Glycotranscriptome study reveals an enzymatic switch modulating glycosaminoglycan synthesis during B-cell development and activation



B-cell fate and responses are modulated by soluble mediators and direct cellular interactions. Migration properties also vary during differentiation, commitment and activation. In many cells, modulation of responses to stimuli involves cell surface glycans, whose architecture depends on the simultaneous expression of multiple enzymes. By looking at the glycosylation-related gene expression patterns among B-cell populations, we determined in this study that the strongest variations were observed for CSGalNAcT-1 and EXTL1. These are enzymes involved in the biosynthesis of alternative forms of glycosaminoglycans (GAGs), namely chondroitin sulfate and heparan sulfate, respectively. These two enzymes showed inverse fluctuations in progenitors, resting B cells and activated B cells, suggesting a developmentally regulated switch between chondroitin and heparan sulfate synthesis. To explore whether these variations contributed to optimal B-cell differentiation, we overexpressed EXTL1 in the B-cell lineage of transgenic mice, yielding a partial differentiation blockade at the pro-B to pre-B transition. In the periphery, this defect was almost fully compensated for in vivo, with normal-size B-cell compartments and normal serum immunoglobulin levels in the transgenic EXTL1 mice. The peripheral B cells from EXTL1 transgenics were only affected with regard to their in vitro responses to polyclonal activation, showing reduced proliferation. Together the data suggest that despite their low amounts in lymphocytes, the heparan sulfate chains decorating the endogenous GAGs appear to be regulators of B-cell physiology.


B-cell differentiation and responses to antigens rely upon their integration in a complex functional network of stromal cells, immune cells, soluble cytokines and chemokines. In most tissues, biological information to control such interactions is not only stored in protein sequences but also in post-translational modifications. In particular, the structure of accessible carbohydrates present at the cell surface is part of the cell ‘glycome’ and is determined by the coordinated expression of genes including glycosyltransferases, glycosidases and nucleotide-sugar transporters. These enzymes and transporters participate in the functional regulation of a number of other proteins and receptors, such as members of the Wnt and Notch families, galectins, membrane lectins or selectins, whose activities are dependent upon their glycosylation or the glycosylation of their ligands 1–3. Glycosaminoglycans (GAGs) constitute a specific type of glycans of major importance since, depending upon their nature, electrostatic charge and abundance, they modulate the locally available concentrations of various intercellular mediators in the vicinity of the cell surface. Proteoglycans (PGs) are macromolecules composed of a central core protein to which one or more GAG chains are covalently attached. These linearly elongated polysaccharides are abundantly found at the surface of various cells and in the extracellular matrix where they modulate cell adhesion, migration and proliferation 4, 5. By forming a glycocalyx in close proximity to the plasma membrane, PGs act as a sponge for soluble mediators and promote interactions between cellular receptors and factors such as fibroblast growth factor (FGF), hepatocyte growth factor (HGF), transforming growth factor (TGF-β), tumor necrosis factor (TNF), interleukins (IL-3, IL-6, IL-7, IL-8, IFN-γ, etc.) or chemokines (CXCL4, stromal cell-derived factor 1, RANTES, etc.), thus modulating their activities 6–13.

GAGs are composed of repeated [N-acetylhexosamine–hexuronic acid] disaccharide units and are divided into two groups: (i) galactosaminoglycans, which include chondroitin sulfate (CS) and dermatan sulfate and (ii) glucosaminoglycans, which include heparan sulfate (HS), heparin, hyaluronan and keratan. The CS disaccharide repeated unit is made up of N-acetylgalactosamine (GalNAc) and glucuronic acid (GlcA), whereas the HS unit includes N-acetylglucosamine (GlcNAc) and GlcA. CS and HS biosyntheses are initiated by the transfer of GalNAc or GlcNAc, respectively, to the linkage region of a [GlcA-β1,3-Gal-β1,3-Gal-β1,4-Xyl] tetrasaccharide attached to a serine residue of the core protein (Fig. 1A). The production of GAGs mostly takes place in the Golgi apparatus, where their initiation, elongation and modification are controlled by glycosyltransferases and carbohydrate modifying enzymes such as sulfotransferases.

Figure 1.

Schematic roles and variations in the expression of CSGalNAcT-1 and EXTL1A. (A) Schematic diagram detailing the catalytic activities of the glycosyltransferases involved in HS/CS synthesis. Xylosyltransferase (XYLT1 and XYLT2), β1,4-galactosyltransferase VII (B4GalT7), β1,3-galactosyltransferase VI (B3GalT6) and β1,3-glucuronyltransferase-I (GlcAT-I) sequentially build the tetrasaccharide covalently linked to serine residues of the core protein. (Top) The precursor chain of HS is made up of alternating GlcA and GlcNAc residues transferred by the HS polymerases EXT1 and EXT2, which have both GlcAT and GlcNAcT activities, whereas EXTL1, 2, and 3 have only HS-α1,4-N-acetylglucosaminyltransferase activity in chain elongation or initiation. HS can be further modified by N-deacetylation/N-sulfation (NDST), epimerization (GLCE) and O-sulfation (HS2ST, HS6ST and HS3ST). (Bottom) The precursor chain of CS includes alternating GlcA and GalNAc assembled by the chondroitin sulfate synthases, CHSY and CSS2, which also exhibit GalNAcT and GlcAT activities in chain elongation. They work with β1,3-glucuronyltransferase (CSGlcA-T). CS-β1,4-N-acetylgalactosaminyltransferases 1 and 2 (CSGalNAcT-1 and CSGalNAcT-2) exhibit GalNAcT activity in both initiation and elongation. CS is also O-sulfated by C4ST, C6ST. (B) CSGalNAcT-1 and EXTL1 expression during B-cell differentiation. Gene expression in RNA samples representative of the various B-cell maturation stages or from activated B cells was studied by quantitative real-time PCR. The vertical axis is a logarithmic scale. The median line (marked 0) corresponds to expression in resting B cells which was always taken as a reference for the evaluation of the relative expression (see Materials and methods for further details).

The critical enzyme of CS biosynthesis is chondroitin sulfate N-acetylgalactosaminyl-transferase-1 (CSGalNAcT-1), active in both CS initiation and elongation 14. On the other hand, HS synthesis involves glycosyltransferases of the exostosin (EXT) family, named according to their association with hereditary exostoses affecting bone formation 15. Targeted deletion of the EXT1 co-polymerase is embryonically lethal by preventing gastrulation, revealing a major role of HS in development 16. EXTL1 (or EXT-like 1), an α1,4-N-acetylglucosaminyl-transferase, is involved in the elongation of the HS chain 17. All these enzymes have an N-terminal transmembrane domain and reside in the Golgi apparatus. Except for the in vitro biochemical characterization of several enzymes, the precise in vivo regulation of CS and HS biogenesis and their respective functions are still incompletely understood. In lymphoid cells, while the global amount of GAGs is very low and escape detection by direct biochemical methods, a variety of PG core polypeptides were indirectly shown to carry HS/CS chains and are differentially expressed, including syndecans, glypican, serglycin and CD44-HS 18–22.

In this study, we first tried to check whether glycome variations could be appreciated at the transcriptional level during B-cell differentiation. One hundred sixty-two genes encoding both key enzymes of glycome synthesis and proteins either linked to carbohydrate recognition or functionally regulated by glycosylation were studied together with control genes known to vary during B-cell ontogeny. This survey allowed us to draw a short list of genes with variations sufficient for the delineation of a glycotranscriptome signature specific for each B-cell differentiation stage. Within this list, very strong and correlated variations were observed for two transcripts encoding CSGalNAcT-1 and EXTL1, two key enzymes of GAG synthesis involved in CS and HS biogenesis, respectively. These transcripts showed inverse variations from progenitors to mature and activated B cells. Although the level of intrinsic GAG production by lymphocytes is usually considered as very low, this observation suggested that the quantity and quality of endogenously synthesized GAGs may play a functional role and prompted us to additional experiments by trying to deregulate the equilibrium of these enzymes and to get insights into their potential role in the B-cell lineage. We thus designed and studied transgenic mice in which EXTL1 transcripts were overexpressed in a B-cell specific manner, with the aim of disturbing the apparent regulated balance in the expression of HS and CS biosyntheses enzymes during B-cell differentiation.


Gene array study

We used high-throughput real-time quantitative PCR technology to assess simultaneously the expression of 162 genes chosen after the involvement of their products either directly into glycosylation processes or into polysaccharide binding (galectins, siglecs, etc.).

An overview of global gene expression data in purified B lineage cells corresponding to various differentiation stages and lymphoid tissues is provided by a two-way hierarchical clustering analysis of genes after expression level filtering of those 116 genes expressed above a threshold cycle (Ct) <36 in at least 50% of the samples studied (Supporting Information Fig. 1). The 29 samples studied included FACS-sorted primary cells and B lineage cell lines together with control thymocytes and stromal cells.

Among the genes removed from the analysis because of their mostly low or undetectable expression were included galactosyltransferases (such as B3GalT1 and B3GalT5), glucuronyltransferases (B3GAT1 and B3GAT2), sulfotransferases (CHST4, CHST5, CHST8, HS3ST3a, HS6ST2 and HS6ST3), N-deacetylase/N-sulfotransferases (NDST 3 and 4), galectins 7 and 12, enzymes synthesizing galectin ligands (MGAT5, GCNT2, GCNT3), Siglec Mag, sialyltransferases (ST6Gal2, SIAT7A) and CSPG serpin, all of which appeared as virtually not expressed in the B-cell lineage. Three exceptions of rarely expressed genes were not further studied but could deserve special attention: SIAT7A was found exquisitely expressed in pro-B cells and lymph nodes, galectin 2 was found in stromal cells and Peyer patch B cells, while beside B cells, wnt5a was only found in stromal cells.

It was possible to cluster the 116 informative selected genes into functional subfamilies of genes encoding enzymes of GAG metabolism, galectins, siglec and genes involved in the Notch or Wnt signaling pathways. Gene variations were studied in 24 samples representative of different B-cell populations. We then applied a second filter with variations in the relative quantity of template (RQ) above eight in at least two samples to select the most important genes variations along B-cell differentiation. This array proved able to provide signatures typical for each cell subtype (Fig. 2).

Figure 2.

Strategy and confirmation of the generation of transgenic mice overexpressing EXTL1. (A) Diagram of the vector construction (not to scale) showing the cDNAs of interest inserted in-between the pVH promoter and the intronic Eμ enhancer, and followed by a downstream copy of all the enhancers of the 3′IgH regulatory region. The complete expression cassette was flanked with β-globin insulators. (B) Southern analysis of HindIII-digested ES cell clone genomic DNA with the respective cDNA probe.

Detailed analysis of CSGalNAcT-1 and EXTL1 expression

Two genes involved in GAG synthesis (Fig. 1A) more specifically retained our attention due to their marked variations in the same samples but in opposite directions. One encoded the HS-elongating enzyme EXTL1 and was mostly expressed in resting B cells, whereas the other encoded the CS biosynthesis-initiating enzyme CSGalNAcT-1 and was shut-off in resting B cells but expressed in B-cell progenitors, in activated B cells and in plasma cells (Fig. 1B).

Data obtained from RT-PCR arrays regarding these genes were confirmed by repeated individual quantitative RT-PCR assays. Both EXTL1 and CSGalNAcT-1 transcripts were cloned from 129 cells and repeatedly sequenced. Their sequences revealed polymorphisms by comparison with the previously reported C57Bl/6 sequences. EXTL1 differs by seven nucleotide substitutions (five of them resulting in amino acid replacements: 32S→32C; 204I→204Y; 361V→361I; 450P→450L; 540D→540E) from a previously published C57/Bl6 cDNA (GenBank NM_019578), whereas CSGalNAcT-1 differs by three nucleotide substitutions from the previously published C57/Bl6 cDNA [NM_172753-Mus musculus RIKEN cDNA 4732435NO3Rik gene (4732435NO3Rik)], only one of them resulting in a replacement: 393V→393I.

Generation of 5′-Eμ-pVH-EXTL1-LCR-3′ transgenic mice

The EXTL1 cDNA was cloned from B lymphocytes and inserted into an expression vector previously demonstrated to yield B-cell-specific expression from pro-B to terminally differentiated plasma cells 23. The amplified open-reading frame was cloned in-between the Eμ/pVH cassette and the 3′IgH enhancers (Fig. 3A). While all these enhancers have strong and synergic activities, we also flanked our expression cassette with insulators known to protect such transgenes from position effects and to allow the combined 3′IgH regulatory elements to altogether behave as a locus control region 24–27. Two independent ES clones carrying the expression cassette were microinjected into blastocysts to derive chimeras and transgenic mouse strains. Independence of these clones was checked by sequencing junctions of transgenes with chromosomal DNA after amplification of both insertion sites (clone ♯7 showed insertion on chromosome 4, in the fourth intron of the nuclear receptor nrf4 gene, clone 10 showed insertion on chromosome X, in a non-coding region downstream of the phosphoglycerate kinase pgk1 gene; Supporting Information Fig. 2). The same B-cell lineage alterations were noticed in both independent resulting mouse strains. The progeny of transgenic founders was followed for the presence of the transgene by Southern blot (Fig. 3B).

Figure 3.

Glycotranscriptome signatures characterizing B-cell subpopulations. Two-way hierarchical agglomerative clustering was applied to the gene expression matrix consisting of 24 B-cell samples (with sorted splenic IgD+ resting B cells taken as the reference sample) and the selection of those 162 analyzed genes with significant expression in at least 50% of the samples (Ct <36) and with major variations of expression (above 8) between at least two different B-cell samples to discriminate the most important variations. The input consisted of the relative quantity of template (RQ) for each gene and sample. Selected genes included 14 (out of 43 initially assayed) involved in GAG biosynthesis, 5 (out of 20 assayed) related to the galectin family, 4 (out of 7 assayed) encoding siglecs, 12 (out of 38 tested) involved in the Notch signaling pathway and 9 (out of 35 tested) involved in the WNT signaling pathway [black color denotes expression similar to the resting B cells calibrator sample (RQ=1); silver denotes no detectable expression; dark red and bright red, respectively, denote high (RQ>1) or very high (RQ>32) expression by comparison to the calibrator sample; dark green and bright green, respectively, denote low (RQ<1) or very low (RQ<1/32) expression by comparison to the calibrator sample].

EXTL1 expression in transgenic mice

We first used real-time RT-PCR to evaluate transgene expression in whole lymphoid tissues from 4- to 6-wk-old transgenic mice (5–10 mice analyzed for each condition, Fig. 4A). EXTL1 mice showed high levels of EXTL1 transcripts in BM (relative expression of 1479±772) and spleen (relative expression of 48.6±29.2) compared with WT mice (relative expression of 1). The expression of the protein was evaluated by ELISA in cell lysates from sorted spleen B cells and shown to be significantly increased (Fig. 4B).

Figure 4.

EXTL1 levels are increased in transgenic mice. (A) Gene expression of EXTL1 was analyzed by quantitative real-time PCR on RNA samples isolated from BM and spleen tissues from EXTL1 mice (n=7) and WT mice (n=5); data are mean±SEM and the vertical axis is on a logarithmic scale. (B) EXTL1 protein expression in B-cell lysates from EXTL1 transgenic and WT littermates was evaluated by ELISA. Optical density was measured at 492 nm. Statistical significance: *p<0.05, **p<0.01, ***p<0.001 Mann–Whitney test comparing transgenic versus WT animals.

Partial B-cell development blockade in EXTL1 transgenic mice

We analyzed B-cell progenitors and mature B cells in 4–8 wk transgenics. Spleens, lymph nodes and Peyer's patches were of normal sizes.

Early B-cell compartments were analyzed by cell flow cytometry using a set of specific surface markers. The EXTL1 transgenic strain showed a global decrease of B lineage cells in the BM by comparison to WT littermate controls (about 6 million total cells versus about 10 million, respectively). The balance between maturation stages was also altered, and the limited B-cell pool included a normal absolute quantity (resulting in an increased percentage) of the CD43+/B220+ pro-B/early pre-B compartment. Beyond this stage, there was a roughly twofold decrease in the absolute number of CD25+/B220+ late pre-B cells, then resulting in a decreased percentage. These anomalies were associated with a reduction in the immature IgMlow B220+ B-cell compartment, while the recirculating IgMhigh B220+ mature compartment was normal (Fig. 5A and Table 1). To check whether the addition of soluble GAG related to HS would also be able to affect the pro-B to pre-B transition, we carried out in vitro differentiation assays of WT BM pro-B cells. Exogenously added heparin affected neither pro-B-cell proliferation under a high IL-7 concentration nor the pre-B-cell differentiation induced by IL-7 deprivation (Supporting Information Table 1). Altogether, these data indicated a partial developmental blockade at the pro-B to pre-B transition, specifically resulting from endogenous expression of EXTL1 in the B-cell lineage of transgenic animals.

Figure 5.

Partial blockade of pro-B to pre-B transition in transgenic mice. (A) BM cells from 4- to 8-wk-old WT (left) or EXTL1 mice (right) were labeled with PC5-conjugated anti-B220 and PE-conjugated anti-CD117 (top), anti-CD43 (second panel from the top), anti-CD25 (third panel from the top) or FITC-conjugated anti-IgM and PE-conjugated anti-IgD (bottom) to stain pro-B, proB/early preB, late pre-B, immature and recirculating B cells. (B) Splenocytes from 4- to 8-wk-old WT (left) or EXTL1 mice (right) were labeled with PC5-conjugated anti-B220, PE-conjugated anti-IgD and FITC-conjugated anti-IgM antibodies (top). The amount of transitional and mature B cells in the spleen was assessed using AA4.1 labeling (second panel from the top). The ratio of MZ B cells (B220+ CD23low/–CD21high) and follicular B cells (B220+ CD23highCD21int) is shown (third panel from the top). Splenocytes were then stained with FITC-conjugated anti-mouse CD5 and PC5-conjugated anti-B220 to determine the percentage of B1 cells (bottom). Labeled cells were gated first on lymphocytes according to FS/SSC criteria and secondly on B220+ cells. Numbers represent the percentages of cells within gates. The plots are representative of at least three similar experiments.

Table 1. Relative values of the various lymphoid populations in BM and spleena)
  WTEXTL1 clone n 7EXTL1 clone n 10
Total B-cell (B220+) amountN=26/17/49.71±0.866.33±1.05*3.88±0.54*
B-cell subpopulations among B220+ B cells
Pro-B (ckit+ B220+)N=26/17/46.83±0.65% (0.62±0.07)8.30±0.66% (0.46±0.06)8.82±0.82 (0.33±0.025)
Pro-B/pre-B (CD43+ B220+)N=26/17/425.50±1.49% (2.34±0.21)31.63±2.74%* (1.81±0.3)37.4±7.14* (1.35±0.16)
Pre-B (CD25+ B220+)N=26/17/439.53±1.82% (4.16±0.53)32.93±2.77%* (2.21±0.48*)20.25±4.43*** (0.85±0.29*)
Immature B cells (B220low IgMlow)N=26/17/418.91±1.02% (1.78±0.15)18.53±1.31% (1.10±0.17**)18.35±2.78 (0.74±0.18*)
Recirculating B cells (B220high IgMhigh)N=26/17/49.63±0.89% (0.830±0.09)11.82±0.82% (0.69±0.09)13.35±1.47 (0.49±0.019)
T cell (CD5+ B220) amount in spleenN=20/15/420.34±2.8421.57±4.1115.67±2.8
Total B-cell (B220+) amountN=25/22/617.41±1.9013.62±1.4417.13±3.49
B-cell subpopulations among B220+ B cells
Resting B cells (IgD+ IgM+ B220+)N=25/22/680.45±1.03% (14.21±1.59)76.96±1.12%* (10.55±1.15)82.35±1.81 (14.30±3.1)
Transitional B cells (AA4.1+ B220+)N=22/22/618.23±1.48% (3.48±0.49)16.80±1.70% (2.16±0.32*)13.81±1.4 (2.42±0.62)
T1 (AA4.1+ B220+ CD23–/low)N=22/22/67.16±0.74% (1.39±0.23)6.76±0.78% (0.88±0.15)5±0.64 (0.88±0.24)
T2 (AA4.1+ B220+ CD23+)N=22/22/611.11±1.12% (2.08±0.32)10.09±1.13% (1.28±0.19*)8.58±0.89 (1.49±0.38)
Follicular B cells (CD23+ CD21+ B220+)N=25/22/667.62±1.83% (12.29±1.49)63.78±1.71% (8.85±1.04)74.78±1.79 (12.99±2.88)
MZ B cells (CD23–/low CD21+ B220+)N=25/22/69±0.98% (1.34±0.14)10.49±1.33% (1.36±0.22)9.64±1.4 (1.56±0.26)
  1. a

    a) The second column indicates the number (N) of WT/mutant animals from strain # 7/and from strain # 10, used for the calculation. Percentages of the different subpopulations gated on B220+ B cells are in bold; absolute values in millions cells are given between parentheses. Asterisks indicate statistically significant differences with unpaired t-test between values observed in mutant mice and WT littermate controls: *p<0.05, **p<0.01, ***p<0.001. Values are means±SEM.

B1 cells (CD5low B220+)N=20/15/48.85±1.13% (1.33±0.13)9.81±1.41% (1.06±0.18)6.5±2.01 (0.96±0.25)

The spleen architecture was grossly normal in EXTL1 transgenics, with normal-size follicles including B lymphocytes and normal presence of plasma cells into the red pulp (Supporting Information Fig. 3). Cytometric evaluation of splenocytes also showed that the total amount of B cells among splenocytes was normal (Table 1). The follicular (CD23highCD21int) and marginal zone (CD23CD21high) B-cell compartments were of normal size (Fig. 5B). In the blood and peritoneum, all B-cell compartments appeared as normal (data not shown) and finally, serum Ig levels reflecting secretion by terminally differentiated B cells and in vivo Ig accumulation were also normal (Fig. 6).

Figure 6.

Ig production is unaltered in transgenic, as compared with, WT mice. Total endogenous Ig production was determined by ELISA in sera from 4-wk-old unimmunized WT and EXTL1 transgenic mice. The vertical axis is logarithmic; values are indicated as μg/mL. Data are presented as means±SEM of assays carried on ten mice for each experiment. Individual symbols represent individual mice. **p<0.01.

Proliferation of mature B cells upon stimulation and Ig secretion

We tested the proliferation of spleen CD43 B cells during 3-day stimulation assays and found that EXTL1 overexpressing B cells showed a decreased proliferation rate as compared with WT B cells, independent of the stimulation (LPS, anti-CD40 plus IL-4 or BCR cross-linking using an anti-μ antibody) (Fig. 7A). These anomalies were restricted to proliferation while apoptosis did not vary in these cultures (Fig. 7B). Ig secretion levels measured in vitro in such culture conditions (maintained daily at the same cell density) did not significantly differed between transgenic mice and WT controls (Supporting Information Fig. 4).

Figure 7.

Decreased B-cell proliferative responses in transgenic mice. (A) Proliferation of splenic CD43 B cells isolated from EXTL1 transgenic and WT mice at the indicated time points following in vitro stimulation with LPS (top), anti-CD40 plus IL-4 (middle) or anti-μ antibodies (bottom). Proliferation was estimated by the MTS test. Data are presented as means±SEM of assays carried out on cells isolated from at least seven mice (from one representative experiment out of three). *p<0.05, **p<0.01, comparing transgenic mice and WT littermate controls, unpaired t-test. (B) Splenocytes from transgenic and WT mice were stimulated in vitro for 3 days with LPS or anti-CD40 plus IL-4. The percentage of apoptotic CD19+ 7AAD gated cells was estimated on days 0 and 3 by cell cytometry staining for annexin V (BD Pharmingen). Data are presented as means±SEM of assays carried on cells from six mice (from one representative experiment out of three).


Cells of the B-cell lineage go through successive differentiation stages and distribute into compartments with varying responses to their microenvironment and to external stimuli. These responses are regulated along differentiation through programmed variations in the patterns of surface markers, transcription factors and global gene expression 28–31. By contrast, little is known about the potential variations of the ‘glycome’, except for a few cases well documented at the cell biology level, such as the enhanced expression of glycans binding the peanut agglutinin lectin in follicular B cells or such as the regulated expression of the CD22 Siglec molecule 32. Glycosylation is a major modifier of protein function especially in the context of surface receptors, but there is currently no mean to globally follow glycosylation patterns in primary lymphocytes at the biochemical level. To gain insights into a potential regulation of the glycome along B-cell development, we studied the variations of an array of 162 genes involved in glycosylation processes or coding for factors, which control cell fate and are modulated in their activities through glycosylation or interactions with glycans.

In a first step, we observed that this functionally restricted set of genes was sufficient to determine signatures typical of each stage of differentiation and/or activation of B lineage cells. Even an array limited to those 58 genes showing the highest variations in expression allowed a clear distinction between the various samples of primary cells, including BM pro-B and pre-B cells, splenic resting or LPS-activated B cells (with or without added cytokines), peritoneum B cells, lymph node B cells, Peyer patch B cells related or not to germinal centers (PNAhigh or PNAlow), by comparison with thymocytes and non-lymphoid stromal cells. Cell lines representative of a single stage of differentiation (pro-B, pre-B, mature B cells or plasma cells) showed patterns clearly related to primary cells of the same stage. Typical expression profiles were thus established for various B-cell activation or maturation stages, and for cells homing at various locations within mouse lymphoid tissues.

This analysis pinpointed two genes with major and inverse stage-specific variations. Both encoded enzymes synthesizing GAGs, but in the alternative pathways of either HS or CS synthesis. This observation suggests that the balance of HS/CS synthesis might be tightly regulated along B-cell differentiation, with mature resting B cells rather synthesizing HS with the contribution of EXTL1, but down-regulating the initiation of CS synthesis in the absence of CSGalNAcT-1 expression. By contrast, the latter CS-initiating enzyme was up-regulated both in B-cell progenitors and later on, in activated B cells. Although in vivo GAG biosynthesis is not fully understood even in cells producing large amounts of GAGs such as chondrocytes, CSGalNAcT-1 is demonstrated as a key enzyme whose expression level controls the amount of CS chains synthesized in a given cell 14. On a lower scale since GAG biosynthesis is minimal in lymphocytes, variations in the expression enzyme initiating CS synthesis are likely to produce similar effects in lymphocytes as in chondrocytes and to modulate the amount of total CS coupled to GAGs.

Such variations may simply be collateral to broad changes in the transcriptional program of B cells without any functional relevance, or may be part of the B-cell developmental program. To check for such a physiological role, we tried to deregulate the genetic CSGalNAcT-1/EXTL1 switch through a pan-B specific overexpression of EXTL1, and we looked for resulting developmental or phenotypic alterations of B cells.

Overexpression of EXTL1 hampered early B-cell maturation and featured a partial blockade at the pro-B to pre-B transition in the BM, suggesting that optimal differentiation of pre-B cells requires a balanced production of CS and HS. This blockade was likely related to the endogenous synthesis of GAGs associated with membrane PGs, since a similar blockade could not be obtained by checking in vitro the pro-B to pre-B transition in the presence of exogenous heparin. Interestingly, a similar partial blockade at the pro-B/pre-B transition has recently been reported in mice with a B-cell-specific inactivation of the GlcA/GlcNAc co-polymerase-1 (EXT1), another enzyme of HS synthesis 33.

Although this partial early defect did not show up in the normal-sized peripheral B-cell compartments, overexpression of the HS-elongating enzyme EXTL1 also resulted in a decreased proliferation of mature B cells upon LPS activation. Even in lymphoid cells known as low producers of GAGs, the HS/CS switch thus appears as an additional element modulating B-cell fate. CS chains associated with some surface receptors in progenitors and activated B cells might favor cellular interactions and/or the capture of cytokines. By contrast, the addition of HS chains, mostly a feature of resting B cells, might be a mean for limiting proliferative responses to mitogenic stimuli. The molecular dissection of how those GAGs can precisely modulate cell–cell contacts or control the activity of soluble mediators appears as a new and important question towards the understanding of B-cell physiology.

Finally, this study shows that a restricted cluster of glycosylation-related genes is sufficient for the identification of ‘glyco-patterns’ specific for various types of primary B cells representative of a given differentiation/activation stage or residing in a given tissue. B-cell differentiation thus appears to rely not only on the variation of transcription factors and membrane receptors but also on a regulated expression of glycoconjugates. Among these variations, the most dramatic involves two enzymes of GAG synthesis and suggests a developmental switch between two molecular pathways leading to the synthesis of either CS or HS. These two genes, CSGalNAcT-1 and EXTL1, were also regulated upon B-cell activation by mitogens.

This study resulted in the description of previously unsuspected variations along B-cell differentiation in the expression of two genes of GAG biosynthesis. These physiological variations proved important both for an optimal early B-cell differentiation and proliferative responses of mature cells. GAGs (and notably syndecan-1) have been shown in other tissues to be differentially branched either with HS chains, then promoting cell adhesion, or with CS chains then promoting cell migration 34–36. Part of the regulation of these properties may be at the level of expression of the core protein of GAGs, since transfected cells overexpressing syndecan-1 or syndecan-4 display increased intercellular adhesion due to the HS chains of syndecan 37. Our study now shows that in addition to a regulation in the expression of the core polypeptides, the function of GAGs may also be regulated in the lymphoid lineage through a control of the synthesis or elongation of HS versus CS chains decorating these core proteins. Altogether, these data point to GAGs are important modulators of lymphocyte functions.

Materials and methods

Cell collection and cell cytometry

Animal experimentation was performed in accordance with institutional guidelines. Single cell suspensions from spleen and BM from 4- to 8-wk-old mice were stained with the following monoclonal antibodies conjugated to either fluorescein isothiocyanate (FITC), phycoerythrin (PE), allophycocyanin, phycoerythrin-cyanin 7 (PECY7) or phycoerythrin-cyanin 5 (PC5): anti-CD5 (53-7.3), anti-CD19 (6D5), anti-CD23 (B3B4), anti-CD21/CD35 (7G6), anti-CD117/c-kit (2B8), anti-BP1/Ly51 (6C3), anti-CD43 (S7), anti-CD25 (PC61), anti-CD93 (AA4.1), anti-IgM (eB121-15F9), anti-IgD (11-26), anti-B220 (RA3-6B2) (all from BD Pharmingen, Southern Biotech or e-Bioscience) and FITC-conjugated anti-PNA (Sigma). Analyses were made on a Coulter FC500 apparatus and cell sorting was done on a FACSvantage apparatus (Becton Dickinson).

For RNA analysis, primary B cells isolated from normal C57BL/6 or 129SV mice were first magnetically enriched using either CD19+ selection or CD43+ selection (Miltenyi-Biotec) and secondly sorted by flow cytometry. Cell purity was always beyond 96%. The following markers were used for cell sorting:

  • (i)BM cells, Stromal WT: primary stromal cells, and stromal Gal1−/−: primary stromal cells from KO Gal1−/− kindly provided by Claudine Schiff (Marseille, France); proB: CD19+ IgM B220+ CD43high pro-B cells; preB: CD19+ IgM B220+ CD43low pre-B cells.
  • (ii)Splenocytes Spleen IgD+: IgD+ CD19+ resting B cells; spleen CD19+: CD19+ cells; T1: CD19+ IgMhigh IgDlow CD21 CD23low/– type 1 transitional B cells; T2: CD19+ IgMhigh IgDhigh CD21high CD23+ type 2 transitional B cells; FO: CD19+ IgMlow IgDhigh CD21int CD23+, follicular B cells; MZ: CD19+ IgMhigh IgDlow CD21high CD23low/– B cells; spleen LPS: CD43 cells stimulated 3 days with LPS (20 μg/mL, Sigma); spleen LPS+IL-4: CD43 cells stimulated 3 days with LPS (20 μg/mL)+IL-4 (1 ng/mL, PeproTech); spleen LPS+IFN-γ: CD43 cells stimulated 3 days with LPS (20 μg/mL)+IFN-γ (1 ng/mL, PeproTech); spleen LPS+TGF-β: CD43 cells stimulated 3 days with LPS (20 μg/mL)+TGF-β (1 ng/mL, R&D Systems).
  • (iii)Other lymphoid tissues Peritoneal cavity CD19+: CD19+ B cells; PP. CD19+: CD19+ Peyer patches cells; PP. B220+ PNAhigh: B220+ PNAhigh Peyer patches activated cells; PP. B220+ PNAlow: B220+ PNAlow Peyer patches non-activated cells; P. LN CD19+: CD19+ peripheral lymph node cells; M. lymph nodes CD19+: CD19+ mesenteric lymph nodes cells; Thymus: thymocytes.

Several murine cell lines were also used: MS5.1 and OP9 are stromal cell lines; BAF/3 is a pro-B cell line, 70Z/3 and 18.81 are pre-B cell lines; A20 and BCL1 are B lymphoma mature cell lines; NS1, S194, X63Ag8 and sp2/0 are myeloma cell lines.

RNA analysis

Total RNA was prepared by the Tripure technique (Roche), treated with DNase I (Invitrogen), cleaned up on RNeasy Mini Kit (Qiagen) and checked using an Agilent 2100 Bioanalyzer.

TaqMan® low-density arrays (Micro Fluidic Cards, Applied Biosystems) were used in a two-step RT-PCR process. First-strand cDNA was synthesized from total pre-treated RNA using the High-Capacity cDNA Archive Kit (Applied Biosystems). PCR reactions were carried out in Micro Fluidic Cards using an ABI PRISM7900HT apparatus. The 384 wells of each card were preloaded with 96×4 predesigned fluorogenic TaqMan probes and primers. The probes were labeled with the fluorescent reporter dye 6-carboxyfluorescein (FAM) on the 5′ end, and with non-fluorescent quencher on the 3′ end. One hundred nanograms of cDNA combined with 2× TaqMan Universal Master Mix (Applied Biosystems) were loaded into each well (200 ng for each sample). Data were analyzed using the threshold cycle (Ct) relative quantification method. Ct values ranged from 0 to 40 (the latter represents the default upper limit PCR cycle number defining failure to detect a signal). Genes with no or very low signal (Ct>35) in most samples were eliminated. The relative amount of each transcript was normalized to 18S transcript in the same cDNA to obtain the ΔCt. A ΔΔCt was then obtained to calculate the comparative gene expression between a calibrator sample (in this case, resting B cells sorted as IgD+ splenocytes) and the other samples. The formula 2–ΔΔCt was applied to calculate the relative expression of target genes.

Vector construction

The ‘5′-Eμ-pVH-EXTL1-LCR-3′’ transgene (Fig. 3A) included a murine pVH promoter as a 0.2-kb HindIII fragment, the Eμ enhancer as a 1-kb XbaI fragment 25 cloned upstream of pVH, and all four enhancers of the 3′ IgH regulatory region inserted downstream of the polyA site. The hs1,2 enhancer is a 0.6-kb StuI–EcoRV murine genomic fragment 38. It was flanked on both sides by hs3a and hs3b, as two 2.1-kb genomic EcoRI–HindIII fragments, with inverted orientations reproducing their endogenous arrangement 39, while hs4 was inserted farther downstream as a 1.38-kb genomic PstI–HindIII fragment 40. Two copies of the chicken β-globin HS4 insulator, as a 250-bp PCR fragment, were inserted on both sides of the expression cassette. Downstream pVH, the vector included a full-length EXTL1 cDNA (2048 pb) previously amplified from resting B cells by using a mix of conventional Taq and proof-reading polymerase (Takara, Otsu, Japan), with the following primers:





Generation of transgenic mice

The E14 embryonic stem (ES) cell line (129 mouse strain background) was co-transfected by electroporation with each linearized expression vector and with a CMV promoter-neomycin resistance gene cassette (in a 10:1 molar ratio). After G418 selection (400 μg/mL; Sigma) 41, recombinant clones were identified by Southern blot after an HindIII digestion (Fig. 3B), with the cDNA probe and confirmed by PCR using the following primers:



After microinjection in C57Bl/6 blastocysts of two independent recombinant ES clones for Extl1 construction, male chimeras were mated with C57Bl/6 females. Germline transmission of the mutation was assessed by coat color, and the presence of the transgene was checked by PCR with the above-cited primers on genomic tail DNA. All mice studied, either transgenic or non-transgenic littermates, were in the same mixed 129×C57Bl/6 genetic background.

Proliferation and activation assays

Spleen CD43 B cells were magnetically sorted (Miltenyi Biotech) from transgenic mice or littermate controls and cultured for 3 days at 5×105 cells/mL in 10% FCS RPMI medium supplemented with 20 μg/mL of LPS (Salmonella typhimurium, Sigma) or with 2.5 μg/mL of anti-CD40 (R&D Systems) and 50 ng/mL of murine IL-4 (PeproTech) or with 10 μg/mL of goat anti-mouse-μ chain (Southern Biotechnologies). Proliferation was measured by the 3-[4,5-dimethylthiazol-2-yl-5]-[3-carboxymethoxyphenyl]-2-[4-sulfophenyl]-2H (MTS) tetrazolium non-radioactive cell proliferation assay (Promega). Briefly, cells were plated in 96-well plates at a density of 50 000 cells/well in 100 μL medium and allowed to grow for 24, 48 or 72 h; 20 μL/well of MTS solution was then added and absorbance at 490 nm was measured after 4 h incubation in the dark at 37°C. Experiments were performed in triplicate and the results were expressed as the mean optical density (OD)±standard error of the mean (SEM).

Ig secretion was assayed by ELISA

Spleen CD43 B cells were cultured at 5×105 cells/mL in RPMI medium supplemented with 10% FCS with LPS or with anti-CD40+IL-4 for 5 days. On day 3, the cells were numbered and diluted to 5×105 cells/mL. On day 5, the supernatants were harvested and assayed by ELISA for Ig secretion.

On days 0 and 3, cells were analyzed by flow cytometry to evaluate B-cell apoptosis after staining with anti-CD19 antibody, annexin V and 7AAD (BD Pharmingen).

In vitro pre-B cell differentiation assays

BM B-cell progenitors were sorted as B220+ CD117+ Igκ cells; 2×104 sorted progenitors per well were seeded in 24-well plates in 1 mL of complete medium and cultured for 4 days with OP9 feeder stromal cells under high doses of IL-7 (10 ng/mL), with or without 5 μg/mL addition of heparin (Sigma). Further in vitro cultures for 2 days at a lowered IL-7 concentration (0.01 ng/mL) allowed to check differentiation into late pre-B cells, as evaluated by the loss of BP-1 expression and the gain of CD25 expression.

ELISA assays

Ig class and subclass determinations were done using polycarbonate plates (Maxisorb; Nunc) coated overnight at 4°C (100 μl/well) with 1 μg/mL of goat anti-mouse IgM, IgG, IgG1, IgG2a, IgG2b, IgG3 and IgA (Southern Biotechnologies) diluted in 0.05 M sodium bicarbonate buffer. After blocking with 3% BSA/PBS buffer and washing in 0.1% Tween-20/PBS buffer, 100 μL of sera from transgenic mice or normal littermates (first diluted to 1:200 in 1% BSA/PBS buffer) or serial dilutions of standard isotypes were incubated into wells for 2 h at 37°C. After washing, 100 μL/well of isotype-specific alkaline phosphatase conjugated goat anti-Ig (1 μg/mL) (Southern Biotechnologies), diluted in 0.1% Tween-20/PBS, was added for 2 h at 37°C. After washing, alkaline phosphatase activity was assayed on 1 mg/mL alkaline phosphatase substrate (Sigma) and blocked with the addition of 30 μL/well of 3 M NaOH. OD was measured at 405 nm.

EXTL1 expression in B cells was evaluated by ELISA using polycarbonate 96-multiwell plates coated with 1 μg/mL of anti-EXTL1 antibodies (353713, R&D Systems). Lysates of magnetically sorted CD43 splenocytes (Miltenyi Biotech) from three transgenic mice and littermate controls were prepared with RIPA lysis buffer (Santa Cruz Biotechnology) and assayed at 1:2 dilutions starting at 25 μg of cell lysate per wells. Extl1 expression was detected using biotin-conjugated anti-EXTL1 antibodies (1 μg/mL) and extravidin peroxidase (Sigma) After washing, peroxidase activity was assayed on o-phenylenediamine substrate and blocked with addition of 30 μL/well of 3 N HCl. OD was measured at 492 nm.

The results were expressed as mean±SEM, and overall differences between variables were evaluated by unpaired Student's t-test or Mann and Whitney test.


The authors thank Sylvie Desforges for help with animal care, Lionel Forestier for help with QPCR and Dr. Laurent Tesson for advice in cloning transgene insertion sites. This work was supported by Ligue Nationale Contre le Cancer, Institut National du Cancer and Conseil Régional du Limousin. This work was supported by grants from Ligue Nationale contre le Cancer, Institut National du Cancer and Conseil Régional du Limousin. S. D. was supported by a fellowship from Région Limousin and by Fondation pour la Recherche Médicale.

Conflict of interest: The authors declare no financial or commercial conflict of interest.