No Specific Subcellular Localization of Protein Kinase C Is Required for Cytotoxic T Cell Granule Exocytosis*

Cytotoxic T cells kill virus-infected cells and tumor cells by releasing lytic granules that contain cell-killing contents. Exocytosis requires calcium influx and protein kinase C (PKC) activation. Here, we extend our previous finding regarding the lack of isoform specificity of PKCs in the granule release step, showing that mutant constitutively active PKCδ can substitute for phorbol esters and support exocytosis. PKCδ, a novel PKC isoform, was recently shown to play a role in lytic granule reorientation. Surprisingly, however, our results suggested that mutant PKCδ did not localize to the plasma membrane (PM). To test directly whether PKC has to be in the PM to drive exocytosis, we generated mutants of various PKC isoforms that were tethered either to the outer mitochondrial membrane or to the PM. Tethered mutant PKCδs were able to promote exocytosis as effectively as the untethered version. The substrates of PKCs involved in lytic granule exocytosis are currently unknown, but subcellular localization is believed to be a critical factor in determining PKC accessibility to substrates. That there is no requirement for specific PKC localization in lytic granule exocytosis may have important implications for the identity of PKC substrates.

PKCs are serine/threonine protein kinases that are involved in many biologically important processes. There are 10 isoforms of PKC that are classified as classical, novel, or atypical based on their modulators (8). In helper T cells, the novel isoform PKC plays a preferential role in important cell functions (9). It synergizes with the calcium/calmodulin-dependent phosphatase calcineurin to upregulate IL-2 gene transcription via activation of the nuclear factor of activated T-cells (NFAT) (10,11), and also specifically localizes to the immunological synapse (12), a specialized membrane domain that forms during the interaction of T cells with antigen-presenting cells (12) or targets (13). In contrast, it appears at present that there is no isoform specificity for PKCs in the degranulation step of CTL lytic granule exocytosis, since when calcium is elevated with drugs, constitutively active mutants of multiple PKC isoforms can substitute for treatment with DAG analogs such as phorbol myristate acetate (PMA) (14,15). However, there may be isoform specificity in other steps of the lytic interaction. Recently PKC␦, a novel isoform, was shown by Ma et al. (16) to play a specific role in controlling reorientation of lytic granules to the site of contact with target cells in murine CTLs. Further work from this group showed that the effects of PKC␦ on target cell lysis required kinase activity, but that PKC␦ localized to lytic granules in a manner that was independent of kinase activation (17). Although both reorientation of lytic granules and target cell killing were reduced in CTLs from PKC␦ knock-out mice, it is important to note that block of granule reorientation could inhibit target cell killing without necessarily affecting the actual exocytosis of lytic granules, because reorientation precedes exocytosis.
In previous experiments using TALL-104 human leukemic CTLs, we did not detect expression of PKC␦ and so did not determine whether a constitutively active mutant form of it, like PKC␣ and PKC, is capable of supporting exocytosis (14). However, there appear to be problems with the antibody we used. For example, it has been reported that it only detects an unphosphorylated form of PKC␦, not the phosphorylated form capable of catalytic activation (18). In light of the new information provided by Ma et al., we decided to revisit the role of PKC␦ in lytic granule exocytosis. Here, we report that TALL-104 cells do express PKC␦, and that mutant constitutively active PKC␦ can support exocytosis. Wild-type PKC␦ displayed an apparent cytosolic localization, in contrast to what was observed in murine CTLs (17), and it translocated to the plasma membrane upon stimulation with PMA, similar to what we observed for PKC␣ and PKC (14). Strikingly, however, mutant PKC␦ also displayed an apparent cytosolic localization in unstimulated cells, appearing in the PM only after stimulation with thapsi-gargin (TG). Our previous work indicated that mutant PKC␣ and PKC displayed a predominantly PM localization, and we had suspected that this indicated that the PM was the site where PKC activity was required for granule exocytosis. Targeting of PKC isoforms to specific membranes is thought to be important for controlling their access to different substrates. This is thought to underlie the ability of isoforms to serve specific functions (reviewed in Ref. 19), because in general PKC isoforms display relatively little substrate specificity (20). As the emergent consensus appears to be that substrates of a given PKC isoform are likely to be associated with the membrane to which that isoform translocates upon stimulation, the idea that there might be no need for PKC to be localized to the PM to participate in exocytosis was quite surprising to us.
To specifically test whether there is a requirement for PM localization of PKC in granule exocytosis, we tethered constitutively active mutant PKC␦ to the outer membrane of mitochondria using a targeting sequence derived from BCL-x (21). Mitochondrially tethered mutant PKC␦ was able to support granule exocytosis essentially as effectively as the untethered version. A version of the mutant tethered to the PM by a myristoylation sequence was also able to promote exocytosis. These results indicate that no specific subcellular localization of protein kinase C is required for cytotoxic T cell granule exocytosis.

EXPERIMENTAL PROCEDURES
cDNA Constructs and Transfections-Wild-type PKC␦ and the constitutively active PKC␦ mutant DR144/145A (22) were given to us by Dr. Mary Reyland (UCDHSC, Aurora, CO). Standard PCR methods were used to subclone them into pEGFPN1 (Clontech). PKC␣ and PKC mutant constructs have been previously described (14). BCL-x in pEGFP-C3 (21) was kindly provided by Dr. Christoph Borner (Albert-Ludwigs-University Freiburg, Germany). Because we already had mutant PKCs in pEGFP-N1, to facilitate construction of mutant PKC-GFP-BCL-x chimeras, PCR methods were used to subclone the BCL-x sequence after the GFP coding sequence in pEGFP-N1 using the XbaI restriction enzyme site. This construct, which we refer to as GFP-BCL-x, was then used to make mitochondrially tethered mutant PKCs. Coding frames of the constitutively active mutants (all were in pEGFPN1) were double-digested and then ligated into the GFP-BCL-x backbone vector. To make myr-PKC␦-GFP, a sequence coding for the myristoylation motif (GSSKSKPKDPSQR) was added to the 5Ј-end of the mutant PKC␦-GFP using an engineered XhoI restriction site. For heterologous expression experiments, 2.5 ϫ 10 6 TALLs were transfected using an Amaxa Nucleofector (Amaxa Biosystems, Gaithersburg, MD) using program T-20 and solution V. Experiments were performed 6 -7 h post-transfection.
Chemicals, Reagents, Cells, and Solutions-Thapsigargin and PMA were from Alexis Biochemicals (San Diego, CA). Anti-CD107a (clone H4A3) was purchased from BD Biosciences (San Diego, CA), and was conjugated to Alexafluor 647 using a kit from Molecular Probes/Invitrogen (Eugene, OR) according to manufacturer's protocol. The antibodies against the phosphorylated PKC substrate phosphorylation site and the phospho-Thr-505 PKC␦ were from Cell Signaling Technology, Inc (Danvers, MA). Mitotracker Orange CMTMRos and Cy3-wheat germ agglutinin were from Molecular Probes/Invitrogen. Recombinant human IL-2 was provided by the National Cancer Institute's Research Reagent Program (Bethesda, MD). TALL-104 cell culture and Ringer's solution have been previously described (23).
Immunostaining and Flow Cytometry-LAMP externalization was monitored as described previously (14,24). The anti-LAMP-1 antibody was used at a final concentration of 0.9 g/ml. For intracellular staining, CALTAG's fix and perm kit was used following the manufacturer's protocol (CALTAG/Invitrogen). Flow cytometry was performed on a FACSCalibur at the University of Connecticut at Storrs Flow Cytometry and Confocal Microscopy Facility. FlowJo software (TreeStar, Ashland, OR) was used to analyze data offline.
Western Blotting-Whole cell lysates prepared from TALL-104 cells, Jurkat cells or human brain lysates (Imgenex Corp., San Diego, CA) were run on 10% polyacrylamide gels and transferred to nitrocellulose. To assess expression of PKC␦, blots were probed with a rabbit anti-PKC␦ antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) followed by a goat anti-rabbit horseradish peroxidase-conjugated secondary antibody and chemiluminescent detection. To determine whether cleavage of tethered PKC␦ generated untethered catalytically active fragments, blots were probed with the anti-PKC␦ antibody and a goat anti-GFP antibody. Immunoreactivity was detected with Alexafluor 555-labeled donkey anti-rabbit secondary antibodies (PKC␦) or FITC-labeled donkey anti-goat secondary antibodies (GFP). Fluorescent images were acquired in epifluorescence mode with a Kodak IS2000MM (Carestream Health, Rochester NY) with appropriate excitation and emission filters. Bradford assays were performed so as to allow us to load equal amounts of protein.
Statistics-Statistical significance was assessed using repeated measures analysis of variance (Instat, Graphpad Software, San Diego, CA).

TALL-104 Cells Express PKC␦, Which Translocates to the Plasma Membrane upon Stimulation with Soluble Chemical
Agents-We used Western blotting to probe for PKC␦ expression in TALL-104 cells (Fig. 1A), and found that expression levels were comparable to those seen in Jurkat helper T cells and in a commercial human brain lysate. Next, we probed lysates from unstimulated cells and from cells stimulated with 50 nM PMA and 1 M TG with an antibody against PKC␦ phosphorylated on Thr-507 (this residue corresponds to Thr-505 in mouse, the species against which the antibody was raised) (Fig.  1B). Phosphorylation at this residue may be associated with PKC␦ activation. Whereas there was no immunoreactivity in unstimulated cells, stimulation with PMA triggered recognition with the anti-phospho-Thr-505 PKC␦ Ab. We conclude that PKC␦ is not phosphorylated in unstimulated cells, and therefore have no ready explanation for our previous failure to detect PKC␦. However, that PKC␦ is phosphorylated on Thr-507 after treatment with PMA may be consistent with the idea that PMA activates the native enzyme in TALL-104 cells. We next assessed the subcellular localization of PKC␦ in unstimulated cells and in cells stimulated with TG and PMA (drugs that trigger calcium influx and activate PKC respectively) in two ways: heterologous expression of a PKC␦-GFP fusion protein ( Fig. 1C) and immunocytochemical analysis with the same antibody used in Western blotting (Fig. 1D). These probes were then visualized using confocal microscopy. Results with the antibody were somewhat unsatisfying, because cells had to be permeabilized with methanol, which changed their morphology (see brightfield images in Fig. 1D). Nonetheless, both approaches suggest that PKC␦ is found in the cytoplasm in unstimulated cells, and translocates to the PM upon stimulation, similar to what we observed previously with PKC␣ and PKC (14).
A Constitutively Active Mutant PKC␦ Can Substitute for PMA in Promoting Lytic Granule Exocytosis-We next tested whether a constitutively active mutant PKC␦ could promote exocytosis (Fig. 2). We overexpressed the constitutively active PKC␦ mutant fused to GFP, and stimulated cells with TG. We assessed exocytosis by measuring externalization of LAMP-1 using flow cytometry, as described previously (14). Responses of untransfected cells to stimulation with TG or TGϩ PMA are shown in Fig. 2A, panel i. TG treatment alone triggered sub-maximal exocytosis in GFP-transfected cells (used as controls) at all levels of expression ( Fig. 2A, panel iii. As we found previously was the case with constitutively active mutant PKC␣ and PKC (14), cells expressing high levels of constitutively active mutant PKC␦ fused to GFP demonstrated significantly increased levels of anti-LAMP staining when compared with the non-expressing cells or cells expressing equivalent levels of GFP (Fig. 2

, A, panel iv and B).
Anti-LAMP fluorescence intensity increased with increasing levels of expression of mutant PKC␦. This indicates that PKC␦, like PKC␣ and PKC, can substitute for PMA in synergizing with calcium increases to promote TCR-independent lytic granule exocytosis. Interestingly, and in contrast to what we observed with PKC␣ and PKC mutants, ϳ10 -15% of cells expressing high levels of PKC␦ mutant were LAMPpositive in the absence of any overt stimulation (supplemental Fig. S1). This effect was at least partially calciumindependent, as adding 10 mM EGTA to cells 2 h after nucleofection (the earliest time at which GFP expression is detected) reduced the number of LAMP-positive cells, but did not fully prevent the response (supplemental Fig. S1).  SEPTEMBER 11, 2009 • VOLUME 284 • NUMBER 37

JOURNAL OF BIOLOGICAL CHEMISTRY 25109
Mutant PKC␦-GFP Is Not Associated with the Plasma Membrane in Unstimulated Cells-In our previous experiments, we found that both PKC␣ and PKC mutants were constitutively associated with the plasma membrane (14). However, when we examined the subcellular localization of mutant constitutively active PKC␦ using confocal microscopy, we found that in most unstimulated cells it displayed an apparent cytosolic localization (Fig.  3A). As described above, some cells transfected with mutant PKC␦ expose LAMP on their surface in the absence of an overtly stimulated increase in intracellular Ca 2ϩ . When we stained intact mutant PKC␦-transfected TALL-104 cells with anti-LAMP antibody, we found PKC␦ was cytosolic in most of the cells that stained, although occasional examples could be found in which it appeared to reside in the PM. There was thus no apparent correlation between localization of mutant PKC␦ and exocytosis. However, when we treated transfected cells with TG, mutant PKC␦-GFP was found in the plasma membrane in most cells (Fig. 3B). This PM translocation required calcium influx, as it did not occur in the absence of extracellular Ca 2ϩ (supplemental Fig. S2).
Mutant PKC␦ Tethered to the Outer Mitochondrial Membrane Retains Catalytic Activity and Can Promote Exocytosis-Results presented thus far raise the possibility that PKC␦ may not need to be associated with the PM to participate in exocytosis of lytic granules. To test this directly, we created a mutant construct that would be prevented from translocating to the PM. We tethered mutants to the outer mitochondrial membrane using a specific targeting sequence from BCL-x (21) inserted at the 3Ј-end of GFP, which was itself at the 3Ј-end of the mutant PKC (Fig. 4A). In this construct, were cleavage between the regulatory and catalytic domains to occur (a process reviewed in Ref. 25), the catalytic domain would remain tethered. Note that we ini- tially constructed mitochondrially tethered versions of mutant PKC␣, PKC, and PKC␦. However, only mutant PKC␦-GFP-BCL-x expressed at sufficiently high levels to be useful (data not shown), and so this is the only one of the mitochondrially tethered constructs that is described further.
We first confirmed that our targeting strategy tethered PKC␦ to the outer mitochondrial membrane. We transfected TALL-104 cells with mutant PKC␦-GFP-BCL-x, and assessed subcellular localization using confocal microscopy (Fig. 4B). Representative images of cells that were not stimulated are shown in the top row, and images of cells that were treated with TG are shown in the bottom row. In these experiments, we used mitotracker orange to label mitochondria (shown in red in column 3 of Fig. 4B)). The fluorescence of mutant PKC␦-GFP-BCL-x colocalized extensively with mitochondria (a merge is shown in column 4 of Fig. 4B). Fluorescence of PKC␦-GFP-BCL-x did not overlap with anti-LAMP fluorescence (anti-LAMP fluorescence is shown in red in column 5 of Fig. 4B, and a merge between PKC␦-GFP-BCL-x and anti-LAMP is shown in column 6). Note that LAMP staining following TG treatment was not as prominent in these experiments (bottom row of Fig. 4B) as in those shown in Fig. 3B or Fig. 6B, likely due to the fact that cells were treated with mitotracker orange, which we find reduces LAMP staining intensity by ϳ30 -40% (data not shown).
We next tested whether cleavage products of PKC␦-GFP-BCL-x were generated by proteolysis that could yield unteth-ered catalytically active PKC fragments (Fig. 5). We used Western blotting with primary antibodies against GFP and against the catalytic domain of PKC␦ to determine whether there were species that were recognized by the anti-PKC␦ antibody that were not recognized by the anti-GFP antibody, or vice versa. Essentially all of the species reacted with both antibodies, consistent with the idea that all consisted of a fusion of mutant PKC␦ to GFP. Thus, as the fluorescence associated with GFP was localized to mitochondria, these results suggest that all active PKC␦ was likely to be properly tethered.
To confirm that the mitochondrially tethered mutant PKC␦ was catalytically active, we used a commercially available antibody against the phosphorylated PKC consensus phosphorylation site to assess relative levels of PKC-dependent substrate phosphorylation in transfected cells (Fig. 4, C and D). We previously used this antibody in Western blotting experiments to test the effects of PKC inhibitors on PKC activity in TALL-104 cells (14), and evidence that the antibody works in flow cytometry is presented in supplemental Fig. S3. Cells were processed for indirect intracellular immunostaining, and the fluorescence intensity of GFP, mutant PKC␦, or mutant PKC␦-GFP-BCL-x together with anti-phospho PKC substrate antibody staining (detected with a Cy5-labeled secondary antibody) was measured with flow cytometry. For a given level of GFP fluorescence intensity, mutant PKC␦-GFP-BCL-x and untargeted mutant PKC␦-GFP produced comparable levels of anti-phospho PKC substrate antibody staining (Fig. 4, C and D), suggesting equivalent catalytic activity, while expressing GFP alone had no effect.
Next, we tested whether mutant PKC␦-GFP-BCL-x, like untethered constitutively active mutant PKC␦, can substitute for PMA in promoting lytic granule exocytosis (Fig. 4, E and F). We expressed GFP-BCL-x (which demonstrated an identical subcellular localization to PKC␦-GFP-BCL-x, data not shown) or mutant PKC␦-GFP-BCL-x, stimulated cells with TG alone, and assessed exocytosis via LAMP staining using flow cytometry. As was also the case in GFP-transfected cells (see Fig. 2), TG treatment resulted in low levels of exocytosis at all levels of expression of GFP-BCL-x (Fig. 4, E and F). In contrast, TGstimulated cells expressing high levels of mutant PKC␦-GFP-BCL-x showed significantly increased levels of anti-LAMP staining compared with non-expressing cells or cells expressing equivalent levels of GFP-BCL-x. The enhancement of exocytosis with mutant PKC␦-GFP-BCL-x was essentially indistinguishable from that seen with the untethered PKC␦ mutants.
Finally, to confirm that mutant PKC␦ can function to promote lytic granule exocytosis when directed to subcellular locations other than the mitochondria, we examined the effects of tethering it to the plasma membrane via a myristoylation sequence (Fig. 6). Mutant myr-PKC␦-GFP colocalized extensively with Cy3-labeled wheat germ agglutinin (WGA) applied to cells after fixation to label the plasma membrane. However, there was some apparent overlap of myristoylated mutant PKC␦ with anti-LAMP staining (Fig. 6B), and it was clear that not all of the fluorescence of either the myr-mutant-PKC␦ or WGA itself was associated with the PM. Association with LAMP may reflect trafficking between the PM and intracellular organelles, including the lytic granules, and is also likely con- founded by the fact that mutant PKC␦ can promote modest levels of exocytosis without additional stimulation (supplemental Fig. S1), resulting in appearance of LAMP in the PM. As was the case for the mitochondrially tethered PKC␦ mutant, cleavage fragments were detected using Western blotting, but the majority reacted with both anti-GFP and anti-PKC␦ antibodies (Fig. 5). The construct was catalytically active as assessed with the anti-phospho PKC substrate antibody (Fig. 6, C and D). When we transfected cells with GFP alone or with myristoy-lated mutant PKC␦, stimulated them with TG and measured exocytosis via externalization of LAMP, there was a significant enhancement of exocytosis in cells that expressed high levels of the myristoylated mutant compared with cells that expressed comparable levels of GFP or to nonexpressing cells from the same population (Fig. 6, E and F). Similar results were obtained with a construct in which mutant PKC␦ was fused at its N terminus to a truncated form of human CD4 from which most of the cytosolic sequence had been deleted (supplemental Fig. S4). Although neither construct resulted in a pure PM localization, they would not be expected to. In the case of the myristoylated construct, the protein will likely interact with numerous different membrane compartments. In the case of the CD4 construct, the protein would be expected to be distributed throughout the ER, Golgi, and endosomal compartments due to trafficking. Nevertheless, both are clearly distributed differently than mutant PKC␦-GFP-BCL-x, yet they are essentially equally able to promote exocytosis.

DISCUSSION
Two key new findings are presented here. First, mutant PKC␦, like mutant PKC␣ and mutant PKC, can synergize with increases in intracellular calcium to promote lytic granule exocytosis. This was not clear from previous work (16,17), as in those studies granule reorientation was inhibited, and this is likely to inhibit in an obligate manner downstream steps such as exocytosis. That PKC␦ can participate in granule exocytosis further underscores the fact that there is no isoform specificity to this process (14,15). This appears not to be the case with regard to lytic granule polarization, where PKC␦ plays a preferred role (16,17). Interestingly, of the constitutively active mutant PKCs we have examined, only the PKC␦ mutant was able to promote exocytosis independent of calcium influx (supplemental Fig. S1). This effect may somehow be related to the ability of PKC␦ to participate in the reorientation of lytic granules (16,17). Note that Ma et al. showed that PKC␦ colocalized with lytic granules in murine CTLs (17). We did not observe this to be the case in TALL-104 cells. Instead, PKC␦ exhibited a cytosolic appearance in unstimulated cells, and translocated to the membrane upon stimulation with PMA. We are not sure how to account for this discrepancy. It is apparently not due to the fact that we were examining the localization of overexpressed PKC␦ in cells that endogenously express it, whereas Ma et al. were using CTLs from PKC␦ knock-out animals, since they report similar localization using immunocytochemical analysis in wild-type cells (17).
Second, the results we present here, taken together with our previous work (14), demonstrate clearly that PKCs can partici-pate in lytic granule exocytosis regardless of whether or not they are located in the plasma membrane. The result that supports this most directly is the ability of mutant PKC␦-GFP-BCL-x to support exocytosis. This is surprising, as targeting to specific membranes is thought to control the accessibility of PKCs to their substrates (26,27). Underscoring the importance of localization, one of the key initial pieces of evidence supporting a preferential role for PKC in helper T cell functions was its specific localization to what is now called the immunological synapse (28). The most direct evidence to date for the importance of PKC localization in determining interactions with substrates comes from experiments that showed that PKC must be localized to the PM to phosphorylate the PM substrate MARCKs, while only a cytosolic form of PKC could phosphorylate mutant cytosolic MARCKs (29).
Having provided strong evidence that there is no preferred localization for PKCs in promoting exocytosis, it still may be informative to speculate as to how the localization of PKCs might occur. We think that wild-type PKC␣, PKC, and PKC␦ likely translocate to the PM after stimulation with PMA because this phorbol ester accumulates in the PM. Wang et al. (30) have shown that PKC␦ translocates to different membranes when cells are stimulated with different phorbol ester analogs or bryostatin in CHO cells, and the effect seems to be governed by analog hydrophobicity (31). Once at the PM, PKCs might bind to phospholipids, to specific binding proteins such as RACKS (receptors for activated c kinase) and STICKs (substrates that interact with c kinase (see Ref. 26,27), or to both lipids and protein. Membrane binding is then likely associated with activation of catalytic function.
Although we did not realize it at the time, it is not clear why constitutively active PKC␣ and PKC mutants associate with the PM, as we showed previously was the case (14). These mutants are rendered active by amino acid substitutions that displace the pseudosubstrate domain from the catalytic domain, and so do not require membrane binding for activation (22). However, it may be that they still bind lipids with high affinity. If they do, then it seems reasonable to suppose that mutant PKC␣ and PKC translocate to the membrane as a result of lipid binding. If so, then mutant PKC␦ may bind to a protein that prevents membrane association, unless it alone of the isoforms does not bind lipids when rendered constitutively active. Alternatively, if mutant PKC␣ and PKC do not bind lipids, then their membrane association likely arises from binding to RACKs or STICKs. If this is the case, then we can infer that mutant PKC␦ does not interact with the same protein(s). In either scenario, it seems likely that PKC␣ and PKC interact with protein(s) differently than PKC␦. Furthermore, those proteins are unlikely to be substrates for exocytosis, as in either case not all of the mutants that can drive exocytosis bind to them. Thus, our results suggest that there are probably proteins to which PKCs bind in an isoform-specific manner that are not relevant substrate(s) for lytic granule exocytosis. Because mutant PKC␦, which lacks functional Ca 2ϩ -binding C2 domains, appears to translocate to the PM in a manner that depends on Ca 2ϩ , we lean toward the idea that PKC␦ binds to a cytosolic protein that prevents its membrane association, but then translocates to the PM after a Ca 2ϩ rise. We do not yet know whether translocation of the mutant PKC␦ occurs as a result of exocytosis. However, as described above, this protein, if it exists, is not likely to be a substrate important for exocytosis.
Assuming that all of the PKC constructs we have tested have the same substrate(s) relevant for exocytosis, those substrates must be accessible to PKCs residing in the PM, tethered to the mitochondrial membrane, or, in the case of mutant untethered PKC␦, likely resident in the cytoplasm. This would seem to argue against PM-resident proteins as substrates, provided that it is correct that only PM-resident PKCs can phosphorylate PM substrates as has been reported (29). We feel that it will be important to confirm this, but so far have been unable to reliably detect a change in MARCKS localization from the PM to intracellular membranes following stimulation with PMA, and thus have been unable to investigate the effects on MARCKS localization of expressing various tethered mutant PKC␦s.
If the PM is not the location of the substrate(s) involved in granule exocytosis, then what is? Proteins associated with lytic granules might be accessible to cytosolic or to mitochondrially FIGURE 6. Constitutively active myristoylated mutant PKC␦-GFP localizes to the plasma membrane, is catalytically active and can promote granule exocytosis. A, schematic representation of the myristoylation targeting construct. The amino acid sequence of the myristoylation sequence is shown in the top portion of the figure. B, myristoylation sequence targets mutant PKC␦ to the plasma membrane. TALL-104s were transfected with myristoylated mutant PKC␦-GFP (shown in green in column 2). 6 -7 h after transfection, the cells were fixed and stained with Cy3-labeled wheat germ agglutinin (shown in red in column 3) and Alexafluor 647-labeled anti-LAMP antibody (shown in red in column 5) then fixed. The color merge between GFP and WGA (column 4) shows more extensive colocalization than the color merge between GFP and anti-LAMP (column 6). A bright field view is shown in column 1. Images are from an unstimulated cell (top) and from a cell that was treated with TG (bottom). Scale bars are 5 m. C, myristoylated mutant PKC␦-GFP is catalytically active. TALL-104 cells were transfected with GFP (panel i) or myristoylated mutant PKC␦-GFP (panel ii) then 6 -7 h later were fixed, permeabilized, and stained with anti-phospho PKC substrate antibody. Representative dot plots of anti-phospho-PKC substrate antibody staining intensity versus GFP fluorescence intensity are shown. Numbered bars denote gating regions that were used to analyze the data subsequently. D, data from three experiments like the one in C were averaged and plotted for the different gating regions shown in C. Open circles are data obtained with myristoylated mutant PKC␦-GFP. Data from Fig. 4D are replotted, and the symbols used are the same as in that figure. E, myristoylated mutant PKC␦-GFP can substitute for PMA and support lytic granule exocytosis. Cells were either transfected with GFP (panel i) or with myristoylated mutant PKC␦-GFP (panel ii). Representative plots are shown of anti-LAMP antibody staining intensity for cells treated with TG for different gating regions as shown in (C, panels i and ii). Histograms are arranged with untransfected cells at the back, and cells with increasing levels of expression progressively toward the front. F, quantification from three experiments like the one shown in E. Gray bars represent cells transfected with GFP, and black bars represent myristoylated mutant PKC␦-GFP-transfected cells. * and # indicate that cells from gating region 3 expressing myristoylated mutant PKC␦-GFP differ significantly from non-expressers from the same sample (gating region 1) and from cells expressing high levels of GFP (cells from gating region 3).
tethered PKCs, and could become accessible to PM-resident PKCs during docking of granules with the PM. Proteins associated with the lytic granules would obviously also likely be accessible to PKCs associated with granules, as was reported to be the case for PKC␦ in murine CTLs (17). Proteins resident in a recycling endosomal compartment might also be accessible to PKCs in all three locations. Ménager et al. (32) recently provided evidence that a recycling endosomal compartment had to fuse to lytic granules to form a fully functional granule capable of undergoing secretion. Alternatively, the substrate(s) could be cytosolic proteins. If so, they could bind while the PKC is in the cytoplasm. For wild-type PKCs, this might be while the mature enzyme resides in the cytosol prior to activation. For mutant PKC␣ and PKC, it could occur after synthesis but before membrane association. Generally consistent with our speculation, it has recently been proposed that in helper T cells, PKC binds to SPAK, a recently discovered helper T cell substrate, in the cytoplasm, and the two translocate to the PM together upon PKC activation (33).
We believe that the results we present here will aid in several key ways in the identification of PKC substrates important for lytic granule exocytosis. First, as the previous paragraphs indicate, our results suggest that attempts to identify substrates via identification of PKC binding partners will likely be complicated by the fact that there are almost certainly proteins to which PKCs bind that are not important substrates for exocytosis. On the other hand, our results also offer an important discriminating feature of potential substrates identified via binding: the substrates important for exocytosis must bind to wild-type and mutant PKC␣, PKC, and PKC␦, as well as to the mitochondrially tethered and PM-tethered mutant PKC␦. Irrelevant binding partners may not bind to all of the PKC forms. Another possible strategy for identifying PKC substrates could rely on enriching phosphoproteins using tools such as immobilized metal affinity columns, followed by identification via mass spectroscopy. While the anti-phospho PKC substrate antibody we used to assess catalytic activity of mutant PKCs works in Western blotting (see e.g. Ref. 14), the antibody detects so many proteins that the result is a smear or ladder of immunoreactivity, rendering one-dimensional blotting uninformative. Two-dimensional blots might be useful. Our results indicate that PKC substrate(s) involved in lytic granule exocytosis will have to be substrates of all of the different versions of the PKC isoforms. This might serve to narrow down what otherwise could be a very long list of candidates.