|Home | About | Journals | Submit | Contact Us | Français|
Intrinsically disordered cytoplasmic domains of T cell receptor (TCR) signaling subunits including ζcyt and CD3εcyt all contain one or more copies of an immunoreceptor tyrosine-based activation motif (ITAM), tyrosine residues of which are phosphorylated upon receptor triggering. Membrane binding-induced helical folding of ζcyt and CD3εcyt ITAMs is thought to control TCR activation. However, the question whether or not lipid binding of ζcyt and CD3εcyt is necessarily accompanied by a folding transition of ITAMs remains open. In this study, we investigate whether the membrane binding mechanisms of ζcyt and CD3εcyt depend on the membrane model used. Circular dichroic and fluorescence data indicate that binding of ζcyt and CD3εcyt to detergent micelles and unstable vesicles is accompanied by a disorder-to-order transition, whereas upon binding to stable vesicles these proteins remain unfolded. Using electron microscopy and dynamic light scattering, we show that upon protein binding, unstable vesicles fuse and rupture. In contrast, stable vesicles remain intact under these conditions. This suggests different membrane binding modes for ζcyt and CD3εcyt depending on the bilayer stability: 1) coupled binding and folding, and 2) binding without folding. These findings explain the long-standing puzzle in the literature and highlight the importance of the choice of an appropriate membrane model for protein-lipid interactions studies.
Intrinsically disordered proteins (IDPs) are proteins that lack a well-defined three-dimensional structure under physiological conditions . In this context, cytoplasmic regions of signaling subunits of immune receptors, including those of ζ and CD3ε signaling subunits (ζcyt and CD3εcyt, respectively) of T cell receptor (TCR), represent a novel class of IDPs [2–5]. These regions all have one or more copies of an immunoreceptor tyrosine-based activation motif (ITAM) , tyrosine residues of which are phosphorylated upon receptor engagement in an early and obligatory event in the signaling cascade. IDPs are thought to undergo coupled binding and folding upon interaction with their partners . In contrast, random coil ζcyt remains unfolded upon homodimerization [3, 4, 8] or interaction with the well-structured core domain of simian immunodeficiency virus Nef , as shown by circular dichroic (CD) and nuclear magnetic resonance (NMR) spectroscopy. Perhaps even more intriguing is the fact that no chemical shift changes and significant changes in peak intensities are observed in the 1H-15N heteronuclear single quantum coherence spectra of 15N-labeled ζcyt in ζcyt dimer  or ζcyt–Nef complex , thus highlighting unusual biophysical features of this and, possibly, other ITAM-containing proteins. Considering a crucial role of ζcyt and CD3εcyt in TCR signaling and their close proximity to the cell membrane, the question whether or not lipid binding of these IDPs promotes folding of ζcyt and CD3εcyt ITAMs and thus leads to inaccessibility of the ITAM tyrosines for phosphorylation is of fundamental importance. However, little is known about lipid-binding activity of the ITAM-containing cytoplasmic domains and the existing data are contradictory[2, 4, 10].
In 2000 , it has been shown that α-helical folding transition of ζcyt upon binding to acidic phospholipids prevents ITAM phosphorylation. The authors concluded that this folding transition can represent a conformational switch to regulate TCR triggering . Later, the other group of investigators extended these findings to CD3εcyt and showed that binding of this protein to acidic phospholipids is accompanied by folding of ITAM, leading to inaccessibility of the ITAM tyrosines for phosphorylation in vitro . This led the authors [10, 11] to the conclusion that the conformational model of TCR activation previously suggested for ζcyt  can be extended to CD3εcyt. In contrast, we have previously shown that binding of ζcyt and CD3εcyt as well as ITAM-containing cytoplasmic domain of Fc receptor, FcRγcyt, to acidic phospholipids is not accompanied by a disorder-to-order structural transition . This questions the possibility of ITAM folding upon membrane binding in vivo and challenges the suggested model of TCR activation [2, 10, 11].
Lipid bilayers are self-assembled structures, the mechanical properties of which are derived from noncovalent forces such as the hydrophobic effect, steric forces, and electrostatic interactions. In this context, the electrostatic force is of special interest because biological membranes are rich in anionic lipids  and are therefore charged in aqueous solution. Importantly, electrostatic interactions play the critical role in membrane stability . Thus, considering that net charges ofζcyt, CD3εcyt, and FcRγcyt are + 5, + 11, and + 3, respectively, binding of these proteins to acidic phospholipids can potentially destabilize and disrupt lipid bilayers.
In this study, using CD and fluorescence spectroscopy, we demonstrated that α-helical folding ofζcyt, CD3εcyt, and FcRγcyt upon binding to acidic lipids depends on the membrane model structures used (lipid-mimetic detergent micelles vs. lipid vesicles) and the lipid composition of vesicles. For ζcyt, we used synthetic peptides and a lipid-binding assay employing sucrose-loaded large unilamellar vesicles (LUVs) of palmitoyloleoylphosphatidylglycerol (POPG)  to map the lipid-binding region(s) of ζcyt. The assay revealed that not ITAMs but clusters of basic residues outside ITAMs are involved in protein binding to stable acidic phospholipid vesicles. As shown by electron microscopy (EM) and dynamic light scattering (DLS), binding of ζcyt to small unilamellar vesicles (SUVs) and LUVs of dimyristoylphosphatidylglycerol (DMPG) but not POPG promotes vesicle fusion and rupture, highlighting the importance of ensuring the integrity of model membranes upon protein binding.
Peptides A-G (Table 1) were from Sigma. Lysomyristoylphosphatidylglycerol (LMPG), DMPG, and POPG were from Avanti Polar Lipids. Cytoplasmic domains of CD3ε, ζ, and FcRγ were expressed and purified as described previously .
LUVs of DMPG and POPG were made after five freeze-thaw cycles of the hydrated lipids by extruding multilamellar vesicles 10 times through a stack of two polycarbonate filters (100 nm pore diameter) in an Avanti Mini-Extruder (Avanti Polar Lipids). Filters with 30 nm pore diameter were used to prepare extruded SUVs of DMPG and POPG. Alternatively, SUVs were prepared by sonication of the hydrated lipids to clarity in a high intensity bath sonicator (Laboratory Supplies).
Far-UV CD spectra were recorded on an Aviv 202 spectropolarimeter (AVIV Instruments) as described previously . The intrinsic tyrosine fluorescence spectra were taken at 25 °C using a Spex Fluoromax-2 spectrofluorimeter as reported .
Scattering data were collected at 20°C with a DynaPro-MS800 instrument (Protein Solutions) and hydrodynamic radius values were calculated as described previously .
Samples of 3 mM extruded LUVs of POPG and DMPG as well as sonicated SUVs of DMPG alone or in the presence of 10 μM ζcyt were prepared in 20 mM 3-(N-morpholino)propanesulfonic acid (MOPS), 0.1 M NaCl, pH 7.0. Negatively stained grids were prepared by applying 8-μl drops to carbon-coated Formvar 300-mesh gold grids, blotting awayexcess sample after 1 min, then soaking the grid on 6 successive drops of 1% (w/v) uranyl acetate, blotting, and allowing to dry. Micrographs were taken at an initial magnification of 28,500 with a Philips CM10 transmission electron microscope (Philips) operating at 80 kV.
To map the lipid-binding ζcyt sites, we used a sucrose-loaded vesicle-binding assay  and synthetic ζcyt peptides of varying length that cover the entire protein sequence (Fig. 1). The binding data clearly show that the peptides A, C, and E that correspond to the regions outside ITAMs and contain clusters of basic amino acids bind to POPG LUVs while the ITAM peptides (B, D, and F) – do not (Table 1).
These findings experimentally prove our early hypothesis that binding of ζcyt, CD3εcyt, and FcRγcyt to acidic phospholipid LUVs is driven by the clusters of basic amino acid residues rather than the overall net charge . In addition, we specifically show that ITAMs do not contribute to membrane binding by ζcyt. As recently reported , CD3εcyt ITAM is not involved in binding to DMPG LUVs. Thus, taken together, these findings suggest that binding of ITAM-containing ζcyt, CD3εcyt, and highly possibly, FcRγcyt, to acidic phospholipid bilayers involves the protein regions outside ITAMs. This raises an important question: if ITAMs do not contribute substantially in binding, then what determines lipid binding-mediated folding of these functionally important domains and therefore accessibility (or inaccessibility) of the ITAM tyrosines for phosphorylation? To answer this question, we further investigated the induction of secondary structure upon binding of ζcyt, CD3εcyt, and FcRγcyt to acidic micelles and vesicles of different size and composition.
Considering current discrepancies in the literature on whether or not lipid binding of intrinsically disordered ζcyt, CD3εcyt, and FcRγcyt is accompanied by a disorder-to-order transition [2, 4, 10], we hypothesized that membrane binding mode of these proteins can depend on the membrane model used. To test this hypothesis, we examined binding of ζcyt, CD3εcyt, and FcRγcyt to micelles and different vesicles: LMPG micelles, extruded or sonicated DMPG and POPG SUVs as well as LUVs of DMPG and POPG. Among vesicles, POPG LUVs likely represent the best model to mimic the cell membrane because of higher vesicle stability (POPG vs. DMPG) and lower degree of membrane curvature (LUVs vs SUVs).
The CD data clearly show that ζcyt remains unfolded upon binding to POPG SUVs obtained by sonication (Fig. 2a) or extruding (data not shown) and POPG LUVs (Fig. S1a). Similarly, CD studies of CD3εcyt and FcRγcyt do not reveal any detectable secondary structure induction upon binding to POPG SUVs (Figs. S2a and S3a) and LUVs (data not shown). In contrast, the CD spectra of ζcyt taken in the presence of LMPG micelles (Fig. 2a), DMPG SUVs obtained by sonication (Fig. 2a) and extruding (data not shown) as well as DMPG LUVs (Fig. S1a) are typical of proteins with a high helical content. Similar induction of helical conformation is observed for CD3εcyt and FcRγcyt in the presence of DMPG SUVs (Figs. S2a and S3a) and LUVs (data not shown). Thus, in the lipid systems studied, induction of helical conformation of intrinsically disordered ζcyt, CD3εcyt and FcRγcyt upon lipid binding depends on lipid structure (micelles vs. vesicles) and composition (DMPG vs. POPG) but not vesicle size and preparation technique. In addition, for each protein studied, the percentage of helicity induced by LMPG micelles is similar to that by DMPG vesicles, providing indirect evidence for similar mode of protein-lipid interactions in these systems.
Interestingly, in 1998 , using CD and infrared spectroscopy, the ζcyt ITAM peptides have been reported to adopt an α-helical conformation in the presence of trifluoroethanol (TFE). In this study, the ITAM1 peptide contained the smallest percentage of a-helix in TFE, while the ITAM2 and ITAM3 peptides were predominantly helical. Later, similar helical conformational changes of ζcyt ITAMs were observed in NMR studies of full-length ζcyt in the presence of LMPG micelles , showing that micelles and TFE induce similar helical folding of ITAMs but not other domains. Results of our titration studies (not shown) are similar to previously reported data  and indicate that the percentage of helicity reaches a maximal value of ~35–40% at 1:50 ζcyt/LMPG molar ratio. Considering an LMPG micelle aggregation number of approximately 55 detergent molecules, this protein/detergent ratio corresponds to one molecule per micelle. In summary, considering also the results of our mapping studies (Table 1 and Fig. 1), these findings suggest that in contrast to lipid binding that is mediated by clusters of basic amino acids outsideζcyt ITAMs, helical folding occurs within ITAMs and can be induced by physiologically irrelevant agents such as TFE. Similarly, it has been reported recently for CD3εcyt that basic residues outside ITAM contribute to binding of this protein to DMPG LUVs whereas helical folding is induced within ITAM .
Our fluorescence studies demonstrate that changes in intrinsic tyrosine fluorescence accompany binding of ζcyt, CD3εcyt, and FcRγcyt to all membrane models studied: LMPG micelles, DMPG and POPG SUVs and LUVs (Fig. 2b, and Figs. S1b, S2b, and S3b). However, quantitative analysis reveals that fluorescence changes are substantially larger in the presence of LMPG micelles and DMPG vesicles as compare to POPG vesicles (independent on vesicle size and method of preparation). A reasonable molecular explanation for this phenomenon is that ITAMs of ζcyt, CD3εcyt, and FcRγcyt all contain two tyrosines  that upon helical folding of ITAMs, can associate with LMPG and DMPG molecules, resulting in a substantial increase in tyrosine fluorescence. In contrast, binding to POPG that is mediated by the regions outside ITAMs and is not accompanied by a disorder-to-order transition can affect the fluorescence intensity of only non-ITAM tyrosines (1 in ζcyt, 1 in FcRγcyt, and none in CD3εcyt), thus changing an overall tyrosine fluorescence in much less extent. This is well supported by the data obtained (Fig. 2b, and Figs. S1b, S2b, and S3b). In addition, as with the CD data, for each protein studied, the amplitude of the fluorescence changes induced by LMPG micelles is similar to that promoted by DMPG vesicles, suggesting further evidence of similar mode of protein-lipid interactions in these systems.
Thus, the CD and fluorescence results suggest the existence of two different membrane binding modes of intrinsically disordered ζcyt, CD3εcyt, and FcRγcyt, depending on the cell membrane model: binding followed by helical folding of the ITAMs (LMPG micelles and DMPG vesicles) and binding without a disorder-to-order transition (POPG vesicles). Importantly, in both modes, initial binding of these proteins to acidic phospholipids is driven by electrostatic interactions between polar heads and clusters of positively charged amino acids outside ITAMs. In contrast, formation of helices within ITAMs in the presence of LMPG micelles and DMPG vesicles is likely mediated by hydrophobic interactions of these domains with detergent and lipid tails. The observed difference between DMPG and POPG vesicles (independent on vesicle size and preparation procedure) in their ability to induce helical structure formation led us to suggest that binding of ζcyt, CD3εcyt, and FcRγcyt to phospholipid vesicles may or may not affect the bilayer integrity. To test this hypothesis, we used DLS and EM to investigate lipid-binding activity of ζcyt as related to vesicle stability.
Cationic peptides are well known to induce aggregation of negatively charged lipid vesicles . Thus, it is important to assess acidic phospholipid vesicle stability upon binding to positively chargedζcyt, CD3εcyt, and FcRγcyt. DLS is a rapid and nondestructive method to measure vesicle hydrodynamic radius (RH) and size distribution. Our previous DLS studies demonstrated that binding of ζcyt, CD3εcyt, and FcRγcyt does not disturb the lipid bilayer structure and integrity of the sucrose-loaded POPG LUVs used . Here, we describe the results of DLS measurements of phospholipid vesicles of different size and composition in the presence or absence of ζcyt. Importantly, in DLS studies, particle size distributions can be displayed as the percent mass distributions that show how much mass of that particular peak would explain the correlation function or as the percent scattering intensity distributions that reflect the actual contribution of signal to the measurement. Graphically, mass- and intensity-weighted size distributions can be significantly different due to the exponentially stronger scattering from large particles. Thus, conversion of the percent scattering distributions into percent mass distributions allows to correct for the dominance of larger aggregates.
The DLS data do not reveal any substantial changes in the mass- and intensity-weighted size distributions of POPG LUVs and SUVs upon addingζcyt, suggesting that protein binding does not disturb the lipid bilayer structure in these particles and does not cause vesicle fusion (Figs. 2c and 2d). For POPG SUVs, similar results were obtained by using the vesicles prepared by sonication (Fig. 2d) or extruding (Fig. S1c), demonstrating the lack of any significant dependence on the method of vesicle preparation. In contrast, binding of ζcyt to DMPG LUVs and SUVs leads to vesicle fusion and formation of large vesicle aggregates (Figs. 2c and 2d). The results for DMPG SUVs do not depend on the method of preparation, and the vesicles prepared by sonication or extruding fuse and aggregate similarly in the presence of ζcyt (Fig. 2d and Fig. S1c). Interestingly, adding of ζcyt to LMPG micelles results in narrowing the initial RH distributions (Fig. S1d). This can be explained by formation of relatively uniform micelles due to structuring and stabilizing effect of ζcyt. According to the results of our CD titration studies, these micelles contain one protein molecule per particle.
To further investigate structure and morphology changes of vesicles of different size and composition upon adding ζcyt, we used negative staining transmission EM. Our EM studies clearly show that protein binding-dependent membrane disruption is observed for DMPG LUVs and SUVs (Fig. 3a). In contrast, POPG LUVs (Fig. 3a) and SUVs (not shown) remain intact upon binding to ζcyt. These results provide direct evidence that interactions of a positively charged ζcyt with acidic phospholipids can disturb membrane integrity and that this effect depends on vesicle composition rather than size.
In summary, the DLS and EM data clearly demonstrate that binding to ζcyt induces membrane fusion and rupture in DMPG vesicles. Importantly, the destructive effect of protein binding on bilayer lipid membrane is not dependent on vesicle size (SUV vs. LUV) or technique of SUV preparation (sonication vs. extruding). The membrane rupture is known to result in monolayer fusion of the membranes, i.e., in the formation of a bridge connecting the monolayers, which is usually named the stalk or hemifusion intermediate . Interestingly, tight coupling between the loop-to-helix structural transition and stalk formation as a result of deformation of the target and viral membranes has been reported for influenza hemagglutinin [19, 20]. Thus, protein binding-induced membrane perturbation and disruption can represent a molecular basis for formation of the ITAM helixes in the presence of DMPG vesicles.
Here we analyzed membrane binding of intrinsically disorderedζcyt, CD3εcyt, and FcRγcyt, and found that there are two different modes of their lipid-binding activity toward acidic phospholipids (Figs. 3b and S2c): 1) coupled binding and folding that is characteristic for micelles and those vesicles that are unstable upon protein binding (DMPG), and 2) binding without folding that is observed in the presence of stable vesicles (POPG). As we suggest, in both modes, initially, clusters of basic amino acids in the regions outside ITAMs bind to polar heads of acidic phospholipids while the ITAM residues do not contribute to binding at this stage. Then, in micelles (mode I), hydrophobic interactions between ITAMs and detergent tails promote folding of ITAMs, thus making ITAM tyrosines inaccessible for kinases as it has been shown for ζcyt/LMPG micelles system [2, 16]. In vesicles, depending on the bilayer stability, initial protein binding to the membrane may (mode I) or may not (mode II) induce vesicle fusion and rupture and promote formation of ITAM helixes stabilized by hydrophobic interactions with lipid tails in ruptured bilayers.
Instability of the bilayers of DMPG LUVs or POPG/dihexanoylphosphatidylcholine bicelles upon binding to CD3εcyt (net charge of +11) can explainα-helical folding of CD3εcyt ITAM (mode I) and inaccessibility of the ITAM tyrosines for kinases observed in the presence of these membrane models . It should be also noted that in mode II, ITAMs do not participate in binding to lipid bilayers and the ITAM tyrosines are therefore likely to be easily accessible for phosphorylation (Fig. 3b and Fig. S2c).
The existence of two different modes of membrane binding raises an important question: which mode of action is of physiological relevance? Considering that folding of ITAMs is not observed in the presence of those vesicles that are stable upon protein binding, this folding transition is unlikely to play a significant role in transmembrane signaling and cell activation. However, it does not necessarily mean that mode II (binding without folding) is also physiologically irrelevant. For example, in a recently proposed novel model of immune signaling, the signaling chain homooligomerization (SCHOOL) model [21, 22], homointeractions between cytoplasmic domains of the ITAM-containing receptor signaling subunits are suggested to be critical in receptor-mediated cell activation. Thus, membrane binding of ζcyt, CD3εcyt, and FcRγcyt might prevent homooligomerization of these cytoplasmic domains  in receptor clusters on the surface of resting cells and during random encounters of receptors diffusing in the cell membrane.
In summary, our data first provide molecular explanation for current discrepancies in the literature on whether or not lipid binding of ζcyt, CD3εcyt, and FcRγcyt is accompanied by helical folding of ITAMs [2, 4, 10]. Second, we show that the use of lipid vesicles of the same size and surface charge can result in opposite conclusions regarding membrane-binding activity of proteins and its physiological relevance. The questions raised here highlight the importance of the choice of an appropriate membrane model and how substantially our improved understanding of important biological processes such as receptor triggering and cell activation depends upon critical evaluation of the data and observations accumulated to date.
The authors thank Dr. Dikran Aivazian for his assistance in the initial phases of this work. This study was supported in part by the NIH/NIAID, University of Massachusetts Center for AIDS Research (project P30 AI42845-08, to A.B.S.).