Analysis of transient phosphorylation-dependent protein-protein interactions in living mammalian cells using split-TEV
© Wehr et al. 2008
Received: 24 May 2007
Accepted: 13 July 2008
Published: 13 July 2008
Skip to main content
© Wehr et al. 2008
Received: 24 May 2007
Accepted: 13 July 2008
Published: 13 July 2008
Regulated protein-protein interactions (PPIs) are pivotal molecular switches that are important for the regulation of signaling processes within eukaryotic cells. Cellular signaling is altered in various disease conditions and offers interesting options for pharmacological interventions. Constitutive PPIs are usually mediated by large interaction domains. In contrast, stimulus-regulated PPIs often depend on small post-translational modifications and are thus better suited targets for drug development. However, the detection of modification-dependent PPIs with biochemical methods still remains a labour- and material-intensive task, and many pivotal PPIs that are potentially suited for pharmacological intervention most likely remain to be identified. The availability of methods to easily identify and quantify stimulus-dependent, potentially also transient interaction events, is therefore essential. The assays should be applicable to intact mammalian cells, optimally also to primary cells in culture.
In this study, we adapted the split-TEV system to quantify phosphorylation-dependent and transient PPIs that occur at the membrane and in the cytosol of living mammalian cells. Split-TEV is based on a PPI-induced functional complementation of two inactive TEV protease fragments fused to interaction partners of choice. Genetically encoded transcription-coupled and proteolysis-only TEV reporter systems were used to convert the TEV activity into an easily quantifiable readout. We measured the phosphorylation-dependent interaction between the pro-apoptotic protein Bad and the adapter proteins 14-3-3ε and ζ in NIH-3T3 fibroblasts and in primary cultured neurons. Using split-TEV assays, we show that Bad specifically interacts with 14-3-3 isoforms when phosphorylated by protein kinase Akt-1/PKB at Ser136. We also measured the phosphorylation-dependent Bad/14-3-3 interactions mediated by endogenous and transient Akt-1 activity. We furthermore applied split-TEV assays to measure the phosphorylation-dependent interactions of Neuregulin-1-stimulated ErbB4 receptors with several adapter proteins.
Split-TEV assays are well suited to measure phosphorylation-dependent and transient PPIs that occur specifically at the membrane and in the cytosol of heterologous and primary cultured mammalian cells. Given the high sensitivity of the split-TEV system, all assays were performed in multi-plate formats and could be adapted for higher throughput to screen for pharmacologically active substances.
Constitutive and regulated PPIs are the main organizing principles within signaling cascades the integration of which results in an adaptive cellular behaviour. Modification-dependent PPIs are often positioned at pivotal positions within signaling pathways, and are thus central to signaling processes at the membrane and in the cytosol of living mammalian cells . Phosphorylation of specific serine or threonine residues by kinases represents the prototype and most abundant type of post-translational protein modifications . In light of the fact that cellular signaling is altered in many disease conditions, functional subunits of signaling processes are the focus of intense research, since they represent attractive targets for pharmacological intervention . In contrast to constitutive PPIs, stimulus-regulated PPIs often depend on small post-translational modifications, and are thus better suited targets for drug development . However, the detection of modification-dependent PPIs with biochemical methods still remains a labour- and material-intensive task, and many pivotal PPIs potentially suited for pharmacological perturbation most likely still remain to be identified. Therefore, the availability of methods to easily screen and identify stimulus-dependent, potentially transient, interaction events is essential. Ideally, the assays should be applicable to intact mammalian cells, including cultured primary cells.
Recently, we reported the development of the split-TEV approach that allowed us to monitor the ligand-induced dimerisation of ErbB receptors at the membrane of mammalian cells . Split-TEV is based on the functional complementation of two inactive TEV protease fragments fused to interacting proteins. The PPI-dependent TEV protease activity can be followed by several reporters, which either rely on a fluorescent or a luminescent readout . In this study, we wanted to adapt the split-TEV system to analyse constitutive and phosphorylation-dependent interactions of full-length proteins that occur in the cytosol and at the membrane. For the technical proof-of-principle for cytosolic interactions, we chose the interactions between Bad and 14-3-3 isoforms as a model system [6, 7]. Both, the 14-3-3 and Bad proteins are involved in the regulation of apoptosis and survival signaling [8, 9]. Bad is a pro-apoptotic protein exerting its action by binding to the anti-apoptotic, mitochondrially localised proteins Bcl-XL and Bcl2, thereby inactivating the Bcl proteins . However, upon phosphorylation at serine 136 by protein kinase Akt-1/PKB, Bad can be complexed by 14-3-3 proteins in the cytosol, thus preventing the association with the Bcl proteins and inhibiting apoptosis [6, 11]. 14-3-3 proteins were shown to be involved in sequestering functions through binding to phoshorylated proteins and consequently influencing signaling events [9, 11]. There are seven 14-3-3 genes giving rise to the seven isoforms β, γ, ε, η, σ, τ (or θ) and ζ. The 14-3-3 isoforms can functionally compensate for each other, but can also mediate specific cellular functions: the σ isoform for example is implicated in cancer and cell cycle regulation [9, 12], whereas the isoforms ε and ζ are highly expressed in postmitotic cells of the brain . Additionally, 14-3-3 proteins can homo- and heterodimerise .
To demonstrate the applicability of the split-TEV system to analyse phosphorylation-dependent interactions at the membrane, we chose stimulus-dependent interactions of the ErbB4 receptor with various cytosolic adapter proteins. ErbB4 belongs to the family of ErbB receptor tyrosine kinases, which are involved in diverse signaling mechanisms ranging from proliferation to differentiation and neuronal specification [14–16]. Upon ligand binding, ErbB4 homo- or heterodimerises, followed by an autophosphorylation in trans, which then leads to the recruitment of SH2 domain-containing adaptor proteins, such as Grb2, Shc1 and the regulatory subunit of PI3K (PI3Kp85) [20, 21]. Neuregulin-1 (Nrg1) represents the best characterized ErbB4 ligand and has been shown to be implicated several diseases, including cancer and schizophrenia [17–19].
In this report, we measured the homo- and heterodimeric interactions of 14-3-3 isoforms and the modification-dependent interaction between full-length cytosolic Bad and 14-3-3 isoforms ε and ζ in heterologous NIH-3T3 cells and primary neurons using the split-TEV system. Moreover, we measured the Nrg1-induced interactions of several SH2-adapter proteins with phosphorylated ErbB4 in living cells.
For the split-TEV assays, we expressed a Nrg-1 isoform (Nrg1-typeII-β1a) as a full-length protein . To ensure that ErbB4 is specifically activated at the cell membrane, Nrg1-typeII-β1a was separately transfected into PC12 cells (population 1), whereas ErbB4, the adapters and the GV-dependent reporter were transfected into a second batch of PC12 cells (population 2) (Fig. 8b). 20 h post transfection cell populations 1 and 2 were mixed, and 24 h later the adapter/ErbB4 receptor interactions were monitored. The experimental setup of this 2-population assay was confirmed by co-expressing EYFP in population 1 and ECFP in population 2. Fluorescence microscopy of YFP and CFP revealed the existence of two separate cellular populations with non-overlapping yellow and cyan positive cells being in close contact (Fig. 8c).
Split-TEV assays show that ErbB4-N-TEV-tevS-GV interacts with the adapters PI3Kp85α-C-TEV, PI3Kp85β-C-TEV, Grb2-C-TEV and Shc1-C-TEV only if Nrg1-typeII-β1a was expressed in neighboring cells (Fig. 8d). The cytosolic protein FK506-binding protein-C-TEV fusion protein (FKBP-C-TEV) served as a negative control showing no activation in the presence or absence of Nrg1-typeII-β1a (Fig. 8d). The corresponding Renilla luciferase readings are highly similar between all assays showing that transfection efficiencies were similar, and that secondary stimulatory effects may have induced the assays (Fig. 8e). Thus, inter- and intracellular signaling events can be monitored with appropriately designed split-TEV assays in living cells.
Protein modification-dependent PPIs serve complex regulatory functions in cellular signaling cascades . However, the quantitative analysis of these key cellular events still remains a challenging task. Here, we apply split-TEV assays to monitor constitutive and transient phosphorylation-dependent PPIs that occur in the cytosol and at the membrane of living cells. Thus, split-TEV assays represent an alternative option towards the goal of analysing regulated PPIs of full-length proteins within signaling cascades that occur both at the membrane and in the cytosol of living cells (this study and ).
Using split-TEV, we analyzed the regulated interaction of Bad with 14-3-3 isoforms in primary cultured neurons. Cellular assays in most primary cultured cells, including neurons, still represent a major challenge due to lower transfection efficiencies and limited availability. With our transfection protocol, we routinely obtained transfection efficiencies of primary cultured neurons that did not exceed 10%. Nonetheless, we were able to robustly measure the phosphorylation-dependent Bad/14-3-3 interaction in our standard 96-well format. Since observations in primary cultures may be more relevant to understand cell-type specific cellular signaling events, split-TEV assays represent a particularly valuable tool to determine modification-regulated interaction profiles in these cells. Although we mainly focused on luciferase-based reporters in this study, the fluorescent split-TEV reporter RedERnuc displayed a high signal-to-noise ratio and may represent an additional readout option when cell numbers are limiting and/or transfection efficiencies are low.
In this report, the association of the pro-apoptotic Bad and the adapter protein 14-3-3, which only binds to phosphorylated Bad, was quantified in heterologous cells and primary cultured neurons along with the effects on the cellular state evoked by the interaction itself. Although the results obtained in NIH-3T3 cells and primary cultured neurons showed the same tendency, the effect was less pronounced in primary neurons. This may be caused by the lower transfection efficiencies and/or expression levels obtained with primary neurons. In our assays, the transfection efficiencies of NIH-3T3 cells vary between 20–30%, whereas the relative number of neurons transfected is generally lower than 10%. The cell type depending differences in transfection efficiencies may also explain the results that we obtained by FACS analyses with HEK293 and NIH-3T3 cells using the fluorescent EYFPnuc reporter. Although the results between the cell types and the different reporter systems (luciferase versus fluorescence) were highly similar, the relative robustness of the readout varied. In HEK293, the Akt-1 dependent interaction of 14-3-3ζ-C-TEV with N-TEV-Bad led to an activation of the reporter in almost 20% of all cells, whereas the maximum level of activation in NIH-3T3 was below 2%. However, the relative induction ratios over controls (interaction of 14-3-3ζ-C-TEV with N-TEV-Bad without Akt-1 or with the phosphorylation mutant BadS136A variants) were comparable between cell lines. This difference could be due to variable numbers of plasmids taken up by individual cells, as up to five plasmids are required for the split-TEV assays used. In NIH-3T3 cells, split-TEV was also successfully applied to measure endogenous phosphorylation levels of Akt-1, proving the high sensitivity of split-TEV assays.
Also, we monitored phosphorylation-dependent interactions between activated ErbB4 receptor tyrosine kinases and several cytosolic adapter proteins in PC12 cells. In the presented 2-cell batch assays, ErbB4 was stimulated by a full-length Nrg1 variant expressed exclusively in neighbouring cells. The chosen isoform of Nrg-1 (Nrg1-typeII-β1a) requires proteolytic cleavage by extracellulary proteases to yield a soluble, active form . We conclude that the split-TEV technique may be particularly suited to study weak and eventually transient protein interactions with a simple readout format even if complex membrane signaling is under investigation. Given the enormous complexity of mutually dependent inter- and intracellular signaling processes, the need of scalable assay systems, such as the split-TEV, is apparent. For example, the NRG1 gene generates at least 31 isoforms, which may have different signaling capabilities and/or different receptor affinities . Nrg1 isoforms can bind to ErbB3 and ErbB4 receptors with ErbB4 having four splice variants itself. ErbB receptors can form homo- and heterodimeric complexes exerting different signaling properties, making a differential receptor/adapter association analysis in living cells a valuable and highly challenging task. We have shown that split-TEV assays can be designed to study pivotal inter- and intracellular signaling events in living cells with high sensitivity. Given its flexibility and scalability, the split-TEV technique may help to unravel the complexity of cellular signaling in the future.
Split-TEV assays may complement existing techniques to study phosphorylation-dependent and transient PPIs that were induced by intrinsic kinase activities in living cells. The interactions can be monitored with full-length cytosolic proteins in heterologous cell lines as well as in primary cultured neurons. All assays were performed in multi-plate formats and could therefore be adapted to higher throughput to screen for pharmacological substances interfering with pivotal PPIs within cellular signaling cascades.
Expression plasmids were constructed with PCR techniques using proofreading polymerases (Pfu and EasyA, Stratagene). PCR products were either TA-overhang sub-cloned in pGEM-T (Promega) using standard molecular cloning procedures or were modified with attB1/2 recombination sites for the 'Gateway' recombination cloning (Invitrogen). All protocols were performed according to the manufacturers instruction (Invitrogen). Reporter plasmids used have been described recently .
COS1 cells were transfected with 5 μg of plasmid DNA per sample, cultured for 40 h and lysed in RIPA buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS), including protease inhibitors (Complete kit, Roche). The lysates were sonicated and the cell debris was removed by centrifugation. Equal amounts of lysate containing LDS sample buffer were loaded on precasted gels (Invitrogen) and blotted on PVDF membranes (Amersham Biosciences). The membranes were blocked in blocking buffer (5% milkpowder in TBS-T) and probed with primary (αFlagM2 1:5000 in blocking buffer, Sigma; αHA 1:1000, Roche) and secondary antibodies (HRP goat-α-mouse 1:5000, HRP goat-α-rabbit 1:5000, HRP goat-α-rat 1:1000, Jackson Immuno Research Labs). Anti-phospho-Akt (Ser473) (Cell Signaling Technology, #9272, 1:1000), anti-Akt-1 (Cell Signaling, #9271, 1:1000) and anti-Bad antibodies (Cell Signaling, #9292, 1:1000) were used to determine exogenous, activated and total amounts of Akt-1 and Bad in transfected and stimulated NIH-3T3 cells. For luminescence detection, the Western Lightning kit (Perkin Elmer Life Science) was used and ECL-films (Amersham) were developed in a Kodak X-OMAT 1000. Band intensities of endogenous and transfected proteins were determined using ImageJ 1.37. Scanned images were computationally inverted, the bands in question were equally boxed, and the intensities were measured and converted to mean values. For comparison, the band intensities representing the endogenous protein were set to 100% and plotted as histogram.
NIH-3T3, COS1 and PC12 cells were cultured in DMEM with 1 g/l glucose (BioWhittaker) and supplemented with 5% fetal calf serum (FCS, Invitrogen) (NIH-3T3 and COS1 cells) or 10% FCS and 10% horse serum (HS, Invitrogen) (PC12 cells) and 200 mM Glutamax (Invitrogen). For luciferase experiments, NIH-3T3 cells were cultured in 1% FCS and analysed 24 h (using proteolysis-only reporters) or 24 to 48 h (using 'transcription-coupled' reporters) post transfection. In 2-population cell assays, PC12 cells were cultured with 5% FCS and 5% HS. Therefore, PC12 cells were separated in two populations and transfected with plasmids encoding for ligands only (population 1) or membrane and cytosolic proteins along with reporters (population 2). Populations 1 and 2 were allowed to individually express the proteins for 20 h. Subsequently, the cells were washed, the populations were mixed and the final assay was performed 24 h later. Cells were transfected with Lipofectamine2000 (Invitrogen) according to the manufacture's instructions. For specific stimulation of the PI3K-Akt pathway, NIH-3T3 cells were treated with PDBF-BB (50 ng/ml, PeproTech).
Primary neurons were prepared from the cortex and hippocampus of E17 old mouse embryos. After preparation of the hippocampus and the cortex, both tissues were incubated in HBSS (supplemented with 10 mM Hepes, NaHCO3 and 1× penicillin/streptomycin) containing 0.5× trypsin-EDTA and 0.1 mg/ml DNAseI for 10 min at 37°C. The cells were further dissociated by gentle trituration until a homogenous dispersion was achieved. The cells were pelleted (800 rpm for 10 min) and resuspended in Neurobasal medium supplemented with B27-supplement (1:50, Gibco), L-glutamine (1:100, 2 mM), penicillin (100 U/ml), streptomycin (100 μg/ml), MK-801 (10 μM, Tocris), NGF (0.1 ng/μl, Promega) and β-FGF (0.1 ng/μl, PeproTech). Finally, cells were counted and plated for the luciferase assays (96-well plate: 6 × 104/well). For transient transfections, the primary neurons were cultured for 3 to 4 days, followed by the removal of 60% of the medium, and the addition of the Lipofectamine2000/plasmid mix. Six hours later, 75% of the medium was removed, and replaced by fresh medium and analysed 24 h post transfection.
Luciferase assays were performed in 96-well plates using the Dual-Luciferase kit (Promega). For standardization of transfection variability and increased readout stability, a mix of plasmids coding for renilla-luciferase (RL) was used (CMV-RL, TK-RL and SV40-RL; 1:2:10 molar proportions, respectively). DNA amounts were adjusted to equal levels with pcDNA3 or pCMV-EYFP. Cells were treated with lysis reagent (Promega) according to the manufacturers instructions. Coupled firefly and Renilla-luciferase assays were performed in a Lumat LB96V reader (Berthold Technologies). Results were given as Renilla-normalized relative luciferase units (RLUs) ± standard deviation (SD). All experiments were repeated at least three times with a minimum of three replicates for each data point.
For FACS analysis, HEK293 and NIH-3T3 cells were seeded on day 1 in 24 well plates with 3 × 105 and 105 cells per well, respectively. On day 2, cells were transfected with expression constructs along with the GV2ER plasmid and a nuclear targeted Gal4-dependent EYFP reporter (G5-EYFPnuc) using Lipofectamin 2000 (Invitrogen). 30 h later, cells were stained for 30 min with the nuclear dyes Hoechst 33342 (Hoe, cell permeable, labels all cells) and Propidium Iodide (PI, cell impermeable, labels dying cells) (Invitrogen) according to manufacturer specifications. Single cell suspensions were prepared by trypsin treatment and thorough pipetting in PBS containing 5 mM EDTA. Before FACS analysis, cells were filtered through 50 μm nylon meshes and kept on ice. Hoe, YFP and PI analyses were performed with standard filter settings, gates were adjusted with negative and positive control samples. All experiments were performed with three independent replicates for each data point. Results were given as numbers of YFP-positive cells per 10000 (HEK293) or 5000 (NIH-3T3) Hoe-positive stained cells. FACS measurements were performed with a FACS-Aria (Becton-Dickinson). Errors are given as SEM.
We would like to acknowledge the help of W. Hinrichs for cloning of expression constructs. This work was supported by BMBF (FKZ01GS0498, FKZ0315180A) grants.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.