Mucosal surfaces are the largest and most important interface between the human body and its environment, comprising a total area of approximately 300 m² [1]. Interaction with the environment is of vital importance for the uptake of nutrients and oxygen, but the external interaction of mucosal surfaces also poses a threat the body. These threats include antigenic, mitogenic and toxic stimuli in food and in the air as well as pathogenic bacteria and viruses. To overcome these constant challenges, the immune system has developed extensive innate and adaptive responses.

The oral tissues, a part of the mucosal immune system, are constantly covered by saliva, which harbors a similar set of antimicrobial proteins as other mucosal fluids [2]. A major role of saliva is to maintain the natural oral microbial ecosystem. The function and secretion of saliva can be disturbed after radiation therapy for head and neck cancer, by auto-immune diseases affecting glandular tissues such as Sjögren’s syndrome [3,4] or as a side-effect of numerous drugs. A reduced saliva secretion leads to a significant increase in oral bacteria as well as to a shift in composition of the oral microflora. At the same time, the oral complications increase in number and severity [5]. Reduced saliva secretion may also affect general health. For instance, the majority of sedated patients in intensive-care units show a shift in the oral microflora from Gram-positive to Gram-negative species, which subsequently may spread into the respiratory tract causing pulmonary afflictions [6].

As a result of the continuous flow of saliva in our mouth, planktonic bacteria are swallowed, making microbial adherence one of the key selective criteria in the oral cavity [7,8]. Salivary proteins in solution may inhibit adherence by competition with bacterial binding sites on proteins coated at dental surfaces. Binding of salivary proteins to bacteria may result in bacterial agglutination which is, therefore, considered as a mechanism for clearance of bacteria and protection against dental caries [9–11]. One of the important bacteria agglutinating proteins is Deleted in Malignant Brain Tumors-1 protein (DMBT1) also known as Salivary Agglutinin (DMBT1^SAG) [12,13].

Streptococcus mutans is considered the major cause for dental caries. In search for S. mutans agglutinating substances, Ericson and Rundegren isolated a 300–400 kDa glycoprotein from parotid saliva [12] that showed calcium-dependent binding to S. mutans. This protein was named salivary agglutinin (DMBT1^SAG). It also binds a number of both Gram-positive and Gram-negative bacteria, as well as to host proteins including IgA, complement factor C1q, bovine and human lactoferrin, albumin and lysozyme [14–19]. Several studies have shown that the protein core of DMBT1^SAG is identical with lung glycoprotein 340 (DMBT1^GP340) and Deleted in Malignant Brain Tumors-1 (DMBT1) [20–22]. The amino acid sequences characterized so far are identical to the one deduced from the DMBT1 gene [23,24]. DMBT1^SAG and DMBT1^gp-340 cross react with monoclonal antibodies raised against the DMBT1s isolated from the different sources. Both proteins show calcium-dependent binding to S. mutans and SP-D [21]. DMBT1^GP340 was purified from bronchoalveolar lavage fluid of patients with alveolar proteinosis [20]. It binds to surfactant proteins A (SP-A) and D (SP-D) [25], members of the collectin family that play an important role in innate immunity by binding to specific carbohydrate structures on the surface of pathogenic micro-organisms [26]. DMBT1^SAG and DMBT1^GP340 thus represent DMBT1 isoforms that are encoded by the DMBT1 gene on chromosome 10q26.13. The DMBT1 gene shows frequent genomic rearrangements and/or loss of expression in two common types of malignant brain tumors, but also in diverse epithelial cancer types, including lung, gastric, esophageal, colon, breast, and skin cancer [27–36]. A search through the human genome revealed only one copy of the gene. Genetic polymorphism results in variants with different numbers of SRCR exons. In addition, different isoforms may arise through alternative splicing or differential post-translational modifications such as glycosylation.

The most typical feature of DMBT1 is that it consists of protein domains (Figure 1). DMBT1 is a member of the scavenger receptor cysteine-rich (SRCR) superfamily, which is characterized by the presence of one or more SRCR domains [23,24]. In between each of the multiple SRCR domains, all located to the N-terminal end of the DMBT1s, are short serine-threonine-rich amino acid motifs of 20–24 amino acids in length, called SRCR interspersed domains (SIDs). The largest naturally occurring allele known to date contains 13 of these SRCR-SID units, while the shortest known displays only eight. This genetic variability in normal individuals raises the question if size variations of DMBT1^GP340, DMBT1^SAG and variants from other tissues are due to alternative splicing and post-translational processing or are alternatively determined by genetic differences.

The stretch of tandem repeated SRCR domains with their interspersed SIDs are followed by two CUB domains separated by a 14th SRCR domain, and at the C-terminal end of the gene is a Zona Pellucida domain. Except for the SIDs, these domains are all known to be involved in ligand binding [37–39]. DMBT1 orthologs have been identified in various mammalian organisms such as mouse (dmbt1, CRP-ductin, vomeroglandin, muclin, apactin), rat (dmbt1, ebnerin, pancrin), rabbit (hensin), cow (bovine gall bladder mucin), pig (dmbt1) and rhesus monkey (H3) (Table 1). These proteins all have in common that they consist of SRCR, CUB and ZP domains although the number and order of these domains may be different (Figure 1).

The most prevalent domains in DMBT1 are the SRCR domains [38]. The SRCR superfamily consists of both membrane bound and secreted proteins which play a role in ligand binding [40]. SRCR proteins are present in multicellular animals along the entire animal kingdom, with their earliest appearance in sponges [38,41,42]. In addition, they also appear in protozoan parasites such as Cryptosporidium, Toxoplasma and Plasmodium and algae of the genus Chlamidomonas [43,44]. The organism with the highest number of SRCR domains is the purple sea urchin, Strongylocentrotus purpuratus, which contains 218 genes comprising altogether 1,095 SRCR domains [45]. For these SRCR proteins, a role as innate immune receptors was suggested since this organism also has a high number of innate immune receptors such as Toll-like receptors. Despite their presence in a large variety of animals [42], SRCR domains have retained a high degree of conservation. SRCR domains are ∼100–110 amino acids long and can be subdivided into two groups: members of group A contain SRCR domains with six cysteine residues, which are encoded by two exons; members of group B, to which DMBT1 belongs, usually contain SRCR domains with eight cysteine residues and are encoded by a single exon [38]. The position of the cysteines and their disulfide bond pattern are well conserved within each SRCR domain. The disulfide bond pattern of a group B SRCR domain is C1–C4, C2–C7, C3–C8 and C5–C6 [42].

The CUB domain is a 100–110 residue-spanning extracellular domain named after the three proteins, in which it was first recognized: Complement subcomponents (C1s/C1r), embryonic sea urchin protein (Uegf; sea urchin epidermal growth factor), and bone morphogenetic protein 1 (Bmp1). CUB domains are found in numerous proteins that are mainly involved in developmental processes [39]. Almost all CUB domains contain four conserved cysteines, which probably form two disulfide bridges (C1–C2, C3–C4). The structure of the CUB domain has been predicted to be a beta-barrel similar to that of immunoglobulins [46].

The zona pellucida (ZP) domain was first recognized in the sperm receptor proteins ZP2 and ZP3 [39,47]. These proteins are, along with ZP1, responsible for sperm adhesion to the zona pellucida, the glycoprotein membrane surrounding the plasma membrane of the oocyte. The ZP domain is a ∼260 amino acid module which contains eight conserved cysteines, forming four disulfide bridges. The disulfide bonding pattern suggests that the ZP domain consists of two subdomains. In addition to the conserved cysteines, a few aromatic or hydrophobic amino acids are absolutely invariant [48]. ZP domains are usually present in glycosylated proteins containing other domains and usually located at the C-terminal of the polypeptide, which is also the case for DMBT1s. The functions of proteins containing a ZP domain vary tremendously. They are found in glycoproteins involved in development, acoustic perception, immunity, and cancer [49]. ZP domains have been suggested to play a role in protein oligomerization [50]. This is consistent with the model that DMBT1^SAG forms aggregates [51,52].

The carbohydrate part of DMBT1^SAG accounts for 25–40% of the molecular weight [23,52]. DMBT1s contains up to 14 potential N-linked glycosylation sites [23], and numerous potential O-linked sites situated primarily in the SIDs. The SRCR domains, however, contain only few potential O-glycosylation sites (Ser and Thr residues). Presumably, the high density of glycans forces the SIDs in an extended conformation, as has been demonstrated for mucins [53], thereby creating a molecule with alternating stretched SIDs and globular SRCR domains.

The dominating O-linked structures in DMBT1^SAG are extended core 1 and core 2 oligosaccharides, either neutral or monosialylated (Table 2). These structures are extended by fucosylated oligo-N-acetyllactosamine units. Sialylation and fucosylation are found terminating the N-acetyllactosamine chains as Sialyl-Le^x or Le^b/Le^y, respectively. Sialylated structures tend to be shorter than fucosylated structures, suggesting that sialylation regulates the N-acetyllactosamine extension.

Glycosylation of proteins is controlled by genetically encoded glycosyltransferases [54]. The Se gene, which determines the presence of blood group antigens in secretory fluids, encodes a α1,2-fucosyltransferase. This enzyme couples a fucose to galactose (Fucα1-2Gal), which is the first step in the generation of blood group antigens. Consequently, non-secretors have a different carbohydrate composition compared to secretors. Non-secretors have Le^a and Le^x structures on DMBT^SAG; secretors also have Le^b and Le^y and ABH structures, in addition to Le^a and Le^x, on DMBT1^SAG [13,55]. DMBT1^SAG from secretors has a higher molecular mass than DMBT1^SAG from non-secretors [55]. Although evidence has been obtained that blood group antigens such as ABH and the Le^a antigens function as ligands for bacterial receptors, and thus might be involved in bacterial binding [56–58], only a few papers describe a correlation between blood group status and the susceptibility to caries [59].

Glycosylation of lung DMBT1^GP340 is different from that of salivary DMBT1^SAG. Lung DMBT1^GP340 completely lacks ABH and Le antigens [55] and its content of α2,3 linked sialic acid residues seems to be lower than that of salivary DMBT1^SAG [60]. It is not clear, however, to what extent this difference is related to individual variation in glycosylation, as the preparations were from different donors. Differences are also found for tear DMBT1 when compared to saliva DMBT1^SAG. In contrast to saliva, DMBT1 in tears had no sialyl-Le^x. [61].

Also, time-dependent variations in glycosylation patterns have been reported. The reactivity of DMBT1^SAG with antibody MECA-79, which recognizes the L-selectin ligand SO₃-6GlcNAc [62,63], fluctuates during the ovulation cycle. It is low shortly after menstruation and reaches a maximum a few days after ovulation. Also, during pregnancy and lactation, reactivity is high. These results suggest that sulfation of DMBT1^SAG is hormonally regulated [63].

Besides bacteria, DMBT1^SAG also binds viruses, including HIV-1 [92,103] and influenza A virus [60,104]. HIV-1 infects host cells by binding to the CD4 receptor through the viral envelope glycoprotein gp120, which leads to conformational changes in gp120 allowing high affinity interaction with chemokine receptors [105,106]. DMBT1^SAG shows a calcium-dependent interaction with the virus gp120. The DMBT1^SAG binding region on gp120 is localized to a linear, highly conserved sequence (CTRPNYNKRKR) near the stem of the V3 loop that is critical for chemokine receptor interaction during viral binding and infection [92]. Thus, DMBT1^SAG most probably blocks the access of gp120 to the chemokine receptor. The N-terminal sequence of DMBT1, including the first SRCR domain and one-half of the first SID, binds in a Ca²⁺-dependent manner to the same HIV-1 V3 sequences previously identified to interact with full-length SAG [107].

DMBT1^SAG and DMBT1^GP340 also inhibit the hemagglutination activity and infectivity of Influenza A virus (IAV) [104], which is responsible for outbreaks of influenza. IAV infects cells of the respiratory tract by binding to sialic acid residues present at the cell surface. Different influenza strains recognize specifically α(2,3)-linked or α(2,6)-linked sialic acid. DMBT1 has broad antiviral activity against human, equine and porcine IAV strains by sialic acid mediated binding to IAV, which results in virus neutralization [60]. A variant of DMBT1^SAG had significantly greater inhibitory activity against avian-like IAV strains which preferentially bound sialic acids in α(2,3) linkage, than DMBT1^SAG from another donor or three lung preparations of DMBT1^GP340 [60]. In line with this, a greater density of α(2,3)-linked sialic acids was present on the DMBT1^SAG from this donor as compared with the other samples. Hence, the specificity of sialic acid linkages on DMBT1 is an important determinant of anti-IAV activity. In addition to these direct virus-neutralizing properties, DMBT1^GP340 displays cooperative interactions with SP-D in viral neutralization and aggregation assays. On the other hand, in some cases DMBT1^SAG strongly antagonizes the antiviral activities of SP-D by binding to its carbohydrate recognition domain, which is involved in virus binding [60].

DMBT1 binds to a number of endogenous proteins such as secretory IgA, MUC5B, surfactant proteins A and D, complement factor C1q, lactoferrin and albumin [16–18,21,25,89,108,109]. DMBT1^SAG binds to secretory IgA in a calcium-dependent manner, resulting in a cooperative effect on bacterial aggregation [15,110]. Binding of DMBT1^SAG to IgA is mediated by the same 11-mer peptide that is responsible for the broad-spectrum bacteria binding of DMBT1^SAG [18]. Binding of a DMBT1^SAG-IgA complex to the Pac molecule of S. mutans was demonstrated [52]. After chemical reduction, this complex was dissociated into sIgA and DMBT1^SAG and concomitantly binding to Pac disappeared.

DMBT1^SAG also binds to C1q of the complement system, a system of plasma proteins that may react with one another to opsonize or permeabilize pathogens and induce a series of inflammatory responses [109]. By binding of SAG to C1q, the complement system is activated through the classical pathway [16]. Although DMBT1^SAG is secreted in saliva and C1q is a serum component, these two fluids may mix in the oral cavity under conditions of oral inflammation, e.g., periodontal disease, or mucosal damage, thus enabling a local complement activation.

DMBT1 also binds to bovine and human lactoferrin, an 80 kDa iron binding protein belonging to the transferrin family [17,19]. Lactoferrin exhibits various functions in the innate immune system. It sequesters iron from the local environment thus inhibiting microbial growth and it prevents the formation of Pseudomonas biofilms [111]. In addition, antimicrobial peptides are released upon proteolytic degradation of lactoferrin [112]. Bovine lactoferrin inhibits DMBT1^SAG binding to S. mutans protein antigen Pac, which belongs to the antigen I/II family [17]. The SRCRP2 peptide of DMBT1^SAG that is responsible for S. mutans binding also mediates binding of lactoferrin [17,89]. Bovine lactoferrin residues 480–492 (SCAFDEFFSQSCA) are important for DMBT1^SAG binding. Although the homologous sequence in human lactoferrin is slightly different (SCKFDEYFSQSCA), human lactoferrin also binds to DMBT1 [19].

DMBT1^GP340, which originally was isolated as a soluble receptor for SP-D, also binds SP-A [20,25]. SP-A and SP-D are collagen-containing (C-type) calcium-dependent lectins called collectins [113]. SP-D forms oligomers of 4–8 subunits. Each subunit is composed of three identical polypeptides of 43 kDa held together by disulfide bonds and non-covalent interactions at the N-terminal ends of the chains. Each polypeptide consists of a short N-terminal region, followed by a collagen-like sequence, a short α-helical sequence and the carbohydrate recognition domain. DMBT1^GP340 binds to the carbohydrate recognition domain of SP-D through a calcium-dependent protein-protein interaction [20]. SP-A and SP-D are involved in a range of immune functions including viral neutralization, aggregation and killing of bacteria and fungi, and clearance of apoptotic and necrotic cells.

Due to their broad pathogen binding spectrum, DMBT1 and its mouse homolog CRP-ductin are considered pattern recognition receptors (PRRs) [101], an ancient germline encoded defense system against pathogen invasion in both plants and animals [114,115]. PRRs recognize conserved structures of the microbial cell that are vital for survival, so–called pathogen-associated molecular patterns (PAMPs). PAMPs include mannans and zymosan of the yeast cell wall, various bacterial cell-wall components, such as lipoteichoic acid (LTA) of Gram-positive bacteria or lipopolysaccharide (LPS) of Gram-negative bacteria, and bacterial DNA [116,117]. In the human body, PRRs are found on host–pathogen interacting surfaces and expressed by epithelial cells, macrophages, granulocytes and dendritic cells or as secretory products present in mucosal fluids [115,118]. These characteristics are also applicable to DMBT1. PRRs form a diverse group of proteins including mannose-binding lectin, collectins, CD14, Toll-like receptors and Nucleotide-binding oligomerization domain (Nod) proteins [70,115,119,120]. The SRCR proteins SR-AI and SR-AII, MARCO and Spα, also belong to this group.

Conclusively, DMBT1 plays a role in innate immunity by binding to a large panel of exogenous and endogenous ligands. SRCRP2 on the SRCR domains and sialic acids on the glycan chains have been identified as ligand binding sites, but DMBT1 may harbor many other potential ligand binding sites. These many ligand binding sites and its structural heterogeneity—both in the different tissues and between different individuals—are only two of the many surprises hidden in this molecule. Future research should focus on the precise role of DMBT1 in the innate immune system and the physiological consequences of its heterogeneity.