Subunit Α Receptor Shedding of the Cell Surface GM-CSF Monocytes

Granulocyte-Macrophage Colony-Stimulating Factor (GM-CSF) and Inflammatory Stimuli Up-Regulate Secretion of the Soluble GM-CSF Receptor in Human This information is current as Monocytes: Evidence for Ectodomain of September 25, 2021. Shedding of the Cell Surface GM-CSF Receptor α Subunit Jay M. Prevost, Jennifer L. Pelley, Weibin Zhu, Gianni E. D'Egidio, Paul P. Beaudry, Carin Pihl, Graham G. Neely, Downloaded from Emmanuel Claret, John Wijdenes and Christopher B. Brown J Immunol 2002; 169:5679-5688; ; doi: 10.4049/jimmunol.169.10.5679 http://www.jimmunol.org/content/169/10/5679 http://www.jimmunol.org/

References This article cites 39 articles, 19 of which you can access for free at: http://www.jimmunol.org/content/169/10/5679.full#ref-list-1

Why The JI? Submit online. by guest on September 25, 2021 • Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2002 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Granulocyte-Macrophage Colony-Stimulating Factor (GM-CSF) and Inﬂammatory Stimuli Up-Regulate Secretion of the Soluble GM-CSF Receptor in Human Monocytes: Evidence for Ectodomain Shedding of the Cell Surface GM-CSF Receptor ␣ Subunit1

Jay M. Prevost,2* Jennifer L. Pelley,2*† Weibin Zhu,* Gianni E. D’Egidio,* Paul P. Beaudry,‡ Carin Pihl,* Graham G. Neely,¤ Emmanuel Claret,¶ John Wijdenes,¶ and 3 ʈ

Christopher B. Brown * Downloaded from

Soluble GM-CSF receptor ␣ subunit (sGMR␣) is a soluble isoform of the GMR␣ that is believed to arise exclusively through alternative splicing of the GMR␣ gene product. The sGMR␣ mRNA is expressed in a variety of tissues, but it is not clear which cells are capable of secreting the protein. We show here that normal human monocytes, but not lymphocytes, constitutively secrete sGMR␣. Stimulation of monocytes with GM-CSF, LPS, PMA, or A23187 rapidly up-regulates the secretion of sGMR␣ in a http://www.jimmunol.org/ dose-dependent manner, demonstrating that secretion is also regulated. To determine whether sGMR␣ arose exclusively through alternative splicing of the GMR␣ gene product, or whether it could also be generated through ectodomain shedding of GMR␣, we engineered a murine pro-B cell line (Ba/F3) to express exclusively the cDNA for cell surface GMR␣ (Ba/F3.GMR␣). The Ba/F3.GMR␣ cell line, but not the parental Ba/F3 cell line, constitutively shed a sGMR␣-like protein that bound specifically to GM-CSF, was equivalent in size to recombinant alternatively spliced sGMR␣ (60 kDa), and was recognized specifically by a mAb raised against the ectodomain of GMR␣. Furthermore, a broad-spectrum metalloprotease inhibitor (BB94) reduced constitutive and PMA-, A23187-, and LPS-induced secretion of sGMR␣ by monocytes, suggesting that shedding of GMR␣ by monocytes may be mediated in part through the activity of metalloproteases. Taken together, these observations demonstrate that sGMR␣ is constitutively secreted by monocytes, that GM-CSF and inflammatory mediators up-regulate sGMR␣ secretion, and that sGMR␣ by guest on September 25, 2021 arises not only through alternative splicing but also through ectodomain shedding of cell surface GMR␣. The Journal of Im- munology, 2002, 169: 5679Ð5688.

M-CSF is a multifunctional cytokine that mediates its GMR␣ is also expressed as an alternatively spliced soluble iso- hematopoietic and inﬂammatory activities (1) through a form, soluble GMR␣ (sGMR␣). sGMR␣ arises through alternative membrane-spanning cell surface receptor that consists of splicing of the parental GMR␣ gene product (4Ð7). Exclusion of G 4 a ligand-binding subunit, GM-CSF receptor ␣ subunit (GMR␣) the exon that encodes the transmembrane domain of GMR␣ results (2), and a signal-transducing subunit, ␤c (3). GMR␣ is a 378-aa in translation of a protein that is missing both the membrane-span- protein that consists of a 298-aa extracellular domain, a 26-aa ning domain of GMR␣ and the cytoplasmic domain, but that re- membrane-spanning domain, and a 54-aa cytoplasmic domain (2). tains the 295-aa GMR␣ ectodomain. In addition, the alternative splicing event generates a frameshift in the coding sequence that results in the addition of 16-aa to the carboxy terminus of the *Cancer Biology Research Group, Southern Alberta Cancer Research Center, De- sGMR␣ protein. partments of †Biochemistry and Molecular Biology, ‡Surgery, and ¤Medical Science, sGMR␣ has been cloned and expressed as a recombinant protein University of Calgary, Calgary, Alberta, Canada; ¶Diaclone Research, Besancon, ʈ ␣ France; and Alberta Bone Marrow and Stem Cell Transplant Program (5Ð7). Recombinant sGMR is a heavily glycosylated 60-kDa pro- Received for publication January 16, 2002. Accepted for publication September tein that binds GM-CSF in solution with the same afﬁnity as cell ␣ ϭ 17, 2002. surface GMR (Kd 2Ð10 nM) (2, 7). However, unlike cell sur- The costs of publication of this article were defrayed in part by the payment of page face GMR␣, which binds GM-CSF then associates with ␤c, an charges. This article must therefore be hereby marked advertisement in accordance exogenous source of sGMR␣ added to ␤c-expressing cells will not with 18 U.S.C. Section 1734 solely to indicate this fact. associate with ␤c, whether in the presence or absence of GM-CSF 1 This work was supported by grants from the Canadian Institutes of Health Research and the Arthritis Society of Canada (to C.B.B.). J.L.P. is the recipient of a studentship (8, 9). Instead, sGMR␣ antagonizes the function of GM-CSF in award from the Canadian Institutes of Health Research. vitro by binding to it and preventing it from interacting with the 2 J.M.P. and J.L.P. contributed equally to this work. cell surface GMR␣/␤c complex (7). sGMR␣ has also been shown 3 Address correspondence and reprint requests to Dr. Christopher B. Brown, Depart- to inhibit the proliferation of GM-CSF-dependent cell lines (5, 6, ments of Medicine and Oncology, University of Calgary, Heritage Medical Research Building, Room 311, 3330 Hospital Drive NW, Calgary, Alberta, Canada T2N 4N1. 10) and to inhibit bone marrow colony formation (7). E-mail address: [email protected] The alternatively spliced GMR␣ message is expressed by a va- 4 Abbreviations used in this paper: GMR␣, GM-CSF receptor ␣ subunit; s, soluble. riety of cell types, including human placental tissue (7), myeloid

leukemic cell lines (5, 6, 11), bone marrow progenitors, monocyte/ sity centrifugation over Ficoll. The enriched mononuclear cell layer typi- ϩ macrophages, and synovial fibroblasts (6). Less is known about the cally consisted of 75Ð85% CD14 monocytes. ␣ expression of the alternatively spliced sGMR protein. A soluble Flow cytometry GM-CSF binding protein was identified in medium conditioned by a human choriocarcinoma cell line (JEG-3) (12), but no immuno- The percentage of monocytes in either the PBMC preparations or the monocyte-enriched/depleted preparations was determined by CD14 stain- logical or biochemical data were presented as to whether this pro- ing and FACS. Briefly, 106 cells were stained for 15 min on ice with 0.5 tein was equivalent to alternatively spliced sGMR␣. However, a ␮g of a FITC-labeled mouse anti-human CD14 mAb or an IgG2a-FITC 60-kDa soluble GM-CSF binding protein was recently identified in isotype control mAb (BD PharMingen, San Diego, CA). The cells were medium conditioned by human myeloid leukemic cell lines (U937, washed and fixed in PBS containing 1% PBS-buffered formalin. Data were THP-1, and HL-60), by normal human granulocytes, and in normal acquired using a FACStation flow cytometer (BD Biosciences) and were analyzed using Flowjo (Treestar, San Carlos, CA). The expression of human plasma (13). This protein migrated as a 60-kDa band by GMR␣ on the cell surface was determined in a similar manner using 1 ␮g SDS-PAGE, was recognized by a mAb that was raised against the of a mouse anti-human GMR␣ mAb (8G6; a gift of Dr. A. Lopez, Hansen ectodomain of GMR␣, and bound GM-CSF with the same affinity Center for Cancer Research, Adelaide, Australia) or an equivalent murine ␮ as recombinant alternatively spliced sGMR␣, suggesting that these IgG1 isotype control (Sigma-Aldrich), followed by staining with 0.5 gof a PE-labeled F(abЈ) goat anti-mouse Ab (Calbiochem). two proteins were equivalent. 2 We were interested in determining whether normal human Detection of total sGMR␣ protein by ELISA ␣ monocytes, which express both cell surface GMR and the mRNA Supernatants from cell culture experiments were screened by ELISA for for sGMR␣, could also secrete the alternatively spliced sGMR␣ the presence of all sGMR␣ protein (sCD116 ELISA; Diaclone Research, protein. We were also interested in determining whether the se- Besancon, France). The sCD116 ELISA uses a capture mAb raised against Downloaded from cretion of alternatively spliced sGMR␣ by monocytes could be the ectodomain of GMR␣ (SCO6), whereas the detection mAb (SCO4) was also raised against the ectdomain of GMR␣. Because of this, the sCD116 regulated by GM-CSF and other stimuli. Using an ELISA that ␣ ␣ ELISA is not specific for alternatively spliced sGMR but instead recog- recognized all soluble isoforms of GMR , we found that mono- nizes any sGMR␣ isoform that shares the common ectodomain of GMR␣. cytes but not lymphocytes could constitutively secrete a sGMR␣- Linear regression analysis of ELISA data was performed using Graph Pad like protein. We also found that stimulation of monocytes with (Prism, San Diego, CA). All cultures were performed in duplicate and were measured again in duplicate by ELISA. Experiments were repeated with GM-CSF and mediators such as LPS, PMA, and the calcium iono- http://www.jimmunol.org/ cells from an individual donor or from different donors, as indicated. phore A23187 rapidly up-regulates the secretion of this sGMR␣- like protein. Importantly, this protein did not arise exclusively Engineering of the pBabePuro3/GMR␣ retroviral construct ␣ through the alternative splicing of GMR , but was also derived ␣ ␣ The cDNA for GMR was generated by PCR from the previously de- through ectodomain shedding of cell surface GMR , demonstrat- scribed ␭-gt11/GMR␣ clone (7) with primers designed to amplify the en- ing for the first time that sGMR␣ can arise through both alternative tire coding sequence. The blunt-end PCR product was cloned into the pCR- ϩ splicing and proteolytic cleavage of cell surface GMR␣. Script Amp SK cloning vector (Stratagene, La Jolla, CA). The full sequence of the cDNA was confirmed by sequencing and was subse- quently subcloned into the XhoI-ClaI sites of the retrovirus expression Materials and Methods vector pBabePuro3 (gift of Dr. S. Robbins). by guest on September 25, 2021 Cytokines, reagents, and cell culture Retroviral infection of the Ba/F3 cell line Recombinant human GM-CSF (a gift of Cangene, Mississauga, ON, Can- ada) and G-CSF (Amgen Canada, Mississauga, ON, Canada) were recon- The murine IL-3-dependent pro-B cell line Ba/F3 was kindly provided by stituted in sterile PBS. LPS (Escherichia coli serotype 0111:B4; Sigma- Dr. K. Kaushansky (University of Washington, Seattle, WA) and was Aldrich, St. Louis, MO) was reconstituted in PBS. PMA (Sigma-Aldrich) maintained in RPMI 1640 medium (Life Technologies) with 10% FBS ␺ and A23187 (Calbiochem, La Jolla, CA) were solubilized in DMSO (Sig- supplemented with 1 ng/ml murine IL-3 (BD PharMingen). The 2 eco- ma-Aldrich) and diluted in PBS. The broad-spectrum hydroxamic acid- tropic retroviral packaging cell line was maintained in DMEM (Life Tech- based inhibitor of metalloprotease activity (BB94; British Biotechnology, nologies) plus 10% FBS. Retroviral infection was performed using stably ␺ ␺ Oxford, U.K.) was solubilized in DMSO and used directly. Recombinant transfected 2 packaging cells as follows: 2 cells were transfected with ␣ alternatively spliced sGMR␣ was purified from medium conditioned by a pBabePuro3/GMR using Effectene Reagent (Qiagen, Valencia, CA) and ␮ baby hamster kidney fibroblast cell line engineered to express the alterna- stable transfectants were selected with 3 g/ml puromycin. These retrovi- ␺ ϫ 6 tively spliced sGMR␣ cDNA (BHK.sGMR␣) as previously described (7). rus-producing 2 cells were grown to subconfluence (5 10 cells/10-cm Similarly, a recombinant isoform of sGMR␣ that was missing the unique dish), and 10 ml of fresh medium was added. The viral supernatant was ␮ 16-aa C-terminal domain of alternatively spliced sGMR␣ was purified harvested 18 h later and filtered through a 0.45- m membrane. Six micro- ϫ 5 from medium conditioned by the BHK.ECD␣ cell line (9, 14). Unless liters of the supernatant was applied to 5 10 cells in a 10-cm dish ␮ otherwise indicated, experiments were performed in sterile 1.5-ml Eppen- containing 6 ml of fresh medium and 8 g/ml Polybrene. The infected cells ␮ dorf tubes in HEPES-modified RPMI 1640 (Sigma-Aldrich) containing 1% were grown for 48 h and then were selected in medium containing 4 g/ml ␣ ␣ penicillin/streptomycin (Life Technologies, Rockville, MD) and supple- puromycin. Cell surface expression of GMR on the Ba/F3.GMR cell mented with 10% heat-inactivated FBS (Sigma-Aldrich) (complete me- line, or lack thereof on the Ba/F3 cell line, was confirmed by flow cytometry. dium) in a 37¡C incubator containing 5% CO2. Cell isolation GM-CSF ligand-affinity chromatography GM-CSF binding proteins were isolated from medium conditioned by the Blood was collected from healthy volunteers by venipuncture after in- Ba/F3 cell line (240 ml) or the Ba/F3.GMR␣ cell line (120 ml) using formed consent. Blood was collected into vacuum containers containing affinity chromatography as previously described (7Ð9, 13Ð15). Briefly, cell- sodium heparin (Vacutainer; BD Biosciences, Mountain View, CA). conditioned medium was passed over a Sepharose 4B column (Pharmacia) PBMCs were isolated by density centrifugation of diluted blood over a that had been coupled to recombinant human GM-CSF. The column was cushion of Ficoll (Ficoll-Paque Plus; Pharmacia, Peapack, NJ) followed by washed extensively with PBS and the adsorbed proteins were eluted with osmotic lysis of residual erythrocytes. PBMC viability was typically 98% a 0.1 M glycine buffer (pH 2.5). The eluted fractions were immediately and contaminating granulocytes accounted for Ͻ3% of cells as assessed by neutralized with 1 M Tris buffer. The neutralized fractions were volume light scatter and flow cytometry. For lymphocyte isolation, PBMCs were reduced to ϳ5 ␮l by centrifugal filtration (Ultrafree, 5 kDa; Millipore, depleted of monocytes by plating them on polystyrene petri dishes. The Bedford, MA) and were analyzed by SDS-PAGE and Western blotting. nonadherent cells were decanted and the procedure was repeated a total of ϩ three times. The lymphocyte preparations contained Ͻ1% CD14 cells as SDS-PAGE and Western blotting assessed by flow cytometry. Monocytes were isolated from whole blood using a negative selection procedure (RosetteSep Monocyte Isolation Kit; The concentrated eluents were boiled in an equal volume of 2ϫ SDS- StemCell Technologies, Vancouver, British Columbia, Canada) and den- PAGE loading buffer in the presence of the reducing agent DTT. The The Journal of Immunology 5681 solubilized and denatured proteins were electrophoresed on 8% SDS poly- acrylamide minigels and transferred to polyvinylidene fluoride membranes (BioTrace PVDF; Pall Corporation, Ann Arbor, MI). The blots were blocked in 5% skim milk in TBST and were probed with a mouse mAb raised against the ectodomain of GMR␣ (8D10; a gift of Dr. A. Lopez) followed by an HRP-conjugated rabbit anti-mouse IgG secondary Ab (Bio/ Can Scientific, Mississauga, Ontario, Canada). Proteins on the blots were visualized by ECL (Amersham Life Sciences, Oakville, Ontario, Canada) and exposure to x-ray film. Preparation of rabbit splicing-specific antiserum against the 16-aa C-terminal tail of alternatively spliced sGMR␣ Peptide corresponding to the predicted 16-aa sequence of the alternatively spliced sGMR␣ C-terminal tail (LGYSGCSRQFHRSKTN) was synthe- sized and coupled to the carrier protein KLH by the Peptide Synthesis Core Facility of the University of Calgary. Female New Zealand white rabbits were immunized by bilateral i.m. injection with a total of 50 ␮g of KLH- conjugated peptide in CFA (200 ␮l total volume). Animals were boosted every 3Ð4 wk with an identical dose of peptide in IFA. A total of five injections were administered before the initial test bleed. An affinity column was prepared by covalently attaching non-KLH-coupled 16-aa peptide

to N-hydroxysuccinimide-activated Sepharose 4 Fast Flow resin (Pharma- Downloaded from cia), according to the manufacturer’s instructions. The rabbit antiserum was then affinity purified by passage first over a Sepharose 4B sham column and then passage over the peptide-conjugated affinity column. Affin- ity-purified Ab was then eluted from the column using a 0.1 M glycine buffer (pH 2.5) and then immediately neutralized with 1 M Tris. Statistics Data from representative and replicate experiments are presented as the http://www.jimmunol.org/ mean Ϯ SEM, whereas the significance of the differences between groups was determined by paired or unpaired t test, as applicable to the assay condition, with p Ͻ 0.05 being deemed a significant change. Results sGMR␣ is secreted constitutively by monocytes but not by lymphocytes Mature human monocytes express GMR␣ on their cell surface whereas lymphocytes do not (16Ð18), suggesting that monocytes by guest on September 25, 2021 but not lymphocytes might also secrete sGMR␣. To test this hypothesis, we first confirmed that monocytes could express GMR␣ on their cell surface. PBMCs were stained with an anti-CD14- FITC mAb and an anti-GMR␣ mAb (8G6) or with an IgG1 isotype control and were analyzed by flow cytometry. The majority of the CD14ϩ monocytes, but none of the CD14Ϫ lymphocytes, stained positively with the anti-GMR␣ Ab but not with the IgG1 isotype control, confirming that monocytes but not lymphocytes express GMR␣ on their cell surface (Fig. 1A). To determine whether monocytes could also secrete the sGMR␣ protein, we incubated 6 PBMCs from healthy donors at a density of 10 /ml (equivalent to FIGURE 1. sGMR␣ is secreted constitutively by monocytes but not by ϳ ϫ 5 3 10 monocytes/ml) for 24 h in complete medium. The con- lymphocytes. A, PBMCs were stained with a FITC-labeled mouse anti- ditioned medium was harvested and assayed for the presence of human CD14 mAb and a mouse anti-human GMR␣ mAb (8G6) or an ␣ Ј sGMR by sCD116 ELISA (Diaclone Research). PBMCs from 16 isotype-matched IgG1, followed by a PE-labeled F(ab )2 goat anti-mouse different donors secreted an average of 165 Ϯ 26 pg/ml of sGMR␣ Ab. The cells were analyzed for the cell surface expression of CD14 and during the 24-h culture period (Fig. 1B). Monocytes were then GMR␣ by flow cytometry. B, PBMCs (106/ml, n ϭ 16 donors), monocytes isolated from PBMCs (ϳ3 ϫ 105/ml monocytes) and cultured for (3 ϫ 105/ml, n ϭ 3 donors), or lymphocytes (107/ml, n ϭ 5 donors) were 24 h. The monocyte-enriched population secreted 136 Ϯ 58 pg/ml cultured in duplicate for 24 h in complete medium. The supernatants were ␣ of sGMR␣ during the 24-h culture period, which is consistent with harvested and screened for the presence of sGMR by sCD116 ELISA. Values shown are the mean Ϯ SEM (ns, not statistically different). PBMCs the concentration of sGMR␣ that was secreted by the unfraction- (106/ml) (C) and monocytes (3 ϫ 105/ml) (D) were cultured in duplicate ated PBMCs, suggesting that it was the monocytes and not the for 0Ð96 h and the cell supernatants were screened every 24 h for the ␣ ,p Ͻ 0.05 ء .lymphocytes that were secreting sGMR .Toconfirm this, PBMC presence of sGMR␣ by ELISA. Shown are the means Ϯ SEM preparations were depleted of monocytes by adherence to plastic as compared with the previous time point; representative of n ϭ 3 indi- 7 and were cultured for 24 h. Lymphocytes cultured at 10 /ml, a vidual experiments. log-fold higher concentration than was used with the unfraction- ated PBMCs, showed no detectable secretion of sGMR␣, demonstrating that it is monocytes and not lymphocytes that secrete sGMR␣. To test this hypothesis, we cultured PBMCs at 106/ml for sGMR␣. 96 h in the absence of stimuli, and the cell supernatants were Secretion of sGMR␣ by monocytes in the absence of stimula- screened for the presence of sGMR␣ by sCD116 ELISA every tion suggested that monocytes could constitutively secrete 24 h. PBMCs secreted increasing concentrations of sGMR␣ for up 5682 ECTODOMAIN SHEDDING OF GMR␣ IN MONOCYTES to 96 h in culture with significant differences occurring between 0 and 24 h and between 24 and 48 h, suggesting that monocytes could constitutively secrete sGMR␣ (Fig. 1C). Monocytes did not require the presence of lymphocytes to secrete sGMR␣, because purified monocytes also constitutively secreted sGMR␣ for up to 96 h (Fig. 1D). These observations demonstrate that monocytes constitutively secrete sGMR␣ in the absence of external stimuli. Induction of sGMR␣ secretion by GM-CSF, LPS, PMA, and A23187 GM-CSF, LPS, PMA, and A23187 alter the ability of monocytes and neutrophils to bind GM-CSF on their cell surface, suggesting a role for them in regulating the expression of cell surface GMR␣ (19Ð23). Because sGMR␣ arises via alternative splicing of the GMR␣ transcript, it was possible that these stimuli might also regulate the secretion of sGMR␣. To test this hypothesis, we incubated PBMCs from individual donors with increasing concentrations of GM-CSF, LPS, PMA, or A23187 and analyzed the conditioned medium for the presence of sGMR␣. GM-CSF increased Downloaded from the secretion of sGMR␣ in a dose-dependent manner, with half- maximal activity occurring at 262 pg/ml (Fig. 2A). LPS also up- regulated sGMR␣ secretion in a dose-dependent manner with 1.5 ng/ml inducing a half-maximal response (Fig. 2B). PMA and A23187 also induced a dose-dependent increase in sGMR␣ secre-

tion with half-maximal induction occurring with 8 ng/ml PMA and http://www.jimmunol.org/ 1 ng/ml A23187 (Fig. 2, C and D, respectively). DMSO, used at a maximal dilution of 1/1000 (0.1%) in the dose response experiments, had no effect on sGMR␣ secretion by monocytes (data not shown). These results demonstrate that GM-CSF and other mediators increase the secretion of sGMR␣ by monocytes in a dose- dependent manner. To further investigate the kinetics of the up-regulation of sGMR␣ secretion, we incubated PBMCs with 100 ng/ml PMA for 10, 20, or 30 min and analyzed the harvested cell-conditioned me- by guest on September 25, 2021 dium for the presence of sGMR␣ using the Diaclone sCD116 ELISA. PMA up-regulated sGMR␣ secretion by PBMCs within 10 min (Fig. 2E), demonstrating that induction of sGMR␣ secretion occurs rapidly after cell stimulation. FIGURE 2. Induction of sGMR␣ secretion by GM-CSF, LPS, PMA, Assay development for the specific detection of alternatively and A23187. PBMCs were incubated for 1 h with increasing concentrations spliced sGMR␣ (log-fold dilutions) of GM-CSF (A), LPS (B), PMA (C), or A23187 (D). sGMR␣ was believed to arise exclusively through alternative The conditioned medium was harvested and screened for the presence of ␣ splicing of the GMR␣ gene product (4Ð7). However, the rapid sGMR by ELISA. The results are expressed as the level of induction of ␣ ␣ sGMR secretion (pg/ml) above the unstimulated control. Shown are the ء secretion of sGMR from monocytes in response to stimulation Ϯ ϭ ␣ means SEM for one representative of n 2 individual experiments ( , with PMA (Fig. 2E) suggested that sGMR may arise in part p Ͻ 0.05, as compared with unstimulated control). E, PBMCs were incu- through a mechanism other than de novo translation of the alter- bated for 10, 20, or 30 min with or without 100 ng/ml PMA, and the natively spliced mRNA. Because available Abs were not specific conditioned medium was screened for the presence of sGMR␣ using the for the alternatively spliced sGMR␣ protein but instead recognized sCD116 ELISA. Shown are the mean Ϯ SEM of data obtained from five p Ͻ 0.05, meaning ,ء) the shared ectodomain of GMR␣, we endeavored to produce an Ab independent experiments using five different donors that was specific for the alternatively spliced sGMR␣ isoform. To that there is a significant difference between the PMA-stimulated and un- this end, we immunized rabbits with a 16-aa peptide that corre- stimulated control at this time point). sponded to the unique C-terminal tail of the alternatively spliced sGMR␣ protein. The rabbit antiserum was purified by affinity chromatography using the 16-aa peptide as a ligand. The specific- with the splicing-specific antiserum (Fig. 3A). Furthermore, pre- ity of the purified antiserum (splicing-specific antiserum for alter- incubating the splicing-specific antiserum with an excess of the natively spliced sGMR␣ was then tested by SDS-PAGE and im- 16-aa peptide Ag (splicing specific ϩ peptide) completely inhib- munoblotting. Both recombinant alternatively spliced sGMR␣ ited the ability of the splicing-specific antiserum to recognize al- protein, which contains the 16-aa C-terminal domain, and recom- ternatively spliced sGMR␣. Taken together, these results demon- binant sGMR␣ lacking the 16-aa C-terminal tail (non-alternatively strate that the rabbit splicing-specific antiserum specifically spliced) were recognized by a mAb (8D10) that was raised against recognizes only the alternatively spliced sGMR␣ isoform, which the shared ectodomain of GMR␣ (Fig. 3A). Neither the alterna- contains the unique 16-aa C-terminal domain. tively spliced nor the non-alternatively spliced recombinant We were also interested in developing an ELISA that would sGMR␣ proteins were recognized using rabbit preimmune serum, allow us to detect alternatively spliced sGMR␣ in conditioned me- and only the alternatively spliced sGMR␣ protein was recognized dium. The detection Ab (SCO4) supplied in the commercially The Journal of Immunology 5683

strates the specificity of the splicing-specific ELISA for alternatively spliced sGMR␣. The splicing-specific ELISA also failed to detect a 1000-fold higher amount of the non-alternatively spliced sGMR␣ variant protein (data not shown). Together, these results demonstrate the successful modification of the sCD116 ELISA using the splicing-specific antiserum to detect only the alternatively spliced sGMR␣ protein. However, although the splicing- specific and sCD116 ELISAs are equally effective at detecting recombinant purified alternatively spliced sGMR␣, the splicing- specific ELISA presently has a background signal when evaluating alternatively spliced sGMR␣ in PBMC-conditioned medium. Therefore, although our new splicing-specific ELISA is an excellent tool for determining the relative levels of alternatively spliced soluble receptor secreted by monocytes, it does not at present allow us to make any conclusions about the absolute amounts of alternatively spliced sGMR␣ secreted by monocytes.

␣

sGMR is released by stimulated monocytes through both Downloaded from alternative splicing and proteolytic cleavage Because the presence of alternatively spliced sGMR␣ mRNA in monocytes had previously been demonstrated (6), we anticipated that the sGMR␣ protein being produced by monocytes would rep- resent the alternatively spliced isoform. However, the rapid up-

regulation of sGMR␣ secretion by PBMCs in response to stimu- http://www.jimmunol.org/ lation led us to hypothesize that monocytes might also secrete a sGMR␣ variant generated through ectodomain shedding. With the development of the splicing-specific ELISA, we were now able to test this hypothesis. Because most known sheddases are metalloproteases (24), we decided to treat unstimulated and stimulated monocytes with the broad-spectrum metalloprotease inhibitor BB94 (batimastat) and to look at the effect of this inhibitor on the secretion of total and alternatively spliced sGMR␣ by monocytes. by guest on September 25, 2021 FIGURE 3. Production of rabbit antiserum against the 16-aa C-terminal To this end, we pretreated isolated PBMCs (from n ϭ 7 individual domain of alternatively spliced sGMR␣. A, Recombinant sGMR␣ protein donors) for2hat37¡C with 100 ␮M of either the metalloprotease containing the 16-aa C-terminal domain (A, alternatively spliced) or a re- inhibitor BB94 or the appropriate vehicle control (1% DMSO), combinant version of sGMR␣ that is missing the 16-aa epitope (NA, non- pelleted the cells, and resuspended them in fresh medium contain- alternatively spliced) was fractionated by SDS-PAGE and transferred to ing either 1% DMSO or 100 ␮M BB94. Cell viability was assessed polyvinylidene difluoride membranes for immunoblotting. The blots were by trypan blue dye exclusion, and no significant difference be- ␣ probed with a mAb raised against the shared domain of cell surface GMR tween the viability of the DMSO- or BB94-treated cells was noted (8D10), rabbit preimmune serum, affinity-purified alternatively spliced (data not shown). The PBMCs were then either left unstimulated or sGMR␣ rabbit antiserum (splicing-specific antiserum), or rabbit antiserum that was preincubated with an excess of the 16-aa peptide used as Ag stimulated for an additional hour using concentrations of PMA, (splicing specific ϩ peptide). B, 1 ng/ml purified recombinant alternatively A23187, LPS, or GM-CSF, which were predicted to induce max- spliced (A) or non-alternatively spliced (NA) sGMR␣ protein was loaded imal sGMR␣ secretion based on the dose response curves in Fig. onto a SCO6-coated ELISA plate. Total sGMR␣ protein was detected us- 2, AÐD. This conditioned medium was then collected and screened ing the biotinylated mAb SCO4, whereas alternatively spliced sGMR␣ in parallel on three different ELISAs: the Diaclone sCD116 ELISA protein was detected using biotinylated splicing-specific antiserum. was used to quantitate total sGMR␣ levels (Fig. 4A), the splicing- specific ELISA was used to look at levels of only the alternatively spliced sGMR␣ (Fig. 4B), and the samples were also screened on available sGMR␣ ELISA (sCD116 ELISA; Diaclone Research) an IL-8 ELISA. Importantly, treatment of PBMCs with 100 ␮M recognizes the extracellular domain of GMR␣, and consequently, BB94 led to a significant decrease in the amount of total sGMR␣ this ELISA is predicted to detect total sGMR␣. We modified the secreted constitutively ( p ϭ 0.02) and in response to PMA ( p ϭ sCD116 ELISA to detect exclusively the alternatively spliced 0.004), A23187 ( p ϭ 0.001), and LPS ( p Ͻ 0.001), but it did not sGMR␣ protein by using the splicing-specific antiserum as the have a significant effect on the amounts of secreted alternatively detection Ab (splicing-specific ELISA). The splicing-specific spliced sGMR␣ (Fig. 4B) or IL-8 (data not shown). The lack of ELISA was comparable to the Diaclone sCD116 ELISA in terms effect on either alternatively spliced sGMR␣ or IL-8 suggests that of its sensitivity and range. This was demonstrated by loading BB94 does not have a nonspecific effect on cell viability or protein 1000 pg/ml purified recombinant alternatively spliced sGMR␣ or secretion by PBMCs. This implies that a metalloprotease-mediated non-alternatively spliced sGMR␣ variant (which is missing the cleavage of GMR␣ from the surface of human monocytes can lead 16-aa C-terminal tail) onto the sCD116 ELISA and probing using to the generation of a shed sGMR␣ variant. either biotinylated mAb SCO4 or splicing-specific antiserum. Both Because we were concerned about the presence of 1% DMSO in ELISAs detected similar amounts of alternatively spliced protein, all of these samples, in a separate experiment we looked at the but importantly only the Diaclone sCD116 detected the non-alter- effect of 1% DMSO on secretion of sGMR␣ by PBMCs. The pres- natively spliced sGMR␣ variant protein (Fig. 3B), which demon- ence of 1% DMSO led to a statistically significant decrease in the 5684 ECTODOMAIN SHEDDING OF GMR␣ IN MONOCYTES

whereas treatment with GM-CSF did not. Importantly, pretreat- ment of PBMCs with BB94 appears to have abrogated the ability of these cells to up-regulate secretion of total sGMR␣ in response to LPS. Therefore, there was no signiﬁcant difference between the amount of total sGMR␣ secreted by the BB94-treated unstimulated control and LPS-treated cells (Fig. 4A). In contrast, production of total sGMR␣ by the PMA- and A23187-stimulated BB94- treated PBMCs remained signiﬁcantly up-regulated vs the unstimulated control ( p ϭ 0.01 and p ϭ 0.02, respectively). In contrast, when looking at the production of alternatively spliced sGMR␣ by monocytes (Fig. 4B), we note that alternatively spliced soluble receptor levels were also induced in response to both PMA ( p ϭ 0.03) and A23187 ( p ϭ 0.003), but not in response to either LPS or GM-CSF. Ectodomain shedding of GMR␣ The observed inhibition of sGMR␣ secretion by PBMCs in response to treatment with a metalloprotease inhibitor strongly sug-

gested that a sGMR␣ variant could arise through ectodomain shed- Downloaded from ding. To conﬁrm that shedding of cell surface GMR␣ could occur, we engineered a murine pro-B cell line (Ba/F3) to express the cDNA for human transmembrane GMR␣ (Ba/F3.GMR␣). Be- cause this is a murine cell line transfected with human cDNA, the Ba/F3.GMR␣ cell line was incapable of secreting human alterna-

tively spliced sGMR␣. http://www.jimmunol.org/ Expression of GMR␣ on the cell surface of the Ba/F3.GMR␣ cell line, but not on the surface of the parental Ba/F3 cell line, was confirmed by flow cytometry using the anti-GMR␣ mAb 8G6 or an IgG1 isotype control (Fig. 5A). The Ba/F3 and the Ba/ F3.GMR␣ cell lines were then cultured at a density of 107/ml in complete medium for 2 h. The cell-conditioned medium was harvested and screened for the presence of total sGMR␣ protein by sCD116 ELISA. There was no detectable sGMR␣ in the Ba/F3 conditioned medium, but the Ba/F3.GMR␣ cell line secreted by guest on September 25, 2021 157 Ϯ 13 pg/ml sGMR␣ (Fig. 5B), demonstrating that the ectodo- FIGURE 4. BB94 inhibits secretion of total sGMR␣ but not alternatively main of GMR␣ could be shed from the cell surface. We were also ␣ spliced sGMR by PBMCs. PBMCs from healthy donors were resuspended in interested in determining whether the ectodomain shedding of ϫ 7 complete medium at a concentration of 1Ð2 10 cells/ml and were incubated sGMR␣ from the Ba/F3.GMR␣ cell line could also be induced for2hat37¡C in the presence of either 1% (v/v) DMSO (vehicle control) or using the stimuli shown previously to up-regulate sGMR␣ secre- 100 ␮M BB94. Cells were then pelleted by centrifugation and resuspended at a concentration of 107/ml in fresh medium containing either DMSO or BB94. tion by monocytes. The Ba/F3 cell line is not LPS-responsive (25) ␤ PBMCs were then incubated for an additional hour in the absence of stimulus and lacks the c subunit of the human GM-CSF receptor. There- (control) or in the presence of 100 ng/ml PMA, 1 ␮g/ml A23187, 1 ␮g/ml fore, we looked only at the effect of PMA and A23187 on the LPS, or 100 ng/ml GM-CSF. Cells were pelleted and conditioned medium was shedding of sGMR␣ from the Ba/F3.GMR␣ cell line. Neither collected and then screened by ELISA to quantitate either the amount of total stimulus appeared to have an effect on the shedding of sGMR␣ sGMR␣ using the Diaclone sCD116 ELISA (A) or the amount of alternatively from the surface of these cells (data not shown). spliced sGMR␣ using the splicing-specific ELISA described in this paper (B). To verify that the sGMR␣ produced by the Ba/F3.GMR␣ cell ϭ Pooled results from n 7 individual donors are presented as percent change line was in fact being derived by enzymatic cleavage rather than by Ͻ ء compared with the unstimulated 1% DMSO vehicle control. , p 0.05, nonspecific shearing from the cell surface, we pretreated Ba/ comparing DMSO control- and BB94-treated conditions; ns, not statistically F3.GMR␣ cells with BB94 (100 ␮m) or with a 1% DMSO vehicle different. control for 2 h, then pelleted the cells, resuspended them in fresh medium, and allowed them to condition medium for an additional hour. The cell-conditioned medium was then screened using the amount of total sGMR␣ secreted by PBMCs ( p Ͻ 0.05) (data not sGMR␣ sCD116 ELISA. Treatment of cells with BB94 com- shown). pletely inhibited the production of sGMR␣ by these cells (Fig. 5C), further supporting the hypothesis that a sGMR␣ variant can Up-regulated secretion of both alternatively spliced and total arise through metalloprotease-mediated ectodomain shedding. ␣ sGMR in response to stimulation There was no significant difference in the viability of the DMSO In Fig. 4, one can also compare the effect of the various stimuli vehicle control or BB94-treated Ba/F3.GMR␣ cells after 3 h, as (PMA, A23187, LPS, and GM-CSF) on the regulated secretion of measured by trypan blue dye exclusion (data not shown). either the alternatively spliced or shed sGMR␣ variants. When To determine the molecular mass of the constitutively shed looking at the production of total sGMR␣ in Fig. 4A, we note that sGMR␣ protein, we attempted to purify the sGMR␣-like protein stimulation of DMSO control PBMCs led to a significant up-reg- out of medium conditioned by the Ba/F3.GMR␣ cell line. To this ulation of total sGMR␣ secretion (vs unstimulated cells) in the end, Ba/F3 or Ba/F3.GMR␣ conditioned medium was passed over case of PMA ( p ϭ 0.02), A23187 ( p ϭ 0.03), and LPS ( p ϭ 0.05), a GM-CSF-Sepharose 4B ligand affinity column and the adsorbed The Journal of Immunology 5685

proteins were analyzed by SDS-PAGE and Western blotting. No sGMR␣-like proteins were purified from the Ba/F3 cell-conditioned medium; however, a distinct protein band was present in the Ba/F3.GMR␣ column eluent (Fig. 5D). This 60-kDa protein was the same size as recombinant alternatively spliced sGMR␣ and was specifically recognized by the anti-GMR␣ mAb (8D10) (Fig. 5D), but not by the anti-alternatively spliced sGMR␣ rabbit splicing-specific antiserum (Fig. 5E), demonstrating that the sGMR␣ protein released from the Ba/F3.GMR␣ cells arose via shedding of cell surface GMR␣. Importantly, the fact that we were able to enrich this shed sGMR␣ variant using a GM-CSF ligand affinity column suggests that the shed sGMR␣ protein retained the ability to specifically bind to GM-CSF.

Discussion The alternatively spliced sGMR␣ mRNA has been identiﬁed in many cell types that express GMR␣, including human placental tissue (4, 7), myeloid leukemic cell lines (5, 11), monocytes, macrophages, bone marrow progenitors, and synovial ﬁbroblasts (6). Downloaded from However, there is little information available about which primary human cells secrete the protein. In this paper, we demonstrate that monocytes constitutively secrete sGMR␣ but lymphocytes do not (Fig. 1). More importantly, we also demonstrate that secretion of sGMR␣ by monocytes can be rapidly up-regulated by GM-CSF,

LPS, PMA, and A23187 (Fig. 2). We further describe the produc- http://www.jimmunol.org/ tion of a rabbit antiserum that specifically recognizes the 16-aa tail of the predicted alternatively spliced sGMR␣ protein (Fig. 3), and using this new “splicing-specific” antiserum in an ELISA, we demonstrate for the first time that monocytes do indeed secrete the predicted alternatively spliced sGMR␣ protein (Fig. 4B). We also demonstrate that the secretion of total sGMR␣ (Fig. 4A), but not of alternatively spliced sGMR␣ (Fig. 4B) or IL-8 (data not shown), can be inhibited using the broad-spectrum metalloprotease inhibitor BB94. Together, these results show for the first time that by guest on September 25, 2021 monocytes secrete sGMR␣ protein that represents a mixed population of alternatively spliced and proteolytically cleaved species. Monocytes but not lymphocytes express GMR␣ on their cell surface (Fig. 1A), suggesting that monocytes might also secrete sGMR␣. PBMCs and an equivalent number of purified monocytes secreted similar concentrations of sGMR␣ in vitro, whereas lymphocytes secreted none (Fig. 1B). Secretion of sGMR␣ by human monocytes is consistent with previous findings that demonstrated expression of the alternatively spliced sGMR␣ mRNA in human monocytes (6) and secretion of a sGMR␣ protein by myeloid leukemic cell lines (13). The inability of lymphocytes to secrete sGMR␣ is consistent with the lack of sGMR␣ mRNA in lymphocytes (11) and the absence of GMR␣ expression on the cell surface of normal human lymphocytes (Fig. 1A). Together, these results show that normal human monocytes but not lymphocytes can secrete a sGMR␣ protein. In the absence of stimulation, human monocytes secreted FIGURE 5. Ectodomain shedding of GMR␣ from a GMR␣-expressing sGMR␣ for up to 4 days in culture (Fig. 1, C and D), suggesting recombinant cell line. A, Ba/F3 and Ba/F3.GMR␣ cells were stained with a mouse anti-human GMR␣ mAb (8D10) or an isotype-matched IgG1, Ј followed by PE-labeled F(ab )2 goat anti-mouse Ab. The cells were ana- conditioned medium was then screened for the presence of the shed lyzed for the cell surface expression of GMR␣ by flow cytometry. B, The sGMR␣ variant using the Diaclone sCD116 ELISA. Results shown are the ␣p Ͻ 0.05). D and E, Ba/F3 (240 ml) or Ba/F3.GMR ,ء) Ba/F3 and the Ba/F3.GMR␣ cell lines were cultured in duplicate at 107/ml mean Ϯ SEM in complete medium containing 1 ng/ml murine IL-3 for 2 h. The condi- (120 ml) conditioned medium was passed over a GM-CSF-Sepharose 4B tioned medium was harvested and screened for the presence of sGMR␣ by affinity column. The column was washed extensively with PBS and the adsorbed proteins were eluted with a 0.1 M glycine buffer (pH 2.5). The ,ء) ELISA. Shown is the mean Ϯ SEM of three independent experiments p Ͻ 0.05). C, Ba/F3.GMR␣ cells were incubated at 37¡Cfor2hata eluted fractions were pH neutralized, volume reduced, and fractionated by concentration of 107/ml in complete medium supplemented with murine SDS-PAGE, along with recombinant alternatively spliced sGMR␣. The IL-3 in the presence of 1% DMSO (vehicle control) or 100 ␮m BB94, and blot was probed using a mouse anti-human GMR␣ mAb (8D10) (D) and then the cells were pelleted and resuspended in fresh medium containing then was stripped and reprobed with the alternatively spliced sGMR␣ DMSO or BB94 and incubated in triplicate for an additional hour. The splicing-specific antiserum (E). 5686 ECTODOMAIN SHEDDING OF GMR␣ IN MONOCYTES that monocytes constitutively secrete sGMR␣. Other soluble cy- was an additional sGMR␣ variant being produced by monocytes. tokine receptors, such as sTNFR-p55/75 (26) and sIL-6R, are also Because metalloprotease-mediated cleavage of cell surface recep- constitutively secreted in vitro by normal human monocytes and tors is a common mechanism through which other soluble recep- monocyte-like cell lines (27, 28). Furthermore, sIL-1RII (29, 30), tors arise (24, 37, 38), we wondered whether our non-alternatively along with sTNFR-p55 (31), sIL-6R (32), and sGMR␣ (13), are spliced sGMR␣ protein may also have arisen via metalloprotease- also constitutively present in normal human plasma, suggesting mediated ectodomain shedding of cell surface GMR␣. Using a that sGMR␣ may also be constitutively secreted in vivo. The rea- broad-spectrum metallprotease inhibitor (BB94), we demonstrated son for the constitutive secretion of sGMR␣ is unclear. It is pos- that the amount of total sGMR␣ protein released by monocytes sible that sGMR␣ is constitutively secreted to modulate the activ- was reduced in the presence of BB94 (vs a 1% DMSO vehicle ity of GM-CSF during homeostasis. However, GM-CSF, like most control) (Fig. 4A). Importantly, release of the alternatively spliced other cytokines, is secreted in a tightly regulated manner and does sGMR␣ protein (Fig. 4B) or of IL-8 (data not shown) from PBMC not normally circulate in human plasma in the absence of inflam- cultures was not inhibited by BB94, demonstrating that the ob- mation or disease. Elucidation of the role of sGMR␣ during ho- served inhibition of the non-alternatively spliced sGMR␣ protein meostasis will have to await further investigation. production was not due to a nonspecific down-regulation of protein Stimulation of monocytes with GM-CSF, LPS, PMA, and secretion. There was no difference in the viability of the 1% A23187 up-regulated the secretion of sGMR␣ in a dose-dependent DMSO vehicle control and BB94-treated PBMCs after 2 h. How- manner (Fig. 2, AÐD), demonstrating that sGMR␣ secretion could ever, the presence of 1% DMSO in PBMC cultures did signifi- be regulated by its cognate ligand, proinflammatory stimuli (LPS), cantly reduce the amount of sGMR␣ secreted by PBMCs as com- and heterogeneous chemical stimuli (PMA and A23187). DMSO, pared with PBMCs in the absence of DMSO. Downloaded from used at a maximum dilution of 1/1000 (0.1%) in the dose response We had expected that the sGMR␣ produced constitutively by experiments, had no effect on sGMR␣ secretion (data not shown). monocytes would be the product of alternative splicing and that Together, these results demonstrate that stimulation of monocytes shedding would account for the additional sGMR␣ protein pro- can increase the secretion of sGMR␣. duced in response to stimulation, as has been previously demon- The regulation of secretion of a soluble cytokine receptor by its strated for sIL-6R in THP-1 cells (28). Using our new splicing-

cognate ligand has been demonstrated in vitro and in vivo for specific ELISA, we have now demonstrated that alternatively http://www.jimmunol.org/ TNF-␣ (33Ð35), where stimulation of cells with TNF-␣ or i.v. spliced sGMR␣ can be secreted constitutively by monocytes (Fig. injection of recombinant TNF-␣ into mice leads to the rapid up- 4B). However, both PMA and A23187 treatment led to a statisti- regulation of sTNFR-p55/p75 secretion. Although the signal-trans- cally significant up-regulation of the alternatively spliced sGMR␣ duction pathways responsible for ligand-induced secretion of sol- variant (Fig. 4B), which suggests that additional alternatively uble receptors are unclear, it is likely that secretion is initiated after spliced sGMR␣ can be inducibly secreted by monocytes upon ex- ligand-induced phosphorylation of the cognate cell surface recep- posure to chemical stimuli. We conclude that alternatively spliced tor complex. This suggests that soluble cytokine receptors such as sGMR␣ can be produced by monocytes both constitutively and in sGMR␣ and sTNFR-p55/p75 may play a critical role in modulat- response to stimulation. ing the activity of their respective cytokines during inflammation Similarly, shedding of GMR␣ from the surface of monocytes by guest on September 25, 2021 and hematopoiesis. It is interesting to note that recombinant also appears to occur constitutively, because BB94 treatment of sGMR␣ anatagonizes GM-CSF activity in vitro, preventing GM- unstimulated cells led to a decrease in secretion of total sGMR␣. CSF-induced cell proliferation (5, 10), bone marrow colony for- At present, because of a background signal observed when ana- mation (7), and neutrophil functional activity (J.M.P. and C.B.B., lyzing PBMC-conditioned medium on the splicing-specific unpublished data), suggesting that sGMR␣ secretion in response to ELISA, the sCD116 and splicing-specific ELISAs can be used to GM-CSF stimulation may indeed occur to moderate the activity of measure relative levels of total and alternatively spliced sGMR␣, GM-CSF during inflammation and/or hematopoiesis. but cannot yet be used to reliably quantitate the amount of shed Proinflammatory mediators such as LPS, along with chemical sGMR␣ being produced by monocytes. It is therefore somewhat stimuli such as PMA and A23187 that work through divergent more difficult to make conclusions about whether the shedding of signaling pathways that do not necessarily result in phosphoryla- GMR␣ from the surface of monocytes is also a regulated process. tion of ␤c, also up-regulated monocyte secretion of sGMR␣ (Fig. However, we would argue that because LPS up-regulates the se- 2), demonstrating a direct role for inflammatory mediators in reg- cretion of total sGMR␣ in the absence of BB94 ( p ϭ 0.05) but not ulating sGMR␣ secretion. LPS, PMA, and A23187 have also been when BB94 is present, this implies that LPS might be specifically shown to up-regulate the secretion of other soluble cytokine re- up-regulating the shedding of sGMR␣. This conclusion is further ceptors such as sIL-1RII (29, 36), sIL-6R (27, 28), and sTNFR- supported by the observation that LPS did not significantly up- p55/p75 (27), highlighting a common role for inflammatory me- regulate the amount of alternatively spliced sGMR␣ produced by diators in regulating the secretion of not only proinflammatory monocytes (Fig. 4B). Therefore, it appears that LPS may be acting cytokines, but also their respective soluble cytokine receptors. to induce total sGMR␣ protein production by specifically up-reg- The rapid release of sGMR␣ from monocytes in response to ulating the proteolytic activity of the relevant sheddase. As men- stimulation with PMA (Fig. 2E) made us wonder whether sGMR␣ tioned previously, PMA and A23187 both led to the up-regulated may in fact arise through proteolytic cleavage of cell surface secretion of both total and alternatively spliced sGMR␣ by mono- GMR␣ rather than, or in addition to, alternative splicing of the cytes. On the basis of the data presented in Fig. 4A, we would GMR␣ gene product. To this end, we generated rabbit antiserum suggest that the degree of the up-regulated secretion in response to that specifically recognized the 16-aa C terminus of alternatively PMA and A23187 seems to be reduced by BB94 treatment. Thus, spliced sGMR␣, allowing us to differentiate between sGMR␣ that PMA and A23187 may up-regulate the shedding of GMR␣ from arose via alternative splicing and sGMR␣ that arose via another the surface of monocytes in addition to inducing the secretion of mechanism (Fig. 3A). Using this reagent, we developed a splicing- alternatively spliced sGMR␣. However, we are cautious about specific ELISA that allowed us to demonstrate conclusively for the making conclusions as to the effect of PMA and A23187 on shed- first time that monocytes could secrete alternatively spliced ding of sGMR␣ until we are able to specifically quantitate the sGMR␣ protein (Fig. 4B). However, it also became clear that there amount of shed sGMR␣ being produced by monocytes or are able The Journal of Immunology 5687 to directly measure the effect of PMA and A23187 on the activity University of Calgary Hybridoma Facility, and Diane Teoh for excellent of the relevant sheddase. technical assistance. We would also like to thank Dr. Angel Lopez for The ability of GMR␣ to be shed from the cell surface was con- mAbs 8G6 and 8D10, Dr. Ken Kaushansky for the Ba/F3 cell line, and Dr. firmed using a murine pro-B cell line that we engineered to express Stephen Robbins for thoughtful discussions during this project and in the cell surface GMR␣ but not alternatively spliced sGMR␣ (Ba/ preparation of this manuscript. F3.GMR␣; Fig. 5A). A sGMR␣-like protein was detected by ELISA in medium conditioned by the Ba/F3.GMR␣ cell line but References not from the parental Ba/F3 cells (Fig. 5B) or pBabe vector control 1. Gasson, J. C. 1991. Molecular physiology of granulocyte-macrophage colony- cells (data not shown). Importantly, treatment of Ba/F3.GMR␣ stimulating factor. Blood 77:1131. 2. Gearing, D. P., J. A. King, N. M. Gough, and N. A. Nicola. 1989. Expression cells with BB94, but not with the DMSO vehicle control, led to a cloning of a receptor for human granulocyte-macrophage colony-stimulating fac- complete inhibition of sGMR␣ by these cells (Fig. 5C), confirming tor. EMBO J. 8:3667. that the sGMR␣ released by these cells is generated by proteolytic 3. Hayashida, K., T. Kitamura, D. M. Gorman, K. Arai, T. Yokota, and A. Miyajima. 1990. Molecular cloning of a second subunit of the receptor for cleavage rather than by vesicle budding or nonspecific shearing of human granulocyte-macrophage colony-stimulating factor (GM-CSF): reconsti- the receptor from the cell surface. tution of a high-affinity GM-CSF receptor. Proc. Natl. Acad. Sci. USA 87:9655. The sGMR␣ variant was then purified from Ba/F3.GMR␣-con- 4. Ashworth, A., and A. Kraft. 1990. Cloning of a potentially soluble receptor for human GM-CSF. Nucleic Acids Res. 18:7178. ditioned medium using a GM-CSF-Sepharose affinity column, 5. Raines, M. A., L. Liu, S. G. Quan, V. Joe, J. F. DiPersio, and D. W. Golde. 1991. which indicates that this shed sGMR␣ protein retains the ability to Identification and molecular cloning of a soluble human granulocyte-macrophage bind specifically to GM-CSF. This was not surprising because we colony-stimulating factor receptor. Proc. Natl. Acad. Sci. USA 88:8203. 6. Williams, W. V., J. M. VonFeldt, H. Rosenbaum, K. E. Ugen, and D. B. Weiner. ␣ Downloaded from have previously demonstrated that purified recombinant GMR 1994. Molecular cloning of a soluble form of the granulocyte-macrophage col- extracellular domain binds to GM-CSF with an affinity comparable ony-stimulating factor receptor ␣ chain from a myelomonocytic cell line: expres- to the interaction between GM-CSF and alternatively spliced sion, biologic activity, and preliminary analysis of transcript distribution. Arthri- ␣ ␣ tis Rheum. 37:1468. sGMR (9, 14). Furthermore, the 60-kDa shed Ba/F3.GMR vari- 7. Brown, C. B., P. Beaudry, T. D. Laing, S. Shoemaker, and K. Kaushansky. 1995. ant was recognized specifically by a mAb raised against the In vitro characterization of the human recombinant soluble granulocyte-macroph- ectodomain of GMR␣ (Fig. 5D), but not by our splicing-specific age colony-stimulating factor receptor. Blood 85:1488. 8. Murray, E. W., C. Pihl, A. Morcos, and C. B. Brown. 1996. Ligand-independent antiserum (Fig. 5E). The similar electropheretic mobility of the cell surface expression of the human soluble granulocyte-macrophage colony- http://www.jimmunol.org/ shed and alternatively spliced isoforms (Fig. 5D) is, at least in part, stimulating factor receptor ␣ subunit depends on co-expression of the membrane- ␤ accounted for by the heterogeneous glycosylation of the extracel- associated receptor subunit. J. Biol. Chem. 271:15330. 9. Prevost, J. M., P. J. Farrell, K. Iatrou, and C. B. Brown. 2000. Determinants of lular domain of GMR␣ (7) that would blur the differences in the the functional interaction between the soluble GM-CSF receptor and the GM- length of the primary sequence of the two isoforms. The similar CSF receptor ␤-subunit. Cytokine 12:187. mobility may also provide a hint as to the cleavage site of the shed 10. Monfardini, C., M. Ramamoorthy, H. Rosenbaum, Q. Fang, P. A. Godillot, G. Canziani, I. M. Chaiken, and W. V. Williams. 1998. Construction and binding GMR␣. The apparent 60-kDa molecular mass of the shed sGMR␣ kinetics of a soluble granulocyte-macrophage colony-stimulating factor receptor suggests that the cleavage site lies very close to the membrane. If ␣-chain-Fc fusion protein. J. Biol. Chem. 273:7657. we assume that the WSSWS motif conserved among cytokine re- 11. Heaney, M. L., J. C. Vera, M. A. Raines, and D. W. Golde. 1995. Membrane- associated and soluble granulocyte/macrophage-colony-stimulating factor recep- by guest on September 25, 2021 ceptors and spanning residues 305Ð310 in GMR␣ is essential for tor ␣ subunits are independently regulated in HL-60 cells. Proc. Natl. Acad. Sci. the presentation and integrity of GMR␣, then the cleavage may USA 92:2365. occur between residue 311 and the putative membrane entrance at 12. Sasaki, K., S. Chiba, H. Mano, Y. Yazaki, and H. Hirai. 1992. Identification of a soluble GM-CSF binding protein in the supernatant of a human choriocarci- residue 320. This would predict a relatively small difference be- noma cell line. Biochem. Biophys. Res. Commun. 183:252. tween the number of amino acids in either the shed or alternatively 13. Sayani, F., F. A. Montero-Julian, V. Ranchin, J. M. Prevost, S. Flavetta, W. Zhu, spliced receptor variants. Importantly, the fact that the shed and R. C. Woodman, H. Brailly, and C. B. Brown. 2000. Identification of the soluble granulocyte-macrophage colony stimulating factor receptor protein in vivo. alternatively spliced species are indistinguishable by SDS-PAGE Blood 95:461. likely explains why we did not detect the second sGMR␣ species 14. Murray, E. W., C. Pihl, S. M. Robbins, J. Prevost, A. Mokashi, S. M. Bloomfield, ␣ and C. B. Brown. 1998. The soluble granulocyte-macrophage colony-stimulating when we previously purified sGMR from normal human plasma factor receptor’s carboxyl-terminal domain mediates retention of the soluble re- (13). Determination of the exact cleavage site will await further ceptor on the cell surface through interaction with the granulocyte-macrophage study. colony-stimulating factor receptor ␤-subunit. Biochemistry 37:14113. 15. Brown, C. B., C. E. Pihl, and E. W. Murray. 1997. Oligomerization of the soluble In conclusion, we have demonstrated that a soluble GM-CSF re- granulocyte-macrophage colony-stimulating factor receptor: identification of the ceptor is secreted constitutively by normal human monocytes but not functional ligand-binding species. Cytokine 9:219. by lymphocytes, and that GM-CSF and other inflammatory mediators 16. Park, L. S., D. Friend, S. Gillis, and D. L. Urdal. 1986. Characterization of the cell surface receptor for human granulocyte/macrophage colony-stimulating fac- can rapidly up-regulate its secretion. We have developed an Ab that tor. J. Exp. Med. 164:251. specifically recognizes the predicted alternatively spliced sGMR␣ 17. Wognum, A. W., Y. Westerman, T. P. Visser, and G. Wagemaker. 1994. Dis- tribution of receptors for granulocyte-macrophage colony-stimulating factor on protein and have used it in an ELISA to demonstrate that monocytes ϩ ␣ immature CD34 bone marrow cells, differentiating monomyeloid progenitors do in fact secrete alternatively spliced sGMR . Furthermore, we have and mature blood cell subsets. Blood 84:764. demonstrated for the first time that the ectodomain of cell surface 18. Yamada, T., Q. Sun, K. Zeibecoglou, J. Bungre, J. North, A. B. Kay, A. F. Lopez, GMR␣ can be shed in vitro and that shedding is mediated in part and D. S. Robinson. 1998. IL-3, IL-5, granulocyte-macrophage colony-stimulating factor receptor ␣-subunit, and common ␤-subunit expression by peripheral through the activity of metalloproteases. This dual mechanism of sol- leukocytes and blood dendritic cells. J. Allergy Clin. Immunol. 101:677. uble receptor production has also been seen for other cytokine recep- 19. Cannistra, S. A., P. Groshek, R. Garlick, J. Miller, and J. D. Griffin. 1990. Reg- tors such as sIL-6R␣, sIL-4R, and the growth hormone receptor (re- ulation of surface expression of the granulocyte/macrophage colony-stimulating factor receptor in normal human myeloid cells. Proc. Natl. Acad. Sci. USA 87:93. viewed in Ref. 39). It will be of interest to determine whether the 20. Cannistra, S. A., M. Koenigsmann, J. DiCarlo, P. Groshek, and J. D. Griffin. alternatively spliced and shed sGMR␣ proteins have similar or unique 1990. Differentiation-associated expression of two functionally distinct classes of biological functions in mediating GM-CSF-induced inflammation granulocyte-macrophage colony-stimulating factor receptors by human myeloid cells. J. Biol. Chem. 265:12656. and/or hematopoiesis. 21. Brizzi, M. F., C. Arduino, G. C. Avanzi, F. Bussolino, and L. Pegoraro. 1991. GM-CSF and phorbol esters modulate GM-CSF receptor expression by indepen- Acknowledgments dent mechanisms. J. Cell. Physiol. 148:24. 22. Khwaja, A., J. Carver, H. M. Jones, and D. C. Linch. 1993. Dynamic modulation We thank Laurie Robertson of the Alberta Cancer Board/University of of the cell surface expression of the granulocyte-macrophage colony-stimulating Calgary Flow Cytometry Facility, the staff of the Alberta Cancer Board/ factor receptor. Br. J. Haematol. 85:42. 5688 ECTODOMAIN SHEDDING OF GMR␣ IN MONOCYTES

23. Stacchini, A., L. Fubini, E. Gatti, L. Mutti, W. Piacibello, F. Sanavio, 31. Shapiro, L., B. D. Clark, S. F. Orencole, D. D. Poutsiaka, E. V. Granowitz, and G. Scagliotti, E. Pozzi, and M. Aglietta. 1995. In vivo effect of human granulo- C. A. Dinarello. 1993. Detection of tumor necrosis factor soluble receptor p55 in cyte-macrophage colony-stimulating factor (GM-CSF) on neutrophil GM-CSF blood samples from healthy and endotoxemic humans. J. Infect. Dis. 167:1344. receptors. Leukemia 9:665. 32. Muller-Newen, G., C. Kohne, R. Keul, U. Hemmann, W. Muller-Esterl, 24. Hooper, N. M., E. H. Karran, and A. J. Turner. 1997. Membrane protein secre- J. Wijdenes, J. P. Brakenhoff, M. H. Hart, and P. C. Heinrich. 1996. Purification tases. Biochem. J. 321:265. and characterization of the soluble interleukin-6 receptor from human plasma and 25. Shimazu, R., S. Akashi, H. Ogata, Y. Nagai, K. Fukudome, K. Miyake, and identification of an isoform generated through alternative splicing. Eur. J. Bio- M. Kimoto. 1999. MD-2, a molecule that confers lipopolysaccharide responsive- chem. 236:837. ness on Toll-like receptor 4. J. Exp. Med. 189:1777. 33. Pinckard, J. K., K. C. Sheehan, and R. D. Schreiber. 1997. Ligand-induced for- 26. Hino, T., H. Nakamura, S. Abe, H. Saito, M. Inage, K. Terashita, S. Kato, and mation of p55 and p75 tumor necrosis factor receptor heterocomplexes on intact H. Tomoike. 1999. Hydrogen peroxide enhances shedding of type I soluble tumor cells. J. Biol. Chem. 272:10784. 34. Orlando, S., C. Matteucci, E. J. Fadlon, W. A. Buurman, M. T. Bardella, necrosis factor receptor from pulmonary epithelial cells. Am. J. Respir. Cell Mol. ␣ Biol. 20:122. F. Colotta, M. Introna, and A. Mantovani. 1997. TNF- , unlike other pro- and anti-inflammatory cytokines, induces rapid release of the IL-1 type II decoy re- 27. Gallea-Robache, S., V. Morand, S. Millet, J. M. Bruneau, N. Bhatnagar, ceptor in human myelomonocytic cells. J. Immunol. 158:3861. S. Chouaib, and S. Roman-Roman. 1997. A metalloproteinase inhibitor blocks 35. Bjornberg, F., M. Lantz, I. Olsson, and U. Gullberg. 1994. Mechanisms involved the shedding of soluble cytokine receptors and processing of transmembrane in the processing of the p55 and the p75 tumor necrosis factor (TNF) receptors cytokine precursors in human monocytic cells. Cytokine 9:340. to soluble receptor forms. Lymphokine Cytokine Res. 13:203. 28. Jones, S. A., S. Horiuchi, D. Novick, N. Yamamoto, and G. M. Fuller. 1998. 36. Penton-Rol, G., S. Orlando, N. Polentarutti, S. Bernasconi, M. Muzio, M. Introna, 2ϩ Shedding of the soluble IL-6 receptor is triggered by Ca mobilization, while and A. Mantovani. 1999. Bacterial lipopolysaccharide causes rapid shedding, basal release is predominantly the product of differential mRNA splicing in followed by inhibition of mRNA expression, of the IL-1 type II receptor, with THP-1 cells. Eur. J. Immunol. 28:3514. concomitant up-regulation of the type I receptor and induction of incompletely 29. Giri, J. G., J. Wells, S. K. Dower, C. E. McCall, R. N. Guzman, J. Slack, spliced transcripts. J. Immunol. 162:2931. T. A. Bird, K. Shanebeck, K. H. Grabstein, J. E. Sims, et al. 1994. Elevated levels 37. Mullberg, J., K. Althoff, T. Jostock, and S. Rose-John. 2000. The importance of of shed type II IL-1 receptor in sepsis: potential role for type II receptor in shedding of membrane proteins for cytokine biology. Eur. Cytokine Network regulation of IL-1 responses. J. Immunol. 153:5802. 11:27. Downloaded from 30. Pruitt, J. H., M. B. Welborn, P. D. Edwards, T. R. Harward, J. W. Seeger, 38. Kiessling, L. L., and E. J. Gordon. 1998. Transforming the cell surface through T. D. Martin, C. Smith, J. A. Kenney, R. I. Wesdorp, S. Meijer, et al. 1996. proteolysis. Chem. Biol. 5:R49. Increased soluble interleukin-1 type II receptor concentrations in postoperative 39. Heaney, M. L., and D. W. Golde. 1998. Soluble receptors in human disease. patients and in patients with sepsis syndrome. Blood 87:3282. J. Leukocyte Biol. 64:135. http://www.jimmunol.org/ by guest on September 25, 2021