; Zhang et al., 1998) and aggregation of lipid rafts has beenproposed as a mechanism for generating activation-induced signaling complexes (Harder,2004; Zhang et al., 1998). However, other studies favored the role of an extended network ofprotein-protein interactions as the formation mechanism (Bunnell et al., 2002; Douglass andVale, 2005; Houtman et al., 2006). Our discovery of a LAT nanoscale distribution led us tosearch for the molecular requirements for LAT nanoclustering. For this reason, weconducted two-color PALM imaging of wildtype (WT) LAT-Dronpa with various LATconstructs conjugated to PAmCherry (Figure 4A). We examined the clustering of a mutantthat is incapable of phosphorylation-dependent protein-protein interactions due to thereplacement of the four distal tyrosines of LAT with phenylalanines (Bunnell et al., 2006)(LAT4YF-PAmCherry) and a mutant in which the absence of two juxtamembrane cysteinesblocks palmitoylation and lipid raft recruitment (Zhang et al., 1998) (LAT2CA-PAmCherry). In a cell expressing both WTLAT-Dronpa and WTLAT-PAmCherry, bothmolecules were incorporated into nanoclusters (Figure 4B, Figure S3A) with homogeneousmixing as determined by bivariate analysis, indicating that the two chimeras reside in thesame clusters (Figure 4C, Figure S3D). In contrast, there was only a slight mixing ofWTLAT and LAT4YF (Figure 4D,E, Figure S3D). Moreover, there was no interaction insamples imaged on αCD45-coated coverslips (Figure S3E,F).

Recent studies have shown that LAT palmitoylation is required for normal transport of LATto the PM (Hundt et al., 2009). However, our TIRF-based assay allowed us to image thefraction of LAT2CA molecules that resided in the membrane. We observed completeindependence (see Analyses section in SI) of WTLAT and LAT2CA (Figure 4F,G, FigureS3D). The univariate statistics also revealed interesting information. LAT4YF molecules

Sherman et al. Page 5

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

preferred to self-associate rather than mix with WTLAT, as they showed residual nanoscaleclustering by themselves under both stimulating (Figure S3B) and non-stimulating (FigureS3E) conditions. In contrast, LAT2CA molecules were randomly located at the PM with nosignificant self-association (Figure S3C), suggesting an essential role of the cysteineresidues, perhaps by their ability to become palmitoylated, in the nucleation of LATnanoclusters. These results together show that mutation of LAT at residues required foreither protein-protein interactions or protein-lipid interactions blocked full integration intointact nanoclusters.

LAT clusters of all sizes can recruit signaling molecules and can be phosphorylated uponTCR stimulation

LAT nanoclusters ranged in size from a few, predominantly, to hundreds of detectedmolecules. The question arose whether signaling occurs only in large LAT aggregates orwhether the entire distribution of LAT nanoclusters is capable of signal transduction uponTCR stimulation. To this end, we conducted two-color PALM imaging of LAT clusters andGrb2. Grb2 is a key adapter molecule that binds to any of the three LAT distalphosphotyrosines upon cell activation, and in turn recruits molecules such as the guaninenucleotide exchange factor SOS1, resulting in activation of the Ras pathway (Houtman etal., 2006) or the ubiquitin ligase Cbl, resulting in protein recycling and degradation(Balagopalan et al., 2007). In addition, the Grb2-SOS1 complex has been proposed as amediator of LAT clustering (Houtman et al., 2006). PALM images of LAT and Grb2showed no co-localization under non-stimulating conditions (Figure 5D,E). Grb2-Dronpawas barely seen at the PM under these conditions although it was well expressed in thecytosol, as observed by epifluorescence. However, following TCR stimulation and LATphosphorylation , LAT and Grb2 showed high colocalization that was homogeneous withLAT cluster size (Figure 5A,B). The extent of mixing between LAT and Grb2 averaged93±12% (Figure 5I). Because LAT cluster size distribution is dominated by very smallclusters (Figure 5C), we conclude that LAT phosphorylation and downstream signaling donot require the formation of large assemblies of LAT molecules.

PLC-γ1 is highly co-localized with LAT in isolated clustersPLC-γ1 is another signaling molecule that is recruited to phosphorylated LAT upon TCRactivation (Braiman et al., 2006; Sommers et al., 2002). Upon activation it cleavesphosphoinositides to products that regulate the elevation of intracellular Ca++ and activationof critical enzymes such as protein kinase C and RasGRP. As expected, PALM showed thatTCR stimulation led to the recruitment of PLC-γ1 to LAT (Figure 5F) while there was noco-localization of LAT and PLC-γ1 under non-stimulating conditions (Figure 5G). Closerexamination reveals that, in contrast to Grb2, PLC-γ1 was not distributed evenly with LAT,but was located in isolated clusters, within LAT-rich lamellae and microclusters (Figure 5F,inset).

LAT and PLC-γ1 maintained a high extent of co-localization, as revealed by the proximityof their bivariate correlation function to the model of random labeling (Figure 5H). Theextent of mixing between the two molecules averaged 87 ±10 %. Nevertheless, several cells(4 out of 8) demonstrated bivariate correlation curves above the 95% confidence levels ofthe random-labeling model (Figure 5I), suggesting that PLC-γ1 molecules can preferentiallybecome closer to LAT than expected by the model. On the other hand, the fact that somecells demonstrate an extent of mixing significantly below 100% indicates that occasionallyPLC-γ1 can be recruited to the PM at sites without LAT or that its position relative to LATis also controlled by other molecules.

Sherman et al. Page 6

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Nanoscale organization of SLP-76 is revealed at the rim of LAT clusters upon TCRstimulation

SLP-76 is a cytosolic adapter protein that is recruited to LAT phosphotyrosines via bindingof the adapter, Gads (Bunnell et al., 2006). By imaging Jurkat T cells expressing LAT-Dronpa and SLP-76-PAmCherry with two-color PALM, we found that SLP-76 wasefficiently recruited to the PM and to LAT nanoclusters only upon cell activation (Figure6A,B), but in a pattern markedly different from either Grb2 or PLC-γ1. The bivariatecorrelation function of LAT and SLP-76 strongly departed from the homogeneous mixingmodel at short distances below 160nm, and agreed with this model only at longer distances(Figure 6C). Taking a closer look at LAT clusters at this short length-scale, we found thatSLP-76 was enriched at the rims of LAT clusters (Figure 6D-F) with an intensity contrast(SLPout/SLPin)(LATin/LATout) of about 16-fold (Figure 6E). Thus, LAT-nucleated clustersexhibit nanoscale organization as defined by the distinctive localization of SLP-76.Importantly, a similar contrast was found by two-color PALM imaging of LAT and SLP-76with reversed colors, i.e. SLP-76-Dronpa and LAT-PAmCherry (data not shown). Theextent of mixing between the two molecules averaged at 33 ±7 % (Figure 6F), furthershowing partial molecular segregation and indicating that most of the SLP-76 molecules lieoutside of LAT microclusters. Finally, we used the clustering algorithm to demonstrate thatLAT clusters recruit SLP-76 with a linear efficiency as a function of LAT cluster size(Figure 6G). However, we could not determine whether SLP-76 molecules arrange at theperiphery of the predominant and smallest LAT clusters as the number of SLP-76 moleculesthat were associated with these clusters was too small. Two-color PALM imaging of SLP-76and TCRζ showed no mixing of these two molecules, thus demonstrating that the nanoscalestructure observed between LAT and SLP-76 is specific to them (Figure S4A). Additionally,by imaging either Grb2-Dronpa or SLP-76-Dronpa with LAT-2CA-PAmCherry, we foundthat LAT-2CA did not mix with these proteins (Figure S4B). Since LAT-2CA did notincorporate into intact LAT clusters (Figure 4F,G), this observation shows thatphosphorylation-dependent interactions of LAT require localization within intact clusters atthe PM.

The nanoscale organization of SLP-76 depends on actin polymerization and facilitatesSLP-76 cluster dynamics

SLP-76 links LAT clusters to the process of actin polymerization via interactions with theeffector proteins Nck, Vav-1 and Wasp (Barda-Saad et al., 2010; Bubeck Wardenburg et al.,1998). To study the possible relationship between the nanoscale organization of SLP-76 andLAT with actin, we disrupted SLP-76 and actin interactions in two different ways. First, wetreated Jurkat E6.1 cells with Latrunculin A, a drug that inhibits actin polymerization(Morton et al., 2000). Second, we mutated three tyrosines on SLP76, namely Y113, 128 and145, to phenylalanines, thus blocking both phosphorylation of these sites and binding ofNck, Vav-1 and the tyrosine kinase Itk (Bunnell et al., 2006). We observed that in bothexperiments, SLP-76 failed to organize at the periphery of LAT clusters (Figure 7A,B) andthe two molecules showed a very low extent of mixing (Figure 7C,D,E) although univariateanalysis showed that LAT and SLP-76 could still cluster individually at the PM (FigureS5A,B). Interestingly, treatment with Latrunculin significantly affected LAT clustering(Figure S5B) but not clustering of SLP-76 (Figure S5A). Thus, actin polymerization isrequired for establishment of the LAT-SLP-76 nanostructure.

Abrogating SLP-76 cluster dynamics, and in particular, blocking SLP-76 internalization ordecreasing its persistence in clusters has been correlated with impaired TCR-dependent cellfunction, including defects in TCR induced Ca++ flux, CD69 upregulation, and nuclearfactor of activated T cells (NF-AT) activation (Barr et al., 2006; Bunnell et al., 2006). Tostudy the functional role of the nanoscale organization of SLP-76 and LAT we tracked

Sherman et al. Page 7

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

SLP-76-3YF clusters at the PM of cells and compared their dynamics to the dynamics ofWTSLP-76 clusters by confocal microscopy. By imaging SLP-76-3YF-YFP together withimmunofluorescence staining for pSLP-76 on Y145, we also found that WTSLP-76 and theSLP-76-3YF mutant co-localized to the same clusters (Figure 7F). Strikingly, clusterscontaining the SLP-76-3YF mutant were less mobile and less persistent than clusters withonly WTSLP-76 clusters (Figure 7G,H and supplement movies M1 and M2). By univariateanalysis we found a small increase in clustering of SLP-76-3YF compared to WTSLP-76(Figure S5C), and an insignificant decrease in LAT clustering (Figure S5D). These dataindicate that the SLP-76-3YF mutant has a dominant negative effect and could abrogate boththe intact nanostructure of LAT and SLP-76 as well as the dynamics of the clusters in spiteof the presence of WTSLP-76 in these clusters. Thus, in combination with previous studies(Barr et al., 2006; Bunnell et al., 2006), our results strongly suggest that LAT-SLP-76nanostructure is required for intact TCR-dependent cellular function.

DiscussionUsing PALM imaging of Jurkat, and PBT, and live and fixed cells together with extensivespatial analyses, we observed a continuous size distribution of LAT in nanoclusters. Mostnanoclusters were comprised of a few molecules each and accounted for the majority ofLAT molecules at the PM, either within or outside microclusters. Thus, we set out to studythe formation mechanism, role in signaling, and organization of upstream and downstreammolecules of the newly identified LAT nanoclusters.

Two-color PALM imaging and bivariate correlation analyses enabled us to resolve complexnanoscale relationships between several different proteins. We found that LAT is clusteredwithin lamellae or membrane folds that first engage the activating surface of the coverslip.Indeed, previous scanning electron micrograph (SEM) images of the interface between Tcells and antigen presenting cells (APCs) have captured similar protrusions that mediateearly contact between the cells (Gomez et al., 2007). This complexity of contact morphologygives rise to apparent first order clustering, which has not been previously appreciated(Lillemeier et al., 2010; Lillemeier et al., 2006; Wilson et al., 2001). The clustering of LATand other proteins of interest discussed throughout this study is thus considered a second-order clustering mechanism observed at the sites of first cellular contact with the activatingsurface.

We further defined the role of critical LAT amino acids that control protein-proteininteractions and protein-lipid interactions in the formation of LAT nanoclusters. Recentstudies have emphasized the role of protein-protein interactions in LAT clustering leavingthe transport of LAT to the cell surface as the sole function of protein-lipid interactions(Hundt et al., 2009). We show here the complementary role of both protein-protein andprotein-lipid interactions in the sorting of LAT into intact LAT clusters at the PM. Asexpected, mutations that block cluster formation also block interactions of downstreamsignaling molecules. We propose that the full size distribution of LAT is dependent on intactnanocluster assembly since the mutations we studied also fail to form microclusters (Bunnellet al., 2006; Hundt et al., 2009). Finally, the self-association of LAT4YF mutants or ofTCRζ suggests the ability of different molecules to sort into separate nanoclusters in the PM.

Previous studies of TCR and LAT distribution using PALM and electron microscopy (EM)have suggested that TCR and LAT form domains ('protein islands') at the PM with largersizes than the smallest nanoclusters reported here (Lillemeier et al., 2010; Lillemeier et al.,2006; Wilson et al., 2001). This interpretation is not supported by our results. We notedifferences in how signal intensity was used to threshold and thus exclude molecules inthese other studies. The choice of statistical techniques also resulted in differences in scoring

Sherman et al. Page 8

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

signaling clusters. Additionally, we believe that consideration of the heterogeneity ofmembrane contact with the coverslip has consequences for the analysis. These importantdifferences are further elaborated in the Analyses part of the Methods section of the SI. Weconclude that, in several ways, the previous analyses made it difficult to recognize thedominance of very small clusters (namely dimers, trimers, etc.) over the cluster sizedistributions, while overestimating the importance of considerably larger clusters. Otherstudies (Purbhoo et al., 2010; Williamson et al., 2011) focus on the presence of LAT in largevesicular structures but we note that the timing of activation in these studies (10 min) is farlonger than in ours. We focussed on the basal state of molecules and their initial changeswithin seconds to a few minutes of activation, when phosphorylation of LAT and its bindingpartners peak.

Analysis of the PM distribution of the TCR using biochemical and EM techniques hasshown a range of size distribution down to the nanoscale much as we now see with LAT andas we confirm for the TCRζ chain under non-stimulating conditions (Schamel et al., 2005).The relationship of immunuoreceptor (TCR and FcRε) clusters to LAT clusters has beenstudied extensively using immuno-EM (Lillemeier et al., 2010; Lillemeier et al., 2006;Wilson et al., 2001). In contrast to our observation about the TCR and LAT, these studiesshowed that before cell activation LAT and the two immunoreceptors exist in segregateddomains that come together upon cell activation without mixing. However, multipleconcerns have been raised regarding the interpretation of immuno-EM results in molecularclustering (D'Amico and Skarmoutsou, 2008). These concerns include the possibility of falseclustering of nano-gold particles due to the attachment of multiple primary or secondaryantibodies to single target molecules, the steric hindrance by one type of nano-gold label onanother in experiments that use multiple immuno-gold labels, and alterations of moleculardistributions and PM structure due to procedures of sample preparation such as fixation withglutaraldehyde, sample drying and membrane ripping. PALM microscopy of intact cells hasbeen suggested as a way to overcome these caveats.

Using two-color PALM, we also observed separate pools of LAT and TCR. However, incontrast to previous studies, we identified mixed pools of LAT and TCR molecules prior toand then following cell activation. Importantly, these areas of overlap and mixture havefunctional significance, as they are likely to be the most efficient locations for thephosphorylation of LAT by activated ZAP-70. We further addressed this finding by imagingZAP-70 with either TCRζ or LAT. Indeed, TCRζ and ZAP-70 showed very high co-localization consistent with the binding of ZAP-70 to phosphorylated TCRζ. LAT andZAP-70, in contrast, were in spatially separated pools with partial mixing similar to thespatial separation of TCRζ and LAT. Thus we suggest that these areas in which LAToverlaps with TCRζ and activated ZAP-70 act as critical “hot spots” for LATphosphorylation by ZAP-70 bound to TCRζ or in near proximity to it. These sites may verywell change dynamically with time and might also serve as the location where proteinbinding to LAT occurs. They thus may be the sites of signal relay and amplification. Itshould be mentioned that in our imaging assay LAT can freely diffuse at the PM of T cellsand ZAP-70 can freely diffuse in the cytosol and bind or detach from TCRζ (Douglass andVale, 2005; Sloan-Lancaster et al., 1998). TCRζ could be potentially mobile with partialindependence from the rest of the TCR chains (Ono et al., 1995), although the CD3 chainsof the TCR molecules at the PM are immobilized on the αCD3-coated coverslip. Thus, ourfindings regarding the nanoscale organization of TCRζ, ZAP-70 and LAT relative to eachother directly relate to physiologic function.

We further showed that LAT nanoclusters were viable for signaling by demonstrating thatthe downstream adapter Grb2 mixed with LAT in a homogeneous manner. Since thedistribution of LAT is dominated by clusters as small as dimers and trimers, Grb2 must

Sherman et al. Page 9

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

associate with the very smallest of nanoclusters upon T cell activation. This conclusioncontrasts with previous studies that emphasized the role of large microcluster formation forefficient LAT signaling (Lillemeier et al., 2010). If LAT becomes activated at ‘hot spots’proximal to TCR clusters, one might wonder how activated LAT (marked here by Grb2recruitment) homogeneously distributes between LAT clusters. An obvious explanation isthat the high mobility of LAT, which has been shown at the PM of both non-activated andactivated T cells (Douglass and Vale, 2005) leads to the rapid exchange of LAT moleculesbetween clusters of all sizes. The enzyme PLC-γ1, whose activity downstream of LAT iscrucial to T cell activation, is also efficiently incorporated into LAT clusters. However,PLC-γ1 seemed to get recruited into isolated clusters without mixing with all of the LAT-rich domains at the PM. The reason for this preferential clustering might relate to thecooperative binding of PLC-γ1 to LAT with molecules such as SLP-76 and Gads (Braimanet al., 2006; Houtman et al., 2006); however, a detailed account of PLC-γ1 organizationawaits further study.

Surprisingly, using two-color PALM imaging we found a nanoscale organization of SLP-76at the rim of LAT microclusters. To our knowledge, this is the first observation of multi-molecular nanoscale organization in signaling complexes. Previous studies have suggested arole of the cytoskeleton in the organization of signaling molecules at the PM (Harwood andBatista, 2010a). Indeed, we further showed that the nanoscale organization of LAT andSLP-76 depended on polymerized actin, since the organization was disrupted byperturbations that affected the link between SLP-76 and actin. Finally, we showed thatabrogating nanoscale organization of LAT and SLP-76 correlated with impaired mobility ofSLP-76 microclusters at the cell surface. Based on these findings and previous work (Barr etal., 2006; Bunnell et al., 2006), we conclude that the nanoscale organization LAT andSLP-76 is required for intact TCR-dependent cell functions. We predict that other signalingsystems at the PM are organized in nanoclusters and expect that further super-resolutionstudies will lead to important insights into the structure and function of these criticalsignaling aggregates.

To conclude, using state-of-the art imaging techniques, we were able to show that LATresides in nanoclusters at the PM under non-stimulating conditions and observe a modestgrowth in cluster size upon cell stimulation. We also showed unexpected results regardingthe formation mechanism, functional role and structure of the newly identified LATnanoclusters. Some of these observations include: (i) the distribution of the TCR and LAT atthe PM demonstrating mixing in the basal state with a decrease in mixing following T cellactivation, (ii) the distribution of ZAP-70 compared to that of TCRζ and LAT suggests thatthe sites of partial mixing of the latter two molecules are where ZAP-70 phosphorylatesLAT, (iii) the complementary role of protein-protein and protein-lipid interactions in LATnanocluster formation, (iv) the ability of LAT nanoclusters, as small as dimers, to signalwithout large-scale aggregation, (v) the colocalization of PLC-γ1 with LAT in isolatedclusters, (vi) the localization of SLP-76 to the periphery of clusters indicating that thesenanoclusters have structure, and (vii) the dependence of this structure on actin and therelationship of LAT-SLP-76 nanostructure to cluster dynamics. Our observations show thecomplexity of organization of signaling complexes and the mechanisms that shape theirarrangement at the PM at the level of single molecules. These findings greatly extend ourunderstanding of signal transduction at the PM. In addition, these observations can serve asa model for many signaling processes that occur at the PM and other cellular compartments.

Sherman et al. Page 10

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

MethodsSample preparation

Fluorescent conjugates of proteins of interest and the photoactivatable proteins Dronpa orPAmCherry were created by the replacement of previously tagged constructs by digestionand ligation of new inserts. DNA was introduced into E6.1 Jurkat T cells or PBTs, isolatedfrom healthy donors (Barda-Saad et al., 2005), using the LONZA nucleofector shuttlesystem. Transiently transfected cells were sorted for positive expression of PAmCherry orDronpa chimeras and imaged within 24 hours (PBTs) or 48-72 hours (Jurkats) fromtransfection or used to establish stable cell lines (Jurkats). Sorting was performed by the NCIflow cytometry core facility.

The preparation of coverslips for imaging spread cells followed a previously describedtechnique (Bunnell et al., 2003) and as further described in the SI. Briefly, sorted cells weredropped on glass coverslips coated with 0.01% poly-L-lysine (Sigma) and either stimulatoryor non-stimulatory antibodies at a concentration of 10 µg/ml (unless specified otherwise).Cells were resuspended in imaging buffer and dropped onto the coated coverslips, incubatedat 37°C for the specific spreading time (typically 3 min) and fixed with 2.4% PFA for 30min at 37°C.

ImagingConfocal images were taken with a 510 LSCM confocal microscope using a 63X, 1.4 NAobjective (Zeiss). PALM imaging was performed on a previously described homemadeTIRF microscope (Betzig et al., 2006) and on a second system based on a TIRF microscope(Nikon). For live cell PALM imaging, the set-up was modified for the delivery of highexcitation power (x16 times the power flux used for fixed cell imaging) and fast readout (at800 frame/sec), as described in the SI. PALM images were analyzed using a previouslydescribed algorithm to identify peaks and group them into functions that reflect the positionsof single molecules (Betzig et al., 2006). The acquisition sequence typically consisted of5-10 min of PALM image acquisition in a single channel or 15-20 min in two channels, or5min at 800 frame/sec for live cell PALM imaging. Custom algorithms were then applied tofurther study the position map of the identified molecules, as detailed in the SI.

To image molecules with two-color PALM, we created chimeric constructs of moleculeswith Dronpa or PAmCherry. We made use of the ~50-fold difference in activation intensitiesbetween Dronpa and PAmCherry to image them sequentially. We first activated Dronpawith ~0.3mW of 405nm illumination. After the Dronpa population was completely bleached,we switched detection channels and activated PAmCherry with up to 10mW of 405nm light.Fiduciary markers (0.1 μm gold particles or TetraSpec microspheres; Invitrogen) and affinetransformations were used to register images taken in the two channels.

Further information regarding materials, methods and algorithms can be found in the SI.

Supplementary MaterialRefer to Web version on PubMed Central for supplementary material.

AcknowledgmentsThe authors would like to thank B. J. Taylor at the NCI Flow Cytometry Core Facility, Zeiss, H. Hess (HHMI,Janelia Farm) for providing the PALM software, W. Losert (University of Maryland) for multiple discussions ondata analyses and T. Wiegand (Helmholtz Centre for Environmental Research - UFZ) for providing us his point-pattern analyses software. We thank P. Sengupta, M. Renz and T. Jovanovic-Talisman (NIH/NICHD) for readingthe manuscript. This research was supported by the Intramural Research Programs of the National Cancer Institute

Sherman et al. Page 11

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

(The Center for Cancer Research) and of the Eunice Kennedy Shriver National Institute of Child Health and HumanDevelopment.

ReferencesAndo R, Mizuno H, Miyawaki A. Regulated fast nucleocytoplasmic shuttling observed by reversible

protein highlighting. Science. 2004; 306:1370–1373. [PubMed: 15550670]Balagopalan L, Barr VA, Sommers CL, Barda-Saad M, Goyal A, Isakowitz MS, Samelson LE. c-Cbl-

mediated regulation of LAT-nucleated signaling complexes. Molecular and Cellular Biology. 2007;27:8622–8636. [PubMed: 17938199]

Balagopalan L, Coussens NP, Sherman E, Samelson LE, Sommers CL. The LAT story: a tale ofcooperativity, coordination, and choreography. Cold Spring Harb Perspect Biol. 2010; 2:a005512.[PubMed: 20610546]

Barda-Saad M, Braiman A, Titerence R, Bunnell SC, Barr VA, Samelson LE. Dynamic molecularinteractions linking the T cell antigen receptor to the actin cytoskeleton. Nature Immunology. 2005;6:80–89. [PubMed: 15558067]

Barda-Saad M, Shirasu N, Pauker MH, Hassan N, Perl O, Balbo A, Yamaguchi H, Houtman JC,Appella E, Schuck P, et al. Cooperativ interactions at the SLP-76 complex are critical for actinpolymerization. EMBO J. 2010; 29:2315–2328. [PubMed: 20562827]

Barr VA, Balagopalan L, Barda-Saad M, Polishchuk R, Boukari H, Bunnell SC, Bernot KM, Toda Y,Nossal R, Samelson LE. T-cell antigen receptor-induced signaling complexes: Internalization via acholesterol-dependent endocytic pathway. Traffic. 2006; 7:1143–1162. [PubMed: 16919152]

Betzig E, Patterson GH, Sougrat R, Lindwasser OW, Olenych S, Bonifacino JS, Davidson MW,Lippincott-Schwartz J, Hess HF. Imaging intracellular fluorescent proteins at nanometer resolution.Science. 2006; 313:1642–1645. [PubMed: 16902090]

Braiman A, Barda-Saad M, Sommers CL, Samelson LE. Recruitment and activation of PLC gamma 1in T cells: a new insight into old domains. Embo Journal. 2006; 25:774–784. [PubMed: 16467851]

Bubeck Wardenburg J, Pappu R, Bu JY, Mayer B, Chernoff J, Straus D, Chan AC. Regulation of PAKactivation and the T cell cytoskeleton by the linker protein SLP-76. Immunity. 1998; 9:607–616.[PubMed: 9846482]

Bunnell SC, Barr VA, Fuller CL, Samelson LE. High-Resolution Multicolor Imaging of DynamicSignaling Complexes in T Cells Stimulated by Planar Substrates. Sci STKE. 2003:P18.

Bunnell SC, Hong DI, Kardon JR, Yamazaki T, McGlade CJ, Barr VA, Samelson LE. T cell receptorligation induces the formation of dynamically regulated signaling assemblies. Journal of CellBiology. 2002; 158:1263–1275. [PubMed: 12356870]

Bunnell SC, Singer AL, Hong DI, Jacque BH, Jordan MS, Seminario MC, Barr VA, Koretzky GA,Samelson LE. Persistence of cooperatively stabilized signaling clusters drives T-cell activation.Molecular and Cellular Biology. 2006; 26:7155–7166. [PubMed: 16980618]

Campi G, Varma R, Dustin ML. Actin and agonist MHC-peptide complex-dependent T cell receptormicroclusters as scaffolds for signaling. J Exp Med. 2005; 202:1031–1036. [PubMed: 16216891]

Cebecauer M, Spitaler M, Serge A, Magee AI. Signalling complexes and clusters: functionaladvantages and methodological hurdles. Journal of Cell Science. 2010; 123:309–320. [PubMed:20130139]

D'Amico F, Skarmoutsou E. Quantifying immunogold labelling in transmission electron microscopy. JMicrosc. 2008; 230:9–15. [PubMed: 18387034]

Douglass AD, Vale RD. Single-molecule microscopy reveals plasma membrane microdomains createdby protein-protein networks that exclude or trap signaling molecules in T cells. Cell. 2005;121:937–950. [PubMed: 15960980]

Gomez TS, Kumar K, Mecleiros RB, Shimizu Y, Leibson PJ, Billadeau DD. Formins regulate theactin-related protein 2/3 complex-independent polarization of the centrosome to theimmunological synapse. Immunity. 2007; 26:177–190. [PubMed: 17306570]

Harder T. Lipid raft domains and protein networks in T-cell receptor signal transduction. CurrentOpinion in Immunology. 2004; 16:353–359. [PubMed: 15134785]

Sherman et al. Page 12

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Harding A, Hancock JF. Ras nanoclusters: Combining digital and analog signaling. Cell Cycle. 2008;7:127–134. [PubMed: 18212529]

Harwood NE, Batista FD. The Cytoskeleton Coordinates the Early Events of B-cell Activation. ColdSpring Harb Perspect Biol. 2010a

Harwood NE, Batista FD. Early events in B cell activation. Annu Rev Immunol. 2010b; 28:185–210.[PubMed: 20192804]

Houtman JC, Yamaguchi H, Barda-Saad M, Braiman A, Bowden B, Appella E, Schuck P, SamelsonLE. Oligomerization of signaling complexes by the multipoint binding of GRB2 to both LAT andSOS1. Nat Struct Mol Biol. 2006; 13:798–805. [PubMed: 16906159]

Hundt M, Harada Y, De Giorgio L, Tanimura N, Zhang W, Altman A. Palmitoylation-dependentplasma membrane transport but lipid raft-independent signaling by linker for activation of T cells.J Immunol. 2009; 183:1685–1694. [PubMed: 19592663]

Lillemeier BF, Mortelmaier MA, Forstner MB, Huppa JB, Groves JT, Davis MM. TCR and Lat areexpressed on separate protein islands on T cell membranes and concatenate during activation. NatImmunol. 2010; 11:90–96. [PubMed: 20010844]

Lillemeier BF, Pfeiffer JR, Surviladze Z, Wilson BS, Davis MM. Plasma membrane-associatedproteins are clustered into islands attached to the cytoskeleton. Proc Natl Acad Sci U S A. 2006;103:18992–18997. [PubMed: 17146050]

Morton WM, Ayscough KR, McLaughlin PJ. Latrunculin alters the actin-monomer subunit interface toprevent polymerization. Nat Cell Biol. 2000; 2:376–378. [PubMed: 10854330]

Ono S, Ohno H, Saito T. Rapid Turnover of the Cd3-Zeta Chain Independent of the Tcr-Cd3 Complexin Normal T-Cells. Immunity. 1995; 2:639–644. [PubMed: 7796297]

Purbhoo MA, Liu HB, Oddos S, Owen DM, Neil MAA, Pageon SV, French PMW, Rudd CE, DavisDM. Dynamics of Subsynaptic Vesicles and Surface Microclusters at the Immunological Synapse.Science Signaling. 2010; 3

Schamel WWA, Arechaga I, Risueno RM, van Santen HM, Cabezas P, Risco C, Valpuesta JM,Alarcon B. Coexistence of multivalent and monovalent TCRs explains high sensitivity and widerange of response. J Exp Med. 2005; 202:493–503. [PubMed: 16087711]

Schlessinger J. Cell signaling by receptor tyrosine kinases. Cell. 2000; 103:211–225. [PubMed:11057895]

Sloan-Lancaster J, Presley J, Ellenberg J, Yamazaki T, Lippincott-Schwartz J, Samelson LE. ZAP-70association with T cell receptor zeta (TCR zeta): Fluorescence imaging of dynamic changes uponcellular stimulation. Journal of Cell Biology. 1998; 143:613–624. [PubMed: 9813084]

Sommers CL, Park CS, Lee J, Feng CG, Fuller CL, Grinberg A, Hildebrand JA, Lacana E, Menon RK,Shores EW, et al. A LAT mutation that inhibits T cell development yet induceslymphoproliferation. Science. 2002; 296:2040–2043. [PubMed: 12065840]

Subach FV, Patterson GH, Manley S, Gillette JM, Lippincott-Schwartz J, Verkhusha VV.Photoactivatable mCherry for high-resolution two-color fluorescence microscopy (vol 6, pg 153,2009). Nature Methods. 2009; 6:311–311.

Wang H, Kadlecek TA, Au-Yeung BB, Goodfellow HE, Hsu LY, Freedman TS, Weiss A. ZAP-70: anessential kinase in T-cell signaling. Cold Spring Harb Perspect Biol. 2010; 2:a002279. [PubMed:20452964]

Williamson DJ, Owen DM, Rossy J, Magenau A, Wehrmann M, Gooding JJ, Gaus K. Pre-existingclusters of the adaptor Lat do not participate in early T cell signaling events. Nature Immunology.2011; 12:655–662. [PubMed: 21642986]

Wilson BS, Pfeiffer JR, Surviladze Z, Gaudet EA, Oliver JM. High resolution mapping of mast cellmembranes reveals primary and secondary domains of Fc epsilon RI and LAT. Journal of CellBiology. 2001; 154:645–658. [PubMed: 11489921]

Wilson BS, Steinberg SL, Liederman K, Pfeiffer JR, Surviladze Z, Zhang J, Samelson LE, Yang LH,Kotula PG, Oliver JM. Markers for detergent-resistant lipid rafts occupy distinct and dynamicdomains in native membranes. Molecular Biology of the Cell. 2004; 15:2580–2592. [PubMed:15034144]

Yokosuka T, Sakata-Sogawa K, Kobayashi W, Hiroshima M, Hashimoto-Tane A, Tokunaga M,Dustin ML, Saito T. Newly generated T cell receptor microclusters initiate and sustain T cell

Sherman et al. Page 13

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

activation by recruitment of Zap70 and SLP-76. Nat Immunol. 2005; 6:1253–1262. [PubMed:16273097]

Zhang WG, Trible RP, Samelson LE. LAT palmitoylation: Its essential role in membranemicrodomain targeting and tyrosine phosphorylation during T cell activation. Immunity. 1998;9:239–246. [PubMed: 9729044]

Sherman et al. Page 14

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1. LAT clusters at the PM are mainly in small nanoclusters(A) Confocal images of Jurkat E6.1 T cells on αCD3- or αCD45-coated coverslips. Cellswere stained for pY (blue), pLAT(pY191, red) and stably expressed LAT-Dronpa (green).Bar - 10 µm. Insets show a magnified view of a microcluster.(B) PALM images of Jurkat T cells expressing LAT-PAmCherry spread on αCD3 or spreadon αCD45. Insets show nano-scale organization of individual molecules within apparentmicroclusters. Color codes (heat map, white is highest) for overlapping probability densityfunctions of individual molecules, with a maximal value of 410 molecules/µm2 on αCD3and 360 molecules/µm2 on αCD45. Bars – 0.5 µm and 1 µm, respectively. (C) Paircorrelation function (PCF) of LAT-PAmCherry molecules (n=7 cells for stimulating andnon-stimulating conditions; dashed lines indicate upper and lower 95% confidence levels ofa Heterogeneous Poisson process; see Analyses section in SI for further details). (D) Resultsof a clustering algorithm to resolve individual clusters and generate cumulative size-distribution curves of the LAT clusters seen under stimulating and non-stimulatingconditions. (n=12 cells for stimulating and 10 for non-stimulating conditions; dashed linesrepresent clustering of random sets; the two dashed lines overlap; see SI for further detailson statistical analyses). (E) PALM images of peripheral blood lymphocytes (PBLs)expressing LAT-PAmCherry spread on either αCD3+αCD28- or on αCD45-coatedcoverslips. Insets show nano-scale organization of individual molecules within apparentmicroclusters. Maximal density values are 740 molecules/µm2 and 490 molecules/µm2,respectively. Bars – 1 µm (left) and 2 µm (right). (F) The PCF of LAT-PAmCherrymolecules (n=8 cells for stimulating and non-stimulating conditions; dashed lines defined asin panel C). (G) The cluster analyses showing cumulative size-distribution curves. (n=12cells for stimulating and n=10 for non-stimulating conditions; dashed lines representclustering of random sets; the two dashed lines overlap; see SI for further details onstatistical analyses). (H) PALM image of a live Jurkat T cell expressing LAT-PAmCherryspread on αCD3- or on αCD45-coated surface. Maximal density value 50 molecules/µm2 forboth panels. Bars - 2 µm (I) PCF of LAT-PAmCherry molecules (n=3 cells for stimulatingand non-stimulating conditions; dashed lines defined as in panel C). (J) Clustering analyses(n=3 cells for stimulating and non-stimulating conditions). See Materials and Methodssection in SI and Figure S1K for further details regarding live cell imaging with PALM. (K)Peripheral blood T (PBT) cells expressing LAT-PAmCherry plated onto αCD43-, αCD18-or αCD28-coated coverslips. Bars - 2µm. (L) PCFs of LAT-PAmCherry molecules (dashedlines defined as in panel C).

Sherman et al. Page 15

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 2. Two-color PALM captures the mutual organization and interactions of signalingmolecules at the PM(A,B,C) LAT is recruited to early contact areas within lamellae. Two-color PALM imagesof Jurkat cells expressing LAT-Dronpa (green) and TAC-PAmCherry (red) understimulating conditions, showing (A) partial or (B) complete spreading, or (C) cells undernon-stimulating conditions showing partial spreading. Right panels show a zoomed image ofthe region of interest, a single color rendering of TAC-PAmcherry (labelled red) and LAT-dronpa (labelled green). Maximal probability density values (as detailed in Figure 1B) forLAT-Dronpa and TAC-PAmCherry – (A) 270 and 1340 molecules/µm2, respectively; (B)190 and 10 molecules/µm2; (C) 126 and 59 molecules/µm2. Bars – (A) 2µm, (B) 5µm, (C)0.5µm. (D) The PCF of LAT-Dronpa (red line) and TAC-PAmCherry (green line) understimulating conditions (n=5 cells; Dashed lines indicate upper and lower 95% confidencelevels of a Heterogeneous Poisson process). See Figure S2F,G for the related PCFs for non-stimulating conditions.(E,F,G,H) ZAP-70 is well co-localized with TCRζ in activated T cells. Two-color PALMimages of Jurkat T cells expressing TCRζ-Dronpa and ZAP-70-PAmCherry on (E) αCD3-and (F) αCD45-coated coverslips. Insets show zoomed images of individual TCRζ andZAP-70 clusters (bars – 200nm). (G) Bivariate correlation curves of TCRζ and ZAP-70 forαCD3 (red lines) and αCD45 (green lines). Dashed lines indicate upper and lower 95%confidence levels of a random labeling (homogeneous mixing) model (H) The extent ofmixing between TCRζ and ZAP-70 (as defined in the text) are compared between activatingand non-activating conditions for multiple cells expressing TCRζ and ZAP-70 (n=13 forαCD3, and n=5 for αCD45). Maximal probability density values (as detailed in Figure 1B)for TCRζ-Dronpa and ZAP-70-PAmCherry – (E) 1580 and 1630 molecules/µm2,respectively, (F) 1070 and 1610 molecules/µm2. Bars – 2µm.

Sherman et al. Page 16

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 3. LAT and TCRζ exist in overlapping pools, where nanoscale domains could function ashot spots of T cell activationTwo-color PALM images of a LAT-Dronpa- and TCRζ-PAmCherry-expressing Jurkat Tcell (A) on αCD3- or (B) αCD45-coated coverslips. Zoomed images (insets) of individualTCRζ and LAT clusters are shown (bars – 100nm). (C) Bivariate correlation curves forαCD3 (Dashed lines indicate upper and lower 95% confidence levels of a HeterogeneousPoisson process). (D) Bivariate correlation curve for non-activating conditions (Dashed linesdefined as in panel C). Maximal probability density values (as detailed in Figure 1B) forLAT-Dronpa and TCRζ-PAmCherry – (A) 1380 and 1120 molecules/µm2, respectively; (D)510 and 400 molecules/µm2. Bars – 2µm.(E,F,G,H) ZAP-70 is poorly co-localized with LAT. Two-color PALM images of Jurkat Tcells expressing LAT-Dronpa and ZAP-70-PAmCherry on (E) αCD3- or (F) αCD45-coatedcoverslips. Zoomed images (insets) of individual LAT and ZAP-70 clusters (bars – 200nm).(G) Bivariate correlation curves of LAT and ZAP-70 for αCD3 (red lines) and for αCD45(green lines). Dashed lines defined as in panel C. (H) The extent of mixing between LATand TCRζ or LAT and ZAP-70 are compared between activating and non-activatingconditions for multiple cells (LAT and TCRζ - n=19 for αCD3, and n=14 for αCD45; LATand ZAP-70 - n=13 for αCD3, and n=4 for αCD45). See definition of the mixing coefficientin the text. Maximal probability density values (as detailed in Figure 1B) for TCRζ-Dronpaand ZAP-70-PAmCherry – (A) 1580 and 1630 molecules/µm2, respectively, (B) 1070 and1610 molecules/µm2 (C) 1500 and 1260 molecules/µm2, (D) 1400 and 440 molecules/µm2.Bars – 2µm.

Sherman et al. Page 17

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 4. Protein-protein and protein-lipid interactions are both required for LAT intactnanocluster formation(A) A cartoon showing the position of LAT mutations and their effect on LAT interactions.The 4YF mutations prevent the binding of adapter proteins such as Grb2 and the 2CAmutations prevent LAT palmitoylation and recruitment to lipid rafts. Two-color PALMimages of activated Jurkat T cells expressing (B) WTLAT-Dronpa and WTLAT-PAmCherry and (C) their bivariate analyses or Jurkat cells expressing (D,E) WTLAT-Dronpa and 4YF-LAT-PAmCherry or (F,G) WTLAT-Dronpa and 2CA-LAT-PAmCherry.In each panel, the PALM images show two-color rendering of WTLAT-Dronpa (green),WT- or mutated-LAT-PAmCherry (red) and a zoomed image in the inset. The univariate

Sherman et al. Page 18

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

PCF curves of the cells are shown in Figure S3A-C. Dashed grey lines in panels C, E and Gindicate upper and lower 95% confidence levels of the random labeling model. See Analysessection in SI for further details. Color codes of PALM images (bright is highest) forindividual probability density functions of single molecules, with maximal values forWTLAT-Dronpa and mutant or WTLAT-PAmCherry of – (B) 100 and 100 molecules/µm2,respectively; (D) 180 and 120 molecules/µm2; (F) 310 and 470 molecules/µm2, respectively.Bars – 5µm

Sherman et al. Page 19

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 5. LAT clusters of all sizes can recruit Grb2 upon TCR stimulation, while PLC-γ1 is mostefficiently recruited in isolated clusters(A) PALM of Jurkat cells expressing Grb2-Dronpa and WTLAT-PAmCherry on αCD3-coated coverslips. (B) Bivariate PCFs indicate colocalization of Grb2 and LAT on αCD3-coated coverslips (in black; random labeling model in grey, upper and lower confidencelevels of 95%). (C) Cumulative probability function (n=7 cells, as detailed in Figure 1). (D)PALM of Jurkat cells expressing Grb2-Dronpa and WTLAT-PAmCherry on αCD45-coatedcoverslips. (E) Bivariate PCF between LAT and Grb2 on αCD45-coated coverslips (dashedgrey lines defined as in panel B). (F-I) PALM analysis of Jurkat cells expressing LAT-Dronpa and PLC-γ1-PAmCherry spread on (F) αCD3- or (G) αCD45-coated coverslips. (H)

Sherman et al. Page 20

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Bivariate PCFs indicate colocalization of PLC-γ1 and LAT on αCD3-coated coverslips (inblack; dashed grey lines defined as in panel B). (I) The extent of mixing between of LAT-PAmCherry and Grb2 Dronpa or PLC-γ1 and LAT (n=5 for LAT-PAmCherry and Grb2-Dronpa, and n=8 for LAT-Dronpa and PLCγ1-PAmCherry). Maximal probability densityvalues (as detailed in Figure 1B) for Grb2-Dronpa and LAT-PAmCherry – (A) 640 and 260molecules/µm2, respectively; (D) 540 and 860 molecules/µm2. Bars – (A) 0.5 µm, (D) 2 µm.Maximal probability density values (as detailed in Figure 1B) for LAT-Dronpa and PLC-γ1-PAmCherry – (F) 730 and 290 molecules/µm2, respectively; (G) 370 and 120 molecules/µm2. Bars – (F) 2 µm, (G) 5 µm.

Sherman et al. Page 21

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 6. Nano-scale organization of SLP-76 is revealed at the rim of LAT clusters upon TCRstimulationTwo-color PALM images of a LAT-Dronpa and SLP-76-PAmCherry expressing Jurkat Tcell (A) on αCD3- (B) αCD45-coated coverslips and (C) Bivariate correlation curve (dashedgrey lines indicate 95% confidence levels of a random labeling model). (D) Zoomed imagesof individual LAT clusters showing preferential organization of SLP-76 at the rims of LATclusters. Bars – 200nm. (E) Top - ratio of LAT and SLP-76 brightness in the rim vs. innerpart of the cluster, as identified by image processing (n=10 clusters). Bottom - density ratiosof LAT and SLP-76 as calculated by dividing the brightness measurements with thematching areas of the rim and inner part of the cluster (n=10 clusters). (F) The extent of

Sherman et al. Page 22

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

mixing between LAT and SLP-76 (n=10 cells). (G) SLP-76 molecules were associated withLAT clusters using the clustering algorithm (see Analyses section in SI and Figure S1M,Nfor further details). The number of SLP-76 molecules in LAT clusters is plotted as afunction of LAT cluster size (in copy number). A linear fit indicates linear efficiency ofrecruiting SLP-76 molecules to LAT clusters as a function of LAT cluster size. Maximalprobability density values (as detailed in Figure 1B) for LAT-Dronpa and SLP-76-PAmCherry – (A) 310 and 310 molecules/µm2, respectively; (D) 320 and 310 molecules/µm2. Bars – 2 µm.

Sherman et al. Page 23

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 7. Nano-scale organization of SLP-76 depends on actin and facilitates regulation ofSLP-76 clustersTwo-color PALM images of (A) a LAT-Dronpa and SLP-76-PAmCherry-expressing JurkatT cells treated with 300nM Latrunculin A and (B) a LAT-Dronpa and SLP-76-3YF-PAmCherry-expressing Jurkat T cells. (A,B) cells were imaged on αCD3-coated coverslips.(C,D) The bivariate correlation curves related to cells in panels A and B in cells treated with(C) Latrunculin A or (D) between LAT and the SLP76-3YF mutant (Dashed grey linesindicate 95% confidence level of a random labeling model). (E) The extent of mixingderived from imaging multiple cells as in panel A (n=8) and panel B (n=10) and cellsexpressing LAT-Dronpa and SLP-76-PAmCherry (n=5). (F,G,H) Jurkat T cells, transientlyexpressing SLP-76-YFP or SLP-76-3YF-YFP, were imaged by confocal microscopy as theyspread on αCD3-coated coverslips. (F) Confocal images of fixed cells expressing SLP-76-YFP or SLP-76-3YF-YFP and stained with αphosphoSLP-76(Y145). (G) Images are shownsumming all frames from Maximal intensity projection (MIP) movies (see related MoviesS1 and S2). These images describe SLP-76 cluster movement across the plasma membranefor SLP-76 or SLP-76-3YF. (H) Individual clusters containing SLP-76 or SLP-76-3YF wereidentified and tracked in the MIP movies using SlideBook (3i) as described in the methodssection. An average of the mean squared displacement (MSD) over time was calculated formultiple trajectories from multiple cells (SLP-76, n=2496 from 19 cells; SLP-76-3YF,n=233 from 10 cells as a significantly lower number of clusters could be detected for theSLP-76-3YF expressing cells). Maximal probability density values (as detailed in Figure1B) for LAT-Dronpa and SLP-76-PAmCherry – (A) 400 and 240 molecules/µm2,respectively; (B) 640 and 520 molecules/µm2. Bars – 2 µm.

Sherman et al. Page 24

Immunity. Author manuscript; available in PMC 2012 November 23.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript