Analysis of the Membrane Proteome of Ciprofloxacin-Resistant Macrophages by Stable Isotope Labeling with Amino Acids in Cell Culture (SILAC)

Overexpression of multidrug transporters is a well-established mechanism of resistance to chemotherapy, but other changes may be co-selected upon exposure to drugs that contribute to resistance. Using a model of J774 macrophages made resistant to the fluoroquinolone antibiotic ciprofloxacin and comparing it with the wild-type parent cell line, we performed a quantitative proteomic analysis using the stable isotope labeling with amino acids in cell culture technology coupled with liquid chromatography electrospray ionization Fourier transform tandem mass spectrometry (LC-ESI-FT-MS/MS) on 2 samples enriched in membrane proteins (fractions F1 and F2 collected from discontinuous sucrose gradient). Nine hundred proteins were identified with at least 3 unique peptides in these 2 pooled fractions among which 61 (F1) and 69 (F2) showed a significantly modified abundance among the 2 cell lines. The multidrug resistance associated protein Abcc4, known as the ciprofloxacin efflux transporter in these cells, was the most upregulated, together with Dnajc3, a protein encoded by a gene located downstream of Abcc4. The other modulated proteins are involved in transport functions, cell adhesion and cytoskeleton organization, immune response, signal transduction, and metabolism. This indicates that the antibiotic ciprofloxacin is able to trigger a pleiotropic adaptative response in macrophages that includes the overexpression of its efflux transporter.


Introduction
The plasma membrane is a receptacle to key molecules vital for the cellular integrity, and critical activities such as self-recognition, environment sensing, and communication are mediated by membrane proteins. Substance transport across membranes is fundamental to the maintenance of cellular homeostasis and is accomplished by ATP-binding cassette (ABC) transporters, one of the most conserved family of proteins present in all cell types [1,2]. Under environmental changes, cells respond by altering their gene expression to restore homeostasis and survive. This is well illustrated in cells exposed to potentially toxic drugs in which diverse cytoprotective mechanisms can be set off conferring to them a resistant phenotype. A commonly observed biological strategy is the reduction of the intracellular drug accumulation by overexpression of multidrug efflux transporters of the ABC superfamily [3]. Yet, drug-resistant cells may present other less evident mechanisms operating synergistically. The fact that drug-resistant cell lines present higher levels of resistance than their transfected counterparts suggests indeed that other uncharacterized functions might contribute to the resistance [3]. In this respect, profiling of protein abundance may help to uncover the extent of the impact of chronic exposure to a toxic drug.
Previous work from our laboratory has shown that the fluoroquinolone antibiotic ciprofloxacin can select for a resistant phenotype in J774 mouse macrophages that had been exposed chronically to high concentrations of this drug [4]. This phenotype is characterized by a reduced accumulation and an increased efflux of the drug that we could attribute to an increase in the expression of the multidrug transporter Mrp4 encoded by the Abcc4 gene located on chromosome 14 [5]. Molecular cytogenetic experiments showed that this overexpression is linked to Abcc4 gene overrepresentation, grading from a partial trisomy of chromosome 14 at the first step of selection, to low-level amplifications of Abcc4 locus, up to high-level amplification of Abcc4 as homogeneous staining region (hsr), inserted on 3 different derivative chromosomes [6].
To examine whether exposure and resistance to the toxic effects of ciprofloxacin could be associated with other, unanticipated changes in the membrane proteome, we have now performed a global quantitative analysis using stable isotope labeling with amino acids in cell culture (SILAC) technology to compare the proteins expressed in membranes isolated from ciprofloxacinresistant (CIP-R) J774 murine macrophages vs. their wild-type (WT) counterparts. SILAC is a potent technique for direct qualitative as well as quantitative comparison of proteomes [7]. It has been successfully used to evidence biological pathways altered in cisplatin-resistant cells [8]. Here, we have focused our analysis on enriched membrane fractions isolated from sucrose gradient interfaces, which allowed us to confirm a large abundance of Mrp4 in CIP-R cells. Yet, we also evidenced a modulation in the expression of multiple surface and intracellular membrane proteins with variety of functions. The data therefore suggest that along with the acquisition of a targeted-resistant mechanism (Mrp4 overexpression), cells surviving the chronic stress imposed by high ciprofloxacin concentrations simultaneously modify multiple pro-survival pathways to reach a new homeostatic equilibrium. In a broader context, they demonstrate the interest of the SILAC approach for unraveling pleiotropic changes occurring in cells upon drug exposure.

Results and Discussion
Membrane Proteins Preparation Figure 1 shows the isolation of samples enriched in plasma membrane by the discontinuous sucrose gradient method. The multipass plasma membrane transporter MRP1 was used as an indicator for enrichment in cell surface proteins and MDCKII-MRP1 cells overexpressing this protein were used to set up the fractionation conditions and the optimized protocol was applied to J774 macrophages. MRP1 levels were monitored by Western blotting of the fractions isolated from 5 sucrose density-interphases (F1 to F5). The mitochondrial marker prohibitin was used to evaluate the purity of each fraction, given that this organelle is the most common contaminant in this kind of preparations [9]. When using MDCKII-MRP1 cells, protein samples collected from interphases F1 to F3 showed increased MRP1 levels with a concomitant rise in prohibitin. Samples from WT-J774 macrophages presented an enrichment of Mrp1 in the equivalent fractions, with F1 and F2 displaying the lowest mitochondrial contamination. Based on these results, membrane samples isolated at the sucrose density-interphases F1 and F2 were selected for comparative and quantitative proteomic analysis, using pooled F1 (F1 WT +F1 CIP-R ) and F2 (F2 WT +F2 CIP-R ) membrane samples as depicted in Figure 2. A possible drawback in this experimental design is that membranes samples for each WT and CIP-R cells, had to be prepared and isolated independently (because of the intensive labor and instrumental constraints), and mixed only prior to the separation by SDS-PAGE. We can therefore not exclude slight variations in the purification conditions.

Protein Identification by LC-MS/MS
After LC-MS analysis of the SDS-PAGE gel bands, a total of 122,542 and 111,580 MS/MS spectra were acquired from the pooled membrane fractions F1 and F2, respectively. For each pooled fraction, after protein database searching and automatic validation using TPP's PeptideProphet and ProteinProphet tools, 45,947 and 49,268 peptides were identified, accounting for 1061 and 1144 proteins in pooled membrane fractions F1 and F2, respectively. At peptide level, PeptideProphet applied a 70% confidence threshold for positive identifications. During the further processing of the ProteinProphet results, EXCELH exported protein lists of individual LC-MS experiments were reorganized using in-house written VBA-scripts. Only protein hits with probabilities higher than 0.95 were retained, and redundant hits, appearing in multiple, often adjacent gel bands, were grouped. No decoy proteins were identified with more than 2 unique peptide sequences. Therefore, a list of proteins with at least 3 unique peptides identified within an individual LC-experiment was generated reflecting very significant and unambiguously identified proteins (Table S1). A total of 651 and 735 proteins were unambiguously identified in the pooled F1 sample and pooled F2 sample, respectively, out of which 486 were present in both fractions, leading to a total number of 900 unique mouse proteins recognized. Additionally, 386 and 381 proteins represented by only 1 or 2 peptides were recovered in pooled fraction F1 and F2 with a respectively false discovery rate of 3.5% and 4.5%, correlating well with the 0.95 significance threshold used by ProteinProphet (Table S2).
To assess the contribution of proteins from plasma membrane and other cellular organelles to the total protein hits, an analysis according to the cellular compartment and biological function was performed based on the protein's Gene Ontology (GO) terms. From all identified proteins, 17% were plasma membrane proteins. Proteins from the endomembrane system accounted for 44% of the sample proteins (15% from ER/Golgi, 5% from ribosomes, 7% from mitochondria, 8% from endosomes/lysosomes, and 9% from nucleus) whereas 17% were assigned as unknown membrane proteins. Finally, 6% of the sample corresponded to structural proteins of the cytoskeleton, 7% to cytosolic and 4% to extracellular proteins (Figure 3, left). Based on their biological function ( Figure 3, right), most of these proteins are involved in transmembrane or intracellular transport mechanisms (27%) or metabolic processes (30%; among which 17% in the metabolism of proteins; 5%, of nucleic acids; 4%, of carbohydrates; and 4%, of lipids).

Quantitative SILAC-based Analysis
Final protein abundance ratios were calculated by averaging the logarithmic based 2 transformed L/H ratios of a particular protein scattered in multiple, often adjacent, gel bands. After normalization towards the median, the resulting histograms fit a Gaussian distribution for both pooled membrane fractions ( Figure 4). Therefore, albeit that pooling of light (CIP-R) and heavy (WT) labeled proteins was done after purification of the membrane fractions, the mixture was composed of a 1:1 proportion, validating the experimental method and parameters. In pooled fraction F1, we detected 61 unambiguously identified proteins that show a significant (P-value ,0.05) difference in protein abundance ratios between WT and CIP-R cells (Table 1). Similarly, pooled fraction F2 data demonstrated 69 proteins with altered protein abundance between both proteomes (Table 2). Several proteins display a significant change in expression level in only one of these two fractions. These differences are expected in our approach, where pooled F1 and F2 fractions differ in their enrichment in plasma membrane and constitute therefore partially different proteomes.
When examining the function of the differentially expressed proteins (Figure 4), we observed a small enrichment of proteins involved in cell adhesion, motility and immune response (11%, vs. 5% of identified proteins) in comparison with the totality of proteins identified. In contrast, proteins involved in structural  Left: WT-J774 and CIP-R cells were cultured in media containing 13 C 6 -Lys/ 13 C 6 -Arg and 12 C 6 -Lys/ 12 C 6 -Arg, respectively. Following isolation of membrane samples for each strain independently, equal amounts from each matching fraction were pooled (F1 wt +F1 CIP and F2 wt +F2 CIP ) and then separated by SDS-PAGE. Right: Lanes of pooled sample fractions were first cut in sections (A-K for pooled F1 and A-J for pooled F2) and further sliced in thinner, numbered gel plugs which were processed for Trypsine in-gel digestion as described in the experimental procedures. For pooled F1 gel line, section A includes gel plugs from 1 to 8, B 9 to 14 functions were modulated in lesser extent (3%, vs. 7% of identified proteins associated with cytoskeleton).

Proteins Modulated Upon Chronic Exposure to Ciprofloxacin
Overexpression of the multidrug transporter Abcc4 involved in ciprofloxacin efflux. In both membrane fractions, the multidrug transporter Abcc4/Mrp4 (Q3TZN9) belonging to the ABC superfamily [10] was the most abundant protein in CIP-R cells (see Figure 5 for an example of a MS/MS spectrum which contributed to the identification of Mrp4). This observation is coherent with our previous observation that Mrp4 is the main ciprofloxacin transporter in these cells and is markedly overexpressed at both the protein and mRNA levels in CIP-R cells [5]. As we were interested in multidrug resistance, we also looked for other ABC multidrug transporters. Only Abcb1b (Mdr1) (one of the two murine P-gp isoforms), and Abcc1 were detected, but without any difference of expression between the two cell lines (See Tables S1 and S2). Abcc2 (Mrp2), for which we have already demonstrated an overexpression in resistant cells, was not detected here, probably due to its low abundance [5]. Mrp4 has a natural function in the export of cyclic nucleotides like cAMP and proinflammatory mediator leukotriene [11,12]. Massive production of Mrp4 may therefore have an impact on intra-en extracellular signaling as will be demonstrated below.
Overexpression of p58 IPK /Dnajc3 and evaluation of resistance to ER stress. In both membrane fractions, P58 IPK /Dnajc3 (Q91YW3) is also strongly more abundant in CIP-R macrophages. This was confirmed by Western-Blot ( Figure 6). The corresponding gene is found 250 kB downstream of the Abcc4 gene, located on mouse chromosome 14E4, suggesting that overproduction of the corresponding proteins is due to gene amplification of a chromosomal fragment. This hypothesis has been recently confirmed by FISH experiments demonstrating a co-amplication of these two genes [6]. P58 IPK /Dnajc3 is a member of the DnaJ/Hsp40 (heat shock protein 40) cochaperones family [13] and is considered to be an ER stress inducible chaperone and attenuator of the unfolding protein response (UPR [14,15]). The massive expression of Mrp4 at the plasma membrane might generate a burden to the ER where membrane protein folding and maturation takes place. We therefore compared the response to ER stress in the two cell lines by measuring the expression of the ER chaperone Bip/Grp78 also known as Hspa5 heat shock protein 5 [16] in control conditions or after 24 h exposure to 0.2 mM ciprofloxacin or 0.5 mg/ml of tunicamycin, an UPR-stress inducer. Our SILAC analysis did not reveal any change in the expression of this protein (see Table S1) in standard conditions of culture (i.e. control medium for WT cells and medium added by 0.2 mM of ciprofloxacin for CIP-R cells), which was further confirmed by immunoblotting of whole cell lysates ( Figure 6) or of membrane protein samples (not shown) from both cell lines. Yet, both ciprofloxacin and tunicamycin triggered a sustained over-expression of Grp78 in WT cells, but only tunicamycin in CIP-R cells ( Figure 7). The absence of effect of ciprofloxacin in CIP-R cells could easily be attributed to the reduced cellular accumulation of the drug resulting from the overexpression of its efflux transporter [4]. Yet, the absence of protection against tunicamycin-induced stress in CIP-R cells rather suggests that the overexpression of the two proteins, Mrp4 and Dnajc3, is coincidental and related to their close location on the chromosome [6]. Most likely, P58 IPK /Dnajc3 does not play a particular role in the resistant phenotype. This needs however to be further explored by examining other proteins involved in the ER stress cascade.
Moreover, P58 IPK /DNAJC3 protects cells against apoptosis by preventing the activation of NF-kB signal cascades [17]. Of interest, ciprofloxacin has been shown capable of causing cell death by apoptosis in specific models (see [18] [25] was down regulated. Modulation of lipid synthesis. One of the physiological functions attributed to Mrp4 is the transport of lipid mediators like leukotrienes and prostaglandins [11,26]. In this context, our analysis also highlights changes in the expression of proteins involved in lipid metabolism, which could have significant impact in the composition of the membranes and important consequences for intracellular transduction activities. In particular, the lipid phosphate phosphatase 1 (LPP-1or PAP-2a [Q61469]) and the  [27,28], 2 proteins participating in the regulation of the activity of bioactive lipids and mediators of the intracellular phosphoinositide-based signaling [29], were upregulated. The low-density receptor protein LDRL [P35951] was upregulated as well [30]. On the contrary, two lysosomal hydrolases acting on lipids, namely glucosylceramidase [P17439] [31] and acid ceramidase [Q9WV54] [32] as well as the glycosyltransferase beta-1,4 N-acetylgalactosaminyltransferase 1 (GalNAc-T; Q09200) involved in the biosynthesis of gangliosides GM2, GD2 and GA2 [33] were downregulated in CIP-R cells. Some of these changes could be attributed to reduced cAMP levels, as for example for acid ceramidase [34].
Modulation of intracellular protein transport and recycling. Driving increased amounts of Mrp4 at the cell surface may require modulation of the intracellular pathways of protein transport and/or recycling. Although not optimized to detect proteins associated to intracellular membranes, our proteomic analysis detected changes in the expression of proteins implicated in intracellular trafficking and recycling in CIP-R macrophages. We found RAB11 family interacting protein 1 class 1 (Rab11-FIP1; Q9D620) positively modulated. This protein had been proposed to regulate Rab11 localization and to recruit additional factors to different endocytic compartments [35]. In contrast, the Rab4-interacting protein (RabIP4; Q8BIJ7), a protein that in conjunction with Rab4 seems to be involved in early endosomal traffic [36] was found down regulated in CIP-R cells. Apparently, RabIP4 could provide directionality to a backward traffic from the recycling endosomes to sorting endosomes [37]. The upregulated serine/threonine kinase lemur tyrosine kinase 2 (LMTK2; Q3TYD6) is involved in the transition  of membrane vesicles from early endosomes to recycling endosomes and interacts with the actin-based retrograde motor protein myosin VI [38] involved in the movement of those vesicles [39]. Two Dnajc13 homologues, different from p58IPK (RME-8 [Q8BNL1] and Q3UQJ), involved in endosomal trafficking [40], were less expressed in CIP-R cells. In cells knock-out for RME-8, the level of cargos, such as the epidermal growth factor receptor (EGFR), is decreased while receptors that primarily recycle to the plasma membrane remain unaffected [41]. Two other proteins with a role in membrane trafficking were also modulated in the resistant cells, namely copine [Q8C166] [42,43] and myoferlin [Q69ZN7] [44]. The link between these changes and functionality of Mrp4 at the cell membrane remains however to be established. Modulation of adhesion and motility. Several integrins or associated proteins were found modified in our proteomic experiment. Thus, intregrin aL [P24063] was upregulated in CIP-R cells, while a9 [Q3UT74] and the leukocyte surface antigen CD47 [Q61735] (45) were downregulated. In addition, various adhesion molecules implicated in tumorigenesis were found up-regulated in ciprofloxacin-resistant macrophages. These include neuropilin-1 [P97333], which is implicated in cell proliferation, survival, and migration and protects cancer cells from apoptosis [46], the cell adhesion molecule NrCAM [Q810U4] that promotes cell growth and motility [47,48], the extracellular matrix component and tumor progression marker osteopontin [P10923] [49], and the metalloproteinase-disintegrin ADAM8 [Q05910] [50].
Alteration of lysosomal proteome. Although the samples were selected to be enriched in pericellular membrane proteins, we also detected proteins associated to organelle membranes. In particular, we found modifications in the expression of a series of lysosomal proteins. Besides the lysosomal hydrolases acting on lipids mentioned above, many other proteins were found overexpressed in CIP-R cells, mainly: (i) precursor peptides of cathepsin A [P16675], D [Q3U651] and L1 [P06797] that intervene not only in proteolysis but also in inflammatory processes and apoptosis [51] (ii) legumain [O89017], a member of the C13 family of cysteine proteases, also found highly upregulated in many murine and human tumors [52] and (iii) prosaposin [Q61207], the precursor of lysosomal saposins A, B, C and D that has been associated with cancer [53,54] and is described as having growth, migration, and invasion promoting activities and also anti-apoptotic effects [55][56][57]. It is established that lysosomal enzymes are first translated as proenzymes that become proteolytically activated within the lysosomes. Our GeLC approach therefore is able to separately identify and quantify the proteins processing states. Focusing on cathepsin B (P10605) for example, our data indicate a significant up-regulation of the cathepsin B precursor (MW 37.3 kDa), whereas the processed enzyme (MW 27.5 kDa) shows no regulation between WT and ciprofloxacin resistant cells ( Figure S1 and Table S3). Averaging the ratios for the different peptides, however, results in a clear but not significant up-regulation of cathepsin B. Similar results are obtained for cathepsins L1 and D. In addition, six out of the eight (A to H) peripheral V1 subunits of the V-type ATPase [58] were significantly downregulated in CIP-R. Of interest, we noted previously that the basal level of ATP was increased in CIP-R cells [4], possibly to insure the increased need in energy needed for Mrp4 activity.
Alteration of cell signaling. A series of receptors playing a role in macrophage differentiation or activation were expressed to higher levels in CIP-R cells. This is the case for (i) the cytokine receptor common subunit b (CSF2RB, CD131) which is a shared component of Type I cytokine receptors (IL 3, IL 5 and GM-CSF) [59], (ii) the a and b chains of tumor necrosis factor receptor (CD40; P27512), a member of the tumor necrosis factor receptor (TNF) receptor family, (iii) the insulin-like growth factor 1 receptor (IGR-1 [Q60751] [60], a survival factor of malignant cells [61], and (iv) single Ig IL-1 related receptor (Sigirr) or TIR8 [Q9JLZ8],  which regulates cancer-related inflammation and autoimmunity [62]. On the contrary, the Toll-like receptor 7 (TLR 7; P58681) was the lower expressed protein in resistant cells, in both analyzed fractions (see also Figure 6 for validation by western-blot). This could be consecutive to the dowregulation of Sigirr, a negative regulator of the IL-1 receptor/TLR system [62]. In the context of this study, an interesting characteristic of this protein is that it recognizes nucleosides and nucleotides [63] as does Mrp4.
Among proteins involved in differentiation, we found an increased expression of the histocompatibility H-2 class I antigen chain precursors (P01902). Interestingly, a proteomics study reports on the induction of major histocompatibility complex (MHC) class I molecules in activated macrophages [64]. On the contrary, we evidenced a reduction in the expression of a F4/F80 receptor [Q61549], a specific macrophage cell-surface marker for murine macrophages [65] and of the precursor of the macrophage scavenger receptors (MSR) I and II [P30204, which are expressed during differentiation of monocytes into macrophages [66].
Modulation of other cell surface proteins. Among other surface proteins, we also detected in CIP-R cells a down-regulation of Mpeg-1 [A1L314], a macrophage porin-like protein [67], and an up-regulation of semaphorin-4D (CD100 [O09126]) [68], of collectin-12 [Q3TYT8], a scavenger receptor [69], and of the C-C chemokine receptor type 1 CD191 [P51675] [70]. Peculiar outcomes were the overexpression of the amyloid beta (A4) precursor protein (App [P12023]; validated by Western-Blot, see Figure 6), the MLV related proviral Env polyprotein precursor [P10404], and the IgE binding protein or intracisternal A-Particle (IAP [P03975]). The IAP are encoded by retrovirus-like mobile elements present in the mouse genome [71] where they can transpose and act as mutagenic agents [72]. For instance, it has been shown that P-gp overexpression in a murine leukemic tumor cell line (gene Abcb1a) is caused by gene amplification and also transcriptional activation due to the insertion of a IAP in the promoter region of the gene [73]. Whether this is also the mechanism having led to gene amplification in our model needs to be determined experimentally. We also detected in CIP-R cells an increased amount of p42-MAPK [P63085], a serine/threonine kinase that controls a wide range of cellular activities [74]. Yet, peptides were located in bands corresponding to a much higher molecular weight than that expected for the denatured p42-MAPK polypeptide (band 6, .150 kDa).

Conclusion
This study is, to the best of our knowledge, the first one to perform a SILAC analysis on purified membrane fractions of eukaryotic cells, constituting a potentially useful technological advance that may have several applications in biology or pharmacology.  Thus, the analysis performed here clearly shows that exposure of J774 macrophages to high concentrations of ciprofloxacin to obtain cells that resist to drug concentrations that would kill parent cells, induces pleiotropic alterations in protein expression. This observation is in the line of recent proteomic studies of drugresistant cell lines that all highlight impressive modifications in the profile of protein expression, most often related to a diversity of metabolic pathways [8,[75][76][77]. In the present case, the most obvious change to explain the resistance phenotype is the overexpression of the ciprofloxacin efflux transporter, a mechanism observed for many other cytotoxic drugs. Quite intriguingly, however, co-amplification of P58 IPK /Dnajc3 does not seem to contribute to resistance. Many of the other differentially expressed cell surface receptors and cell surface signaling elements evidenced in CIP-R macrophages are suggestive of activated macrophages, or contribute to cell motility, adhesion, cytoskeleton, immune response, activation of proinflammatory cytokines, angiogenesis, differentiation, lipid metabolism, and cell proliferation. Thus, virtually all of the regulated surface proteins in CIP-R cells have been linked to malignancies and/or in a broader context, conditions promoting cell growth and motility. It remains nevertheless to be determined whether these changes are contributing to resistance or are simply secondary to alterations in regulatory pathways, for example as a consequence of altered cAMP homeostasis. This would represent a piece of work by itself for each group of modified proteins associated with a specific biological function. Yet, confirming the significance of our results, transcriptomic studies (including microarray analysis) of human lymphocytes exposed to supratherapeutic concentrations of ciprofloxacin similar to those used here, also revealed modulations in mRNA of proteins belonging to major gene programs (like signal transduction factors, cytokines, adhesion and apoptosis related molecules), which share many commonalities with our own data [78,79].
In a broader context, the fact that many of the proteins altered in their expression upon exposure to ciprofloxacin were similar to those modified with other cytotoxic agents (including chemotherapeutic agents), indicates that they are part of a global adaptative response to xenobiotics.

Cell Culture
J774 mouse macrophages (referred to as WT cells) were grown in RPMI-1640 medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum in 5% CO 2 atmosphere, as described previously [80]. J774 macrophages resistant to ciprofloxacin (referred to as CIP-R cells) were obtained as described earlier [4]. In brief, wild-type cells were cultivated in the presence of increasing concentrations of ciprofloxacin (from 0.1 mM to 0.2 mM) for about 50 passages (approx. 8 months), after which cells were maintained in the presence of 0.2 mM ciprofloxacin. MDCKII-MRP1 cells [81] were received from Prof. P. Borst (The Netherlands Cancer Institute, Amsterdam, The Netherlands), and grown in DMEM medium supplemented with 10% fetal bovine serum in 5% CO 2 atmosphere.

Metabolic Labeling with Stable Isotopes
Before performing the SILAC experiment, a time course of incorporation of labeled amino acids was performed to establish the minimum time required for complete labeling of proteins. Both WT-and CIP-R macrophages were grown in the double labeled [U-13 C 6 ] L-Lysine, [U-13 C 6 ] L-Arginine SILAC medium (+6 Da shift) in 20 cm 2 dishes seeded at 1610 5 cells/cm 2 (RPMI-1640 SILAC TM Protein Identification and Quantitation Media Kit, Invitrogen). Whole cell lysates were prepared at various time points in 10 mM Tris-HCl, pH 7.4 plus 0.1% SDS and random peptides were analyzed by MS to evaluate incorporation of the heavy amino acids. Two passages were sufficient for the WT-J774 cells to reach complete incorporation of the heavy isotopes. SILAC labeling was thus performed with WT cells, while CIP-R cells were grown in medium with light amino acids.

Preparation of Enriched Plasma Membrane Fractions
Cell monolayers (10 Petri dishes of 165 cm 2 , seeded at approx. 1.75610 5 cells/cm 2 ) were rinsed with cold PBS then detached with policeman and collected in the same buffer by centrifugation at 2806g for 10 min at 4uC. Sets of 5 monolayers were pooled and resuspended in 5 mL of 10 mM Tris-HCl, pH 7.4 plus proteaseinhibitor cocktail-EDTA free (one tablet/10 ml; F. Hoffmann-La Roche Ltd Diagnostics Division, Basel, Switzerland). Cell suspensions were transferred to a Dounce B homogenizer (glass/glass, tight pestle) and stroked 25 times. The homogenates were centrifuged at 6006g for 10 min at 4uC. The post-nuclear supernatants were layered on the top of discontinuous sucrose gradients designed to separate membranes from the other main cell constituents [82]. Sucrose solutions were prepared in 10 mM Tris-HCl, pH 7.4 plus 0.5 mM phenylmethanesulphonyl fluoride (PMSF) and stored overnight at 4uC. Sucrose step gradients were assembled on ice, layering cold solutions of sucrose using a syringe with long needle in the following order: 1.5 ml of 1.10 density solution, 1 ml of 1.13 density solution, 1 ml of 1.15 density solution, 2 ml of 1.17 density solution and 2 ml of 1.19 density solution. Gradients were ultracentrifuged in a Beckman SW40 rotor at 280.0006g for 2 h at 4uC. The resulting visible bands at each interphase (Fig. 1A) constituted 5 different membrane fractions (F1 to F5). The five fractions were collected on ice and diluted to 6 ml with cold 0.1 M Na 2 CO 3 , pH 11 and centrifuged in a Type 50 Ti rotor at 100,0006g for 30 min at 4uC. The membrane pellets were then resuspended in 10 mM Tris-HCl,  pH 7.4 plus protease-inhibitor cocktail-EDTA free and aliquots were removed for protein content determination using the bicinchoninic acid protein assay (Pierce, Rockford, IL). Membrane samples were then flash-frozen in an ethanol-dry ice bath and stored at -80uC for downstream analysis.

SDS-PAGE and In-Gel Digestion
Membrane fractions F1 and F2 were separately collected for each cell line ( Fig. 2A). To extract proteins from the membrane fractions, samples were resuspended in the extraction/solubilization buffer (7M urea, 2 M thiourea, 2% C7BzO and 5 mM tributyl phosphine [TBP]). Matching fractions (F1 wt +F1 CIP and F2 wt +F2 CIP ) were mixed at concentration ratios of 1:1 ( 12 C-light: 13 C-heavy) and subjected to separation by SDS-PAGE as a single sample. The mixed fractions are denominated pooled F1 and pooled F2, respectively. After separation by SDS-PAGE, the pooled F1 and F2 membrane fractions complete lanes were sliced into small gel bands (Fig. 2B). Each band was in-gel digested as previously described [84]. In short, gel bands were first destained by two consecutive washes in 150 ml buffer containing 200 mM ammonium bicarbonate in 50% (v/v) acetonitrile in water for 30 min at 30uC. After drying the gel pieces in a Speedvac (Thermo Savant, Holbrook, NY), 10 ml trypsin solution (0.002 mg/ ml in a 50 mM ammonium bicarbonate buffer solution) was added and allowed to be absorbed by the gel for 40 min on ice. Gel bands were completely immersed by adding additional buffer solution and incubated at 37uC. After overnight digestion, tryptic peptides were extracted by 2 consecutive washes in 50 ml buffer (60% (v/v) acetonitrile, 0.1% (v/v) formic acid in water). Pooled peptide extractions were subsequently dried in a speedvac, dissolved in 15 ml 2% (v/v) acetonitrile, 0.1% (v/v) formic acid in water and stored at -20uC for further analysis.

Liquid Chromatography Electrospray Ionization Fourier Transform Tandem Mass Spectrometry (LC-ESI-FT-MS/ MS)
Tryptic peptides from each gel band were individually analyzed via a fully automated LC-MS/MS setup. Peptides were first separated on an Agilent 1200 chromatographic system (Agilent, Santa Clara, CA) and on-line measured on a LTQ-FT Ultra mass spectrometer (Thermo Fisher Scientific, Waltham, MA). Peptides in 5 ml solution were initially concentrated and desalted on a Zorbax 300SB-C18 trapping column, 5 mm60.3 mm, at a 4 ml/ min flow rate using a 2% (v/v) acetonitrile, 0.1% formic acid in water buffer. After valve switching, eluted peptides were separated on a Zorbax 300SB-C18 analytical column, 150 mm675 mm (Agilent), by a 50 min linear gradient ranging from 2% (v/v) to 80% (v/v) acetonitrile, 0.1% formic acid in water at a 250 nl/min flow rate. Eluting peptides were analyzed on-line in the mass spectrometer via nano-ESI. Protein digests from fraction F2 were measured using the LTQ-FT Ultra nano-ESI source and applying a 1.1 kV voltage on PicoTip emitters (New Objective, Woburn, MA). For fraction F1 digests, LC-effluent was directly coupled to a Triversa NanoMate ESI source (Advion, Ithaca, NY), working in nano-LC mode and equipped with D-chips whereon a 1.55 kV voltage was supplied. The LTQ-FT Ultra mass spectrometer was tuned and calibrated with caffeine, MRFA (met-arg-phe-ala peptide) and UltraMark before measurement. The Fouriertransform ion cyclotron resonance (FT-ICR) mass analyzer acquired MS scans at 100,000 resolution during the LC separation. The 3 most intense precursor peptides for each MS scan were automatically selected and fragmented by the LTQ ion trap mass analyzer, and after 2 occurrences, precursor masses were excluded for 90 sec.

Data Processing and Analysis
Raw LC-MS/MS data were analyzed with the Sequest database-searching algorithm [85] implemented in the Bioworks v3.3.1 software (Thermo Fisher Scientific). MS/MS spectra were searched against the mouse uniprot protein database concatenated with a shuffled decoy database generated by the Decoy Database Builder software tool [86] Propionamide on cysteines (+71.037114) and SILAC labels in arginine and lysine (+6.020129) were allowed as variable modifications. Mass tolerances were set to 10 ppm and 0.5 amu for peptide mass and fragment mass respectively. Two miscleavages and only b-and y-ions were considered. Raw LC-MS/MS data and Sequest result files were loaded into the TPP software v3.4 (Seattle Proteome Center [87]). Peptide identifications were validated with PeptideProphet filtering out peptide hits with probabilities below 0.70. Retained peptide identifications were subsequently quantified via the ASAPRatio tool using a fixed scan range for Light and Heavy labeled peptides and a 6.02013 Da label mass for both arginine and lysine. Mass tolerance was set to 0.01 Da. For each individual LC-run, peptide identifications and quantifications were grouped via the ProteinProphet tool and afterwards exported as excel files for further processing. All peptide and protein L/H ratios were manually controlled and if necessary adjusted. Manually validated protein ratios for each individual LC-run were grouped for redundant protein identifications via in-house written VBA (Visual Basic Scripting Edition)scripts. Only proteins with a probability of 0.95 or higher were retained in the script. Final protein abundance ratios were subsequently calculated by averaging the logarithm base 2 protein ratios of an individual protein present in multiple gel bands. The final proteins ratios were normalized towards the median and their p-values calculated based on Z-statistics as previously described [88]. Unambiguously identified and quantified proteins from pooled fractions F1 and F2 were submitted to the online available DAVID Bioinformatics tool [89] and analyzed via their respectively GO terms for cellular localization and biological process. All MS data are available on the EBI's PRIDE mass spectral depository (accession numbers 17116-17117). All samples were run in duplicate, as previously done by other groups using the SILAC technique [8,90]. A main advantage of this approach resides indeed in the fact that it allows a relative quantification of proteins in a single sample obtained by mixing material collected from two different experimental conditions (in our case, WT and CIP-R cells), reducing the need of replicated samples. Moreover, quantitation is based on ratio's obtained from different peptides per protein for which we used stringent conditions to accept a ratio as significant.