ras/pi3kinase/cofilin-independent activation of human cd45ra+ and cd45ro+ t cells by superagonistic...

Ras/PI3Kinase/cofilin-independent activation of humanCD45RA+ and CD45RO+ T cells by superagonistic CD28stimulation

Urban Sester*,**, Guido H. Wabnitz*, Henning Kirchgessner andYvonne Samstag

Institute for Immunology, Ruprecht-Karls-University, Heidelberg, Germany

T cell activation requires costimulation of TCR/CD3 plus accessory receptors (e.g.CD28). A hallmark of costimulation is the dynamic reorganization of the actincytoskeleton, important for receptor polarization in the immunological synapse. Theclassical model of Tcell costimulation was challenged by the detection of superagonisticanti-CD28 antibodies. These induce T cell proliferation and – as demonstrated here –production of IFN-c, CD25 and CD69 even in the absence of TCR/CD3 coligation. Here,we analyzed whether superagonistic CD28 stimulation induces costimulatory signalingevents. Costimulation leads to phosphorylation of the actin-bundling protein L-plastinand dephosphorylation of the actin-reorganizing protein cofilin. Cofilin binds to F-actinonly in its dephosphorylated form. Binding of cofilin to F-actin leads to depolymeriza-tion or severing of F-actin. The latter ends up in smaller F-actin fragments, which can beelongated at the free barbed ends. This results in enhanced actin polymerization.Dephosphorylation of cofilin requires activation of Ras and PI3Kinase. Interestingly,superagonistic CD28 stimulation activates human peripheral blood T cells indepen-dently of Ras and PI3Kinase. Accordingly, it does not lead to cofilin dephosphorylationand receptor polarization. Likewise, L-plastin is not phosphorylated. Thus, super-agonistic CD28 stimulation does not mimic costimulation. Instead, it leads to a Ras/PI3Kinase/cofilin-independent state of “unpolarized T cell activation”.

Introduction

Adaptive T cell responses are believed to requirerecognition of (foreign) antigens that are presentedon antigen-presenting cells (APC) toward the T cell

receptor (TCR) (signal 1). In addition, second signalstransmitted through costimulatory receptors like CD28or CD2 are essential to induce expression of thefunctional T cell repertoire [1, 2]. One major functionof costimulation is the dynamic rearrangement of theactin cytoskeleton, which in turn is crucial for signaltransduction and cell polarization like formation of theimmunological synapse (IS). Although the molecularmechanisms that promote costimulation-induced actinreorganization are of fundamental interest, only little isknown about the molecular integrators that connectcostimulation with the actin cytoskeleton. We havepreviously demonstrated that the actin-remodeling

* The first two authors equally contributed to this work.

Correspondence: Yvonne Samstag, INF305, Institute for Im-munology, Ruprecht-Karls-University, Im NeuenheimerFeld 305, 69120 Heidelberg, GermanyFax: +49-6221-565549e-mail: [email protected]

Received 22/2/07Revised 9/7/07

Accepted 14/8/07

[DOI 10.1002/eji.200737206]

Key words:Antibodies

� Costimulation� Human � T cells

Abbreviations: BAPTA-AM: 1,2-bis(o-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid acetoxymethyl ester �EGFP: enhanced green fluorescent protein � IS: immunologicalsynapse � PBT: peripheral blood T lymphocytes � PP2: 4-amino-5-(4-chlorophenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine

Eur. J. Immunol. 2007. 37: 2881–2891 Molecular immunology 2881

f 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.eji-journal.eu

** Current address: Medical Department IV, Nephrology, Uni-versity of the Saarland, Homburg, Germany

protein cofilin, a central factor for dynamic reorganiza-tion of actin, is activated through costimulation [3–5]. Inquiescent Tcells, cofilin is constitutively phosphorylatedon Ser3. Costimulation leads to its dephosphorylation[4, 6], which is mediated via the small GTPase Ras andits downstream effector PI3Kinase [7]. Cofilin dephos-phorylation enables its binding to actin. The importanceof the cofilin/actin interaction for T cell activation wasdemonstrated by blocking the actin binding sites ofcofilin through cofilin-derived cell-permeable peptides(CPH). These peptides strongly interfere with cofilinaccumulation at the Tcell/APC interface and prevent thematuration of the IS as well as T cell activation [3, 5]. Inaddition, costimulation induces phosphorylation of theactin-bundling protein L-plastin on Ser5. Experimen-tally, T cells can also be activated with antibodiesdirected against the accessory receptors CD2 or CD28without the need of TCR/CD3 coligation [8, 9].Stimulation of the “alternative pathway” of T cellactivation via a combination of antibodies directedagainst different epitopes of the CD2 molecule is able tomimic costimulation with regard to cofilin dephos-phorylation on Ser3 and L-plastin phosphorylation onSer5 [10, 11]. Superagonistic T cell activation throughCD28 is induced by individual antibodies directedagainst the CD loop of the CD28 molecule [12]. Theyare called superagonistic anti-CD28 antibodies [9, 13]since they induce a strong T cell proliferation withoutthe need of TCR/CD3 coligation. Thus, experimentsperformed in the rat system revealed that TCR-associated tyrosine phosphorylation of TCRf, ZAP-70and LAT are not induced by superagonistic CD28stimulation [14]. However, this type of stimulationinduces activation of JNK [14], PLCc, calcium flux aswell as membrane recruitment and activation of PKCh[15]. The latter results in activation of the transcriptionfactor NF-jB. Interestingly, although ZAP-70 is notphosphorylated by superagonistic T cell activation, theexpression and tonic activity of ZAP-70 are crucial forsuperagonistic stimulation of Jurkat lymphoma orT hybridoma cells [16].

Up to now, the effects of superagonistic CD28stimulation on costimulatory signals are largely un-known. Here, we demonstrate that CD28 superagonisticstimulation of both primary human CD45RA+ andCD45RO+ T cells does not induce cofilin dephosphory-lation on Ser3 and L-plastin phosphorylation on Ser5.Moreover, the signaling pathways leading to cofilindephosphorylation following costimulation, i.e. activa-tion of Ras and PI3Kinase, are not activated throughsuperagonistic anti-CD28 antibodies. Hence, super-agonistic CD28 does not mimic classical costimulation.

Results

Stimulation of human T cells through anti-CD28superagonistic antibody does not mimiccostimulation

The aim of this study was to analyze whether super-agonistic CD28 stimulation of untransformed humanperipheral blood T cells (PBT) using 5.11A1 antibody[17] mimics costimulation via anti-CD3 plus conven-tional anti-CD28 antibody, referred to as CD28.2. Sincefunctional heterogeneities exist between naive(CD45RA+) and memory (CD45RO+) T cells withrespect to their dependence on costimulation [18–20],these subpopulations were analyzed separately.

As reported earlier [4], the actin-reorganizingprotein cofilin is constitutively phosphorylated inquiescent PBT and undergoes dephosphorylation uponcostimulation through anti-CD3 plus -CD28 antibodies(CD3 � CD28). Here, we compared this mode of T cellactivation with superagonistic CD28 stimulation. To thisend, cells were stimulated through the differentantibodies as described in Materials and methods andthe phosphorylation state of cofilin was visualized byWestern blotting using phospho-specific anti-cofilinantibodies (Fig. 1AI). For each experiment, the extentof cofilin (de)phosphorylation was quantified bydensitometry (Fig. 1AII). The calculated phosphoryla-tion indices (Ph-Index = p-cofilin/cofilin as assessed bydensitometry) were normalized to the respective valueof unstimulated, CD45RA+ T cells. These experimentsrevealed that stimulation via 5.11A1 alone – althoughinducing T cell activation (compare Fig. 2) – did notresult in cofilin dephosphorylation. In contrast, costi-mulation via CD3 in combination with either 5.11A1(CD3 � 5.11A1) or CD28.2 (CD3 � CD28.2) led todephosphorylation of cofilin.

To extend the characterization of the effects of thesuperagonistic 5.11A1 antibody on costimulation-re-lated signaling pathways, next the phosphorylation stateof L-plastin was analyzed. L-Plastin is an actin-bundlingprotein that is phosphorylated on Ser5 after T cellcostimulation but not after TCR/CD3 triggering alone[11]. Phosphorylated and unphosphorylated forms ofL-plastin were detected by 2D SDS-PAGE followed byWestern blotting. The ratios of phospho-L-plastin to totalL-plastinwere determined following densitometry of therespective protein spots. Like costimulation throughCD3 � CD28.2, CD3 � 5.11A1 led to an increase inL-plastin phosphorylation in both CD45RA+ andCD45RO+ T cells, while stimulation with anti-CD3antibody alone had no effect (Fig. 1B). Importantly, soletriggering with superagonistic 5.11A1 antibody did notinduce L-plastin phosphorylation.

Urban Sester et al. Eur. J. Immunol. 2007. 37: 2881–28912882


Superagonistic antibody 5.11A1 effectivelyinduces Bcl-xL expression and activation of T cells

Costimulation via CD3� CD28 augments the expressionof Bcl-xL, a member of the Bcl-2 family which protectsT cells from apoptosis [21]. Bcl-xL was also described tobe up-regulated after T cell stimulation with super-agonistic anti-CD28 antibodies [13]. To formallyexclude that the lack of cofilin Ser3 dephosphorylationand L-plastin Ser5 phosphorylation was due to ineffec-tive 5.11A1 antibody stimulation, the ability of thesuperagonistic antibody 5.11A1 to provoke Bcl-xLexpression was evaluated in our experimental setting.Western blot analysis after 24 h of stimulation with5.11A1 revealed a marked up-regulation of Bcl-xL(Fig. 1C). The levels of Bcl-xL expression were compar-able in CD45RA+ and CD45RO+ T cells, demonstratingthat both T cell subsets were equally activated. In linewith data obtained with pan T cells [13, 21], CD3stimulation alone led only to a marginal induction of

Bcl-xL in both CD45RA+ naive and CD45RO+ memoryT cells.

Next, we analyzed the activation of CD45RA+ andCD45RO+ T cells, i.e. the production of cytokines (IL-2and IFN-c), proliferation ([3H]thymidine uptake), andexpression of inducible surface receptors (CD25 andCD69) following 5.11A1 stimulation. To this end, T cellswere stimulated for 6 h by 5.11A1, CD3� 5.11A1, CD3�CD28.2 and the respective controls as described in Fig. 1.Subsequently, cells were stained for expression of theT cell activation molecules CD25 and CD69 (Fig. 2A–D)as well as for expression of IL-2 and IFN-c (Fig. 2E, F andI, J) and analyzed by flow cytometry. 5.11A1 stimulationled to an increase of CD69 and CD25 expression withinthe CD45RA+ and CD45RO+ T cell populations, whichwas comparable to the effects of CD3 � CD28.2stimulation (Fig. 2A–D). Note that superagonistic5.11A1 in combination with anti-CD3 antibodies (CD3� 5.11A1) induced the highest expression of CD25 andCD69 in both subpopulations. As expected, stimulation

Figure 1. Sole 5.11A1 stimulation does notmimic classical costimulation (CD3� CD28). (A) CD45RA+ naive (left part) and CD45RO+

memory T cells were stimulated for 30 min. The phosphorylation state of cofilin in whole cell lysates was analyzed via p-cofilin(Ser3) Western blotting (I). After stripping, the total amount of cofilin was determined with a pan cofilin antiserum (cofilin).(II) Densitometric evaluation of five independent experiments. To compare different donors and cell populations, the cofilinphosphorylation value of unstimulated CD45RA+ naive T cell was set as 1 (n = 5; * p <0.05; ** p <0.01). (B) Cells were stimulated asdescribed in (A). L-Plastin phosphorylation was determined by 2DWestern blotting. The phosphorylation state was quantified bydensitometry and L-plastin phosphorylation is depicted as quotient of phospho-L-plastin and total L-plastin (n = 3; * p <0.05;** p <0.01, compared to the respective unstimulated control). (C) Cells were stimulated via anti-CD3, conventional anti-CD28,superagonistic anti-CD28 or anti-CD3 antibodies in combinationwith either conventional or superagonistic anti-CD28 antibodies.After 24 h, cells were lysed and Bcl-xL expression was analyzed in 1DWestern blots. The figure is representative of three donors.Note that naive cells of only one donor did not up-regulate Bcl-xL upon costimulation via anti-CD3 in combination withconventional anti-CD28 antibody.



with conventional anti-CD28 antibody (CD28.2) alonedid not induce expression of any of these activationreceptors.

Comparable to CD3 � CD28.2 stimulation, IL-2production was significantly increased upon 5.11A1stimulation of CD45RO+ T cells (Fig. 2F). In CD45RA+

T cells, IL-2 production remained at a lower level, butwas still comparable with CD3 � CD28.2 stimulation(Fig. 2E). Importantly, stimulation via 5.11A1 incombination with TCR/CD3 stimulation resulted in4–5-fold higher amounts of IL-2-positive CD45RA+ orCD45RO+ Tcells as compared to classical CD3� CD28.2costimulation or 5.11A1 stimulation. Accordingly,costimulation through CD3 � CD5.11A1 was an evenstronger stimulus for cell proliferation than costimula-tion through CD3 � CD28.2 (Fig. 2G, H). In contrast toIL-2 production, in CD45RA+ Tcells, 5.11A1 stimulationresulted in a much weaker cell proliferation thancostimulation (Fig. 2G).

Similar effects were observed with regard to theproduction of the effector cytokine IFN-c in CD45RO+

T cells (Fig. 2J). In CD45RA+ T cells, sole CD3stimulation as well as costimulation via CD3 �CD28.2 and CD3 � 5.11A1 induced equal amounts ofIFN-c (Fig. 2I).

To sum up this part, stimulation via 5.11A1 alone issufficient to induce signaling events leading to Bcl-xLexpression and T cell activation, but it does not inducedephosphorylation of cofilin or phosphorylation ofL-plastin which are early signaling events observedafter classical costimulation. Only when the anti-CD28superagonistic 5.11A1 antibodies were used in combina-tion with TCR/CD3 triggering, these signaling eventswere induced. Surprisingly, when 5.11A1 was used ascostimulator together with anti-CD3 antibodies, T cellactivation was even more pronounced as compared toclassical costimulation through TCR/CD3 plus theconventional anti-CD28 antibody CD28.2.

Superagonistic CD28 stimulation is independentof Ras and PI3Kinase

Following conventional costimulation through CD3 plusCD28.2 of primary human (pan) T cells, cofilin isdephosphorylated via a Ras-PI3Kinase-dependent sig-naling pathway [7]. Since cofilin was not depho-sphorylated after sole 5.11A1 stimulation in CD45RA+

and CD45RO+ T cells (compare Fig. 1A), we askedwhether this might be due to a failure of activation of theRas-PI3Kinase pathway. To this end, the activity of thesmall GTPase Ras was analyzed through precipitation ofthe activated GTP-bound form of Ras from cell lysates(Fig. 3). Indeed, the activated form of Ras could only bedetected in costimulated T cells, i.e. CD3 � CD28.2 orCD3� 5.11A1 stimulation, and – to a lesser extent – afterCD3 triggering (Fig. 3I). In marked contrast, solesuperagonistic 5.11A1 triggering did not induce Rasactivation.

Similar results were achieved for the activation of theRas effector PI3Kinase, as assessed by analysis of thephosphorylation state of the PI3Kinase effector Akt.

Figure 2. Up-regulation of activation-induced receptors, pro-liferation and cytokine production by sole 5.11A1 stimulation.The up-regulation of CD69 (A, B), CD25 (C, D), production of thecytokines IL-2 (E, F) and IFN-c (I, J) were determined after 6 h ofstimulation. Cytokine secretion was blocked with brefeldin Aand cytokines were determined with an intracellular stainingprotocol. Shown are the means (� SEM) of percent positivecells of three independent experiments. To assess proliferation(G, H), T cells were stimulated for 48 h and subsequentlypulsed with 1 lCi [3H]thymidine for another 24 h. Proliferationindices were calculated by dividing mean values fromstimulated samples by mean values from the respectiveunstimulated controls (n = 4, mean � SEM).



Costimulation via both CD3 � CD28.2 and CD3 �5.11A1 provoked a phosphorylation of Akt, while onlyvery little amounts of phosphorylated Akt were observedafter 5.11A1 or CD28.2 stimulation alone (Fig. 3II, III).We therefore conclude that T cell activation via super-agonistic anti-CD28-antibodies occurs independently ofthe Ras-PI3Kinase-cofilin pathway.

To finally confirm this assumption by an independentexperimental approach, 5.11A1-mediated activation ofPBT was performed after introduction of a dominant-negative Ras mutant (H-Ras-N17; Fig. 4A). For theseexperiments, pan T cells were used since no differenceswere recognized between CD45RA+ and CD45RO+

T cells concerning Ras and PI3Kinase activation. Assurrogate marker for T cell activation the expression ofthe activation-induced receptor CD69 was determined.The dominant-negative form of Ras (H-Ras-N17), whichis able to inhibit PI3Kinase activation and cofilindephosphorylation in primary human T cells [7], wasused as an enhanced green fluorescent protein (EGFP)fusion protein. Expression of CD69 indeed remainedunaffected in 5.11A1-stimulated T cells expressing thedominant-negative H-Ras-N17 (Fig. 4A). In contrast and

as expected, in costimulated T cells expressing H-Ras-N17, up-regulation of CD69 was significantly lower thanin the corresponding EGFP-expressing control cells(Fig. 4A; CD3 � CD28, p <0.05). In line with theseresults, wortmannin (25 nM) and Ly294002 (12.5 lM),chemical reagents with different modes of PI3Kinaseinhibition, potently inhibited CD69 up-regulation uponCD3 � CD28.2 stimulation (Fig. 4B). However, noinhibitory effect on CD69 up-regulation could bedetected in 5.11A1-treated T cells.

Given that Ras and PI3Kinase were not involved, wenext asked which additional pathways are activated bysuperagonistic CD28 stimulation. It was tempting tospeculate that PKC, known to be triggered by super-agonistic T cell activation [15], may directly activate theRaf-1-MAPKinase pathway without the need of Ras[22]. To test this hypothesis, we pre-incubated panT cells with inhibitors for PKC (Ro31–8220), Raf-1(Raf-1-Inhibitor) and ERK1/2 (UO126) for 30 min priorto 5.11A1 stimulation. Fig. 4C demonstrates thatinhibiton of PKC, Raf-1 and Erk1/2 indeed led to astrong reduction in the activation of T cells after 5.11A1stimulation as detected by CD69 expression. Further-more, in line with data obtained in the rat system [15],inhibition of the calcium flux [1,2-bis(o-aminophenox-y)ethane-N,N,N',N'-tetraacetic acid acetoxymethyl ester(BAPTA-AM)] or Src kinases [4-amino-5-(4-chlorophe-nyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP2)] pre-vented the 5.11A1-induced CD69 expression in primaryhuman T cells.

In conclusion, activation of human T cells via thesuperagonistic anti-CD28 antibody 5.11A1 occurs in-dependently of activation of the Ras-PI3Kinase signalingpathway, whereas activation of Src kinases, PKC, Raf-1,and MAPKinases as well as calcium flux are cruciallyinvolved.

Incomplete receptor clustering after 5.11A1stimulation

One hallmark of costimulation is the coordination ofreceptor polarization through dynamic rearrangementsof the actin cytoskeleton. Cofilin, a key regulator of actindynamics [3], plays an important role for receptorpolarization [5]. Since cofilin was not dephosphorylatedby 5.11A1 stimulation (compare Fig. 1A), we analyzedwhether 5.11A1 is able to induce receptor polarization.To this end, CD45RA+ naive and CD45RO+ memoryT cells were incubated with 5.11A1 or a combination ofCD3 and CD28.2. Receptor clustering was induced byadding fluorescence-labeled anti-mouse IgG. Fig. 5demonstrates that, as expected, cross-linking of CD3and CD28 induced polarized large receptor clusters(caps) in CD45RA+ and CD45RO+ T cells (Fig. 5A). Incontrast, 5.11A1 cross-linking provoked only small

Figure 3. Ras and PI3Kinase are not activated by sole 5.11A1stimulation. (I) Activated Ras (Ras-GTP; upper panel) wasimmunoprecipitated from lysates of CD45RA+ naive T cells(left part) and CD45RO+ memory T cells (right part) anddetected as described in Materials and methods. (II) Thephosphorylation of the PI3Kinase effector Akt was assessedin parallel via 1D Western blots of whole cell lysates andquantified via densitometry (III) (n = 3, mean � SEM; * p <0.05,compared to the respective unstimulated control).



Figure 4. A1-mediated CD69 up-regulation is insensitive to dominant-negative Ras (H-Ras-N17) expression and PI3Kinaseinhibitors. (A) (I) Human PBTwere transfected with EGFP or EGFP-tagged dominant-negative H-RasN17. Subsequently, cells werecostimulated via CD3� CD28.2 or triggeredwith 5.11A1 for 6 h. After harvesting, cellswere surface-stained for CD69 and analyzedby flow cytometry. For CD69quantification, eventsweregated on EGFP-positive cells [note that all constructswere expressed at thesame level (data not shown)]. (II) Depicted is the CD69 expression (calculated from percent positive cells) of three independentexperiments (� SEM). The value for 5.11A1-stimulated EGFP-expressing cells was set as 100%. (B) (I) T cells were left untreated orpre-incubatedwith Ly294002 (12.5 lM) or wortmannin (25 nm) for 30 min. Subsequently, cellswere stimulatedwith 5.11A1 or CD3� CD28.2 for 6 h. Then, cells were harvested and CD69 expression was assessed by flow cytometry. (II) Evaluation of threeexperiments. The value obtained after 5.11A1 stimulation in the presence of DMSOwas set as 100% (n = 3, mean � SEM). (C) (I, II)T cells were left untreated or pre-incubatedwith the inhibitors PP2 (5 lM), BAPTA-AM (60 lM), Ro31–8220 (10 nM), Raf-1-Inhibitor(10 nM) or UO126 (20 lM) for 30 minutes. The CD69 expressionwas detected after 6 h of stimulationwith 5.11A1 as described in (B)(n = 3, mean � SEM). All presented compounds inhibited the CD69 expression significantly (p <0.01).



receptor clusters that failed to coalesce at one pole of thecell (Fig. 5B).

In conclusion, our data demonstrate that super-agonistic CD28 stimulation of untransformed humanT cells does not mimic classical costimulation withregard to the induction of signaling pathways control-ling the dynamic rearrangement of the actin cytoske-leton and receptor polarization.

Discussion

T cell costimulation via TCR/CD3 and accessoryreceptors is essential for the activation of restingTcells viaAPC. It promotes survival, cytokine productionand T cell proliferation. However, the development ofstimulating antibodies that specifically recognize acces-sory receptors, e.g. CD2 or CD28, has strikinglydemonstrated that T cell activation can be achievedvia alternative pathways without coligation of the TCR/CD3 complex [8, 9]. Although the signaling pathwaysunderlying T cell costimulation and CD2 stimulation(alternative pathway) have been explored extensively in

the last decade, only little is known about the molecularmechanisms underlying T cell activation via anti-CD28superagonistic antibodies (e.g. sole 5.11A1 stimulation).We demonstrate here that superagonistic anti-CD28antibody-mediated Tcell activation does notmimic Tcellcostimulation: It does not lead to dephosphorylation ofcofilin on Ser3 or phosphorylation of L-plastin on Ser5and functions via a Ras/PI3Kinase-independent signal-ing pathway.

Costimulation (e.g. CD3 � CD28 costimulation) ofuntransformed human T lymphocytes activates thesmall GTPase Ras [7]. This activation is mediated viaPKC and/or RasGRP [23]. A loss of Ras function resultsin a reduced antigen-dependent T cell activation [24,25]. Surprisingly, superagonistic CD28 stimulationalone induced T cell proliferation without prior activa-tion of Ras and PI3Kinase. Moreover, introduction of adominant-negative Ras mutant (H-Ras-N17) did notinterfere with superagonistic CD28 stimulation. Theseresults demonstrate that Ras activity is not indispensablefor T cell activation. The requirement for Ras activationrather depends on the mode of T cell stimulation.

PI3Kinase is believed to be essential for antigen-dependent T cell activation. The precise function ofPI3Kinase during costimulation is still unresolved.Uncontrolled PI3Kinase signaling contributes to auto-immunity [26]. In human T cells, the costimulatoryreceptor CD28 associates with PI3Kinase through theYXXMmotif [27–30]. It was recently demonstrated that,following costimulation, the binding of PI3Kinase toCD28 is essential for IL-2 gene transcription, but IL-2mRNA stability is mediated via a PI3Kinase-independentpathway [31]. Inhibition of PI3Kinase in freshly isolatedhuman T cells abrogates IL-2 production and prolifera-tion [32]. In contrast, PI3Kinase inhibitors have no effecton IL-2 production after costimulation of the humanT lymphoma cell line Jurkat [33] and they can evenenhance it after PMA/CD28 stimulation [34]. Thesediscrepancies can – at least in part – be explained by thenature of the different human T cell types used sinceJurkat T cells do not express the PIP3 phosphatases andnegative regulators of PI3Kinase PTEN and SHIP1[34–37]. In addition, we demonstrate here that both aRas/PI3Kinase-dependent mode of T cell activation viaCD3 � CD28 and a Ras/PI3Kinase-independent activa-tion via superagonistic anti-CD28 antibodies can beinduced within the same human cell type, namelyfreshly isolated naive and memory T cells. Thus,inhibition of PI3Kinase even enhanced the superago-nistic CD28-driven T cell activation, whereas costimula-tion-derived T cell activation was significantly inhibited.Given that superagonistic CD28 stimulation inducesT cell activation, IL-2 production and proliferationindependent of PI3Kinase, an alternative signalingpathway must exist that bypasses Ras and PI3Kinase

Figure 5. Superagonistic 5.11A1 stimulation induces receptormicroclusters. Receptor caps were induced by cross-linkingCD3 and CD28.2 (A) or 5.11A1 (B). The left panels show receptorstaining, the right panels an overlay of the receptor stainingand transmission images. The figure is representative of threeindependent experiments.



activation. This pathway likely also regulates glucosemetabolism, which is crucial for Tcell proliferation [38].The exact nature of this pathway remains to bedetermined. The PLCc-PKCh-NF-jB pathway and cal-cium flux [15] as well as PI3K-independent activation ofJNK [14], known to be induced following superagonisticCD28 activation of rat T cells, may represent importantcomponents of this pathway. Moreover, our data showthat, following superagonistic CD28 stimulation inhuman primary T cells, activation of the MAPKinasepathway [14] is independent of Ras. Instead, it might betriggered through direct activation of Raf-1 by PKC.Such a pathway was also suggested for leukotriene D4-dependent activation of intestinal epithelial cells [22].

One major function of Ras and PI3Kinase is thereorganization of the actin cytoskeleton. Dynamic actinpolymerization and depolymerization is crucial forantigen-dependent T cell activation (for review see[3]). Hence, interfering with actin integrity by cytocha-lasins leads to impaired T cell activation. After T cellcostimulation, this actin dynamics is at least in partmediated by cofilin [4, 5, 7], a key player for dynamicactin reorganization and cell polarization. As a con-sequence of the missing Ras and PI3Kinase activation inhuman CD45RA+ or CD45RO+ T cells, superagonisticCD28 stimulation does not result in cofilin depho-sphorylation, which is accompanied by insufficientreceptor polarization. Thus, receptor clusters that wereinduced with superagonistic anti-CD28 antibodies failedto coalesce and remained as microclusters.

Although 5.11A1 stimulation did not result inreceptor cap formation, the cells were fully activatedand produced IL-2. Whether later events like cytokinerelease occur in a polarized manner remains to bedetermined. Interestingly, in line with these results,PI3Kinase mutant mice expressing a dominant-negativeform of the catalytic PI3Kinase subunit p110d(p110dD910A/D910A) displayed a defect in lipid raftrecruitment towards CD3 � CD28-coated plastic beads.Yet these cells still showed normal proliferation and IL-2production after CD3 � CD28 stimulation.

In conclusion, our data show that superagonisticCD28 stimulation does not mimic classical costimula-tion. It rather induces a Ras/PI3Kinase/cofilin-indepen-dent state of “unpolarized T cell activation”.

Materials and methods

Antibodies and reagents

Rabbit antisera against cofilin and against bacterially ex-pressed recombinant human L-plastin were raised in ourlaboratory [10]. Antibodies against the following human cellsurface or intracellular antigens were used; (i) monoclonalantibodies: CD3 (SK7); CD45RA (HI100) and CD45RO

(UCHL-1), CD69 (FN50), CD25 (2A3), IL-2 (5344.111), IFN-c(25723.11) (all from BD Biosciences Heidelberg, Germany)and Akt (5G3) from Cell Signaling (Beverly, MA);(ii) polyclonal antibodies: rabbit anti-phospho-cofilin (Ser3)antibody (Cell Signaling) and pAkt (Ser473) from CellSignaling. Polyclonal secondary antibodies goat anti-mouseIgG (unconjugated or Cy-3 labeled) and peroxidase-conju-gated goat anti-rabbit IgG were from Dianova (Hamburg,Germany). The stimulatory anti-CD3e mouse antibody(BMA030, IgG2a) was a kind gift from R. Kurrle (Behring-werke, Marburg, Germany), conventional anti-CD28 mouseantibody (CD28.2, IgG1) was purchased from BD BiosciencesHeidelberg, and superagonistic anti-CD28 mouse antibody(5.11A1) was a kind gift from Thomas Hanke (TeGenero,ImmunoTherapeutics AG, W�rzburg, Germany). FITC-labeledphalloidin was from Sigma. PI3Kinase inhibitors used were:Ly294002 (Sigma) and wortmannin (Sigma). UO126 (Pro-mega), PP2, Ro31–8220 and Raf-1-Inhibitor were fromCalbiochem, BAPTA-AM from Molecular Probes.

Flow cytometry

Flow cytometry was performed with a FACSCalibur (BDBiosciences) and data were analyzed with the CellQuestProsoftware (BD Biosciences). For the detection of cytokines, thesecretion of the cytokines was stopped by adding brefeldin A(10 lg/mL) for the last 4 h of stimulation as described earlier[39]. For the staining, 5 � 105 PBT were incubated withfluorescence-labeled antibodies as described [5]. Briefly, cellswere incubated for 5 min with 4% paraformaldehyde at 37�C.Fixation was stopped by adding two volumes of FACS buffer(0.5% BSA, 5% FCS and 0.07% NaN3). Cells were then spundown. For intracellular staining, cells were permeabilized for10 min at room temperature with 0.1% saponin in FACS buffer.All following washing steps and antibody incubations weredone in the presence of 0.1% saponin. Surface staining wasperformed using the same protocol but in the absence ofsaponin.

Isolation of T cells

Human PBMC were obtained by Ficoll-Hypaque (Linaris,Wertheim-Bettingen, Germany) density gradient centrifuga-tion of heparinized blood from healthy volunteers. T cells werepurified with negative magnetic bead selection using the “PanT cell isolation Kit” (Miltenyi Biotech, Bergisch Gladbach,Germany) as described in the manufacturer's instructions.

CD45RA+ and CD45RO+ T cells were isolated by using the“Pan T cell isolation Kit” and CD45RA or CD45RO microbeads(Miltenyi Biotec). Labeling of cells and isolation wereperformed according to the manufacturer's instructions. Purityof isolated CD45RA+ or CD45RO+ T cells was generally above90%.

T cell stimulation

Classical costimulation (CD3� CD28.2) and CD3 stimulation

Wells of 96- (proliferation assay) or 24-well (lysate prepara-tion) plates (Nunc) were coated at 4�C overnight with 7.2 lg/



mL goat anti-mouse antibodies in PBS. After thorough washingwith PBS-1% BSA, surface receptor antibodies were allowed tobind during 2 h of incubation at RT as follows: IgG control –IgG2a (1 lg/mL) + IgG1 (10 lg/mL); CD3 – BMA030 (1 lg/mL) + IgG1 (10 lg/mL); CD28.2 – IgG2a (1 lg/mL) + CD28.2(10 lg/mL); CD3 � CD28.2 costimulation – BMA030 (1 lg/mL) + CD28.2 (10 lg/mL). Thereafter, plates were washedagain and T cells were added, i.e. for lysate preparation:2.5 � 106 cells/400 lL/well and for proliferation assays0.1 � 106 cells/100 lL/well. Stimulation was started bycentrifugation of the plates at 600 � g for 5 min andcontinued at 37�C and 5% CO2 for the indicated times.

Superagonistic CD28 stimulation and CD3 � 5.11A1costimulation

As described above for classical costimulation, wells of 96- or24-well plates were coated at 4�C overnight with goat anti-mouse antibodies, since stimulation with soluble superago-nistic anti-CD28 antibody without cross-linking was notsufficient to induce full T cell activation (data not shown).For costimulation with superagonistic anti-CD28 antibody,BMA030 (1 lg/mL) was added for 2 h at RT, while super-agonistic anti-CD28 (5.11A1, 0.2 lg/mL) antibody wastransferred into the respective wells together with the cellsas described earlier [12]. The optimal concentration of anti-CD28 antibody clone 5.11A1 was determined by proliferationassays (data not shown). All experiments with cells from oneindividual donor were done in parallel.

Proliferation assays

T cells were stimulated in 96-well plates in triplicates asdescribed above. To label synthesized DNA, 1 lCi [3H]thymi-dine was added at 48 h after stimulation. After a total time of72 h, incorporated [3H]thymidine was determined using aTomtec harvester (Perkin Elmer Wallac GmbH, Freiburg,Germany) and a Topcount counter (Canberra Packard GmbH,Schwadorf, Austria) according to the manufacturers' instruc-tions. Proliferation indices were calculated by dividing meanvalues from stimulated samples by mean values from therespective unstimulated controls.

Capping

Purified CD45RA+ and CD45RO+ T cells were adhered for15 min on adhesion slides (Marienfeld, Lauda-K�nigshofen,Germany). Adhesion wells were washed twice with RPMI 1640and blocked with RPMI 1640 + 10% FCS. Cells were incubatedat room temperature for 15 min with 10 lg/mL antibodies asindicated. Subsequently, unbound antibodies were removed bywashing with RPMI 1640, and receptor cross-linking wasinduced by 30 min of incubation with Cy-3-labeled anti-mouseIgG (1.5 lg/mL). Cells were fixed with 4% paraformaldehydefor 30 min. After two washing steps in PBS, slides were rinsedin ddH2O and mounted with MOWIOL 4-88 (Calbiochem,Schwalbach, Germany; in PBS/glycerol, pH 8.0). Digitizedimages were generated with a confocal laser-scanningmicroscope (Leica DMRBE with TCS NT; Bensheim, Germany).Images were stored separately for fluorescence and transmis-

sion light and then overlaid electronically using AdobePhotoshop 7.0.

Preparation of lysates and 1D Western-blotting

Coating of 24-well plates and cell stimulationwas performed asdescribed above. Then, plates were placed on ice to stopstimulation and all following steps were performed on ice.Cells were detached by thorough resuspension and transferredinto 1.5-mL tubes. After centrifugation, pellets were washedoncewith PBS and cells were lysed for 5 min in lysis buffer (2%SDS, 63 mM Tris pH 6.8, 50 mM dithiothreitol and 0.03%bromphenol blue) at 95�C. These SDS-lysates were either usedimmediately or stored at –70�C. All electrophoresis, blotting,staining and scanning procedures from one experiment wereperformed in parallel.

For the determination of cofilin dephosphorylation, 1DWestern-blotting was performed. Phosphorylation indices (Ph-Index) of cofilin were calculated from the optical densities(OD) of the band corresponding to phosphorylated cofilin(staining with phospho-cofilin Ser3-specific antibodies) di-vided by the OD of the band corresponding to total cofilinobtained after stripping [incubation for 30 min at 50�C instripping buffer (100 mM 2-mercaptoethanol, 2% SDS,62.5 mM Tris-Cl pH 6.7)] and restaining with a cofilin-specificantiserum.

2D Western-blotting

The phosphorylation state of L-plastin was analyzed by 2DWestern blotting. Briefly, lysates corresponding to 2.25 � 105

cells were diluted in 300 lL rehydration sample buffer [7 Murea, 2 M thiourea, 1% CHAPS, 65 mM DTT and 0.2% (w/v)Bio-Lytes (Bio-Rad)] and subjected to isoelectric focusing (Bio-Rad Protean IEF Cell) using immobilized pH gradient stripes(Bio-Rad ReadyStrip IPG Stripes, pH 4–7, 17 cm). Stripes wereoverlaid with mineral oil and rehydrated at 50 V for 12 h.Immediately after rehydration, protein separation was per-formed for 15 min at 250 V followed by a linear voltage slopeto 10 000 V for 5 h and a rapid voltage slope to 10 000 V for anadditional 6 h. Prior to SDS-PAGE, IEF stripes were equili-brated for 30 min in equilibration buffer (6 M urea, 2% SDS,20% glycerol, 0.375 M Tris-HCl pH 8.8) containing 75 mMDTT and an additional 30 min in equilibration buffersupplemented with 216 mM iodoacetamide and a traceamount of bromphenol blue. After separation by SDS-PAGE,proteins were blotted onto PVDF membranes. Staining wasperformed as described for 1D Western-blotting employing anL-plastin-specific antiserum. The relative amount of phos-phorylated L-plastin was calculated by the ratio of the OD ofthe phosphorylated spot and the sum of both the phosphory-lated and unphosphorylated spots.

Transfections

The pEGFP-C1-H-RasN17 expression vector was cloned in ourlaboratory [7]. Expression of EGFP or EGFP fusion proteinswas achieved through nucleofection of 5 � 106 PBTwith 2 lgof the respective plasmid according to the protocol supplied bythe manufacturer of the nucleofector (Amaxa). The transfec-



tion method led to about 50% EGFP-expressing cells withmean fluorescence intensities (MFI) between 200 and 300.

Determination of Ras activity

The Ras activation assay was performed as described [7] usinga commercially available system (Upstate). Briefly, cells(10 � 106/sample) were lysed. Cleared lysates were incubatedwith bead-bound Ras binding domain (RBD) protein toprecipitate Ras-GTP. The precipitates were subjected to SDS-PAGE and Western blotting. Staining of the blots with an anti-Ras antibody reveals the level of Ras activation in the lysate.

Software and statistics

Data management and calculations were performed usingMicrosoft Excel 2000, and graphs were generated usingGraphPad Prism 4.01. Values are expressed as mean �standard error of the mean (SEM). Significant differencesbetween two modes of stimulation were calculated using two-sided t-tests for paired observations.

Acknowledgements: The authors thank Stefan Meuerand Thomas H�nig for critical reading of the manu-script. This work was supported by grants from theDeutsche Forschungsgemeinschaft [SE1082/11 (to U.S.),DFG SA 393/3-3 and SFB405/A4 (to Y.S.)].

References

1 Mueller, D. L., Jenkins, M. K. and Schwartz, R. H., Clonal expansion versusfunctional clonal inactivation: A costimulatory signalling pathway deter-mines the outcome of T cell antigen receptor occupancy. Annu. Rev.Immunol. 1989. 7: 445–480.

2 Robey, E. and Allison, J. P., T-cell activation: Integration of signals from theantigen receptor and costimulatory molecules. Immunol. Today 1995. 16:306–310.

3 Samstag, Y., Eibert, S. M., Klemke, M. and Wabnitz, G. H., Actincytoskeletal dynamics in T lymphocyte activation and migration. J. Leukoc.Biol. 2003. 73: 30–48.

4 Lee, K. H., Meuer, S. C. and Samstag, Y., Cofilin: A missing link betweenT cell costimulation and rearrangement of the actin cytoskeleton. Eur. J.Immunol. 2000. 30: 892–899.

5 Eibert, S. M., Lee, K. H., Pipkorn, R., Sester, U., Wabnitz, G. H., Giese, T.,Meuer, S. C. and Samstag, Y., Cofilin peptide homologues interfere withimmunological synapse formation and T cell activation. Proc. Nat. Acad. Sci.USA 2004. 101: 1957–1962.

6 Samstag, Y., Henning, S. W., Bader, A. and Meuer, S. C., Depho-sphorylation of pp19: A common second signal for human T cell activationmediated through different accessory molecules. Int. Immunol. 1992. 4:1255–1262.

7 Wabnitz, G. H., Nebl, G., Klemke, M., Schroder, A. J. and Samstag, Y.,Phosphatidylinositol 3-kinase functions as a Ras effector in the signalingcascade that regulates dephosphorylation of the actin-remodeling proteincofilin after costimulation of untransformed human T lymphocytes. J.Immunol. 2006. 176: 1668–1674.

8 Meuer, S. C., Hussey, R. E., Fabbi, M., Fox, D., Acuto, O., Fitzgerald, K. A.,Hodgdon, J. C. et al., Alternative pathway of T cell activation: A functionalrole for the 50kD T11 sheep erythrocyte receptor protein. Cell 1984. 36:897–906.

9 Tacke, M., Hanke, G., Hanke, T. and Hunig, T., CD28-mediated inductionof proliferation in resting T cells in vitro and in vivo without engagement of

the T cell receptor: Evidence for functionally distinct forms of CD28. Eur. J.Immunol. 1997. 27: 239–247.

10 Samstag, Y., Eckerskorn, C., Wesselborg, S., Henning, S., Wallich, R. andMeuer, S. C., Costimulatory signals for human T cell activation inducenuclear translocation of pp19/cofilin. Proc. Nat. Acad. Sci. USA 1994. 91:4494–4498.

11 Henning, S. W., Meuer, S. C. and Samstag, Y., Serine phosphorylation of a67 kDa protein in human T lymphocytes represents an accessory receptormediated signalling event. J. Immunol. 1994. 152: 4808–4815.

12 Luhder, F., Huang, Y., Dennehy, K. M., Guntermann, C., Muller, I.,Winkler, E., Kerkau, T. et al., Topological requirements and signalingproperties of T cell-activating, anti-CD28 antibody superagonists. J. Exp.Med. 2003. 197: 955–966.

13 Kerstan, A. and Hunig, T., Cutting Edge: Distinct TCR- and CD28-derivedsignals regulate CD95L, Bcl-xL, and the survival of primary T cells. J.Immunol. 2004. 172: 1341–1345.

14 Bischof, A., Hara, T., Lin, C. H., Beyers, A. D. and Hunig, T., Autonomousinduction of proliferation, JNK and NF-alphaB activation in primary restingT cells by mobilized CD28. Eur. J. Immunol. 2000. 30: 876–882.

15 Dennehy, K. M., Kerstan, A., Bischof, A., Park, J. H., Na, S. Y. and Hunig,T., Mitogenic signals through CD28 activate the protein kinase Ctheta-NF-kappaB pathway in primary peripheral T cells. Int. Immunol. 2003. 15:655–663.

16 Dennehy, K. M., Elias, F., Na, S. Y., Fischer, K. D., Hunig, T. and Luhder,F.,Mitogenic CD28 signals require the exchange factor Vav1 to enhance TCRsignaling at the SLP-76-Vav-Itk signalosome. J. Immunol. 2007. 178:1363–1371.

17 Evans, E. J., Esnouf, R. M., Manso-Sancho, R., Gilbert, R. J., James, J. R.,Yu, C., Fennelly, J. A. et al., Crystal structure of a soluble CD28-Fabcomplex. Nat. Immunol. 2005. 6: 271–279.

18 Croft, M., Bradley, L. M. and Swain, S. L., Naive versusmemory CD4 Tcellresponse to antigen. Memory cells are less dependent on accessory cellcostimulation and can respond to many antigen-presenting cell typesincluding resting B cells. J. Immunol. 1994. 152: 2675–2685.

19 Fischer, H., Gjorloff, A., Hedlund, G., Hedman, H., Lundgren, E.,Kalland, T., Sjogren, H. O. and Dohlsten, M., Stimulation of human naiveand memory T helper cells with bacterial superantigen. Naive CD4+45RA+

T cells require a costimulatory signal mediated through the LFA-1/ICAM-1pathway. J. Immunol. 1992. 148: 1993–1998.

20 Perrin, P. J., Lovett-Racke, A., Phillips, S. M. and Racke, M. K.,Differential requirements of naive and memory T cells for CD28costimulation in autoimmune pathogenesis. Histol. Histopathol. 1999. 14:1269–1276.

21 Boise, L. H., Minn, A. J., Noel, P. J., June, C. H., Accavitti, M. A., Lindsten,T. and Thompson, C. B., CD28 costimulation can promote Tcell survival byenhancing the expression of Bcl-xL. Immunity 1995. 3: 87–98.

22 Paruchuri, S., Hallberg, B., Juhas, M., Larsson, C. and Sjolander, A.,Leukotriene D(4) activates MAPK through a Ras-independent but PKCepsi-lon-dependent pathway in intestinal epithelial cells. J. Cell Sci. 2002. 115:1883–1893.

23 Ebinu, J. O., Stang, S. L., Teixeira, C., Bottorff, D. A., Hooton, J.,Blumberg, P. M., Barry, M. et al., RasGRP links T-cell receptor signaling toRas. Blood 2000. 95: 3199–3203.

24 Genot, E. and Cantrell, D. A., Ras regulation and function in lymphocytes.Curr. Opin. Immunol. 2000. 12: 289–294.

25 Cantrell, D. A., GTPases and T cell activation. Immunol. Rev. 2003. 192:122–130.

26 Borlado, L. R., Redondo, C., Alvarez, B., Jimenez, C., Criado, L. M.,Flores, J., Marcos, M. A. et al., Increased phosphoinositide 3-kinase activityinduces a lymphoproliferative disorder and contributes to tumor generationin vivo. FASEB J. 2000. 14: 895–903.

27 Fruman, D. A., Meyers, R. E. and Cantley, L. C., Phosphoinositide kinases.Annu. Rev. Biochem. 1998. 67: 481–507.

28 Stein, P. H., Fraser, J. D. andWeiss, A., The cytoplasmic domain of CD28 isboth necessary and sufficient for costimulation of interleukin-2 secretionand association with phosphatidylinositol 3-kinase.Mol. Cell. Biol. 1994. 14:3392–3402.



29 Truitt, K. E., Hicks, C. M. and Imboden, J. B., Stimulation of CD28 triggersan association between CD28 and phosphatidylinositol 3-kinase in JurkatT cells. J. Exp. Med. 1994. 179: 1071–1076.

30 Prasad, K. V., Cai, Y. C., Raab, M., Duckworth, B., Cantley, L., Shoelson,S. E. and Rudd, C. E., T-cell antigen CD28 interacts with the lipid kinasephosphatidylinositol 3-kinase by a cytoplasmic Tyr(P)-Met-Xaa-Met motif.Proc. Natl. Acad. Sci. USA 1994. 91: 2834–2838.

31 Sanchez-Lockhart, M. and Miller, J., Engagement of CD28 outside of theimmunological synapse results in up-regulation of IL-2 mRNA stability butnot IL-2 transcription. J. Immunol. 2006. 176: 4778–4784.

32 Fruman, D. A. and Cantley, L. C., Phosphoinositide 3-kinase inimmunological systems. Semin. Immunol. 2002. 14: 7–18.

33 Ueda, Y., Levine, B. L., Huang, M. L., Freeman, G. J., Nadler, L. M., June,C. H. and Ward, S. G., Both CD28 ligands CD80 (B7-1) and CD86 (B7-2)activate phosphatidylinositol 3-kinase, and wortmannin reveals hetero-geneity in the regulation of T cell IL-2 secretion. Int. Immunol. 1995. 7:957–966.

34 Harada, Y., Tanabe, E., Watanabe, R., Weiss, B. D., Matsumoto, A., Ariga,H., Koiwai, O., Fukui, Y. et al., Novel role of phosphatidylinositol 3-kinasein CD28-mediated costimulation. J. Biol. Chem. 2001. 276: 9003–9008.

35 Shan, X., Czar, M. J., Bunnell, S. C., Liu, P., Liu, Y., Schwartzberg, P. L.and Wange, R. L., Deficiency of PTEN in Jurkat T cells causes constitutivelocalization of Itk to the plasma membrane and hyperresponsiveness to CD3stimulation. Mol. Cell. Biol. 2000. 20: 6945–6957.

36 Astoul, E., Edmunds, C., Cantrell, D. A. and Ward, S. G., PI 3-K and T-cellactivation: Limitations of T-leukemic cell lines as signaling models. TrendsImmunol. 2001. 22: 490–496.

37 Kane, L. P. andWeiss, A., The PI-3 kinase/Akt pathway and Tcell activation:Pleiotropic pathways downstream of PIP3. Immunol. Rev. 2003. 192: 7–20.

38 Frauwirth, K. A., Riley, J. L., Harris, M. H., Parry, R. V., Rathmell, J. C.,Plas, D. R., Elstrom, R. L. et al., The CD28 signaling pathway regulatesglucose metabolism. Immunity 2002. 16: 769–777.

39 Waldrop, S. L., Pitcher, C. J., Peterson, D. M., Maino, V. C. and Picker, L.J., Determination of antigen-specific memory/effector CD4+ T cell fre-quencies by flow cytometry: Evidence for a novel, antigen-specifichomeostatic mechanism in HIV-associated immunodeficiency. J. Clin. Invest.1997. 99: 1739–1750.



ras/pi3kinase/cofilin-independent activation of human cd45ra+ and cd45ro+ t cells by superagonistic...

Documents