A proteomic approach for the elucidation of the specificity of ectodomain shedding

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

Available online at www.sciencedirect.com

ScienceDirect www.elsevier.com/locate/jprot

A proteomic approach for the elucidation of the specificity of ectodomain shedding Kyoko Shirakabea,b,⁎, Yoshio Shibagakic , Akihiko Yoshimuraa , Shigeo Koyasua,d , Seisuke Hattoric a

Department of Microbiology and Immunology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan PRESTO, Japan Science and Technology Agency, 4-1-8 Honcho, Kawaguchi, Saitama 332-0012, Japan c Division of Biochemistry, School of Pharmaceutical Science, Kitasato University, 5-9-1 Shirokane, Minato-ku, Tokyo 108-8641, Japan d Laboratory for Immune Cell System, RCAI, RIKEN Center for Integrative Medical Sciences (IMS), 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama City, Kanagawa 230-0045, Japan b

AR TIC LE I N FO

ABS TR ACT

Article history:

Ectodomain shedding (shedding) is a posttranslational modification mechanism, which

Received 2 October 2013

liberates extracellular domains of membrane proteins through juxtamembrane processing.

Accepted 8 January 2014

Because shedding alters cell characteristics in a rapid and irreversible manner, it must

Available online 20 January 2014

be strictly regulated. However, the regulatory mechanisms of shedding in response to environmental changes remain obscure. To evaluate the regulatory mechanisms of endogenous

Keywords:

shedding, we previously developed a proteomic screening system to identify shedding targets.

Ectodomain shedding

This system revealed a comprehensive picture of membrane proteins shed under defined

ADAMs

conditions. In this study, we have improved the screening system to compare the shedding

2D-DIGE

patterns in a mouse macrophage cell line treated with two different shedding inducers,

Macrophage

lipopolysaccharide (LPS) and 12-O-tetradecanoylphorbol 13-acetate (TPA). We show here that LPS

Mac-1

simultaneously activates the shedding of multiple membrane proteins. We further show that TPA specifically activates the shedding of αM/β2 integrin (Mac-1), which was not shed upon LPS-stimulation of macrophages. These results clearly demonstrate that the regulation of endogenous membrane protein shedding is both stimulus- and substrate-specific. Biological significance The shedding targets reported to date play pivotal roles in a variety of biological phenomena, including the immunological response, cell growth, cell adhesion and cell movement. In addition, several disease-related membrane proteins are shedding targets. Thus, understanding the regulation of shedding is important for the elucidation of pathogenesis and the development of therapeutic strategies. We submit that a comprehensive characterization of endogenous shedding is indispensable for understanding the regulatory mechanisms of shedding, and thus have developed a proteomic screening system to identify shedding targets. In this study, using our screening system, we demonstrate that different extracellular stimuli

Abbreviations: ADAM, a disintegrin and metalloprotease; LPS, lipopolysaccharide; TPA, 12-O-tetradecanoylphorbol 13-acetate; GPCR, G-protein coupled receptor; TLR, Toll-like receptor; M-CSF, macrophages colony stimulating factor; 2D-DIGE, two-dimensional difference gel electrophoresis; LBP, lipopolysaccharide-binding protein; MMP, matrix metalloproteinase; SILAC, stable isotope labeling using amino acids in cell culture. ⁎ Corresponding author at: Department of Microbiology and Immunology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan. Tel.: +81 3 5363 3769; fax: +81 3 5361 7658. E-mail address: [email protected] (K. Shirakabe). 1874-3919/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jprot.2014.01.012

234

J O U RN A L OF P ROTE O M IC S 9 8 ( 2 01 4 ) 2 3 3 –24 3

activate different types of shedding, even in a single cell. Our results prove that this proteomic approach is quite effective for the elucidation of the regulatory mechanisms of shedding. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Ectodomain shedding, also simply called shedding, is a posttranslational modification mechanism for membrane proteins. Shedding liberates the extracellular domain (ectodomain) of membrane proteins through juxtamembrane processing, which is executed mainly by A Disintegrin And Metalloproteases (ADAMs), a family of membrane-bound metalloproteases [1–4]. Shedding can solubilize growth factors and cytokines that are expressed as extracellular domains of transmembrane proteins, thus drastically expanding their effective working area. On the other hand, shedding can decrease the amount of cell surface receptors and adhesion molecules, and alter the responsiveness of cells to the corresponding ligands. Taken together, shedding is an effective mechanism for regulating in a rapid and irreversible manner not only the membrane proteins to be shed (hereafter called shedding targets) but also the functions of cells expressing shedding targets on their surface. Since shedding is such an effective and immediate mechanism, it should be tightly regulated. However, the molecular mechanisms ensuring the specificity of shedding are unclear. First of all, because shedding targets reported to date have no consensus cleavage site sequence [1,2], it is unclear how those membrane proteins are selectively shed. Secondly, while extracellular stimuli, including lipopolysaccharide (LPS), 12-Otetradecanoylphorbol 13-acetate (TPA) and G-protein coupled receptor ligands have been reported to activate shedding [2], the intracellular signaling pathways activated by these stimuli are completely different, and molecular mechanisms linking these stimuli and the activation of shedding remain obscure. Furthermore, these stimuli are assumed to activate different types of shedding (e.g. LPS, pro-inflammatory cytokine and its receptors; TPA, almost all reported shedding targets; G-protein coupled receptor ligands, epidermal growth factor family members) through the activation of different signaling pathways [2], but this assumption is mainly based on analysis using stimulussensitive cells ectopically expressing a shedding target of interest. It is thus unclear whether different stimuli can activate different types of endogenous shedding machineries even in a single cell. Based on this background, comprehensive analysis of endogenous shedding is indispensable in order to evaluate the specificity of shedding. Since unbiased proteomic screening approaches have been successful in identifying endogenous substrates of several types of proteases [5–9], we have developed a proteomic screening system for shedding targets to reveal the comprehensive picture of endogenous membrane proteins shed under defined conditions. In a previous study, we screened for shedding targets in a LPS-stimulated mouse macrophage cell line and identified vesicular integral membrane protein 36 kDa (VIP36) as a new shedding target [10]. We further showed that VIP36-activated phagocytosis of macrophages is shedding-dependent, confirming that our screening system could identify a shedding target that executed a physiological function in the cell type subjected to the screening, i.e. phagocytosis in activated macrophages.

In this study, we further improved our proteomic screening system for shedding targets and succeeded in comparing the endogenous shedding occurring in macrophages stimulated with two different shedding inducers, LPS and TPA. We show that LPS simultaneously activates the shedding of multiple membrane proteins that are weakly shed even in the absence of extracellular stimuli. We further show that TPA specifically activates the shedding of Mac-1, a bimolecular complex of integrin αM and β2 [11], as well as the membrane proteins shed in LPS-stimulated macrophages. These results clearly demonstrate that different extracellular stimuli can activate different types of endogenous shedding even in a single cell, and prove that shedding of endogenous membrane proteins is regulated in a both stimulus- and substrate-specific manner.

2. Materials and methods 2.1. Antibodies, plasmids, and chemicals Antibodies were purchased from Santa Cruz Biotechnology (anti-M-CSF receptor, sc-692), Promega (anti-HaloTag, G921A), BioLegend (anti-CD11b/integrin αM, M1/70), and SouthernBiotech (biotin conjugate anti-CD18/integrin β2, C71/16). To construct a plasmid expressing N-terminally HaloTag-fused membrane proteins of interest, the signal sequence of IL-6 and coding sequence of HaloTag 7 (Promega) were conjugated using a PCR-based method and subcloned into pcDNA3.1/Zeo(−) (Life Technologies). The coding sequences of human Sema4D (pF1KSDBB1316, Kazusa DNA Research Institute) and mouse M-CSF receptor (M-CSFR) were subcloned into the HaloTagexpressing plasmid after removal of its own signal sequence (Met1-Ala21 and Met1-Gly19, respectively) by PCR. LPS was purchased from Sigma. TPA was purchased from Calbiochem. BB94 was provided by Vernalis.

2.2. Cell line and transfection The Raw 264.7 line was purchased from ATCC and cultured in high glucose DMEM supplemented with 10% fetal bovine serum, 50 μM 2-mercaptoethanol, and antibiotics at 37 °C with 5% CO2. Transfections were performed using FuGENE HD (Promega).

2.3. Sample preparations for Western blotting Cell extracts were prepared using an extraction buffer (20 mM Tris–HCl (pH 7.4), 150 mM NaCl, 2 mM EDTA, 1% Nonidet P-40, 0.1% SDS) supplemented with a protease inhibitor mixture (Sigma) and 10 μM BB94. When conditioned culture media were prepared, cells were washed twice with serum-free medium and cultured in serum-free medium with or without 20 μM BB94, 1 μg/ml LPS, or 200 ng/ml TPA for 60 min at 37 °C with 5% CO2. Media were centrifuged at 20,400 ×g for 10 min to remove cells and debris, and proteins were precipitated by

235

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

the addition of 0.25 volumes of 100% trichloroacetic acid and briefly washed with −20 °C acetone.

2.4. Two-dimensional DIGE screening for shedding targets and protein identification To completely remove serum from conditioned medium, twenty million Raw 264.7 cells were washed twice with serum-free medium and seeded into a single 10-cm culture dish filled with serum-free medium the day before sample preparation. Cells were cultured in fresh serum-free medium with or without 20 μM BB94, 1 μg/ml LPS, or 200 ng/ml TPA for 60 min at 37 °C with 5% CO2. Conditioned media were collected and centrifuged at 9100 ×g for 10 min to remove cells and debris, and proteins were concentrated using VIVASPIN (Sartorius Stedim) up to ~200-fold, purified using a 2-D clean-up kit (GE Healthcare), and solubilized in protein extraction reagent type 4 (Sigma, C0356). Protein concentrations were determined using a protein assay solution (Bio-Rad). Approximately 15 μg of protein was obtained from 2 × 107 cells. 45 μg of each sample was labeled with the IC3-OSu or IC5-OSu fluorescent labeling reagent (Dojindo), combined, and two-dimensionally separated using Immobiline™ DryStrip, pH 4–7, 24 cm (GE Healthcare) and 10% SDS-PAGE gels (20 × 25 cm). Fluorescence images were taken using a Typhoon Trio scanner (GE Healthcare), and differential analysis in single gel and quantitative analysis between multiple gels were performed using DeCyder software version 5.0.1 (GE Healthcare) and ImageMaster 2D Platinum software version 6.0 (GE Healthcare), respectively. For mass spectrometric analysis of the spots of interest, gels were stained using SYPRO Ruby protein gel stain (Life Technologies), and the spots were excised and subjected to in-gel digestion using sequencing grade Trypsin (Promega). The digested peptides were analyzed by nano-LCESI-MS/MS using a DiNa nano-LC system (KYA Technologies) with an L-column 2 ODS 0.075 mm × 100 mm, 3 μm (CERI, Japan) coupled to a QSTAR Elite hybrid liquid chromatography tandem mass spectrometry (LC/MS/MS) system (Applied Biosystems). Accurate mass and precursor ion scanning data were processed using ProteinPilot software version 4.0.8085 (Applied Biosystems). The paragon search method for protein identification was used and all search parameters were as follows: cysteine alkylation by iodoacetamide, ID focus on biological modifications, taxonomy set to Mus musculus. In all cases the data were searched against the UniProt_Sprot_contaminants + iso_08092010 database. The

accuracy tolerance for both peptides and peptide fragments was set to 0.10 Da with a 95% confidence threshold. A protein was considered to be identified when two or more peptides, each with a confidence >95%, were positively matched in the database. Proteins identified in this study had 7 to 11 confident peptides (Table 1).

2.5. ELISA for soluble Mac-1, complex of integrin αM and β2 Conditioned media for ELISA were prepared using serumcontaining medium since serum stabilized the results probably through the inhibition of nonspecific antibody binding. Raw 264.7 cells were cultured in serum-containing medium with or without 20 μM BB94, 1 μg/ml LPS, or 200 ng/ml TPA for 60 min at 37 °C with 5% CO2. Conditioned media were collected and centrifuged at 20,400 ×g for 10 min to remove cells and debris. Maxisorp™ (Nunc) plates were coated with 5 μg/ml of anti-CD11b/integrin αM antibody (M1/70, BioLegend) in 0.1 M carbonate buffer solution (pH 9.6) overnight at 4 °C, washed with PBS containing 0.05% Tween 20 (PBST) five times, and blocked with 5% BSA in PBST overnight at 4 °C. The plates were incubated with samples for 1 h at room temperature, washed with PBST five times, incubated with 5 μg/ml of biotin conjugated anti-CD18/integrin β2 antibody (C71/16, SouthernBiotech) for 30 min at room temperature, washed, and incubated with 5000-fold-diluted HRP-streptavidin (GE Healthcare) for 30 min at room temperature. Following a final wash step, TMB substrate (BD Biosciences) was added and, after the addition of H2SO4, absorbance at 450 nm was measured.

3. Results 3.1. High-dose LPS induces ectodomain shedding in serum-free conditions Fig. 1 shows the strategy of our proteomic screening system for shedding targets. In this system, we screened proteins abundant in the conditioned media of the shedding-positive sample as compared to the shedding-negative sample. In our screening system, the only difference between shedding-positive and shedding-negative samples is the presence of BB94, a hydroxamic acid-derived broad metalloprotease inhibitor. Since BB94 inhibits all metalloprotease-dependent shedding events, we prepared

Table 1 – Identified shedding targets in LPS- and TPA-stimulated macrophages. The fluorescent volume ratio of each spot was calculated in three images each of −/BB94, LPS/LPS + BB94, and TPA/TPA + BB94. The mean ± S.D. values are shown. Protein name VIP36 Sema4D MHC class l α chain L MHC class l α chain D Integrin αM Integrin β2

Accession Calc. mass of Calc. pl of Number of Sequence number ectodomain ectodomain match coverage peptides (%)

Fluorescent volume ratio (−/BB94)

(LPS/LPS + BB94) (TPA/TPA + BB94)

Q9DBH5 O09126 P01897

31.6 78.8 32.9

5.65 8.53 5.19

11 8 11

37.2 14.9 40.9

1.48 ± 0.056 1.24 ± 0.197 1.96 ± 0.361

1.90 ± 0.222 1.87 ± 0.045 2.97 ± 0.161

1.74 ± 0.197 2.12 ± 0.342 3.17 ± 1.058

P01900

33.2

5.52

8

36.2

1.45 ± 0.573

2.24 ± 0.452

2.10 ± 0.223

P05555 P11835

120.7 75.0

6.67 6.32

11 7

26.3 30.6

ND ND

ND ND

2.27 ± 0.447 1.93 ± 0.407

ND means “not detectable”.

236

J O U RN A L OF P ROTE O M IC S 9 8 ( 2 01 4 ) 2 3 3 –24 3

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

shedding-positive sample in the absence of BB94 (Fig. 1, left side) and shedding-negative sample in the presence of BB94 (Fig. 1, right side). For example, when we screened for shedding targets in LPS-stimulated macrophages, LPS-treated (shedding-positive) and LPS + BB94-treated (shedding-negative) macrophages were compared. Thus, the effects of LPS-induced events other than shedding were excluded as both samples were treated with LPS. The two samples were compared by two-dimensional difference gel electrophoresis (2D-DIGE). We labeled proteins in conditioned media from shedding-positive and shedding-negative samples with different fluorescent dyes, IC3 and IC5, combined the samples, and separated proteins on a single two-dimensional gel. The fluorescent signals originating from each sample can be perfectly combined into a single image since the samples were separated in a single gel. We then searched for the proteins more abundant in the shedding-positive sample than in the sheddingnegative sample. Such proteins appear in red in the bottom image of Fig. 1, because shedding targets should accumulate in the conditioned media from shedding-positive samples in the absence of BB94. Since serum contains abundant proteins that interfere with proteomic screening, we modified our previous protocol to completely remove serum proteins from conditioned media. Raw 264.7 cells were washed twice with serum-free media and then pre-cultured in serum-free media for one day. However, we noticed during this process that the removal of serum weakens the activation of shedding by LPS. Western blotting using cell extracts showed that the amount of full-length macrophage colony-stimulating factor (M-CSF) receptor (c-fms), a well-known shedding target in LPS-stimulated macrophages [12], was decreased by LPS treatment in the presence of serum (Fig. 2A, +). On the other hand, in the absence of serum, the amount of full-length M-CSF receptor was effectively unchanged by LPS treatment (Fig. 2A, −). Meanwhile, the amount of full-length M-CSF receptor was decreased by TPA treatment, a different shedding inducer, regardless of whether serum was present or not (Fig. 2A). These results show that serum is indispensable for the effective activation of shedding by LPS but not for the shedding of M-CSF receptor itself. Therefore, we established conditions where LPS can fully activate shedding even in serum-free media to improve our screening system for shedding targets. Because the serum protein lipopolysaccharide-binding protein (LBP) binds to LPS and assists the efficient activation of its receptor, Toll-like receptor 4 (TLR4) [13–15], it is possible that removal of serum might cause a deficiency in LBP resulting in insufficient activation of TLR4 by LPS. If so, a high-dose addition of LPS might overcome this affect. To this end, we evaluated whether 1 μg/ml LPS, a 10-fold higher concentration than used in the previous screening method, could effectively induce shedding even in serum-free conditioned media. Western blotting using cell extracts showed that 1 μg/ml LPS significantly decreased the level of full-length M-CSF receptor, even in serum-free media, to levels similar to those achieved by

237

100 ng/ml LPS in serum-containing media (Fig. 2B). We further examined the release of extracellular domain of M-CSF receptor using N-terminally Halo (Promega)-tagged M-CSF receptor (Halo-M-CSFR) and found that 1 μg/ml LPS potentiates the release of soluble Halo-M-CSFR as same as TPA in serum-free media (Fig. 3E). Thus, high-dose LPS can fully activate shedding even in serum-free media.

3.2. Proteomic screening of shedding targets in LPS-stimulated macrophages We next performed proteomic screening of shedding targets in LPS-stimulated macrophages using high-dose LPS under serum-free conditions. The Raw 264.7 macrophage cell line was cultured in serum-free media containing 1 μg/ml LPS with or without 20 μM BB94 for 60 min, and proteins in the harvested, conditioned media were concentrated and compared using 2D-DIGE as described above. Throughout this study, all of the 2D-DIGE images are presented in a pseudocolored format so that spots of the proteins abundant in shedding-positive samples, i.e., candidates for shedding targets, appear red. We carefully compared conditioned media from LPS-treated samples (LPS) with LPS + BB94-treated samples (LPS + BB94) by exchanging the two fluorescent dyes (Fig. 2C, left panel and Fig. 4A, left panel) in order to avoid any bias derived from a difference in labeling efficiency. Predictably, we found two red spots in both images: that of M-CSF receptor (open arrowhead #1, Figs. 2 and 4) and VIP36 (open arrowhead #2, Figs. 2 and 4), as found in our previous screening assays. Moreover, we found additional red spots in both images and thus performed differential analysis using DeCyder software (GE Healthcare), which calculates the fluorescent volume ratio of the LPS-treated sample to the LPS + BB94-treated sample for individual spots. Spots from both images having a fluorescent volume ratio of more than 1.5 were screened, and the spots that were common between the two images were selected manually. We identified two regions of red spots (Fig. 2C, left panel, white rectangles #a and c, enlarged in Fig. 2D and E) that are common between the two images. Mass spectrometric analysis of the red spots in those regions identified peptides corresponding to the extracellular domain of three different type I membrane proteins, Sema4D (Fig. 2D, open arrowhead #3), MHC class I α chain L (Fig. 2E, open arrowhead #4), and MHC class I α chain D (Fig. 2E, open arrowhead #5), all of which are reported shedding targets [16–18]. The fluorescent volume ratios of these spots confirmed that those proteins were shed in LPS-stimulated Raw 264.7 cells in a metalloprotease-dependent manner (Table 1). These results demonstrated that multiple membrane proteins are simultaneously shed in LPS-stimulated macrophages, and proteomic screening assays using high-dose LPS are much more comprehensive than previous screening assays using lower doses of LPS. To confirm that the shedding targets identified in this study are shed in an LPS-dependent manner, we next performed

Fig. 1 – Strategy for proteomic screening of shedding targets. A mouse macrophage cell line, Raw 264.7, was cultured in the serum-free media containing the indicated combinations of 1 μg/ml LPS, 200 ng/ml TPA, or 20 μM BB94 for 60 min. Conditioned media were harvested after culture and proteins in the media of the different cultures were compared using 2D-DIGE. We then searched for the proteins abundant in shedding-positive samples that appear in red in the bottom image, and identified those proteins by MS analysis.

238

J O U RN A L OF P ROTE O M IC S 9 8 ( 2 01 4 ) 2 3 3 –24 3

proteomic screening of shedding targets in unstimulated macrophages. We compared conditioned media from non-treated samples (−) with BB94-treated samples (BB94) using 2D-DIGE and found multiple red spots corresponding to those found in LPS/LPS + BB94 images in the −/BB94 image, too (Fig. 2C, right panel), even though their red colors appeared to be weaker than those in LPS/LPS + BB94 images. Differential analysis using

DeCyder software confirmed that the fluorescent volume ratios of these spots were smaller than in the LPS/LPS + BB94 images (Table 1). These observations suggested that those proteins are shed even in unstimulated macrophages but their shedding is enhanced by LPS stimulation. We thus performed quantitative analyses across the different images using ImageMaster 2D Platinum software (GE Healthcare), which calculates the relative

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

volume (%Vol) of individual spots. This value indicates the abundance ratio of the fluorescence volume of each spot in relation to the summation of the volume of whole spots in the sample, thus reflecting the relative amount of each spot in the sample. Since all samples contain the same amount of proteins in this study, %Vol is an efficient measure for comparing the amount of individual spots across the different images. Even though the spot of M-CSF receptor was too broad to calculate %Vol without contamination from other proteins, we were able to calculate the %Vol of VIP36 (Fig. 2C, open arrowheads #2 and 7, Fig. 4A open arrowhead #2), Sema4D (Fig. 2D, open arrowheads #3 and 8, Fig. 4A open arrowhead #3), MHC class I α chain L (Fig. 2E, open arrowheads #4 and 9, Fig. 4A open arrowhead #4), and MHC class I α chain D (Fig. 2E, open arrowheads #5 and 10, Fig. 4A open arrowhead #5). We found that the %Vol of VIP36, Sema4D, and the two MHC class I α chains were considerably higher in LPS-treated samples than in non-treated samples (Fig. 3A–D, black bars). In addition, %Vol of those spots was decreased by BB94 (Fig. 3A–D, white bars), confirming that those proteins were shed in a metalloprotease-dependent manner. Taken together, these results showed that the above proteins are continuously shed in macrophages and their shedding is enhanced by LPS stimulation. To confirm the enhancing effect of LPS, we next evaluated the shedding of ectopically expressed Sema4D, as a representative shedding target identified in LPS-stimulated macrophages. Sema4D is a member of the semaphorin family, originally identified as axon-guidance factors in neuronal development [19]. Among family members, Sema4D is known to be an “immune semaphorin” expressed mainly in T cells [20,21], and is shed upon activation of T cells [17,18]. To examine whether shedding of Sema4D is activated in LPS-stimulated macrophages, we expressed N-terminally Halo-tagged Sema4D (Halo-Sema4D) in Raw 264.7 cells. Cells were treated with 1 μg/ml LPS in the presence or absence of 20 μM BB94, and conditioned media and cell extracts were subjected to Western blotting using an anti-Halo antibody (Fig. 3E). Halo-tagged, soluble Sema4D with an apparent molecular mass corresponding to the entire extracellular domain of Sema4D was released into conditioned medium in the absence of LPS, and LPS potentiated the release of soluble Halo-Sema4D (Fig. 3E, top panel, arrowhead). BB94 completely inhibited the release of soluble Halo-Sema4D (Fig. 3E, top panel, arrowhead), demonstrating that the release is metalloprotease-dependent. We confirmed that neither LPS nor BB94 treatment altered the amount of full-length

239

Halo-Sema4D in the cell (Fig. 3E, middle panel). As expected, LPS activated the shedding of M-CSF receptors in the same experiment (Fig. 3E, lower panel, arrowhead). These results confirm that Sema4D is constantly shed and LPS enhances its shedding. Our results collectively suggest that LPS globally activates the shedding of various targets in macrophages.

3.3. Comparison of shedding targets in LPS-stimulated macrophages with those in TPA-stimulated macrophages We next performed proteomic screening of shedding targets in TPA-stimulated macrophages to evaluate whether different stimuli can activate different types of shedding in the same Raw 264.7 cells. We compared conditioned media from TPA-treated samples (TPA) with TPA + BB94-treated samples (TPA + BB94) and found that all of the red spots found in LPS/LPS + BB94 images appeared in red in the TPA/TPA + BB94 image. Furthermore, the red colors of those spots appeared to be as strong as those in the LPS/LPS + BB94 image (Fig. 4A, right panel). Differential analysis using DeCyder software showed that VIP36 (open arrowhead #7), Sema4D (open arrowhead #8), MHC class I α chain L (open arrowhead #9), and MHC class I α chain D (open arrowhead #10) were more abundant in the TPA-treated sample compared to the TPA + BB94-treated sample (Table 1), confirming that those proteins were also shed in TPA-stimulated macrophages in a metalloprotease-dependent manner. In addition, quantitative analysis demonstrated that the %Vol of these spots was higher in TPA-treated samples than in non-treated samples (Fig. 3A–D, black bars), confirming that the shedding of those proteins is similarly activated by TPA. Correspondingly, the release of soluble Halo-Sema4D was considerably potentiated by TPA treatment (Fig. 3E, top panel, arrowhead). Taken together, these results demonstrated that TPA, like LPS, also activates shedding. In addition to the shedding targets observed in LPSstimulation, we found two rows of red spots in the TPA/ TPA + BB94 image (Fig. 4A, right panel, white rectangle #b, enlarged in Fig. 4B), and mass spectrometric analysis of these spots identified peptides corresponding to the extracellular domain of integrin αM (Fig. 4B, open arrowhead #11) and integrin β2 (Fig. 4B, open arrowhead #12). The fluorescent volume ratio of these spots confirmed that these proteins were shed in TPA-stimulated Raw 264.7 cells in a metalloprotease-dependent manner (Table 1). Since integrin αM and β2 together constitute a

Fig. 2 – Proteomic screening of shedding targets in LPS-stimulated macrophages. (A) Raw 264.7 cells were treated with 100 ng/ml LPS or 200 ng/ml TPA for 30 min in the presence (+) or absence (−) of serum, and cell extracts were subjected to Western blotting (IB) with an antibody against the M-CSF receptor (M-CSFR), a known shedding target. The closed arrowhead indicates the full-length M-CSF receptor. (B) Raw 264.7 cells were treated with 100 ng/ml LPS in the presence of serum (serum+) or 1 μg/ml LPS in the absence of serum (serum−) for the indicated times, and cell extracts were subjected to Western blotting with anti-M-CSF receptor antibody. A closed arrowhead indicates the full-length M-CSF receptor. (C) Conditioned media from LPS-treated (LPS) and LPS + BB94-treated (LPS + BB94) Raw 264.7 cells were collected separately and subjected to 2D-DIGE to screen shedding targets (left panel). Conditioned media from non-treated (−) and BB94-treated (BB94) Raw 264.7 cells were also compared (right panel). The images were pseudo-colored so that shedding targets appear in red. White boxes indicate the regions enlarged in D and E. Open arrowheads indicate the spots of M-CSF receptor (#1 and #6) and VIP36 (#2 and #7), which were subjected to protein identification, differential analysis and quantitative analysis. (D and E) Regions containing red spots are enlarged. Arrowheads indicate all red spots in the images, and among them, open arrowheads with numbers indicate the spots subjected to protein identification, differential analysis and quantitative analysis (Sema4D (#3 and #8), MHC class I α chain L (#4 and #9), and MHC class I α chain D (#5 and #10)).

240

J O U RN A L OF P ROTE O M IC S 9 8 ( 2 01 4 ) 2 3 3 –24 3

cell adhesion molecule, Mac-1 [11], these results suggested that TPA specifically activates the shedding of Mac-1. To evaluate whether Mac-1 is shed as a complex of integrin αM and β2, we

performed an enzyme-linked immunoassay (ELISA) using two different antibodies recognizing integrin αM and β2. Raw 264.7 cells were treated with 1 μg/ml LPS or 200 ng/ml TPA in the

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

presence or absence of 20 μM BB94, and conditioned media were subjected to ELISA using anti-integrin αM monoclonal antibody (M1/70) as a capture antibody and anti-integrin β2 monoclonal antibody (C71/16) as a detection antibody [22]. As shown in Fig. 5, TPA considerably increased the amount of soluble Mac-1 in the conditioned media and BB94 canceled this increase, indicating that Mac-1 was shed in TPA-stimulated macrophages in a metalloprotease-dependent manner. In contrast, LPS did not increase the amount of soluble Mac-1. Western blotting of full-length M-CSF receptor confirmed that both LPS and TPA activated the shedding of M-CSF receptor in a metalloproteasedependent manner in the same experiment (Fig. 5, lower panel). These results clearly show that the shedding of Mac-1 is specifically activated by TPA but not by LPS, and demonstrate that different extracellular stimuli activate different types of shedding.

4. Discussion and conclusions In this study, we have improved our proteomic screening system for the identification of shedding targets and succeeded in acquiring a comprehensive picture of endogenous membrane proteins shed under different conditions. Our observations provide a wealth of information concerning the regulatory mechanisms of shedding. First, we reveal that multiple proteins are constitutively shed in the absence of extracellular stimuli, demonstrating that shedding is not regulated in an all-or-none fashion. In addition, we reveal that both LPS and TPA enhance shedding of the same set of proteins that are constantly being shed. These results raise the possibility that each cell has an intrinsic pattern of constant shedding that is further activated by different extracellular stimuli. Furthermore, we discovered that TPA can specifically activate additional shedding, the shedding of Mac-1, that is not activated by LPS, demonstrating that different extracellular stimuli can activate different types of shedding even in a single cell. The physiological stimuli leading to the shedding of Mac-1 remain to be determined. Our study proves that shedding is strictly regulated in both a stimulus- and substrate-specific manner. How does TPA specifically activate the shedding of Mac-1? Recently, several molecular mechanisms enabling stimulusand substrate-specific shedding were reported. Those include interaction between ADAMs and a membrane protein, iRhom2 [23], and, in one case, phosphorylation of the cytoplasmic domain of a shedding target, neuregulin [24]. In the case of Mac-1, integrin β2, a component of Mac-1, has been identified as a substrate of MMP-9 (matrix metalloproteinase-9) through

241

the proteomic screening of proteins released into conditioned media by the overexpression of MMP-9 in Raw 264.7 cells [9]. Thus it is possible that TPA, but not LPS, activates MMP-9, and activated MMP-9 specifically causes shedding of Mac-1. TPA is known to be a strong shedding inducer that can activate almost all reported shedding in most cell types. Such a strong shedding-inducible ability of TPA is likely due to its capability to activate a wide range of metalloproteases. Our results indicate that Mac-1 is shed as a functional complex of integrin αM and β2 in TPA-stimulated Raw 264.7 cells. Firstly, the spots of integrin αM and β2 were simultaneously found in the 2D-DIGE images comparing TPA-treated and TPA + BB94-treated conditioned media. Secondly, ELISA using two different antibodies against integrin αM and β2 detected the increase of soluble Mac-1 complexes in TPAtreated conditioned media. Intriguingly, however, flow cytometric analysis revealed that cell surface levels of Mac-1 were barely decreased by TPA-treatment, whereas levels of M-CSF receptor were markedly decreased (data not shown). In addition, we found that the expression level of MMP-9, a potential shedding protease of Mac-1, was not changed by TPA-treatment in whole cell extract (data not shown). Taken together, we now speculate that MMP-9 is activated through some posttranslational modification mechanisms in restricted region of the cell, and thus shedding of Mac-1 also occurs in a restricted region. One possible region is the podosome, an actin-rich protrusion of the cell serving as a site of attachment and degradation of the extracellular matrix [25]. The podosome regionally regulates the adhesion of macrophages to the extracellular matrix. On the other hand, since soluble Mac-1 reportedly has the capacity to bind its ligands [22], shedding of Mac-1 may regulate the migratory capacity of macrophages through the release of soluble Mac-1, which would serve as a competitive antagonist of cell surface Mac-1. How do both LPS and TPA activate the same set of constant shedding, even though they activate different intracellular signaling pathways? We found in this study that a serum protein(s), probably LBP, is necessary for the full activation of shedding by LPS, indicating that the activation of TLR4 receptor is necessary for the effective activation of shedding by LPS. Identification of the signaling pathways activating shedding under the control of TLR4 should be the first step to examine how those different stimuli activate the same set of shedding. Proteomic screening of shedding targets in the presence of specific inhibitors of TLR4 signaling pathways would be helpful. In this study, we demonstrated that the proteomic approach is effective in elucidating the stimulus- and substrate-specificity

Fig. 3 – Both LPS and TPA activate constant shedding of the same proteins. (A–D) %Vol of the spots of VIP36 (A), Sema4D (B), MHC class I α chain L (C), and MHC class I α chain D (D) was calculated from samples prepared in the presence (white bars) and absence (black bars) of BB94 using ImageMaster 2D Platinum software from three images each of −/BB94 (−), LPS/LPS + BB94 (LPS), and TPA/TPA + BB94 (TPA). Error bars indicate the S.D. (E) N-terminally Halo-tagged Sema4D (top three panels) or M-CSF receptor (bottom panel) was expressed in Raw 264.7 cells and cells were treated with 1 μg/ml LPS or 200 ng/ml TPA in the presence or absence of 20 μM BB94. Both cell extracts (second and third panels) and culture supernatants (top and bottom panels) were subjected to Western blotting with anti-Halo antibody (top two panels and bottom panel) or anti-M-CSF receptor antibody (third panel). Closed arrowheads indicate soluble Halo-Sema4D (top panel), full-length M-CSF receptor (third panel), and soluble Halo-M-CSF receptor (bottom panel). Similar results were observed in at least three independent experiments.

242

J O U RN A L OF P ROTE O M IC S 9 8 ( 2 01 4 ) 2 3 3 –24 3

Fig. 4 – Comparison of shedding targets in LPS-stimulated versus TPA-stimulated macrophages. (A) Conditioned media from LPS-treated (LPS) and LPS + BB94-treated (LPS + BB94) Raw 264.7 cells were collected separately and subjected to 2D-DIGE to screen shedding targets (left panel). Conditioned media from TPA-treated (TPA) and TPA + BB94-treated (TPA + BB94) Raw 264.7 cells were also compared (right panel). The images were pseudo-colored so that shedding targets appear in red. Red spots are indicated by arrowheads, and among them, open arrowheads indicate the spots of M-CSF receptor (#1 and #6), VIP36 (#2 and #7), Sema4D (#3 and #8), MHC class I α chain L (#4 and #9), and MHC class I α chain D (#5 and #10), which were subjected to differential analysis and quantitative analysis. White boxes indicate the regions enlarged in B. (B) Two rows of red spots are indicated by arrowheads, and among them, open arrowheads indicate the spots of integrin αM (#11) and β2 (#12), which were further subjected to protein identification, differential analysis, and quantitative analysis. White dots indicated the spots of Sema4D appearing in both images.

of endogenous shedding. We have also used a different screening method, SILAC (stable isotope labeling using amino acids in cell culture) using conditioned media prepared in the same manner as in this study, and succeeded in identifying additional shedding targets in LPS-stimulated macrophages (data not shown). Since 2D-DIGE and SILAC identified different sets of shedding targets,

the combination of these two methods should enable an even more comprehensive screening of shedding targets. In addition, further screening using different stimuli and/or different cell types would give us a more comprehensive understanding of the stimulus- and substrate-specificity of endogenous shedding and the regulatory mechanisms of endogenous shedding in future.

J O U RN A L OF P ROT EO M IC S 9 8 ( 2 01 4 ) 2 3 3 –2 43

Fig. 5 – TPA specifically activates the shedding of Mac-1. (A) Raw 264.7 cells were treated with 1 μg/ml LPS or 200 ng/ml TPA in the presence (white bars) or absence (black bars) of 20 μM BB94, and conditioned media were subjected to ELISA using two different antibodies recognizing integrin αM and β2. The experiments were repeated three times. Error bars indicate the S.D. p values were calculated by Student's t test. *, p < 0.005 and **, p < 0.05. (B) In the same experiments, cell extracts were subjected to Western blotting with anti-M-CSF receptor antibody (bottom panel). The closed arrowhead indicates the full-length M-CSF receptor.

Acknowledgments We thank Ms. Akiko Minowa and Dr. Hideki Fujii for the technical advices about ELISA. This work was supported in part by the Improvement of Research Environment for Young Researchers, and the Funds for the Development of Human Resources in Science and Technology (to K.S.) from MEXT. S.K. is a consultant for Medical and Biological Laboratories, Co. Ltd. The authors otherwise declare no conflict of interest.

REFERENCES [1] Blobel CP. ADAMs: key components in EGFR signalling and development. Nat Rev Mol Cell Biol 2005;6:32–43. [2] Huovila AP, Turner AJ, Pelto-Huikko M, Karkkainen I, Ortiz RM. Shedding light on ADAM metalloproteinases. Trends Biochem Sci 2005;30:413–22. [3] Kheradmand F, Werb Z. Shedding light on sheddases: role in growth and development. Bioessays 2002;24:8–12.

243

[4] Seals DF, Courtneidge SA. The ADAMs family of metalloproteases: multidomain proteins with multiple functions. Genes Dev 2003;17:7–30. [5] Guo L, Eisenman JR, Mahimkar RM, Peschon JJ, Paxton RJ, Black RA, et al. A proteomic approach for the identification of cell-surface proteins shed by metalloproteases. Mol Cell Proteomics 2002;1:30–6. [6] Morrison CJ, Butler GS, Rodriguez D, Overall CM. Matrix metalloproteinase proteomics: substrates, targets, and therapy. Curr Opin Cell Biol 2009;21:645–53. [7] Overall CM, Blobel CP. In search of partners: linking extracellular proteases to substrates. Nat Rev Mol Cell Biol 2007;8:245–57. [8] Schilling O, Overall CM. Proteomic discovery of protease substrates. Curr Opin Chem Biol 2007;11:36–45. [9] Vaisar T, Kassim SY, Gomez IG, Green PS, Hargarten S, Gough PJ, et al. MMP-9 sheds the beta2 integrin subunit (CD18) from macrophages. Mol Cell Proteomics 2009;8:1044–60. [10] Shirakabe K, Hattori S, Seiki M, Koyasu S, Okada Y. VIP36 protein is a target of ectodomain shedding and regulates phagocytosis in macrophage Raw 264.7 cells. J Biol Chem 2011;286:43154–63. [11] Hynes RO. Integrins: bidirectional, allosteric signaling machines. Cell 2002;110:673–87. [12] Rovida E, Paccagnini A, Del Rosso M, Peschon J, Dello Sbarba P. TNF-alpha-converting enzyme cleaves the macrophage colony-stimulating factor receptor in macrophages undergoing activation. J Immunol 2001;166:1583–9. [13] Miyake K. Roles for accessory molecules in microbial recognition by Toll-like receptors. J Endotoxin Res 2006;12:195–204. [14] Tobias PS, Mathison JC, Ulevitch RJ. A family of lipopolysaccharide binding proteins involved in responses to gram-negative sepsis. J Biol Chem 1988;263:13479–81. [15] Tobias PS, Soldau K, Ulevitch RJ. Identification of a lipid A binding site in the acute phase reactant lipopolysaccharide binding protein. J Biol Chem 1989;264:10867–71. [16] Demaria S, Schwab R, Gottesman SR, Bushkin Y. Soluble beta 2-microglobulin-free class I heavy chains are released from the surface of activated and leukemia cells by a metalloprotease. J Biol Chem 1994;269:6689–94. [17] Wang X, Kumanogoh A, Watanabe C, Shi W, Yoshida K, Kikutani H. Functional soluble CD100/Sema4D released from activated lymphocytes: possible role in normal and pathologic immune responses. Blood 2001;97:3498–504. [18] Elhabazi A, Delaire S, Bensussan A, Boumsell L, Bismuth G. Biological activity of soluble CD100. I. The extracellular region of CD100 is released from the surface of T lymphocytes by regulated proteolysis. J Immunol 2001;166:4341–7. [19] Kolodkin AL, Matthes DJ, Goodman CS. The semaphorin genes encode a family of transmembrane and secreted growth cone guidance molecules. Cell 1993;75:1389–99. [20] Delaire S, Elhabazi A, Bensussan A, Boumsell L. CD100 is a leukocyte semaphorin. Cell Mol Life Sci 1998;54:1265–76. [21] Takamatsu H, Kumanogoh A. Diverse roles for semaphorin–plexin signaling in the immune system. Trends Immunol 2012;33:127–35. [22] Gomez IG, Tang J, Wilson CL, Yan W, Heinecke JW, Harlan JM, et al. Metalloproteinase-mediated Shedding of Integrin beta2 promotes macrophage efflux from inflammatory sites. J Biol Chem 2012;287:4581–9. [23] Maretzky T, McIlwain DR, Issuree PD, Li X, Malapeira J, Amin S, et al. iRhom2 controls the substrate selectivity of stimulated ADAM17-dependent ectodomain shedding. Proc Natl Acad Sci U S A 2013;110:11433–8. [24] Dang M, Armbruster N, Miller MA, Cermeno E, Hartmann M, Bell GW, et al. Regulated ADAM17-dependent EGF family ligand release by substrate-selecting signaling pathways. Proc Natl Acad Sci U S A 2013;110:9776–81. [25] Linder S, Aepfelbacher M. Podosomes: adhesion hot-spots of invasive cells. Trends Cell Biol 2003;13:376–85.