Glycoconjugates play various roles in biological processes. In particular, oligosaccharides on the surface of animal cells are involved in virus infection and cell-cell communication. Inhibitors of carbohydrate-protein interactions are potential antiviral drugs. Several anti-influenza drugs such as oseltamivir and zanamivir are derivatives of sialic acid, which inhibits neuraminidase. However, it is very difficult to prepare a diverse range of sugar derivatives by chemical synthesis or by the isolation of natural products. In addition, the pathogenic capsular polysaccharides of bacteria are carbohydrate antigens, for which a safe and efficacious method of vaccination is required. Phage-display technology has been improved to enable the identification of peptides that bind to carbohydrate-binding proteins, such as lectins and antibodies, from a large repertoire of peptide sequences. These peptides are known as “carbohydrate-mimetic peptides (CMPs)” because they mimic carbohydrate structures. Compared to carbohydrate derivatives, it is easy to prepare mono- and multivalent peptides and then to modify them to create various derivatives. Such mimetic peptides are available as peptide inhibitors of carbohydrate-protein interactions and peptide mimotopes that are conjugated with adjuvant for vaccination.

1. Introduction

A variety of glycoconjugate carbohydrate structures on the cell surface are important for biological events [1]. Carbohydrate structures on the cell surface change according to cell status, for example, during development, differentiation, and malignant alteration. Several glycoconjugates, including stage-specific embryonic antigen (SSEA)-3, SSEA-4, and tumor-rejection antigen (TRA)-1-60, are used as molecular makers of pluripotency to control the quality of induced pluripotent stem (iPS) cells [2]. Carbohydrate-protein interactions are the first cell surface events in cell-cell communication, following which processes such as infection and signal transduction occur. However, the reasons for the changes in carbohydrate structures on the cell surface are not clear. In addition, most receptors for glycoconjugates have not been identified. To investigate the biological roles of carbohydrates, sets of carbohydrates and their corresponding carbohydrate-binding proteins are required.

Carbohydrate-binding proteins such as plant lectins, bacterial toxins, and anticarbohydrate antibodies are available for studying carbohydrate-protein interactions [3, 4]. However, the repertoire of carbohydrate structures recognized by these proteins is limited and insufficient to cover the majority of structures. In addition, because carbohydrates are ubiquitous components of cell membranes and bio(macro)molecules, the immune response stimulated by glycoconjugates is negligible [5, 6], that is, high affinity carbohydrate-specific IgG-isotype antibodies are not easily obtained. Even if anticarbohydrate antibodies are generated, IgG comprises no more than 28% of the antibodies (74 IgGs in a total of 268 antibodies, with the remainder being IgMs) [7]. Therefore, while anticarbohydrate antibodies of the IgG isotype are preferred for carbohydrate research, IgM-antibodies with low affinity have been often used. Moreover, obtaining pure and homogeneous carbohydrates (or glycoconjugates) is very difficult. This is because regioselective protection of the hydroxy groups of the monosaccharide is required. Programmable one-pot oligosaccharide synthesis is widely performed using protected monosaccharides and/or oligosaccharides [810]. Enzyme-catalyzed oligosaccharide synthesis has been also developed [1012]. Several oligosaccharides such as KH-1 antigen (nonasaccharide of - ), globo-H hexasaccharide, and the core pentamannosides have been prepared by automated solid-phase oligosaccharide synthesis [8]. However, due to the complicated procedures of carbohydrate preparation, a general methodology for their chemical synthesis is not yet established.

To compensate for the lack of synthetic carbohydrates and to overcome their inherent weak immunogenicity, short peptides that bind to carbohydrate-binding proteins have been identified from phage-display libraries (Figure 1). These peptides mimic carbohydrate structures [13] and are called “carbohydrate-mimetic peptides (CMPs)” or “peptide mimotopes.” It is predicted that CMPs, as well as carbohydrates, are recognized by carbohydrate-binding proteins. Small molecules such as biotin and carbohydrate mimotope (Glycotope) mimicking peptides have been frequently identified, and a number of reviews focusing on different aspects of their properties and uses have been published [1416]. In this paper, recent studies on the selection and application of CMPs are surveyed and summarized according to the classification of target carbohydrate-binding proteins.

2. Peptide Selection from Phage Display Libraries

Phage display is an efficient selection (and screening) system for the identification of target-specific peptides and proteins from a large number of candidates [2022]. A filamentous virus (M13 and fd, etc.) that infects E. coli is frequently used in phage display technology. When DNA encoding foreign sequences is inserted into the coat protein (pIII or pVIII) region in the virus genome (M13 phage vector, etc.), the corresponding sequence is fused with the coat protein of the viral particle (Figure 2(a)) [20]. The foreign sequence is “displayed” on the viral particle and is able to interact with various types of target molecules.

In the case of peptide libraries, the length of the peptides is often 5–20 amino acids. There are two types of peptide library: linear peptide libraries and cyclic peptide libraries (Figure 2(b)). The randomized region of cyclic peptide libraries is surrounded by two cysteines (e.g., CX7C) to restrict the peptide conformation via disulfide bonds. The diversity of a peptide library is often 108-109, which is sufficient to cover a combination of hexapeptide libraries (X6; 206 = 6.4 × 107). Several kinds of peptide libraries (e.g., Ph.D. Phage Display Peptide Library Kits, New England Biolabs) and customizable phage vectors (Ph.D. Peptide Display Cloning System) are commercially available.

To isolate phage clones that have high affinity for a target molecule, a set of procedures called “affinity selection (biopanning)” is performed (Figure 2(c)). First, the target molecule is incubated with the phage library in order to bind to specific peptide sequences. After removal of excess phages by washing, the bound phages are eluted by incubation with a known ligand for the target or an acidic buffer. The phages are amplified by infection of hosts (E. coli), and the phage pool is subjected to another round of biopanning. By repeating these steps, target-binding phages are enriched, and, finally, phage clones are obtained. The peptides with high affinity for the target molecule are identified by DNA sequencing of individual phage clones. Huang and coworkers established a mimotope database MimoDB (http://immunet.cn/mimodb/) that contains the results of biopanning experiments including the phage libraries used and the peptide sequences identified [23, 24]. This database will help in the development of therapeutic molecules and the identification of superior peptide mimotopes for vaccination.

3. CMPs against Lectins

3.1. Monosaccharide-Mimetic Peptides

Most lectins recognize monosaccharides and disaccharides [4]. Concanavalin A (ConA) is a lectin from jack-bean (Canavalia ensiformis) that binds to α-mannose (α-Man) and α-glucose (α-Glc). ConA is a famous lectin that is commercially available for the biological investigation of glycoconjugates. The first CMPs were selected from a random peptide library against ConA simultaneously by Oldenburg et al. (octapeptide library) [25] and Scott et al. (hexapeptide library) [13] (Table 1). Peptides containing the consensus sequence, Tyr-Pro-Tyr (YPY), showed high affinity for ConA with a dissociation constant ( ) of 46 μM, and the for methyl α-Man was 89 μM. The peptides are considered to mimic the structure of carbohydrates because the ConA-peptide interaction was inhibited by α-Man.

To obtain Man/Glc-mimetic peptides, Yu et al. used three lectins, including ConA, Lens culinaris agglutinin (LCA) from lentil, and Pisum sativum agglutinin (PSA) from pea [31]. Two cyclic peptides, CNTPLTSRC and CSRILTAAC, were selected from a cyclic heptapeptide library, but these peptides did not contain the YPY motif. Docking simulation of the peptide-lectin interaction suggested that the cyclic peptides bound to an alternative binding site, not to the sugar-binding site that is recognized by YPY-containing peptides. In another screen using monosaccharide-binding lectins, Eggink and Hoober identified a GalNAc/Gal-mimetic dodecapeptide, VQATQSNQHTPR, that was selected against Helix pomatia (HPA) lectin [32]. A tetrameric dendrimer of the peptide, [(VQATQSNQHTPR)2 K]2 K, was synthesized chemically (Figure 3), which was shown to stimulate the secretion of interleukin (IL)-8 and IL-21 from human peripheral blood mononuclear cells (PBMCs).

3.2. Disaccharide-Mimetic Peptides

The Galα1-3Gal disaccharide is recognized by Griffonia simplicifolia I-B4 (GS-I-B4) and Bandeiraea simplicifolia isolectin B4 (BS-I-B4) (Figure 4). The Galα1-3Gal structure is a major carbohydrate antigen recognized by human anti-pig antibodies, and inhibitors of human natural antibodies may be useful in pig-to-human xenotransplantation. Kooyman et al. identified a peptide sequence, SSLRGF, that binds to GS-I-B4 lectin from a hexapeptide library [27]. Zhan et al. identified a peptide, NCVSPYWCEPLAPSARA, by selection with BS-I-B4 lectin [28]. These peptides, SSLRGF and NCVSPYWCEPLAPSARA, inhibited the agglutination of pig red blood cells (RBCs) by human serum. Two peptides, FHENWPS and FHEFWPT, that inhibit the agglutination of RBCs were identified by selection against anti-Gal antibody by Lang et al. [42]. However, the peptides identified from three selections contained no obvious consensus sequence.

Influenza virus hemagglutinin (HA) recognizes sialylgalactose structures (Neu5Ac-Gal) in glycoproteins and glycolipids on the cell surface in the initial stage of the infection process (Figure 4). Matsubara et al. identified CMPs from a pentadecapeptide library by selection with HAs of the H1 and H3 subtypes [17]. A HA-binding peptide, ARLSPTMVHPNGAQP, was identified from the first selection, and mutational sublibraries were prepared. A secondary selection was performed to improve the binding affinity for HAs, and the peptide was matured to peptide s2, ARLPRTMVHPKPAQP. The peptide was modified with a stearoyl group, and a molecular assembly of the alkylated peptides inhibited the infection of Madin-Darby canine kidney cells by influenza virus (Figure 3). Finally, a pentapeptide fragment from the N-terminal of s2, ARLPR [s2(1–5)], was found to show the highest inhibitory activity. A docking study of the interaction between the peptide s2(1–5) and HA suggested that the peptide is recognized by the Neu5Ac-Gal receptor-binding pocket (Figure 5(a)). The figure indicates that three side chains of H3HA (Ser 136, Asn137, and Glu190) have the potential to interact with the peptide instead of Neu5Ac, and hydrophobic residues (Leu194, Leu226, and Trp222) are close to the peptide (Figure 5(b)).

4. CMPs against Oligosaccharide-Binding Antibodies

4.1. Oligosaccharide-Mimetic Peptides for Inhibition

Glycoproteins and glycosphingolipids have unique oligosaccharide structures at their nonreducing termini [1]. Cell-cell communication is performed by oligosaccharides that are recognized by families of cell adhesion proteins such as selectins and sialic acid-binding immunoglobulin- (Ig-) like lectins (siglecs). Pathogenic viruses, toxins, and bacteria also recognize oligosaccharide structures [3]. Because an abundant variety of oligosaccharide structures relates to many carbohydrate-protein interactions, oligosaccharide-mimetic peptides mediate many kinds of inhibitory activities.

The sialyl- ( ) structure, Neu5Acα2-3Galβ1-4(Fucα1-3)GlcNAc, is recognized by E-selectin and is a famous carbohydrate antigen (Figure 4). -mimetic peptides were identified by selection against E-selectin [29, 30] and anti- antibody [36] (Tables 1 and 2). Martens et al. identified the HITWDQLWNVMN peptide and further optimized the sequence as DITWDQLWDLMK using a mutagenesis library [29]. The binding affinity of the synthetic peptide for E-selectin was improved 100-fold by this optimization (IC50 for binding to E-selectin; from 420 nM to 4 nM). The DITWDQLWDLMK peptide inhibited the adhesion of HL-60 cells and reduced neutrophil rolling on lipopolysaccharide- (LPS-) stimulated human umbilical vein endothelial cells. Qiu et al. designed WRY-containing peptides from the -mimetic peptide sequences, but these peptides cross-reacted with anti- antibody. Octameric multiple antigen peptides (MAPs) were conjugated with QS-21 adjuvant, which resulted in cytotoxic IgM and IgG antibodies (Figures 3 and 6). MAPs, in which peptides are attached to an octabranched amino acid backbone, are used to generate antibodies against a synthetic peptide, which is useful for the design of vaccines [94]. Katagihallimath et al. selected a cyclic CSRLNYLHC peptide against anti- antibody [37]. The trisaccharide structure is known as CD15 or SSEA-1, and this structure is expressed in the developing and adult murine central nervous system. The mimetic peptide inhibited CD24-induced neurite outgrowth.

Neutral glycosphingolipid Lc4Cer-mimetic peptides showed unique activity [46] (Table 3). Lc4Cer contains Galβ1-3GlcNAcβ1-3Galβ1-4Glc tetrasaccharide that is linked to ceramide (Figure 4), and Jack bean β-galactosidase digests a nonreducing terminus β-Gal to give Lc3Cer. The Lc4Cer-mimetic peptides inhibited digestion by β-galactosidase at a high concentration of enzyme, whereas the peptides enhanced the digestion of Lc4Cer at lower concentration of enzyme. This unique activity of the peptides was also shown in the digestion of nLc4Cer. This group also identified WHW-containing peptides such as WHWRHRIPLQLAAGR by selection with anti-GD1α antibody [47]. The ganglioside GD1α is cell adhesion molecule of murine metastatic large cell lymphoma (RAW117-H10 cells) that binds to endothelial cells. GD1α-mimetic peptides inhibited the adhesion between RAW117-H10 cells and hepatic sinusoidal endothelial (HSE) cells. Furthermore, the metastasis of RAW117-H10 cells to lung and spleen was completely inhibited by the intravenous injection of the peptide. Subsequently, WHW was found to be a minimal sequence that mimics the GD1α structure [48]. To modify the liposome surface with the WHW peptide, the WHW tripeptide was conjugated to alkyl groups such as palmitoyl or stearoyl groups (Figure 3). Coating of liposomes with peptides is often performed in drug delivery systems. The WHW-modified liposomes inhibited the adhesion between RAW117-H10 cells and HSE cells.

Tryptophan/tyrosine-containing tripeptides (YPY for ConA, WRY for , and WHW for GD1α) may comprise a key sequence that mimics oligosaccharide structure. Although Gb3 (Galα1-4Galβ1-4Glc trisaccharide) is dissimilar to the disaccharide (Galβ1-3GlcNAcβ) structure of Lc4 at the nonreducing terminus, Miura et al. identified a WHW-containing peptide (WHWTWLSEY) that mimics the Gb3 structure [60]. Gb3 is well known as a receptor for Shiga toxin (Stx). The Gb3-mimetic peptide showed neutralization activity against Stxs (Stx-1 and Stx-2) in a HeLa cell cytotoxicity assay. The binding affinity of the Gb3-mimetic peptide for Stx-1 was also investigated by surface plasmon resonance analysis ( = 1.4 nM).

4.2. Oligosaccharide-Mimetic Peptides for Vaccination

The immunogenicity of oligosaccharides is weak because oligosaccharides are ubiquitous components of cell membranes in tissues throughout the human body. When antioligosaccharide antibodies are generated, they attack these tissues and cause the risk of autoimmune disease. For example, lipopolysaccharides of Campylobacter jejuni isolated from GBS patients contain ganglioside-like epitopes such as GM1, GM1b, GD1a, and GalNAc-GDla, and these epitopes induce Guillain-Barre syndrome [95]. However, this low immunogenicity interferes with the preparation of antioligosaccharide antibodies that are useful for the investigation of glycoconjugate function.

To improve the binding affinity, specificity and cytotoxicity of antibodies, oligosaccharide-mimetic peptides are applied as peptide mimotopes of carbohydrate antigens for vaccination (Figure 6). Oligosaccharide-mimetic peptides were identified by selection against [34, 35, 37], [36, 50], GD2 [36, 5056], GD3 [36, 50, 59], lipooligosaccharide (LOS) [38, 39], β-1,2-oligomannoside [40], N-acetylglucosaminyl-β1,4-N-acetylmuramyl-alanyl-D-isoglutamine (GMDP) [43], and high-mannose oligosaccharide (Man9GlcNAc2 for HIV-1 gp120) [45]. The oligosaccharide-mimetic peptides were chemically synthesized and conjugated with adjuvant. To enhance the immunogenicity of the peptides, MAPs were prepared and resulted in dimeric, tetrameric, and octameric dendrimers (Figure 3). The peptide-adjuvant conjugates were vaccinated, with the adjuvants used being keyhole limpet hemocyanin (KLH) [39, 40, 53, 54], QS-21 [36, 50, 54], diphtheria toxoid (DT) [38], ovalbumin (OVA) [43], or very small size proteoliposomes (VSSP) [59] (Figure 6, Tables 2 and 3). In some cases, DNA vaccination was also performed [55, 56]. The CMP-induced antibodies are able to bind to peptide mimotopes and carbohydrate antigens.

5. CMPs against Polysaccharide-Binding Antibodies

Most polysaccharide-mimetic peptides to be applied for vaccination are identified as peptide mimotopes of carbohydrate antigens (Figure 6). Capsular polysaccharides of microorganisms are carbohydrate antigens, and it is known that these polysaccharides cause meningoencephalitis in immunocompromised patients, particularly those with AIDS (polysaccharide from Cryptococcus neoformans), pneumonia and bacteremia (Streptococcus pneumoniae), bacterial meningitis (Neisseria meningitidis), cholera (Vibrio cholerae), tuberculosis (Mycobacterium tuberculosis), and so forth (Table 4). These peptide mimotopes are potential antigens for safe vaccination and are expected to produce highly cytotoxic antibodies.

The typical methodology for vaccination uses a CMP-conjugated adjuvant. Valadon et al. identified CMPs that bind to anticryptococcal polysaccharide (glucuronoxylomannan, GXM) monoclonal antibody 2H1 [64]. The CMPs shared four motifs, for example, (E)TPXWM/LM/L and W/YXWM/LYE, and the dodecapeptide, GLQYTPSWMLVG (PA1) was found to bind 2H1 with a of 295 nM [64]. The three-dimensional structure of 2H1 has been solved in a complex with PA1 [65]. The peptide PA1 was improved by selection from a PA1-based sublibrary, which identified the peptide P206-1 (FGGETFTPDWMMEVAIDNE) [66]. The affinity of peptide 206-1 for 2H1 was 80-fold higher than that of PA1 ( of 3.7 nM). Immunization of mice with P206-1-tetanus toxoid (TT), but not PA1 or P601E (DGASYSWMYEA), induced an anti-GXM response [66, 67].

Although antibodies against the capsular polysaccharide of the same species (e.g., Neisseria meningitidis serogroup B) were used, the CMPs identified were different and shared no consensus motif [73, 7580] (Table 4). This may be due to the different antibodies used (HmenB3, 9-2-L3, 30H12, Seam3, or 13D9), different primary peptide libraries (CX6C, X9, CX9C, X12, or X15), or different selection conditions. Harris et al. also concluded that the CMPs identified by each antibody possessed distinct consensus motifs [72]. A variety of peptide-conjugating adjuvants such as KLH, TT, BSA, OVA, proteasome, and thyroglobulin have been used. In some cases, phage particles were directly used for vaccination [80, 86, 88], and a high level of the subtype in the response against CMPs was shown [80].

Theillet et al. clarified the structural mimicry of O-antigen oligosaccharide by CMPs [19]. Figure 5(c) shows a structural representation of the antibody-peptide complex in which the sugar chains were replaced by amino acids. Glc and GlcNAc were replaced by Tyr or Asp, and one or more hydrogen bonds are indicated. On the other hand, high-mannose oligosaccharide-mimic peptide (2G12-1 peptide) binds to a neighboring pocket of the oligosaccharide (Table 2) [45]. The binding site for the DVFYPYPYASGS peptide, which was selected against ConA, was different from the mannose/trimannose-binding site [26]. However, the peptide inhibits α-mannopyranoside binding to ConA [25], indicating that this peptide shows functional mimicry rather than structural mimicry.

6. Conclusion

Anticarbohydrate antibodies are necessary for clarifying the biological functions of carbohydrates, the detection of carbohydrates during etiological diagnosis, and therapy for carbohydrate-related diseases [7, 96]. Due to the difficulty in obtaining homogeneous glycoconjugates and carbohydrate-binding proteins, phage display libraries have been applied for the identification of peptide mimotopes. In this paper, the selection of CMPs was classified according to the types of target carbohydrates. The first selection was performed against lectins, and then the selections were performed against anticarbohydrate antibodies. To apply the peptide mimotopes for vaccination, this methodology is becoming more widespread.


The preparation of this paper was supported in part by the Keio Gijuku Fukuzawa Memorial Fund for the Advancement of Education and Research.