Domain of the CD164 Receptor Proliferation Interact with the First

Home , CD164, CD43, CD44, MUC1, Selectin

CD164 Monoclonal Antibodies That Block Hemopoietic Progenitor Cell Adhesion and Proliferation Interact with the First Mucin Domain of the CD164 Receptor This information is current as of September 28, 2021. Regis Doyonnas, James Yi-Hsin Chan, Lisa H. Butler, Irene Rappold, Jane E. Lee-Prudhoe, Andrew C. W. Zannettino, Paul J. Simmons, Hans-Jörg Bühring, Jean-Pierre Levesque and Suzanne M. Watt

J Immunol 2000; 165:840-851; ; Downloaded from doi: 10.4049/jimmunol.165.2.840 http://www.jimmunol.org/content/165/2/840 http://www.jimmunol.org/ References This article cites 41 articles, 27 of which you can access for free at: http://www.jimmunol.org/content/165/2/840.full#ref-list-1

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists by guest on September 28, 2021

• 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 © 2000 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. CD164 Monoclonal Antibodies That Block Hemopoietic Progenitor Cell Adhesion and Proliferation Interact with the First Mucin Domain of the CD164 Receptor1

Regis Doyonnas,* James Yi-Hsin Chan,* Lisa H. Butler,* Irene Rappold,* Jane E. Lee-Prudhoe,* Andrew C. W. Zannettino,† Paul J. Simmons,‡ Hans-Jo¨rg Bu¨hring,§ Jean-Pierre Levesque,‡ and Suzanne M. Watt2§

The novel sialomucin, CD164, functions as both an adhesion receptor on human CD34؉ cell subsets in bone marrow and as a potent negative regulator of CD34؉ hemopoietic progenitor cell proliferation. These diverse effects are mediated by at least two functional epitopes defined by the mAbs, 103B2/9E10 and 105A5. We report here the precise epitope mapping of these mAbs together with that of two other CD164 mAbs, N6B6 and 67D2. Using newly defined CD164 splice variants and a set of soluble Downloaded from recombinant chimeric proteins encoded by exons 1–6 of the CD164 gene, we demonstrate that the 105A5 and 103B2/9E10 functional epitopes map to distinct glycosylated regions within the first mucin domain of CD164. The N6B6 and 67D2 mAbs, in contrast, recognize closely associated and complex epitopes that rely on the conformational integrity of the CD164 molecule and encompass the cysteine-rich regions encoded by exons 2 and 3. On the basis of their sensitivities to reducing agents and to sialidase, O-sialoglycoprotease, and N-glycanase treatments, we have characterized CD164 epitopes and grouped them into three classes by analogy with CD34 epitope classification. The class I 105A5 epitope is sialidase, O-glycosidase, and O-sialoglycoprotease sensitive; http://www.jimmunol.org/ the class II 103B2/9E10 epitope is N-glycanase, O-glycosidase, and O-sialoglycoprotease sensitive; and the class III N6B6 and 67D2 epitopes are not removed by such enzyme treatments. Collectively, this study indicates that the previously observed differential expression of CD164 epitopes in adult tissues is linked with cell type specific post-translational modifications and suggests a role for epitope-associated carbohydrate structures in CD164 function. The Journal of Immunology, 2000, 165: 840–851.

ecently, we have identified and cloned human CD164, a al., manuscript in preparation) and prevents the recruitment of novel 80- to 100-kDa type 1 transmembrane sialomucin CD34ϩCD38low/Ϫ cells into cycle in response to the cytokines, ϩ 3 R that is highly expressed on primitive CD34 hemopoietic IL-3, IL-6, stem cell factor (SCF), and G-CSF (1). All these re- by guest on September 28, 2021 progenitor cells (1–4) (J.Y.-H. Chan et al., manuscript in prepa- sults suggest that CD164 may act as a potent negative signaling ration). Analyses of transfectants expressing human CD164 have molecule for hemopoietic progenitor cell proliferation. The CD164 allowed the identification of at least four mAbs, 103B2/9E10, Ag, when analyzed with any of the four CD164 mAbs described 105A5, N6B6, and 67D2, that specifically recognize this sialomu- above, is expressed during ontogeny on CD34ϩ intra-aortic cell cin (1–4). Of these, the interaction of the 103B2/9E10 mAb with clusters in human wk 4–5 embryos and has been shown to be the CD164 receptor inhibits the adhesion of CD34ϩ cells to bone expressed on primitive human hemopoietic progenitors from fetal marrow stromal cells in vitro (1). Interestingly, similar interactions liver, cord blood, and bone marrow (4). The highest cell surface with the 103B2/9E10 or 105A5 mAbs inhibit the proliferation and expression of CD164 epitopes on these cells occurs on the more differentiation of primitive CD34ϩ erythroid and granulocyte- primitive subset of CD34ϩ cells (CD34high, AC133high, monocyte progenitors in colony forming assays (J.Y.-H. Chan et CD38low). It is also expressed on the vast majority of the linϪCD34low/ϪCD38low/Ϫ cells with the capacity for long term repopulation of hemopoiesis in an in vivo fetal sheep model (4, 5). *Medical Research Council Molecular Haematology Unit, Institute of Molecular CD164 expression is maintained at a lower level on the surface of Medicine, John Radcliffe Hospital, Headington, Oxford, United Kingdom; †Hanson all the committed myeloid and erythroid progenitors, with low or ‡ Centre for Cancer Research, Adelaide, Australia; Stem Cell Laboratory, Peter Mac- negligible levels of expression on mature peripheral blood neutro- Callum Cancer Institute, Melbourne, Australia; and §Medizinische Universita¨tsklinik II, University of Tubingen, Tubingen, Germany phils and erythrocytes (4). In contrast to their common distribution Received for publication December 15, 1999. Accepted for publication April pattern on hemopoietic progenitor cells, the CD164 epitopes de- 25, 2000. fined by the 105A5 and 103B2/9E10 mAbs are differentially and The costs of publication of this article were defrayed in part by the payment of page often reciprocally expressed on lymphoid cells, endothelia, postcap- charges. This article must therefore be hereby marked advertisement in accordance illary high endothelial venules, and basal/subcapsular epithelia in he- with 18 U.S.C. Section 1734 solely to indicate this fact. mopoietic and nonhemopoietic tissues, while the N6B6 and 67D2 1 This work was supported by the United Kingdom Medical Research Council, the ϩ ϩ United Kingdom Leukaemia Research Fund, SmithKline Beecham, INTAS/RFBR, mAbs react with both the 103B2/9E10 and 105A5 cell subsets (4). E. U. Biotech. Framework 4, and Taiwan Government Grants (to R.D., J.Y.-H.C., Differential epitope expression has also been described for other L.H.B., I.R., J.L.-P., and S.M.W.), National Health and Medical Research Council of members of the sialomucin adhesion receptor family, to which Australia (to A.C.W.Z., P.J.S., and J.-P.L.), and Deutsche Forschungsgemeinschaft SFB 510, Project A1 (to H.J.B.). 2 Address correspondence and reprint requests to Dr. Suzanne M. Watt, Medical 3 Abbreviations used in this paper: SCF, stem cell factor; GAG, glycosamino- Research Council Molecular Hematology Unit, Institute of Molecular Medicine, The glycan; PCLP, podocalyxin-like protein; PSGL-1, P-selectin glycoprotein ligand-1; John Radcliffe Hospital, Headington, Oxford, United Kingdom OX3 9DS. E-mail MAdCAM-1, mucosal addressin cell adhesion molecule-1; GlyCAM-1, glycosyla- address: [email protected] tion-dependent cell adhesion molecule-1.

CD164 belongs. The expanding family of sialomucin receptors Human bone marrow stromal cells, CD34ϩ cell isolation, and includes CD34, PCLP, PSGL-1 (CD162), CD45RA, MAdCAM,-1, cell lines Sgp200, GlyCAM-1, and CD43 (reviewed in Refs. 6–10). These molecules are expressed on hemopoietic progenitor cells and/or on All human cell samples were obtained with patient permission and with associated stromal, macrophage, T lymphoid, and/or endothelial ethical consent of the institutions or hospitals concerned. Human bone marrow stromal cells were prepared and stained with the CD164 mAbs cells in hemopoietic microenvironments, where they function in using the immunofluorescence technique previously described (2, 4). Hu- regulating hemopoiesis, leukocyte trafficking, inflammatory re- man CD34ϩ cells (Ͼ90% purity) were purified from fresh cord samples sponses, or T cell activation. They are all serine and threonine rich, provided by Prof. J. Hows (Southmead Hospital, Bristol, U.K.), using the allowing the potential for extensive O-linked glycosylation. They Miltenyi Biotech (Bergish Gladbach, Germany) miniMACS CD34 stem cell isolation kit (2). The human KG1a, KG1B, THP-1, U937, CEM, are either secreted or transmembrane molecules with the ability to RPMI-1788, TF1, and 293T and the mouse MS.5 cell lines were cultured extend well above the glycocalyx to promote ligand interactions. as previously described (2, 4). The diversity of these sialomucin receptors is further enhanced by alternative splicing and by cell-specific glycosyltransferase-medi- Sialidase and O-sialoglycoprotease treatment of cell lines ated sialyl, fucosyl, or sulfate modifications of their oligosaccha- Clostridium perfringens sialidase was purchased from Roche (Mannheim, ride side chains, which alter mucin function and allow it to be Germany). Pasteurella haemolytica O-sialoglycoprotein endopeptidase regulated independently of the rest of the molecule (reviewed in (O-sialoglycoprotease) was prepared by Cedarlane Laboratories (Hornby, Refs. 6–10). Epitope characterization of these receptors has been Canada) and was purchased from Accurate Chemical & Scientific Corp. (Westbury, NY). For enzymatic treatment, 7 ϫ 106 KG1a cells were in- very helpful in defining the relationship between the post-transla- ␮

cubated in 300 l of PBS or RPMI with or without 0.1 U of C. perfringens Downloaded from tional modifications occurring on these molecules and their spe- sialidase or 120 ␮gofO-sialoglycoprotease for 60 min at 37°C. Cells were cific distribution and functions in various tissues. Some of the sia- then washed with PBS containing 0.2% (w/v) BSA (PBS-BSA) and 0.1% lomucins interact with selectin ligands in vitro, a process that may (w/v) sodium azide before flow cytometric analyses. be abrogated by sialidase or O-siaolglycoprotease treatment of the sialomucin (reviewed in Refs. 6–10). Their ligand specificities de- Flow cytometric analyses pend on post-translational modifications of the peptide or oligo- All analyses were conducted at 4°C. The sialidase- or O-sialoglycopro- saccharide side chains, which are tissue specific. For example, cor- tease-treated and the untreated cell lines were blocked with FcR-blocking http://www.jimmunol.org/ rect sulfation and glycosylation allow CD34, PCLP, GlyCAM-1, agent (Miltenyi Biotech) according to the manufacturer’s instructions and Sgp200, and MAdCAM-1, to act as high affinity ligands for L- then labeled with the CD164 mAbs; with the CD34 mAbs, Tu¨k-3 and My10; or with isotype-matched control mAbs followed by FITC-anti- selectin on high endothelial venules of peripheral lymph nodes or isotype-specific secondary Abs (Southern Biotechnology Associates) as endothelia of Peyer’s patches (reviewed in Refs. 6–10). Like detailed above. Cells were also stained with PE-conjugated QBEND-10 or CD164, the PSGL-1, CD34, and CD43 sialomucins may also func- an irrelevant PE-mIgG1 according to the manufacturer’s protocol. After the ␮ tion as signaling molecules that regulate cell proliferation (11–18), addition of 2 g/ml propidium iodide (Sigma, St. Louis, MO), cells were analyzed on a FACSCalibur using CellQuest software (both from Becton possibly by enabling other receptor-ligand interactions to occur. Dickinson, Sunnyvale, CA) (2). Experiments were repeated on at least These diverse functions are thought to be the result, at least in part, of three independent occasions. by guest on September 28, 2021 cell type- and stage-specific oligosaccharide modifications, particularly those involving sialylation (19–22; reviewed in Refs. 6–10). Generation of full-length CD164 cDNA splice variant constructs In this article we report for the first time the identification of CD164 isoforms were PCR amplified from templates subcloned in the novel isoforms of CD164 and of three classes of epitopes on the pMOS-Blue-T vector (Amersham, Aylesbury, U.K.) after the RT-PCR CD164 sialomucin. Two of these, the class I and II epitopes, have analyses.4 The CD164(E1–6) and CD164(E⌬5) cDNA templates were de- been shown previously to encompass sites that regulate the adhe- rived from human kidney, while the CD164(E⌬4) cDNA template was sion and proliferation of CD34ϩ cell subsets in vitro (1) (J. Y.-H. derived from human spleen. PCR amplifications were conducted using the Expand high fidelity PCR system (Roche) with 100 ng of the CD164(E1– Chan et al., manuscript in preparation). These epitopes are differ- 6), CD164(E⌬5), or CD164(E⌬4) cDNAs, and 1 ␮M concentrations of the entially glycosylated regions located on the N-terminal mucin-like F164 forward primer (5Ј-3Ј GATCGCGGCCGCCGCTGAGGACAC domain of the CD164 molecule that is encoded by exon 1 of the GATGTCGCGG) containing a NotI restriction enzyme site and the reverse CD164 gene. This study together with our previous analysis suggest R6BCD164 SpeI primer (5Ј-3Ј GGACTAGTTTACAGAGT GTGGTA that these glycosylation-dependent epitopes are regulated through cell ATTTCGT) containing a SpeI restriction site after the stop codon. The samples were digested with NotI and SpeI restriction enzymes and subtype-specific post-translational modifications of CD164. cloned into a similarly digested pEFBos-HPC4-TM vector (24) that removed the HPC4-TM sequence. Positive clones were sequenced as described below, and Maxipreps of each cDNA were prepared using the Promega Megaprep separation protocol (Promega, Madison, WI) according to the manufacturer’s instructions. Materials and Methods Primary Abs Production of transient transfectants expressing CD164 cDNA The murine CD164 mAbs, 103B2/9E10 (mIgG3), 105A5 (mIgM), N6B6 splice variants (mIgG2a), and 67D2 (mIgG1), were prepared as previously described The CD164 splice variant cDNAs were transfected into MS.5 mouse stro- (1–4). The CD66 mAb, clone D14-HD11 (mIgG1), was obtained from the mal cells at 40–50% confluence using calcium phosphate as a facilitator Fifth Leucocyte Culture Conference (23). The CD33 mAb, clone WM-54 (2–10 ␮g of plasmid DNA/well of a 24-well tissue culture plate). At 2 days (mIgG1), was obtained from Dakopatts (Glostrup, Denmark). The PE-con- post-transfection, cells were resuspended and washed twice in PBS and jugated CD34 mAb, QBEND-10 (mIgG1), was purchased from Cambridge lysed directly in 4ϫ modified Laemmli reducing sample buffer (0.05% Biosciences (Cambridge, U.K.). The CD34 mAbs, clone My10 (mIgG1) (w/v) bromophenol blue, 10% (v/v) glycerol, 0.5% (w/v) SDS, and 5 mM and Tu¨k-3 (mIgG3), were purchased from Becton Dickinson (San Jose, DTT in 0.1 M Tris-HCl, pH 6.8) containing 1ϫ Complete protease inhib- CA) or were provided by Prof. D. Mason (LRF Center, Department of itors (Roche) and resolved by 6 or 10% SDS-PAGE before immunoblotting Cellular Sciences, Oxford, U.K.), respectively. As comparative negative as described below. Transiently transfected cells were also fixed in situ in controls, irrelevant mAbs of the mIgG1, mIgG2a, mIgM (Dakopatts), and mIgG3 (Southern Biotechnology Associates, Birmingham, AL) isotypes, unconjugated or PE conjugated, were used in place of primary Abs. Abs 4 J. Y.-H. Chan, J. E. Lee-Prudhoe, B. Jorgensen, G. Ihrke, R. Doyonnas, A. C. W. were used as tissue culture supernatants or purified Ig fractions. Zannettino, P. J. Simmons, V. J. Buekle, and S. M. Watt. Submitted for publication. 842 EPITOPE MAPPING OF CD164 FUNCTIONAL DOMAINS

24-well plates using a 1-ml 50/50 mixture of acetone/methanol and stained in the culture supernatants were affinity isolated on protein A-Sepharose (4 with each CD164 mAb or with an irrelevant isotype-matched control mAb Fast Flow; Pharmacia Biotec, Piscataway, NJ) and analyzed as described followed by HRP-conjugated goat anti-mouse Ig (Dakopatts) at a 1/1000 previously (26). dilution (4). Cells were counterstained with Harris’ hematoxylin (Surgi- path, Eynesbury, U.K.) and viewed under a Leitz inverted microscope Purification of CD164 native molecules from KG1a cells (Leica U.K. Ltd., Milton Keynes, U.K.). Images were captured on a JVC A soluble fraction of detergent lysate was prepared from 2 ϫ 108 KG1a 3-CCD color video camera using the Neotech JVC application (Datacell, cells as described for Triton X-100 insolubility studies. This material was Maidenhead, U.K.). passed twice over a 1-ml column of protein A-Sepharose. The unbound Triton X-100 insolubility studies material was then passed over a 1-ml protein A-Sepharose to which 1 mg of purified N6B6 mAb had been bound and covalently coupled with di- KG1a cells (8 ϫ 104) were resuspended in 1% (v/v) Triton X-100 lysis methyl pimelimidate (Pierce, Rockford, IL). After washing with PBS con- buffer (20 mM Tris-HCl, 1% (v/v) Triton X-100, 150 mM NaCl, and 1ϫ taining 0.1% Triton X-100, the bound material was eluted with 100 mM Complete protease inhibitors) and incubated for 30 min at 4°C. After cen- triethylamine containing 0.1% Triton X-100 and neutralized with 0.1 vol of trifugation at 14,000 rpm for 15 min at 4°C, the supernatant or soluble 3 M Tris (pH 6.8); a 1/200 dilution of the resulting material was analyzed fraction and the pellet or insoluble fraction, resuspended by the addition of on SDS-PAGE followed by Western blotting with CD164 mAbs. 250 ␮l of nonreducing 1ϫ Laemmli buffer, were collected. After the addition of Laemmli loading buffer containing 5 mM DTT, the lysate proteins Enzymatic treatment and immunoblotting of modified soluble from both fractions were boiled for 5 min, resolved on 8% SDS-PAGE, and constructs immunoblotted as described above. Five-microgram aliquots of the lyophilized CD164(E1–3)-Fc* and Generation of recombinant soluble chimeric proteins CD164(E1–6a)-Fc* constructs or a 1/50 dilution of lyophilized CD164 purified from KG1a cells were left untreated or were treated for 16 h at Soluble extracellular domain deletion cDNA constructs prepared on the 37°C with N-glycosidase F (200 ␮U/ml; Flavobacterium meningosepticum Downloaded from basis of the exon organization and containing regions encoded by exon 1 enzyme; Roche) in 10 mM sodium phosphate buffer, pH 6; with sialidase (CD164(E1)), exons 1 and 2 (CD164(E1–2)), exons 1–3 (CD164(E1–3)), (500 ␮U/ml C. perfringens enzyme; Roche) in 10 mM sodium phosphate exons 1–4 (CD164(E1–4)), and exons 1–4 plus the extracellular region of buffer, pH 6; with O-glycosidase (50 ␮U/ml Streptococcus pneumoniae exon 6 (6a; CD164(E⌬5)) were generated by PCR amplification of the enzyme; Oxford Glycosystems, Abingdon, U.K.) in 100 mM sodium ci- corresponding cDNA fragments from the CD164(E⌬5) cDNA in the trate/phosphate buffer, pH 6; with ␣-fucosidase (250 ␮U/ml bovine kidney pGEM-T vector (1). Constructs containing exons 1–5 (CD164(E1–5)) and enzyme; Oxford Glycosystems) in 100 mM sodium citrate/phosphate exons 1–6a (CD164(E1–6a)) were produced from a template generated by buffer, pH 6, separately or in combination for the removal of N- and O- http://www.jimmunol.org/ PCR amplification of the CD164(E1–6) cDNA (J. Y.-H. Chan et al., manu- linked carbohydrates. The soluble constructs CD164(E1–3)Fc* and script in preparation) derived from a normal human colon sample (Clon- CD164(E1–6a)-Fc* were also treated with 250 ␮g/ml O-sialoglycopro- tech, Palo Alto, CA). PCR amplifications were conducted using the Expand tease in PBS containing 1 mM CaCl2, pH 7.2. The CD33-Fc construct was high fidelity PCR system (Roche). Oligonucleotide primers (Genosys Bio- used as a negative control. All samples were heated for 5 min at 95°C in technologies Europe, Cambridge, U.K.) containing NotIorXhoI restriction 2ϫ Laemmli sample buffer (5% (v/v) glycerol, 0.25% (w/v) SDS, and enzyme sites (as underlined below) were used for PCR and were: F164 0.025% (w/v) bromophenol blue in 0.05 M Tris-HCl, pH 6.8) with 5 mM forward primer (5Ј-3Ј), GATCGCGGCCGCCGCT GAG GAC ACG ATG DTT and electrophoresed on 10% SDS-PAGE gels. One gel was stained TCG CGG for all the PCR amplifications plus one of the following reverse with Coomassie blue. Four gels were transferred onto polyvinylidene di- amplification primers for each exon; R164(E1), ATCCCTCGAGGG TGC fluoride Immobilon membranes (Millipore, Watford, U.K.) at9Vfor1h CGG AGT GGT GAC CAG; R164(E2), ATCCCTCGAGTC TTT ACA using a Semiphor semi-dry blot apparatus (Pharmacia Biotech) following

TTC TAT CCA AAA; R164(E3), ATCCCTCGAGAC GGA ACA GAA the manufacturer’s directions. The membranes were blocked overnight at by guest on September 28, 2021 GTC TGT CGT; R164(E4), ATCCCTCGAGGT AGA ATTGGC TGT 4°C in PBS-T (PBS with 0.05% (v/v) Tween-20) buffer plus 5% (w/v) TGG CAC; R164(E5), ATCCCTCGAGGT TGT ACC TGA TGT AGT nonfat powdered milk and then incubated with primary Ab for 30 min at AAC; and R164(E6a), ATCCCTCGAGAA GGT AGA CTT TCG CAC room temperature. After washing in PBS-T, a peroxidase-conjugated goat AGG. The CD164(E1,2,4) and CD164(E1,3,4) cDNA constructs were pro- anti-mouse Ig (Dakopatts) Ab diluted 1/5000 in PBS-T was applied for a duced using a two-step PCR strategy. In the first step, exon 1 (E1), exons further 30 min. Following extensive washing in PBS-T, blots were devel- 1 and 2 (E1,2), exons 3 and 4 (E3,4), and exon 4 (E4) were amplified oped using the ECL system (Amersham) as described by the manufacturer. individually using as the respective forward primers: F164 F(E1, 3), ACC ACT CCG GCA CCA GAT GAG AGC TAT TGT TCA; and F(E2, 4), ELISA analysis of proteins TGG ATA GAA TGT AAA GTT TCC ACG GCC ACT CCA; and as the respective reverse primers: R(E1, 3), TGA ACA ATA GCT CTC ATC Maxisorp 96-well Nunc (Life Technologies) or Immulon 4 (Dynatech, Dy- nal, Oslo, Norway) flat-bottom ELISA plates were coated with untreated or TGG TGC CGG AGT GGT; and R(E2, 4) TGG AGT GGC CGT GGA ␮ AAC TTT ACA TTC TAT CCA) and R164 (E4). To generate the enzyme-treated chimeric proteins (10 g/ml) in PBS overnight at 4°C, CD164(E1,2,4) cDNA, 5 ␮l of the E1, E2, and E4 PCR products were washed, and then blocked with PBS containing 2% (w/v) BSA (fraction V; allowed to anneal together for 10 min at 66°C and thus provided the tem- Sigma) and 0.02% (v/v) Tween-20 before incubation with CD164, CD66a, plate for the second step PCR. For the CD164(E1, 3,4) cDNA, 5 ␮lofE1 and CD33 or appropriate isotype-matched negative control mAbs. The as- and E3,4 PCR products were annealed as described above and provided the says were developed with alkaline phosphatase-conjugated goat anti- template for subsequent PCR amplification. These PCR amplifications mouse Ig (1/4000 dilution; Dakopatts) and para-nitrophenylpentene (Sig- were conducted as described above using the F164 forward and R164(E4) ma), and the absorbance was read at 405 nm in a Bio-Rad model 450 plate reverse primers for both constructs. Soluble recombinant chimeric extra- reader (Bio-Rad Laboratories, Hercules, CA) (2). cellular domain cDNA Fc-mutated (Fc*) constructs were prepared by di- Competitive binding assays gesting the PCR with NotI and XhoI restriction enzymes and subcloning these into the similarly digested IgMu/pEFBOS vector (24), which was Maxisorp plates were coated with CD164(E1–3)-Fc* and washed using the provided by Prof. P. Kincade (University of Oklahoma, Norman, OK) and same conditions as for the ELISA analysis. CD164(E1–3)-Fc*-coated con- was designed to prevent FcR binding. The non-Fc-mutated constructs, structs were then blocked with 103B2/9E10, 105A5, N6B6, or 67D2 as hCD66a-Fc (25) and hCD33(VC)-Fc (26), were prepared in the pIG vector undiluted tissue culture supernatants or with isotype-matched negative con- as previously described and used as controls in ELISA and Western blot- trol Abs at 10 ␮g/ml. After washing, each CD164 mAb was added to the ting analyses. The cDNA inserts were sequenced on a Perkin-Elmer ABI- preblocked constructs, and their reactivities were determined by the addi- PRISM 377 DNA sequencer (Perkin-Elmer-Applied Biosystems, Foster tion of FITC-conjugated anti Ig isotype-specific secondary Abs. The flu- City, CA) according to the manufacturer’s protocol using pEFBos forward orescence was detected on a Cytofluor II microplate fluorescence reader primer at position 1701–1718 bp (CTCAAGCCTCAGACAGTG), pEFBos (PerSeptive Biosystems, Hertford, U.K.) using a wavelength of 485 nm for reverse primer at position 2845–2828 bp (GGGAGACCTGATACTCTC), excitation. The percent binding was calculated as follows: 100 Ϫ [(fluo- pIG specific forward and reverse primers (25, 26), or primers specific for rescence reading for each CD164 mAb binding to the CD164(E1–3)-Fc* the cDNA sequences. The sequences were analyzed using Sequencher and protein in the presence of the blocking CD164 mAb) divided by (fluores- MacVector software programs (Oxford Molecular, Oxford, U.K.). The Fc cence reading for each CD164 mAb binding to the CD164(E1–3)-Fc* pro- and mutated Fc* fusion plasmids were transfected into 293T cells at 70– tein in the presence of the appropriate isotype matched negative control 80% confluence using calcium phosphate as a facilitator (25 ␮g of plasmid mAb) ϫ 100]. Results are presented as the mean Ϯ SD of triplicate de- DNA/15-cm diameter tissue culture plate). The Fc/Fc* chimeras produced terminations, and the experiment was repeated twice. The Journal of Immunology 843

FIGURE 1. Structure and sequences of the CD164 isoforms. A, Schematic representations of the CD164(E1–6), CD164(E⌬4), and CD164(E⌬5) sialomucins. u, mucin domains; Ⅺ, the cysteine-rich domain; ⅜, potential N-linked carbohydrates; horizontal bars with or without arrows, potential O-linked carbohydrates; arrows, potential sialic acid motifs on O-linked carbohydrates. B, cDNA of CD164(E1–6) and its deduced amino acid sequences indicated Downloaded from in triplet codons. E1 to E6 represent exon-encoded domains. E6a, E6b, and E6c are regions of exon 6 that encode the extracellular domain, the transmembrane region (TM), and the cytoplasmic tail, respectively. The two dark gray boxes represent the two mucin do- http://www.jimmunol.org/ mains. ‰, the cysteine residues; ⅜ , the predicted O-linked carbohydrate residues; u , the predicted N-linked carbohydrate residues. The transmembrane region is underlined and shadowed. The putative glycosaminoglycan attachment site is indicated in the lightly shaded box, and the potential tyrosine kinase

phosphorylation site is shown by an as- by guest on September 28, 2021 terisk. The GenBank accession number is AF263279.

Immunoblotting of cell lysates isoform that lacks exon 5, and the CD164(E⌬4) variant that has The equivalent of 8 ϫ 104 exponentially growing cell lines or 1 ϫ 104 exon 4 spliced out. The full-length isoforms are shown diagram- cultured human bone marrow stromal reticular, cord blood CD34ϩ purified matically in Fig. 1A. The peptide encoded by exon 1 is predicted cells or bone marrow mononuclear cells were resuspended in 1ϫ nonre- to be heavily glycosylated with three potential N-linked glycosyl- ducing Laemmli loading buffer (62.5 mM Tris-HCl, 2% (w/v) SDS, 10% ation sites and nine potential O-linked glycosylation sites (Fig. 1). ϫ (v/v) glycerol, and 0.1% (w/v) bromophenol blue, pH 6.8) containing 1 This exon 1 generates the first mucin-like domain of CD164. Pep- Complete protease inhibitors (Roche) plus 5 mM DTT as described above and boiled for 5 min, and lysate proteins were fractionated using 6 or 10% tides encoded by exons 2 and 3 do not contain any predicted O- SDS-PAGE. The proteins were transferred to polyvinylidene difluoride Im- linked glycosylation sites, but each possesses two potential mobilon membranes and immunoblotted with either the CD164 mAbs or N-linked glycosylation sites, and they contain all eight cysteine isotype-matched negative controls as described above. residues from the extracellular domain. This defines the cysteine-rich region that separates the two mucin domains. The second mucin-like Results domain is defined by peptides derived from exons 4, 5, and 6, which, Identification of three isoforms of CD164 like the exon 1-encoded peptide, is predicted to be very highly O- We have defined the genomic structure of human CD164 (J. Y.-H. glycosylated, with six potential sites encoded on exon 4, 10 on exon Chan et al., manuscript in preparation) and have shown that this 5, and seven on exon 6. In addition, peptides encoded by exons 4 and gene comprises six exons (E1–6) that undergo alternative splicing 6 have one potential N-linked glycosylation site each (Fig. 1). Thus, to generate at least three isoforms. These are the full-length isoform, CD164(E⌬4) and CD164(E⌬5) have a smaller second mucin domain CD164(E1–6), the originally identified CD164 or CD164(E⌬5) comparatively to the CD164(E1–6) molecule. This CD164(E1–6) 844 EPITOPE MAPPING OF CD164 FUNCTIONAL DOMAINS

molecule contains a putative membrane-proximal glycosaminoglycan attachment site situated at the splice junction between peptides derived from exons 5 and 6. This is missing in the CD164(E⌬5) isoform.

The CD164 mAbs do not distinguish between the different CD164 splice variants but recognize distinct domains on CD164 Our initial studies identiﬁed four mAbs, 103B2/9E10, 105A5, N6B6, and 67D2, that recognize human CD164 when expressed in FDCP-1 transfectants (1–4). Two of the mAbs, 103B2/9E10 and 105A5, mediate functional effects in vitro (1) (J. Y.-H. Chan et al., manuscript in preparation). These effects, which are summarized in Fig. 2, indicate that while the 103B2/9E10 mAb can partially inhibit the adhesion of CD34ϩ cells to bone marrow stroma (Fig. 2A), both the 103B2/9E10 and 105A5 mAbs inhibit nucleated cell production in liquid cultures (Fig. 2B) and colony formation by primitive granulocyte-monocyte (Fig. 2C) and erythroid (Fig. 2 D) precursors in clonogenic assays from CD34ϩ cells. From single

cell studies, the 103B2/9E10 mAb has been shown to prevent re- Downloaded from cruitment of CD34ϩCD38low/Ϫ cells into cycle in the presence of IL-3, IL-6, G-CSF, and SCF (Fig. 2E).

Reactivity of mAbs with splice variants of CD164 All four CD164 mAbs stain CD34ϩ hemopoietic cell subsets (1–

4), and as shown in Fig. 3A, they all stain cultured human bone http://www.jimmunol.org/ marrow stromal reticular cells. cDNAs encoding the three CD164 splice variants, CD164(E1–6), CD164(E⌬5), and CD164(E⌬4), were transiently transfected into the mouse stromal cell line, MS.5, and their protein products were analyzed by immunohistochemis- try and immunoblotting. All the CD164 mAbs reacted with the splice variants produced by these cells, indicating that the epitopes recognized by the 103B2/9E10, 105A5, N6B6, and 67D2 mAbs FIGURE 2. Functional effects of the CD164 mAbs, 103B2/9E10 and were not located on or did not encompass peptides encoded by

ϩ by guest on September 28, 2021 105A5, in vitro. A, Human bone marrow CD34 cells (104) labeled with exons 4 and 5. This is illustrated in Fig. 3B for two of the CD164 51 Cr were incubated at 4°C in RPMI 1640 medium with 2% FCS contain- mAbs, N6B6 and 103B2/9E10. On Western blots probed with ␮ ␮ ing 2 g/100 l of the CD164 mAbs, 103B2/9E10 and 105A5, or with a 103B2/9E10, 105A5, or N6B6, the apparent m.w. of the proteins cocktail of mAbs to the VLA-4 and VLA-5 integrins, P4C2 and PHM2, or expressed by the splice variants varied slightly, but fell within the a cocktail of isotype-matched mIgG3 and mIgM nonbinding negative con- range of 80–100 kDa that is observed for CD164 on human bone trol mAbs and then transferred to each well of a 96-well plate containing ϩ 104 allogeneic cultured human bone marrow stromal reticular cells per marrow, on bone marrow stromal reticular cells, on CD34 he- well. Unbound cells were removed, and adhesion was quantitated by liquid mopoietic progenitors, and on a set of hemopoietic cell lines rep- scintillation counting of Triton X-100-solubilized lysates. Data are pre- resenting different lineages. Examples of the protein products de- sented as the mean Ϯ SEM for three experiments as a percentage of the tected with these mAbs after SDS lysis of hemopoietic cell lines control adhesion in the presence of the nonbinding mAb control, which has are shown in Fig. 3C (lanes 1–7) and Fig. 4. These mAbs identified ϩ been normalized to 100%. B, Human bone marrow CD34 cells (103) were the different CD164 epitopes on cell lines representing different cultured in triplicate in serum-deprived medium containing 10 ng/ml of hemopoietic lineages to differing degrees, but were all strongly ␤ each of the recombinant human cytokines, IL-1 , IL-3, IL-6, G-CSF, GM- reactive with the most immature CD34ϩ hemopoietic multipoten- CSF, and SCF, in the presence of 10 ␮g/ml of the CD164 mAbs, 103B2/ tial progenitor cell line, KG1a (Fig. 4A). This is consistent with our 9E10 and 105A5, or with isotype-matched mIgG3- and mIgM-negative findings that all CD164 epitopes identified to date are highly ex- control mAbs. Additional factors and mAbs were added on days 7, 14, and ϩ 21 of culture. Nucleated cells were harvested on day 28 and counted. Data pressed on the most primitive CD34 cell subsets from normal are the mean Ϯ SEM (n ϭ 3) of nucleated cells generated in the presence bone marrow, cord blood, and fetal liver (1–4). Some variability in of the negative isotype-matched control mAbs, which have been normal- apparent m.w. was also apparent among the cell lines, with the ized to 100%. C and D, Human bone marrow CD34ϩ cells (103) were promonocytic cell line, THP-1 (Fig. 4A), and the myelomonocytic incubated with 30 ␮g/ml of the CD164 mAbs, 103B2/9E10 and 105A5, or cell line, HL60 (data not shown), exhibiting the lowest electro- the isotype-matched mIgG3 or mIgM negative controls for1hat4°C. Cells phoretic mobilities of SDS-PAGE. It is unclear from the present were plated in 0.9% methocel supplemented with 10 ng/ml each of recom- studies if this molecular mass variability is due to glycosylation binant human IL-1␤, IL-3, IL-6, G-CSF, GM-CSF, erythropoietin, and SCF. Results are the mean Ϯ SEM for three experiments, expressed as a percentage of the value in day 14 clonogenic cells, CFU-GM (C), or eryth- rocyte blast-forming units (BFU-E) (D), from cultures containing CD164 mAbs over those containing the negative isotype-matched mAbs, which ng/ml), G-CSF (100 ng/ml), and SCF (100 ng/ml). One hundred twenty have been normalized to 100%. E, Terasaki wells were coated with 0.5 ␮l wells were analyzed for each set of conditions. Plates were monitored on of 30 ␮g/ml CD164 mAb, 103B2/9E10, with an isotype-matched negative day 10 of culture to determine which cells underwent at least one cell ␤ Ϯ control mIgG3 mAb, or with an anti- 1 integrin mAb, P4C2. Single division. The data are expressed as the mean SD percentage of dividing CD34ϩCD38low/Ϫ human bone marrow cells were added to each well and cells compared with the mIgG3-negative control culture, which was nor- cultured in serum-deprived medium containing IL-3 (10 ng/ml), IL-6 (10 malized to 100%. The Journal of Immunology 845

FIGURE 3. Expression of CD164 epitopes by stromal reticular cells and MS.5 transfectants. A, Immunoﬂuores- cence staining of cultured human bone marrow stromal reticular cells with CD164 mAbs: mAb 103B2/9E10 (a), mAb 105A5 (c), mAb N6B6 (e), and mAb 67D2 (g). Isotype-matched negative control mAbs for each CD164 mAb are shown as insets (b, d, f, and h). B, MS.5 cells were transfected with CD164(E1–6) (a and e), CD164(E⌬4) (b and f), CD164(E⌬5) (c and g), or left untrans- fected (d and h). Cells were stained with either N6B6 (a–d) or 103B2/9E10 (e–h) using the immunoperoxidase technique. C, Immunoblot analyses of CD164. Non- transfected MS.5 cells (lane 1) or MS.5 Downloaded from cells transfected with the different CD164 splice variants are shown: CD164(E1–6) (lane 2), CD164(E⌬4) (lane 3), and CD164(E⌬5) (lane 4) were lysed in Lae- mmli SDS lysis buffer containing 5 mM DTT and immunoblotted with N6B6 (lanes 1–4). Human bone marrow cells http://www.jimmunol.org/ (lanes 5 and 8), CD34ϩ puriﬁed cord blood cells (lanes 6 and 9), or cultured bone marrow stromal reticular cells (lanes 7 and 10) were lysed and immunoblotted with N6B6 (lanes 5–7) or 67D2 (lanes 8–10). Molecular mass markers were myosin (220 kDa), phosphorylase b (97.4 kDa), BSA (69 kDa), OVA (46

kDa), and carbonic anhydrase (31 kDa). by guest on September 28, 2021

FIGURE 4. Immunoblot analyses of CD164 from hemopoietic cell lines. A, TF1 (lane 1), KG1a (lane 2), KG1b (lane 3), U937 (lane 4), THP1 (lane 5), CEM (lane 6), and RPMI 1788 (lane 7) cells, were lysed in SDS-Laemmli lysis buffer containing 5 mM DTT and resolved by 10% SDS-PAGE before immunoblotting with N6B6, 103B2/9E10, or 105A5 as indicated below each blot. B, KG1a cells were lysed with 1% SDS (lane 1) or 1% Triton X-100 (lanes 2 and 3). The Triton X-100-insoluble fraction was solubilized in SDS before analysis. The distribution of the CD164 monomer and dimer and the Ͼ220-kDa complex between the Tri- ton X-100-soluble (lane 3) and Triton X-100-insoluble (lane 2) fractions was assessed by immunoblotting with the four CD164 mAbs as indicated

below each blot. The Mr markers were the same as those shown in Fig. 3C. 846 EPITOPE MAPPING OF CD164 FUNCTIONAL DOMAINS

differences among CD164 molecules on different hemopoietic lineages or reﬂects the expression of different CD164 splice variants. These studies are currently under investigation. It was of further interest to note that while the monomeric form of CD164 (80–100 kDa) was found after SDS lysis of cells with all four CD164 mAbs, Ͼ the 67D2 mAb also detected a band with an apparent Mr 220 kDa. This is illustrated in Fig. 3C (lanes 8–10) on a blot of human bone marrow, of human cord blood CD34ϩ hemopoietic cells, and of cultured bone marrow stromal reticular cells. This additional high molecular mass band may represent a Triton X-100-insoluble form of the CD164 molecule caused by multimeric association, cytoskeletal interaction with its cytoplasmic tail, or glycosaminoglycan (GAG) modiﬁcation. This insoluble form remains accessi- ble to binding by the 67D2 mAb, but not by the other three CD164 mAbs. To test this possibility, we lysed KG1a cells with Triton X-100 and collected the Triton X-100-soluble and -insoluble fractions. These fractions were then boiled with SDS lysis buffer containing 5 mM DTT and subjected to SDS-PAGE followed by im-

munoblotting with the different CD164 mAbs. As indicated in Fig. Downloaded from 4B, all CD164 mAbs reacted with the 80- to 100-kDa CD164 monomer and with the 160- to 180-kDa CD164 dimer in the Triton X-100-soluble fraction as has been documented previously for Tri- ton X-100 lysis conditions (1). In contrast, 67D2 was the only Ͼ CD164 mAb to react with a Mr species 220 kDa that was derived

from the Triton X-100-insoluble fraction (Fig. 4B, 67D2, lane 2). http://www.jimmunol.org/ Since this insoluble form of CD164 has been identiﬁed at 320 kDa, it could represent a tertrameric form of the molecule. Further studies are in progress to identify this hypothetical tetrameric associ- FIGURE 5. N6B6 and 67D2 mAbs recognize conformationally depen- ation and to characterize possible cytoskeletal elements or GAGs dent epitopes. A, CD164-Fc* domain truncated chimeric proteins. Two that might bind to the CD164 sialomucin. micrograms of each puriﬁed protein were resolved by 10% SDS-PAGE under reducing conditions and visualized with Coomassie blue. Approximate Reactivity of mAbs with CD164 domain truncation mutants. molecular masses deduced from electrophoretic mobilities are: CD164(E1)-

Fc*, 50 kDa (lane 1); CD164(E1–2)-Fc*, 65 kDa (lane 2); CD164(E1–3)-Fc*, To characterize the epitopes recognized by the different CD164 by guest on September 28, 2021 80 kDa (lane 3); CD164(E1–4)-Fc*, 90 kDa (lane 4); CD164(E1–5)-Fc*, 95 mAbs, a set of nine domain truncation mutants were produced in kDa (lane 5); CD164(E⌬5)-Fc*, 95 kDa (lane 6); CD164(E1–6a)-Fc*, 100 293T cells: CD164(E1)-Fc*, CD164(E1–2)-Fc*, CD164(E1–3)-Fc*, kDa (lane 7); CD164(E1, 2, 4)-Fc*, 80-kDa (lane 8), and CD164(E1, 3, 4)- CD164(E1–4)-Fc*, CD164(E1–5)-Fc*, CD164(E⌬5)-Fc*, Fc*, 75 kDa (lane 9). B, Aliquots of the CD164(E1–3)-Fc* soluble chimeric CD164(E1–6a)-Fc*, CD164(E1, 2, 4)-Fc*, and CD164(E1, 3, 4)-Fc* proteins were treated with different concentrations of DTT (0–100 mM), an- (Fig. 5A). On SDS-PAGE, all puriﬁed soluble constructs were re- alyzed by 10% SDS-PAGE, and immunoblotted with the CD164 mAbs 103B2/9E10, 105A5, N6B6, and 67D2. Human CD33-Fc was resolved with 5 solved as single glycosylated protein bands, except CD164(E1)- mM DTT and immunoblotted with each CD164 mAb and is represented a Fc*, which occurred as two isoforms: one highly glycosylated and negative control (Co). The molecular mass markers were the same as those one lacking some oligosaccharide side chains (Fig. 5A). This was shown in Fig. 3C. conﬁrmed by the fact that the additional lower molecular mass band was not recognized by 103B2/9E10 (data not shown), which is dependent on both N- and O-linked glycosylation for binding, but is detected by the 105A5 mAb, which binds sialic acid moieties

Table I. Reactivity of CD164 mAbs with domain truncation-soluble recombinant CD164 proteinsa

mAbs

Soluble Constructs 103B2/9E10 105A5 N6B6 67D2 Ig-negative control

CD164(E1)-Fc* 1.276 Ϯ 0.152 0.916 Ϯ 0.221 0.150 Ϯ 0.009 0.150 Ϯ 0.009 0.159 Ϯ 0.011 CD164(E1–2)-Fc* 1.217 Ϯ 0.189 0.599 Ϯ 0.059 0.147 Ϯ 0.006 0.136 Ϯ 0.002 0.147 Ϯ 0.010 CD164(E1–3)-Fc* 1.098 Ϯ 0.199 0.814 Ϯ 0.045 1.392 Ϯ 0.194 0.790 Ϯ 0.143 0.138 Ϯ 0.009 CD164(E1–4)-Fc* 1.318 Ϯ 0.021 0.710 Ϯ 0.012 1.508 Ϯ 0.000 0.992 Ϯ 0.125 0.159 Ϯ 0.071 CD164(E1–5)-Fc* 1.421 Ϯ 0.017 0.720 Ϯ 0.010 1.452 Ϯ 0.127 0.934 Ϯ 0.142 0.136 Ϯ 0.001 CD164(E⌬5)-Fc* 1.428 Ϯ 0.039 0.750 Ϯ 0.065 1.472 Ϯ 0.147 0.832 Ϯ 0.170 0.145 Ϯ 0.001 CD164(E1–6a)-Fc* 1.254 Ϯ 0.294 0.741 Ϯ 0.126 1.249 Ϯ 0.262 0.857 Ϯ 0.132 0.164 Ϯ 0.007 CD164(E1,2,4)-Fc* 1.269 Ϯ 0.128 0.734 Ϯ 0.058 0.163 Ϯ 0.006 0.159 Ϯ 0.006 0.133 Ϯ 0.002 CD164(E1,3,4)-Fc* 1.531 Ϯ 0.037 0.625 Ϯ 0.049 0.198 Ϯ 0.013 0.150 Ϯ 0.006 0.135 Ϯ 0.001 CD66a-Fc 0.156 Ϯ 0.018 0.145 Ϯ 0.011 0.146 Ϯ 0.016 0.138 Ϯ 0.004 0.145 Ϯ 0.011

a The ELISA was performed as described in Materials and Methods, with the CD66a-Fc protein acting as a negative control for binding. The Ig-negative control represents a mixture of irrelevant isotope-matched mIgG1, mIgG2a, mIgG3, and mIgM mAbs that do not react with CD164. Results shown are means Ϯ SD triplicate measurements from one experiment and were consistent when repeated three times. The CD66a-Fc soluble protein, but not the CD164-Fc* domain deletion constructs, reacted with the CD66 mAb, D14-HD11, giving an ELISA reading of 1.872 Ϯ 0.140. The Journal of Immunology 847

FIGURE 6. Sialidase and O-sialoglycoprotease treatment of KG1a cells identifies three classes of CD164 epitopes. KG1a cells were untreated or enzymatically digested for1hat37°C, then stained with CD164 mAbs (unbroken lines) or isotype-matched negative control mAbs (dotted Downloaded from lines). CD34 mAbs class I (My10), class II (QBEND-10), and class III (Tu¨k3) were used in parallel to stain KG1a cells, and values are shown in the table with their isotype-matched negative controls. Cells were analyzed for fluorescence staining using a FACScalibur flow cytometer. For

each Ab, the upper histogram represents the stain- http://www.jimmunol.org/ ing of untreated cells; the middle histogram shows the staining of sialidase-treated cells; and the lower histogram shows the staining of O-sialoglycoprotease-treated cells. Values represent the me- dian ﬂuorescence intensity (MFI) Ϯ SD for three individual experiments. by guest on September 28, 2021

on O-linked glycans (see below). To localize the epitopes for these mAbs reacted with both these proteins. However, the N6B6 and mAbs more precisely, the 103B2/9E10, 105A5, N6B6, and 67D2 67D2 mAbs did not (Table I). These results were verified by im- mAbs were analyzed in a solid phase ELISA assay for their reac- munoblotting of the CD164-Fc* series with the CD164 mAbs tivities with the soluble CD164-Fc* domain deletion constructs. (data not shown) and demonstrate that the latter mAbs require the The 103B2/9E10 and 105A5 mAbs recognized all nine soluble exon 3-encoded peptide for epitope recognition, but that they are proteins, indicating that they react with the region encoded by exon unable to identify the exon 3-derived peptide in the absence of the 1 (Table I). N6B6 and 67D2 recognized the CD164(E1–3)-Fc*, exon 2-encoded region. In competitive binding assays, the 103B2/ CD164(E1–4)-Fc*, CD164(E1–5)-Fc*, CD164(E1–6a)-Fc*, and 9E10 and 105A5 mAbs did not significantly compete with one CD164(E⌬5)-Fc* proteins, but not the CD164(E1)-Fc* and another or with the N6B6 or 67D2 mAbs for binding to the CD164(E1–2)-Fc* constructs (Table I), suggesting either that they CD164(E1–3)-Fc*-soluble protein. For example, when the 103B2/ reacted minimally with epitopes encoded by exon 3 or with spe- 9E10 mAb was used to block binding of the 105A5, N6B6, and cific epitopes created by tertiary folding of exons 1–3. To deter- 67D2 mAbs to the CD164(E1–3)-Fc* construct, no inhibition was mine whether the N6B6 and 67D2 mAbs reacted exclusively with observed (0.5 Ϯ 0.4, 1.6 Ϯ 0.4, and 2.4 Ϯ 1.3% inhibition, re- epitopes on exon 3 or with more complex epitopes on multiple spectively). Similarly, the 105A5 mAb did not block the binding of exons, two additional soluble proteins CD164(E1, 3, 4)-Fc* and 103B2/9E10, N6B6, or 67D2 mAbs (0.3 Ϯ 1, 2.5 Ϯ 0.4, and 1.9 Ϯ CD164(E1, 2, 4)-Fc* were generated. These were encoded by ex- 0.5% blocking, respectively). However, N6B6 partially blocked ons 1, 2, and 4 or exons 1, 3, and 4 and linked to the mutated Fc the binding of 67D2 and vice versa (73.8 Ϯ 0.5 and 47.1 Ϯ 0.9% region of human IgG1. As expected, the 103B2/9E10 and 105A5 inhibition, respectively), but did not substantially block that of 848 EPITOPE MAPPING OF CD164 FUNCTIONAL DOMAINS

while 103B2/9E10 and 105A5 epitopes were not affected by DTT even at high concentrations (Fig. 5B). These results indicate that the N6B6 and 67D2 mAbs recognize conformationally dependent epitopes, involving disulfide bond formation between exons 2 and 3 as shown in Fig. 1A. N- and O-linked oligosaccharide side chains are involved in epitope recognition by CD164 mAbs Deglycosylation experiments were conducted to determine whether oligosaccharide residues contribute to the CD164 epitopes recognized by the 103B2/9E10, 105A5, 67D2, and N6B6 mAbs. In the first set of experiments the KG1a cell line was treated with sialidase or O-sialoglycoprotease and analyzed by flow cytometry for CD164 mAb binding (Fig. 6). The CD34 mAbs, My10, QBEND-10, and Tu¨k-3, which recognize epitopes that are differentially sensitive to sialidase and O-sialoglycoprotease, served as controls for this analysis (Fig. 6). In the second set of experiments, CD164(E1–3)-Fc* soluble proteins and CD164 purified from

KG1a cells (CD164(KG1a)) were subjected to N-glycanase, O- Downloaded from glycosidase, sialidase, ␣-fucosidase, and O-sialoglycoprotease treatments, either separately or together, as indicated in Fig. 7. In these latter experiments O-sialoglycoprotease and N-glycanase treatment alone or in combination with the other enzymes dramat- ically decreased the apparent molecular mass of the soluble con-

structs. However, sialidase and O-sialoglycoprotease treatment of http://www.jimmunol.org/ the native CD164(KG1a) molecule reduced its mobility in SDS- PAGE due to the removal of negative charges present on sialic acid (Fig. 7, E–H). After O-glycosidase treatment only partial deglycosylation was observed on CD164(E1–3)-Fc*, since two bands,

the original CD164 and a lower Mr band of 60-kDa, were detected after electrophoresis. Only the 105A5 epitope is C. perfringens sialidase sensitive

FIGURE 7. Enzymatic treatment of the soluble CD164(E1–3)-Fc* pro- Our results show that by treating hemopoietic cell lines, such as by guest on September 28, 2021 tein and the native CD164 protein purified from KG1a cells distinguishes KG1a, which expresses the CD164 epitopes, with C. perfringens the CD164 epitopes. Chimeric CD164(E1–3)-Fc* (A–D) or native CD164- sialidase, we were able to abrogate binding by the 105A5 mAb, but KG1a (E–H) purified proteins were lyophilized and treated with different not with the other CD164 mAbs tested (Fig. 6). The failure of the glycosidases. Proteins were incubated with N-glycanase (N), O-glycosi- 105A5 mAb to immunoblot either the CD164(E1–3)-Fc* soluble ␣ dase (O), sialidase (S), -fucosidase (F), O-sialoglycoprotease (P), O-gly- protein or the CD164(KG1a) protein after treatment with C. per- ␣ cosidase and sialidase (OS), or sialidase and -fucosidase (OSF) as indi- fringens sialidase confirmed that the 105A5 epitope was sialic acid cated above the immunoblots. The untreated proteins (U) were incubated dependent (Fig. 7, B and F). Using the fact that N-glycanase did under the same conditions as the treated molecules. Each sample was electrophoresed on 10% SDS-PAGE gels (0.2 ␮g protein/lane) before immu- not remove the 105A5 epitope from CD164(E1–3)-Fc* or noblotting with CD164 mAbs 103B2/9E10 (A and E), 105A5 (B and F), CD164(KG1a) (Fig. 6, B and F) and that O-glycosidase is unable N6B6 (C and G), and 67D2 (D and H). The molecular mass markers were to digest long chain O-linked glycans without prior sialidase treat- the same as those shown in Fig. 3C. ment, it appears that the sialic acids intrinsic to the 105A5 epitope are most likely situated on O-glycosylated chains attached to the exon 1-encoded peptide and not on N-linked oligosaccharides (Fig. 105A5 or 103B2/9E10 (14.7 Ϯ 0.6 and 1.2 Ϯ 2.5% inhibition for 1B). In contrast to the 105A5 epitope, the 103B2/9E10, N6B6, and N6B6, and 15.1 Ϯ 0.7 and 0% inhibition for 67D2, respectively). 67D2 epitopes were not affected by C. perfringens sialidase treatment of either KG1a cells (Fig. 6), or CD164(KG1a) protein (Fig. The N6B6 and 67D2 epitopes are conformationally dependent 7, E, G, and H) or the soluble CD164(E1–3)-Fc* protein (Fig. 7, In view of the fact that exons 2 and 3 contain all eight cysteine A, C, and D). This was confirmed in an ELISA analysis in which residues that occur in the extracellular domain (Fig. 1B), we con- the four CD164 mAbs were examined for their ability to bind to sidered the possibility that disulfide bridges may strongly influence the native or sialidase-treated soluble CD164(E1–3)-Fc* construct the conformation of CD164 and thus be intrinsic to the epitopes attached to microtiter wells. In these experiments, sialidase treat- recognized by the CD164 mAbs. As shown in Fig. 5B,inthe ment reduced the binding of the 105A5 mAb by 76 Ϯ 1%, but did absence of reducing agents, the soluble CD164(E1–3)Fc* con- not reduce the binding of the other CD164 mAbs (data not shown). struct formed dimers via disulfide linkages at the hinge region of the human IgG1 Fc, while in 5 mM DTT or above, the protein was Both the 105A5 and 103B2/9E10 epitopes are O- monomeric. Treatment of the CD164(E1–3)-Fc* soluble recombi- sialoglycoprotease sensitive nant protein with increasing concentrations of DTT resulted in the Preincubation of KG1a cells with O-sialoglycoprotease signifi- loss of the epitopes recognized by N6B6 and 67D2. Reducing con- cantly reduced the binding of the 103B2/9E10 and 105A5 mAbs as ditions (Ն20 mM DTT) were sufficient to perturb the N6B6 and measured by flow cytometry (Fig. 6). The other CD164 epitopes 67D2 epitope reactivities on the CD164(E1–3)-Fc* construct, were not affected by this treatment. To confirm these studies, the The Journal of Immunology 849 Downloaded from http://www.jimmunol.org/

FIGURE 8. Schematic representation of CD164 epitopes deduced from glycosidase treatments.

CD164(E1–3)-Fc* protein or the CD164(KG1a) protein was The N6B6 and 67D2 mAbs bind deglycosylated CD164 treated with O-sialoglycoprotease, and the resulting protein was Our results demonstrate that the N6B6 and 67D2 epitopes on analyzed in the presence of 5 mM DTT on SDS-PAGE followed CD164(E1–3)-Fc* or on CD164(KG1a) are not removed by the by guest on September 28, 2021 by immunoblotting with the CD164 mAbs. By Coomassie blue deglycosylation procedures used, since N6B6 and 67D2 still rec- analysis, this treatment of the soluble protein reduced its apparent ognize the different deglycosylated forms (Fig. 7, C, D, G, and H). Mr to approximately 65 kDa (data not shown). Neither the 103B2/ Only N-glycanase and O-sialoglycoprotease treatments of the sol- 9E10 nor the 105A5 mAbs bound to this 65-kDa fragment, uble chimeric molecule appear to reach complete deglycosylation, whereas both N6B6 and 67D2 were found to bind (Fig. 7, C and D, since treatment with O-glycosidase only partially removed the O- respectively). These experiments were repeated with CD164(E1–6a)- linked carbohydrates. This is evidenced by the fact that the 80-kDa Fc*, and the same decrease in the apparent molecular mass was ob- molecule is the major band detected by Coomassie blue staining served (data not shown), indicating than the first mucin domain (data not shown), with an additional weaker band at approximately contained the only cleavage site for the O-sialoglycoprotease en- 60 kDa being detected by immunoblotting with the N6B6 and zyme. The sensitivity of CD164 epitopes to O-sialoglycoprotease 67D2 mAbs (Fig. 7, C and D), but not with 103B2/9E10 or 105A5 was similar to that of purified CD164(KG1a) (Fig. 7, E–H). These mAbs (Fig. 7, A and B). This partial digestion with O-glycosidase studies demonstrate the partial removal of the exon 1-encoded re- was improved by the prior addition of exoglycosidases such as gion identified with 105A5 and 103B2/9E10 mAbs from CD164 sialidase or ␣-fucosidase (Fig. 7, C and D, lanes OS and OSF), but on KG1a cells, and its complete removal from the soluble even in the presence of these enzymes, O-glycosidase did not com- CD164(E1–3)-Fc* molecule by O-sialoglycoprotease treatment. pletely digest the original CD164(E1–3)-Fc* protein. Furthermore, they indicate that the region encoded by exon 1 is not essential for epitope recognition by the N6B6 and 67D2 mAbs. Identification of three classes of CD164 epitopes Recognition of the 103B2/9E10 epitope requires N-linked The CD34 mAbs have been classified into three classes based on carbohydrate attachment their sensitivities to sialidase and O-sialoglycoprotease (27). Thus, The 103B2/9E10 epitope (but not the 105A5, N6B6, or 67D2 by comparing CD164 mAbs with the CD34 mAb classes, it has epitopes) was sensitive to N-glycanase treatment either on soluble been possible to subtype the CD164 mAbs into three analogous CD164(E1–3)-Fc* protein or CD164(KG1a) protein (Fig. 7). This categories. Like the CD34 epitope, My10, the CD164 epitope, was confirmed in an ELISA analysis in which the four CD164 mAbs 105A5, is sensitive to both C. perfringens sialidase and O-sialo- were examined for the ability to bind to the native or N-glycanase- glycoprotease treatments and can be classified as a class I epitope treated soluble CD164(E1–3)-Fc* construct attached to microtiter (Fig. 8). The CD164 epitope, 103B2/9E10, is similar to the CD34 wells. In these experiments removal of N-linked carbohydrates re- epitope, QBEND 10, in that it is sensitive to O-sialoglycoprotease, duced the binding of the 103B2/9E10 mAb by 63.1 Ϯ 6.9%, but did but not to C. perfringens sialidase, and can be classified as a class not reduce the binding of the other CD164 mAbs (data not shown). II epitope (Fig. 8). Interestingly, this 103B2/9E10 epitope is also Hence, the 103B2/9E10 epitope is dependent on the N-linked carbo- sensitive to N-glycanase digestion. The CD164 epitopes, N6B6 hydrates of exon 1 (Fig. 1). and 67D2, and the CD34 epitope, Tu¨k3, are insensitive to both 850 EPITOPE MAPPING OF CD164 FUNCTIONAL DOMAINS

C. perfringens sialidase and O-sialoglycoprotease enzymes and hemopoietic progenitor cell differentiation (33). More signifi- can therefore be classified as class III epitopes (Fig. 8). From our cantly, on high endothelial venules, the CD34 isoform displays the results on the differential binding of the N6B6 and 67D2 mAbs to class II and III, but not class I epitopes (34), thereby implicating the Triton X-100-insoluble cell fraction (Fig. 4), we are able to the class II rather than the class I epitopes in the high affinity group the class III epitopes into two subclasses. The subclass IIIA adhesion of these cells to L-selectin on lymphocytes. mAb N6B6 does not bind to the 320-kDa Triton X-100-insoluble The variable glycosylation of CD164 observed here for different material, whereas the subclass IIIB mAb 67D2 reacts with the cell types is by no means uncommon. One feature of many gly- Triton X-100-insoluble material. coproteins is microheterogeneity, which is due at least in part to the attached glycan chains. This heterogeneity is nonrandom and Discussion reproducible for a given protein synthesized by a specific cell type under defined conditions, a feature reflected when the physiolog- Our previous studies have demonstrated that CD164 is expressed ical relevance of protein glycosylation is considered. While it is ϩ on CD34 hemopoietic progenitor cells and bone marrow stroma interesting to examine the differential glycosylation of the CD164 (1–4). Evidence that 103B2/9E10 mAb inhibits the adhesion of molecule in different cell lineages and tissues, examination of the ϩ CD34 cells to bone marrow stroma and that this same mAb or the glycosylation pattern by enzymatic treatment (Fig. 8) provides in- 105A5 mAb prevents recruitment of quiescent hemopoietic pro- sight into the possible functional relevance of post-translational genitor cells into cell cycle (1) has heightened interest in this mol- processing. Glycans can serve as recognition determinants for or as ecule. In this paper we have identified for the first time three modulators of cell-cell, cell-matrix, and protein-(glyco)protein in- classes of epitopes on CD164: the class I 105A5, the class II teractions. They can also be involved in either adhesive or anti- Downloaded from 103B2/9E10, and the class III N6B6 and 67D2 epitopes. We have adhesive interactions. Both roles may be played by the same mol- shown previously that these epitopes are differentially distributed ecule depending on the tissues in which these glycoproteins are on distinct cell types in the adult (4). The variation in epitope expressed and on the type of specific carbohydrate modifications distribution observed may be explained by the experiments con- that have been processed. This is further demonstrated by the wide ducted in this paper. Analyses of mAb binding to splice variants diversity of glycosyltransferases and protein machinery present in and to soluble domain deletion constructs of human CD164 as well a particular cell type, necessary for the production of a molecule http://www.jimmunol.org/ as analyses of the differential sensitivity of CD164 epitopes to with functionally relevant glycosylation (35). In many cases partial sialidase, O-sialoglycoprotease, and N-glycanase digestions indi- occupancy of potential glycosylation sites has correlated effects on cate that both the class I and class II epitopes of CD164 encompass physiological attributes. Particularly striking examples of the com- different oligosaccharide modifications of the first mucin domain plexities of variable glycosylation site occupancy upon the biolog- of CD164. The class I epitope is associated with long chain sia- ical attributes of a protein are illustrated by GM-CSF and CD44. lylated O-linked glycans, while the 103B2/9E10 class II epitope is Human GM-CSF exhibits variable N-linked glycosylation site oc- dependent on both N- and O-linked glycosylation of CD164. The cupancy, which plays an important role in its biological activity. class III epitopes, N6B6 and 67D2, appear to be more dependent

Indeed, it has been demonstrated that there is an inverse correlation by guest on September 28, 2021 on the peptide backbone than on carbohydrate modifications. between biological activity and the extent of N-glycosylation, sug- These latter epitopes are not removed by any of the glycosidases gesting that N-linked glycans down-regulate the bioactivity of the used, but encompass the cysteine-rich domain that is encoded by molecule (36, 37). The 85-kDa isoform of CD44 on hemopoietic exons 2 and 3 of the CD164 gene and that forms a link between mucin domains I and II. Moreover, 67D2 appears to recognize a cells, on the other hand, acts as a ligand for hyaluronan produced Triton X-100-insoluble form of CD164 that could represent a tet- by endothelial cells when it is sulfated on O-linked oligosaccha- ␣ rameric form of CD164, an aggregate formed by interaction with rides in response to TNF- stimulation (38). Other examples of cytoskeletal elements, or a CD164 isoform modified by GAG at- post-translational modification influencing biological activity are tachment as described previously for CD44 (28, 29) and syndecan by no means rare. Modifications involving O-linked oligosaccha- (30). Thus, N6B6 and 67D2 mAbs appear to predominantly iden- ride or tyrosine sulfation on CD34, PCLP, GlyCAM-1, and tify the core CD164 peptide upon which specific oligosaccharide PSGL-1 on high endothelial venules or on specific leukocyte types modifications encompassing the 103B2/9E10 and 105A5 epitopes are responsible for their high affinity specificity for L-selectin in are arrayed and are responsible for both their more restricted patterns vitro (reviewed in Ref. 10). Since the interaction of sialomucins of expression and the functional significance of this molecule. with selectin ligands generally promotes the rapid proadhesive The classification of the CD164 epitopes presented in this paper tethering of leukocytes to endothelia under conditions of flow in is reminiscent of certain structural features of the CD34 molecule. vitro, it has been postulated that these interactions result in tissue- Three classes of epitopes on CD34 have been defined on the basis specific homing and recirculation of lymphocytes to high endothe- of their sensitivities to sialidase and O-sialoglycoprotease treat- lial venules in lymph nodes and mucosal lymphoid tissues and the ments (27, 31, 32). Like CD164, the CD34 class I epitopes are accumulation of leukocytes at sites of inflammation. Despite this, sialidase/O-sialoglycoprotease sensitive, the class II epitopes are controversy still surrounds the ligand specificity of these sialomu- removed by O-sialoglycoprotease, and the class III epitopes are cins and the functional significance of such sialomucin-ligand in- insensitive to digestion by both enzymes. Furthermore, although teractions in vivo (35). For example, L- and E-selectins function as not as dramatic as the differential tissue distribution of the class I ligands for CD34, yet both sialidase/O-sialoglycoprotease-depen- and II epitopes of CD164, there are some reports on the differential dent and O-sialoglycoprotease-independent adhesion of leukocytes expression of CD34 epitope classes. For example, while the three to high endothelial venules have been described (9; reviewed in classes of CD34 epitopes are equally expressed on immature he- Ref. 10). Finally, gene knockout studies in mice indicate that Gly- mopoietic progenitor cells and immature leukemic blasts (AML- CAM-1 and CD34, at least by themselves, are not responsible for M0/1), class I and class II epitopes are less likely to be expressed L-selectin-mediated lymphocyte recruitment into peripheral on more mature progenitors and on the AML-M3 and -M4/5 leu- lymph nodes, although eosinophil recruitment into the lung fol- kemic blasts than class III epitopes. This suggests a more rapid lowing allergen challenge is down-regulated in CD34-deficient down-regulation of the CD34 class I and II epitopes during normal mice (14, 39). The Journal of Immunology 851

Our previous studies have demonstrated that while all the 17. Hu, M. C., and S. L. Chien. 1998. The cytoplasmic domain of stem cell antigen CD164 epitopes discussed here are expressed on the phenotypi- CD34 is essential for cytoadhesion signaling but not sufficient for proliferation signaling. Blood 91:1152. cally most primitive hemopoietic progenitor cells, it is the class II 18. Krause, D. S., M. J. Fackler, C. I. Civin, and W. S. May. 1996 CD34: structure, epitope, 103B2/9E10, that has been shown to elicit a very potent biology and clinical utility. Blood 87:1. regulatory effect on stem cell proliferation/adhesion in in vitro sys- 19. Ardman, B., M. A. Sikorski, and D. E. Staunton. 1992. CD43 interferes with T-lymphocyte adhesion .Proc. Natl. Acad. Sci. USA 89:5001. tems. In this respect, CD164 resembles other sialomucins, such as 20. Zhang, K., D. Baeckstrom, H. Brevinge, and G. C. Hansson. 1997. Comparison PSGL-1, CD34, and CD43, in that interaction of both sialomucins of sialyl-Lewis a-carrying CD43 and MUC1 mucins secreted from a colon car- with specific mAbs regulates cell proliferation. In the case of cinoma cell line for E-selectin binding and inhibition of leukocyte adhesion. Tumor Biol. 18:175. CD43 and PSGL-1, this receptor binding is functional and specific 21. Baum, L. G., M. Pang, N. L. Perillo, T. Wu, A. Delegeane, C. H. Uittenbogaart, to a particular progenitor cell stage of differentiation (11, 13, 16, M. Fukuda, and J. J. Seilhamer. 1995. Human thymic epithelial cells express an 40–43). Whether the biochemical mechanisms regulating cell pro- endogenous lectin, galectin-1, which binds to core 2 O-glycans on thymocytes and T lymphoblastoid cells. J. Exp. Med. 181:877. liferation following the engagement of CD164 receptor on 22. Tsuboi, S., and M. Fukuda. 1997. Branched O-linked oligosaccharides ectopi- ϩ Ϫ CD34 CD38low/ cells are similar to those observed for CD34, cally expressed intransgenic mice reduce primary T-cell immune responses. CD43, and PSGL-1 is as yet unknown (11, 13, 15–18, 40–43). EMBO J. 16:6364. 23. Skubitz, K. M., K. Micklem, and E. van der Schoot. 1995. CD66 and CD67 However, as indicated in this paper and from our current research, cluster workshop report. In Leucocyte Typing V. S. F. Schlossman, L. Boumsell, with the characterization and better understanding of functional W. Gilks, J. M. Harlan, T. Kishimoto, M. Morimoto, J. Ritz, S. Shaw, R. Sil- epitopes on progenitor cells, new tools are now available that will verstein, T. Springer, T. F. Tedder, and R. F. Todd, eds. Oxford University Press, Oxford, U.K., p. 889. allow us to answer such questions. 24. Oritani, K., and P. W. Kincade. 1996. Identification of stromal cell products that

interact with pre-B cells. J. Cell. Biol. 134:771. Downloaded from 25. Teixeira, A. M., J. Fawcett, D. L. Simmons, and S. M. Watt. 1994. The N-domain Acknowledgments of the biliary glycoprotein (BGP) adhesion molecule mediates homotypic bind- We thank Profs. Sir D. J. Weatherall and L. Kanz for their support. ing: domain interactions and epitope analysis of BGPc. Blood 84:211. 26. Watt, S. M., J. Williamson, H. Genevier, J. Fawcett, D. L. Simmons, A. Hatzfeld, S. A. Nesbitt, and D. R. Coombe. 1993. The heparin binding PECAM-1 adhesion References molecule is expressed by CD34ϩ hematopoietic precursor cells with early myeloid and B-lymphoid cell phenotypes. Blood 82:2649. 1. Zannettino, A. C. W., H.-J. Bu¨hring, S. Niutta, S. M. Watt, M. A. Benton, and

27. Sutherland, D. R., K. M. Abdullah, P. Cyopick, and A. Mellors. 1992. Cleavage http://www.jimmunol.org/ P. J. Simmons. 1998. The sialomucin CD164 (MGC-24v) is an adhesive glyco- of the cell-surface O-sialoglycoproteins CD34, CD43, CD44, and CD45 by a protein expressed by human hematopoietic progenitors and bone marrow stromal novel glycoprotease from Pasteurella haemolytica. J. Immunol. 148:1458. cells which serves as a potent negative regulator of hematopoiesis. Blood 28. Perschl, A., J. Lesley, N. English, R. Hyman, and I. S. Trowbridge. 1995. Trans- 92:2613. membrane domain of CD44 is required for its detergent insolubility in fibroblasts. 2. Watt, S. M., H.-J. Bu¨hring, I. Rappold, J. Y.-H. Chan, J. Lee-Prudhoe, T. Jones, J. Cell Sci. 108:1033. A. C. W. Zannettino, P. J. Simmons, D. Sheer, R. Doyonnas, et al. 1998. CD164 ϩ 29. Neame, S. J., C. R. Uff, H. Sheikh, S. C. Wheatley, and C. M. Isacke. 1995. CD44 a novel sialomucin on CD34 and erythroid subsets is located on human chro- exhibits a cell type dependent interaction with Triton X-100 insoluble, lipid rich, mosome 6q21. Blood 92:849. plasma membrane domains. J. Cell Sci. 108:3127. 3. Zannettino, A. C. W., I. Rappold, H.-J. Bu¨hring, S. M. Watt, M. A. Benton, S. Nuitta, and P. J. Simmons. 1997. CD164 (MGC-24v) workshop panel report. 30. Carey, D. J., K. M. Bendt, and R. C. Stahl. 1996. The cytoplasmic domain of In Leucocyte Typing VI. T. Kishimoto, H. Kikutani, A. E. G. Ch. von den Borne, syndecan-1 is required for cytoskeleton association but not detergent insolubility. S. M. Goyert, D. Y. Mason, M. Miyafaka, L. Moretta, K. Okumura, S. Shaw, T. J. Biol. Chem. 271:15253. by guest on September 28, 2021 A. Springer, K. Sugamura, and H. Zola, eds. Garland Publishing, New York, 31. Sutherland, D. R., S. M. Watt, G. Dowden, K. Karhi, M. A. Baker, M. F. Greaves, p. 456. and J. E. Smart. 1988. Structural and partial amino acid sequence analysis of the 4. Watt, S. M., L. H. Butler, M. Tavian, H.-J. Bu¨hring, I. Rappold, P. J. Simmons, human hemopoietic progenitor cell antigen CD34. Leukemia 2:793. A. C. W. Zannettino, D. Buck, A. Fuchs, R. Doyonnas, et al. 2000. Functionally 32. Watt, S. M., K. Karhi, K. Gatter, A. J. W. Furley, F. E. Katz, L. E. Healy, defined CD164 epitopes are expressed on CD34ϩ cells throughout ontogeny but L. J. Altass, N. J. Bradley, D. R. Sutherland, R. Levinsky, et al. 1987. Distribution display distinct distribution patterns in adult hematopoietic and non-hematopoi- and epitope analysis of the cell surface membrane glycoprotein (HPCA-1) asso- etic tissues. Blood 95:3113. ciated with human hemopoietic progenitor cells. Leukemia 1:417. 5. Almeida-Porada, G., H.-J. Bu¨hring, S. M. Watt, P. J. Simmons, G. Rathke, 33. Steen, R., G. E. Tjonnefjord, G. Gaudernack, L. Brinch, and T. Egeland. 1996. S. Scheding, L. Kanz, W. Brugger, and E. D. Zanjani. 1999. CD164 defines an Differences in the distribution of CD34 epitopes on normal haemopoietic pro- immature subset of human bone marrow CD34-negative stem cells. Blood genitor cells and leukaemic blast cells. Br. J. Haematol. 94:579. 94:462a. 34. Baumhueter, S., M. S. Singer, W. Henzel, S. Hemmerich, M. Renz, S. D. Rosen, 6. Varki, A. 1997. Perspectives series: cell adhesion in vascular biology. J. Clin. and L. A. Lasky. 1993. Binding of L-selectin to the vascular sialomucin CD34. Invest. 99:158. Science 262:436. 7. McEver, R. P., and R. D. Cummings. 1997. Perspectives series: cell adhesion in 35. Lowe, J. B. 1997. Selectin ligands, leukocyte trafficking, and fucosyltransferase vascular biology. J. Clin. Invest. 100:485. genes. Kidney Int. 51:1418. 8. Ostberg, J. R., R. K. Barth, and J. G. Frelinger. 1998. The Roman god Janus: a 36. Ding, D. X.-H., J. C. Vera, M. L. Heaney, and D. W. Golde. 1995. N-glycosyl- paradigm for the function of CD43. Immunol. Today 19:546. ation of the human granulocyte-macrophage colony-stimulating factor receptor ␣ 9. Clark, R. A., R. C. Fuhlbrigge, and T. A. Springer. 1998. L-selectin ligands that subunit is essential for ligand binding and signal transduction. J. Biol. Chem. are O-glycoprotease resistant and distinct from MECA-79 antigen are sufficient 270:24580. for tethering and rolling of lymphocytes on human endothelial venules. J. Cell 37. Cebon, J., N. Nicola, M. Ward, I. Gardner, P. Dempsey, J. Layton, U. Duhrsen, Biol. 140:721. A. W. Burgess, E. Nice, and G. Morstyn. 1990. Granulocyte-macrophage colony 10. Gonzalez-Amaro, R., and F. Sanchez-Madrid. 1999. Cell adhesion molecules: stimulating factor from human lymphocytes: the effect of glycosylation on re- selectins and integrins. Crit. Rev. Immunol. 19:389. ceptor binding and biological activity. J. Biol. Chem. 265:4483. 11. Bazil, V., J. Brandt, S. Chen, M. Roeding, K. Luens, A. Tsukamoto, and 38. Maiti, A., G. Maki, and P. Johnson. 1998. TNF-␣ induction of CD44-mediated R. Hoffman. 1996. A monoclonal antibody recognizing CD43 (leukosialin) ini- leukocyte adhesion by sulfation. Science 282:941. tiates apoptosis of human hematopoietic progenitor cells but not stem cells. Blood 87:1272. 39. Suzuki, A., D. P. Adnrew, J.-A. Gonzalo, M. Fukumoto, J. Spellberg, 12. Wiken, M., P. Bjorck, B. Axelsson, and P. Perlmann. 1989. Enhancement of M. Hashiyar, H. Takimoto, N. Gerwin, I. Webb, G. Molineux, et al. 1996. CD34- human B-cell proliferation by a monoclonal antibody to CD43. Scand. J. Immu- deficient mice have reduced eosinophil accumulation after allergen exposure and nol. 29:363. show a novel crossreactive 90-kD protein. Blood 87:3550. 13. Levesque, J. P., A. C. W. Zannettino, M. Pudney, S. Niutta, D. N. Haylock, 40. Moore, T., S. Huang, L. W. Terstappen, M. Bennett, and V. Kumar. 1994. Ex- K. R. Snapp, G. S. Kansas, M. C. Berndt, and P. J. Simmons. 1999. PSGL-1- pression of CD43 on murine and human pluripotent hematopoietic stem cells mediated adhesion of human hematopoietic progenitors to P-selectin results in J. Immunol. 153:4978. suppression of hematopoiesis. Immunity 11:369. 41. Woodman, R. C., B. Johnston, M. J. Hickey, D. Teoh, P. Reinhardt, B. Y. Poon, 14. Cheng, J., S. Baumhueter, G. Cacalano, K. Carver-Moore, H. Thibodeaux, and P. Kubes. 1998. The functional paradox of CD43 in leukocyte recruitment: R. Thomas, H. E. Broxmeyer, S. Cooper, N. Hague, M. Moore, et al. 1996. a study using CD43-deficient mice. J. Exp. Med. 188:2181. Hematopoietic defects in mice lacking the sialomucin CD34. Blood 87:479. 42. Anzai, N., A. Gotoh, H. Shibayama, and H. E. Broxmeyer. 1999. Modulation of 15. Bazil, V., J. Brandt, A. Tsukamoto, and R. Hoffman. 1995. Apoptosis of human integrin function in hematopoietic progenitor cells by CD43 engagement: possi- hematopoietic progenitor cells induced by crosslinking of surface CD43, the ma- ble involvement of protein tyrosine kinase and phospholipase C-␥. Blood 93: jor sialoglycoprotein of leukocytes. Blood 86:502. 3317. 16. Bazil, V., J. E. Brandt, and R. Hoffman. 1997. Resistance of human hematopoi- 43. Rosenstein, Y., A. Santana, and G. Pedraza-Alva. 1999. CD43, a molecule with etic stem cells to a monoclonal antibody recognizing CD43. Stem Cells 15:13. multiple functions. Immunol. Res. 20:89.