Quantitative Proteomics Analysis of Macrophage Rafts Reveals Compartmentalized Activation of the Proteasome and of Proteasome-mediated ERK Activation in Response to Lipopolysaccharide^,

Abstract

Lipopolysaccharide (LPS), a glycolipid component of the outer membrane of Gram-negative bacteria, is a potent initiator of the innate immune response of the macrophage. LPS triggers downstream signaling by selectively recruiting and activating proteins in cholesterol-rich membrane microdomains called lipid rafts. We applied proteomics analysis to macrophage detergent-resistant membranes (DRMs) during an LPS exposure time course in an effort to identify and validate novel events occurring in macrophage rafts. Following metabolic incorporation in cell culture of heavy isotopes of amino acids arginine and lysine ([¹³C₆]Arg and [¹³C₆]Lys) or their light counterparts, a SILAC (stable isotope labeling with amino acids in cell culture)-based quantitative, liquid chromatography-tandem mass spectrometry proteomics approach was used to profile LPS-induced changes in the lipid raft proteome of RAW 264.7 macrophages. Unsupervised network analysis of the proteomics data set revealed a marked representation of the ubiquitin-proteasome system as well as changes in proteasome subunit composition following LPS challenge. Functional analysis of DRMs confirmed that LPS causes selective activation of the proteasome in macrophage rafts and proteasome inactivation outside of rafts. Given previous reports of an essential role for proteasomal degradation of IκB kinase-phosphorylated p105 in LPS activation of ERK mitogen-activated protein kinase, we tested for a role of rafts in compartmentalization of these events. Immunoblotting of DRMs revealed proteasome-dependent activation of MEK and ERK specifically occurring in lipid rafts as well as proteasomal activity upon raft-localized p105 that was enhanced by LPS. Cholesterol extraction from the intact macrophage with methyl-β-cyclodextrin was sufficient to activate ERK, recapitulating the LPS-IκB kinase-p105-MEK-ERK cascade, whereas both it and the alternate raft-disrupting agent nystatin blocked subsequent LPS activation of the ERK cascade. Taken together, our findings indicate a critical, selective role for raft compartmentalization and regulation of proteasome activity in activation of the MEK-ERK pathway.

Lipopolysaccharide (LPS),1 a glycolipid component of the outer membrane of Gram-negative bacteria, is a potent initiator of the inflammatory response of the macrophage (1–4). LPS signals through Toll-like receptor 4 (TLR4) (5), activating a variety of signaling molecules including protein-tyrosine kinases, Rho GTPases, IκB kinase (IκK), and the MAPKs ERK, p38, and JNK (6). This results in the downstream activation of transcription factors, such as NF-κB and AP-1, and the consequent induction of proinflammatory cytokines, including tumor necrosis factor-α and interferon-β (7, 8). LPS has been linked to a number of diseases, including sepsis, acute lung injury, asthma, and atherosclerosis (9–11); thus, the signals it elicits in the host cell, although incompletely described, are of great biomedical significance.

The LPS receptor TLR4 and its proximal signaling adapter molecules (e.g. MyD88) are recruited to and assembled in the lipid raft microdomain of leukocytes following LPS exposure (12). Lipid rafts are cholesterol-enriched membrane microdomains thought to dynamically organize cellular signaling events triggered by extracellular stimuli (13, 14). Signaling receptors and other interacting molecules use the lipid raft as a platform to selectively assemble and facilitate signal transduction (12–16). The dynamic remodeling of lipid rafts during proximal LPS signaling is believed to actively recruit non-raft proteins to rafts as well as to enrich selected raft-resident proteins. Moreover rafts are reported to regulate downstream proinflammatory cellular functions triggered by LPS (17, 18). Rafts, which may, as an approximation, be biochemically isolated in the form of detergent-resistant membranes (DRMs) (13), thus offer an opportune “window” to study proximal signaling events regulating the induction of the LPS signal. Although several receptor-level regulators of LPS signaling have been reported already, the current “state-of-the-art” understanding of LPS signaling is almost certainly incomplete and oversimplified because of the limitations of traditional biochemical and genetic approaches.

Proteomics provides a global, relatively unbiased approach for examining changes in protein expression and thus offers the opportunity for discovery of novel signaling events and targets in biology. We and others have used proteomics analysis to identify novel signals in the leukocyte triggered by LPS exposure (12, 19). In the current work, we undertook complementary qualitative and quantitative stable isotope labeling with amino acids in cell culture (SILAC)-based proteomics approaches to study macrophage DRMs during an LPS exposure time course. We reasoned that the former approach would globally identify both basal raft-resident proteins and those non-raft-resident targets that are newly recruited to rafts upon LPS exposure and that this would be complemented by the identification, using the latter approach, of quantitative enrichment of select raft-resident proteins during LPS exposure. Unsupervised network analysis of the proteomics data set revealed marked representation of the ubiquitin-proteasome system (UPS) in macrophage DRMs, including most of the subunits of the 26 S proteasome. Expression of several proteasome subunits and ubiquitin ligases in DRMs was sensitive to LPS treatment, suggesting that LPS may induce remodeling of the proteasome in rafts and respecification of its targets. The 26 S proteasome, a large multiprotein complex that degrades proteins that have been ubiquitin-tagged by ubiquitin ligases, is reported to regulate activation of both ERK and NF-κB by LPS (20, 21). Generally described as cytoplasmic in location, a few reports have indicated that the proteasome may be membrane-associated, although what, if any, role rafts play in compartmentalization of the proteasome and its function remains unclear (22–25). One group has suggested that LPS may directly bind specific proteasome subunits in macrophage membranes and activate the proteasome, although neither rafts nor activity of the proteasome within membrane fractions was examined (26).

Herein in validation of our proteomics findings, we confirmed that LPS selectively activates the 26 S proteasome in macrophage DRMs. To our knowledge, this is the first direct demonstration of proteasome function in rafts. Moreover we implicate the proteasome in raft-compartmentalized activation of ERK by LPS through its upstream degradation of IκK-phosphorylated p105 (20, 21, 26), and we report that the proteasome-dependent signaling cascade to ERK requires raft integrity. Extending the importance of cholesterol-dependent membrane microdomains for proteasome-dependent regulation of the ERK pathway as well as the basic link between cholesterol and function of the UPS in the macrophage, we furthermore demonstrate that the LPS pathway to ERK activation can largely be recapitulated by cholesterol extraction from the macrophage.

MATERIALS AND METHODS

Chemicals and Reagents—

RAW 264.7 cells were acquired from ATCC (Manassas, VA). Custom Dulbecco's modified Eagle's medium lacking arginine and lysine and the regular Dulbecco's modified Eagle's medium (DMEM) were purchased from Invitrogen, fetal bovine serum was purchased from Atlanta Biologicals (Atlanta, GA), and dialyzed fetal calf serum was purchased from Sigma-Aldrich. Stable isotope-containing amino acids, [¹³C₆]arginine and [¹³C₆]lysine were purchased from Invitrogen. Escherichia coli 0111:B4 LPS was purchased from List Biological Laboratories (Campbell, CA). Sequencing grade porcine trypsin was purchased from Promega Corp. (Madison, WI). Solvents for liquid chromatography were purchased as follows: HPLC-grade acetonitrile (Fisher Scientific) and formic acid (Sigma-Aldrich). Raft-disrupting agents nystatin and methyl-β-cyclodextrin were purchased from Sigma-Aldrich. Bay 11-7082 (IκK inhibitor), PD98059 (MEK inhibitor), MG-132 (proteasome inhibitor), and lactacystin (proteasome inhibitor) were purchased from Calbiochem. DSMO was purchased from Sigma-Aldrich.

Cell Culture and Treatment—

RAW 264.7 cells were cultured in DMEM supplemented with 10% heat-inactivated endotoxin-free fetal bovine serum, 2 mm l-glutamine, 100 μg/ml streptomycin, and 100 units/ml penicillin under a humidified 5% CO₂ atmosphere at 37 °C. Inhibitors other than polymyxin B and methyl-β-cyclodextrin (mβCD) were dissolved in DMSO, and cells were treated with inhibitor or vehicle in parallel for a final vehicle concentration of 0.1% (v/v). Cells were treated with mβCD and mβCD-cholesterol complex as described previously (12) with the exception that endotoxin-free Hanks’ balanced salt solution was used in place of Krebs-Ringer phosphate dextrose buffer.

Stable Isotope Labeling with Amino Acids in Cell Culture—

RAW 264.7 murine macrophages were adapted for SILAC during growth in DMEM containing “light” [¹²C₆]arginine and [¹²C₆]lysine or “heavy” [¹³C₆]arginine and [¹³C₆]lysine supplemented with 10% dialyzed fetal calf serum and antibiotics. Heavy arginine and lysine were used together in the medium, and the cells were allowed to double at least five times for full incorporation of heavy amino acids in the cell as reported previously (27). Full isotopic incorporation was confirmed by MALDI-MS analysis of a panel of protein targets (data not shown).

LPS Exposure and DRM Isolation—

Heavy SILAC-labeled RAW 264.7 macrophages (10⁷ cells) were stimulated with freshly prepared E. coli 0111:B4 LPS (100 ng/ml), and light labeled macrophages (10⁷ cells) were treated with endotoxin-free PBS (control). Following stimulation, DRMs were isolated by using a modification to the method described by Fessler et al. (12). Briefly cells were washed three times with ice-cold PBS. Following the wash, cells were centrifuged (1000 × g, 10 min, 4 °C) and resuspended in 1 ml of ice-cold lysis buffer (25 mm MES (pH 6.5), 150 mm NaCl, 1% Triton X-100, 1 mm PMSF, 1 mm NaF, 0.1 mm Na₃VO₄, 5 μg/ml leupeptin, 5 μg/ml aprotinin (all reagents purchased from Sigma-Aldrich)). Cells were vortexed for 15 s on ice and allowed to sit on ice for 10 min. This step was repeated three times. High density insoluble debris were removed by centrifugation (1000 × g, 10 min, 4 °C). The supernatant (∼1 ml) was added to an equal volume of 80% sucrose in MBS (25 mm MES (pH 6.5), 150 mm NaCl), gently mixed by vortexing, and then centrifuged through a 5–30% continuous sucrose gradient using an SW41 rotor (20 h, 37,000 rpm, 4 °C) as reported previously (28). Following centrifugation, 12 1-ml fractions were harvested from the bottom by puncturing the tube with an 18-gauge needle. Protein concentrations in each fraction were determined by the Bradford protein assay. 10 μg of protein from each fraction was used to screen for rafts by immunoblotting for raft (flotillin-1) and non-raft (transferrin receptor) markers. Raft fractions were then pooled, and the total protein was quantified using protein assay. Heavy and light labeled proteins were combined 1:1, resolved by SDS-PAGE, and stained with SYPRO Ruby. Each SDS-PAGE minigel lane was loaded with 100 μg of total protein (50 μg of heavy isotope-labeled protein, 50 μg of light isotope-labeled protein).

Protein Digestion—

Each SYPRO Ruby-stained minigel lane was cut out into 24 equal slices and digested in a 96-well tray using a Progest robotic digester (Genomic Solutions Inc., Ann Arbor, MI). Briefly each gel slice was minced and placed in one of the 96 wells. During digestion, gel bands were incubated twice for 15 min in 100 μl of 25 mm ammonium bicarbonate, 50% (v/v) acetonitrile. Following this, they were dehydrated by incubation in 100 μl of acetonitrile for 20 min and dried under a nitrogen stream. Trypsin (250 ng) was added to each well, and the sample was allowed to incubate at 37 °C for 8 h. The supernatants from the digests were stored at −80 °C until analysis. Additionally the gel was re-extracted three times: once with 50 μl of water for 20 min and twice with 20-min incubations in 50 μl of 5% (v/v) formic acid, 50% (v/v) acetonitrile. Supernatants were pooled, lyophilized, and resuspended in 40 μl of 0.1% formic acid.

Nano-LC-MS/MS Analysis and Database Analysis—

All nano-LC-MS/MS experiments were performed on an Agilent 1100 nano-LC system on line with an Agilent XCT Ultra ion trap mass spectrometer with the chip cube interface (Agilent Technologies). Peptides were separated by reverse-phase LC with a trapping column/analytical column nanoflow setup (HPLC-Chip cube; Agilent Technologies) and analyzed in both the full-scan MS and in the CID MS/MS modes. Briefly 20 μl of the peptide digest was loaded onto an Agilent C₁₈ chip (75 μm × 43 mm) followed by a 15-min wash with 5% (v/v) acetonitrile, 0.1% (v/v) formic acid. Peptides were eluted with a linear gradient of increasing concentration of acetonitrile (in 0.1% (v/v) formic acid) as follows: 0–45 min, 5–50% (v/v) acetonitrile; 45–50 min, acetonitrile increased to 95% (v/v); and 50–60 min acetonitrile maintained at 95%. The mass spectrometer was used in the positive ion, standard enhanced mode and included settings of a mass range from 200 to 2200 m/z, an ionization potential of 2.1 kV, an ion change control smart target (number of ions in the trap prior to scan-out) of 100,000 or 200 ms of accumulation, and a 1.0-V fragmentation amplitude (CID collision energy). MS/MS data were acquired using a data-dependent acquisition format with the six most abundant ions from each MS scan further interrogated by MS/MS. The automated switching for MS/MS required a threshold of 5000 counts. Dynamic ion exclusion of two spectra or 1.0-min ion trap resident time was used during MS/MS data acquisition. During SILAC data acquisition, the isolation width was set to 3 m/z to eliminate the isotope bleed-over effect, whereas for all other experiments it was set to 4 m/z.

Peak lists were generated from the data obtained from each nano-LC-ESI-MS/MS analysis using the data extractor feature of the SpectrumMill Proteomics Workbench Version A.03.03.078 (Agilent Technologies). The MS and MS/MS data were extracted by limiting the data search to deconvolved ions observed between 300 and 6000 Da and a retention time between 10 and 60 min. MS scans with the same precursor mass (±1.5 m/z) and retention time within 15 s were merged. Moreover of the remaining MS/MS spectra, only spectra that contained sequence tag information of four or more residues were submitted for database searching. The resulting extracted data were searched against the rodent subset of the National Center for Biotechnology Information (NCBI) RefSeq non-redundant (nr) protein database (release date January 24, 2006 with a total of 3,229,765 protein entries) using the MS/MS Search function of SpectrumMill with SpectrumMill as the search engine. All searches allowed for partial oxidation of methionine residues, formation of glutamic acid, and allowance for one trypsin missed cleavage with mass tolerance set to ±1.5 Da for precursor mass and ±0.7 Da for product ion masses and a minimum matched spectral intensity of 70%. Confident protein identification was based on three criteria: 1) matching of ≥2 tryptic peptides, 2) a SpectrumMill score >17, and 3) a molecular weight of the protein consistent with the gel region from which the protein was identified as indicated by molecular weight standards. All peptides with a score >17 were automatically validated, whereas MS/MS spectra of all the remaining peptides were manually inspected and validated. For SILAC experiments, all analyses were carried out as described above with the inclusion of [¹³C₆]arginine and [¹³C₆]lysine as modifications. SILAC ratios for heavy and light peptides were calculated from the area under the extracted ion chromatogram from the MS data. If an identical tryptic peptide was identified from the heavy and the light samples, it was treated as one peptide for protein identification purposes; the peptide ratios were calculated to determine the protein ratio.

Western Blot—

RAW 264.7 macrophages were washed once in ice-cold 1× PBS, then lysed with 1× Laemmli buffer supplemented with 20 mm DTT, and boiled for 8 min. Lysates were resolved on 8% (for p105 detection) or 10% SDS-PAGE gels, transferred to nitrocellulose (Bio-Rad), and then probed with primary antibodies. Rabbit anti-PO₄-ERK1/2, rabbit anti-p38, goat anti-p105, and mouse anti-tubulin were from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-PO₄-p105, rabbit anti-PO₄-MEK1/2, rabbit anti-PO₄-p38, rabbit anti-PO₄-JNK, rabbit anti-JNK, and rabbit anti-ERK1/2 were from Cell Signaling Technology (Danvers, MA). Mouse anti-transferrin receptor was from Zymed Laboratories Inc. (San Francisco, CA), and mouse anti-flotillin-1 was from BD Biosciences. Membranes were then washed in Tween-TBS (TTBS) and exposed for 60 min to a 1:5000 dilution of species-specific, horseradish peroxidase-conjugated secondary antibody (GE Healthcare) in 5% milk, TTBS. Following further TTBS washes, signal was detected with 60 s of exposure to ECL™ Western blot detection reagents (GE Healthcare) followed by application to film (GE Healthcare).

26 S Proteasome Activity Assay—

Raft and non-raft fractions were isolated from 3 × 10⁷ cells, pooled separately, and then spin-concentrated/buffer-exchanged (Centriprep YM-10, 10,000 molecular weight cutoff, Millipore) into 10 mm Tris (pH 7.2) containing 40 mm KCl, 5 mm MgCl₂. 50-μg aliquots (Bradford assay) of concentrate on ice were then supplemented with fresh stock solutions for final concentrations of 1 mm DTT, 0.5 mm ATP, 0.5 mg/ml BSA, and 100 μm succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin substrate as well as 10 μm lactacystin or DMSO vehicle. Fluorescence of the cleaved product, indicating proteasomal chymotrypsin-like activity (26, 29, 30), was measured in triplicate in three independent experiments after 30 min of incubation at 37 °C with slow rotational mixing (Synergy 2 microplate reader, BioTek Instruments, Winooski, VT). Specific chymotrypsin-like activity was derived by subtracting end point fluorescence values obtained in the absence of lactacystin from those in its presence (29, 30).

Bioinformatics Network Analysis—

Ingenuity Pathway Analysis 6 (Ingenuity Systems Inc.) was used to conduct a knowledge-based network and canonical pathway analysis of the proteomics data. Proteins identified from no-LPS controls and LPS-treated samples were first individually analyzed. Following analysis, the data for the control and treated sample were pooled and reanalyzed.

Statistical Analysis—

The statistical significance of changes in the number of unique DRM proteins and of changes in proteasome activity were determined using two-way student's t test (Graphpad Prism). A p value <0.05 was considered statistically significant. The SILAC ratios are reported as mean ratios (heavy:light) ± S.D. calculated for the heavy and light peptide pair that contributed to the identification of the protein. The standard deviation reported for a SILAC ratio of a given protein is the S.D. in the ratio of the heavy:light peptides for that particular protein.

RESULTS

Qualitative Proteomics Analysis of Lipid Rafts before and after LPS Treatment—

A descriptive proteomics approach was first undertaken to profile the lipid raft proteome in resting and LPS-treated RAW 264.7 macrophages. To test the possibility that LPS induces recruitment of non-raft proteins to the lipid raft, macrophages were treated with endotoxin-free PBS (control) or LPS (100 ng/ml) for 5 or 30 min. As maximal LPS-induced MAPK activation occurs at ∼15 min in RAW 264.7 macrophages (31–33), we selected 5 min as a time point that should enrich for early, “upstream” regulatory targets and 30 min as a time point enriching for “downstream” function-regulatory proteins. Following treatment, the cells were lysed, and DRMs were isolated using ultracentrifugation over a continuous sucrose gradient (5–30%). Raft fractions were identified by immunoblotting for flotillin-1 (raft marker) and transferrin receptor (non-raft marker) (Fig. 1). Raft fractions were then pooled, and 100 μg of pooled raft protein was resolved by SDS-PAGE and detected with SYPRO Ruby (supplemental Fig. S1). Each gel lane was excised into 24 approximately equal sized pieces, and individual LC-MS/MS experiments were performed on the tryptic peptides extracted from each gel piece. A typical ion chromatogram obtained during the LC-MS/MS analysis of a single in-gel digested band is shown in supplemental Fig. S2. Similarly typical MS spectra of a heavy and light SILAC peptide pair and their corresponding MS/MS spectra are summarized in supplemental Fig. S3. The data from the MS/MS fragmentation study were used to identify the peptide sequence, and subsequently the proteins were identified by searching against the NCBInr mouse database.

Fig. 1.

Western blot confirmation of raft subcellular fractions using flotillin-1 (raft marker) and transferrin receptor (non-raft marker). Following cell lysis and sucrose density gradient ultracentrifugation, 12 1-ml fractions were collected. Equal aliquots (10 μg) were analyzed for flotillin-1 and transferrin receptor (TfR) expression with the respective antibodies.

Three LPS-treated biological samples and three corresponding, paired untreated control samples were generated for each time point, 5-min LPS exposure or 30-min LPS exposure. In total, six samples (three control and three LPS-treated) were analyzed for 5-min LPS exposure, and a similar second set of six samples was analyzed for 30-min LPS exposure. A separate set of control samples was generated for each treatment condition to minimize biological variation during sample preparation. The non-redundant proteins identified for the control-treated pair for the 5-min time point are listed in supplemental Tables S1–S6; those corresponding to the 30-min time point are listed in supplemental Tables S7–S12. In the paired 5-min analysis, a mean of 383 (±79) proteins was identified in the control sample, and a mean of 463 (±71) proteins was identified in the LPS-treated sample. Similarly in the 30-min paired analysis, a mean of 241 (±9) proteins was identified in the control sample, and a mean of 409 (±106) proteins was identified for 30-min LPS-treated samples. A tight standard deviation in the number of proteins identified between individual analytical runs indicated good analytical reproducibility in LC-MS/MS analysis. Several functional categories were represented among the raft proteins in each experimental condition (supplemental Fig. S4). The major functional groups include cell movement, cell signaling, cell-to-cell signaling and interaction, cellular assembly and organization, molecular transport, lipid metabolism, post-translational modification, and small molecule biochemistry. Representation of some functional categories was noted to change with LPS treatment; notable was a relative depletion of cellular movement proteins after 5 min of LPS (supplemental Fig. S4).

Consistently for both time points (5 and 30 min), a larger number of proteins were identified in DRMs from the LPS-treated macrophages compared with the control samples, suggesting an active recruitment of basally raft-excluded proteins to the lipid raft upon LPS treatment. This increase in number of raft proteins from base line with LPS exposure was statistically significant (p < 0.05) at 30 min (Fig. 2). Table I represents a condensed list of select proteins that were newly recruited to the lipid raft following LPS treatment. The term “de novo” is used to indicate these newly recruited proteins: “de novo 3” indicates those proteins present in all three experiments in samples from 5 or 30 min after LPS exposure but absent in all three control samples; the “de novo 2” list represents those proteins present in two of the three experiments in samples from 5 or 30 min after LPS exposure but absent in all three control samples. Using these stringent criteria, the repertoire of putative raft-resident proteins in the resting macrophage and those proteins that are newly recruited to lipid rafts upon LPS exposure were thus identified.

Fig. 2.

Increase in number of unique proteins identified in DRMs following LPS exposure. The number of distinct proteins identified in DRMs from control and LPS-treated macrophages is shown for 5- (A) and 30-min (B) exposures. Data shown represent three independent MS/MS technical analyses of three biological samples. *, p < 0.05; Error bars represent standard deviation of mean.

Table I.

A condensed list of representative proteins newly recruited to the macrophage lipid raft upon LPS treatment

NACHT, nucleotide binding oligomerization domain; PYD, pyrin-domain; CBL-B, Cas-Br-M (murine) ecotropic retroviral transforming sequence b; CRA, sequence derived from the Celera database.

5-min LPS treatment		30-min LPS treatment
NCBInr accession number (gi no.)	Protein name	NCBInr accession number (gi no.)	Protein name
De novo 3a
2564072	Calcium-activated potassium channel rsk3	110625671	Apolipoprotein B
148702862	Fatty acid synthase, isoform CRA_b	78214312	ATP synthase, H+-transporting, mitochondrial F0 complex, subunit b, isoform 1
7305189	Integrin α3	7514005	Major vault protein
42475968	Leucine-rich repeats and immunoglobulin-like domains 3	124517716	Solute carrier family 12, member 2
111380589	NADH dehydrogenase subunit 1	18875360	Toll-like receptor 7
108744609	N-Glycan processing α-mannosidase iix	146325812	Vacuolar ATP synthase subunit d 2
31543784	Serine/threonine kinase 16	123173782	Voltage-gated channel like 1
De novo 2b
88911266	E3 ubiquitin-protein ligase CBL-B	18700004	Acetyl-coenzyme A acyltransferase 1
22122701	Leucine-rich repeat protein 4, neuronal	148690825	Adaptor protein complex AP-2, α 1 subunit
148665020	Mitogen-activated protein kinase 1, isoform CRA_b	27545388	ATP-binding cassette, subfamily A (ABC1), member 5
124486847	Mitogen-activated protein kinase kinase kinase 10	148702433	ATP-binding cassette, subfamily A (ABC1), member 6
34870816	Predicted: similar to mitogen-activated protein kinase kinase 3	1143305	Cytosolic phospholipase A2
109460509	Predicted: similar to mitogen-activated protein kinase kinase kinase 5	3335569	Fatty acid transport protein 4; FATP4
149273657	Predicted: similar to ubiquitin-activating enzyme E1	148673922	Heat shock protein 110, isoform CRA_a
109458840	Predicted: similar to ubiquitin-protein ligase E3A isoform 2	149029498	Lipoxygenase homology domains 1 (predicted)
125987791	Rho guanine nucleotide exchange factor 7	148699347	NACHT, leucine-rich repeat, and PYD-containing 2, isoform CRA_b
21312550	Solute carrier family 25, member 35	109458474	Predicted: similar to ryanodine receptor 1 (skeletal)
28893565	Solute carrier family 41, member 2	51948448	RAB21, member RAS oncogene family
148683226	Ubiquitin-associated protein 2-like, isoform CRA_e	56205310	Rho-interacting protein 3
126032299	Ubiquitin-specific protease 32	2144101	Tricarboxylate carrier
149031228	Ubiquitin-specific protease 54	33469031	Ubiquitin-specific peptidase 53
148706006	Ubiquitin-activating enzyme E1-like 2, isoform CRA_a	76443681	Ubiquitin-specific protease 11

^aDe novo 3 indicates proteins present in LPS-treated samples from three of three experiments but absent in paired control samples from three of three experiments.

^bDe novo 2 indicates proteins present in LPS-treated samples from two of three experiments but absent in paired control samples from three of three experiments.

An unsupervised network analysis (Ingenuity Pathway Analysis 6, Ingenuity Systems Inc.) of the qualitative DRM proteomics data set indicated that the UPS was the best populated pathway with a striking representation of its components (p = 0.006). A total of 33 proteins present along the UPS pathway were identified. The overall UPS pathway and, within it, pathway members identified in the DRM analysis are mapped in supplemental Fig. S5. Specific proteasome subunits identified in DRMs are illustrated in Fig. 3. Several proteins involved in protein ubiquitination were noted to be newly recruited into rafts upon LPS treatment (Table I). Moreover 10 distinct proteasome subunits were identified in rafts at base line, whereas three additional proteasome subunits were newly recruited to rafts upon LPS treatment (5 and 30 min combined) (Table II and Fig. 3). Of the several proteasome subunits identified in the analysis, subunit β type 4 has been reported to bind LPS (26). Taken together, these identifications suggest a potential role for the raft-localized proteasome in LPS signaling as well as the possibility that LPS might physically and functionally remodel the proteasome and proteins targeted for it within lipid rafts.

Fig. 3.

An enlarged section of the ubiquitin-proteasome system canonical pathway highlighting proteasome subunits and their reported IFNγ-induced exchange in formation of the “immunoproteasome” (65, 66). The gray-shaded subunits were identified in macrophage DRMs. PSM, proteasome subunit; THIMET, thimet oligopeptidase 1; TAP, transporter 1 ATP-binding cassette (TAP1 and TAP2); IFNG, interferon-γ; ER, endoplasmic reticulum; MHC, major histocompatibility complex. The proteasome subunits with the asterisks represent components observed in more than one biological sample, and the double lined shapes represent a group of proteins/protein complex.

Table II.

Proteasome subunits identified in lipid rafts of the resting or LPS-stimulated macrophage

CRA, sequence derived from the Celera database.

NCBInr accession number (gi no.)	Protein name
Proteasome subunits in lipid raft at base line
149053606	Proteasome 26 S subunit, non-ATPase, 11 (predicted), isoform CRA_b
149030731	Proteasome subunit, β type 4
148704781	Proteasome subunit, α type 6, isoform CRA_b
94388799	Proteasome subunit, α type 5
74193793	Proteasome subunit, α type 2
74185161	Proteasome 26 S subunit, ATPase 3, isoform CRA_c
16758370	Proteasome subunit, β type 1
8394079	Proteasome subunit, β type 2
6755202	Proteasome β 3 subunit
972986	20 S proteasome subunit Lmp7
Proteasome subunits newly recruited to lipid raft on LPS treatment (5 or 30 min)
84028244	Proteasome subunit β type 6 precursor (proteasome δ chain) (macropain δ chain) (multicatalytic endopeptidase complex δ chain) (proteasome subunit Y)a
74315975	Proteasome 26 S non-ATPase subunit 1b
62641588	Proteasome 26 S non-ATPase subunit 13a

^a30-min LPS stimulation.

^b5-min LPS stimulation.

To confirm proteasomal function in lipid rafts and to test for a possible associated modification of proteasomal activity in the setting of LPS exposure, 26 S proteasomal chymotrypsin-like activity upon the fluorogenic substrate succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin was measured (34) in both raft and non-raft subcellular fractions. Chymotrypsin-like activity is one of the three well described prototypical specificities exercised by the proteasome upon protein substrates (34). As shown in Fig. 4, LPS induced a modest but statistically significant increase in proteasome activity in rafts in contrast to a marked reduction in proteasomal activity outside rafts. Taken together, these findings indicate that a compartmentalized effect of LPS upon proteasome function occurs within lipid rafts. To better define the LPS responsiveness of proteasome expression within rafts, we next undertook a quantitative proteomics approach to determine relative changes in expression of specific subunits within the raft upon LPS treatment.

Fig. 4.

LPS selectively activates the proteasome in macrophage rafts. RAW 264.7 macrophages were treated with PBS or LPS (100 ng/ml), and then non-raft and raft pools were isolated (see “Materials and Methods”). Pools were concentrated/buffer-exchanged (Centriprep, 10,000 molecular weight cutoff, Millipore) into proteasome assay buffer. A 50-μg aliquot (Bradford assay) of each pool was then assayed for chymotryptic-like proteasome activity upon the substrate succinyl-Leu-Leu-Val-Tyr-aminomethylcoumarin. Results are representative of three independent experiments analyzed in triplicate (*, p < 0.01; **, p = 0.0001).

Quantitative Proteomics Analysis of Lipid Rafts Reveals Compartmentalized Remodeling and Activation of the Proteasome by LPS—

LPS-induced changes to the lipid raft can involve quantitative enrichment of raft-resident proteins in addition to new recruitment to rafts of proteins that are basally raft-excluded (12). Hence qualitative identification of protein presence in rafts may fail to detect more subtle, but nonetheless biologically important quantitative changes in expression. A SILAC-based stable isotope labeling approach (Fig. 5) was thus used to quantify LPS-induced changes in relative protein expression in rafts as reported previously (35, 36). Within this global quantitative analysis, our specific focus was to profile possible LPS-induced changes in expression of raft-resident proteasome subunits. Following culture of the macrophages in light [¹²C₆]arginine and [¹²C₆]lysine or heavy [¹³C₆]arginine and [¹³C₆]lysine medium, light labeled cells were treated with PBS (control), whereas heavy labeled cells were exposed to LPS. As in the qualitative analysis, paired buffer-LPS exposures were performed for both 5 and 30 min of LPS exposure. Each exposure was carried out in triplicate, resulting in three biological SILAC samples for each time point. Following raft isolation, the pooled raft isolates from the light and heavy labeled cell were combined 1:1, separated by SDS-PAGE, and analyzed by LC-MS/MS. The b- and y-series ions from the MS/MS fragmentation along the peptide backbone were used to identify the peptide sequence, and the light:heavy peptide ratios were determined from the MS experiment by calculating the area under the peak of the extracted ion chromatogram. A mean of 143 and 145 proteins were identified from the 0–5-min and 0–30-min SILAC experiments, respectively. During SILAC data analysis, an identical light and heavy peptide was treated as a single unique peptide for protein identification purposes; this constraint resulted in a lower number of proteins reported as identified in the SILAC experiments compared with the non-SILAC experiments described above. This is due to reduced sensitivity as a result of dividing the total amount of protein (same total protein amount analyzed in each experiment) by two.

Fig. 5.

SILAC and mass spectrometry-based strategy used for quantitative analysis of lipid raft proteins upon LPS exposure of RAW 264.7 macrophages.

The proteins identified in the SILAC experiments along with the calculated light:heavy SILAC ratios are summarized in the supplemental Tables S13–S18. Proteins identified from only their heavy peptide were arbitrarily assigned the largest ratio determined in that data set as reported previously (37). Similarly proteins that were identified with only the light peptide were arbitrarily assigned the smallest ratio determined in that data set (37). Overall modest protein ratios were observed, consistent with previous reports applying SILAC methodology to short cell exposures like those used in the present study (38). The standard deviations calculated for the heavy:light SILAC ratios were consistently tight across all six experiments, with a mean standard deviation in the ratio across all six experiments of 0.16. As in previous SILAC analyses (35, 37), a protein ratio exceeding two standard deviations above unity was defined as indicating up-regulation. Conversely a ratio lower than two standard deviations below unity was defined as indicating down-regulation. Similarity in the number of proteins identified between three different biological samples and three different technical analyses along with the calculated mean standard deviation of 0.16 in the SILAC ratios suggests good biological and analytical reproducibility.

Using these criteria, the SILAC analysis revealed statistically significant LPS-induced enrichment in rafts of LPS receptor components, including CD14 and HSP70, consistent with other reports (18, 39) (supplemental Table S19). TLR4 was not detected in resting or LPS-stimulated rafts. Also noted was LPS-induced quantitative enrichment in rafts of tyrosine kinases (lyn and hck) and adhesion molecules (CD44) reported to associate with TLR4 or its signaling (40–42). Two enzymes, acid sphingomyelinase and gp91 phox, enriched in rafts at both 5 and 30 min of LPS exposure have been reported to regulate recruitment of LPS receptor components to rafts (43, 44).

In addition, seven different proteasome subunits were detected in rafts (Table III). Six of the seven subunits were consistently enriched upon LPS treatment, whereas proteasome subunit α type 6 was depleted upon LPS treatment (5 min) from the lipid raft in all three experiments. Findings were consistent between the qualitative and quantitative proteomics approaches. Proteasome subunit β type 6 was found to be recruited to rafts de novo at 30 min in the qualitative proteomics approach and to be enriched >3.13-fold in the SILAC-based approach. Similarly the proteasome 26 S non-ATPase subunit 1 was found to be recruited de novo at 5 min in the qualitative analysis and to be up-regulated 2.83-fold at 5 min in the SILAC-based approach. Of interest, the catalytic proteasome subunit β type 1, noted to be present in rafts at base line in the qualitative analysis, was revealed by the SILAC approach to undergo time-dependent up-regulation in rafts after LPS challenge with a >2.42-fold up-regulation from control at 5 min and a mean 1.37-fold up-regulation from control at 30 min of LPS.

Table III.

LPS-induced changes in proteasome subunit expression in macrophage lipid rafts

CRA, sequence derived from the Celera database.

NCBInr accession number (gi no.)	Protein name	SILAC ratios
5-min LPS
16758370	Proteasome subunit, β type 1	>2.42
149047100	Proteasome subunit, β type 1, isoform CRA_b	>2.27
109468228	Predicted: similar to 26 S proteasome non-ATPase regulatory subunit 14	H and La
74315975	Proteasome 26 S subunit, non-ATPase, 1	2.83
148704781	Proteasome subunit, α type 6, isoform CRA_b	0.63 (0.18)b,c
148704620	Proteasome subunit, α type 3, isoform CRA_e	1.82
30-min LPS
38051991	Proteasome subunit, α type 2	1.20d
37231712	Proteasome subunit, β type 1	1.53d
31982099	Proteasome subunit, β type 6	>3.13

^aH, heavy and L, light.

^bRepresents standard deviation for the calculated ratio.

^cObserved in three of three biological samples.

^dObserved in two of three biological samples.

The marked preponderance of proteasome subunits and other UPS-associated proteins in the lipid raft and the observed sensitivity of expression and function to LPS suggested that the proteasome might play an upstream regulatory role in LPS signaling at the level of the membrane. Experiments were thus next undertaken to investigate this hypothesis. Given previous reports that the proteasome regulates early, upstream events in LPS activation of the ERK MAPK pathway (20, 21, 26) we tested for a regulatory role of rafts in functional compartmentalization of these events.

Rafts Couple the Proteasome to LPS-induced Activation of ERK—

The proteasome is reported to regulate LPS-induced ERK activation in macrophages (Fig. 6A) (20, 21, 26). Following LPS exposure, IκK phosphorylates p105, targeting it for ubiquitination and proteasomal degradation. Proteasomal degradation releases the inhibitory binding interaction of p105 upon the MEK kinase Tpl2, thereby allowing consequent MEK and then ERK activation. Nevertheless the subcellular localization and organizing principle of these events that involve the proteasome upstream of ERK remain undescribed. Before investigating a possible role for rafts in compartmentalization of these events, we first confirmed roles for IκK and the proteasome in LPS-induced activation of ERK in our system. Consistent with a previous report (21), LPS-induced activation of ERK and its upstream kinase MEK were attenuated by pretreatment of macrophages with the proteasome inhibitor MG-132 (Fig. 6B). Similar results were found with the proteasome inhibitor lactacystin (data not shown). By contrast, activation of the alternate MAPKs p38 and JNK was not reduced by MG-132 (Fig. 6B) or lactacystin (data not shown), suggesting a specific role for the proteasome in ERK activation. IκK inhibition with Bay 11-7082 blocked LPS-induced phosphorylation of p105, MEK, and ERK and attenuated LPS-induced degradation of the former consistent with its localization upstream of the p105-MEK-ERK cascade (Fig. 6C). Degradation of p105 following LPS was attenuated by MG-132 (Fig. 6D), also consistent with the reported role of the proteasome in p105 degradation (21). Taken together, these findings confirm previous reports (20, 21, 26) of an upstream role for the proteasome in the ERK cascade with a hierarchy of sequential events: 1) IκK phosphorylation of p105, 2) proteasomal degradation of p105, 3) MEK activation, and 4) ERK activation.

Fig. 6.

Selective regulation of LPS-ERK pathway by the proteasome through activity upon p105. A, schematic representation of the role of the proteasome in the regulation of LPS-ERK pathway. B, RAW 264.7 macrophages were pretreated with 5 μm MG-132 or 0.1% DMSO vehicle (60 min) and then either left untreated or stimulated with 10 ng/ml LPS or buffer for the indicated durations. Cells were then washed, lysed, resolved by 10% SDS-PAGE, transferred to nitrocellulose, and probed with the antibodies indicated. C, RAW 264.7 macrophages were pretreated with Bay 11-7082 (Bay; 10 μm, 60 min) or 0.1% DMSO and then treated with LPS, lysed, and probed for the indicated targets. D, RAW 264.7 macrophages were treated as in A, and lysates were then probed for p105, p50, and tubulin. Results are representative of four independent experiments. ERK1/2, p38, JNK, p50, and tubulin were detected as loading controls.

To investigate a role for rafts in the signaling events upstream of LPS-induced ERK activation and to confirm a role for the raft-localized proteasome in them, we pretreated RAW 264.7 macrophages with MG-132 (or vehicle) followed by LPS (or buffer) and then isolated DRM and non-DRM fractions for targeted immunoblotting. As shown in Fig. 7, despite expression of the majority of cellular ERK outside of rafts, activated (i.e. phosphorylated) ERK was detected only in rafts following LPS, indicating high specific activity in this compartment. Phosphorylated MEK was similarly detected only in raft fractions following LPS. Although the majority of cellular p105 was expressed outside of rafts, a minor fraction was detected in rafts. Expression of raft and non-raft p105 was enhanced following MG-132 treatment, suggesting that p105 in both compartments is a substrate for the proteasome. Moreover raft p105 was down-regulated by LPS, consistent with promotion by LPS of p105 proteasomal degradation in rafts, as observed in cell lysates (Fig. 6, C and D). Taken together, these findings indicate that compartmentalized, proteasome-dependent activation of the ERK cascade occurs in rafts and suggest that p105 is a target for the proteasome in rafts.

Fig. 7.

LPS selectively activates the proteasome and the p105-MEK-ERK pathway in rafts. RAW 264.7 macrophages were pretreated with 5 μm MG-132 or 0.1% DMSO vehicle (60 min) and then exposed to buffer or 10 ng/ml LPS (15 min). Cells were then washed and lysed, and rafts were isolated as described under “Materials and Methods.” Non-raft and raft fractions were individually pooled from all conditions, and 20-μg protein aliquots (Bradford assay) from each were then subjected to SDS-PAGE, nitrocellulose transfer, and immunoblotting for the targets shown. Results are representative of two independent experiments. TfR, transferrin receptor.

To further confirm a regulatory role for rafts in organization of the p105-MEK-ERK cascade, we tested the effect of two alternate, structurally dissimilar agents that have been used to perturb/disrupt raft microdomain organization by different mechanisms, mβCD (45–47) and nystatin (48, 49). mβCD is thought to disorganize rafts through extraction of cholesterol, whereas nystatin is thought to sequester membrane cholesterol. p105 phosphorylation and degradation and ERK phosphorylation were selected as upstream and downstream readouts, respectively, to assess the effect of these raft-perturbing agents on the LPS pathway. LPS-induced p105 phosphorylation and degradation and ERK phosphorylation were all attenuated by both mβCD and nystatin (Fig. 8A). Of interest, in the absence of LPS, mβCD but not nystatin was sufficient to induce phosphorylation of p105 and ERK. This stimulating activity of mβCD on ERK was not due to LPS contamination as ERK activation by LPS but not mβCD was abolished by the LPS-sequestering agent polymyxin B (50) (Fig. 8B). Moreover consistent with mβCD acting through cholesterol extraction, its activity upon ERK was abolished by preloading with cholesterol as reported previously (12). As mβCD has been reported by us and others to recapitulate several of the features of LPS signaling (12, 51), we investigated the mechanism of its activation of ERK for comparison with that of LPS. Like LPS (Fig. 6), mβCD-induced activation of both ERK and its upstream kinase MEK was inhibited by both proteasome inhibition and IκK inhibition (Fig. 8C), indicating that the proteasome and IκK are upstream of MEK-ERK activation. Also similar to LPS, mβCD-induced phosphorylation and degradation of p105 appear to be mediated by IκK as they were both inhibited by Bay 11-7082. Finally as with LPS, mβCD-induced p105 degradation was blocked by pretreatment with MG-132, consistent with its mediation by the proteasome.

Fig. 8.

Raft-disrupting agents modulate the LPS-p105-MEK-ERK pathway. A, RAW 264.7 macrophages were left untreated or pretreated with mβCD (CD) (10 mm, 30 min) or nystatin (25 μg/ml, 30 min). Cells were then exposed to 10 ng/ml LPS, lysed, electrophoresed, and probed for the indicated protein targets. B, RAW 264.7 macrophages were pretreated with buffer or polymyxin B (25 μg/ml, 30 min) and then left untreated or treated with mβCD (10 mm, 30 min), LPS (10 ng/ml, 30 min), or mβCD-cholesterol complex. Lysates were then probed for the targets shown. C, RAW 264.7 macrophages were pretreated for 60 min with 0.1% DMSO vehicle, 5 μm MG-132 (MG), or 10 μm Bay 11-7082 (Bay) and then either left untreated or exposed to mβCD (10 mm, 30 min). Lysates were probed for the targets shown. Results are representative of four independent experiments.

DISCUSSION

Complementary qualitative and quantitative proteomics approaches were used to probe the dynamics of the lipid raft proteome in RAW 264.7 macrophages during LPS stimulation, together allowing for both global and targeted investigations. Using strict criteria, the qualitative profiling approach revealed that during LPS exposure a substantial number of new (i.e. basally raft-excluded) proteins are recruited to rafts in a time-dependent manner. We are unaware of any previous report that has documented a stimulus-induced statistically significant global increase in the number of distinct proteins in rafts. Reproducibly distinct sets of proteins were recruited at the two time points (5 and 30 min) following LPS challenge, supporting our selection of these two post-LPS time points as biologically distinct and informative. Classification of raft proteins from resting and LPS-stimulated macrophages revealed a variety of functional categories as well as relative changes in class representation during the LPS time course; of particular note was a relative depletion of cellular movement proteins noted 5 but not 30 min following LPS exposure. A more focused network analysis of the qualitative proteomics data set then revealed the UPS to be the best represented pathway in rafts. De novo recruitment to rafts of several UPS members following LPS suggested that the UPS might be LPS-responsive, a hypothesis that was supported by quantitative proteomics measurements of proteasome subunits. In final validation, it was confirmed that LPS induces functional changes in proteasome activity specific to rafts and that the proteasome plays an important role in raft-compartmentalized activation of the ERK pathway by LPS.

Although not proving physical complex formation among the proteasome subunits identified, the data collectively provide firm evidence for the localization of proteasome subunits to rafts, their reorganization within rafts following LPS exposure, and an associated functional change in proteasomal activity specific to rafts that occurs during LPS signaling. Ubiquitination of a target protein and its subsequent proteolysis by the 26 S proteasome is a well documented process in biology. The “canonical” NF-κB pathway is one such well characterized pathway that involves the rapid ubiquitination of IκB and its degradation by the 26 S proteasome, resulting in release of NF-κB dimers into the nucleus and consequent gene transcription (52, 53). Proteasomal degradation of p105 has also been shown to occur upstream of LPS-stimulated ERK activation (54, 55). We now show the cascade of LPS-triggered events extending from proteasomal degradation of p105 down to activation of ERK to be compartmentalized to rafts. Moreover as previously reported for LPS (12, 51) and other raft-dependent signaling pathways (e.g. epidermal growth factor receptor (56)), we used mβCD to show that raft “integrity” may play a role in both basal quiescence and ligand-stimulated activation of signaling cascades. Previous reports have indicated that plasma membrane remodeling induced by mβCD (57) may in certain cases trigger “ligand-independent” signaling phenomena perhaps either through liberation of raft-sequestered signals or through microdomain remodeling (51). Taken together, the findings in the present report suggest that raft integrity is required for feed-forward LPS-induced activation of the IκK-p105-proteasome-MEK-ERK cascade and that primary, ligand-independent lipid perturbations of this membrane compartment may nevertheless suffice to trigger these same signaling cascades.

DRMs have been studied extensively in the literature as biochemical approximations of in vivo rafts; nevertheless they are not identical to rafts in cellular membranes in vivo (58, 59). Hence the descriptive list of DRM protein identifications we provide should be considered as hypothesis-generating in nature. Independent approaches, such as confocal microscopy and co-immunoprecipitation, may be used to complement DRM analysis in confirmation of localization of specific proteins to rafts in less global approaches (60, 61). Nevertheless approaches like the one herein that target stimulus-induced changes in DRM protein expression and activity are more specific for the in vivo raft proteome (14, 47) particularly when coupled with confirmatory functional data using raft-perturbing agents such as mβCD and nystatin.

It has been postulated that distinct subcellular pools of the proteasome, characterized by differing subunit composition and activity profiles, may possibly subserve distinct roles in cell biology (62–64). We are unaware of any previous reports of stimulus-sensitive compartmentalization of proteasome activity to lipid rafts or of LPS-induced dynamic exchange of proteasome subunits in the macrophage. LPS has been reported to bind directly to specific proteasome subunits and to directly activate the proteasome in vitro. LPS is also thought to be “targeted” to lipid rafts in host cells in part through its interactions with raft-resident proteins (i.e. CD14). Taken together, it is interesting to speculate that these features may account for the specificity of LPS for activation of the raft-localized proteasome in our study. To what extent LPS promotes proteasomal degradation of p105 upstream of ERK through activation of the proteasome as compared with specific ubiquitin targeting of p105 is uncertain. Future efforts will be needed to investigate the means of localization of the proteasome to rafts, the functional effects specific to the proteasome subunits noted to change with LPS in this study, and additional roles for the 26 S proteasome in raft biology.