![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,
Akemi Suzuki,5 and
Yasunori Kozutsumi1,5,7*
Laboratory of Membrane Biochemistry and Biophysics, Graduate School of Biostudies,1 Department of Biological Chemistry,2 Department of Genomic Drug Discovery, Graduate School of Pharmaceutical Sciences,3 Center for Genomic Medicine, Graduate School of Medicine, Kyoto University, Sakyo, Kyoto 606-8501, Japan,4 Supra-Biomolecular System Research Group, RIKEN Frontier Research System,5 Laboratory for Behavioral Genetics, RIKEN Brain Science Institute, RIKEN, Wako, Saitama 351-0198, Japan,6 CREST, Japan Science and Technology, Kawaguchi, Saitama, Japan7
Received 2 November 2006/ Returned for modification 9 January 2007/ Accepted 30 January 2007
 |
ABSTRACT |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2,6-linked N-acetylneuraminic acid (Neu5Ac) on a lactosamine glycan chain(s), in both Sia modification- and Sia linkage-dependent manners. In mouse germinal center B cells, the expression of the GL7 epitope was upregulated due to the in situ repression of CMP-Neu5Ac hydroxylase (Cmah), the enzyme responsible for Sia modification of Neu5Ac to Neu5Gc. Such Cmah repression caused activation-dependent dynamic reduction of CD22 ligand expression without losing 2,6-linked sialylation in germinal centers. The in vivo function of Cmah was analyzed using gene-disrupted mice. Phenotypic analyses showed that Neu5Gc glycan functions as a negative regulator for B-cell activation in assays of T-cell-independent immunization response and splenic B-cell proliferation. Thus, Neu5Gc is required for optimal negative regulation, and the reaction is specifically suppressed in activated B cells, i.e., germinal center B cells. |
INTRODUCTION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In the present study, we found that GL7 recognizes a glycan moiety containing terminal sialic acid (Sia) in both linkage- and modification-dependent manners. Sia is a family of acidic nine-carbon sugars that often occupies the nonreducing terminus of mammalian glycan chains (47), and Sia is essential for early development of mice (49). The localization of Sia-bearing glycan chains on the cell surface makes sialylated molecules seem to be likely targets for various cellular and molecular recognition molecules, such as the mammalian lectins that are abundant in the immune system (61). A family of enzymes, sialyltransferases, is responsible for the formation of the Sia linkage to the underlying glycan chains. To determine the linkage specificity of GL7 recognition, we used the gene expression profiles of sialylation-related genes obtained by DNA microarray analysis to screen for a responsible sialyltransferase gene for the biosynthesis of the GL7 determinant.
Apart from the linkage variations, Sia also occurs in various molecular species as a result of modifications at its C-4, C-5, C-7, C-8, and C-9 positions; these modifications are spatially and temporary regulated (60). We also found that the determinant recognition by GL7 is specific to a Sia modification at the C-5 position. In mice, Sia occurs in two main forms with respect to the moiety at the C-5 position: N-acetylneuraminic acid (Neu5Ac), which is a precursor form of the diverse Sia family, and its major modified form, N-glycolylneuraminic acid (Neu5Gc). The structural difference between Neu5Ac and Neu5Gc is a single oxygen atom in the C-5 position. The modification reaction that produces Neu5Gc is catalyzed at the sugar-nucleotide level in the cytosol by the enzyme CMP-Neu5Ac hydroxylase (Cmah) (24, 53). Cmah determines the cell surface expression ratio of these two Sia species, as the cytosolic Cmah reaction occurs prior to the sialyltransferase reaction, which takes place in the Golgi apparatus during the biosynthesis of glycoconjugates. We found that GL7 recognizes only Neu5Ac-bearing glycans and that the reduction of Cmah expression plays a major role in the formation of the GL7 epitope in activated B cells in the germinal center, which was in sharp contrast to the dominant expression of Neu5Gc in mouse lymphocytes.
To examine the in vivo function of Neu5Gc-bearing glycans, we disrupted the Cmah gene in mice. Cmah disruption is expected to modify the Sia-mediated Sia species-specific recognition event without affecting overall sialylation, which can affect the behavior of the protein in various ways. We primarily focused on the phenotypic consequences of Cmah disruption in B cells since Cmah is regulated in B cells, especially in response to activation. Cmah-null mice exhibited hyperresponsive B cell phenotypes in assays measuring B-cell functions, i.e., antibody production and proliferation.
|
MATERIALS AND METHODS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Antibodies and lectins. The antibodies used were as follows: donkey F(ab')2 against mouse IgM (Jackson ImmunoResearch, West Grove, PA); R-phycoerythrin (R-PE)-conjugated anti-mouse B220 (RA3-6B2); R-PE-conjugated goat F(ab')2 anti-human IgG-Fc; R-PE-conjugated streptavidin (CALTAG Laboratories, Burlingame, CA); fluorescein isothiocyanate (FITC)-conjugated and purified GL7; FITC-conjugated anti-mouse B220 (RA3-6B2); R-PE-conjugated anti-mouse I-A/I-E (M5/114.15.2); biotin-conjugated anti-CD22 (Cy34.1) (BD Pharmingen, San Diego, CA); horseradish peroxidase (HRP)-conjugated goat anti-rat IgM; alkaline phosphatase-conjugated isotype-specific goat anti-mouse IgA, IgG1, IgG3, and IgM; unlabeled isotype-specific goat anti-mouse IgA and IgG3; R-PE-conjugated anti-mouse IgM (1B4B1); biotin-conjugated anti-mouse CD22 (2D6) (Southern Biotechnology Associates, AL); anti-mouse polyvalent Igs; HRP-conjugated PT-66 (an antiphosphotyrosine MAb; Sigma, St. Louis, MO); CD90 (Thy1.2) MicroBeads; anti-FITC MicroBeads (Miltenyi Biotec, Bergisch Gladbach, Germany); rabbit anti-mouse CD22 serum (Chemicon, Temecula, CA), HRP-conjugated donkey F(ab')2 anti-rabbit Ig (Amersham Life Science, Buckinghamshire, United Kingdom); antiactin (Santa Cruz Biotechnology, Santa Cruz, CA); HRP-conjugated goat anti-mouse IgG; HRP-conjugated rabbit anti-goat IgG (ZYMED Lab, South San Francisco, CA). Anti-CD22 MAb (Cy34.1) was purified from the culture supernatant of hybridoma Cy34.1 (ATCC). Biotinylated A. hypogaea PNA was obtained from HONEN (Tokyo, Japan), and FITC-conjugated Sambucus sieboldiana agglutinin (SSA) was obtained from Seikagaku Kogyo (Tokyo, Japan).
Preparation of Fc fusion proteins of sialoadhesin and CD22. Recombinant soluble forms of the amino-terminal domains (domains 1 to 3) of mouse sialoadhesin/Siglec-1, mouse CD22/Siglec-2, and human CD22/Siglec-2 fused to the Fc region of human IgG1 (mSn-Fc, mCD22-Fc, and hCD22-Fc, respectively) were produced in stably transfected Lec2 cells, a cell line deficient in protein sialylation. The production of the Siglec (Sia-binding Ig superfamily lectin)-Fc fusion probe in the Lec2 cell line resulted in considerably enhanced binding to the ligand, which allowed the identification of changes in ligand expression. The Siglec-Fc probes were purified from the culture supernatant using protein A-Sepharose columns (Pierce, Rockford, IL).
Flow cytometry. Cell labeling was carried out in fluorescence-activated cell sorter buffer (1% bovine serum albumin [BSA] and 0.1% NaN3 in phosphate-buffered saline [PBS]). Data were acquired using a FACScan (Becton Dickinson, Franklin Lakes, NJ) instrument and analyzed using FlowJo software (Tree Star, San Carlos, CA). For comparison with the microarray data, B lymphoma cells (1 x 105) were stained with FITC-conjugated GL7 (dilution, 1:100) for 1 h. This staining condition was determined using the criterion that the strongest staining did not reach a plateau. Mean fluorescence intensity (MFI) of GL7 staining was acquired using a FACScan at settings under which unstained control cells gave a signal of around 5 on the FL-1 channel. The mean FL-1 signal of each stained sample was divided by that of the unstained sample to produce the relative staining profiles on flow cytometry to be compared with the cDNA microarray profiles of relative gene expression. For mSn-Fc, mCD22-Fc, and hCD22-Fc staining, these Fc fusion proteins were precomplexed with R-PE-conjugated goat F(ab')2 anti-human IgG.
Sialidase treatment. Sialidase treatment was carried out in 100 mM sodium acetate (pH 5.2) for 30 min at room temperature prior to the staining for flow cytometry. Sialidase from Arthrobacter ureafaciens (Calbiochem, San Diego, CA) and sialidase from Salmonella enterica serovar Typhimurium (Takara, Kusatsu, Japan) were used.
Immunoblotting with GL7. The cells were sonicated in detergent-free lysis buffer (25 mM Tris-HCl [pH 7.6], 1 mM dithiothreitol, protease inhibitor cocktail [Nacalai Tesque]). The pellets (membrane fractions) were collected by ultracentrifugation and solubilized in NP-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 25 mM HEPES [pH 7.4], protease inhibitor cocktail). The extracts were subjected to immunoblotting with GL7 in the presence or absence of 100 mM Neu5Ac.
Development of cDNA microarray for glycan-related genes. The RIKEN Frontier Human Glyco-gene cDNA microarray, version 2, which was spotted by Takara, consisted of 888 genes, which included glycosyltransferase genes and genes related to sugar metabolism, glycan modification, glycan recognition, and lipid metabolism.
Use of cDNA microarray for identification of glycan-related genes. Poly(A)+ RNA samples were isolated from mid-log-phase cells using the mTRAP system (Activemotif, Carlsbad, CA) and were quality checked using a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). One microgram of poly(A)+ RNA from the B-cell lines (rRNA contamination subtracted) and universal reference RNA (Clontech, Mountain View, CA) were labeled using a CyScribe first-strand cDNA labeling kit (Amersham). Competitive hybridization was performed on the microarray, and data were obtained using an Affymetrix 428 array scanner. To achieve a fair cross-cell line comparison, we fixed Cy3 as the signal for the universal reference RNA and Cy5 for the RNA from the B-cell lines. Microarray data were background corrected using a smoothing function and then Lowess normalized using linear models for microarray data. This readout was sigma normalized to avoid variation among microarray replicates. Then, the Cy5 signal from the B-cell lines was divided by the Cy3 signal to obtain the relative expression profile for each gene in the six cell lines as expression ratios relative to the universal reference RNA (1, 16, 29, 40). The gene expression profiles were compared with the GL7 staining profiles from flow cytometry. The similarity between the profiles was evaluated with Pearson's correlation coefficient, and probability values (P values) were calculated by the correlation coefficient test. For the correlation coefficient test of a sample size of six, a coefficient of 0.81 indicates a statistical significance level of 5%.
Transfection. CHO-K1 cells were stably transfected with pIRES (where IRES is internal ribosome entry site) vector (Clontech), either with or without rat cDNA for St6gal1. Transfected cells were selected with G418 (1 mg/ml), and multiple stable clones were established.
Enzyme-linked immunosorbent assay (ELISA). In 96-well assay plates, GL7 antibody was immobilized in wells coated with the capturing antibody, purified anti-rat IgM. The wells were washed and incubated with streptavidin-conjugated sugar chain probes (50 µM), prepared as previously reported (65). The captured probes were detected with biotinylated alkaline phosphatase (Vector Laboratories, Burlingame, CA) and p-nitrophenyl phosphate by measuring the absorbance at 405 nm.
Spleen sectioning and immunohistochemistry. Mice were immunized intraperitoneally with 3 x 108 sheep red blood cells (SRBC) in 100 µl of saline. Spleens were removed 8 or 10 days after immunization and embedded in Tissue-Tek OCT (22-oxyacalcitriol) compound (Sakura Finetechnical, Tokyo, Japan). Spleen sections were cut at a 6-µm thickness on a cryostat microtome (Leica Geosystems, Heerbrugg, Switzerland), thaw-mounted onto Matsunari adhesive silane-coated slides, and fixed in acetone. After rehydration in Tris-buffered saline and blocking in Tris-buffered saline with 5% BSA and 0.05% Tween 20, the sections were stained with GL7, PNA, or mCD22-Fc precomplexed with R-PE-conjugated anti-human IgG. The stained sections were analyzed under a confocal laser-scanning microscope (Olympus, Tokyo, Japan).
Magnetic sorting preparation of splenic B-cell-enriched fraction. B-cell-enriched fractions were obtained by Thy1.2 depletion of splenocytes on a MACS (magnetic cell sorter) depletion column (Miltenyi Biotec). Thy1.2-depleted fractions were stained with B220 to confirm B-cell enrichment. To avoid Neu5Gc contamination in the experimental systems, RPMI 1640 medium (Invitrogen, Carlsbad, CA) containing 10% human serum (Uniglobe, Reseda, CA) or chicken serum (JRH Biosciences, Lenaxa, KS), rather than fetal bovine serum (FBS) (JRH), was used in most of the experiments. In addition, sodium pyruvate (Invitrogen), nonessential amino acids solution (Invitrogen), L-glutamine, and 2-mercaptoethanol were added to the medium.
Germinal center B-cell analyses.
Splenic B cells from SRBC-immunized mice were incubated with FITC-conjugated GL7 and then with anti-FITC MicroBeads. The labeled cells were collected as germinal center B cells using a MACS LS column (Miltenyi Biotec). The germinal center, nongerminal center, and control (untreated) B cells were lysed by sonication in detergent-free lysis buffer (described above), and the lysates were separated by ultracentrifugation. The supernatant (cytosolic fraction) was used for immunoblotting with anti-Cmah antibody, and the pellet (membrane fraction) was used for the analysis of Sia species by high-pressure liquid chromatography (HPLC). Immunoblotting was performed using rabbit N8 antiserum against mouse Cmah, as previously reported (27). The ratios of Neu5Gc were determined by derivatizing Sia with 1,2-diamino-4,5-methylenedioxybenzene (DMB), a fluorescent compound for -keto acids, as previously described (27). In brief, Sia was released by incubating the pellet in 2 M acetic acid at 80°C, derivatized with DMB (Dojindo, Mashiki, Japan), and analyzed on a reverse-phase column (TSK-gel ODS-80Tm; Tosoh, Tokyo, Japan) using a Shimadzu LC10 HPLC system.
Detection of Sia in tissues. The ratios of Neu5Ac and Neu5Gc were determined as above. Sia was released by incubating tissues in 100 mM sulfuric acid (which also destroys the O-acetyl group often found on the C-7 to C-9 positions of Sia molecules), derivatized with DMB, and analyzed by HPLC.
Real-time RT-PCR analysis. Real time reverse transcription-PCR (RT-PCR) experiments were performed using a QuantiTect SYBR Green PCR kit (QIAGEN Japan, Tokyo, Japan) and an ABI 7700 sequence detection system (Applied Biosystems Japan, Tokyo, Japan). Total RNA was purified from untreated or lipopolysaccharide (LPS)-stimulated mouse splenic B cells, and 2 µg was used for reverse transcription. The amplification cycle was as follows: 15 min at 95°C, followed by up to 40 cycles of 15 s at 94°C, 30 s at 58°C/50°C, and 30 s at 72°C. The PCR primers used for amplification were: ZP-5, 5'-AGATTTACAAGGATTCC-3'; ZP-E, 5'-CTTAAATCCAGCCCA-3' (Cmah); PS-mCD22-6, 5'-CCTCCACTCCTCAGGCCAGA-3'; PS-mCD22-E, 5'-GCCTATCCCATTGGTCCCT-3' (Cd22); PS-ST6Gal-1, 5'-TCTTCGAGAAGAATATGGTG-3'; PS-ST6Gal-A, 5'-GACTTATGGAGAAGGATGAG-3' (St6gal1); PS-GAPDH-1 (where GAPDH is glyceraldehyde-3-phosphate dehydrogenase), 5'-GTGGAGATTGTTGCCATCAACG-3'; PS-GAPDH-A, 5'-TCTCGTGGTTCACACCCATCAC-3' (Gapdh); PS-BACTIN-1, 5'-ACGATATCGCTGCGCTGGTC-3'; and PS-BACTIN-A, 5'-CATGAGGTAGTCTGTCAGGT C-3' (Actb). Each sample was analyzed in more than three wells. Relative mRNA abundance was calculated using the comparative cycle threshold method and expressed as a ratio to the nonstimulated sample.
Retrovirus preparation and infection. Cmah cDNA was cloned into the modified mouse stem cell virus vector, which expresses Cmah and the extracellular domain of human CD4 by means of an internal ribosome entry site. Plasmids were transiently transfected into Plat-E packaging cells (35), and retrovirus-containing supernatants were collected. After stimulation with LPS for 12 to 14 h, splenic B cells were spin infected (at 32°C for 90 min) with the retrovirus in the presence of N[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate (DOTAP; Roche Diagnostics, Mannheim, Germany). The retrovirus-infected B cells were cultured in the presence of 30 µg/ml LPS for 2 to 2.5 days, and then human CD4-positive cells were enriched with a MACS system using MACSelect 4 MicroBeads (Miltenyi Biotec). The sorted cells were subjected to flow cytometry or a proliferation assay (described below).
Targeting construct and embryonic stem (ES) cells. The Cmah targeting vector was assembled from a 129/Sv genomic clone containing exons 4 and 5 of this gene and a neomycin resistance gene driven by the phosphoglycerate kinase 1 promoter (PGK-neoR) as well as a diphtheria toxin A gene fragment driven by the MC1 promoter (DT-A) as positive and negative selection markers, respectively. The construct was created by inserting the PGK-neoR cassette into the NspV site of exon 5 of the Cmah gene. The DT-A cassette was then ligated adjacent to the 3' terminus of the construct.
Generation of mutant mice. Gene targeting and generation of mutant mice were performed essentially as described previously (23). In brief, E14 cells were electroporated with a Bio-Rad Gene Pulser (0.8 kV; 3 µF) using 30 µg of NotI-linearized targeting vector. The electroporated cells were selected in medium containing G418 (125 µg/ml) and screened for homologous recombination by Southern blot analysis of genomic DNA digested with BglI, using both radiolabeled 5' internal and 3' external probes. The mutant cells were microinjected into 3.5-day-old C57BL/6J blastocysts, and the embryos were transferred into the uteri of pseudopregnant ICR mice. Mice were used for the determination of immunological features after more than seven backcrosses to the C57BL/6J strain. All mice examined in this study were housed in a specific-pathogen-free facility.
Serum isotype-specific antibody measurement. Serum samples from nonimmunized mice at 8 to 12 weeks of age were subjected to isotype-specific ELISAs. Isotype-specific capturing antibodies were coated onto 96-well ELISA plates, and nonspecific binding was blocked with 1% BSA-supplemented PBS. A serially diluted standard MAb of each isotype (Ancell, Bayport, MN) and diluted serum samples were captured on the wells. The captured Abs were detected with alkaline phosphatase-conjugated isotype-specific goat IgG using a 1420 ARVO SXc (Wallac, Turku, Finland) luminometer.
Determination of antibody production in immunized mice. Eight-week-old mice were immunized after preimmune serum was obtained. Freund's complete adjuvant containing 100 µg of dinitrophenyl (DNP)-keyhole limpet hemocyanin (KLH) was used for primary T-dependent immunization by intraperitoneal injection, and a second boost was performed with the antigen in incomplete adjuvant. For T-independent immunization, 10 µg of DNP-Ficoll in PBS was injected. The anti-DNP titer was measured essentially as above, except that DNP-BSA was used for antibody capture, and a mixed pool of DNP-KLH-immunized serum was used as the standard. The value relative to that of the pooled serum was used to normalize the values obtained from different plates.
B-cell proliferation analysis. In 96-well plates, 100-µl aliquots of B cells at 1 x 105 cells/ml were stimulated in RPMI 1640 medium containing the indicated concentrations of stimulation reagents. After 24 h of incubation, bromodeoxyuridine (BrdU) was added, and the incubation was continued overnight. Incorporated BrdU was detected using a chemiluminescent ELISA system (Roche Diagnostic GmbH) with an 1420 ARVO SXc luminometer.
Immunoblotting and immunoprecipitation of CD22. Splenic B cells were adjusted to 5 x 105 cells/50 µl in RPMI 1640 medium. After preincubation at 37°C, the B cells were stimulated with F(ab')2 anti-mouse IgM (10 µg per 5 x 105 cells) at 37°C. To detect the pattern of tyrosine phosphorylation, cells were lysed in sodium dodecyl sulfate-polyacrylamide gel electrophoresis sample buffer (50 mM Tris-HCl [pH 7.6], 2% sodium dodecyl sulfate, 0.1% pyronin G, 10% glycerol, 2-mercaptoethanol). For immunoprecipitation studies, the stimulated B cells were lysed in NP-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 25 mM HEPES [pH 7.4], 5 mM NaF, 2 mM sodium orthovanadate, protease inhibitor cocktail [Nacalai Tesque]), and CD22 was immunoprecipitated with anti-CD22 (Cy34.1) antibody and protein G-Sepharose beads (Amersham Biosciences). In the CD22 immunoprecipitation studies, after a probing step with PT-66, the membrane was reprobed with anti-CD22 polyclonal antibody.
Experimental animals. The studies presented here were performed in accordance with animal care guidelines and were approved by the animal experimental committee of Kyoto University Graduate School of Biostudies.
Microarray data accession numbers. The GEO platform (GPL3465) and experimental results are registered in the Gene Expression Omnibus database under accession number GSE4407.
|
RESULTS |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2,3-linked Sia, had no effect (Fig. 1B). To assess the role of Sia and other sugars in GL7 reactivity, we analyzed the inhibitory effects of sugar on GL7 binding. The results clearly showed specificity of Neu5Ac for inhibition (Fig. 1C), and the inhibition was dependent on the Neu5Ac concentration (Fig. 1D). Neu5Ac is a major form of Sia in human cells. GL7 binding was decreased with a metabolic N-glycosylation inhibitor, tunicamycin (see Fig. S1 in the supplemental material). Multiple bands were detected in immunoblotting experiments using the membrane fraction of Daudi cells (Fig. 1E). Thus, it is likely that GL7 recognizes some glycan epitopes, including Sia, rather than some specific protein(s).
|
2,6 linkage (42), and B cells have been shown to express this enzyme (20, 64). This indicates that the terminal transferase reaction by ST6GAL1, but not the supply of the substrate, is the rate-limiting step in GL7 epitope biosynthesis in these cells. Interestingly, a negative correlation was found between GL7 staining and the expression of SIAE, a gene encoding Sia 9-O-acetylesterase (Table 1). Although Sia 9-O-acetylesterase cleaves the O-acetyl group of Sia, SIAE is expressed in cell types expressing its substrate, 9-O-acetylated Sia (57). If the degree of 9-O acetylation were to correspond with the level of SIAE expression, GL7 binding might be negatively affected by 9-O-acetyl modification, similar to CD22 (56).
|
|
2,6-linked Sia on N-glycan and that the expression of the GL7 epitope on human B cells depends mainly on ST6GAL1 expression. To evaluate these findings, we explored the ST6GAL1 expression dependence of GL7 epitope expression. CHO-K1 cells are known to lack 2,6-linked Sia on their cell surfaces. As expected, the parental CHO-K1 cells were GL7 negative (data not shown), as were vector-transfected CHO-K1 cells (Fig. 2B). In contrast, rat ST6GAL1 (rSt6galI)-transfected CHO-K1 cells showed a marked increase in GL7 staining (Fig. 2B). The increase in GL7 staining upon rSt6gal1 expression coincided with the increase in staining by SSA, a plant lectin which reacts with Sia 2,6-Gal/GalNAc on glycans. As CHO-K1 cells are nonimmune cells, GL7 seemed to recognize 2,6-linked Neu5Ac-containing sugar chains on various proteins. Immunoblotting analysis of these stable clones further clarified that the introduction of rSt6gal1 was sufficient to give rise to bands on the blot. The membrane fractions of both CHO-K1 stable clones and human B-cell lines resulted in multiple bands (data not shown).
Glycan-binding assay of GL7.
To confirm that GL7 is an antiglycan antibody that recognizes 2,6-linked Sia and also to determine the fine specificity of the epitope, we examined GL7 binding to various glycan probes (65) by ELISA. GL7 bound to LSTc (Neu5Ac2-6Galß1-4GlcNAcß1-3Galß1-4Glc) but not to its structural isomer with 2-3 linked Neu5Ac, LSTa (Neu5Ac2-3Galß1-3GlcNAcß1-3Galß1-4Glc) (Fig. 2C). Interestingly, GL7 did not bind to Neu5Ac2-6Galß1-4Glc (sialyllactose) in spite of the existence of 2,6-linked Neu5Ac in the probe. The glucose (Glc) of the reducing terminal was destroyed during probe preparation for coupling with streptavidin. Thus, it is likely that the structure of Neu5Ac2-6Gal is not sufficient for GL7 binding but that the binding requires at least a trisaccharide for optimal recognition or GlcNAc in the underlying lactosamine. Taking all of the results into consideration, we concluded that GL7 recognizes 2,6-linked Sia-containing glycan chains that are often found on N-glycans of various proteins.
A shift in the major Sia species, Neu5Gc to Neu5Ac, in the mouse germinal center reaction.
It was still not clear why GL7 failed to react with mouse mature B cells, given that these cells abundantly express 2,6-linked sialoglycans, as St6gal1 is also expressed in these cells (20, 64). The dominant difference in sialylation between mice and humans occurs in the Sia modification at the C-5 position (60). Humans predominantly express Neu5Ac, whereas the major Sia in mice is Neu5Gc (Fig. 3A). It is possible that the change in GL7 reactivity could be a consequence of the change in sia modification. Neu5Gc modification in biosynthesis is regulated by the Cmah reaction in the cytosol, which metabolically gives rise to the donor, CMP-Neu5Gc, for a subsequent sialyltransferase reaction(s) in the Golgi apparatus (Fig. 3B) (24, 25). We therefore asked whether mouse B cells undergo a change in Sia species, from Neu5Gc to Neu5Ac, in GL7-positive cells. We first stained the germinal centers with GL7 and the lectin domain of mouse CD22 (mCD22-Fc), because mouse CD22 demonstrates a marked preference for Neu5Gc-bearing over Neu5Ac-bearing 2,6-linked sialoglycan ligands (26, 44, 50). As shown in Fig. 3C, in the SRBC-immunized mouse spleen, GL7-positive germinal centers were specifically excluded by mCD22-Fc recognition. This complementarity of staining appeared to be the result of the probe preferences, Neu5Ac for GL7 and Neu5Gc for mCD22-Fc, respectively. We then assessed Cmah expression and the Neu5Ac-Neu5Gc ratio in GL7-positive germinal center B cells. Germinal center (GL7-bound) cells showed severely reduced expression of Cmah, and this reduction coincided with the loss of Neu5Gc in the membrane fraction of the cells (Fig. 3D). In contrast, GL7-negative SRBC-immunized B cells were not significantly different from nonimmunized splenic B cells. Thus, the gain of GL7 staining reflected the loss of the CD22 ligand in germinal center B cells due to the repression of Cmah.
|
2,6-linked Sia-containing glycan probed with SSA, St6gal1 expression showed a subtle reduction in activated B cells after 48 h (Fig. 4A and B). Cmah reduction appears to play a prominent role in the appearance of the GL7 epitope in activated B cells. Retrovirus-mediated ectopic Cmah expression consistently reduced the expression of the GL7 epitope in LPS-stimulated B blasts (Fig. 4C), further confirming the responsibility of Cmah for the repression of the appearance of the GL7 epitope. After 48 h of stimulation with LPS, Gapdh expression increased by about 30% (Fig. 4A). This may be attributable to the blastic transformation of LPS-stimulated proliferating B cells (B blasts), which produce much more cytosolic space and subsequent metabolism than resting B cells. GL7 staining of LPS-stimulated B cells showed heterogeneity in the degree of staining. Thus, cells used to prepare RNA for this real-time PCR experiment were a mixture of GL7high and GL7low cells. When these finding are taken into consideration, the reduction of Cmah expression in GL7high germinal center B cells could be more drastic. The expression of Cd22, an 2,6-linked Neu5Gc binding protein, on B cells was reduced to around 40% after 48 h, even though its cell surface expression was still comparable to that of unstimulated cells in flow cytometry (Fig. 4A and B).
|
|
2,6-linked Neu5Ac on LacNAc (Fig. 2C). Serum Ig measurements using the sandwich ELISA method revealed a significant (P = 0.074) increase in the serum IgG1 level of the Cmah-null population (Table 2).
|
|
|
|
2,3-linked Neu5Ac on galactose as a ligand (10), and CD22 (Siglec-2), which has a strong preference for Neu5Gc over Neu5Ac in the 2,6 linkage to LacNAc in mice (3, 26, 44, 50). To explore the change in ligand expression for Siglecs in Cmah-null mice, we prepared Siglec-Fc fusion probes that were free from intramolecular sialylation. In null B cells, the expression of the CD22 ligand was reduced roughly 20-fold compared with that in wild-type cells (Fig. 9A). We also histochemically examined the expression of the CD22 ligand on spleen sections from Cmah-null mice. Regardless of immunization, the mCD22-Fc probe failed to detect any staining in the sections of Cmah-null spleen, as in the germinal centers of immunized wild-type mice (Fig. 9B). Therefore, Cmah disruption caused the reduction of the optimal ligand for CD22. At the same time, ligand expression for sialoadhesin was greatly increased in Cmah-null mice (Fig. 9A). Sialoadhesin is expressed on macrophages, whereas CD22 is expressed on B cells. Ligand(s) for Siglec-G, another Siglec molecule presumably expressed on B cells, was not detected on B cells (data not shown); thus, the Siglec-related effects in Cmah-null B cells could be a loss of CD22 ligand.
|
|
DISCUSSION |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2-6LacNAc-containing N-glycan, which is prominently expressed in activated B cells upon the repression of Cmah. Gain of GL7 epitope expression coincided with the loss of optimal ligand expression of CD22 in germinal center B cells, presumably centrocytes. Considering the rather strong degree of GL7 positivity in germinal center B cells in comparison with in vitro stimulated B-cell blasts, the degree of Cmah reduction might have been severe in these cells. In general, it is thought that Neu5Gc is easy to accumulate but difficult to turn over in cells. This is attributable to the one-way direction of the metabolic pathway; Neu5Gc is biosynthesized by Cmah from Neu5Ac (24, 36, 54), whereas no conversion activity was found to biosynthesize Neu5Ac from Neu5Gc. Therefore, the reduction of Neu5Gc found in the GL7-enriched germinal center cells is remarkable. Such rapid clearance of Neu5Gc could be attributable to several characteristics of germinal center cells. Most importantly, as shown in Fig. 3D, these cells repressed Cmah, the enzyme responsible for the de novo biosynthesis of Neu5Gc. Moreover, because lymphocytes are small cells with limited cytosolic space, the cytosolic pool of Sia in these cells is likely limited and easily turned over. In addition, centrocytes undergo extremely fast cell cycles (66), which probably leads to rapid passive dilution of the cytosolic pool in these cells. At the same time, new protein synthesis should be a primary event that happens in germinal center B cells, as shown by cDNA microarray analysis (51). The transcriptional repression of Cmah, together with these features of germinal center cells, could contribute to the efficient conversion of the major Sia species from Neu5Gc to Neu5Ac. Negative regulation of B-cell activation by Cmah and its product, Neu5Gc. To clarify the biological role of Neu5Gc in vivo, we disrupted the Cmah gene in mice and examined their B-cell activation phenotypes. Cmah-null mice showed a hyperreactive B-cell phenotype to T-independent stimulation. In contrast, the T-dependent immunization response was similar to that in wild-type mice. This is consistent with the findings that Cmah expression is severely repressed in the germinal centers of wild-type spleen upon T-dependent immunization and that Cmah-null mice could develop follicles stained with PNA, another marker for germinal centers. Forced expression of Cmah caused repression of the proliferative response of Cmah-null B cells, indicating that Neu5Gc-containing sialoglycan functions to suppress B-cell reactivity though the mechanism is still unknown. This suppression via Neu5Gc-containing sialoglycan appears to be canceled by Cmah repression in germinal center B cells that are "activation committed" or "activation competent." The hyperreactive B-cell phenotypes observed in Cmah-null B cells could mirror differences in cellular reactivity between germinal center and nongerminal center B cells, as indicated by differential cell surface expression of the GL7 epitope (5).
Possible change in sialoglycan-receptor interaction in Cmah-null mice.
As Cmah disruption results in a single oxygen atom change in these mice, it is expected that this mutation leaves both the Sia amount and Sia linkage intact in terms of sialoglycans, which could change the stability or turnover of the proteins modified with Sia (14). Although only limited information is available, sialyltransferases that biosynthesize sialylated glycans in the Golgi apparatus do not show strong preferences for CMP-Neu5Ac or CMP-Neu5Gc as substrates (59). When we probed linkage-specific protein sialylation by using 2,6-linked Sia-binding plant lectins such as Sambucus nigra agglutinin, we did not observe a change (data not shown). Thus, the molecular event affected in Cmah-null mice is likely to be lectin recognition of a single oxygen atom on sialoglycans expressed on the cell surface, although a single responsible lectin may not explain the phenotype. One of the candidate lectins as the receptor of sialoglycans is the Siglec family (9, 12, 62), though a yet-to-be-characterized Sia-binding molecule could be affected.
When ligand expression for Siglecs was detected using Siglec-Fc probes, Cmah-null mice lost optimal ligand expression for CD22 (Siglec-2). The ligand function of CD22 in a mouse model has been addressed in two different ways. One study was done using St6gal1-knockout mice (20), and another study analyzed gene-targeted mice expressing mutant CD22 molecules that do not interact with ligands (43). The phenotypes found in Cmah-null mice are considerably different from these two previous studies; therefore, Cmah-null phenotypes might be caused by the combination of loss/gain of a Sia-mediated interaction. Additional studies using a combination of various knockout strains related to sialoglycan recognition are required to address such possibilities.
Apart from the phenotypic contribution of CD22 to the assays in the present study, CD22 ligand expression is not static but is, instead, a regulated event during in vivo B-cell activation. We showed that mCD22-Fc probe staining was downregulated in germinal centers. Moreover, it was reported that in vitro activated human B cells unmask CD22 from a cis-ligand (45). Thus, the regulation of CD22 ligand expression could be an important event to modulate B-cell activation in vivo.
Loss of Neu5Gc in relation to human deficiency for the CMAH gene. Homo sapiens is the sole mammalian species that lacks Neu5Gc expression throughout the body; indeed, Neu5Gc is antigenic to humans (31). This is a striking difference between humans and chimpanzees, which express Neu5Gc as the major species of Sia throughout their bodies. Recently, it was shown that, unlike gene expression in the extant great apes, the CMAH gene is inactivated in humans (6, 22). Here, we demonstrated that Cmah is the sole enzyme responsible for the production of Neu5Gc in cells since our mouse model reproduced the human-like deficiency in Neu5Gc biosynthesis. This result confirmed that a genetic mutation in the human lineage caused the lack of Neu5Gc in humans.
Sia is commonly used in the host recognition system of microbes, and human-specific microbes are reported to recognize epitope(s) containing Neu5Ac on human cells. The mouse described here is thus the first mammalian line that could be used as an animal model system to assess Sia-targeted human infectious diseases (15).
|
ACKNOWLEDGMENTS |
|---|
This work was supported by CREST, Japanese Science and Technology; a grant-in-aid program from the Ministry of Education, Culture, Sports, Science, and Technology of Japan; and RIKEN.
|
FOOTNOTES |
|---|
Published ahead of print on 12 February 2007.
Supplemental material for this article may be found at http://mcb.asm.org/.
Present address: Research Center for Glycobiotechnology, Ritsumeikan University, Kyoto, Japan.
|
REFERENCES |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2. Baum, L. G., K. Derbin, N. L. Perillo, T. Wu, M. Pang, and C. Uittenbogaart. 1996. Characterization of terminal sialic acid linkages on human thymocytes. Correlation between lectin-binding phenotype and sialyltransferase expression. J. Biol. Chem. 271:10793-10799.
3. Blixt, O., B. E. Collins, I. M. Van Den Nieuwenhof, P. R. Crocker, and J. C. Paulson. 2003. Sialoside specificity of the Siglec family assessed using novel multivalent probes: identification of potent inhibitors of myelin associated glycoproteins. J. Biol. Chem. 278:31007-31019.
4. Buhl, A. M., and J. C. Cambier. 1997. Co-receptor and accessory regulation of B-cell antigen receptor signal transduction. Immunol. Rev. 160:127-138.[CrossRef][Medline]
67-kDa band (arrowhead) in wild-type and heterozygote livers represents the signal of Cmah, which is not detectable in Cmah-null liver samples. (D) Loss of Neu5Gc production throughout the body in mutated mice. Acid-hydrolyzed Sia from the indicated tissues was derivatized using DMB, and the ratios of Neu5Ac and Neu5Gc to total Sia were measured by reverse-phase HPLC. Solid columns represent the percentage of Neu5Gc in various tissues, and open columns represent the percentage of Neu5Ac. The detection limit for Neu5Gc in this assay was around 0.1%. (E) Flow cytometry profile of Cmah-null mice splenocytes. The expression of IgM, MHC-II (I-A and I-E), and CD22 on splenocytes from wild-type and Cmah-null mice was detected by flow cytometry. In anti-MHC-II and anti-CD22 staining, splenocytes were costained with anti-B220, a marker for B cells. (F) Strong expression of the GL7 epitope on Cmah-null mice B cells. Splenocytes from wild-type and Cmah-null mice were costained with anti-B220 and GL7 and subjected to flow cytometry.
