GFP-tagged First Cys-Domains of PKC-γ Expressed in RBL Cells by RNA Transfection
The conventional isoforms of PKC: PKC-α, PKC-β1, PKC-β2, and PKC-γ (cPKCs), each contain two Cys-domains in their regulatory region (Fig. A
). For PKC-γ, previous in vitro studies have shown that fusion constructs of glutathione S transferase (GST) with either the first or second Cys-domain bind to lipid vesicles in the presence of phorbol ester (i.e., Quest and Bell, 1994
). To investigate the functions of diacylglycerol and phorbol ester in intact cells, we tagged the first Cys-domain of PKC-γ with GFP. In vitro translation showed that the protein encoded by the construct (Cys1–GFP) has the expected molecular mass (Fig. B
) and can bind to lipid vesicles in the presence of phorbol ester (Fig. C
). When a conserved proline residue within the Cys-domain (Pro 46) was replaced by a glycine residue (mCys1–GFP), the Cys-domain showed a markedly reduced ability to bind lipid vesicles in the presence of phorbol ester (Fig. C
). Such an important function of the conserved proline residue in phorbol ester binding has been predicted by a sequence comparison of all Cys-domains that bind phorbol ester in vitro (Kazanietz et al., 1994
Figure 1 Receptor-mediated plasma membrane translocation of the GFP-tagged first Cys-domain from PKC-γ. (A) Schematic representation of the domain composition of conventional PKC and of the Cys1–GFP fusion construct used in the experiments. The (more ...)
Cys1–GFP (Fig. , D
), mCys1–GFP (Fig. F
), or GFP alone (Fig. G
) were expressed in RBL cells by transfection of in vitro transcribed and polyadenylated mRNA using a microporation device (Yokoe and Meyer, 1996
; Teruel and Meyer, 1997
). Measurements of the translocation of GFP-tagged Cys-domains were typically made within 3–12 h after RNA transfection.
Receptor-mediated Transient Translocation of Cys1–GFP from the Cytosol to the Plasma Membrane
When expressed in unstimulated cells, Cys1–GFP appeared homogeneously distributed across the cytosol and nucleus (Fig. , D
, left column
). To test the response to cell stimulation, two alternative receptor pathways were used to activate either PLC-γ or PLC-β. Cross-linking of the IgE receptors (FcεRI) in RBL cells by addition of DNP-BSA (Fig. D
) has been shown to activate PLC-γ1 (Schneider et al., 1992
), whereas stimulation through transfected PAF receptor (Fig. E
) has been shown to activate PLC-β (Ali et al., 1995
). In response to both receptor stimuli, Cys1–GFP translocated from the cytosol to the plasma membrane within <1 min (the middle images were taken after 1 min) and dissociated from the membrane after several minutes (the right images were taken after 5 min). The striking dynamics of this transient translocation process can be more clearly visualized in a movie of DNP-BSA–stimulated RBL cells (the movie can be viewed at http://note.cellbio.duke.edu/Faculty/~Meyer/PKC)
. As can be seen in Fig. , D
), a variable smaller fraction of the Cys1–GFP did not participate in the translocation process but remained localized in the nucleoplasm and to a lesser extent in the cytoplasm.
When the same two stimulation protocols were applied to cells expressing the proline-mutated Cys1–GFP or GFP alone, no significant changes in the localization of mCys1– GFP and GFP could be observed (Fig. , F and G show the response to PAF activation). This suggests that the plasma membrane translocation of Cys1–GFP is likely mediated by the phorbol ester binding site of the Cys-domain.
Overall, these results demonstrate that an individual Cys-domain can act as a plasma membrane–targeting module in response to stimuli that activate either PLC-β or PLC-γ. In addition, these results show that the Cys- domain lacking the conserved proline residue is largely ineffective as a translocation module.
To determine the time course of the transient translocation of the Cys-domain to the plasma membrane more quantitatively, time series of confocal fluorescence images of cells expressing Cys1–GFP domains were recorded before and after receptor stimulation (Fig. A
). The relative change in the plasma membrane fluorescence intensity was determined in each image using line intensity profiles across each one of the cells (Fig. B
). The relative increase in the plasma membrane versus the cytosolic fluorescence intensity was calculated by measuring the amplitude of the fluorescence signal at the plasma membrane and dividing it by the average intracellular fluorescence intensity (Imb
. Fig. C
shows the average intensity ratio for each time point as a function of time. The time course for translocation significantly varied between cells, and a few cells did not exhibit a measurable translocation. This variability between cells is reminiscent of the cell-to-cell variability observed for IgE receptor–mediated calcium signaling (i.e., Millard et al., 1988
Figure 2 Comparison of the time course of plasma membrane translocation of Cys1–GFP in response to activation of IgE or PAF receptors. (A) Sequential images of RBL cells expressing Cys1–GFP taken immediately before and 40, 80, and 200 s after (more ...)
Whereas the activation of both receptor types induced a near uniform association of Cys1–GFP with the plasma membrane, the time course of translocation differed depending on whether cells were stimulated by activating IgE or PAF receptors. The beginning of Cys1–GFP translocation was typically delayed by 30 s after IgE receptor activation but started immediately after PAF receptor activation. The same difference in the delay time between PAF and IgE receptor–induced activation is also observed for the induction of calcium spikes after activation of the two receptors (data not shown). Taken together, these observations are consistent with the hypothesis that the initial membrane translocation of Cys1–GFP is mediated by the phospholipase C–mediated production of diacylglycerol.
PMA and DiC8 Mimic the Receptor-induced Plasma Membrane Translocation of Cys-Domains
As discussed in the introduction, PMA can potently activate cPKCs by directly binding to their Cys-domains. Therefore, we tested the effect of PMA on expressed Cys1–GFP. In response to extracellular addition of PMA, Cys1–GFP translocated from the cytosol to the plasma membrane (Fig. A). The left panel in this figure shows a differential interference contrast (DIC) image of a group of RBL cells, the middle panel shows a confocal fluorescence image of the initially homogenous distribution of Cys1–GFP and the right panel shows the plasma membrane distribution of Cys1–GFP 5 min after PMA addition. As a control for the specificity of PMA-induced plasma membrane translocation, the addition of the bioinactive 4α isomer of PMA, instead of the bioactive 4β isomer, did not translocate Cys1–GFP to the plasma membrane (data not shown).
Figure 3 Translocation of Cys1–GFP in response to the addition of PMA or DiC8. Cells expressing Cys1–GFP were stimulated with either 1 μM PMA (A) or 100 μg/ml DiC8 (B). The left panels show DIC images of the cells before stimulation. (more ...)
In addition to PMA, the extracellular addition of 1,2- dioctanoyl sn
-glycerol (DiC8), a diacylglycerol analog with short fatty acid chains, also induced translocation of Cys1– GFP to the plasma membrane (Fig. B
). Again, the left panel shows a DIC image, the middle panel the Cys1–GFP distribution in unstimulated cells and the right panel the Cys1–GFP distribution after DiC8 addition. A similar translocation of Cys-domains to the plasma membrane was observed after addition of extracellular bacterial phosphatidylcholine phospholipase C (PC-PLC; data not shown). This enzyme generates diacylglycerol by cleaving phosphatidylcholine in the outer leaflet of the plasma membrane, with diacylglycerol exerting its biological function at the inner leaflet by randomization (Besterman et al., 1986
Interestingly, addition of PMA to the Cys-domain mutated on the conserved proline residue led to a smaller but measurable translocation to the plasma membrane (Fig. C). This residual plasma membrane translocation is consistent with the small phorbol ester mediated vesicle binding shown in Fig. D.
A quantitative analysis of the concentration dependence of the plasma membrane translocation in response to PMA and DiC8 showed that half-maximal translocation occurred at 40 nM of PMA and 10 μg/ml of DiC8 (Figs. , D and E, respectively). Interestingly, an analysis of the kinetics of translocation showed that PMA-mediated translocation is much slower than that mediated by DiC8. While PMA-mediated translocation was half-maximal after ~60 s (n = 14; Fig. F), translocation in response to DiC8 only required ~6 s (n = 12; Fig. G).
How Tight Are the Plasma Membrane Binding Interactions of Cys1–GFP?
Whereas PMA, DiC8, and bacterial PC-PLC addition all led to a similar translocation of Cys1–GFP to the plasma membrane, photobleaching recovery measurements suggested that the dissociation time and the diffusion coefficient for the membrane-associated Cys1–GFP was markedly different for the three stimuli (Fig. ). In these experiments, a small spot of plasma membrane localized Cys1–GFP was photobleached by a short laser pulse, and the recovery of fluorescence was monitored as a function of time using sequential imaging (Fig. A shows an example of cells stimulated with PC-PLC). The plasma membrane–bound Cys1– GFP had recovery times of one second in cells stimulated by addition of PC-PLC or DiC8 (see Table ). In contrast, the recovery time after PMA-induced localization was typically 10 s and a variable fraction of the membrane-associated Cys1–GFP was completely immobile (examples of the recovery curves are shown in Fig. B).
Figure 4 Comparison of the apparent lateral membrane diffusion coefficient and apparent plasma membrane dissociation time of Cys1–GFP in response to the addition of PMA, PC-PLC, or DiC8. Fluorescence recovery after photobleaching was used to determine (more ...)
Apparent Plasma Membrane Diffusion Coefficients and Dissociation Times of Cys1–GFP
We separated a lateral membrane diffusion component of the recovery process from a membrane dissociation component by measuring one-dimensional line intensity profiles along the plasma membrane in each of a series of images. Since the laser used for photobleaching had an approximately Gaussian bleach profile, we fit the profiles in each of the images by Gaussian functions (Fig. C; see Materials and Methods section). During the recovery process, the replacement of bleached Cys1–GFP with unbleached Cys1– GFP due to membrane diffusion is expected to lead to a widening of the Gaussian bleach profile, whereas the dissociation of bleached Cys1–GFP would not widen the bleach profile. Thus, the widening of the bleach profile as a function of time can be used to determine an apparent lateral diffusion coefficient of Cys1–GFP within the plasma membrane (Fig. D). For PMA, an apparent membrane diffusion coefficient of D = 0.14 ± 0.04 μm2/s (n = 17) was calculated, whereas Cys1–GFP bound to the plasma membrane in response to DiC8 and PLC addition had a much faster apparent membrane diffusion coefficient: D = 0.97 ± 0.14 μm2/s (n = 18) for DiC8 and D = 1.19 ± 0.19 μm2/s (n = 15) for PC-PLC (see also Table ).
In a second analysis of the recovery process, an apparent dissociation time constant of Cys1–GFP from the plasma membrane was determined by subtracting the recovery component that results from membrane diffusion (see Materials and Methods). This analysis shows that Cys1–GFP bound to the plasma membrane by PMA has an apparent dissociation time of 98.6 ± 19 s (n = 17), whereas Cys1– GFP localized by DiC8 and by externally added PC-PLC has apparent dissociation times of 8.0 ± 1.6 s (n = 18) and 3.5 ± 0.5 s (n = 15), respectively (Fig. E and Table ). It should be noted that the values for diffusion coefficients and dissociation times obtained in these analysis procedures can be affected by the particular cell geometry and are most useful as a means to compare membrane binding interactions within the same cell type.
Overall, this analysis suggests that Cys1–GFP is reversibly bound to the plasma membrane in response to increases in diacylglycerol concentration and can diffuse rapidly within the plasma membrane. However, the same Cys1–GFP probe not only has a much slower dissociation time in the presence of PMA but also shows a markedly reduced lateral membrane diffusion coefficient. As an additional result, these measurements suggest that short chain diacylglycerols (DiC8) are more effective in binding Cys1–GFP to the plasma membrane than diacylglycerol produced by extracellular addition of PC-PLC.
Identification of the Nuclear Membrane as a Second Target for Cys1–GFP
The studies described above have shown that the plasma membrane is a primary target of Cys1–GFP in response to IgE and PAF receptor–mediated production of diacylglycerol or addition of PMA, DiC8, or extracellular PC-PLC. PDBu is a smaller and less hydrophobic phorbol ester analogue than PMA that is expected to equilibrate more rapidly between the plasma membrane and internal membranes. These properties make PDBu an ideal tool to identify potential other intracellular membranes as targets for Cys-domains. Strikingly, extracellular addition of PDBu led to a rapid translocation of Cys1–GFP to the plasma as well as nuclear membranes (Fig. A
). The left panel shows the distribution of Cys1–GFP before, the middle image shows the distribution 1 min after, and the right panel shows the distribution 10 min after PDBu addition. A nuclear membrane localization of Cys1–GFP can be seen in the middle image. In RBL cells, the nucleus is typically bean shaped with distinct membrane invaginations (i.e., Subramanian and Meyer, 1997
). Confocal analysis of a large number of cells showed that significantly less fluorescence was associated with other cytosolic membranes than with the nuclear or plasma membrane, suggesting that nuclear and plasma membrane are preferential targets for Cys1–GFP. Interestingly, several minutes after PDBu addition, the association of Cys1–GFP with the nuclear membrane was significantly reduced (i.e., Fig. A
Figure 5 Addition of PDBu identifies the nuclear membrane as a selective target for Cys1–GFP translocation. (A) Series of three confocal fluorescence images of RBL cells expressing Cys1–GFP and stimulated by addition of PDBu (1 μM). The (more ...)
An analysis of the concentration dependence of translocation showed that 30 nM PDBu induced half-maximal plasma membrane translocation, whereas 400 nM was required for half-maximal nuclear membrane translocation (Fig. B). The time course of membrane translocation to the plasma membrane was as rapid as the one observed above for DiC8 (Fig. C). For the nuclear membrane, a rapid nuclear translocation was typically followed by a slower reduction in nuclear membrane staining.
Photobleaching recovery measurements showed that the half-maximal recovery time was faster for nuclear membrane–bound Cys-domains compared with plasma membrane-bound Cys-domains (Fig. D). While the apparent diffusion coefficients in the plasma and nuclear membrane (Fig. E) were similar: D = 0.25 ± 0.02 μm2/s (n = 10) and D = 0.34 ± 0.05 μm2/s (n = 10), the apparent dissociation time at the plasma membrane (Fig. F) was markedly slower than the one at the nuclear membrane: 35.9 ± 0.5 s for the plasma membrane dissociation time compared with 12.9 ± 3.6 s for the nuclear one (see also Table for a comparison of the plasma membrane parameters). Thus, the titration with PDBu and the photobleaching recovery analysis both suggest that the affinity of Cys1– GFP for the nuclear membrane is lower than that for the plasma membrane. It is therefore conceivable that reversibly bound nuclear constructs will slowly diffuse away and bind to the plasma membrane, explaining the transient association of Cys1–GFP with the nuclear membrane.
Addition of PDBu was different from addition of PMA in that the concentration of initially homogeneously distributed nuclear Cys1–GFP rapidly decreased in parallel with an increase in the nuclear membrane staining (on a time scale of 10 s; Fig. C). Thus, it is likely that the nuclear Cys1–GFP rapidly binds to the inner nuclear membrane after PDBu addition. The subsequent slow reduction in nuclear membrane staining may be a result of equilibration, since the nuclear membrane has likely a lower affinity than the plasma membrane.
Arachidonic Acid Prevents the Diacylglycerol-mediated Plasma Membrane Translocation of Cys-Domains
Previous studies have suggested that different protein kinase C isoforms can also be regulated by other lipid second messengers such as ceramide and free fatty acids. To investigate a potential effect of ceramide and free fatty acids on Cys-domains, the relative increase in plasma membrane fluorescence of Cys1–GFP was measured after addition of ceramide or free fatty acid. Using the analysis procedure shown in Fig. B, DiC8 and PMA were found to induce a relative increase of plasma membrane fluorescence of ~100 and 240%, respectively (Fig. A). In contrast, ceramide, oleic acid, and arachidonic acid had little or no effect on the translocation of Cys1–GFP to the plasma membrane.
To investigate whether these messengers can suppress or enhance diacylglycerol induced localization of Cys1– GFP, different lipid second messengers were added 5 min before diacylglycerol addition. Ceramide and oleic acid had no significant effect on the diacylglycerol induced translocation of the probe. In contrast, 100 μM arachidonic acid almost completely abolished the ability of DiC8 to localize Cys1–GFP to the plasma membrane (Fig. B).
Confocal imaging analysis of cells stimulated by arachidonic acid showed that this fatty acid induces an initial association of the Cys1–GFP probe with internal structures (Fig. C). This is at least suggested from a less homogenous and punctuate distribution of Cys1–GFP after arachidonic acid addition. After the subsequent addition of DiC8, the probe became partially nuclear excluded and more prominently associated with nonuniform cytosolic structures. This suggests that Cys-domains are localized to internal membranes or other targets in response to a combined increase in diacylglycerol and arachidonic acid concentration.
To further test this hypothesis, we measured the mobility of Cys1–GFP in the cytosol before and after arachidonic acid and diacylglycerol addition (Fig. D). Photobleaching recovery was used to determine the diffusion coefficient of Cys1–GFP in the cytosol. The apparent diffusion coefficient of the Cys1–GFP probe in the cytosol before stimulation was D = 7.0 ± 1.5 μm2/s. Addition of arachidonic acid (100 μM) led to a decrease of the diffusion coefficient to D = 4.2 ± 0.8 μm2/s. Addition of DiC8 to the arachidonic acid–treated cells led to a further decrease in the mobility of the probe in the cytosol to D = 2.9 ± 1.1 μm2/s. These results suggest that arachidonic acid enhances the binding interaction of Cys-domains with membranes and reduces the preferential plasma membrane localization of Cys1–GFP in response to diacylglycerol increases.
Receptor and PMA-induced Plasma Membrane Translocation of GFP-tagged Cys2-Domains, Cys1Cys2 Tandem Domains and Full-Length PKC-γ
In vitro studies have shown that the second Cys-domain of PKC-γ, the Cys1Cys2 tandem domains and the full-length PKC are also phorbol ester sensitive (Quest et al., 1994). Therefore, we tested whether these constructs can also be used as GFP-tagged probes to study diacylglycerol-mediated signal transduction (Fig. , A and B). As for the Cys1– GFP, the GFP-fusion constructs were expressed in RBL cells by RNA transfection (Fig. , C to E, left images). The expressed Cys2–GFP and Cys1Cys2–GFP were uniformly expressed in the cytosol and were in most cells enriched in the nucleoplasm (Fig. , C and D, left images). The expressed GFP-tagged full-length PKC-γ was largely cytosolically localized (Fig. E, left image).
Figure 7 Translocation of Cys2–GFP, Cys1Cys2–GFP, and full-length PKC-γ–GFP to the plasma membrane in response to receptor activation. (A) Schematic representation of the GFP-tagged constructs used in these experiments: Cys2–GFP, (more ...)
Activation of the PAF receptor led to a marked translocation of cytosolically localized Cys1Cys2–GFP and PKC-γ–GFP to the plasma membrane. A much smaller or no plasma membrane translocation was observed for Cys2– GFP. In Fig. C, the cellular redistribution of the Cys2– GFP is shown 1 and 5 min after a maximal stimulation of PAF receptors. Fig. , D and E, show the plasma membrane translocation of Cys1Cys2–GFP and PKC-γ–GFP, respectively. Similar transient translocation events were observed for Cys1Cys2–GFP and PKC-γ–GFP after activation of IgE receptors, while no significant membrane translocation was observed for Cys2–GFP in response to IgE receptor activation (data not shown).
Nevertheless, in vitro translated Cys2–GFP, Cys1Cys2– GFP and PKC-γ–GFP all bound lipid vesicles in a phorbol ester-dependent manner (Fig. A). Furthermore, all three constructs showed marked plasma membrane translocation in response to PMA (Fig. , B–D). Only the initially nuclear prelocalized Cys2–GFP and Cys1Cys2–GFP molecules was not significantly affected by the addition of PMA. In contrast, addition of PDBu induced a translocation of Cys2–GFP and Cys1Cys2–GFP to the plasma as well as nuclear membrane (data not shown, similar observations were made for Cys1–GFP in Fig. ). No significant nuclear membrane localization of PKC-γ–GFP was observed after addition of PDBu, possibly because the full-length PKC-γ was largely nuclear excluded. While these measurements give additional insights into the function of the two Cys-domains in the context of the plasma membrane translocation of PKC-γ holoenzyme, they also suggest that the Cys1–GFP probe is better suited as a fluorescent indicator for studying diacylglycerol signaling.
Figure 8 Phorbol ester sensitivity of Cys2–GFP, Cys1Cys2–GFP and full-length PKC-γ–GFP. (A) In vitro binding of Cys2–GFP, Cys1Cys2–GFP, and PKC-γ–GFP to lipid vesicles in the presence of phorbol (more ...)