Categorizing biases in high-confidence high-throughput protein-protein interaction data sets

We characterized and evaluated the functional attributes of three yeast high-confidence protein-protein interaction data sets derived from affinity purification/mass spectrometry, protein-fragment complementation assay, and yeast two-hybrid experiments. The interacting proteins retrieved from these data sets formed distinct, partially overlapping sets with different protein-protein interaction characteristics. These differences were primarily a function of the deployed experimental technologies used to recover these interactions. This affected the total coverage of interactions and was especially evident in the recovery of interactions among different functional classes of proteins. We found that the interaction data obtained by the yeast two-hybrid method was the least biased toward any particular functional characterization. In contrast, interacting proteins in the affinity purification/mass spectrometry and protein-fragment complementation assay data sets were over- and under-represented among distinct and different functional categories. We delineated how these differences affected protein complex organization in the network of interactions, in particular for strongly interacting complexes (e.g. RNA and protein synthesis) versus weak and transient interacting complexes (e.g. protein transport). We quantified methodological differences in detecting protein interactions from larger protein complexes, in the correlation of protein abundance among interacting proteins, and in their connectivity of essential proteins. In the latter case, we showed that minimizing inherent methodology biases removed many of the ambiguous conclusions about protein essentiality and protein connectivity. We used these findings to rationalize how biological insights obtained by analyzing data sets originating from different sources sometimes do not agree or may even contradict each other. An important corollary of this work was that discrepancies in biological insights did not necessarily imply that one detection methodology was better or worse, but rather that, to a large extent, the insights reflected the methodological biases themselves. Consequently, interpreting the protein interaction data within their experimental or cellular context provided the best avenue for overcoming biases and inferring biological knowledge.     

One-step Purification of Twin-Strep-tagged Proteins and Their Complexes on Strep-Tactin Resin Cross-linked With Bis(sulfosuccinimidyl) Suberate (BS3)

Affinity purification of Strep-tagged fusion proteins on resins carrying an engineered streptavidin (Strep-Tactin) has become a widely used method for isolation of protein complexes under physiological conditions. Fusion proteins containing two copies of Strep-tag II, designated twin-Strep-tag or SIII-tag, have the advantage of higher affinity for Strep-Tactin compared to those containing only a single Strep-tag, thus allowing more efficient protein purification. However, this advantage is offset by the fact that elution of twin-Strep-tagged proteins with biotin may be incomplete, leading to low protein recovery. The recovery can be dramatically improved by using denaturing elution with sodium dodecyl sulfate (SDS), but this leads to sample contamination with Strep-Tactin released from the resin, making the assay incompatible with downstream proteomic analysis. To overcome this limitation, we have developed a method whereby resin-coupled tetramer of Strep-Tactin is first stabilized by covalent cross-linking with Bis(sulfosuccinimidyl) suberate (BS3) and the resulting cross-linked resin is then used to purify target protein complexes in a single batch purification step. Efficient elution with SDS ensures good protein recovery, while the absence of contaminating Strep-Tactin allows downstream protein analysis by mass spectrometry. As a proof of concept, we describe here a protocol for purification of SIII-tagged viral protein VPg-Pro from nuclei of virus-infected N. benthamiana plants using the Strep-Tactin polymethacrylate resin cross-linked with BS3. The same protocol can be used to purify any twin-Strep-tagged protein of interest and characterize its physiological binding partners.     

Strategy for identifying protein-protein interactions of gel-separated proteins and complexes by mass spectrometry

A strategy for identifying and characterizing protein interactions among gel-separated proteins and complexes has been developed and tested. The method involves the efficient recovery of proteins or complexes from native gels without affecting their conformational integrity. The use of limited proteolysis of protein complexes, isolated from the gel or formed from the interaction of gel-recovered proteins with potential binding partners, has enabled local binding domains to be efficiently identified using a combination of microfiltration and mass spectrometric analysis. The application of mass spectrometry affords high detection sensitivities, enabling the strategy to be applied to low levels of protein and protein mixtures. The approach is demonstrated for both antigen-antibody and peptide-protein complexes for which protein-binding regions are characterized among simple peptide mixtures and proteolytic digests. The strategy can be easily adapted to achieve high sample throughput and automation using gel-excision robotics and provides a means to study protein interactions in complex biological mixtures and extracts.     

Engineering a high-affinity scaffold for non-chromatographic protein purification via intein-mediated cleavage

While protein purification has long been dominated by standard chromatography, the relatively high cost and complex scale‐up have promoted the development of alternative non‐chromatographic separation methods. Here we developed a new non‐chromatographic affinity method for the purification of proteins expressed in Escherichia coli. The approach is to genetically fuse the target proteins with an affinity tag. Direct purification and recovery can be achieved using a thermo‐responsive elastin‐like protein (ELP) scaffold containing the capturing domain. Naturally occurring cohesin–dockerin pairs, which are high‐affinity protein complex responsible for the formation of cellulosome in anaerobic bacteria, were used as the model. By exploiting the highly specific interaction between the dockerin and cohesin domain from Clostridium thermocellum and the reversible aggregation property of ELP, highly purified and active dockerin‐tagged proteins, such as the endoglucanase CelA, chloramphenicol acetyl transferase (CAT), and enhanced green fluorescence protein (EGFP), were recovered directly from crude cell extracts in a single thermal precipitation step with yields achieving over 90%. Incorporation of a self‐cleaving intein domain enabled rapid removal of the affinity tag from the target proteins, which was subsequently removed by another cycle of thermal precipitation. This method offers great flexibility as a wide range of affinity tags and ligands can be used. Biotechnol. Bioeng. 2012; 109: 2829–2835. © 2012 Wiley Periodicals, Inc.     

Dephosphorylation of photosystem II proteins and phosphorylation of CP29 in barley photosynthetic membranes as a response to water stress

Kinetic studies of protein dephosphorylation in barley thylakoid membranes revealed accelerated dephosphorylation of photosystem II (PSII) proteins, and meanwhile rapidly induced phosphorylation of a light-harvesting complex (LHCII) b4, CP29 under water stress. Inhibition of dephosphorylation aggravates stress damages and hampers photosystem recovery after rewatering. This increased dephosphorylation is catalyzed by both intrinsic and extrinsic membrane protein phosphatase. Water stress did not cause any thylakoid destacking, and the lateral migration from granum membranes to stroma-exposed lamellae was only found to CP29, but not other PSII proteins. Activation of plastid proteases and release of TLP40, an inhibitor of the membrane phosphatases, were also enhanced during water stress. Phosphorylation of CP29 may facilitate disassociation of LHCII from PSII complex, disassembly of the LHCII trimer and its subsequent degradation, while general dephosphorylation of PSII proteins may be involved in repair cycle of PSII proteins and stress-response-signaling.     

Separation methods in the analysis of protein membrane complexes

The separation of membrane protein complexes can be divided into two categories. One category, which is operated on a relatively large scale, aims to purify the membrane protein complex from membrane fractions while retaining its native form, mainly to characterize its nature. The other category aims to analyze the constituents of the membrane protein complex, usually on a small scale. Both of these face the difficulty of isolating the membrane protein complex without interference originating from the hydrophobic nature of membrane proteins or from the close association with membrane lipids. To overcome this difficulty, many methods have been employed. Crystallized membrane protein complexes are the most successful example of the former category. In these purification methods, special efforts are made in the steps prior to the column chromatography to enrich the target membrane protein complexes. Although there are specific aspects for each complex, the most popular method for isolating these membrane protein complexes is anion-exchange column chromatography, especially using weak anion-exchange columns. Another remarkable trend is metal affinity column chromatography, which purifies the membrane protein complex as an intact complex in one step. Such protein complexes contain subunit proteins which are genetically engineered so as to include multiple-histidine tags at carboxyl- or amino-termini. The key to these successes for multi-subunit complex isolation is the idea of keeping the expression at its physiological level, rather than overexpression. On the other hand, affinity purification using the Fv fragment, in which a Strep tag is genetically introduced, is ideal because this method does not introduce any change to the target protein. These purification methods supported by affinity interaction can be applied to minor membrane protein complexes in the membrane system. Isoelectric focusing (IEF) and blue native (BN) electrophoresis have also been employed to prepare membrane protein complexes. Generally, a combination of two or more chromatographic and/or electrophoretic methods is conducted to separate membrane protein complexes. IEF or BN electrophoresis followed by 2nd dimension electrophoresis serve as useful tools for analytical demand. However, some problems still exist in the 2D electrophoresis using IEF. To resolve such problems, many attempts have been made, e.g. introduction of new chaotropes, surfactants, reductants or supporting matrices. This review will focus in particular on two topics: the preparative methods that achieved purification of membrane protein complexes in the native (intact) form, and the analytical methods oriented to resolve the membrane proteins. The characteristics of these purification and analytical methods will be discussed along with plausible future developments taking into account the nature of membrane protein complexes.     

Detection of antibody to M protein of measles virus in patients with subacute sclerosing panencephalitis: a comparative study on immunoprecipitation

Consistent results have not been obtained yet on the presence of antibody to the M protein of measles virus in the sera of patients with subacute sclerosing panencephalitis (SSPE). We performed a comparative study on various immunoprecipitation systems which appeared in the literature and found that the difference in the composition of the solubilizing buffer produced a large variety of results on the immunoprecipitation. [35S]Methionine-labeled Vero cells infected with the Edmonston strain of measles virus were solubilized by 10 different buffers and reacted with hyperimmune rabbit serum to whole virus, monospecific antisera to H, NP, and M proteins of the virus, normal adults' sera, and the sera from 16 SSPE patients. The immune complex was absorbed by protein A and both solubilization and precipitation rates were compared with each viral protein. Although viral proteins were solubilized by all buffers, the solubilization rate varied considerably. M protein was solubilized and was not coprecipitated nonspecifically with any of the other viral proteins. Purified protein A conjugated to Sepharose was preferable to Staphylococcus aureus for absorption of the immune complex since the latter absorbed both viral and host proteins nonspecifically. The precipitation rates of the viral proteins also varied according to the buffers. Better solubilization of the viral proteins seemed to reduce their rate of precipitation for which the presence of SDS may be responsible, and the presence of the protease inhibitors may also affect the results of immunoprecipitation. Detection of M protein in the immunoprecipitates was largely influenced by the kind of buffer used: some buffers could detect it clearly, but others could not defect it at all. Among the solubilizing buffers tested, Saleh's buffer (Virology 93: 369-376 (1979)),, which contains 0.5% DOC and 0.5% Triton X-100, was most reliable for detection of the anti-M antibody in the rabbit serum, because it showed a high solubilization and high precipitation rates of viral proteins without nonspecific absorption by protein A or coprecipitation of M proteins with any of the other proteins. Using this buffer, we could definitely detect M proteins in the immunoprecipitates from the sera of all six healthy adults and 15 out of 16 patients with SSPE. It was found, however, that the amount of M proteins in SSPE patients was lower than that in healthy adults and varied considerably.     

Downstream protein separation by surfactant precipitation: a review

In a conventional protein downstream processing (DSP) scheme, chromatography is the single most expensive step. Despite being highly effective, it often has a low process throughput due to its semibatch nature, sometimes with nonreproducible results and relatively complex process development. Hence, more work is required to develop alternative purification methods that are more cost-effective, but exhibiting nearly comparable performance. In recent years, surfactant precipitation has been heralded as a promising new method for primary protein recovery that meets these criteria and is a simple and cost-effective method that purifies and concentrates. The method requires the direct addition of a surfactant to a complex solution (e.g. a fermentation broth) containing the protein of interest, where the final surfactant concentration is maintained below its critical micelle concentration (CMC) in order to allow for electrostatic and hydrophobic interactions between the surfactant and the target protein. An insoluble (hydrophobic) protein–surfactant complex is formed and backextraction of the target protein from the precipitate into a new aqueous phase is then carried out using either solvent extraction, or addition of a counter-ionic surfactant. Importantly, as highlighted by past researchers, the recovered proteins maintain their activity and structural integrity, as determined by circular dichroism (CD). In this review, various aspects of surfactant precipitation with respect to its general methodology and process mechanism, system parameters influencing performance, protein recovery, process selectivity and process advantages will be highlighted. Moreover, comparisons will be made to reverse micellar extraction, and the current drawbacks/challenges of surfactant precipitation will also be discussed. Finally, promising directions of future work with this separation technique will be highlighted.     

Role of Nonspecific Interactions in Molecular Chaperones through Model-Based Bioinformatics

Molecular chaperones are large proteins or protein complexes from which many proteins require assistance in order to fold. One unique property of molecular chaperones is the cavity they provide in which proteins fold. The interior surface residues which make up the cavities of molecular chaperone complexes from different organisms has recently been identified, including the well-studied GroEL-GroES chaperonin complex found in Escherichia coli. It was found that the interior of these protein complexes is significantly different than other protein surfaces and that the residues found on the protein surface are able to resist protein adsorption when immobilized on a surface. Yet it remains unknown if these residues passively resist protein binding inside GroEL-GroEs (as demonstrated by experiments that created synthetic mimics of the interior cavity) or if the interior also actively stabilizes protein folding. To answer this question, we have extended entropic models of substrate protein folding inside GroEL-GroES to include interaction energies between substrate proteins and the GroEL-GroES chaperone complex. This model was tested on a set of 528 proteins and the results qualitatively match experimental observations. The interior residues were found to strongly discourage the exposure of any hydrophobic residues, providing an enhanced hydrophobic effect inside the cavity that actively influences protein folding. This work provides both a mechanism for active protein stabilization in GroEL-GroES and a model that matches contemporary understanding of the chaperone protein.     

Repair of nitric oxide-modified ferredoxin [2Fe-2S] cluster by cysteine desulfurase (IscS)

Iron-sulfur proteins are among the sensitive targets of the nitric oxide cytotoxicity. When Escherichia coli cells are exposed to nitric oxide, iron-sulfur clusters are modified forming protein-bound dinitrosyl iron complexes. Such modified protein dinitrosyl iron complexes are stable in vitro but are efficiently repaired in aerobically growing E. coli cells even without any new protein synthesis. Here we show that cysteine desulfurase encoded by the gene iscS of E. coli can directly convert the ferredoxin dinitrosyl iron complex to the ferredoxin [2Fe-2S] cluster in the presence of L-cysteine in vitro. A reassembly of the [2Fe-2S] cluster in the ferredoxin dinitrosyl iron complex does not require any addition of iron or other protein components. Furthermore, a complete removal of the dinitrosyl iron complex from ferredoxin prevents reassembly of the [2Fe-2S] cluster in the protein. The results suggest that cysteine desulfurase (IscS) together with L-cysteine can efficiently repair the nitric oxide-modified ferredoxin [2Fe-2S] cluster and that the iron center in the dinitrosyl iron complex may be recycled for the reassembly of iron-sulfur clusters in proteins.     

Solid-phase assay for determination of binding parameters of ligand-protein complexes with high dissociation rates

The binding parameters, the affinity constant (Ka) and binding capacity (Q), of a protein possessing ligand-protein complexes with a high dissociation rate (Sex Steroid Binding protein from Bufo arenarum) were determined using a solid-phase method. The technique is based upon the adsorption of the steroid-protein complex to DEAE-cellulose. This method was compared with a nonequilibrium method (charcoal adsorption of free ligand), and the latter resulted in underestimation of both binding parameters, Ka and Q. The solid-phase method reported here is appropriate to determine the binding parameters of proteins with high dissociation rates because the results are independent of the complex half-time. The method also possesses advantages compared to other equilibrium assays such as dialysis or steady-state electrophoresis. With minor modifications, it may be useful to characterize different proteins, particularly those possessing ligand-protein complexes with very high dissociation rates.     

An in vivo Crosslinking Approach to Isolate Protein Complexes From Drosophila Embryos

Many cellular processes are controlled by multisubunit protein complexes. Frequently these complexes form transiently and require native environment to assemble. Therefore, to identify these functional protein complexes, it is important to stabilize them in vivo before cell lysis and subsequent purification. Here we describe a method used to isolate large bona fide protein complexes from Drosophila embryos. This method is based on embryo permeabilization and stabilization of the complexes inside the embryos by in vivo crosslinking using a low concentration of formaldehyde, which can easily cross the cell membrane. Subsequently, the protein complex of interest is immunopurified followed by gel purification and analyzed by mass spectrometry. We illustrate this method using purification of a Tudor protein complex, which is essential for germline development. Tudor is a large protein, which contains multiple Tudor domains - small modules that interact with methylated arginines or lysines of target proteins. This method can be adapted for isolation of native protein complexes from different organisms and tissues.     

A two-cycle immunoprecipitation procedure for reducing nonspecific protein contamination.

A two-cycle immunoprecipitation procedure is described that markedly reduces nonspecific protein contamination occurring during the precipitation of hepatic lipase from rat H4 hepatoma cells. In this method, the precipitation of immune complexes during both cycles is achieved by utilizing a sodium dodecyl sulfate (SDS)-washed preparation of lyophilized Staphylococcus aureus cells (Staph A); this washed preparation effectively removes Staph A contaminants without compromising the ability to bind immune complexes. Following initial immunoprecipitation of the antigen, the Staph A/IgG/antigen complex containing coprecipitated nonspecific proteins was dissociated with SDS. Triton X-100 was added to the dissociated immunoprecipitate at a concentration (by weight) of at least 5 parts Triton X-100 to 1 part SDS. A second cycle of immunoprecipitation was then initiated by addition of fresh antibody, followed by Staph A precipitation of immune complexes and analysis by SDS-polyacrylamide gel electrophoresis. The two-cycle procedure is shown to be reproducible and suitable for the quantitative determination of relative amounts of hepatic lipase. The procedure described here is generally applicable to the immunoprecipitation of other antigens.     

Sampling strategy for protein complex prediction using cluster size frequency

In this paper we propose a Markov chain Monte Carlo sampling method for predicting protein complexes from protein–protein interactions (PPIs). Many of the existing tools for this problem are designed more or less based on a density measure of a subgraph of the PPI network. This kind of measures is less effective for smaller complexes. On the other hand, it can be found that the number of complexes of a size in a database of protein complexes follows a power-law. Thus, most of the complexes are small-sized. For example, in CYC2008, a database of curated protein complexes of yeast, 42% of the complexes are heterodimeric, i.e., a complex consisting of two different proteins. In this work, we propose a protein complex prediction algorithm, called PPSampler (Proteins' Partition Sampler), which is designed based on the Metropolis–Hastings algorithm using a parameter representing a target value of the relative frequency of the number of predicted protein complexes of a particular size. In a performance comparison, PPSampler outperforms other existing algorithms. Furthermore, about half of the predicted clusters that are not matched with any known complexes in CYC2008 are statistically significant by Gene Ontology terms. Some of them can be expected to be true complexes.     

Affinity purification of plasmid DNA by temperature-triggered precipitation

This protocol presents a new method to purify plasmid DNA using temperature-triggered precipitation. The principle is based on the specific DNA-binding affinity of a bacterial metalloregulatory (MerR) protein to its cognate DNA sequence and the temperature responsiveness of elastin-like protein (ELP). A bifunctional ELP-MerR fusion protein is created to enable the precipitation of plasmid DNA, designed to contain the MerR recognition sequence, by a simple temperature trigger. The protocol covers all stages of the process from the design of ELP-MerR fusion proteins and MerR-binding plasmids, to the isolation of plasmid DNA from Escherichia coli cultures after boiling lysis, the subsequent temperature-triggered precipitation of plasmid DNA-fusion protein complexes and final elution of plasmid DNA by mild heating. This protocol is well suited to laboratory research-scale applications, producing plasmid DNA of better purity and similar yield as one of the most commonly used laboratory methods, standard alkaline lysis (known as the midiprep procedure). The protocol takes approximately 30 min to obtain pure plasmid DNA from cell cultures using the temperature-triggered precipitation method.     

Detection of circulating immune complexes in liver diseases, systemic lupus erythematosus and glomerulonephritis by polyethylene glycol precipitation

A method for detection and quantitation of circulating immune complexes using precipitation of the complexes by polyethylene glycol (PEG) has been reexamined to determine the influence of pH on the recovery and the reproducibility of the results. Results showed that the pH optimum for these determinations was 7.8. The recovery percentages range from 57.8-146.5% at lower immune complex concentrations, and from 73.9-101.3% at higher concentrations. The reproducibility of the method seems reasonably acceptable with a percent coefficient of variation ranging from 0.5-9.5. This method for quantitation of circulating immune complexes by polyethylene glycol precipitation is consistent and relatively reliable. Using this method, the levels of circulating immune complexes in sera in patients with hepatitis, liver cirrhosis, hepatoma, acute post-streptococcal glomerulonephritis (before and after treatment) and systemic lupus erythematosus have been examined. The results showed that except the patients with treated acute post-streptococcal glomerulonephritis who had a similar amount of immune complexes with normal controls, the level of immune complexes in patients with other types of diseases were all higher than the control. In addition, the composition of IgG, IgA, IgM, C3 and C4 of the precipitable complexes in sera of patients with three types of liver disease has been analyzed and demonstrated that the percentages of IgM were higher than the normal control. However, C3 and C4 in hepatitis and liver cirrhosis patients were lower than those of the control.     

Isolation of specific RNA-binding proteins using the streptomycin-binding RNA aptamer

Here we report a simple and cheap one-step affinity purification protocol for isolating RNAs or proteins that interact with selected functional RNAs. The streptomycin-binding aptamer, termed 'StreptoTag,' is embedded in or fused to either end of any RNA of interest. The resulting hybrid RNA can then be immobilized on a streptomycin affinity matrix. When a complex protein mixture or total cellular lysate is applied to the matrix, subsequent elution with free streptomycin allows efficient recovery of specific ribonucleoprotein or RNA-RNA complexes. The method was successfully used to purify yeast and phage RNA-binding proteins and group II intron, viral and bacterial noncoding RNA (ncRNA)-binding proteins. The selective enrichment of bacterial mRNAs that bind ncRNAs has also been demonstrated. Once the affinity matrix, the RNA construct and the protein extracts have been prepared, the experimental procedure can be performed in 1-2 h.     

Localization of 13 one-helix integral membrane proteins in photosystem II subcomplexes

Photosystem II is a multimeric protein complex of the thylakoid membrane in chloroplasts. Approximately half of the at least 26 different integral membrane protein subunits have molecular masses lower than 10 kDa. After one-dimensional (1D) or two-dimensional (2D) polyacrylamide gel electrophoresis (PAGE) separation, followed by enzymatic digestion of detected proteins, hardly any of these low-molecular-weight (LMW) subunits are detectable. Therefore, we developed a method for the analysis of highly hydrophobic LMW proteins. Intact proteins are extracted from acrylamide gels using a mixture of formic acid and organic solvent, precipitated with acetone, and analyzed by “top-down” mass spectrometry (MS). After offline nanoESI (electrospray ionization) MS, all LMW one-helix proteins from photosystem II were detected. In the four detected photosystem II supercomplexes of Nicotiana tabacum wild-type plants, 11 different one-helix proteins were identified as PsbE, -F, -H, -I, -K, -L, -M, -Tc, -W, and two isoforms of PsbX. The proteins PsbJ, -Y1, and -Y2 were localized in the buffer front after blue native (BN) PAGE, indicating their release during solubilization. Assembled PsbW is detected exclusively in supercomplexes, whereas it is absent in photosystem II core complexes, corroborating the protein’s function for assembly of the light-harvesting complexes. This approach will substantiate gel-blot immunoanalysis for localization and identification of LMW protein subunits in any membrane protein complex.     

Identification of novel protein-protein interactions using a versatile mammalian tandem affinity purification expression system

Identification of protein-protein interactions is essential for elucidating the biochemical mechanism of signal transduction. Purification and identification of individual proteins in mammalian cells have been difficult, however, due to the sheer complexity of protein mixtures obtained from cellular extracts. Recently, a tandem affinity purification (TAP) method has been developed as a tool that allows rapid purification of native protein complexes expressed at their natural level in engineered yeast cells. To adapt this method to mammalian cells, we have created a TAP tag retroviral expression vector to allow stable expression of the TAP-tagged protein at close to physiological levels. To demonstrate the utility of this vector, we have fused a TAP tag, consisting of a protein A tag, a cleavage site for the tobacco etch virus (TEV) protease, and the FLAG epitope, to the N terminus of human SMAD3 and SMAD4. We have stably expressed these proteins in mammalian cells at desirable levels by retroviral gene transfer and purified native SMAD3 protein complexes from cell lysates. The combination of two different affinity tags greatly reduced the number of nonspecific proteins in the mixture. We have identified HSP70 as a specific interacting protein of SMAD3. We demonstrated that SMAD3, but not SMAD1, binds HSP70 in vivo, validating the TAP purification approach. This method is applicable to virtually any protein and provides an efficient way to purify unknown proteins to homogeneity from the complex mixtures found in mammalian cell lysates in preparation for identification by mass spectrometry.     

Inferring the lifetime of endosomal protein complexes by fluorescence recovery after photobleaching

Cellular signal transduction is dynamic, with signaling proteins continually associating and dissociating into and from protein complexes. Here we present a fluorescence recovery after photobleaching technique to determine the lifetime of protein complexes on intracellular vesicles. We use Bayesian inference based on a model that includes the diffusion of cytosolic proteins and their interaction with membrane-bound receptors. Our analysis is general: we incorporate prior information on protein diffusion, measurement error in determining fluorescence intensities, corrections for photobleaching, and variation in the concentration of receptors between vesicles. We apply our method to the complexes formed on endosomes by G-protein-coupled receptors and the protein β-arrestin. The lifetime of these complexes determines the recycling rate of the receptors. We find in mammalian cells that the bradykinin type 2 receptor and β-arrestin2 complex has a lifetime of ∼2 min, while the angiotensin II type 1A receptor and β-arrestin2 complex has a lifetime of ∼6 min. As well as allowing quantitative comparisons between experiments, our method provides in vivo parameters for systems biology simulations of signaling networks.