An initial and rapid step of lytic granule secretion precedes microtubule organizing center polarization at the cytotoxic T lymphocyte/target cell synapse

Florie Bertranda,b,c,1, Sabina Müllera,b,c,1, Kyung-Ho Rohd,e, Camille Laurenta,b,c, Loïc Dupréa,b,c, and Salvatore Valituttia,b,c,2

aSection Dynamique Moléculaire des Interactions Lymphocytaires, Institut National de la Santé et de la Recherche Médicale, Unité Mixte de Recherche 1043, Centre de Physiopathologie de Toulouse Purpan, 31024 Toulouse, France; bUniversité Toulouse III Paul-Sabatier, 31062 Toulouse, France; cLaboratoire d’Excellence Toulouse Cancer, Toulouse, France; and dDepartment of Microbiology and Immunology, and eHoward Hughes Medical Institute, Stanford University School of Medicine, Stanford, CA 94305-5323

Edited by Mark M. Davis, Stanford University School of Medicine, Stanford, CA, and approved March 5, 2013 (received for review November 1, 2012) It is presently assumed that lethal hit delivery by cytotoxic T (MTOC) and the centrosome beneath the IS and by a microtubule lymphocytes (CTLs) is mechanistically linked to centrosome polari- cytoskeleton–directed movement of lytic granules toward the target zation toward target cells, leading to dedicated release of lytic cells. These observations suggest a mechanistic link between cen- granules within a confined secretory domain. Here we provide three trosome docking to the plasma membrane and lytic granule se- lines of evidence showing that this mechanism might not apply as cretion (11, 12). Such a mechanism is considered instrumental to a general paradigm for lethal hit delivery. First, in CTLs stimulated ensure confined secretion of lytic molecules within the synaptic with immobilized peptide–MHC complexes, lytic granules and mi- cleft, resulting in selective killing of cognate target cells (1). crotubule organizing center localization into synaptic areas are spa- Although parallel movement of lytic granules and MTOC to- tio-temporally dissociated, as detected by total internal reflection ward the IS has been thoroughly documented, whether a mech- fluorescence microscopy. Second, in many CTL/target cell conju- anistic link exists between MTOC polarization and lytic granule gates, lytic granule secretion precedes microtubule polarization secretion is still unknown. In particular, it is not clear whether and can be detected during the first minute after cell–cell contact. lytic granule secretion is initiated at the CTL/target cell contact IMMUNOLOGY Third, inhibition of microtubule organizing center and centrosome site immediately after a cell–cell encounter, independently of polarization impairs neither lytic granule release at the CTL synapse MTOC repolarization. nor killing efficiency. Our results broaden current views of CTL bi- In the present study performed with untransformed human ology by revealing an extremely rapid step of lytic granule secretion CTLs, we show that lytic granule enrichment at the synaptic area and by showing that microtubule organizing center polarization is and their secretion precede MTOC polarization in ∼50% of the fi dispensable for ef cient lethal hit delivery. analyzed CTLs, as detected by total internal reflection fluorescence microscopy (TIRFM) and time-lapse microscopy. In addition, se- ζ immunological synapse | T-cell antigen receptor | protein kinase C | lective inhibition of MTOC/centrosome polarization inhibited signal transduction neither lytic granule secretion at the IS nor cytotoxic activity. Taken together, our results show that CTLs can secrete lytic ytotoxic T lymphocytes (CTLs) are central actors of the granules at the synaptic area as early as 40–60 s after target cell Cadaptive immune response implicated in the elimination of encounter and before MTOC polarization. They also show that infected cells and tumor cells. On encounter with cognate target lytic granule secretion can occur in conditions in which MTOC cells, CTLs activate different cytotoxic mechanisms, leading to polarization is inhibited. These unexpectedly rapid and MTOC- target cell annihilation. Among those mechanisms, the most rapid independent secretory events provide a basis for the described and efficient pathway of CTL-mediated cytotoxicity is the release capacity of CTL to behave as multiple killers of different target of the pore-forming protein perforin together with granzymes and cells encountered simultaneously. other proteolytic enzymes (all stored in cytosolic granules, named lytic granules) at the CTL/target cell immunological synapse (IS) Results (1–3). In human CTL and natural killer (NK) cells, the perforin/ Monitoring MTOC and Lytic Granule Dynamics in Antigen-Stimulated granzyme pathway plays a central role in immune surveillance, as CTLs. To study the dynamics of lytic granules and of the MTOC indicated by the severe immunodeficiency phenotype of patients within the IS area, we first used TIRFM to restrict the spatial exhibiting genetic mutations of perforin or of molecules implicated window of observation within a few hundred nanometers proxi- in lytic granule docking and fusion to plasma membrane (4). mal to the IS. HLA-A2–restricted human CTLs specific for the Cytotoxicity is a highly dynamic phenomenon, based on the CMV pp65 peptide were loaded with LysoTracker-red and formation of rapid and multiple cellular contacts. CTL have been TubulinTracker-green to monitor in parallel lytic granules and shown to kill target cells displaying on their surface extremely low microtubule dynamics and were then seeded into microchambers antigen densities within a few minutes after initial contact (5–9). Moreover, CTL have been reported to kill multiple target cells either serially, by bouncing from one target to another, or simul- Author contributions: F.B., S.M., and S.V. designed research; F.B., S.M., K.-H.R., and C.L. taneously (9, 10). performed research; F.B., S.M., and S.V. analyzed data; and F.B., S.M., L.D., and S.V. wrote Despite detailed knowledge of the immunological function of the paper. CTLs and of the numerous molecular steps involved in the re- The authors declare no conflict of interest. lease of lytic granule content after T-cell antigen receptor (TCR) This article is a PNAS Direct Submission. engagement, we do not have a complete understanding of the Freely available online through the PNAS open access option. process of lytic granule secretion at the CTL/target cell IS. 1F.B. and S.M. contributed equally to this work. The established model of CTL-mediated cytotoxicity is based 2To whom correspondence should be addressed. E-mail: [email protected]. on the observations that lethal hit delivery to target cells is par- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. alleled by the repositioning of the microtubule-organizing center 1073/pnas.1218640110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1218640110 PNAS Early Edition | 1of6 Downloaded by guest on September 24, 2021 2+ coated with specific peptide-MHC complexes (pMHCs). In CTLs 45% of CTL/target cell conjugates, [Ca ]i increase in target cells stimulated by immobilized specific pMHCs, lytic granules were was detected simultaneously with CTL MTOC polarization (32 of 2+ detected in the TIRF plane during the entire time of acquisition (a 72 cells scored). Conversely, in 55% of the conjugates, [Ca ]i total of 59 CTLs were scored for 180–360 s). On the contrary, increase preceded CTL MTOC polarization at the synaptic area MTOC was detected much less frequently in the TIRF plane. (40 of 72 cells scored; Fig. 2B; Movies S11 and S12). Fig. 2C shows fi 2+ Indeed, depending on the cells analyzed, we identi ed different the comparison of the time required for [Ca ]i increase in target types of MTOC dynamics: (i) MTOC did not appear in the TIRF cells and for CTL MTOC polarization in typical CTL/target cell plane (18 cells, 30% of total, Fig. 1A; Movie S1); (ii) microtubules conjugates exhibiting the latter phenotype. This single cell analysis 2+ and lytic granules appeared almost simultaneously, yet MTOC was showed that in these CTL/target cell conjugates, [Ca ]i increase detected in the TIRF plane later than lytic granules and stayed in clearly preceded MTOC polarization of the conjugated CTLs. B 2+ that plane only very transiently (24 cells, 41% of total; Fig. 1 ; [Ca ]i increase in target cells could be detected as early as a few Movie S2); (iii) MTOC appeared at the beginning of the movie but tens of seconds after initial CTL/target cell contact (Fig. 2C). then disappeared from the TIRF plane, whereas lytic granules Taken together these results show that lytic granule dynamics stayed in that plane for all of the acquisition time (7 cells, 12% of and secretion in antigen-stimulated CTLs does not coincide with total; Fig. 1C; Movie S3); and (iv) microtubules with or without MTOC dynamics and that lytic granule secretion can be initiated a distinguishable MTOC structure were detected in the TIRF before MTOC polarization. plane at the beginning of the acquisition and disappeared during the acquisition (10 cells, 17% of total; Fig. 1D; Movie S4). In cells Lytic Granule Secretion by CTLs Can Occur in the Absence of Detectable in which MTOC was clearly detectable in the TIRF plane, al- MTOC Synaptic Recruitment. We next attempted to prevent CTL though lytic granules could be observed in close proximity of the MTOC polarization to extend the window of time whereby MTOC- MTOC, they were frequently spread in the entire TIRF plane and independent lytic granule dynamics could be observed. To do so, found up to 5 μm away from the MTOC. Control experiments we targeted protein kinase Cζ (PKCζ) function in CTLs. Indeed, showed that, under the above-described conditions, CTLs were we and others recently showed that the activation of this ancestral specifically and rapidly stimulated (SI Results; Fig. S1; Movies S5, polarizing enzyme at the IS is required for MTOC polarization in + S6, S7, and S8). Taken together, the above results indicate that, in CD4 helper T cells interacting with dendritic cells (14, 15). To antigen-stimulated CTLs, the dynamics of lytic granule entry into test whether this pathway is also implicated in MTOC polariza- the TIRF plane is uncoupled from MTOC dynamics. They thus tion in CTLs, we monitored PKCζ activation at the IS by staining raise the question of whether lytic granules secretion might be CTL/target cell conjugates with antibodies directed against the initiated independently of MTOC polarization and contact to the phosphorylated form of PKCζ (p-PKCζ) (14). As shown in Fig. + + plasma membrane. S2, both in CD8 Vβ2 human primary T cells interacting with To address this question, we took advantage of the observation Epstein-Barr virus (EBV)-transformed B cells pulsed with toxic 2+ A B that lethal hit delivery can be visualized by a rapid [Ca ]i increase shock syndrome toxin-1 (TSST-1) (Fig. S2 and ) and in HLA- 2+ – fi in target cells. This [Ca ]i increase cannot be used to predict the A2 restricted human CTLs speci c for the CMV pp65 peptide time of target cell death, yet it can be used as an early marker of (Fig. S2C), p-PKCζ staining was increased at the IS where it lytic granule reception (9, 10, 13). We thus performed time-lapse paralleled phospho-tyrosine (p-Tyr) staining. These results in- confocal laser scanning microscopy to monitor in parallel micro- dicated that in CTLs, similarly to TH cells (14), the PKCζ pathway tubule dynamics, lytic granule delivery to the IS, and lethal hit is active at the IS. We thus treated CTLs with a selective pseu- reception by target cells. This approach showed that lethal hit dosubstrate inhibitor of PKCζ (PKCζ-PS) (16) to investigate the reception by target cells could precede a detectable polarization of impact of PKCζ signaling blockade on MTOC polarization. the MTOC (11 of 19 cells analyzed; Fig. 2A; Movies S9 and S10). PKCζ-PS treatment did not affect CTL viability (Fig. S3A), TCR Although this approach allowed us to monitor lytic granule down-regulation (Fig. S3B), or IFN-γ production (Fig. S3C), 2+ ζ and MTOC dynamics together with [Ca ]i increase, it had the indicating that inhibition of PKC function did not affect the limitation of a relatively slow image acquisition setup (1 image stability of CTL/target cell interaction or productive TCR en- every 7 s). It was therefore not possible to exclude the possibility gagement. Conversely, PKCζ-PS treatment of CTL inhibited that CTL MTOC might move back and forth to the IS with MTOC polarization toward target cells both in TSST-1–specific a rapidity not detectable with the time resolution used. and in pp65-specific CTL as depicted in Fig. 3A and Fig. S4A, To limit this possibility we used a single color, fast (one image respectively. Scoring of a significant number of individual con- per 0.5 s) but low-resolution image acquisition using confocal laser jugates showed that, in both TSST-1– and antigen-stimulated scanning microscopy. Using this approach, we observed that in CTLs, the distance between the MTOC and the center of the IS

Fig. 1. TIRFM analysis shows dissociation between MTOC and lytic granule dynamics. (A–D) Sequence of snapshots of TIRFM depicting the interaction between pp65-specific CTL and pMHC-coated sur- faces. CTLs were loaded with LysoTracker-red and TubulinTracker-green. Results are representative of 59 cells analyzed over four independent experi- ments. (A) Snapshots are from Movie S1.(B) Snap- shots are from Movie S2.(C) Snapshots are from Movie S3.(D) Snapshots are from Movie S4.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1218640110 Bertrand et al. Downloaded by guest on September 24, 2021 Fig. 2. Spatiotemporal dissociation of lytic granule secretion and MTOC polarization during CTL/target cell interaction. (A) Sequence of snapshots depicting the interaction between pp65-specific CTL and tar- get cells pulsed with antigenic peptide. CTLs were labeled with LysoTracker-red and TubulinTracker- green. Target cells were loaded with Fluo-4 AM to 2+ detect [Ca ]i increase as a marker of lethal hit de- livery. Image acquisition (one image per 7 s) was performed using a LSM 510 confocal microscope. Snapshots are from Movie S9. Results are from one representative conjugate of 11 (57% of a total 19 cells analyzed). (B) Sequence of snapshots depicting the interaction between TubulinTracker-green loaded pp65-specific CTLs and target cells pulsed with anti- genic peptide and loaded with Fluo-4 AM. Image acquisition (one image per 0.5 s) was performed using a LSM 710 confocal microscope. Snapshots are from Movie S11. Results are from one representative conjugate of 40 (55% of a total 72 cells analyzed). (C) Comparison of the time required for lethal hit reception by target cells and CTL MTOC polarization in individual CTL/target cells conjugates. Statistical significance of difference between groups was evaluated by a paired Student’s t test using GraphPad Prism software. ***P < 0.001.

was significantly increased when PKCζ was inhibited, although (Fig. 4A). Confocal microscopy was then used to define whether MTOC position was not reverted to that of unstimulated CTLs polarization-independent CD107a exposure might occur within the + (Fig. 3B; Fig. S4B). Comparable results were obtained when CTL confined CTL/target cell contact site. In untreated Vβ2 CTLs polarization was investigated by monitoring γ-tubulin (a micro- interacting for 3 min with cognate target cells, CD107a was exposed tubule nucleator used to detect centrosome position) together on the CTL surface in parallel with MTOC polarization. Moreover,

with α-tubulin (to detect microtubule organization) in CTL/target in CTL/target cell conjugates, CD107a exposure was preferentially IMMUNOLOGY cell conjugates using four-color confocal microscopy (Fig. S5). detected at the synaptic area after such short time of interaction Having observed that treatment with PKCζ-PS strongly reduced (Fig. 4B). Interestingly, in conditions in which PKCζ-dependent MTOC polarization in CTL without affecting productive TCR MTOC polarization was blocked, a similar exposure of CD107a engagement, we used this selective PKC-ζ inhibitor to investigate was found in the CTL/target cell conjugates (Fig. 4 B and C). whether lytic granule secretion might take place at the CTL/target The inhibition of MTOC polarization was confirmed by mea- cell IS in conditions in which MTOC polarization was inhibited. suring, in the same conjugates, the distance between MTOC and We first measured by FACS analysis CD107a exposure on the center of the IS (Fig. S6). These results indicate that CD107a CTLs (either untreated or pretreated with PKCζ-PS) interact- synaptic exposure occurs independently of MTOC polarization. ing with target cells. This analysis showed that the surface up- Three-dimensional confocal microscopy analysis showed that regulation of this lysosomal/lytic granule-associated membrane lytic granule release could still occur within a central region of the protein was not affected by the inhibition of PKCζ function IS, defined by the exclusion of CD45 (17), in conditions in which

+ Fig. 3. Inhibition of PKCζ function affects CTL polarization toward target cells. (A) Target cells unpulsed or TSST-1–pulsed were conjugated with Vβ2 CTLs pretreated or not with PKCζ-PS. After 5 min at 37 °C, cells were stained for α-tubulin (green) and perforin (blue). (B) Measurement of distances between MTOC and the center of the CTL/target cell contact site; 82 unpulsed, 85 pulsed, and 71 pulsed conjugates with PKCζ-PS–pretreated CTLs from from three in- dependent experiments (performed with cells form three different donors) were scored. Bars indicate mean values. Statistical significance of difference between groups was evaluated by an unpaired Student’s t test using GraphPad Prism software. ***P < 0.001.

Bertrand et al. PNAS Early Edition | 3of6 Downloaded by guest on September 24, 2021 + Fig. 4. Synaptic exposure of CD107a can occur in the absence of MTOC polarization. (A)Vβ2 CTLs pretreated or not with PKCζ-PS were conjugated with unpulsed or TSST-1–pulsed target cells. Surface expression of CD107a on CTLs was measured by FACS analysis at the indicated times. Data are from one representative experiment of three independent experiments performed in duplicate. (B)TSST-1–pulsed target cells were conjugated with Vβ2+ CTLs pretreated or not with PKCζ-PS. After 3 min, cells were fixed and stained for CD107a. Cells were then permeabilized and stained for α-tubulin. (C)Quantification of the intensity of CD107a staining per micrometer at the IS and at the distal-IS; 90 untreated conjugates and 83 conjugates in which CTLs were pretreated with PKCζ-PS were scored from three different donors. Statistical significance of difference between groups was evaluated by an unpaired Student’s t test using GraphPad Prism software. ***P < 0.001; nsP > 0.05.

MTOC polarization to the synaptic area was prevented by the We cannot exclude the possibility that, during a standard 4-h blockade of PKCζ activity (SI Results; Fig. S7, Movies S13, S14, cytotoxicity assay, a transient polarization of the MTOC toward S15,andS16). the IS might occur in PKCζ-inhibited CTLs. However, the obser- These results were supported by time-lapse confocal laser vation that in different CTL models using different E:T ratios and 2+ scanning microscopy showing a [Ca ]i increase in a significant antigen doses, PKCζ-PS pretreatment affected neither cytotoxicity fraction of target cells interacting with PKCζ-PS–treated CTLs nor caspase-3 activation kinetics strongly suggests that microtu- (32 of 66 cells analyzed; Fig. S8; Movies S17 and S18). bule organizing center polarization is dispensable for efficient le- These results extend our observation that lytic granule release thal hit delivery. can precede MTOC polarization by showing that preventing MTOC polarization does not abort lytic granule exocytosis. Discussion In the present work, we investigated the mechanistic link be- CTLs Elicit Cytotoxicity in Conditions in Which MTOC Polarization Is tween MTOC/centrosome polarization at the IS and lethal hit Inhibited. The above results indicated that secretion of lytic gran- ules could be mechanistically dissociated from MTOC apposition to the IS. We therefore asked whether CTLs treated with PKCζ-PS might elicit cytotoxicity. For this analysis, we took advantage of the observation that in helper T-cell/antigen-presenting cell (APC) conjugates, treatment with PKCζ-PS resulted in an inhibition of the secretory machinery toward the APC that lasted at least 6 h, a time longer than a standard cytotoxicity assay (14). In TSST-1– + stimulated Vβ2 CTLs, inhibition of MTOC polarization did not affect cytotoxicity at the different effector:target (E:T) ratios (Fig. 5A) and TSST-1 concentrations (Fig. S9A) used. These results suggested that efficient lethal hit delivery can occur in conditions in which MTOC positioning beneath the IS is strongly reduced. Comparable results were observed when cytotoxicity was mea- sured in antigen-stimulated CTLs at different E:T ratios (Fig. 5B) and peptide concentrations (Fig. S9B). We also investigated Fig. 5. CTL polarization responses are dispensable for cytotoxicity. (A and B) whether the time needed from initial CTL/target cell contact to Target cells either unpulsed or TSST-1–pulsed (A) or unpulsed and peptide- ζ pulsed (B) were cocultured for 4 h with Vβ2+ CTLs (A) or pp65-specific CTLs target cell death was altered when PKC activity was blocked by ζ measuring the time kinetics of caspase-3 activation in target cells at (B) pretreated or not with PKC -PS at the indicated effector/target ratios. Cytotoxicity was evaluated by flow cytometry using 7-AAD uptake by target different times after conjugation with CTLs. This analysis showed ζ cells. Data are from three independent experiments performed in duplicate that inhibition of PKC function did not affect the time kinetics of and are represented as mean ± SEM. Statistical significance of difference caspase-3 activation in target cells, indicating that the time needed between groups was evaluated by an unpaired Student’s t test using for induction of target cell death was not altered (Fig. S10). GraphPad Prism software. nsP > 0.05.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1218640110 Bertrand et al. Downloaded by guest on September 24, 2021 2+ delivery by CTLs. We show that lytic granule release at the CTL/ results showing that synaptic expression of CD107a and [Ca ]i in target cell synapse can precede MTOC polarization and can target cells was not inhibited in conditions in which MTOC po- occur in conditions in which MTOC polarization is impaired. larization was impaired. Moreover, results presented in Fig. S7 An initial observation indicating that lytic granule dynamics and Movies S13, S14, S15, and S16 unequivocally showed that lytic might be, at least in part, dissociated from MTOC dynamics came granules could route toward the center of the synaptic area in from the analysis of the kinetics of lytic granule and MTOC entry conditions in which MTOC polarization was inhibited. in the TIRF plane in CTL interacting with immobilized specific It should be noted that CTLs do not need synaptic contacts and pMHCs. This analysis showed that lytic granules and MTOC dy- polarization of MTOC/centrosome to secrete their lytic granules. namics were heterogeneous and clearly not synchronized in the Lytic granule secretion happens, for instance, when CTLs are various cells analyzed. The dynamics of lytic granules and MTOC triggered by soluble stimuli such as phorbol-12-myristate-13-acetate was recently studied using TIRFM in human CTL (18) and mouse and ionomycin (22). Our results are in line with these observations NK cells (19) stimulated by immobilized anti-CD3 or anti-NKp30 and provide further support to the notion that docking and fusion of and anti-CD18, respectively. In human CTLs, it was found that lytic granules to the CTL plasma membrane can occur indepen- the MTOC did not enter in the TIRF plane (18); conversely, in dently of microtubule polarization events. mouse NK cells, it was found that MTOC appeared in the TIRF Our results raise the question of what could be the in vivo impact plane and lytic granules mostly associated with MTOC (19). In of the above-described dynamics of CTL secretory mechanisms. It line with the latter observation, superresolution microscopy is conceivable that although CTLs might indifferently use the rapid analysis of the NK secretory domain showed that MTOC polarizes (MTOC-preceding) and sustained (MTOC-accompanied) path- toward the granule penetrable synaptic areas of the actin mesh- ways of lytic granule secretion to annihilate sensitive target cells, work (20). the sustained delivery of lytic granules to the IS might be required Our results reconcile these previous observations by identifying to optimize killing of resistant target cells such as tumor cells (13). different phenotypes for lytic granules and MTOC dynamics. If the role of the MTOC is not directly involved in the control of Using rapid time-lapse microscopy, we also show that, during lytic granule release, it is tempting to speculate that the numerous initial CTL/target cell interaction, CTLs can secrete cytotoxic lytic granules that follow the MTOC constitute a reserve pool that granules before MTOC polarization. In these conditions, the can fuel the pool of exocytosis-competent granules. It is also 2+ rapid elevation of [Ca ]i in target cells was instrumental to detect possible that the MTOC-driven polarization of a large fraction of lytic granule secretion before MTOC polarization. lytic granules at the IS could limit CTL cytotoxic potential in tis-

Thus, in the present study, we reveal that MTOC/centrosome sues by sequestering a large fraction of lytic granules at the contact IMMUNOLOGY polarization is not strictly required for lytic granule secretion at the site with one target cell, thus preventing excessive elimination of CTL/target cell contact site during the first minutes after CTL/ adjacent cells. A combination of these mechanisms might function target cell encounter. At later time points (as those corresponding in vivo to improve the efficiency of target cell killing while con- to Fig. 3 and Fig. S4), we observed parallel polarization of MTOC fining CTL biological function. and lytic granules at the IS in agreement with previous results Our results do not exclude the possibility that microtubule in- obtained using electron microscopy on fixed CTL/target cell con- tegrity is required for efficient cytotoxicity, as indicated by previous jugates (11, 21). studies showing that functional microtubules and microtubule Our findings are complementary and extend the current model motor proteins are important for CTL lethal hit delivery (23, 24). of lytic granule secretion established by Griffiths and coworkers (1, Further research based on more recent technical approaches such 11). The model states that CTLs use a secretory mechanism by as time-lapse microscopy, 3D confocal microscopy, and super- which the centrosome contacts the plasma membrane and delivers resolution microscopy is required to investigate the role played by secretory granules at the center of the IS. Taken together, our microtubules and microtubule motors on lytic granule dynamics TIRFM and time lapse-microscopy results indicate that during the and synaptic localization. very first steps of CTL activation, lytic granule secretion is initiated All in all, by showing that lytic granule secretion can be mech- independently of MTOC polarization (not excluding that the anistically dissociated from the dynamics of MTOC reorientation, MTOC might enter in contact with the CTL plasma membrane at our results are complementary and extend the current model of later time points). In addition, our TIRFM data do not rule out the lytic granule secretion (11). Taken together with the established possibility that during the CTL activation process the MTOC model, our results indicate that on the encounter with a cognate might lie just beyond the TIRF plane and that it might bring the target cell, lytic granule secretion is swiftly triggered in CTLs by major lytic granule pool in a position facilitating further mobili- TCR engagement as a default mechanism of immediate response zation to the synaptic area. Thus, our results provide further un- to antigenic stimuli. Polarization of CTL lytic machinery follows to derstanding of the dynamics of lytic granule delivery. Although ensure the delivery of additional lytic granules to the IS, resulting they do not disprove the concept that MTOC polarization can in prolonged and confined secretion of lytic molecules toward one favor the movement of lytic granules toward the IS (6, 11), they target cell. When multiple cognate targets are encountered si- reveal the existence of a previously undetected early step of lytic multaneously, lytic granules are secreted at the different contact granule secretion, which is independent of MTOC polarization. sites resulting in multiple killing (9). The described early and We also provide further evidence that lytic granule secretion MTOC-independent step of lytic granule secretion is likely to be might be, under certain circumstances, uncoupled from MTOC instrumental for the exquisite rapidity of single and multiple target dynamics by blocking MTOC polarization with a selective in- cell annihilation exhibited by CTLs. hibitor of the ancestral polarity protein PKCζ (14, 15). Our results show that in conditions in which MTOC polarization is inhibited, Materials and Methods + + no major effect on cytotoxicity and on the activation of caspase-3 T Cells and APC. Either CD8 Vβ2 T cells or an HLA-A2–restricted T-cell line specific for a peptide of the CMVpp65 protein was used as human CTL models. in target cells could be detected. In interpreting these results, it + + ζ For CD8 Vβ2 T cells, T cells were isolated from whole blood of healthy donors should be noted that PKC inhibition does not completely revert + (Centre Hospitalier Universitaire Purpan, Toulouse, France). The CD8 T-cell the position of the MTOC in antigen stimulated CTLs to that of + + fraction was sorted from whole blood using RosetteSep (StemCell). CD8 Vβ2 unstimulated CTLs. However, our results show that in cognate T cells were isolated by positive selection using an anti-Vβ2 antibody (clone CTL/target cell conjugates in which PKCζ function was inhibited fi MPB2D5; Beckman Coulter) and goat anti-mouse IgG microbeads (Miltenyi a signi cant distancing of MTOC from the synaptic area was in- Biotec). Cell purity was assessed by FACS analysis (Facscalibur; Becton Dick- duced, yet no effect on cytotoxicity could be detected in parallel inson) using phycoerythrin (PE)-labeled anti-CD8 mAb (clone RPA-T8; BD experiments. The above phenomenon is directly illustrated by PharMingen) and FITC-labeled anti-Vβ2 mAb (clone MPB2D5; Beckman

Bertrand et al. PNAS Early Edition | 5of6 Downloaded by guest on September 24, 2021 + + Coulter). Freshly isolated CD8 Vβ2 T-cell populations were cultured in RPMI Measurement of TCR Down-Regulation, IFN-γ Production, CD107a Expression, 1640 medium, 5% (vol/vol) human serum, and IL-2 (250 U/mL) in the presence and Active Caspase-3 by FACS Analysis. Target cells were either unpulsed or of anti-CD3/CD28 mAb-coated Dynabeads (Invitrogen) at a ratio of one bead pulsed with 10 ng/mL of the bacterial superantigen TSST-1 for 1 h or with for 10 T cells. Blood samples from healthy donors were obtained following 10 μM CMV peptide for 2 h at 37 °C in RPMI and 5% FCS/Hepes. During the standard ethical procedures (Helsinki protocol) and with the approval of the last 15 min of pulsing, target cells were stained with CMTMR-orange (Mo- concerned Internal Review Boards. The HLA-A2–restricted T-cell line lecular Probes) or CellTracker™ Green CMFDA (5-Chloromethylfluorescein (CMVpp65) specific for the peptide NLVPMVATV of the human cytomegalo- Diacetate) (Molecular Probes) to discriminate them from CTLs. After washing, virus protein pp65 was used (25). × β + HLA-A2–matched EBV-transformed B cells (JY) were used as target cells cells were conjugated by 1-min centrifugation at 370 g with V 2 CTLs at and pulsed either with TSST-1 (Toxin Technology) when cocultured with a 1:1 ratio. After different times of culture, cells were either stained with anti- + + CD8 Vβ2 T cells or with specific peptide pp65 when cocultured with CMV- CD3 antibody (OKT3; ATCC) or with anti-CD107a. In some experiments, 10 μg/ specific CTL lines. mL Brefeldin A (Sigma) was added to the culture, and after 4 h of coculture, cells were fixed with paraformaldehyde, permeabilized with 0.1% saponin (in Confocal Microscopy. Target cells were either unpulsed or pulsed with 10 ng/mL PBS/3% BSA/Hepes), and stained with anti–IFN-γ mAb (clone B27; BD Bio- of the bacterial superantigen TSST-1 for 1 h or with 10 μM CMV peptide for sciences). Primary antibodies were followed by isotype-matched Alexa-con- 2 h at 37 °C in RPMI 5% FCS/Hepes and washed three times. During the last jugated secondary antibodies. For active caspase-3 detection, target cells 15 min of pulsing, in some experiments, target cells were stained with were left unstained, whereas T cells were stained with CMTMR-orange to be CellTracker™ Orange CMTMR [5-(and-6)-([(4-Chloromethyl)Benzoyl]Amino) excluded from the FACS analysis. After different times of culture, cells were Tetramethylrhodamine] (Molecular Probes). In some experiments, CTLs were stained with an anti-active caspase-3 rabbit Ab (clone C92-605; BD Pharmin- pretreated or not with a myristoylated PKCζ pseudosubstrate peptide (PKCζ- gen), followed by an isotype-matched Alexa-conjugated secondary antibody. PS) at 10 μM (Invitrogen) for 1 h at 37 °C and washed. After washing, cells were conjugated by 1-min centrifugation at a ratio of one CTL for one target cell. At different times after conjugation, cells were Cytotoxic Assays. Target cells were either unpulsed or pulsed with 10 ng/mL of fixed with 3% paraformaldehyde, permeabilized with 0.1% saponin (in PBS/ the bacterial superantigen TSST-1 for 1 h (or with 10 μM pp65 peptide for 2 h) 3% BSA/Hepes), and stained with the following primary antibodies: anti– at 37 °C in RPMI/5% FCS/Hepes and washed. CTLs were either untreated or α-tubulin mAb (clone DM1A, Sigma) or anti–α-tubulin rabbit Ab (ab15246; pretreated with 10 μM PKCζ-PS for 1 h at 37 °C and washed. CTLs were Abcam), anti–γ-tubulin rabbit Ab (ab11317; Abcam), anti-human perforin conjugated with target cells at different E:T cell ratios for 4 h. To distinguish mAb (clone δG9; BD Pharmingen), anti–phospho-PKCζ (p-PKCζ) rabbit Ab (sc- CTLs from target cells in the analysis, different approaches were used. Target 12894-R; Santa Cruz Biotechnology), anti-phosphotyrosine (p-Tyr) mAb cells were loaded with 1 μM CellTrace Far Red DDAO (N,N-Dimethyldecyl- (clone sc-7020; Santa Cruz), or anti-CD45 mAb (clone 9.4; ATCC). The staining amine-N-oxide) (DDAO-SE) Far Red (Molecular Probes) in RPMI for 15 min at for CD107a (clone H4A3; BD Pharmingen) was performed before per- 37 °C, prior conjugation with CTLs. Alternatively, CTLs were labeled before meabilization and followed by intracellular staining with anti–α-tubulin conjugation with 1 μM CMFDA (Molecular Probes) for 15 min at 37 °C. Im- rabbit Ab. Primary Abs were followed by goat anti-mouse isotype-specific Ab or goat anti-rabbit Ab labeled with Alexa 350, Alexa 488, Alexa 546, mediately before FACS analysis, 7-Amino-actinomycin D (7-AAD) was added Alexa 633, or Alexa 647 (Molecular Probes). The samples were mounted in to each sample to stain dead cells. 90% glycerol-PBS containing 2.5% DABCO (Fluka) and examined using ei- ther a LSM 510 or a 710 confocal microscope (Zeiss) with a 63× Plan-Apo- ACKNOWLEDGMENTS. We thank Nathalie Joncker and Mark M. Davis for chromat objective (1.4 oil). For live cell imaging, target cells were loaded discussion, Daniel Dunia for discussion and critical reading of the manuscript, with 1μM Fluo-4 AM for 30 min at 37 °C to detect [Ca2+] increase as a marker Magda Rodrigues for help in image analysis, and Renaud Poincloux at the i “plateau technique d’imagerie” (Institute of Pharmacology and Structural of early cellular damage, and CTLs were loaded with LysoTracker-red and Biology of Toulouse) for help in TIRFM experiments. We also thank the TubulinTracker-green (Molecular Probes) for 30 min at 37 °C in RPMI 1640 “plateau technique de cytométrie et de microscopie,” Institut National de medium and 5% FCS/Hepes. Target cells were seeded into microchambers la Santé et de la Recherche Médicale, Unité Mixte de Recherche 1043. This (Lab-Tek Chambered coverglass; Nalge Nunc) previously coated with poly-D- work was supported by grants from the Association pour la Recherche sur le lysine (Sigma), and CTLs were added at the beginning of the recording. Cancer and from the Institut National du Cancer.

1. Griffiths GM, Tsun A, Stinchcombe JC (2010) The immunological synapse: A focal point 14. Bertrand F, et al. (2010) Activation of the ancestral polarity regulator protein kinase C for endocytosis and exocytosis. J Cell Biol 189(3):399–406. zeta at the immunological synapse drives polarization of Th cell secretory machinery 2. Huse M, Quann EJ, Davis MM (2008) Shouts, whispers and the kiss of death: Di- toward APCs. J Immunol 185(5):2887–2894. rectional secretion in T cells. Nat Immunol 9(10):1105–1111. 15. Tourret M, et al. (2010) T cell polarity at the immunological synapse is required for 3. Law RH, et al. (2010) The structural basis for membrane binding and pore formation CD154-dependent IL-12 secretion by dendritic cells. J Immunol 185(11):6809–6818. by lymphocyte perforin. Nature 468(7322):447–451. 16. Etienne-Manneville S, Manneville JB, Nicholls S, Ferenczi MA, Hall A (2005) Cdc42 and 4. de Saint Basile G, Ménasché G, Fischer A (2010) Molecular mechanisms of biogenesis Par6-PKCzeta regulate the spatially localized association of Dlg1 and APC to control and exocytosis of cytotoxic granules. Nat Rev Immunol 10(8):568–579. cell polarization. J Cell Biol 170(6):895–901. 5. Purbhoo MA, Irvine DJ, Huppa JB, Davis MM (2004) T cell killing does not require the 17. Leupin O, Zaru R, Laroche T, Müller S, Valitutti S (2000) Exclusion of CD45 from the formation of a stable mature immunological synapse. Nat Immunol 5(5):524–530. T-cell receptor signaling area in antigen-stimulated T lymphocytes. Curr Biol 10(5):277–280. 6. Stinchcombe JC, Bossi G, Booth S, Griffiths GM (2001) The immunological synapse of 18. Qu B, et al. (2011) Docking of lytic granules at the immunological synapse in human CTL contains a secretory domain and membrane bridges. Immunity 15(5):751–761. CTL requires Vti1b-dependent pairing with CD3 endosomes. J Immunol 186(12):6894–6904. 7. Sykulev Y, Joo M, Vturina I, Tsomides TJ, Eisen HN (1996) Evidence that a single 19. Rak GD, Mace EM, Banerjee PP, Svitkina T, Orange JS (2011) Natural killer cell lytic peptide-MHC complex on a target cell can elicit a cytolytic T cell response. Immunity granule secretion occurs through a pervasive actin network at the immune synapse. 4(6):565–571. PLoS Biol 9(9):e1001151. 8. Valitutti S, Müller S, Dessing M, Lanzavecchia A (1996) Different responses are elicited 20. Brown AC, et al. (2011) Remodelling of cortical actin where lytic granules dock at in cytotoxic T lymphocytes by different levels of T cell receptor occupancy. J Exp Med natural killer cell immune synapses revealed by super-resolution microscopy. PLoS Biol 183(4):1917–1921. 9(9):e1001152. 9. Wiedemann A, Depoil D, Faroudi M, Valitutti S (2006) Cytotoxic T lymphocytes kill 21. Ueda H, Morphew MK, McIntosh JR, Davis MM (2011) CD4+ T-cell synapses involve multiple targets simultaneously via spatiotemporal uncoupling of lytic and stimula- multiple distinct stages. Proc Natl Acad Sci USA 108(41):17099–17104. tory synapses. Proc Natl Acad Sci USA 103(29):10985–10990. 22. Pores-Fernando AT, Bauer RA, Wurth GA, Zweifach A (2005) Exocytic responses of 10. Poenie M, Tsien RY, Schmitt-Verhulst AM (1987) Sequential activation and lethal hit single leukaemic human cytotoxic T lymphocytes stimulated by agents that bypass the measured by [Ca2+]i in individual cytolytic T cells and targets. EMBO J 6(8):2223–2232. T cell receptor. J Physiol 567(Pt 3):891–903. 11. Stinchcombe JC, Majorovits E, Bossi G, Fuller S, Griffiths GM (2006) Centrosome 23. Burkhardt JK, McIlvain JM, Jr., Sheetz MP, Argon Y (1993) Lytic granules from cyto- polarization delivers secretory granules to the immunological synapse. Nature toxic T cells exhibit kinesin-dependent motility on microtubules in vitro. J Cell Sci 104 443(7110):462–465. (Pt 1):151–162. 12. Stinchcombe JC, et al. (2011) Centriole polarisation to the immunological synapse 24. Kupfer A, Dennert G (1984) Reorientation of the microtubule-organizing center and directs secretion from cytolytic cells of both the innate and adaptive immune systems. the Golgi apparatus in cloned cytotoxic lymphocytes triggered by binding to lysable BMC Biol 9:45. target cells. J Immunol 133(5):2762–2766. 13. Caramalho I, Faroudi M, Padovan E, Müller S, Valitutti S (2009) Visualizing CTL/mel- 25. Faroudi M, et al. (2003) Lytic versus stimulatory synapse in cytotoxic T lymphocyte/ anoma cell interactions: Multiple hits must be delivered for tumour cell annihilation. J target cell interaction: Manifestation of a dual activation threshold. Proc Natl Acad Cell Mol Med 13(9B):3834–3846. Sci USA 100(24):14145–14150.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1218640110 Bertrand et al. Downloaded by guest on September 24, 2021