Thesaurus

  • Antigen-presenting cells – cells process endocytosed foreign antigen into antigenic peptides for association with major histocompatibility complex (MHC) and presentation to T- lymphocytes

  • Classical (or conventional) dendritic cells (cDCs) – the DCs that are considered to be the most potent antigen-presenting cells, and therefore also referred to as “professional antigen-presenting cells”.

  • Plasmacytoid dendritic cells (pDC) – the DCs which resemble plasmacytoid cells, and are recognized as a major source of interferon-α/β following their induction.

  • Monocytes (Mo) – the cell population of myeloid origin characterized by phenotype as CD16neg wherein the CD14 expression can be described relatively as CD14high.

  • CD14low/-CD16+ cells – the blood mononuclear cells characterized by phenotype as CD14low/-CD16+

  • moDCs – the dendritic cells of monocyte origin, resulting from the differentiation of CD14+ monocytes stimulated with cytokines (GM-CSF and IL-4)

  • SSC – side scatter – parameter in flow cytometry analysis that characterizes inner complexity of the cells: shape of the nucleus, the amount and type of cytoplasm granules or the membrane roughness.

Introduction

Dendritic cells (DCs) play a key role in initiation of the innate and adaptive immune response [1,2,3]. They are the so-called “professional” antigen-presenting, due to their particular efficiency at endocytosing and processing antigen for presentation of the derived peptides in complex with MHC molecules, both to prime naive T-cells and re-activate memory T-cells (specifically responsive to the presented peptide) [1,2,3]. Human DCs can express CD1d, allowing them to activate NKT cells by presenting antigenic glycolipids [4]. In contrast to other antigen-presenting cells, subsets of DCs are highly potent at performing cross-presentation – presenting the antigenic peptides derived from processing endocytosed exogenous antigen in complex with MHC class I for the generation of cytotoxic lymphocytes. The latter in turn are capable of attacking cells of “altered self” – such as tumor cells and virus-infected cells – bearing the same antigenic peptide presented by the DCs subset [3].

DCs will present processed antigen derived from an infection, but also following vaccination. The efficacy of vaccine processed by DCs has been shown in cancer therapy [5], as well as viral and bacterial infections including hepatitis C, influenza and HIV [6,7,8,9,10,11]. In this context of vaccination, it is important for the DCs to recognize pathogen-associated molecular patterns (PAMPs) of viruses, bacteria or vaccines. This so-called recognition of “danger” activates different DCs subsets, ultimately promoting maturation of DCs presenting the antigen to T-lymphocytes. DCs possess so-called pattern recognition receptors (PRRs) for response to cognate PAMPs, including Toll-like receptors and certain leсtins [12, 13]. Consequently, choosing appropriate ligands, such as specific glycans could modify interactions with either PRRs or C-type lectin receptors (CLRs) to enhance DCs activation and/or facilitate endocytosis. Both processes would have consequential influences on processing of the antigen for presentation to T-lymphocytes. Indeed, anti-lectin immunoglobulins in combination with an immunogen can influence vaccine efficacy [14]. However, such application of anti-lectin antibodies is limited due to a risk of immune reaction against the targeting agent, which is often an immunoglobulin from a species foreign to the recipient; even allogeneic immunoglobulins could prove problematic due to interaction with Fc receptors on inflammatory cells, leading to induction of immunopathological phenomena. A promising elaboration of this approach would be delivery to DCs with specific glycans [15,16,17,18,19,20,21], primarily due to the lower immunogenicity of glycans compared with immunoglobulins, and absence of allogeneic problems. For instance, glycodendrimers containing 16–32 Leb copies, and liposomes loaded with Leb or Lex glycan residues were found to improve immunogen delivery to DCs, the report proposing an apparent involvement of binding with DC-SIGN [15,16,17,18]. Mannosylation of vaccine particles provided a similar effect, again with implications for the potential binding to DC-SIGN or mannose receptor [19, 20]. Modification of liposomes with 3′-sialyllactose (Neu5Acα2–3Galβ1–4Glcβ) has also facilitated uptake of the liposomes by dendritic cells, in this case with Siglec-mediated processes being proposed [21].

The above examples highlight a critically important property of the DCs family. These cells possess a particularly extensive repertoire of cell surface lectins as compared to other cells. Siglecs, selectins, proteins of CLR families and galectins have been identified [22, 23]. However, this identification of lectins has relied heavily on application of anti-lectin antibodies, or through assessment of their mRNA, the presence of their genes or by solid-phase assay. Such approaches are inadequate for defining cognate interactions of the receptors with glycans expressed on cell surfaces, and certainly mRNA profiles give no insight into activities at the cell surface. It is this knowledge with intact cells that is critical to understand the applicability of receptor-ligand interactions on DCs, particularly in the context of vector delivery. For the latter, the lectin must be assessed in terms of being functionally active, rather than the presence of mRNA – says nothing about the lectin itself – or the epitopes present – may be irrelevant to the functionally active sites on the receptor. Moreover, closely spaced endogenous glycans are capable of preventing lectin binding to “external” ligands, either completely or partially, like it was well documented for siglecs and galectins [24, 25]. Accordingly, the aim of the present research was to profile systematically the glycan binding ability of human DCs, with the ultimate aim of defining glycans for subsequent assessment in the construction of vectors for delivery to the DCs [26, 27]. For this purpose, we focused particularly on blood cells likely to represent the most potent antigen-presenting cells, namely CD14low/-CD16+CD83+.

Materials and methods

Materials

The following antibodies were used for flow cytometry: anti-human CD14-APC-Fluor780 (clone 61D3, eBiosciences, San-Diego, USA), CD83-RPE (clone HB15e Biolegend, San-Diego, USA), CD16-PerCP/Cy5.5 (clone 3G8, Biolegend, San-Diego, USA), CD45-PE-eFluor610 (clone HI30, eBiosciences, San-Diego, USA), CD11c-PE-Cy5 (clone 3.9, eBiosciences, San-Diego, USA), CD86 -PE-Cy7 (clone IT2.2, eBiosciences, San-Diego, USA). Lysis buffer OptiLyse C to remove red blood cells was purchased from Beckman-Coulter Immunotech (Marseille, France). Ficoll-Paque PLUS was from GE Healthcare (Uppsala, Sweden), anti-human CD14 - MicroBeads and MACS MS columns were purchased from Miltenyi Biotec GmbH (Bergisch Gladbach, Germany). RPMI-1640 medium and glutamine were from Life Technologies Co (Paisley, UK), fetal calf serum was from Sigma (St. Lois, USA). Carbohydrate-free bovine serum albumin (BSA) was from Serva (Heidelberg, Germany). Recombinant human cytokines rhGM-CSF (granulocyte macrophage colony stimulating factor) and rhIL-4 were from SciStoreLaboratory, LTD (Moscow, Russia). All the other reagents were from Reachem (Moscow, Russia).

Glycoprobes

Glycoprobes were synthesized as described in [28]. Starting polymer, 30 kDa poly(4-nitrophenylacrylate), is labeled with fluorescein (fluo) (1% mol.) with quantitative yield. Next, Glyc-sp-NH2 (20% mol.) is attached to the fluo-labeled polymer (also ~100% yield); only (Adi)3 ligand content was 7% mol. (Adi)3 cluster was synthesized by methodology described in [29] (see Suppl. Material, Fig. S1 ). Thus, all glycoprobes have identical mole content of fluo label as well as Glyc residues.

Glycoprofiling of blood cells

Peripheral blood was obtained from individual healthy donors at the Cytology Clinical Laboratory at the Russian Children’s Hospital of the Ministry of Health of the Russian Federation (Moscow); the age of donors ranged from 18 to 54 years. Blood collection was performed according to the Federal law “On donation of blood and its components” # 125FZ dated of 20 July 2012.

To 100 μl of blood sample were added 5 μl of Glyc-PAA-fluo (final concentration 100 μM) and 5 μl of fluorescent labeled antibodies against CD45, CD14, CD83 and CD16, samples were vortexed and incubated at 4 °C in the dark for 40 min. To remove erythrocytes, samples were incubated with lysis buffer OptiLyse C (Beckman-Coulter Immunotech, Marseille, France). The lysis was performed according to the recommendation of the manufacturer, as indicated at room temperature (18-25 °C). On analysis of the cells following lysis, the lysed cells were no longer detectable, but interaction of the remaining cells with probes was clear; for example, the 50% of population of cells CD14low/− CD16+CD83+ was positive for Adi (see Results). Briefly, 0.5 ml of lysis buffer was added to the sample, following by incubation for 20 min at room temperature, addition of PBS and incubation for 10 min at the same temperature. After that 1 ml of PBS containing 0.5% formaldehyde was added to the sample following by centrifugation at 350 g for 5 min. The supernatant was removed by aspiration. 2 ml of PBS containing 0.5% formaldehyde was added to the pellet of cells following by centrifugation at the same conditions. Flow cytometry was performed using a Beckman Coulter FC500 (Miami, USA) equipped with the software CPX and Kaluza 1.3. Probes staining <15% of cells were considered as negative. In the inhibitory setting as specificity control, blood samples were incubated with unlabelled Glyc-PAA (200 μM) for 30 min at 4 °C, followed by incubation with Glyc-PAA-fluo (50 μM) as described above.

Probing of CD14−CD16+ - cells and CD14low/-CD16+ cells

PBMCs were isolated over Ficoll-Paque PLUS gradient from whole citrated blood sample (50 ml) at room temperature by centrifugation at 800 g for 30 min. Fractions containing PBMCs was collected, diluted 1:2 with PBS-A and centrifuged at 200 g for 15 min to remove platelets. All the other steps were performed at 4 °C. Fraction containing monocytes and lymphocytes were washed twice in PBS-A at 100 g for 10 min. CD14+ and CD14− cells were sorted using anti-CD14 MicroBeads and a MACS-column. The CD14+and CD14− fractions (1 × 105/100 μl) were resuspended in TBACa (100 μl) to give 1 × 105 cells/100 μl); 5 μl glycoprobe solution and anti-human CD marker antibodies were added as described above. Samples were incubated for 40 min at 4 °C under gentle agitation on Rotomix (Elmi, Riga, Latvia) followed by washing in TBACa and TBSCa, then analyzed by flow cytometry as above.

Probing of CD11c+CD83+CD86+ cells

PBMCs isolated as described above were cultured on Petri dishes (Corning Inc., Oneonta, USA) in RPMI-1640 supplemented with 10% fetal calf serum and 2 mM glutamine at 37 °C in an atmosphere of 5% CO2 for 3 h. After removal of non-adherent cells the remaining adherent cells were cultured with RPMI-1640 containing rhGM-CSF (100 ng/ml) and rhIL-4 (20 ng/ml) for 6 days. On the 2nd and 4th days, 50% of medium were replaced with a fresh portion. The obtained moDCs were removed from the Petri dishes by scraping, and washed with TBACa. Cells (1 × 105 / 100 μl) were mixed in the same buffer with 5 μl glycoprobe solution and anti-human CD11c, CD86 and CD45 antibodies. Following incubation for 40 min at 4 °C under gentle agitation, the cells were washed (twice) in TBACa and TBSCa, and analyzed by flow cytometry as above.

Results

Primary selection of glycoprobes and setup for the glycoprofiling

The primary screening of human blood cells sought to detect the СD14−CD16+ population, which were also labeled for the CD83+ cells. This population contains DCs circulating in peripheral human blood [30,31,32,33,34,35]. The antigen CD16 is also expressed on CD14lowCD16+ cells [30, 31]. Discrimination of these two subpopulations directly in the blood is technically difficult, for which reason the analyses at this stage focused on the combination of the two subpopulations, namely CD14low/-CD16+ CD83+ cells. All cell populations – lymphocytes and monocytes, as well as granulocytes in whole blood – express the common isoform of CD45, which was therefore employed in the gating strategy (see Fig. 1).

Fig. 1
figure 1

Revealing of the СD14low/-CD16+CD83+ population. a Experimental flow-chart. b Phenotypic profiles obtained by flow cytometry. Monocytes/DCs were first gated using SSC vs. FL-anti-CD45, then the CD14low/-CD16+CD83+ subpopulation was identified by gating FL-anti-CD83 vs. FL-anti-CD14 and confirming with FL-anti-CD14 vs. FL-antiCD16. SSC, side scatter; FL, mean fluorescence intensity

Full size image

The monocyte (Mo) and DCs populations were identified by first gating with SSC versus CD45, and then discriminated in terms of CD14 and CD16 expressions. As explained in the Introduction, our main focus was on cells possessing CD14low/-CD16+CD83+ phenotype; the CD14low/-CD83+ cells were first localized, then confirmed that this subpopulation was also CD16+ positive (Fig. 1a and b). Following this gating, the binding of 229 glycoprobes – including ligands of known DCs leсtins (see Suppl. Material, Table S1 ) – were assessed for interaction with this CD14low/-CD16+CD83+ population. Only glycoprobes interacting with >15% of cells (data not shown) were selected for further work.

For discrimination of binding to CD14−CD16+ from binding to CD14lowCD16+, mononuclear cells were isolated from whole blood followed by separation on magnetic beads conjugated with anti-CD14 antibodies (Fig. 2). Monocyte-derived DCs (moDCs) were also generated by culturing isolated CD14+ monocytes with the cytokines rhIL-4 and rhGM-CSF (Fig. 2). Due to the reported expression of CD11c on moDCs, the phenotyping analyses also assessed marker [30,31,32]. It is also reported that during the maturation, DCs lose the macropinocytic capacity as well as increasing expression of co-stimulatory molecules including CD86 [32]. Accordingly, the CD11c+CD83+CD86+ subpopulation was assessed due to the likely presence of more mature moDCs (Fig. 2).

Fig. 2
figure 2

Experimental flow-chart for probing of CD14−CD16+CD83+, CD14lowCD16+, and CD11c+CD83+CD86+ by glycoconjugates. Blood mononuclear cells were obtained by centrifugation of blood from healthy donor over Ficoll-Paques PLUS gradients. The CD14− and CD14+ populations were isolated by positive and negative selection using MACS separation. The moDCs were obtained by culturing plastic-adherent CD14+ cells with cytokines as described in “Materials and Methods”. Subpopulations CD14−CD16+CD83+, CD14lowCD16+ and CD11c+CD83+ CD86+ were gated by staining with the corresponding antibodies and analyzing by flow cytometry

Full size image

Glycoprofiling of CD14low/-CD16+CD83+ cells

The glycoprofiling employed as a quantitative parameter the percentage of gated PBMC (see Fig. 2) binding a particular probe. In this context, 15–25% of the CD14low/-CD16+CD83+ cells were positive for binding of GlcNAcβ1–6(Galβ1–3)GalNAcα (core 2), Neu5Acα2-8Neu5Acα2-8Neu5Acα ((Neu5Acα)3), 4-O-Su-GalNAcβ1–4GlcNAcβ (4’-O-Su-LacdiNAc), and Glcα probes (Fig. 3a). About 30% of the CD14low/-CD16+CD83+ cells were positive for binding Man-containing glycans, namely the trisaccharide Manα1–3(Manα1–6)Manβ and the pentasaccharide Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ (Fig. 3a). This contrasted with only 18% of the cells binding a glycan lacking terminal mannose – (Galβ1–4GlcNAcβ1–2Manα)2–3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ (Fig. 3a). The highest binding value (57%) was observed with GalNAcα1–3Galβ (Adi). The binding is dose-dependent; the population of positive cells is increased from 3 to 50% when concentration of the probe ranged from 5 to 100 μM (see Suppl. Material, Fig. S2 ). In contrast, only 9% of the cells were positive for Galα1–3Galβ (Bdi) binding, which differs from Adi by the absence of the N-acetyl group. This compared with 45% and 35% of CD14+ monocytes binding Adi and Bdi, respectively (see Suppl. Material, Table S1 ). Specificity of the binding of Adi was also confirmed by an inhibition experiment. The population of of Adi-positive CD14low/-CD16+CD83+ cells was inhibited by Adi-PAA, whereas LN-PAA demonstrated absence of inhibition (see Suppl. Material, Fig. S3 ).

Fig. 3
figure 3

Comparison of glycan-binding profiles for a CD14low/-CD16+CD83+ blood cells and b CD14−CD16+ vs CD14lowCD16+ gated cells within isolated PBMC. Cells were probed directly in blood (a) or after magnetic separation (b) by Glyc-PAA-fluo as described in “Materials and Methods”. The binding was assessed by flow cytometry to generate the data for the number of positive cells (x-axis). The complete list of glycans and corresponding percentage bound values is given in Suppl. Material, Table S1 . Probes bind to <15% of cells are not included. The x-axis shows the percentage of gated cells binding the glycoprobe (y-axis). The results include data obtained with cells from four donors; error bars represent the standard deviation

Full size image

By using PBMC, we were then able to compare the CD14− cells (CD14−CD16+CD83+) with the CD14low cells (CD14lowCD16+CD83+). For this purpose, cells were probed with the glycans selected as described above during the screening of CD14low/-CD16+CD83+ cells. In general, the percentage of binding glycans was higher with the CD14−CD16+ cells (Fig. 3b). It was also particularly notable that 50% of CD14−CD16+CD83+ cells were positive for binding of (Galβ1–4GlcNAcβ1–2Manα)2–3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ (Fig. 3b), compared with only 18% of CD14low/-CD16+CD83+ subpopulation probed directly in blood (Fig. 3a). It is considered that this low binding with the cells in blood may reflect masking of DCs by blood IgG possessing biantennary N-chains. At the same time, an inner motif of (Galβ1–4GlcNAcβ1–2Manα)2–3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ, as the probe Manα1–3(Manα1–6)Manβ and Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ did not demonstrate this effect – lower binding in presence of blood glycoproteins. This could be explained as recognition of the (Galβ1–4GlcNAcβ1–2Manα)2–3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ probe by non-mannose-specific DCs lectin.

There were exceptions to the above observation of higher binding on CD14−CD16+ cells (Fig. 3b). The binding of 4’-O-Su-LacdiNAc was higher with CD14lowCD16+ cells than CD14−CD16+ cells, while binding of Glcα (the Adi cluster – see also below) was similar for the two subpopulations (Fig. 3b). Moreover, 23% of CD14lowCD16+ were positive for binding Bdi, whereas binding of this glycoprobe to CD14−CD16+ cells was particularly rare (Fig. 3b).

Considering that the CD14low/-CD16+CD83+ population within human blood showed the highest binding with the Adi probe, a new glycoprobe (Adi)3 was designed having three Adi residues located close to each other (Suppl. Material, Fig. S1 ). The percentage of the CD14lowCD16+ cells positive for binding this (Аdi)3 was higher than that obtained with Аdi (40% vs 26%), Fig. 3b. At the same time difference between binding to (Adi)3 and Adi was not observed in the case of CD14−CD16+CD83+ cells.

Glycoprofiling of moDC

This series of experiments employed the moDCs derived from adherent CD14+ monocytes (Fig. 2), which were gated for the CD11c+CD83+CD86+ subpopulation. The highest percentage of cells binding to glycoprobes was observed when the Man-containing pentasaccharide Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ and the trisaccharides core 2, (Neu5Acα)3 and (Adi)3 were employed (Fig. 4). The percentage of cells positive for binding the other Man-containing probes – namely Manα1–3(Manα1–6)Manβ and (Galβ1–4GlcNAcβ1–2Manα)2–3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ – was only slightly lower than that obtained with the Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAc (42% vs. 34%).

Fig. 4
figure 4

Glycan-binding profile of CD11c+CD83+CD86+, flow cytometry data. The results include data from two donors; error bars represent the standard deviation

Full size image

Comparative glycan binding profiles for CD14−CD16+, CD14lowCD16+ and CD11c+CD83+CD86+ subpopulations

Comparison of the glycan-binding profile of the CD14−CD16+CD83+ subpopulation – which would certainly contain circulating DCs – with the CD14low16+CD83+ subpopulation and CD11c+CD83+CD86+ subset of moDCs highlights the following features. Firstly, CD14lowCD16+ cells generally bound glycoprobes weaker than CD14−CD16+ cells (Fig. 5). The only exception was 4’-O-Su-LacdiNAc capable of interacting with CD14lowCD16+ cells better than the CD14−CD16+ subpopulation or moDCs. Secondly, CD11c+CD83+CD86+ subset bound to Adi, Manα1–3(Manα1–6)Manβ and (Galβ1–4GlcNAcβ1–2Manα)2-3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ more weakly compared with the CD14−CD16+ population. Thirdly, the percentage of cells positive for binding (Neu5Acα)3, trisaccharide core 2, Glcα and Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ was similar for CD11c+CD83+CD86+ and CD14−CD16+CD83+ cells. Finally, the cluster of three Adi residues bound to CD14lowCD16+CD83+ and CD11c+CD83+CD86+ cells better than the single disaccharide.

Fig. 5
figure 5

Comparison of glycan-binding profiles of CD14−CD16+CD83+, CD14lowCD16+CD83+ and CD11c+CD83+CD86+ subpopulations, flow cytometry data

Full size image

Discussion

DCs are prominent among mammalian cells with respect to the variety of lectins synthesized and exposed on their plasma membranes. Yet, the choice of glycan-vector for vaccine particles when employing anti-lectin antibody binding seems to be unproductive. We therefore sought to be more informative by applying an approach based on the lectin function, namely the ability of cells to bind carbohydrate ligands. We anticipated two advantages for this strategy. Firstly, it will allow exclusion of lectins with masked (by endogenous cell glycans) binding sites. Secondly, it facilitated identification of otherwise unexpected glycan-binding activity. Indeed, a broad pre-screening with almost three hundred glycoprobes showed that many of the glycans expected to be reactive from the published literature on anti-lectin antibodies and solid-phase assays did not display efficient binding. At the same time, some glycans with little information available from the literature on likely ligand-binding activity, were found to interact with human blood cell subpopulations containing DCs and/or monocytes.

A major objective of our study was to assess the CD14low/-CD16+ cells of human blood known to contain circulating DCs [35]. In comparison, we also probed moDCs derived from CD14+ monocytes, because moDCs are often used in studies of DCs in vitro. One reason for this is that unlike circulating DCs, moDCs can be generated on a larger scale. Moreover, most of the prior papers [15,16,17,18, 21] about glycans-vectors, with which we need to be comparative, employed moDCs. However, the differentiation of monocytes into moDCs after incubation with cytokines has been shown primarily in vitro. Evidence for an equivalent population in vivo has been difficult to assess, but it is considered that moDCs may reflect more the DCs arising from Mo during the extravasation events occurring with inflammatory processes. This would relate to moDCs showing a closer phenotype to macrophages than to blood DCs. Nonetheless, no data exists on comparison of glycan-binding profiles for circulating blood DCs vs. moDCs. Therefore, we also investigated the glycan-binding profile of the CD11c+CD83+CD86+ cells obtained upon differentiation of CD14+ Mo into moDCs.

The presented results enabled to draw the following conclusions. (1) The glycan-binding profiles of the CD14−CD16+ and CD11c+CD83+CD86+ cells were similar, but not identical. Clearly, the CD11c+CD83+CD86+ moDCs are related to blood DCs, although there are certainly distinct differences – in particular the binding to Adi and Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ. (2) Of the eight glycans selected by the pre-screening (CD14low/-CD16+CD83+ subpopulation in blood, only three bound clearly on CD14−CD16+ and CD14lowCD16+ subpopulations from sorted PBMCs. These were Manα1–3(Manα1–6)Manβ, Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ, and (Adi)3. From these, the latter one and Man-pentasaccharide bound on moDCs. Although mannose-rich glycans are common structures on human cells, this result is surprising. Lectins, in particular the CLRs, are known for their characteristics as PRRs and/or endocytic receptors. As such, one would expect recognition of “alien”, not “self”. Yet, numerous alien ligands typical of bacterial polysaccharides (Rha, Ara, melibiose, chitobiose, etc.), displayed little or no binding (Suppl. Material, Table S1 ).

Lectins potentially binding the identified glycans

CLRs, siglecs and galectins are exposed on DCs [12, 13, 15, 22, 23]. Due to the results obtained above, we sought to determine which lectins would form the more probable partners for the glycans showing clearest interaction with the cell subpopulations under investigation. For this, we compared our binding data with information in the literature gained from gene sequencing, mRNA studies, anti-lectin antibody characterizations and/or solid phase assay results.

Galectins

Although genes encoding galectins −1, −2, −3, −4, −8 and −9 have been identified in blood DCs and CD14lowCD16+ cells [23], we did not observe binding of ligands for galectins identified in the literature using PGA and other methods (Table 1) [25, 36]. This can be explained by the galectin localization to the cytoplasm. Moreover, galectins can be masked by cis-glycans, which would hinder binding of a glycoprobes, as observed with other cells [25]. This has implications for DCs, with which a high density of 3′-sialylated oligolactosamines – potent ligands for tandem- and chimeric galectins – is found in the composition of DCs surface glycoconjugates [37].

Table 1 Suggested attribution to particular lectins on DCs* of glycans identified as positive for binding to the blood cell subpopulations assessed
Full size table

Selectins

Genes encoding E- and L-selectins have also been identified in blood DCs [23], which should contain the CD14lowCD16+ subpopulation. Yet, the reported SiaLea, SiaLex, and 6-O-Su-SiaLex ligands for E- and L-selectins [38] did not bind to the CD14low/-CD16+CD83+ subpopulation. This may relate to the low level of selectin expression on the surface of blood cells from healthy donors (in the absence of inflammation).

Siglecs

Siglecs −3, −5, −7 and −9 were identified on the surface of blood DCs and CD14lowCD16+ cells by corresponding antibodies [39]. The CD14lowCD16+ sorted cells did not bind to the glycans known as ligands of siglecs – 3’SiaLac/3’SiaLN and 6’SiaLac/6’SiaLN (Table 1). The only sialoglycan with which the CD14low/-CD16+CD83+ subpopulation in the blood interacted is the trisaccharide (Neu5Acα)3 (Table 1), known as the best ligand of the siglec-7 [36]. This would suggest an apparent interaction with siglec-7. Siglecs of DCs could be masked by sialoglycans of glycoproteins, the density of which is high [39]. Only siglec-7 is not (or weakly) masked, because its ligand (Neu5Acα)3 is present only in gangliosides incapable of masking due to their membrane proximity.

C-type lectin-like receptors

MMR, DC-SIGN, L-SIGN

Genes encoding MMR (macrophage mannose receptor) and DC-SIGN were apparently absent from circulating blood DCs, being identified only in moDCs [22, 23]. Yet, MMR and DC-SIGN are important receptors in the context of antigen targeting to DCs [16, 17, 40, 41]. Indeed, MMR and DC-SIGN exhibit affinity for Man-containing glycans. Moreover, the cysteine-rich domain of MMR binds to 3’-O-sulfated glycans LN, Lec, Lex, and Lea [42], and to 4’-O-sulfated LacdiNAc [43]. Solid phase assay data has also demonstrated DC-SIGN binding to Leb, Lex, Ley [36, 44, 45]. Nonetheless, 4’-O-Su-LacdiNAc was the only sulfated glycan interacting with CD14low/-CD16+CD83+ cells from blood, in particular the CD14lowCD16+ cells from sorted PBMCs (Table 1). This clearly demonstrates a deficiency in the genetic studies and solid phase assay data that do not give a clear picture of lectin expression on DCs. The data also raises the question of using ligands for MMR and DC-SIGN to target vectored vaccines, although it should be stressed that the reactive 4’-O-SuLacdiNAc may interact with another receptor.

In contrast to the MMR and DC-SIGN gene, the gene encoding L-SIGN was identified in blood DCs [23]. The data on the carbohydrate specificity of L-SIGN are controversial. Some sources report recognition of only Man-containing glycans, while others report binding of Leb and Ley, but not Lex and Lea (Table 1) [36, 44, 45]. Although L-SIGN was considered not to be expressed on the surface of peripheral blood cells, recent data has countered this opinion [46]. However, expression does relate to inflammatory processes, levels being very low with cells from healthy donors. With our study employing blood from healthy donors, it is unlikely that L-SIGN was binding Man-containing probes.

DEC-205

Antibody binding and mRNA analysis has identified DEC-205 (CD205) on the surface of blood DCs [22, 23, 47]. Although it should interact with glycans containing Glcβ/GlcNAcβ (Table 1, [36]), we did not observe any such binding to CD14low/-CD16+CD83+ cells. As with L-SIGN, this lectin is not involved in the observed probe interactions.

Decin-1, dectin-2

Gene or mRNA analysis and the use of corresponding specific antibody binding assay has identified dectin-1 [23, 48] and dectin-2 [23, 49] associated with blood DCs. It should be noted that although dectin-1 has been included in the CLRs family, it is not a classical CLRs due to the fact that its binding does not require the presence of Ca2+. The genes encoding the dectin-2 family members [50] DCIR (dendritic cell immunoreceptor), BDCA-2 (blood dendritic cell antigen-2) and mincle (macrophage inducible C-type lectin) have also been identified in blood DCs [23]. Dectin-1 is reported to recognize 3-O-Su-GalNAcβ [36], and DCIR displays an affinity for Glcα, GlcNAcβ1–6GalNAcα, LN-6Tn, and B (type 1) [36]. However, the CD14low/-CD16+CD83+ cells generally did not interact with these reported ligands, only 20% of the subpopulation bound Glcα. Dectin-2, BDCA-2 and mincle apparently bind Man-containing glycans [36]. Since BDCA-2 and mincle were only revealed on pDC (which are not considered to be classical antigen-presenting cells) [23] we assume that dectin-2 could bind the assessed glycans.

DCAL-1, DCAL-2

Specific antibody binding and mRNA assays have also identified an association of the DCAL family (dendritic cell associated lectins) – DCAL-1 and DCAL-2 – with blood DCs [23, 51, 52]. While the carbohydrate specificity of DCAL-2 is unknown, ManP, LNT, and LNnT are reported ligands of DCAL-1 [36]. Nonetheless, these ligands did not bind with the CD14low/-CD16+CD83+ subpopulation, indicating an absence of at least DCAL-1 involvement in the observed interactions with the assessed glycans.

MGL

MGL (macrophage galactose specific lectin) has also been associated with blood DCs and immature moDC, as shown by mRNA analysis and corresponding antibodies [23, 53]. Information on its specificity being controversial. Some sources suggest a high affinity to Аdi and LacdiNAc [54], whereas others report binding of sulfated derivatives of LacdiNAc and LacNAc equally to Аdi (Table 1) [36]. Our own study may explain this discrepancy. Аdi was the most potent probe interacting with CD14−CD16+CD83+ cells from blood, while 4’-O-Su-LacdiNAc was the most potent probe interacting with CD14lowCD16+ cells from sorted PBMCs (the other reported ligands for MGL did not interact with these cells). Certainly, it would appear that MGL is the most likely candidate for this observed binding Аdi and 4’-O-Su-LacdiNAc.

TLRs

Besides lectins, genes encoding TLRs have been revealed in blood DCs by mRNA-analysis [23], and their expression as either cell-surface TLRs (such as TLR1, 2, 4, 5, 6) or internal TLRs (TLR 3, 7, 8, 9) is well established [55, 56]. TLRs are known to bind a wide range of ligands including polysaccharides, lipopeptides, lipopolysaccharides and nucleic acids [57,58,59,60]. For example, TLR-4 binds mannan [58] and anionic glycans, e.g. heparan sulfate [59] and fragments of hyaluronic acid [60]. Nonetheless, there was no observed binding of hyaluronic acid although we cannot exclude the involvement of other negatively charged glycans, such as 4’-O-Su-LacdiNAc or (Neu5Acα)3.

Conclusions

In this study, we sought to define glycans for their binding efficiency with blood cell subpopulations likely to contain DCs, with the aim of identifying potential vectors to improve vaccine delivery to DCs. In this context, the highest percentage (in the range of 30–50% positive blood-derived CD14low/− CD16+CD83+ cells, which should contain the circulating DCs) was observed for the following glycans: Manα1–3(Manα1–6)Manβ, (Galβ1–4GlcNAcβ1–2Manα)2-3,6-Manβ1–4GlcNAcβ1–4GlcNAcβ, Manα1–3(Manα1–6)Manβ1–4GlcNAcβ1–4GlcNAcβ), (Neu5Acα)3 and Adi. This would suggest an involvement of dectin-2 (binds mannose ligands), siglec-7 (binds trisialoside) and MGL (binds Adi). Indeed, all of these lectins are present on the DC surface, an important consideration when considering the particularly efficient capacity of DCs for endocytosis and internalized cargo processing.

Yet, a clear explanation for some of the observations is still lacking. For example, the higher potency of (Adi)3 comparing to single Adi (see Fig. 5) cannot be explained by two- or trivalent interaction of the former with three-subunit MGL –the distance between the disaccharide residues in the cluster is insufficient. The observation is suggestive of the disaccharide residues close to each other forming a conformationally stable spatial epitope, which would interact with MGL better than single GalNAcα1–3Gal. It cannot be excluded that another lectin interacts with the cluster. This suggestion is consistent with the difference in interaction between mono- and trimeric Adi not being observed for all the cells. Another example still requiring explanation is the binding of CD14low/-CD16+CD83+ and CD11c+CD83+CD86+ cells with the trisaccharide GlcNAcβ1–6(Galβ1–3)GalNAcα. PGA analysis [36] does not allow with certainty definition of the particular lectin with which GlcNAcβ1–6(Galβ1–3)GalNAcα would bind in flow cytometry experiments. At this stage, we can only hypothesize participation of an undefined lectin or another carbohydrate-binding protein, such as a TLR.

Overall, we can say that the three Man-containing glycans mentioned above and the trisialoside (Neu5Acα)3 are typical fragments of glycans on human cells. They are found both in the glycocalyx and amongst circulating blood proteins. Accordingly, endogenous ligands could compete with vaccine vector moieties for binding to DCs. In contrast, Adi is not present among human glycoconjugates, which would make it a more attractive choice as a vaccine vector. On the other hand, MGL expressed on macrophages could bind Adi, which might prove problematic considering that the number of monocytes is higher than the DCs in circulating blood. However, we do not think that this an unfavorable factor, because quiescent macrophages may also function as antigen-presenting cells, or assist the DCs in their function [61]. Therefore, Adi in a vaccine composition would be expected to interact with lectins on DCs and other antigen-presenting cells. In turn this which would prove beneficial for activating both naive and memory T-lymphocytes in lymph nodes (DCs) and resident peripheral memory T-lymphocytes (DCs and other antigen-presenting cells). Moreover, the efficacy of delivery is expected to be increased if the vectored vaccine were equipped with two or more ligands directed at different DCs lectins. In this context, we would propose that the trisialoside (Neu5Acα)3 is best paired with Adi, because it has no endogenous glycoconjugate-competitors on cell surface. We are now focusing our work in this direction to see how the above observations and hypotheses can be best applied in the field of vectored vaccine delivery.