If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
ISREC National Centre of Competence in Research (NCCR) Molecular Oncology, Swiss Institute of Experimental Cancer Research (ISREC), Epalinges, CH-1006 SwitzerlandISREC Swiss Institute of Bioinformatics (SIB), CH-1015 Lausanne, Switzerland
From the Department of Molecular Cell Research, Max Planck Institute for Medical ResearchDepartment of Biological Sciences, Imperial College, London SW7 2AZ, United KingdomDépartment de Biochimie, Faculté des Sciences, Université de Genève, CH-1211 Genève-4, Switzerland
* This work was supported in part by the Max Planck Society, The Wellcome Trust, and the Swiss National Science Foundation (to T. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. S The on-line version of this article (available at http://www.mcponline.org) contains supplemental material. c Both authors contributed equally to this work. h Supported by the National Centre of Competence in Research (NCCR) Molecular Oncology, a research program of the Swiss National Science Foundation.
Phagocytosis, whether of food particles in protozoa or bacteria and cell remnants in the metazoan immune system, is a conserved process. The particles are taken up into phagosomes, which then undergo complex remodeling of their components, called maturation. By using two-dimensional gel electrophoresis and mass spectrometry combined with genomic data, we identified 179 phagosomal proteins in the amoeba Dictyostelium, including components of signal transduction, membrane traffic, and the cytoskeleton. By carrying out this proteomics analysis over the course of maturation, we obtained time profiles for 1,388 spots and thus generated a dynamic record of phagosomal protein composition. Clustering of the time profiles revealed five clusters and 24 functional groups that were mapped onto a flow chart of maturation. Two heterotrimeric G protein subunits, Gα4 and Gβ, appeared at the earliest times. We showed that mutations in the genes encoding these two proteins produce a phagocytic uptake defect in Dictyostelium. This analysis of phagosome protein dynamics provides a reference point for future genetic and functional investigations.
Phagocytosis is the complex process by which eukaryotes ingest large particles of over 200-nm diameter. Many primitive eukaryotic cells phagocytose food particles, whereas in metazoa so-called “professional phagocytes” like the neutrophils and macrophages of the immune system specialize in engulfing invading microorganisms and parasites as well as necrotic and apoptotic cell remnants. Usually phagocytosis is triggered by contact of the particle with the cell surface, which induces its actin-mediated engulfment into a membrane-bound phagosome derived from the plasma membrane. Other endomembrane compartments such as endosomes and the endoplasmic reticulum (ER)
The abbreviations used are: ER, endoplasmic reticulum; 2D, two-dimensional; NSF, N-ethylmaleimide-sensitive factor; SNARE, soluble NSF attachment protein receptor; PMF, peptide mass fingerprint; HSP, heat shock protein.
1The abbreviations used are: ER, endoplasmic reticulum; 2D, two-dimensional; NSF, N-ethylmaleimide-sensitive factor; SNARE, soluble NSF attachment protein receptor; PMF, peptide mass fingerprint; HSP, heat shock protein.
contribute to the newly formed phagosomes. After killing of the engulfed microorganism, the contents of the phagosome are digested, and finally the lysosomal hydrolases are recycled for further use. During this maturation process, the phagosome remodels the protein and lipid components of its membrane as well as the soluble luminal components (
). Maturation of the phagosome is thus a normal part of its function, but the organelle can also be remodeled by some bacterial pathogens and eukaryotic parasites that “hijack” phagosomes and use them as a refuge where they can proliferate and evade the surveillance of the immune system (
). Yet despite a century of study, the mechanisms of phagocytic uptake and maturation are still relatively poorly understood. Exciting progress has been made, however, since the advent of large scale proteomics methods, which have revealed new facets of this organelle (
). This analysis found diverse protein classes including not only the expected lysosomal proteins but also a large variety of proteins involved in regulating membrane trafficking, such as SNAREs and Rab GTPases as well as a subset of heterotrimeric G protein subunits involved in signal transduction and many others (
). Despite these pioneering studies and recent technical advances, time-dependent organelle proteomics is still in its infancy. This is largely due to the paucity of appropriate bioinformatics tools to extract and integrate large scale and time-profiled proteomics data.
The social amoeba Dictyostelium is a very effective phagocyte, and its experimental versatility makes it an ideal candidate for multidisciplinary studies of cell function. Its genome is fully sequenced, assembled, and thoroughly and accurately annotated (
). Proteomics studies of phagosomes isolated from the amoeba Entamoeba histolytica, for example, have revealed several aspects of phagosome signaling, uptake mechanisms, and time-dependent maturation in common with phagocytosis in mammalian professional phagocytes (
Here we used a combination of time-resolved 2D gel electrophoresis and mass spectrometry-assisted protein identification to generate a protein history of the life of a phagosome in Dictyostelium, incorporating 179 phagosomal components. By clustering proteins that appear and disappear from the phagosome at similar times, we defined groups of proteins and functions that can be placed on a flow chart of phagosome maturation. Validating this approach, we found that two heterotrimeric G protein subunits, Gα4 and Gβ belong to two distinct but related groups of proteins present at early times in phagosome maturation. By studying Dictyostelium strains with ablations of the genes encoding Gα4 and Gβ using a flow cytometry-based assay for phagocytic uptake, we demonstrated that both Gα4 and Gβ function in an early step of phagocytosis.
Dictyostelium discoideum cells of wild-type strain Ax2 were grown axenically in HL5c medium (
) supplemented with 10 units/ml penicillin and 10 μg/ml streptomycin on plastic dishes or in shaking culture (at 180 rpm) at 22 °C. Mutant cell lines were selected and grown in HL5c medium supplemented with 5 μg/ml G418.
The antibodies used in this study were mouse monoclonal antibodies and rabbit polyclonal antibodies raised against Dictyostelium proteins as listed in Supplemental Table SII.
) and transfer onto nitrocellulose membranes (Protran, Schleicher & Schuell), immunodetection was performed as described previously using horseradish peroxidase-coupled goat anti-mouse or goat anti-rabbit IgGs (BioRad) at 1:5,000 dilution (
). In brief, freshly prepared phagosome pellets (about 7 × 1010 phagosomes prepared from 1.5 × 109 cells containing an average of 46.7 beads/cell) were resuspended in carbonate buffer (200 mm Na2CO3 (Sigma), pH 11) by vortexing. After homogenization by five passages through the needle of a 1-ml insulin syringe, samples were kept on ice for 1 h. Stripped phagosome membranes were repelleted by ultracentrifugation for 1 h at 100,000 × g in a Beckmann MLA130 rotor. Pellets were resuspended in lysis buffer containing 7 m urea (GE Healthcare), 2 m thiourea (Amersham Biosciences), 2% (w/v) CHAPS (Calbiochem), 2% (w/v) ASB-C8Ø (Calbiochem), 1% (w/v) DTT (Pharmacia Biotech), 2% (v/v) ampholytes (IPG buffer, pH 3–10 non-linear, Amersham Biosciences), and a protease inhibitor mixture (Complete EDTA-free). After sonication, isoelectric focusing gel electrophoresis was carried out as described below. Supplemental information is available.
Sample Preparation and 2D Gel Electrophoresis—
Intact purified phagosomes were resuspended in lysis buffer containing 7 m urea (Merck), 2 m thiourea (Merck), 2% (w/v) CHAPS (Sigma), 1% (w/v) DTT (Sigma), 2% (v/v) Pharmalyte pH 3–10 (Amersham Biosciences), and a protease inhibitor mixture (Complete EDTA-free, Roche Applied Science) (
). Suspensions were sonicated 3 × 10 min at 4 °C in a bath sonicator, incubated at room temperature for 2 h, and centrifuged for 60 min at 75,000 × g in a Beckmann TL120 centrifuge, and supernatants were stored at −80 °C until further use.
Extracts were separated in the first dimension using 18-cm strips with immobilized non-linear gradients from pH 3–10 pH (Amersham Biosciences) followed by standard SDS-PAGE as described previously (
) or with colloidal Coomassie Blue (Novex/Invitrogen) according to the manufacturer’s instructions. The gels were scanned using a Sharp JX-330 scanner and Imagemaster Labscan software. Procedures were carried out under standardized conditions for all gels. Supplemental information is available.
Individual spots were excised from 2D gels, reduced with DTT, alkylated with iodoacetamide, and digested with trypsin as described previously (
). Following digestion, tryptic peptides were extracted from the gel pieces with 50% acetonitrile, 0.1% TFA; concentrated; and analyzed by mass spectrometry.
For peptide fingerprinting by MALDI-TOF mass spectrometry (Ultraflex, Bruker), samples were desalted using ZipTip (Millipore) according to the manufacturer’s instructions and spotted onto a steel target using α-cyano-4-hydroxycinnamic acid as matrix. The peptide mass fingerprint (PMF) was acquired after external calibration (peptide calibration standard II, Bruker) in positive ion reflector mode. For protein identification by PMF, peptide masses were labeled manually using the SNAP algorithm (signal to noise ratio = 3; quality factor threshold, 100) (flexAnalysis, Bruker) by comparison with a control sample taken from a spot of an empty area of the same gel. The PMF was searched against the Dictyostelium database (protein sequences for dictyBase primary features, 13,676 sequences, at dictybase.org) using Mascot version 2.0.5 (Matrix Science). The algorithm was set to use trypsin as the enzyme, allowing for one missed cleavage site and assuming carbamidomethyl as a fixed modification of cysteine and oxidized methionine as a variable modification. Mass tolerance was set to 100 ppm unless otherwise indicated. Protein hits were considered identified if the Mascot score exceeded the significance level (p > 0.05).
For peptide sequencing by ESI Q-TOF mass spectrometry, peptides were desalted and concentrated using custom-made chromatographic columns (Poros 50 R2, Perseptive Biosystems) (
). They were eluted directly into a precoated borosilicate nanoelectrospray needle (MDS Protana, Odense, Denmark). Mass spectrometry was performed on a Q-TOF mass spectrometer (PE Sciex, Weiterstadt, Germany) equipped with a nano-ESI ion source (MDS Protana). A potential of 900 V was applied to the nanoelectrospray needle. Declustering potential and focusing potential were set to 40 and 100, respectively. Fragmentation of selected peptides (unit resolution) was usually performed at three different collision energies (22, 27, and 35 V). The data were processed using the Bioanalyst software (PE Sciex).
Image Processing and Dataset and Statistical Analyses—
The digitalized 2D gels of the time series were analyzed using the Phoretix 2D Evolution (version 2005) software (Nonlinear Dynamics, Newcastle-upon-Tyne, UK) for spot detection, gel matching, and background correction (mode of nonspot, vector size 100 pixel), and these data were normalized to the sum of the total spot volume. Data were exported to a spreadsheet program (Microsoft Excel).
Temporal Profile Data Analysis—
Data analysis was performed in R (cran.r-project.org). For data normalization, the spot intensity vectors per time point were first scaled to a constant sum of intensities, and then the time series vector per spot was rescaled so that the maximum intensity was standardized to a value of 1,000. Spots were classified into groups with similar time profiles with the partitioning around medoids (PAM) algorithm (
) and a predefined number of 24 groups because silhouette width values did not support any specific number of groups in the data. Details about the choice of this value of 24 is available in the supplemental information. Only the 898 spots that were found in at least two time points were used. Of the 490 spots detected at only one time point, only one representative from each of the six time points was taken for analysis. The distance measure was 1 − Pearson correlation for the transformed variable Z = log2 (1 + ratio of the intensity to the mean intensity). Color-coded intensity plots for Z were produced with the function “heatmap.” Relationships between the 24 groups were visualized in a hierarchical clustering dendrogram of their average profiles (hclust function in R, average linkage method, 1 − Pearson correlation as distance). For cross-correlation analysis, the pairwise Pearson correlations were computed between each of the averaged profiles of the 24 groups, and this correlation matrix was represented on a false color scale.
Flow Cytometry-based Uptake Assay—
We used a flow cytometry-based particle uptake assay detailed in the supplemental information. Briefly 107 cells grown on plates were harvested, resuspended in 5 ml of HL5c medium, placed in one well of a 6-well plate, and shaken at 150 rpm. After washing, 2 × 109 1-μm fluorescent beads (Fluoresbrite YG 1-μm microspheres, Polysciences, Inc.) were added to the cells in suspension and incubated at room temperature under constant shaking at 120 rpm. At each time point, 0.5 ml of cells was harvested, and bead uptake was stopped. Then cells were centrifuged at 500 × g for 5 min at 4 °C, resuspended in 0.5 ml of ice-cold Soerensen buffer containing 120 mm sorbitol, and kept on ice until fluorescence-activated cell sorter data acquisition. For each time point, 30,000 fluorescence events were acquired using a FACScan flow cytometer (BD Biosciences), and bead uptake was quantified. Mean fluorescence was calculated by analyzing histograms showing fluorescence versus events. Supplemental information is available.
RESULTS AND DISCUSSION
In analyzing the proteome of Dictyostelium phagosomes, we had two aims: to identify new protein components and to see how phagosome proteins change during maturation. To do so, we prepared large amounts of latex bead phagosomes by using a protocol that includes physiological concentrations of ATP, ensuring high yields and purity (see supplemental information). Briefly using markers for a variety of organelles, we calculated that our phagosome fractions were enriched up to about 100-fold over a crude membrane fraction and were depleted about 10-fold of major contaminants, such as contractile vacuole, endoplasmic reticulum, and mitochondria (
). In addition, our pulse-chase protocol allows purification of phagosome at various times that were shown previously to cover all the steps of Dictyostelium phagosome maturation from uptake through digestion and finally to exocytosis of the undigested remnants (
). We separated phagosome proteins by 2D gel electrophoresis using preparative gels from which we picked spots and identified the corresponding proteins (Fig. 1) and analytical gels of six different time points during maturation from which we computed a time profile for each spot (Figs. 2 and 3).
Identification of Phagosome Proteins—
We began by applying standard protein extraction, solubilization, and 2D gel procedures (see “Experimental Procedures”) to analyze the prepared phagosomes. This resulted in highly reproducible gels with very good spot quality (see for example, Fig. 1A (gel A); see also Ref.
). However, because these procedures do not solubilize hydrophobic and transmembrane proteins very well, we also explored a variety of the latest generation detergents (see supplemental information) as well as various extraction methods and 2D gel protocols. Carbonate extraction (
) yielded excellent results; a preparative gel (gel B) from a phagosome extract prepared according to this method is shown in Fig. 1B.
A total of 180 and 232 spots were picked from gels A and B, respectively, and analyzed by peptide fingerprinting (see “Experimental Procedures” and Supplemental Figs. S1 and S2 for spectra). Among these, 11 spots were keratin, and 72 others could not be identified for technical reasons. In addition, four proteins corresponding to spots 134 and 135 of gel A and spots 136 and 166 of gel B were identified by ESI-MS/MS (see Supplemental Table SI and Figs. S1 and S2 for representative spectra). In total, 137 spots from gel A and 192 spots from gel B were identified as known proteins or predicted open reading frames in the Dictyostelium genome (dictybase.org (
)). Taking into account the proteins common to both gels and counting proteins present as multiple spots, including degradation products, as one protein, a total of 179 different proteins were finally identified. Table I lists all these proteins, their location(s) on each gel, their known or proposed functions and cellular locations, whether they were found previously in the mouse phagosome proteome (
), and whether their presence has been confirmed by our ongoing large scale analysis of Dictyostelium phagosomes by one-dimensional gel electrophoresis and liquid chromatography tandem mass spectrometry.
The phagosomal proteins we identified belonged to a variety of functional classes (Fig. 1C). Reflecting the digestive character of phagosomes and their export of nutrients to the cytosol, 37% of the proteins were members of the “energy, transport, and metabolism” class. The second largest group, comprising 13% of polypeptides, was proteins involved in membrane traffic, reflecting the importance of transport to and from the maturing phagosome. Among them, we found a large collection of small GTPases, some of which are known to play a role in uptake processes in Dictyostelium (
), indicating that the phagocytic mechanisms have been conserved through evolution. In the membrane traffic group, we also detected members of the SNARE machinery responsible for the specificity of membrane fusion. Of these, only α- and γ-SNAP (soluble NSF attachment protein) have been found in mouse phagosomes. We also found dynamin, a protein involved in vesicle fission.
Other large groups of proteins we identified function in signal transduction (9%) and cytoskeleton organization (7%). These signaling proteins included the heterotrimeric G protein subunits Gβ, another Gβ-like protein, and Gα4. Gα2, Gβ1, and Gβ2 have been found previously in mouse phagosomes (
). Not surprisingly, in view of their richly documented role in phagocytosis, we found a large variety of cytoskeleton proteins, some of which have also been found in mouse phagosomes. In addition to the signaling and cytoskeleton functional groups, we found a large group of heat shock proteins (HSPs) and other chaperones (9% of all proteins). HSPs act as chaperones and assist protein synthesis, folding, and breakdown. They also regulate actin polymerization/capping, uncoating of clathrin-coated vesicles, and phagocytosis in macrophages (
). Besides proteins known to play a role in phagocytosis and known proteins for which a role in phagocytosis has not been described, the functions of over 35 conserved proteins we identified (8%) are not known. Additional research will be necessary to assess the role of these novel candidates in phagocytic functions.
Fig. 1D shows the known or predicted subcellular locations of the phagosome components we identified. Cytosolic proteins comprise 34%, endosomal and lysosomal proteins comprise 12%, and plasma membrane proteins comprise 8% of the total. The subcellular locations of 28% of them, however, are not known. This broad variety of functions and subcellular locations reflects the complex processes involved in the uptake of particles and the subsequent maturation of the phagosome into a killing and digestive compartment. It is important to note that it is difficult to judge a priori the specificity of the presence of cytosolic proteins on phagosomes. As detailed in the supplemental information, some abundant cytosolic proteins are represented, but many are not. In addition, the fact that the presence of most proteins follows a complex temporal profile speaks against a simple piggy-backing during purification but in favor of a specific and regulated recruitment.
Many of the proteins we identified have evident roles to play at one or another stage of phagosome maturation (see above), but the presence of others is more surprising and might reveal additional processes linked to the phagosome. One such example is the identification of proteins from the ER, Golgi apparatus, and peroxisomes among the phagosome proteins. The presence of four ribosomal components, four tRNA synthetases, the 54-kDa subunit of the signal recognition particle, and three translation elongation factors (which are actin-binding proteins), all factors involved in protein biosynthesis, may be explained either by protein synthesis taking place on the phagosome itself or substantial association/fusion of phagosomes with the ER (
). This latter hypothesis is supported by the presence of a subset of ER-resident proteins (calreticulin, protein-disulfide isomerase, and a homologue of the immunoglobulin Binding Protein BiP) and ER export rafficking regulators (Sar1 and Rab1D) in the phagosomes.
Dynamics of the Phagosome Proteome—
To generate temporal profiles of phagosome proteins during maturation, we made use of our established pulse-chase protocol (see supplemental information). For each maturation stage we quantitate the number of purified phagosomes by measuring light scattering in the fraction collected from the sucrose gradient. This measure is exquisitely precise and allows us to adjust the phagosome fractions for identical concentration of phagosomes. This is further demonstrated both by the almost identical protein concentration measured in each normalized fraction (data not shown) and the equal loading of total proteins on one-dimensional gels (
). We concluded that the total amount of protein per phagosome does not vary significantly throughout maturation (less than 15%) and is ∼12 μg/109 phagosomes. Therefore, each 2D gel of the series was loaded with an equal amount of phagosomes and of total protein and stained to a similar extent. Fig. 2 presents a gallery of 2D gels of phagosome extracts obtained at six time points including a 5-min pulse (5′/0′), a 15-min pulse (15′/0′), and 15-min (15′/15′), 45-min (15′/45′), 105-min (15′/105′), and 165-min (15′/165′) chases after a 15-min pulse. The colored circles around the spots indicate a difference compared with the preceding time point: red circles indicate spots that increased more than 2-fold in intensity, whereas green circles indicate spots that decreased more than 2-fold in intensity. (The spots that were detected at all time points (in varying amounts) are circled in blue in Supplemental Fig. S1.)
This simple analysis revealed substantial remodeling during the different phases. For example, the 5-min pulse (Fig. 2A) had the greatest number of different spots (1,029); of these, 259 were specific for this time point (more than at any other time point) (Table II). This complexity at early times likely reflects the fact that early phagosomes contain both proteins derived from the plasma membrane and newly recruited phagosome-specific proteins (
). Not surprisingly, comparison of the second time point with the first revealed the disappearance of 469 spots and appearance of 130 others (Table III). This substantial remodeling likely includes sorting to recycle plasma membrane proteins back to the surface (
). Overall the total number of spots detected decreased over time except at the last time point (Table II). Vast remodeling of the phagosomes during maturation was indicated by the fact that, on average, 184 spots appeared and 291 spots disappeared between any two consecutive time points (Table III). Again on average, 35% of the spots detected were present only at one time point, 20% were detected at only two time points, and 7% were detected at five time points. Nevertheless 12% of all spots were present at all six time points (Supplemental Fig. S1 and Table IV).
In summary, this analysis clearly indicates distinct phases in the maturation program. We detected the most remodeling at the beginning and end of the pathway and thus propose that uptake and exocytosis are the most complex “multidisciplinary” stages, requiring integration of signaling, membrane trafficking, and cytoskeleton reorganization. The more “specialized” intermediate stages, we suggest, are more uniquely devoted to successive membrane trafficking steps necessary for efficient accumulation of digestive enzymes and extraction and transport of nutrients and may thus require a simpler repertoire of proteins.
Temporal Changes in Phagosome Components—
The complexity of the maturation process can be seen from Fig. 3. The reference gel (Fig. 3A) shows an overlay of the 1,388 spots detected in the whole time series superimposed on the gel of time point 5′/0′. Spots that were present at two or more time points are circled in blue or red; the 125 spots circled in red correspond to proteins identified from the preparative gels (see Table I, column “Group”). The 490 spots present at only one time point are circled in green. Although these spots include proteins that potentially are the most stage-specific, they are also most prone to artifact (degradation and spots that could not be matched to other gels), therefore they were not all included in the cluster analysis; instead six spots representative of the proteins that appear only at one of each of the six time points were included. The spots circled in blue or red were analyzed further to monitor how their intensity changed over time. Spot intensities were quantified by densitometry, normalized for the complete temporal profile, and depicted as heat maps in which red corresponds to high intensity and green corresponds to low intensity (see “Experimental Procedures”). To extend the 2D gel data, we also analyzed the pulse-chased phagosome preparations for the presence of known endosomal and phagosomal proteins by using quantitative Western blotting (Fig. 3C). This approach also allowed us to compare the data obtained for a selection of seven proteins by both Western blotting and 2D gels (see below). Again signal intensities were quantified by densitometry, normalized for the temporal profile, and depicted as heat maps (Fig. 3D).
Altogether we obtained 925 profiles (for 898 spots present at more than one time point, plus the profiles of the six spots representative of the 490 stage-specific spots, plus the 21 profiles obtained by Western blotting) representing 5,550 individual intensity measurements that are presented in the heat map in Fig. 3B. These “temporal profiles” were submitted to cluster analysis, a method also used to group microarray “expression profiles” according to their degree of similarity. Exploratory analysis using a variety of clustering methods and distance metrics gave fairly robust results and similar clusters and revealed that the profiles do not fall into a well defined number of well separated groups. Therefore, the optimal number of clusters was determined so that it would result in (a) a relatively homogeneous number of profiles per group (Fig. 4A), (b) fairly distinct average group profiles (as judged by inspection and by cross-correlation (Fig. 4B), see below), and (c) rather homogeneous groups (as judged by the standard deviation from the average group profile, Fig. 5 and Supplemental Fig. S3). Finally we used the partition around medoids (PAM) algorithm (
) to classify the 925 profiles into 24 groups (a detailed argument is presented in the supplemental information and Supplemental Fig. S2). At a higher level in the dendrogram there are five major clusters (Fig. 4, I–V) that correspond roughly to different times of maximal abundance of each protein during the maturation program.
Many of the 24 groups had simple average profiles, consistent with the proteins they represent having a major function at one stage of phagocytosis and with a relatively simple, linear model of phagosome maturation (Fig. 4C). For example, groups 1–6 had a major sharp peak of red (high spot intensity) and could easily be aligned in such a sequence in association with groups that have a slightly broader peaks (groups 8, 10, 19, and 22). About a fifth of the groups (groups 7, 9, 12, 13, and 17) comprise proteins that appeared at two stages of maturation, including the first time point for four of them. At the other extreme, some groups (groups 11, 16, 18, 20, and 21) comprise proteins present at most time points, perhaps reflecting an unchanging feature of phagosomes identity or a recurrent function. Finally some groups (groups 14, 15, 23, and 24) comprise proteins that appeared and disappeared from the maturing phagosomes in more complex ways, reflecting either a periodic need for a specific function or the existence of parallel maturation pathways in which they act at different times.
From Temporal Profiling to Functional Grouping—
Cluster analysis revealed that the 24 groups are organized into five major clusters that are apparent on the tree structure (Fig. 4A, clusters I–V) and are also visible on the cross-correlation map as square regions of “hot” colors aligned along the diagonal (Fig. 4B, boxes). The groups inside each cluster share a major peak at a common time point, and this reflects the fact that the clusters (from I to V) can be roughly aligned along an axis of maturation in order of appearance of that major peak. This finding is also consistent with a simple, linear model of phagosome maturation as illustrated in Fig. 4C, but more complex alternative pathways of maturation are discussed below.
A detailed description of clusters, groups, and proposed associated functions in maturation is presented in the supplemental information, and the composition of each group is presented in Supplemental Figs. S2-1 to S2-6. Briefly looking at cluster I (groups 1, 2, 8, and 11) and the earliest time point, the presence of group 1 proteins reflects the involvement of the actin cytoskeleton, membrane trafficking, and molecules that bridge the two functions to trigger and carry out uptake; group 11 includes many enzymes, probably reflecting the early establishment of the degradative phases of phagocytosis, and group 8 includes Gα4, suggesting a signaling function during uptake (see below). Cluster II (groups 3, 9, 14, and 18) comprises a diverse collection of proteins likely reflecting the metabolic role of the phagosome as well as other functions that were proposed recently (see below). Cluster III (groups 4, 7, 12, 16, and 20) reflects functions related to late endosomes and multivesicular body formation, including the necessary signaling, cytoskeleton, and membrane trafficking machinery. The proteins representative of cluster IV (from groups 5, 13, 15, 21, and 22) also reflect late endosomal/lysosomal characteristics featuring digestive components as well as the components of trafficking associated with recycling/exocytosis. Cluster V (groups 6, 17, 19, 23, and 24) completes this series, finishing the evolution started in cluster IV, with the presence of factors typical of late endosomes and recycling/exocytosis, but the prominent digestive character of cluster IV is missing.
Cross-correlation Analysis Reveals Further Relatedness between Functional Groups—
Although clustering is important to integrate the information and reveal order in large datasets, a well known and inherent feature of most clustering methods is that sometimes related profiles end up in relatively distant clusters/groups. Therefore, to extend and strengthen the clustering data, we calculated what we call cross-correlation, that is all the pairwise correlation coefficients between the average profiles across all 24 groups. This analysis quantitates and highlights the strength of the relationship between any two average temporal profiles of contiguous or non-contiguous groups of the hierarchical clustering tree. On the resulting heat map matrix (Fig. 4B), the index of correlation between the average temporal profiles of two groups is indicated at the intersection by a color-coded square. Red (on the diagonal) indicates identity, and decreasing similarity is indicated by colors that become closer to green.
The cross-correlation map reveals additional relatedness between groups outside the major five clusters, visible as isolated or small groups of hot squares off the diagonal (Fig. 4B, emphasized by circles, squares, and diamonds). For example, groups 1, 8, 12, and 13 are highly related (Fig. 4B, circles). In addition to the proteins already mentioned above, group 12 also contains signaling and membrane trafficking proteins, and group 13 also contains some enzymes. Similarly groups 10 and 11 from cluster I are closely related to groups 3, 9, and 18 from cluster II, and finally group 10 from cluster I is related to groups 7 and 20 from cluster III. Overall it appears that group 10 has a remarkable position, being a close relative of many groups, both inside and outside its cluster. Altogether these complex patterns of appearance and disappearance of some protein groups during maturation emphasize that the linear maturation program depicted in Fig. 4C is an oversimplification and should be completed by complex cross-talk between endocytic and phagocytic organelles and/or the existence of alternative parallel maturation pathways (
We also focused on proteins for which we had both Western blotting and spot quantification data. For example, the profiles obtained for vacuolin by Western blotting (with an antibody that recognizes both vacA and vacB (
)) and for the vacA spot both fall into group 19, but the profile for vacB is in group 15. These groups are in two different but related clusters (IV and V) and also show a relatively strong cross-correlation (Fig. 4B, see the intersection of groups 19 and 15). The profiles obtained for coronin by Western blotting (group 1) and for the CorA spot (group 11) fall into different groups of the same cluster (cluster I), but these groups are among those with highly cross-correlated average profiles. Similarly the profile obtained for Gβ by Western blotting (group 1) and the corresponding GpbA spot (group 13) are neither in the same group nor the same cluster (clusters I and IV, respectively) but show highest pairwise cross-correlation. These data demonstrate that the use of different methods to obtain the temporal profiles can result in some degree of discrepancy and thus show some of the limitations of our approach. Nevertheless the discrepancies are small and do not really affect our overall conclusions and the concept presented here.
Our analysis of time-dependent proteomics data has allowed us to establish a model of phagocytic mechanisms that will be useful for further functional analysis in Dictyostelium and other organisms. We next aimed to test a prediction of this model as a proof of principle for future investigations.
Heterotrimeric G Protein Function in Early Phagocytosis—
Many signaling pathways are activated when a ligand binds to its G protein-coupled receptor; this receptor-ligand binding activates the downstream heterotrimeric G protein (consisting of one α, one β, and one γ subunit) thus converting the extracellular signal into an intracellular response. Subunits of trimeric G proteins have been reported in phagosomes from mouse (
) are important for chemotaxis and differentiation, but no Gα has been shown to play a role in phagocytosis. Gβ, on the other hand, is required for chemotactic responses and multicellular development as well as phagocytosis (
On our 2D gels, there was a strong signal from Gα4 at the two earliest time points (Fig. 5A, lower panel) suggesting that it may be involved in phagocytic uptake. Likewise group 13 proteins, including Gβ (GpbA), were also present around the beginning of phagocytosis (Fig. 5B). Interestingly both Gα4 and Gβ also peaked at a later time point, perhaps indicating a dual role both during the early uptake phase and in a late maturation phase that might reflect the documented role of Gβ in actin reorganization (
) and thus might be one missing link upstream of Gβ linking the Gαβγ complex to an unknown receptor. We therefore compared phagocytosis in cells deficient in Gα4 subunit (Gα4-null cells) with wild-type cells (Ax2 cells) and, as a positive control, cells deficient in Gβ subunit (Gβ-null cells) with cells expressing a fully functional His-tagged form of the Gβ (Gβ+). To do so, we used a flow cytometry-based particle uptake assay to monitor the uptake kinetics of brightly fluorescent beads (see “Experimental Procedures”). The knock-out strains grew at rates indistinguishable from their respective controls, demonstrating their relative fitness. Both mutant cell lines had a pronounced defect in bead uptake (Fig. 6). During the 120-min course of the assay, the Gα4-null cell population showed an uptake rate of fluorescent beads 50% lower than that of the wild-type cells (Fig. 6A), and the rate of uptake in the Gβ-null cell population was reduced to 40% of that of the Gβ+ cells (Fig. 6C). All the cells in both mutant strains possessed some phagocytic activity and were able to ingest at least one particle at a rate only slightly slower than their respective controls (Fig. 6, B and D, red curves compared with dark blue curves). By contrast, the phagocytic rate in both mutant strains was markedly slower than in the controls when the proportion of cells containing only one bead was plotted against time. In the wild-type and Gβ+ strains, this population appeared and disappeared rapidly (Fig. 6, B and D, light blue curves), whereas in both mutant strains (yellow curves), the peak of cells with only one bead appeared later and disappeared only slowly, illustrating the lower initial rate of uptake of both the first and second bead. We therefore conclude that both Gα4-null and Gβ-null strains are inefficient in an early step of phagocytosis, unambiguously implicating a Gα in this mechanism. A similar functional analysis can now be performed for the proteins of unknown function in these groups with the aim of both validating our approach and strengthening the understanding of the role of heterotrimeric G proteins in phagocytic uptake. The strategy can equally be expanded to any other group of phagosome proteins.
We thank all the laboratory members who contributed thoughts and comments to this investigation as well as help with phagosome purification, in particular Hans-Jörg Warnatz, Eva Neuhaus, and Régis Dieckmann.