The discovery that DAG plays an important role in T cell MTOC polarization immediately suggested that proteins containing C1 domains were involved in the process. Of these, perhaps the most obvious candidates were the PKCs. It had been known for some time that a combination of phorbol esters (e.g., PMA) and Ca2+
ionophores (e.g., ionomycin) can largely recapitulate the effects of T cell activation independent of the TCR (Chatila et al., 1989
). These reagents directly activate multiple PKCs, strongly implicating this family of proteins in T cell signaling. Consistent with this notion, PKC inhibitors effectively block many TCR-induced responses, including proliferation and the secretion of inflammatory cytokines (Baier and Wagner, 2009
). Although few studies had implicated PKCs in the regulation of lymphocyte architecture, they were known to play an important role in cytoskeletal remodeling in adherent cells such as fibroblasts (Larsson, 2006
The protein kinase C family is typically divided into three subgroups, which can be distinguished by the structure of their N-terminal regulatory regions (Newton, 2010
). Conventional PKCs (cPKCs) contain tandem, DAG-binding C1 domains followed by a C2 domain, which recognizes negatively charged phospholipids in a Ca2+
-dependent manner. Novel PKCs (nPKCs), by contrast, contain a C2 domain at their N-termini that cannot bind to phospholipids due to mutations in its Ca2+
binding sites. The tandem C1 domains that follow have an unusually high affinity for phorbol esters and DAG. Atypical PKCs (aPKCs) lack C2 domains entirely, and contain only one C1 domain that has lost the ability to bind DAG. These differences in domain structure endow each PKC subfamily with distinct regulatory properties: cPKCs require both Ca2+
and DAG for their activation, nPKCs require DAG alone, while aPKCs are largely regulated through protein–protein interactions.
Identifying which of these isoforms contribute to MTOC polarization responses was complicated by the fact that most, if not all, PKCs are expressed in T cells. The importance of localized DAG, however, argued against a role for aPKCs, at least during the early phases of the response. Furthermore, we had shown that Ca2+
signaling was not required for polarization (Quann et al., 2009
), suggesting that cPKCs were not involved. Hence, we chose to focus first on the nPKC subfamily, comprising PKCδ, PKCε, PKCη, and PKCθ. Of these, probably the best studied was PKCθ, which is highly expressed in both developing and mature T cells. T cells lacking PKCθ display marked deficiencies in antigen-induced proliferation, cytokine secretion, and development into the TH
2 lineage (Sun et al., 2000
; Marsland and Kopf, 2008
). PKCθ is thought to mediate many of these effects by activating several key transcription factors, including NF-κB, NFAT, and AP-1, which together account for a significant fraction of TCR-dependent gene expression (Manicassamy et al., 2006
T cell receptor signaling induces the accumulation of PKCθ at the IS (Monks et al., 1997
), where it would presumably be well positioned to promote cytoskeletal polarization. Prior to our work, however, it was unknown whether PKCθ actually contributed to this process. Even less was known about the other nPKCs. Indeed, previous studies implied that PKCε and PKCη were not required for any aspect of T cell activation (Monks et al., 1997
; Gruber et al., 2005
). Hence, we were quite surprised to find that TCR stimulation in both photoactivation experiments and T cell–APC conjugates induced the robust IS recruitment of not only PKCθ, but also PKCε and PKCη (Quann et al., 2011
). Notably, PKCδ was not recruited in this manner, consistent with previous reports indicating that it localizes to intracellular granules instead (Ma et al., 2008
The synaptic accumulation of PKCε, PKCη, and PKCθ preceded reorientation of the MTOC (Figure
), consistent with a role for all three proteins in the process. Indeed, siRNA-mediated suppression of either PKCθ alone or PKCε and PKCη in combination disrupted polarization responses (Quann et al., 2011
). These results indicated that all three proteins participate in MTOC reorientation, but that PKCε and PKCη can functionally compensate for each other. In retrospect, redundancy between PKCε and PKCη should not have been particularly surprising, given the high level of sequence identity (60%) between the two proteins. This may explain why PKCε-deficient T cells display no observable TCR activation phenotype (Gruber et al., 2005
). A more concrete answer will await the analysis of PKCε/PKCη double knockout mice.
Interestingly, simultaneous siRNA knockdown of PKCε and PKCη also inhibited the recruitment of PKCθ, while knockdown of PKCθ did not affect PKCη accumulation (Quann et al., 2011
). Taken together, these results suggested that PKCε and PKCη operate upstream of PKCθ to promote MTOC polarization. Close examination of the recruitment dynamics of all three proteins was consistent with this hypothesis (Quann et al., 2011
). PKCε and PKCη arrived at the region of TCR stimulation first, followed by PKCθ ~10 s later, and MTOC reorientation 5–10 s after that. PKCε and PKCη had the same accumulation pattern, which covered a broad region of plasma membrane encompassing the entire IS. By contrast, PKCθ occupied a more restricted zone that was entirely contained within the lamellipodial actin ring at the periphery (Figure
). Whether and how these distinct PKC recruitment patterns contribute to polarization responses remains unclear. It is tempting to speculate, however, that the broad accumulation of PKCε and PKCη controls early polarization steps, while the more confined PKCθ distribution contributes to positional refinement of the MTOC at later stages.
We found that the distinct recruitment patterns of PKCη and PKCθ could be largely recapitulated by constructs containing the tandem C1 domains of each protein (Quann et al., 2011
). This is remarkable, given that typical C1 domains are all thought to bind to the same ligand, DAG. What then could explain the differences we observed? In vitro
studies have demonstrated that PKCε binds to bilayers containing DAG with ~10-fold higher affinity than does PKCθ (Stahelin et al., 2005
; Melowic et al., 2007
). It is likely that the affinity of PKCη for DAG is similar to that of PKCε, given the close homology between the two proteins. The ability of PKCε and PKCη to bind DAG more tightly than does PKCθ would presumably lead to faster IS recruitment. A higher affinity for DAG could also explain why PKCε and PKCη accumulate in a broader membrane zone than PKCθ, assuming that DAG density declines radially outside of the site of TCR stimulation.
Although differential DAG affinity provides an elegant mechanism for modulating PKC recruitment, other results strongly suggest that there are additional contributing factors. For example, PKCδ recognizes DAG with threefold higher affinity than does PKCθ (Stahelin et al., 2004
), and yet PKCδ is not recruited to the IS. This probably has less to do with DAG itself and more to do with the complex protein and lipid environment at the IS in which DAG accumulates. The context within which DAG recognition takes place can have dramatic effects on membrane binding by C1 domains. PKCθ, for example, binds to mixtures of DAG and the charged lipid phosphatidylserine with 28-fold higher affinity than it does to DAG mixtures containing the uncharged phosphatidylglycerol (Melowic et al., 2007
). The C1 domains of PKCε have been documented to recognize arachidonic acid and ceramide in addition to DAG (Kashiwagi et al., 2002
), and various C1 domains engage in protein–protein interactions in the appropriate environments (Colon-Gonzalez and Kazanietz, 2006
Hence, it seems reasonable to hypothesize that there are either lipid- or protein-based “contextual factors” that contribute in a combinatorial manner to the accumulation of PKCs within the IS. Although the identity of these molecules remains unknown, the potential to modulate C1 domain localization independent of DAG has important implications for the continued use of C1 domains as DAG “biosensors.” Constructs derived from different PKC isoforms as well as protein kinase D are now widely used to monitor in situ DAG production in multiple experimental systems. Although we have learned and will continue to learn much from this approach, it is important to keep in mind moving forward that different C1 domains have been evolutionarily tuned to recognize DAG within distinct lipid and protein environments. Therefore, care must be taken when imaging C1 domain constructs, as changes in localization could reflect either a change in DAG density or a change in other contextual factors.