of April 10, 2019.This information is current as

-9Immunosuppressive Receptors Siglec-7 andLigands for Mucosal Macrophage Colonic Epithelial Cells Express Specific

Yukio Saito, Taeko Dohi and Reiji KannagiToshiyuki Yamaji, Yasuhiro Hashimoto, Akemi Suzuki,Mineko Izawa, Katsuyuki Ohmori, Motoaki Mitsuki, Keiko Miyazaki, Keiichiro Sakuma, Yuki I. Kawamura,

http://www.jimmunol.org/content/188/9/4690doi: 10.4049/jimmunol.1100605March 2012;

2012; 188:4690-4700; Prepublished online 30J Immunol 

MaterialSupplementary

5.DC1http://www.jimmunol.org/content/suppl/2012/03/30/jimmunol.110060

Referenceshttp://www.jimmunol.org/content/188/9/4690.full#ref-list-1

, 16 of which you can access for free at: cites 37 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists, Inc. All rights reserved.Copyright © 2012 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

The Journal of Immunology

Colonic Epithelial Cells Express Specific Ligands for MucosalMacrophage Immunosuppressive Receptors Siglec-7 and -9

Keiko Miyazaki,* Keiichiro Sakuma,*,† Yuki I. Kawamura,‡ Mineko Izawa,†

Katsuyuki Ohmori,x Motoaki Mitsuki,{ Toshiyuki Yamaji,{ Yasuhiro Hashimoto,{

Akemi Suzuki,{ Yukio Saito,‖ Taeko Dohi,‡ and Reiji Kannagi*,‡,#

Immune cells are known to express specific recognition molecules for cell surface glycans. However, mechanisms involved in

glycan-mediated cell–cell interactions in mucosal immunity have largely been left unaccounted for. We found that several glycans

preferentially expressed in nonmalignant colonic epithelial cells serve as ligands for sialic acid-binding Ig-like lectins (siglecs), the

immunosuppressive carbohydrate-recognition receptors carried by immune cells. The siglec ligand glycans in normal colonic

epithelial cells included disialyl Lewisa, which was found to have binding activity to both siglec-7 and -9, and sialyl 6-sulfo Lewisx,

which exhibited significant binding to siglec-7. Expression of these siglec-7/-9 ligands was impaired upon carcinogenesis, and they

were replaced by cancer-associated glycans sialyl Lewisa and sialyl Lewisx, which have no siglec ligand activity. When we

characterized immune cells expressing siglecs in colonic lamina propriae by flow cytometry and confocal microscopy, the majority

of colonic stromal immune cells expressing siglec-7/-9 turned out to be resident macrophages characterized by low expression of

CD14/CD89 and high expression of CD68/CD163. A minor subpopulation of CD8+ T lymphocytes also expressed siglec-7/-9.

Siglec-7/-9 ligation suppressed LPS-induced cyclooxygenase-2 expression and PGE2 production by macrophages. These results

suggest that normal glycans of epithelial cells exert a suppressive effect on cyclooxygenase-2 expression by resident macrophages,

thus maintaining immunological homeostasis in colonic mucosal membranes. Our results also imply that loss of immunosuppres-

sive glycans by impaired glycosylation during colonic carcinogenesis enhances inflammatory mediator production. The Journal

of Immunology, 2012, 188: 4690–4700.

Immune cells are known to express specific recognitionmolecules for cell surface glycans, such as galectins, sialicacid binding Ig-like lectins (siglecs), and selectins (1–3). Such

recognition molecules are proposed to be involved in cell–cellinteraction. However, mechanisms involved in glycan-mediatedcell–cell interactions in mucosal immunity have largely beenleft unaccounted for. In this study, we looked for glycan ligandsspecific to siglecs in normal colonic epithelial cells.Cell surface glycans undergo drastic changes during malignant

transformation (4–6). Some carbohydrate determinants are pref-

erentially expressed in nonmalignant epithelial cells, whereasother determinants are expressed in association with cancers. Thecarbohydrate determinants associated with cancers, such as sialylLewisa or sialyl Lewisx, are clinically used as tumor markers.We showed previously that a carbohydrate determinant, disialylLewisa, is preferentially expressed in nonmalignant epithelial cellsand that its expression is lost upon malignant transformation (7).Epigenetic silencing of one of the sialyltransferase genes involvedin its synthesis led to the synthesis of a well-known cancer-associated glycan, sialyl Lewisa, instead of disialyl Lewisa, incancer cells (7). We identified the disialyl Lewisa determinantexpressed in nonmalignant epithelial cells as a specific ligand forsiglec-7, an immunosuppressive carbohydrate-recognition mole-cule present in immune cells in our previous study (7).Most siglecs are known to have ITIMs and to inhibit excess

activation of immune cells by recruiting tyrosine phosphatasesSrc homology region 2-containing tyrosine phosphatase (SHP)1and SHP2 (2, 8). Chronic inflammatory stimulation of mucosalmembranes renders colonic epithelial cells increasingly suscep-tible to cancer progression both in humans and mice (9–11). Thisprompted us to screen more extensively for any other carbo-hydrate determinants preferentially expressed in nonmalignantcolonic epithelial cells that might serve as ligands for immu-nosuppressive siglecs. Our working hypothesis in this study isthat the interaction of carbohydrate ligands in nonmalignantepithelial cells with siglecs carried by mucosal immune cells, ifany, may be beneficial for maintaining immunological homeo-stasis of mucosal membranes. The loss of such siglec ligandglycans as a result of abnormal glycosylation during carcino-genesis may be implicated in enhanced expression of cyclo-oxygenase (COX)2 and other inflammatory mediators, whichwould further promote carcinogenesis.

*Department of Molecular Pathology, Aichi Cancer Center, Research Institute,Nagoya 464-8681, Japan; †Research Complex for Medical Frontiers, Aichi MedicalUniversity, Nagakute 480-1195, Japan; ‡Department of Gastroenterology, ResearchInstitute, International Medical Center of Japan, Tokyo 162-8655, Japan; xCentralClinical Laboratory, Kyoto University Hospital, Kyoto 606-8507, Japan; {Glyco-Chain Functions Laboratory, Supra-biomolecular System Group, Frontier ResearchSystem, Riken Institute, Saitama 351-0198, Japan; ‖Division of Coloproctology,Department of Surgery, International Medical Center of Japan, Tokyo 162-8655,Japan; and #Institute of Biomedical Sciences, Academia Sinica, Taipei 115, Taiwan

Received for publication March 4, 2011. Accepted for publication February 24, 2012.

This work was supported in part by grants-in-aid from the Ministry of Education,Culture, Sports, Science and Technology, Japan (21590324 and 23112520); grants-in-aid for the Third-Term Comprehensive Ten-Year Strategy for Cancer Control fromthe Ministry of Health and Welfare, Japan; and a grant from the Uehara MemorialFoundation of Japan.

Address correspondence and reprint requests to Dr. Reiji Kannagi, Research Com-plex for Medical Frontiers, Aichi Medical University, #1-1 Karimata, Yazako, Naga-kute 480-1195, Japan. E-mail address: kannagi-gi@umin.ac.jp

The online version of this article contains supplemental material.

Abbreviations used in this article: CMF-HBSS, calcium- and magnesium-free HBSS;COX, cyclooxygenase; LacNAc, N-acetyllactosamine; LPMC, lamina propria mono-nuclear cell; SHP, Src homology region 2-containing tyrosine phosphatase; siglec,sialic acid binding Ig-like lectin; TPA, 12-O-tetradecanoylphorbol-13-acetate.

Copyright� 2012 by TheAmericanAssociation of Immunologists, Inc. 0022-1767/12/$16.00

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1100605

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

For this purpose, we screened ligand activities of two sets ofcarbohydrate determinants, which show unique alterations uponmalignant transformation. One pair was the set of disialyl Lewisa,preferentially expressed in nonmalignant epithelial cells, andsialyl Lewisa, a well-known cancer-associated determinant (7).The other pair was a combination of sialyl 6-sulfo Lewisx, pref-erentially expressed in nonmalignant epithelial cells, and a cancer-associated determinant, sialyl Lewisx (12). We also tried tocharacterize immune cells expressing siglecs in colonic mucosalmembranes and examined the effects of siglec ligation on theirfunctions.

Materials and MethodsCells and Abs

Anti-disialyl Lewisa (FH7, murine IgG3) and anti-disialyl Lewisc (FH9,murine IgG2a) were prepared as described previously (13). The anti-sialylLewisa Ab N19-9 (murine IgG1) was obtained from Alexis Biochemicals(Lausen, Switzerland). The Abs G152 (murine IgM), directed to sialyl 6-sulfo Lewisx, and G72 (murine IgM), reactive to both sialyl 6-sulfo Lac-NAc and sialyl 6-sulfo Lewisx, were prepared as described previously (14).An Ab directed to nonsulfated sialyl Lewisx, CSLEX-1 (murine IgM), wasobtained from American Type Culture Collection (Manassas, VA). Thestable ST6GalNAc6 transfectant cells were established by the transfectionof cDNA for human ST6GalNAc6 to human colon cancer SW1083 cells,as described previously (7). The ECV304 cells stably transfected withFUT7 or GlcNAc6ST-1 cDNA, or cotransfected with both genes, wereprepared as described previously (15).

ELISA for binding of recombinant siglecs to syntheticcarbohydrate determinants

In some experiments, the binding specificity of recombinant siglec-7–Fc orsiglec-9–Fc was ascertained by ELISA using pure carbohydrate determi-nants. The synthetic pure carbohydrate determinants were prepared asdescribed previously (7) and kindly supplied by Professor Makoto Kiso(Gifu University, Gifu, Japan). Recombinant siglec-7–Fc, mutant siglec-7–Fc (R124K), and siglec-9–Fc were prepared as described previously (16).ELISA was performed using synthetic carbohydrate determinants immo-bilized at the bottom of 96-well culture plates by a standard method de-scribed previously (7, 14, 17). GD3 ganglioside (Avanti Polar Lipids) wasused as a positive control for the binding of siglec-7–Fc in ELISA assays.Recombinant siglecs were preincubated with affinity-purified biotinylatedrabbit anti-human IgG (Dako, Glostrup, Denmark), followed by incubationwith peroxidase-streptavidin (Dako), before application to the assay plates.Recombinant mutant siglec-7–Fc lacking sialic acid-binding activity (7)served as a negative control.

Cell-binding assays for recombinant siglecs

Binding of recombinant siglecs to SW1083, ECV304, and their transfectantcells was ascertained by flow cytometric analyses, as described previously(18, 19). For this, recombinant siglecs were preincubated with affinity-purified biotinylated rabbit anti-human IgG, followed by incubation withPE-streptavidin, before application to the staining of cells. Cells werepreincubated with blocking anti-carbohydrate Abs (20 mg/ml) for 30 minat 37˚C, when indicated. Percentage inhibition was calculated from thedecrease in the mean fluorescence intensity of the cells after subtractingthe background fluorescence obtained with negative-control cells. Bindingof cells to immobilized siglec-7 or -9 on 24-well plates was evaluated insome experiments. Recombinant siglecs (2 mg/ml in Tris-HCl buffer [pH9.4]) were immobilized at the bottom of 24-well plates overnight, and thewells were blocked with 2% BSA in PBS. 2979-bis-(carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester-labeled ECV304 cells and theirstable transfectant cells (1 3 106 cells/0.5 ml/well) were added, and theplate was placed on a rotating platform for incubation under shear (90 rpm)for 20 min at room temperature. After nonadherent cells were washed outthree times with PBS, the cells were lysed with 0.5% Nonidet P-40, and theattached cells were counted by measuring fluorescence intensity using anArvo 1420 multilabel counter (Wallac, Gaithersburg, MD).

Cell–cell adhesion experiments mediated by siglecs

Monolayer cell-adhesion experiments were performed as described pre-viously (7, 15, 20). SW1083, ECV304, and their transfectant cells werecultured in monolayer in 24-well plates. The cells were preincubated withblocking anti-carbohydrate Abs (20 mg/ml) for 30 min at 37˚C, when in-

dicated. The U937/siglec-7 or U937/siglec-9 cells (1 3 106 cells/0.5 ml/well), which were prepared and labeled with 2979-bis-(carboxyethyl)-5-(and-6)-carboxyfluorescein acetoxymethyl ester, as described previously(7, 21), were added. The parental U937 cells were used to evaluatebackground binding. The plate was then placed on a rotating platform forincubation under shear (90 rpm) for 20 min at room temperature. Afterthree washings, the adherent cells were lysed with 0.5% Nonidet P-40 andcounted with the Arvo 1420 multilabel counter (Wallac).

Immunohistochemical and flow cytometric analyses ofsiglec-expressing cells

Frozen sections, of 10-mm thickness for immunohistochemical examina-tion, were prepared from surgical specimens obtained from 23 patientswith colorectal cancer (11 originating in colon and 12 originating in rec-tum) at the Aichi Cancer Center Hospital and the Hospital of the Inter-national Medical Center of Japan after informed consent was obtainedfrom all patients. Nine female and 14 male patients were included, with anaverage age of 58.3 y. The stage of the patients varied from Dukes A to D;22 cases were histologically diagnosed as moderately differentiated ade-nocarcinoma, and 1 case was diagnosed as poorly differentiated adeno-carcinoma. No cases of mucinous adenocarcinoma were included. Theavidin–biotin complex technique was used, as described in the instructionsfor the kits (VECTASTAIN; Vector Laboratories, Burlingame, CA), for theimmunohistochemical study (12). For immunohistochemical analyses us-ing recombinant siglec-7–Fc or siglec-9–Fc, the immune complex ofrecombinant proteins, or normal human IgG (R&D Systems, Minneapolis,MN) as negative control, with biotinylated anti-human IgG Ab andperoxidase-streptavidin was prepared before application to the staining offrozen sections. Polyclonal rabbit anti–siglec-7 or -9 Ab [rabbit IgG, raisedagainst recombinant siglec-7 or -9 and affinity purified (7)] was used asa primary Ab for confocal microscopy.

For preparation of lamina propria mononuclear cells (LPMCs) for flowcytometry, fresh colonic tissue specimens were washed twice in calcium-and magnesium-free HBSS (CMF-HBSS) supplemented with penicillin (10U/ml), streptomycin (10 mg/ml), gentamicin (100 mg/ml), and amphoter-icin B (1 mg/ml). The 2 3 2-cm pieces of tissue were incubated for 15 minat room temperature in CMF-HBSS supplemented with antibiotics and1 mM DTT and washed in CMF-HBSS three times. After two similarincubations for 20 min at 37˚C in CMF-HBSS supplemented with anti-biotics, 0.75 mM EDTA, and 1 mM DTT, the tissues were minced andincubated for 15 min at 37˚C in HBSS supplemented with antibiotics, 0.1%BSA, 0.5 mg/ml collagenase, 0.4 mg/ml Dispase, 0.01 mg/ml DNase, and10 mM HEPES. Cells in supernatants were collected and filtered througha 200-mm steel mesh. The cells in suspension were washed twice withCMF-HBSS with antibiotics and centrifuged over a layer of Percoll (100/60/40/30%) at 400 3 g for 25 min at 20˚C. The harvested cells werewashed twice and subjected to flow cytometric analyses. See supplementalmaterials for Abs used for cell surface marker analysis. The unpairedStudent t test was used to compare results of flow cytometric analyses.

Effect of siglecs on LPS-induced production of COX2 andPGE2

The U937 cells transfected with cDNA for siglec-7 (U937/siglec-7 cells) orfor siglec-9 (U937/siglec-9 cells), prepared as described previously (7, 21),were cultured for 3 d with 10 ng/ml 12-O-tetradecanoylphorbol-13-acetate(TPA); after being cultured for one additional day without TPA, they werechallenged with 1 mg/ml LPS for 3 h (22–24). The F(ab9)2 fragment (10mg/ml) of anti–siglec-7 Ab or anti–siglec-9 Ab was added to the culturetogether with TPA or LPS. The F(ab9)2 fragment was prepared by incu-bating neutralizing monoclonal anti–siglec-7 Ab (13- three dimensional,rat IgG2b), or anti–siglec-9 Ab (1-3-A, rat IgG2b) with pepsin beads(Pierce, Rockford, IL) at pH 4.5 for 4–6 h, followed by purification usinganti-rat IgG (Fc-specific) Ab-immobilized agarose beads, as describedpreviously (21). Control rat F(ab9)2 fragment was obtained from JacksonImmunoResearch (West Grove, PA). Cells were collected, and total cel-lular RNA was isolated according to the acid guanidinium thiocyanate-chloroform extraction method using an Isogen kit (Nippon-Gene, Tokyo,Japan). First-strand cDNA was prepared using 300 ng total cellular RNA.Synthesis of cDNA was carried out in a 20-ml reaction volume using theSuperscript Preamplification System (Life Technologies BRL, GrandIsland, NY), according to the manufacturer’s protocol, with oligo-d(T) asinitiation primer. The primers used in RT-PCR analysis for human COX1were 59-TGCCCAGCTCCTGGCCCGCCGCTT-39 and 59-GTGCATCA-ACACAGGCGCCTCTTC-39, which give a 331-bp product; those forCOX2 were 59-TTCAAATGAGATTGTGGGAAAATTGCT-39 and 59-AGATCATCTCTGCCTGAGTATCTT-39, which give a 305-bp product.The primers for G3PDH were 59-AAGGTCATCCATGACAAC-39 and 59-

The Journal of Immunology 4691

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

CACCCTGTTGCTGTAGCCA-39, which give a 489-bp product. Quanti-tative real-time RT-PCR analyses for COX2 were performed using ABIPRISM 7000 (Applied Biosystems), as described previously (7). TheTaqMan probe ID for COX2 was Hs 00153133_m1 (Assays-on-Demand),and TaqMan Endogenous Control (Applied Biosystems) was used forG3PDH. The results were calculated using the comparative Ct method.Relative transcripts were determined by the equation 22DCt. The DCt valuewas determined by subtracting the average GAPDH Ct value from theaverage target Ct value, and it was used to calculate percentage inhibitionof COX2 induction in the LPS-stimulated U937 cells treated with F(ab9)2fragments of anti–siglec-7/-9 Abs. PGE2 in culture supernatant of the cellswas evaluated using a Parameter PGE2 assay kit (KGE004; R&D Sys-tems). The culture supernatants were collected for PGE2 determination2 d after the addition of LPS for the determination of PGE2 production.

ResultsDisialyl Lewisa as a ligand for siglec-7/-9

Normal colonic epithelial cells expressed disialyl Lewisa, and theexpression tended to be decreased in colonic cancer cells. Fig. 1Ashows its carbohydrate structure and a representative pattern of itsexpression in colonic cancer tissues. The loss of disialyl Lewisa

determinant is accompanied by the appearance of sialyl Lewisa

in cancer cells. We previously described preferential expression ofdisialyl Lewisa in nonmalignant colonic epithelial cells comparedwith cancer cells (p , 0.05, n = 6) (25). This was ascertained witha new series of 23 patients in the current study (p = 0.004) (Sup-plemental Fig. 1), and Fig. 1A shows a typical example. Sialyl Lewisa

has a carbohydrate structure quite similar to that of disialyl Lewisa; itlacks one of the two sialic acid residues present in disialyl Lewisa,and this alteration was proposed to be the result of cancer-associatedepigenetic silencing of a gene for an a2,6 sialyltransferase, which isinvolved in the synthesis of disialyl Lewisa (7).We previously showed that the disialyl Lewisa determinant

serves as a ligand for an immunosuppressive receptor, siglec-7 (7).In this study, we show that the determinant also serves as a ligandfor another immunosuppressive receptor, siglec-9. The disialylLewisa was found to bind to the recombinant Fc chimera of siglec-9 or siglec-7 in ELISA-based assays using pure synthetic carbo-hydrate determinants (Fig. 1B). Both siglec-9 and siglec-7 alsobound to the pure disialyl Lewisc, the carbohydrate determinantlacking the fucose residue, indicating that the fucose residue is notessential for the binding of siglec-7/-9. See Supplemental Fig. 2Afor the structure of disialyl Lewisc.We next tried to ascertain whether the binding of siglecs to

disialyl Lewisa actually occurred at the cellular level. To see thebinding at the cellular level, we generated cells expressing disialylLewisa by transfecting a cancer cell line SW1083 with the genefor a2,6 sialyltransferase that synthesizes disialyl Lewisa andscreened for the binding of siglecs to the transfectant cells, first byflow cytometry using several recombinant siglec molecules. Theresults indicated that among the four siglecs of the CD33 family,only siglec-7 and -9 showed binding to the transfectant cells (Fig.1C). None of the siglecs bound to the parental cancer cells, whichexpress sialyl Lewisa, the cancer-associated glycan, but no disialylLewisa (see Supplemental Fig. 2B for expression of disialylLewisa-related glycans in these cells). The binding of siglec-9 tothe transfectant cells was inhibited significantly by anti-disialylLewisa (FH7) or anti-disialyl Lewisc (FH9) Ab; it was almostcompletely inhibited by adding both Abs but was not affected bythe anti-sialyl Lewisa Ab (N19-9) (Fig. 1C).

Cell–cell interaction mediated by disialyl Lewisa andsiglec-7/-9

Interactions of siglecs and their ligands were also found to mediatecell-to-cell adhesion (Fig. 1D). The U937 cells transfected withsiglec-7 or -9 adhered to the cells expressing disialyl Lewisa; this

binding was specifically inhibited by FH7 or more strongly byboth FH7 and FH9 Abs, but it was not affected by the anti-sialylLewisa Ab (N19-9) (Fig. 1D). These results indicated that disialylLewisa, the carbohydrate determinant preferentially expressed innonmalignant colonic epithelial cells, specifically serves as a li-gand for siglec-7 and -9, whereas sialyl Lewisa, the closely relateddeterminant on cancer cells, does not.

Sialyl 6-sulfo Lewisx as a ligand for siglec-7/-9

There is another set of glycans that shows a similar distributionpattern between benign and malignant colonic epithelial cells;sialyl 6-sulfo Lewisx and sialyl Lewisx. Fig. 2A shows the car-bohydrate structure of sialyl 6-sulfo Lewisx, as well as a repre-sentative pattern of its expression in colonic cancer tissues. Thisdeterminant is preferentially expressed on normal colonic epi-thelial cells, and its expression is markedly reduced in coloniccancer cells. We previously reported a statistically significantdecrease in sialyl 6-sulfo Lewisx in colonic cancers, which wasaccompanied by an increase in sialyl Lewisx expression(p = 0.007, n = 23) (12). Sialyl Lewisx is closely related struc-turally to sialyl 6-sulfo Lewisx, and it lacks a sulfate residuecompared with sialyl 6-sulfo Lewisx. Impaired sulfation uponmalignant transformation was proposed to be responsible for thisshift in glycan expression (12).The ELISA-based binding assays of siglecs to the pure sialyl

6-sulfo Lewisx determinant were performed by a consortium forfunctional glycomics; the results are available at http://www.functionalglycomics.org. Siglec-7 was found to be reactive toseveral glycans having diverse carbohydrate structures, includingsialyl 6-sulfo Lewisx, which showed the seventh-best binding,Siglec-9 was shown to be the most reactive to sialyl 6-sulfoLewisx among .200 glycans tested, according to the data postedon the consortium Web site on September 15, 2005. However, thereactivity in ELISA does not always warrant reactivity at thecellular level. Therefore, we tested whether the interaction ofsialyl 6-sulfo Lewisx with siglecs is functional at the cellular levelby generating cells expressing sialyl 6-sulfo Lewisx and controlcells expressing the counterpart glycan, sialyl Lewisx.Our initial screening in flow cytometric analyses of the cells

expressing sialyl 6-sulfo Lewisx using recombinant siglecs indi-cated a strong binding of siglec-7, whereas the binding of siglec-9was negligible (Fig. 2B). The latter finding was at variance with theconsortium results, in which siglec-9 was shown to have a muchbetter binding activity than siglec-7 to sialyl 6-sulfo Lewisx. In thecell-binding analysis using immobilized recombinant siglecs(Fig. 2C), again the strong binding to siglec-7 was detected withthe cells expressing sialyl 6-sulfo Lewisx, which were obtained bycotransfection of FUT7 and GlcNAc6ST-1 cDNAs. Cells trans-fected with FUT7 cDNA only, which expressed sialyl Lewisx butnot sialyl 6-sulfo Lewisx, showed no binding. Unexpectedly, cellstransfected with GlcNAc6ST-1 cDNA alone, which expressedsialyl 6-sulfo N-acetyllactosamine (LacNAc) but not sialyl 6-sulfoLewisx, showed strong binding comparable to that of the cellscotransfected with FUT7 and GlcNAc6ST-1 cDNAs and expressedboth determinants (Fig. 2C; see Supplemental Fig. 2C for glycanexpression of the transfectant cells used in these experiments).Because sialyl 6-sulfo LacNAc lacks the fucose residue, thissuggested that the fucose residue in sialyl 6-sulfo Lewisx is notessential for binding of siglec-7.

Cell–cell interaction mediated by sialyl 6-sulfo Lewisx andsiglec-7/-9

In cell–cell adhesion experiments, the U937/siglec-7 cells stronglyadhered to cells expressing only sialyl 6-sulfo LacNAc (GlcNAc6ST-

4692 IMPAIRED IMMUNOSUPPRESSIVE GLYCAN EXPRESSION IN COLON CANCER

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

FIGURE 1. Siglec ligand activity of disialyl Lewisa glycan, which is preferentially expressed on nonmalignant colonic epithelial cells. (A) Structure of disialyl

Lewisa (upper left panel) and its preferential expression in nonmalignant colonic epithelial cells (upper right panel). The structure of sialyl Lewisa determinant and

its close association with cancer cells in a consecutive section are also shown (lower panels). Disialyl Lewisa determinant was detected by FH7 and sialyl Lewisa

was detected by N19-9 Abs in immunohistochemistry. Scale bar, 100 mm. (B) Specific binding of recombinant siglec-7/-9 to pure carbohydrate determinants.

Binding of recombinant siglec-7/-9-Fc was evaluated by ELISA using immobilized pure synthetic glycolipids bearing disialyl Lewisa and disialyl Lewisc structure

at the bottom of 96-well plates. GD3, the known glycolipid ligand for siglec-7, was used as a positive control for siglec-7. The recombinant siglec-7 mutant-Fc

lacking binding activity served as control. (C) Flow cytometric analyses of binding of recombinant siglecs to cells expressing disialyl Lewisa (left panel) and its

inhibition by specific Abs (right panel). Parental cancer cells SW1083 (parent) expressed sialyl Lewisa but not the disialyl Lewisa determinant. The cells

transfected with ST6GalNAc6 cDNA (ST6-Transfectant) strongly expressed disialyl Lewisa and had severely attenuated expression of sialyl Lewisa, the cancer-

associated determinant (Supplemental Fig. 2B). The recombinant siglec-3–Fc and siglec-5–Fc served as negative controls for rIg-Fc molecules instead of siglec-7

mutant-Fc in flow cytometric analyses, because these siglecs do not bind to these cells. Results of flow cytometric analyses of inhibition of recombinant siglec-7/-9

binding to cells expressing disialyl Lewisa determinant by specific Abs (right panel). The percentage binding was calculated from mean fluorescence intensities

obtained by flow cytometric analyses. The inhibitor Abs used included clone 1-3-A (anti–siglec-9), 13-three dimensional (anti–siglec-7), N19-9 (anti-sialyl Lewisa),

FH7 (anti-disialyl Lewisa), and FH9 (anti-disialyl Lewisc). (D) Cell–cell adhesion mediated by siglec-7 or -9 and its specific glycan ligands. Adhesion of U937

cells transfected with siglec-7 (U937/siglec-7) or siglec-9 cDNA (U937/siglec-9) to the parental SW1083 cells (Parent) or ST6GalNAc6-transfectant cells was

evaluated by nonstatic cell-adhesion assays. Significant adhesion, observedwith the ST6GalNAc6-transfectant cells, was tested for possible inhibition with N19-9,

FH7, or FH7 + FH9Abs. Values aremean6SD from triplicate experiments. The results in (B) and (D) are representative of at least three independent experiments,

and those in (C) are representative of two independent experiments. Ca, Cancer cells; N, nonmalignant epithelial cells.

The Journal of Immunology 4693

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

1 transfectant), as well as to cells expressing both sialyl 6-sulfoLacNAc and sialyl 6-sulfo Lewisx (the cells cotransfected withFUT7 and GlcNAc6ST-1 cDNAs) (Fig. 2D). Adhesion of U937/siglec-7 cells to the cells expressing both sialyl 6-sulfo LacNAc andsialyl 6-sulfo Lewisx was partially inhibited by G152 Ab (Fig. 2D),which is reactive to sialyl 6-sulfo Lewisx, and strongly inhibited byG72 Ab, which is reactive to both sialyl 6-sulfo LacNAc and sialyl6-sulfo Lewisx (14). These results indicated that siglec-7 interactedwith both sialyl 6-sulfo LacNAc and sialyl 6-sulfo Lewisx. U937/siglec-7 cells also strongly adhered to the cells expressing onlysialyl 6-sulfo LacNAc but not sialyl 6-sulfo Lewisx, and this ad-hesion was almost completely inhibited by G72 Ab, which is re-active to sialyl 6-sulfo LacNAc (Fig. 2D). These results confirmedthat sialyl 6-sulfo LacNAc serves as a ligand for siglec-7 as ef-fective as sialyl 6-sulfo Lewisx, and the fucose residue is not es-sential for siglec-7 binding. No inhibition was observed with anti-sialyl Lewisx Ab (CSLEX-1) in either case.Interaction of siglec-9 with cells expressing sialyl Lewisx or

sialyl LacNAc was very weak in cell-binding assays usingimmobilized recombinant siglec-9 (Fig. 2C) and was not statisti-

FIGURE 2. Siglec ligand activity of sialyl 6-sulfo Lewisx glycan, which

is preferentially expressed on nonmalignant colonic epithelial cells. (A)

Structure of sialyl 6-sulfo Lewisx determinant (upper left panel) and its

preferential expression in nonmalignant colonic epithelial cells (upper

right panel). The structure of sialyl Lewisx determinant and its close as-

sociation with cancer cells in a consecutive section are also shown (lower

panels). Sialyl 6-sulfo Lewisx determinant was detected by G152 and sialyl

Lewisx was detected by CSLEX-1 Abs in immunohistochemistry. Scale

bar, 100 mm. (B) Flow cytometric analyses of binding of recombinant

siglecs to cells expressing sialyl 6-sulfo Lewisx. ECV304 cells transfected

with FUT7 cDNA (FUT7 Transfectant) expressed sialyl Lewisx but not the

sialyl 6-sulfo Lewisx determinant. Cells cotransfected with FUT7 and

GlcNAc6ST-1 cDNAs (FUT7+6ST Transfectant) strongly expressed sialyl

6-sulfo Lewisx and had severely attenuated expression of sialyl Lewisx, the

cancer-associated determinant (Supplemental Fig. 2C). The recombinant

siglec-2–Fc served as a negative control instead of siglec-7 mutant-Fc in

flow cytometric analyses, because this siglec does not bind to these cells.

(C) Adhesion of cells expressing sialyl 6-sulfo Lewisx to immobilized

siglec-7 or -9. Recombinant siglec-7 or -9 was immobilized at the bottom

of 24-well plates, and adhesion of ECV304 cells transfected with FUT7

(FUT7 transfectant), GlcNAc6ST-1 (6ST transfectant), or cotransfected

with FUT7 and GlcNAc6ST-1 cDNA (FUT7+6ST transfectant) was tested

in nonstatic cell-binding assays. In these assays, siglec-7 mutant-Fc was

used as a negative control. (D) Cell–cell adhesion mediated by siglec-7 or

-9 and its specific glycan ligands. Adhesion of U937 cells transfected with

siglec-7 (U937/Siglec-7) or siglec-9 cDNA (U937/Siglec-9) to the ECV304

cells transfected with FUT7 (FUT7 transfectant), GlcNAc6ST-1 (6ST trans-

fectant), or cotransfected with FUT7 and GlcNAc6ST-1 cDNA (FUT7+6ST

transfectant) was evaluated by nonstatic cell-adhesion assays. Significant

adhesion, observed with the 6ST transfectant cells and FUT7+6ST trans-

fectant cells, was tested for possible inhibition with anti-glycan Abs

CSLEX-1, G72, G152, or with the combination of G72 and G152 Abs.

Values are mean 6 SD from triplicate experiments in (C) and (D). The

results in (B) and (C) are representative of at least three independent

experiments, and those in (D) are representative of two independent

experiments. Ca, Cancer cells; N, nonmalignant epithelial cells.

FIGURE 3. Immunohistochemical analyses of siglec-7/-9 ligand gly-

cans in nonmalignant epithelial cells and cancer cells in the colon using

recombinant siglec-Fc. Results of immunohistochemical staining using

recombinant siglec-7–Fc (upper panels), siglec-9–Fc (middle panels), or

normal human IgG (lower panel) are shown. The results indicate that both

siglec-7–Fc and siglec-9–Fc preferentially bind to nonmalignant epithelial

cells (labeled “N”), which express cognate ligands, such as disialyl Lewisa

and sialyl 6-sulfo Lewisx), compared with cancer cells (labeled “Ca”). The

results are representative of the immunohistochemical staining of five

colonic tissue samples. Sialidase A from Arthrobacter ureafaciens for the

neuraminidase treatment of sections was obtained from Nacalai Tesque

(Kyoto, Japan). Scale bar, 200 mm.

4694 IMPAIRED IMMUNOSUPPRESSIVE GLYCAN EXPRESSION IN COLON CANCER

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

cally significant in cell–cell adhesion assays using U937/siglec-9cells (Fig. 2D). Other cell lines strongly expressing sialyl 6-sulfoLewisx, such as a subclone of Namalwa cells (18) and SW480cells transfected with 6-sulfotransferase cDNA (26), also failed toshow statistically significant binding to siglec-9 (data not shown).These results indicated that the sialyl 6-sulfo Lewisx determi-

nant, which is specifically expressed in nonmalignant colonicepithelial cells, served as a ligand for siglec-7 at the cellular level.Another glycan, sialyl 6-sulfo LacNAc, which was previously

shown to also be preferentially expressed in nonmalignant epi-thelial cells (12), was found to serve as a ligand for siglec-7. Thepossibility that they serve as ligands for siglec-9 cannot be com-pletely excluded, but their ligand activity to siglec-9 at the cellularlevel seemed to be much weaker than expected from the results ofthe ELISA-based assays using pure glycans. Interestingly, sialylLewisx, which is another glycan that is structurally similar and ispreferentially expressed in cancer cells, failed to interact with anysiglecs.

FIGURE 4. Siglec-7/-9–expressing cells in colonic mucosal membranes. (A) Confocal microscopic observation of cells expressing siglec-7 (upper panel)

or siglec-9 (lower panel) in colonic mucosal membranes. Frozen sections of normal colonic tissues were stained with anti–siglec-7 (upper panel) or with

anti–siglec-9 (lower panel) in green. A cognate ligand for siglec-7/-9, disialyl Lewisa, is stained in red using FH7 Ab. Original magnification 3200. (B)

Morphology and distribution of siglec-7/-9–expressing cells in colonic lamina propria. The frozen sections were stained with anti–siglec-7 Ab in green and

with anti–siglec-9 Ab in red. Arrowheads indicate siglec-9+ cells. Areas of normal colonic epithelial cells are shown by dotted white line and labeled “E” in

the merged panel. Original magnification3100. (C) Flow cytometric analyses of siglec-7/-9+ cells in LPMCs. Frequency of siglec-7/-9+ cells in CD33+ and

CD332 fractions in LPMCs, as detected by flow cytometry, and average percentage obtained from six samples (upper panel). Example of flow cytometric

analysis of CD33+ LPMCs prepared from normal human colon (lower panel). (D) Summary of cell lineage-specific marker analyses of siglec-7/-9–

expressing cells in colonic lamina propriae. Average percentage obtained from five samples is shown. CD3 (UCHT1, IgG1), CD4 (RPA-T4, IgG1), CD8

(DK25, IgG1), CD20 (L26, IgG2a), CD56 (B159, IgG1), CD14 (TUK4, IgG2a), and CD68 (PG-M1, IgG3) were used to characterize the leukocyte

subpopulation in immunohistochemistry. The markers used for evaluation of cell number in each leukocyte subpopulation were CD56 for NK cells, CD8 for

killer T cells, and CD68 for monocytes/macrophages. Abs used for leukocytes in peripheral blood are shown in Supplemental Fig. 3. (E) Representative

flow cytometric analyses of siglec-7 expression in CD8aa+b7-integrin+ cells in PBLs. Two-dimensional distribution of CD8 and siglec-7 in gated CD3+

cells (left panel, three-color analysis, 2.4% of CD3+CD8+ cells express siglec-7), two-dimensional distribution of CD8a and siglec-7 in gated CD3+CD8b2

cells (middle panel, four-color analysis, 27.3% of CD3+CD8a+CD8b2 cells express siglec-7), and two-dimensional distribution of b7-integrin and siglec-7

in gated CD3+ CD8+cells (right panel, four-color analysis, 4.2% of CD3+CD8+b7-integrin+ cells express siglec-7). See Table I for statistical analyses.

The Journal of Immunology 4695

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

When the binding of recombinant siglec-7 and -9 was tested byimmunohistochemistry using frozen sections, the results indicatedthat both siglecs preferentially bind to nonmalignant colonic epi-thelial cells but only weakly to cancer cells (Fig. 3). This confirmedthe preferential expression of the ligand glycans for siglec-7 and-9, such as disialyl Lewisa and sialyl 6-sulfo Lewisx, in nonma-lignant epithelial cells, and its decrease or loss in cancer cells.

Identification of siglec-7/-9–expressing cells in colonicmucosal membranes

To determine the physiological significance of siglec-7/-9 ligandexpression in colonic epithelial cells, we examined the charac-teristics of siglec-7/-9–expressing cells in colonic tissues. Colonicmucosal membranes were found to contain a significant numberof siglec-7/-9–expressing cells, some of which were localizedin lamina propria very close to colonic epithelial cells expressingtheir cognate ligands, suggesting the occurrence of possible epi-thelial–mesenchymal interactions (Fig. 4A). The majority of thecells expressing siglec-7 in colonic lamina propriae had a den-dritic appearance with several spines. Relatively few siglec-9–expressing cells were observed, and they tended to be roundshaped with a less dendritic appearance (Fig. 4B). Flow cytometryof LPMCs indicated that most siglec-7/-9–expressing cells wereCD33+, and there were more siglec-7-expressing cells than siglec-9–expressing cells (Fig. 4C). Siglec-7+, siglec-9+, and siglec-7+

siglec-9+ cells made up 11.3%, 5.6%, and 2.8%, respectively, ofthe CD33+ cells in the LPMC fraction. Further marker analyses byflow cytometry were hindered because of extensive application of

proteases during the course of LPMC preparation; because wenoticed that the recovery of siglec-7+ cells was not as good asexpected, we tried to identify siglec-7/-9–expressing cells in co-lonic mucosal membranes by direct confocal microscopic obser-vation of the frozen sections.Results of immunohistochemical staining using cell lineage-

specific markers indicated that most siglec-7–expressing cellsbelonged to the monocyte/macrophage lineage (76%), with asmaller number of CD8+ T lymphocytes (19%). NK cells madeup ,5% of the total siglec-7+ cells (Fig. 4D). This was in clearcontrast to the siglec-7 distribution in peripheral blood leukocytesof healthy individuals, where the majority (75%) of siglec-7+ cellswere NK cells, followed by CD8+ T lymphocytes (23%). Mostsiglec-9–expressing cells in the colon also belonged to themonocyte/macrophage lineage (64%), with a smaller numberof CD8+ T lymphocytes (8%) and other cells (25%, mostly gran-ulocytes). NK cells accounted for ,4% of total siglec-9+ cells(Fig. 4D). Again, this was in clear contrast to the distributionof siglec-9–expressing cells in peripheral blood leukocytes ofhealthy individuals, where granulocytes were predominant and ac-counted for 58% of siglec-9+ cells, followed by NK cells (27%).Closer scrutiny of peripheral lymphocytes indicated that CD8+

cells contained significantly more siglec-7+ cells than did CD4+

cells (p , 0.01), and CD8aa+ cells contained more siglec-7+ cells(p , 0.02) (Fig. 4E) and siglec-9+ cells (p , 0.05) than didCD8ab+ cells (Table I). The Tcrgd+ cell subset also containeda significantly higher number of siglec-7+ cells (p , 0.05) (Fig.4E) and siglec-9+ cells (p , 0.05) than did the Tcrab+ cell subset

Table I. Siglec-7 and -9–expressing cells in peripheral blood of healthy individuals

Siglec-7 Siglec-9

Subset Mean 6 SDa p Value Subset Mean 6 SDa p Value

NK cell markers NK cell markersSiglec-7/CD16+CD32 75.2 6 16.3 Siglec-9/CD16+CD32 37.9 6 13.2Siglec-7/CD56+CD32 72.7 6 15.6 Siglec-9/CD56+CD32 37.5 6 13.2

B cell marker B cell markerSiglec-7/CD19+ 0.6 6 0.1 Siglec-9/CD19+ 0.3 6 0.1

NKT cell markers NKT cell markersSiglec-7/CD16+CD3+ 8.6 6 6.9 p , 0.05b Siglec-7/CD16+CD3+ 13.8 6 12.2 p , 0.05b

Siglec-7/CD56+CD3+ 13.2 6 5.2 p , 0.002c Siglec-7/CD56+CD3+ 10.5 6 8.7 p , 0.05c

T cell markers T cell markersSiglec-7/CD3+ 1.1 6 0.6 Siglec-9/CD3+ 1.5 6 1.2Siglec-7/TcRab+ 1.3 6 0.8 Siglec-9/TcRab+ 1.0 6 0.6Siglec-7/TcRgd+ 4.8 6 3.1 p , 0.05d Siglec-9/TcRgd+ 8.8 6 7.2 p , 0.05d

Siglec-7/CD45RA+CD3+ 1.6 6 0.7 Siglec-9/CD45RA+CD3+ 1.6 6 1.4Siglec-7/CD45RO+CD3+ 2.2 6 1.7 Siglec-9/CD45RO+CD3+ 2.1 6 1.8Siglec-7/CD4+CD3+ 0.2 6 0.1 Siglec-9/CD4+CD3+ 1.0 6 0.9Siglec-7/CD45RA+CD4+ 0.3 6 0.2 Siglec-9/CD45RA+CD4+ 0.5 6 0.4Siglec-7/CD45RO+CD4+ 0.3 6 0.2 Siglec-9/CD45RO+CD4+ 1.5 6 1.3Siglec-7/CD8+CD3+ 2.6 6 1.0 p , 0.01e Siglec-9/CD8+CD3+ 1.3 6 0.9Siglec-7/CD8ab+CD3+ 1.1 6 0.8 Siglec-9/CD8ab+CD3+ 0.8 6 0.4Siglec-7/CD8aa+CD3+ 14.9 6 7.7 p , 0.02f Siglec-9/CD8aa+CD3+ 5.7 6 5.5 p , 0.05f

Siglec-7/CD45RA+CD8+ 2.5 61.3 Siglec-9/CD45RA+CD8+ 1.4 6 0.9Siglec-7/CD45RO+CD8+ 5.3 6 4.3 Siglec-9/CD45RO+CD8+ 1.3 6 1.3

Gut-homing cell markersSiglec-7/b7-integrin+CD3+ 1.7 6 0.9 Siglec-9/b7-integrin+CD3+ 1.2 6 1.0Siglec-7/b7-integrin+CD4+ 0.3 6 0.3 Siglec-9/b7-integrin+CD4+ 0.3 6 0.3Siglec-7/b7-integrin+CD8+ 2.5 6 1.4 p , 0.05g Siglec-9/b7-integrin+CD8+ 0.8 6 0.6Siglec-7/CCR9+CD3+ 2.8 6 0.3 p , 0.002h Siglec-7/CCR9+CD3+ 2.1 6 1.5Siglec-7/CCR92CD3+ 1.0 6 0.7 Siglec-7/CCR92CD3+ 1.5 6 1.2

aRatio of siglec-7/9+ cells/each subset (%) was calculated from the results of flow cytometry for at least four healthy individuals.bVersus CD162CD3+ subset.cVersus CD562CD3+ subset.dVersus TcRab+ subset.eVersus CD4+CD3+ subset.fVersus CD8ab+CD3+ subset.gVersus b7-integrin+CD4+ subset.hVersus CCR92CD3+ subset.

4696 IMPAIRED IMMUNOSUPPRESSIVE GLYCAN EXPRESSION IN COLON CANCER

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

(Table I, see Supplemental Fig. 3 for additional details). In termsof gut-homing markers, b7-integrin+CD8+ cells contained a sig-nificantly higher number of siglec-7+ cells than did b7-integrin2

CD8+ cells (p , 0.05) (Fig. 4E). CCR9+CD3+ cells also containedsignificantly more siglec-7+ cells than did CCR92CD3+ cells (p,0.002) (Table I). These results showed that siglec-7/-9+ cells weremore frequently expressed in gut-related lymphocyte subsets inperipheral blood. The combination of two sets of four-coloranalyses using a set of CD3, CD8a, CD8b, and anti–siglec-7and another set of CD3, CD8, b7-integrin, and anti–siglec-7(Fig. 4E) indicated that the majority of siglec-7+ cells in theCD8+ subset were CD8aa+b7-integrin+ cells.

Characterization of siglec-7/-9–expressing macrophages incolonic mucosa

The number of siglec-7+ macrophages was ∼4-fold greater than thenumber of siglec-9+ macrophages in colonic lamina propriae (Fig.5A). Most macrophages were of the siglec-7+siglec-92 phenotype(78.96 9.7%), and only 11.96 4.6% were siglec-7+siglec-9+. Thepopulation of siglec-72siglec-9+ cells was minor and accounted foronly 9.3 6 6.2%. This was in clear contrast to the siglec-7/-9–expressing monocytes in the peripheral blood of healthy individ-uals, where the majority showed the siglec-7+siglec-9+ phenotype(69.0 6 11.8%), followed by the siglec-72siglec-9+ phenotype(31.1 6 11.8%). Essentially no siglec-7+siglec-92 cells, which arepredominant in colonic mucosa, were observed in peripheral bloodmonocytes (0.3 6 0.5%). These results indicated that siglec-7–expressing macrophages/monocytes in colonic mucosa have char-

acteristics that are quite different from those in peripheral bloodmonocytes. The predominant macrophages in colonic mucosa wereof the siglec-7+siglec-92 phenotype, and the difference in thissubpopulation between the macrophages in colonic mucosa and themonocytes in peripheral blood was statistically significant at p ,0.001 (78.9 versus 0.3%), as ascertained by the unpaired Student ttest. In contrast, the predominant phenotype in peripheral bloodmonocytes was siglec-7+siglec-9+, and the difference in this sub-population between the peripheral blood monocytes and macro-phages in colonic mucosa was also statistically significant at p ,0.001 (69.0 versus 11.9%). When normal human peripheral mono-cytes were induced to differentiate into macrophage-like cells bya standard technique using M-CSF, the resulting macrophage-likecells lost their expression of siglec-7 and -9 and became siglec-72

siglec-92 after long-term culture (Supplemental Fig. 4). This resultagain suggested that the siglec-expression pattern of the residentmacrophages in colonic lamina propria is different from that inmacrophages derived from peripheral monocytes.Further marker analyses revealed that siglec-7+ macrophages/

monocytes in colonic mucosa strongly express CD68 and CD163and only weakly express CD14 and CD89 (Fig. 5B–D). In con-trast, the siglec-9+ macrophages/monocytes in colonic mucosatended to express CD14 and CD89 more frequently than did thesiglec-7+ macrophages/monocytes (Fig. 5B–D). These findingssuggested that the siglec-7+ cells in colonic mucosa have char-acteristics similar to so-called “resident macrophages” (27, 28),whereas siglec-9+ cells, at least in part, exhibit characteristicsmore like monocytes and activated macrophages. A significant

FIGURE 5. Characterization of macrophages/monocytes in colonic lamina propria expressing siglec-7 and -9. (A) Distribution of siglec-7/-9+ macro-

phages/monocytes in colonic lamina propria and comparison with that in peripheral blood monocytes. The number of cells positive for either siglec-7 or -9 in

CD33+ cells (WM53, IgG1) was taken as 100%. (B) Macrophage/monocyte marker analyses of colonic LPMCs expressing siglec-7 and/or -9. The Abs used

were HLA-DR (TAL.1B5, IgG1), CD11c (KB90, IgG1), CD83 (HB15e, IgG1), and CD209 (DC-SIGN, DZX02, IgG2a). (C) Example of confocal mi-

croscopy after double staining for anti–siglec-7 (upper panel) or siglec-9 (lower panel) Ab and anti-CD14 Ab, showing that siglec-9+ cells frequently express

CD14. (D) Example of confocal microscopy after double staining with anti–siglec-7 (upper panel) or siglec-9 (lower panel) Ab and anti-CD163 Ab, showing

that siglec-7+ cells frequently express CD163. (E) Example of CD209 (DC-SIGN) expression in siglec-7+ macrophages/monocytes in lamina propria of

colonic mucosal membranes. Arrows indicate macrophages/monocytes expressing both siglec-7 and DC-SIGN. (C–E) Original magnification 3200.

The Journal of Immunology 4697

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

number of siglec-7+ cells expressed CD209, known as a markerof dendritic cells (Fig. 5B, 5E).The stroma of colon cancer tissues also contained a significant

number of cells expressing siglec-7 and/or -9. Most of themwere macrophages, followed by neutrophils, CD8+ T cells, andNK cells, and their relative numbers were not statistically sig-nificantly different from those in nonmalignant colonic mucosalmembranes. The majority of macrophages were also of the siglec-7+siglec-92 phenotype in the stroma of the colon cancer tissues(data not shown).

Roles of siglec-7/-9 in macrophage function

Disialyl Lewisa and sialyl 6-sulfo Lewisx, the carbohydrate ligandsfor immunosuppressive siglec-7/-9, are preferentially expressed innormal colonic epithelial cells, and their expression is lost duringcolonic carcinogenesis. In this context, we scrutinized the effect ofsiglec-7/-9 on COX2 expression in macrophages, because the en-hanced COX2 expression in stromal cells is implicated in coloncarcinogenesis (10, 29). Human monocytoid cells U937 are knownto differentiate into macrophage-like cells when cultured withTPA and to produce COX2 upon further LPS stimulation. WhenU937 cells transfected with siglec-7/-9 were treated with TPA andLPS, ligation of siglec-7/-9 with agonistic Abs significantly sup-pressed induction of COX2 mRNA (Fig. 6A). The results ofquantitative real-time RT-PCR analyses performed in parallel withconventional RT-PCR indicated that ligation of siglec-7 on theU937/siglec-7 cells resulted in suppression of COX2 gene tran-scription to 48.3 6 9.1% (mean 6 SD, p , 0.05) of controlculture with control F(ab9)2, and ligation of siglec-9 on the U937/siglec-9 cells suppressed COX2 induction to 22.7 6 11.5% (p ,0.01). Both siglecs conferred suppressive effects on COX2 tran-scription, and suppression of COX2 was also reflected in the re-duction of its major product PGE2 (Fig. 6B).

DiscussionAmong the glycans in colonic epithelial cells, some are specificallyexpressed in normal epithelial cells, and some other glycans arepreferentially expressed in cancer cells. It is noteworthy that onlythe glycans that are expressed in normal epithelial cells serve asligands for siglec-7 and -9, whereas cancer-associated glycans donot, as indicated by the results shown in Figs. 1 and 2. We dem-onstrated that disialyl Lewisa preferentially expressed in nonma-lignant colonic epithelial cells serves as a ligand for siglec-7 and-9, whereas sialyl Lewisa on cancer cells has no binding activity.Further cell-binding experiments indicated that sialyl 6-sulfoLewisx, another glycan preferentially expressed on nonmalignantcolonic epithelial cells, also serves as a ligand for siglec-7,whereas sialyl Lewisx, the corresponding glycan in cancer cells,does not.Cancer-associated carbohydrate determinants, such as sialyl

Lewisa and sialyl Lewisx, are known to serve as ligands forselectins and mediate cancer cell adhesion to vascular beds in thecourse of hematogenous metastasis (4–6, 20). Expression of thesedeterminants is markedly increased in cancer cells compared withnormal epithelial cells. Nonmalignant epithelial cells expressdisialyl Lewisa and sialyl 6-sulfo Lewisx glycans having morecomplex structures than sialyl Lewisa/x (7, 12). Their expressionis lost at the early stage of colon carcinogenesis as result of theepigenetic silencing of glycogenes involved in their synthesis (30,31). Loss of these normal glycans leads to accumulation of less-complex glycans, such as sialyl Lewisa and sialyl Lewisx, incancers. Taken together, our results indicate that colonic epithelialcells lose their siglec ligands upon malignant transformation, andthis is associated with the acquisition of selectin ligand glycans.

Our results also indicated that the majority of immune cellsexpressing siglec-7 and -9 in normal colonic mucosa aremacrophages/monocytes. A large number of tissue macrophagesbearing siglec-7 was found in colonic lamina propria, and theyshowed a surface marker pattern similar to that of so-called “res-

FIGURE 6. Inhibition of inflammatory mediator production by human

monocytoid U937 cells by siglec-7 or -9. (A) Results of RT-PCR analyses

of inhibition of LPS-mediated COX2 induction in U937 cells by ligation of

siglec-7 or siglec-9. U937 cells stably transfected with siglec-7 cDNA

(upper panel) or with siglec-9 cDNA (middle panel) or parental U937 cells

(lower panel) were differentiated into macrophages with TPA and chal-

lenged with LPS. The F(ab9)2 fragment of anti–siglec-7 Ab and/or that of

anti–siglec-9 Ab was added to the culture TPA-induced U937 cells with or

without LPS for receptor ligation. (B) Results of ELISA analyses of in-

hibition of LPS-mediated PGE2 production by U937 cells by ligation of

siglec-7 (left panel) or siglec-9 (right panel). F(ab9)2 fragments were also

used to evaluate PGE2 production by the U937 cells. Values are mean 6SD from quadruplicate experiments. The results in the upper two panels in

(A) are representative of at least three independent experiments, and those

in the lowest panel in (A) and (B) are representative of two independent

experiments.

4698 IMPAIRED IMMUNOSUPPRESSIVE GLYCAN EXPRESSION IN COLON CANCER

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

ident macrophages” (27, 28), characterized by low expression ofCD14 and CD89, and eventually expressed dendritic cell markers.Most of them predominantly expressed siglec-7, and the expres-sion of siglec-9 was less prominent. This type of siglec-7–domi-nant macrophage/monocyte was not identified in peripheral blood.Another notable finding was that a significant population of CD8+

T cells was found to express siglec-7/-9 in colonic lamina propria,indicating that the distribution pattern of siglec-7/-9 among mu-cosal immune cells is quite different from that in normal peripheralblood. Human peripheral blood T cells had been known to containa very small number, if any, of siglec-7/-9+ cells, and the signifi-cance of such siglec-7/-9+ CD8+ T cells has largely been over-looked (32). Our detailed analysis of siglec-7/-9+ T cells in normalperipheral blood indicated that siglec-7/-9 are preferentially ex-pressed in gut-associated minor populations, such as Tcrgd cells,CD8aa+ cells, and the T cells expressing gut-homing receptorsb7-integrin and CCR9. These findings imply that siglec-7/-9 issomehow preferentially expressed in gut-associated T cells (33).Most siglecs, including siglec-7/-9, have ITIMs in their cyto-

plasmic domains, which inhibit immune cell activation byrecruiting tyrosine phosphatases SHP-1 and SHP-2. Siglecs’suppression of killing activity and immune mediator productionof immune cells is well documented in the literature (2, 8, 34–37). The biological significance of the interaction of residentmacrophages/monocytes expressing siglec-7/-9 and their specificglycan ligands on colonic epithelial cells could be the protectionof mucosal membrane from excess activation by immune cells.The endogenous glycan ligands on the resident macrophages/monocytes will also play important roles in physiological set-tings, because cis-ligands are known to inhibit binding of siglecswith exogenous glycan ligands (2, 6).Colonic epithelial cells are rendered increasingly susceptible

to cancer progression both in humans and mice by chronic in-flammatory stimuli in mucosal stroma (9). In this context, COX2,known to be expressed in stromal cells, such as fibroblasts (10, 11)and macrophages (29), has especially attracted researchers’ at-tention. COX2 expression in mucosal stromal cells is known toplay a crucial role in facilitating colonic tumorigenesis in theAPCD716 murine model (10, 11). Administration of COX2 inhib-itors is known to be effective in preventing colonic carcinogenesis(9). Several intrinsic protective mechanisms for suppressing ex-cessive production of COX2 by resident stromal cells must beoperating in normal colonic mucosal membranes. Because onlythe carbohydrate determinants on normal epithelial cells, but notcancer-associated determinants, served as ligands for siglecs, wefocused on the effects of siglec ligation on COX2 production ofmacrophages in the current study.Experiments using human macrophage cell lines transfected with

siglec-7/-9 indicated that the ligation of siglec-7/-9 suppresses LPS-induced COX2 and PGE2 production. These results suggest thatnormal glycans of colonic epithelial cells exert a suppressive effecton tissue macrophage COX2 expression in colonic mucosa, thusmaintaining immunological homeostasis in normal mucosal mem-branes. These results also imply that the cancer-associated im-paired glycosylation of siglec-7 and -9 ligands serves to enhanceCOX2 production by mucosal macrophages. Because colonic tis-sues continuously encounter bacterial stimuli from fecal flora,colonic mucosa is equipped with various modules that suppressexcess activation of mucosal immune cells and prevent inflam-matory damage to the host. The interaction of siglecs on mucosalimmune cells and their specific glycan ligands on nonmalignantepithelial cells might be one of such homeostatic systems, and itsabrogation through cancer-associated abnormal glycosylation pro-vides a plausible mechanism for the progression of colon cancers.

AcknowledgmentsWe thank Prof. Makoto Kiso (Gifu University) for the gifts of synthetic

pure carbohydrate determinants.

DisclosuresThe authors have no financial conflicts of interests.

References1. Liu, F. T., and G. A. Rabinovich. 2010. Galectins: regulators of acute and chronic

inflammation. Ann. N. Y. Acad. Sci. 1183: 158–182.2. Crocker, P. R., J. C. Paulson, and A. Varki. 2007. Siglecs and their roles in the

immune system. Nat. Rev. Immunol. 7: 255–266.3. Kannagi, R. 2002. Regulatory roles of carbohydrate ligands for selectins in the

homing of lymphocytes. Curr. Opin. Struct. Biol. 12: 599–608.4. Hakomori, S. 2002. Glycosylation defining cancer malignancy: new wine in an

old bottle. Proc. Natl. Acad. Sci. USA 99: 10231–10233.5. Kannagi, R., K. Sakuma, K. Miyazaki, K.-T. Lim, A. Yusa, J. Yin, and M. Izawa.

2010. Altered expression of glycan genes in cancers induced by epigenetic si-lencing and tumor hypoxia: clues in the ongoing search for new tumor markers.Cancer Sci. 101: 586–593.

6. Kannagi, R., J. Yin, K. Miyazaki, and M. Izawa. 2008. Current relevance ofincomplete synthesis and neo-synthesis for cancer-associated alteration of car-bohydrate determinants—Hakomori’s concepts revisited. Biochim. Biophys.Acta 1780: 525–531.

7. Miyazaki, K., K. Ohmori, M. Izawa, T. Koike, K. Kumamoto, K. Furukawa,T. Ando, M. Kiso, T. Yamaji, Y. Hashimoto, et al. 2004. Loss of disialyl Lewis(a), the ligand for lymphocyte inhibitory receptor sialic-binding immunoglobulin-like lectin 7 (Siglec-7) associated with increased sialyl Lewis(a) expression onhuman colon cancers. Cancer Res. 64: 4498–4505.

8. Varki, A., and T. Angata. 2006. Siglecs—the major subfamily of I-type lectins.Glycobiology 16: 1R–27R.

9. Wang, D., and R. N. DuBois. 2008. Pro-inflammatory prostaglandins and pro-gression of colorectal cancer. Cancer Lett. 267: 197–203.

10. Taketo, M. M. 2006. Mouse models of gastrointestinal tumors. Cancer Sci. 97:355–361.

11. Taketo, M. M. 2006. Wnt signaling and gastrointestinal tumorigenesis in mousemodels. Oncogene 25: 7522–7530.

12. Izawa, M., K. Kumamoto, C. Mitsuoka, C. Kanamori, A. Kanamori, K. Ohmori,H. Ishida, S. Nakamura, K. Kurata-Miura, K. Sasaki, et al. 2000. Expression ofsialyl 6-sulfo Lewis X is inversely correlated with conventional sialyl Lewis Xexpression in human colorectal cancer. Cancer Res. 60: 1410–1416.

13. Kannagi, R., A. Kitahara, S. Itai, K. Zenita, K. Shigeta, T. Tachikawa, A. Noda,H. Hirano, M. Abe, S. Shin, et al. 1988. Quantitative and qualitative charac-terization of human cancer-associated serum glycoprotein antigens expressingepitopes consisting of sialyl or sialyl-fucosyl type 1 chain. Cancer Res. 48:3856–3863.

14. Mitsuoka, C., M. Sawada-Kasugai, K. Ando-Furui, M. Izawa, H. Nakanishi,S. Nakamura, H. Ishida, M. Kiso, and R. Kannagi. 1998. Identification of a majorcarbohydrate capping group of the L-selectin ligand on high endothelial venules inhuman lymph nodes as 6-sulfo sialyl Lewis X. J. Biol. Chem. 273: 11225–11233.

15. Kimura, N., C. Mitsuoka, A. Kanamori, N. Hiraiwa, K. Uchimura, T. Muramatsu,T. Tamatani, G. S. Kansas, and R. Kannagi. 1999. Reconstitution of functionalL-selectin ligands on a cultured human endothelial cell line by cotransfection ofa1–.3 fucosyltransferase VII and newly cloned GlcNAcbeta:6-sulfotransferasecDNA. Proc. Natl. Acad. Sci. USA 96: 4530–4535.

16. Yamaji, T., T. Teranishi, M. S. Alphey, P. R. Crocker, and Y. Hashimoto. 2002. Asmall region of the natural killer cell receptor, Siglec-7, is responsible for itspreferred binding to a 2,8-disialyl and branched a 2,6-sialyl residues. A com-parison with Siglec-9. J. Biol. Chem. 277: 6324–6332.

17. Kannagi, R. 2000. Monoclonal anti-glycosphingolipid antibodies. MethodsEnzymol. 312: 160–179.

18. Ohmori, K., K. Kanda, C. Mitsuoka, A. Kanamori, K. Kurata-Miura, K. Sasaki,T. Nishi, T. Tamatani, and R. Kannagi. 2000. P- and E-selectins recognize sialyl6-sulfo Lewis X, the recently identified L-selectin ligand. Biochem. Biophys.Res. Commun. 278: 90–96.

19. Ohmori, K., F. Fukui, M. Kiso, T. Imai, O. Yoshie, H. Hasegawa, K. Matsushima,and R. Kannagi. 2006. Identification of cutaneous lymphocyte-associated anti-gen as sialyl 6-sulfo Lewis X, a selectin ligand expressed on a subset of skin-homing helper memory T cells. Blood 107: 3197–3204.

20. Takada, A., K. Ohmori, T. Yoneda, K. Tsuyuoka, A. Hasegawa, M. Kiso, andR. Kannagi. 1993. Contribution of carbohydrate antigens sialyl Lewis A andsialyl Lewis X to adhesion of human cancer cells to vascular endothelium.Cancer Res. 53: 354–361.

21. Yamaji, T., M. Mitsuki, T. Teranishi, and Y. Hashimoto. 2005. Characterizationof inhibitory signaling motifs of the natural killer cell receptor Siglec-7: atten-uated recruitment of phosphatases by the receptor is attributed to two aminoacids in the motifs. Glycobiology 15: 667–676.

22. Barbieri, S. S., S. Eligini, M. Brambilla, E. Tremoli, and S. Colli. 2003. Reactiveoxygen species mediate cyclooxygenase-2 induction during monocyte to mac-rophage differentiation: critical role of NADPH oxidase. Cardiovasc. Res. 60:187–197.

23. Tanji, K., T. Imaizumi, T. Matsumiya, H. Itaya, K. Fujimoto, X. Cui, T. Toki,E. Ito, H. Yoshida, K. Wakabayashi, and K. Satoh. 2001. Desferrioxamine, an

The Journal of Immunology 4699

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from

iron chelator, upregulates cyclooxygenase-2 expression and prostaglandin pro-duction in a human macrophage cell line. Biochim. Biophys. Acta 1530: 227–235.

24. Inoue, H., and T. Tanabe. 1998. Transcriptional role of the nuclear factor k B sitein the induction by lipopolysaccharide and suppression by dexamethasone ofcyclooxygenase-2 in U937 cells. Biochem. Biophys. Res. Commun. 244: 143–148.

25. Itai, S., J. Nishikata, T. Yoneda, K. Ohmori, H. Yamabe, S. Arii, T. Tobe, andR. Kannagi. 1991. Tissue distribution of 2-3 and 2-6 sialyl Lewis A antigens andsignificance of the ratio of two antigens for the differential diagnosis of malig-nant and benign disorders of the digestive tract. Cancer 67: 1576–1587.

26. Kimura, N., K. Ohmori, K. Miyazaki, M. Izawa, Y. Matsuzaki, Y. Yasuda,H. Takematsu, Y. Kozutsumi, A. Moriyama, and R. Kannagi. 2007. Human B-lymphocytes express a2-6-sialylated 6-sulfo-N-acetyllactosamine serving asa preferred ligand for CD22/Siglec-2. J. Biol. Chem. 282: 32200–32207.

27. Smith, P. D., L. E. Smythies, M. Mosteller-Barnum, D. A. Sibley, M. W. Russell,M. Merger, M. T. Sellers, J. M. Orenstein, T. Shimada, M. F. Graham, andH. Kubagawa. 2001. Intestinal macrophages lack CD14 and CD89 and conse-quently are down-regulated for LPS- and IgA-mediated activities. J. Immunol.167: 2651–2656.

28. Rogler, G., M. Hausmann, D. Vogl, E. Aschenbrenner, T. Andus, W. Falk,R. Andreesen, J. Scholmerich, and V. Gross. 1998. Isolation and phenotypiccharacterization of colonic macrophages. Clin. Exp. Immunol. 112: 205–215.

29. Bamba, H., S. Ota, A. Kato, A. Adachi, S. Itoyama, and F. Matsuzaki. 1999.High expression of cyclooxygenase-2 in macrophages of human colonic ade-noma. Int. J. Cancer 83: 470–475.

30. Kim, Y. S., and G. Deng. 2008. Aberrant expression of carbohydrate antigens incancer: the role of genetic and epigenetic regulation.Gastroenterology 135: 305–309.

31. Kawamura, Y. I., M. Toyota, R. Kawashima, T. Hagiwara, H. Suzuki, K. Imai,Y. Shinomura, T. Tokino, R. Kannagi, and T. Dohi. 2008. DNA hypermethylationcontributes to incomplete synthesis of carbohydrate determinants in gastroin-testinal cancer. Gastroenterology 135: 142–151.e3.

32. Nguyen, D. H., N. Hurtado-Ziola, P. Gagneux, and A. Varki. 2006. Loss ofSiglec expression on T lymphocytes during human evolution. Proc. Natl. Acad.Sci. USA 103: 7765–7770.

33. McGhee, J. R., J. Kunisawa, and H. Kiyono. 2007. Gut lymphocyte migration:we are halfway ‘home’. Trends Immunol. 28: 150–153.

34. Falco, M., R. Biassoni, C. Bottino, M. Vitale, S. Sivori, R. Augugliaro,L. Moretta, and A. Moretta. 1999. Identification and molecular cloning of p75/AIRM1, a novel member of the sialoadhesin family that functions as an inhib-itory receptor in human natural killer cells. J. Exp. Med. 190: 793–802.

35. Nicoll, G., J. Ni, D. Liu, P. Klenerman, J. Munday, S. Dubock, M. G. Mattei, andP. R. Crocker. 1999. Identification and characterization of a novel siglec, siglec-7, expressed by human natural killer cells and monocytes. J. Biol. Chem. 274:34089–34095.

36. Zhang, J. Q., G. Nicoll, C. Jones, and P. R. Crocker. 2000. Siglec-9, a novel sialicacid binding member of the immunoglobulin superfamily expressed broadly onhuman blood leukocytes. J. Biol. Chem. 275: 22121–22126.

37. Angata, T., and A. Varki. 2000. Cloning, characterization, and phylogeneticanalysis of siglec-9, a new member of the CD33-related group of siglecs. Evi-dence for co-evolution with sialic acid synthesis pathways. J. Biol. Chem. 275:22127–22135.

4700 IMPAIRED IMMUNOSUPPRESSIVE GLYCAN EXPRESSION IN COLON CANCER

by guest on April 10, 2019

http://ww

w.jim

munol.org/

Dow

nloaded from