J. J. Batenburg, Department of Biochemistry and Cell Biology, Faculty of Veterinary Medicine, Utrecht University, PO Box 80176, 3508 TD Utrecht, the Netherlands. Fax: + 31 30 2535492, Tel.: + 31 30 2535381, E-mail: firstname.lastname@example.org
Collectins are a family of collagenous calcium-dependent defense lectins in animals. Their polypeptide chains consist of four regions: a cysteine-rich N-terminal domain, a collagen-like region, an α-helical coiled-coil neck domain and a C-terminal lectin or carbohydrate-recognition domain. These polypeptide chains form trimers that may assemble into larger oligomers. The best studied family members are the mannan-binding lectin, which is secreted into the blood by the liver, and the surfactant proteins A and D, which are secreted into the pulmonary alveolar and airway lining fluid. The collectins represent an important group of pattern recognition molecules, which bind to oligosaccharide structures and/or lipid moities on the surface of microorganisms. They bind preferentially to monosaccharide units of the mannose type, which present two vicinal hydroxyl groups in an equatorial position. High-affinity interactions between collectins and microorganisms depend, on the one hand, on the high density of the carbohydrate ligands on the microbial surface, and on the other, on the degree of oligomerization of the collectin. Apart from binding to microorganisms, the collectins can interact with receptors on host cells. Binding of collectins to microorganisms may facilitate microbial clearance through aggregation, complement activation, opsonization and activation of phagocytosis, and inhibition of microbial growth. In addition, the collectins can modulate inflammatory and allergic responses, affect apoptotic cell clearance and modulate the adaptive immune system.
Collectins belong to the super family of mammalian C-type lectins, and are believed to be involved in innate defense systems. The following eight collectins have been identified so far: mannan-binding lectin (MBL), surfactant protein A (SP-A), surfactant protein D (SP-D), collectin liver 1 (CL-L1), collectin placenta 1 (CL-P1), conglutinin, collectin of 43 kDa (CL-43) and collectin of 46 kDa (CL-46). As part of the innate immune system, collectins have a key role in the first line of defense against invading microorganisms, as demonstrated by elegant experiments with genetically manipulated mice made deficient in MBL, SP-A or SP-D, which show increased susceptibility to bacterial and viral infections. Apart from CL-L1 and CL-P1, which are found in the cytosol and cell membrane, respectively, all collectins are soluble and secreted proteins. An important property of the collectins is their capability to recognize pathogen-associated molecular patterns on foreign organisms, which involves distinguishing between self and nonself carbohydrate structures. This review gives an overview of what is currently known about the functions of the collectins in host defense, the sites of their production, and their structure and function. Emphasis will be on the molecular basis of their recognition of carbohydrate structures.
Sites of collectin production
MBL is secreted into the bloodstream, and is mainly produced by the liver [1–3]. In rodents [4,5], rabbits [6,7], and rhesus monkeys  two forms of MBL have been found (MBL-A and MBL-C), whereas in humans and chimpanzees only one form was shown to be present . Although the liver is the main site of MBL-A and MBL-C production in mice, mRNA expression has been detected in various tissues. However, substantial expression of MBL-A and MBL-C was only demonstrated in the kidney and small intestine, respectively, where expression could also be demonstrated at the protein level using immunohistochemistry [9,10]. The presence of substantial amounts of protein in the small intestine suggests that MBL acts as a humoral immune factor in the intestine, similar to secretory IgA.
The lung collectins SP-A and SP-D were first shown to be present in the alveolar space of the lung, and it has long been established that alveolar type II cells [11–13] and nonciliated bronchial epithelial cells (Clara cells) [13,14] are the major sites of synthesis. Although the major site of SP-A and SP-D synthesis is the lung, both lung collectins have been detected in extrapulmonary tissues as well. Using RT-PCR, low amounts of SP-A mRNA have been shown to be present in a number of murine tissues, whereas on the protein level there were only indications for the presence of SP-A in the murine uterus . In addition to its presence in the murine uterus, low levels of SP-A have also been detected in the porcine eustachian tube . Whereas extrapulmonary SP-A expression seems to be limited to a few organs, SP-D has been detected in many nonpulmonary tissues, on the mRNA as well as protein level, and tissue distribution was found to depend on the animal species studied [15–17].
Using Northern blot analysis, high levels of CL-L1 mRNA were found in the liver and a weaker signal was demonstrated in the placenta. RT-PCR revealed the presence of low copy numbers of CL-L1 mRNA in most tissues except for skeletal tissue. Although most collectins are secreted, CL-L1 was only detected in the cytosol of hepatocytes, suggesting that this protein might react with intracellular ligands . CL-P1 was detected in vascular endothelial cells, while CL-P1 mRNA could be demonstrated in many tissues. This is the only collectin identified so far that is membrane bound, and contains an intracellular domain .
The serum collectins conglutinin, CL-46 and CL-43 have so far only be detected in bovidae, where the liver is their main site of production . The reason for the presence of this wide array of serum collectins in bovidae is unknown but might be related to the fact that these animals live in symbiosis with an enormous amount of microbes in their rumen. One could speculate that the bovine serum collectins provide a first line of defense against these microbes, when they leak into the bloodstream, without eliciting a general inflammatory reaction involving antibodies, which might be detrimental to the fine host-microbial balance in their rumen. It would be of interest to see whether bovidae are the only ruminants that express these additional serum collectins, and in addition, whether nonruminant herbivores like horses, which rely heavily on the microbial symbiosis in their large appendix, have similar collectin-based serum defense mechanisms.
Protein levels of both SP-A and SP-D in the alveolar compartment increase in response to pulmonary infection with microorganisms , and SP-D levels increase in allergen-induced eosinophilia , indicating that both proteins might function as analogues of acute phase reactants in the lung. Interestingly, hyperoxia also induces an increase in SP-A and SP-D concentrations in the alveolar compartment . As damaged epithelium is more susceptible to infection, this might represent a mechanism by which oxygen-damaged alveolar epithelium protects itself against the increased susceptibility to invading microorganisms.
The recent demonstration of MBL [9,10], SP-A [15,16,24] and SP-D [15–17,24,25] expression at mucosal surfaces suggests that these proteins have a general function in innate immunity at these locations and more specifically in the gastrointestinal tract. In addition, the finding that SP-D expression in the gastric mucosa is significantly increased during Helicobacter pylori infection, further points to the possibility of SP-D having a role in mucosal defense systems outside the lung .
Structure of the collectins
The basic functional unit of collectins is a trimer. The number of trimeric units per collectin molecule differs among the collectins. In the monomeric subunits, four structural domains can be distinguished: an N-terminal cysteine-rich domain, a collagen domain, a coiled-coil neck domain and finally a C-type lectin domain, also known as carbohydrate recognition domain (CRD) (Fig. 1).
The CRDs of collectins are compactly folded protein modules of 115–130 amino acid residues and are located at the C-terminus of the protein . Selective binding of collectins to specific complex carbohydrates is mediated by their CRDs, and requires the presence of calcium [26,27]. The actual carbohydrate binding site can be found in a shallow groove in the CRD [27–29].
Comparison of the CRD domains of soluble collectins has revealed that 22 amino acids are conserved within this domain. Most of these conserved residues, including four cysteine residues, that form intrachain disulfide bridges, are involved in proper folding of the CRD . CRDs contain several calcium binding sites, although the exact number of ligated calcium ions under physiological conditions is as yet not totally clear. Crystallographic analysis showed the presence of three and two calcium ions in the CRD of rat MBL-A and MBL-C, respectively [27,28], whereas MBL-A binding data indicated the presence of only two calcium ions per CRD . It has been suggested that the third calcium ion found in the MBL-A CRD crystal resulted from the excess of calcium in the crystallization buffer (15 mM) . However, it was demonstrated that, although it was crystallized in the presence of only 1 mM calcium, the crystal of the SP-D CRD also contained three calcium ions . Moreover, crystallization of the recombinant homotrimeric fragment of SP-D, comprising the CRD and α-helical neck domain, in the presence of about 2.5 mM calcium, but in the absence of saccharide ligand, even revealed the presence of a fourth calcium ion. This latter calcium ion was found to be present in the funnel formed by the three CRDs and close to the neck–CRD interface . Although several calcium ions are present within the CRDs of collectins, monosaccharide binding by their CRDs occurs through direct coordination of one of the calcium ions and hydrogen bond interactions with side-chains of amino acids that also serve as ligands for this calcium ion [27–29,32]. The observation that the α−helical neck domain of SP-D on its own may bind to LPS and phospholipids, and that this interaction is calcium dependent , suggests that the fourth calcium ion found in SP-D might be involved in ligand interactions as well.
The exact function of each of the two calcium ions – found in the CRD away from the neck region – that are not involved directly in ligand interactions is not exactly known, but there are indications that at least one of them is involved in the correct folding of the CRD in order to allow carbohydrate binding [31,34]. Shrive et al.  hypothesized that the calcium ions not involved directly in monosaccharide binding may be involved in binding more extended ligands, or that they are involved in the recognition of immune cell surface receptors. In addition, the electrostatic potential pattern on the surface of the protein might be altered by the additional calcium ions, thereby influencing the affinity for negatively charged ligands.
As indicated above, collectins are multimeric proteins. The degree of multimerization can greatly affect their function. This has been extensively studied for SP-D. The effects of the degree of oligomerization on various functions of this protein (which will be discussed later in this review) are given in Table 1. For the first step in the oligomerization of the collectins, the trimerization of monomers, the presence of the coiled-coil neck domain is essential [32,33, 44–47]. Recombinant proteins consisting only of the neck and CRD region are still assembled as trimers, whereas isolated CRDs lacking the neck domain are secreted as monomers [33,48], or in the case of MBL, as dimers . Recently, it was demonstrated that specific heptad repeats within the hydrophobic neck domain are required for the formation of stable trimeric SP-D subunits. It is thought that the primary role of the neck domain in molecular assembly is to align the collagen chains and thereby facilitate subsequent ‘zipper-like’ folding of the collagen helix .
Table 1. Effects of the degree of oligomerization and truncation of SP-D on various of its activities. (CRD)1, monomeric CRD; (CRD)3, trimeric CRD/neck domain with or without N-terminus of SP-D; (SP-D)3, trimeric SP-D; (SP-D)12, dodecameric SP-D; (SP-D)m, multimeric SP-D. Where no – or + symbols are given, no data are available. Column [(SP-D)12/(SP-D)m] shows reports in which no distinction was made between dodecameric and multimeric SP-D. Higher magnitude of activity is indicated by a greater number of + symbols.
Activity of SP-D
Homotrimers consisting of CRD/neck and eight Gly-Xaa-Yaa repeats from the collagen region.
The collagen-like region of the collectins consists of repeating motifs of Gly-X-Y, where X and Y can be any amino acid, but frequently are proline or hydroxyproline. The collagen helices of monomers are coiled around each other, to form a stable tensile collagen domain that is relatively resistant to proteases [49,50]. Another interesting structural feature of the collagen domain is that it can be N-glycosylated or O-glycosylated [49,50]. The repeat Gly-X-Y pattern in both MBL and SP-A is interrupted, which is thought to introduce a kink or region of flexibility into the protein, enabling the trimeric subunits to angle away from the central core, to form a structure resembling a bouquet of flowers [51,52] (Fig. 1).
The collagen domain of collectins is thought to have several (distinct) functions. It has been shown for SP-A and MBL that the collagen domain is involved in receptor-mediated effects of both proteins [53,54]. A specific GEKGEP motif within the collagen domain of MBL was shown to be involved in binding to the C1q receptor . Interestingly, the amino acid sequence of the collagen domain of SP-A contains a similar motif  that might also be involved in the demonstrated interaction of SP-A with the C1q receptor [56–58]. SP-D, which does not interact with the C1q receptor, does not contain this motif [59,60]. The collagen domain of MBL is also involved in the binding of two MBL-associated serum proteases (MASP1 and -2), which leads to the subsequent activation of the complement cascade [61,62]. The main function of the relatively large collagen domain in SP-D and the closely related bovine proteins, CL-46 and conglutinin, is thought to be the proper spacing of the separate trimeric subunits in order to be able to cross-link carbohydrate structures present on the surface of separate microorganisms, leading to their subsequent aggregation and neutralization . The positively charged collagen domain of membrane bound CL-P1 was suggested to be involved in the uptake of oxidized LDL particles .
After proper folding of the collagen helix, cysteine residues in the relatively short N-terminal domain (7–25 amino acids) form disulfide bridges between monomers, to stabilize trimeric subunits. The degree of multimerization differs between collectins, and it was demonstrated using chimeric collectin proteins, that the structural requirements for multimerization are located in the N-terminal cysteine-rich and in the collagen domain [63–66]. Deletion of particular cysteine residues within the N-terminal region leads to the formation of trimers only [38,44]. It is thought that in order to form multimers of the trimeric subunits, at least two cysteine residues have to be present in the N-terminal domain [38,44,53,67,68]. This view is supported by the fact that CL-L1, which has only one cysteine residue in this domain, is only present as a trimer . However, CL-43 is secreted as a trimer only, despite having twoN-terminal cysteine residues [69,70]. Moreover, the cysteine residues in CL-43 are found in exactly the same positions as in the highly multimerized SP-D . Therefore, it is likely that in addition to the number of N-terminal cysteine residues, other factors also contribute to the oligomerization of trimeric subunits.
The collectins that form multimers of trimeric subunits can be divided into two groups. MBL and SP-A form octadecamers of six trimeric subunits, with their overall structure resembling a bouquet of flowers [51,72], whereas SP-D and the bovine proteins conglutinin and CL-46 are assembled into dodecamers of four trimeric subunits and form a cruciform-like structure [20,49,73] (Fig. 1). In addition, SP-D can form even higher-order multimers, so-called ‘fuzzy balls’ with a mass of several million kDa . The size of fully assembled collectins ranges from 13 nm for MBL  to about 100 nm for SP-D . These differences in size are determined mainly by the manner in which trimers are assembled into oligomers (bouquet of flowers vs. cruciform), and by the length of the collagen domains of the monomeric subunits. The exact sequences that determine these different arrangements of higher-order multimers remain to be identified.
Structural basis of monosaccharide recognition by collectins
Collectins require a broad monosaccharide specificity in order to recognize a variety of cell surfaces. This broad specificity is achieved by the fact that their CRDs have a very open trough-like binding pocket. This site selects its ligands mainly on the basis of the positioning of two vicinal hydroxyl groups, which form two coordination bonds with ligated calcium, four hydrogen bonds with calcium ligands and a single apolar Van der Waals contact [27,28]. Despite their broad monosaccharide specificity, C-type lectins, to which the collectins belong, can be divided into mannose/glucose-type or galactose-type, based on relative monosaccharide specificity. Specificity of the collectin CRDs for mannose over galactose is determined by three residues (Glu-Pro-Asn) at positions equivalent to the residues Glu185 and Asn187 in MBL [74–76]. Amino acid analysis and monosaccharide inhibition studies indicated that all collectins have mannose-type CRDs [75,77] with one exception, membrane-bound CL-P1, for which the amino acid analysis predicted preference of galactose over mannose . Unfortunately, this predicted preference was not tested . Although SP-A has a preference for mannose over galactose, its CRD contains the motif Glu-Pro-Arg, indicating that the conservation of the last amino acid of the triplet determining relative saccharide affinity is not critical . However, substitution of the Glu-Pro-Asn (or Glu-Pro-Arg in the case of SP-A) triplet with Gln-Pro-Asp changes the CRD specificity from mannose-type to galactose-type , consistent with the fact that the latter triplet is conserved in the CRDs of galactose-recognizing C-type lectins [78,79]. At positions equivalent to the residues Glu185 and Asn187 in MBL, Glu and Ser are found in CL-L1 . However, as extensive sugar binding studies are not yet available for this protein, the effect of the substitution of Asn by a Ser residue within the CRD on monosaccharide specificity is not known. Mutagenesis experiments have revealed that substitution of three amino acids and the insertion of a glycine-rich repeat, is sufficient to establish both high selectivity and affinity for galactose in CRDs normally recognizing mannose-type ligands [74,75,80]. Furthermore, the mode of galactose-binding was similar to the mode of ligand binding of the galactose-recognizing CRD from the asialoglycoprotein receptor .
The molecular basis on which CRDs discriminate between mannose- and galactose-type ligands lies in the presentation of two vicinal hydroxyl groups on the 3− and 4− position of the sugar ring of hexoses. For ligand binding in mannose-type CRDs, these hydroxyl groups need to have an equatorial position, whereas for high-affinity binding by galactose-type CRDs, they have to be placed axially. Interestingly, it is thought that fucose is bound by mannose-type CRDs in a slightly different manner, as this molecule has equatorial hydroxyl groups on its 2− and 3− positions of the sugar ring which, in molecular models, superimpose on the hydroxyl groups on the 3− and 4− position of the sugar ring of mannose [27,28,81]. In addition to fucose, αd-glucose also appears to be oriented differently from mannose within the mannose-type CRD. It was predicted recently, using computational docking studies, that αd-glucose docks into the SP-D CRD via vicinal equatorial hydroxyl groups on the 2− and 3− position of the sugar ring . Although MBL has low affinity for the monosaccharide galactose, crystals of MBL complexed with this monosaccharide revealed that galactose was ligated in the MBL binding site via coordination bonds with equatorial hydroxyl groups at the 1– and 2– position of the sugar ring . This mode of binding excludes the possibility of binding to galactose residues in galactosides, as in this case the hydroxyl group at the 1– position of the sugar ring is involved in glycosidic bonding.
Binding of collectins to polysaccharides
Natural (poly)saccharide ligands for the collectins are normally attached to the surface of microorganisms, resulting in a high local density of collectin binding sites. High-affinity interactions between microorganisms and collectins depend on the density of carbohydrate ligands on the microbial surface  on the one hand, and on the degree of oligomerization of the collectin , on the other.
Clustering of glycoproteins or glycolipids on the surface of microorganisms allows for the simultaneous binding of multiple CRDs of one fully assembled collectin. In an elegant study by Lee et al.  using trimeric CRD/neck domains of MBL, it was shown that the affinity for monosaccharide subunits increased exponentially when these subunits were coupled to BSA, thereby increasing their surface density. The coupling of for instance 23 mannose monosaccharides per molecule BSA resulted in a decrease in the I50 value for this particular monosaccharide of about 85 000 times. In the same study it was found that the I50 values of various coupled monosaccharides differed dramatically: glucose was only slightly less potent than mannose in inhibiting MBL binding to a particular ligand when added as uncoupled monosaccharide, whereas when coupled to BSA, the inhibition potency differed by a factor 10. This clearly demonstrates the shortcomings of the use of monosaccharides in defining CRD specificity.
Biologically relevant interactions by collectins are brought about by the concerted binding to two or more monosaccharide units. It can be hypothesized that for native SP-A and MBL in their fully assembled form, in which the CRDs of multiple trimeric subunits all face the same direction, the affinity might be even further enhanced by simultaneously binding of up to 18 CRDs.
Multiple CRDs can also bind simultaneously to the monosaccharide units of a single polysaccharide chain. This follows from the observation that, when expressed per hexose unit, the mannose-polysaccharide mannan was more potent in inhibiting collectin binding to solid-phase bound ligands than mannose as monosaccharide. Part of this increased affinity may be explained by the interactions of (adjacent) saccharide units outside the CRD binding pocket. For SP-D it was shown by computational docking studies, that flanking saccharide residues in trisaccharides do form additional hydrogen bonds with amino acids outside the CRD binding pocket, and thereby contribute to overall binding energy . The contributions of the flanking saccharides to overall binding energy was different for various trisaccharides, suggesting that amino acids outside the CRD binding pocket might be important in fine-tuning binding specificity of collectins, consistent with the fact that the amino acid residues at these positions are not conserved in collectins.
In addition to the surface density of carbohydrate ligands, the multimerization of collectins is of eminent importance for collectin binding to multivalent ligands. Compared with trimeric collectin subunits, monomers display rather weak affinity for immobilized saccharide ligands : the Kd of the binding of a single C-type CRD with a monosaccharide ligand is in the order of 10−3m, whereas the Kd of binding of collectin trimers and higher-order multimers to polyvalent ligands is in the order of10−8 or 10−11m, respectively [33,83].
Most studies concerning carbohydrate binding of collectins have focused on the binding to terminal carbohydrate residues. However, recently it was reported that the terminal sugar residues on lipopolysaccharide (LPS) of Neisseria gonorrhoeae and Salmonella typhimurium, could not always predict MBL binding . Although direct evidence is lacking, this might suggest that MBL also interacts with internal sugar residues of LPS. In addition, SP-D has been shown recently to bind to nonterminal glucosyl residues of polysaccharides, and binding was shown to be dependent on the nature of the glycosidic linkage between monosaccharide units, as the hydroxyl groups on the 2– and 3– or on the 3– and 4– position had to be available to dock into the CRD . Further studies are needed to see whether the ability to bind to internal saccharide units is a property of all collectins, or that it is specific for SP-D. Interactions of multiple CRDs of SP-D with one polysaccharide chain could be due to binding of two CRDs of one trimeric subunit and/or to binding of CRDs of different trimeric subunits. For instance, to bridge the 51 Å spanning region between CRDs within a trimeric subunit of SP-D, an oligosaccharide of 13 or 14 residues is needed, whereas bridging CRDs of different trimers would require a polysaccharide of up to 280 sugar residues to span the maximum distance of 100 nm between opposite sides of dodecameric SP-D. Binding of the collectins to multivalent ligands most likely requires some flexibility of the protein and/or the polysaccharide. Although it is not yet known whether the CRDs within trimeric subunits display substantial flexibility, electron microscopy pictures of dodecameric SP-D and conglutinin revealed great flexibility of trimeric subunits within these higher-order multimers . For SP-A and MBL it is thought that the kink in the collagen stalk provides these oligomers with additional flexibility in order to bind to microbial surfaces [5,85]. In addition to flexibility on the part of the protein, NMR studies have shown that polysaccharide chains also have considerable flexibility [86,87] that might be of importance for collectin binding to these structures.
Functions of the collectins in host defense
Collectins interact with glycoconjugates and/or lipid moieties present on the surface of a great variety of microorganisms and allergens, and with receptors on host cells. Through these interactions, the collectins play an important role in innate host defense. The following host defense functions have been reported to date (Fig. 2).
Due to the formation of bridges between carbohydrate ligands present on the surface of different microorganisms, the interactions with intact microbes can result in massive aggregation [37,40,88–90]. This, in turn, may result in enhanced mucociliary removal by the respiratory tract, prevention of the attachment of pathogens to cell surfaces, and inhibition of microbial colonization and invasion. It may also facilitate uptake of the microorganisms by phagocytosis, but it should be noted that in some cases phagocytosis is decreased by agglutination [91,92].
Binding of MBL to microorganisms can result in inactivation of the organism by activation of the complement cascade [93,94]. On the other hand, by binding to C1q and thereby preventing association of C1q with C1r and C1s, SP-A can prevent the formation of active C1 complex .
Opsonization and activation of phagocytosis
Collectins may coat microorganisms and act as opsonins. This requires specific interactions of the collectins with receptors on phagocytic cells and may result in increased association, uptake and killing of the microorganisms [96–105]. Binding of MBL can lead to opsonization through complement activation and deposition of C3 , but can also opsonize microorganisms directly  as is the case for SP-A and SP-D. There is increasing evidence that, in addition to opsonization, where coating of microorganisms with collectins increases their uptake by phagocytes, SP-A and SP-D can also have direct, nonopsonic stimulatory effects on the uptake of microorganisms by phagocytic cells [40,97,108]. Binding to specific receptors on the surface of phagocytic cells may be responsible for this activation. At least one mechanism by which SP-A directly stimulates phagocytosis is by up-regulating the activity of the mannose-receptor, a pattern recognition receptor involved in the binding and phagocytosis of microorganisms .
Although in many cases SP-A stimulates phagocytosis and killing of pathogens, some microorganisms may increase the efficiency of their infection by using SP-A as a Trojan horse to gain entry to target cells [110–112]. MBL and conglutinin have been reported to enhance in vivo herpes simplex virus type 2 infection in mice .
Inhibition of microbial growth
Recent data indicate that collectins have direct effects on the survival of microorganisms. SP-A and SP-D were found to have direct effects on the survival of Gram-negative bacteria through mechanisms leading to increased permeability of the bacterial cell membrane . Moreover, exposure of the facultative intracellular fungal pathogen Histoplasma capsulatum to SP-A or SP-D also resulted in increased cell permeability and enhanced killing of the pathogen , whereas SP-D has a pronounced inhibitory effect on the growth and hyphal outgrowth of the fungus Candida albicans.
Modulation of inflammatory responses
A considerable number of in vitro studies have focused on the modulation of inflammatory responses by collectins. Addition of MBL to blood from MBL-deficient donors decreases the secretion of tumor necrosis factor-α (TNF-α) by monocytes in response to Neisseria meningitidis, whereas MBL-induced alteration of interleukin (IL)-6 and IL-8 secretion was found to be concentration-dependent, with stimulation and inhibition by low and high concentrations of MBL, respectively . MBL also inhibits release of TNF-α from human monocytes stimulated by rhamnose glucose polymers from streptococcal cell walls .
Also, SP-A and SP-D can modulate cytokine production [117–119]. These collectins can also modulate the production of reactive oxygen and nitrogen species, an important mechanism for killing of phagocytic cells [117–119]. In addition, SP-A and SP-D can act as chemoattractants for alveolar neutrophils or monocytes and thereby recruit the immune cells to the site of an inflammation [120–122].
Induction of inflammation by LPS or endotoxin, a component of the outer membrane of Gram-negative bacteria which is an important mediator of septic shock and acute respiratory distress syndrome, is dampened by SP-A or SP-D in a number of ways [123–126]. The mechanisms of this dampening include scavenging of the LPS  and binding to the LPS receptor CD14 on macrophages, which blocks LPS-mediated inflammatory responses of macrophages .
A matter of controversy in studies concerning SP-A has been whether this collectin should be considered anti-inflammatory or pro-inflammatory: some groups reported that interaction of SP-A with macrophages stimulates the production of proinflammatory mediators, such as TNF-α and NO, while others observed inhibition by SP-A of the production of these mediators (reviewed in [117–119]). A partial explanation of these conflicting results may come from recent observations that the functional outcome of SP-A exposure is determined by the state of macrophage activation. For example, SP-A enhances LPS-induced production of NO by interferon-γ (IFN-γ)-treated macrophages, while it inhibits LPS-induced NO production in macrophages not treated with IFN-γ. In older experiments in which direct stimulatory effects of SP-A on cytokine release by macrophages was found, the result may have been due to contamination of the SP-A with LPS. However, Guillot et al.  showed that SP-A can stimulate cytokine secretion by macrophages, even when the SP-A has been treated with polymyxin to remove LPS. On the contrary, this was not seen by others using polymyxin-purified SP-A . Differences in cell types and experimental variables may be the cause of this discrepancy.
A recent publication  provided evidence that SP-A and SP-D act in a dual manner to enhance or suppress inhibitory mediator production depending on binding orientation. The data in that paper indicate that SP-A and SP-D bind signal regulating protein α (SIRPα; a transmembrane protein involved in signal transduction) through their CRDs to initiate a signaling pathway that blocks proinflammatory mediator production. In contrast, their collagenous tails stimulate proinflammatory mediator production via binding to calreticulin/CD91. The authors  propose a model in which SP-A and SP-D help maintain a non/anti-inflammatory lung environment by stimulating SIRPα on resident cells via their CRDs. On the other hand, according to this model, interaction of these CRDs with pathogen-associated molecular patterns on foreign organisms or damaged cells and presentation of the collagenous tails in an aggregated state to careticulin/CD91 stimulates phagocytosis and proinflammatory responses.
In vivo studies using mice made deficient in SP-A or SP-D, show that the anti-inflammatory effects of both lung collectins predominate in vivo: exposure of SP-A –/– mice to intact microorganisms [131–133] as well as to LPS  results in increased inflammatory reactions in the lung compared to wild-type mice. Furthermore, increased pulmonary TNF-α concentrations, detected in SP-A –/– mice after exposure to LPS, could be normalized by the administration of exogenous SP-A . In vivo, SP-D is thought to have an anti-inflammatory effect as well, because, compared to wild-type mice, SP-D –/– mice show increased inflammatory reactions in their lungs after infection with bacteria  or viruses .
Modulation of the adaptive immune system
In vitro, both SP-A and SP-D can inhibit the proliferation of T-lymphocytes, associated with a lowered IL-2 production [135,136]. Moreover, while SP-D enhances bacterial antigen presentation by bone marrow-derived dendritic cells , SP-A inhibits the differentiation of immature dendritic cells into mature dendritic cells . In vivo, absence of SP-A in mice has effects on various lymphocyte subgroups .
Modulation of allergic response
The lung collectins SP-A and SP-D have been shown to mediate a number of anti-allergic effects [139–142], including inhibition of IgE binding to allergens, suppression of histamine release from basophils in the early phase of allergen provocation, and inhibition of lymphocyte proliferation in the late phase of bronchial inflammation.
Effects related to apoptosis
SP-A was reported to protect pulmonary alveolar type II epithelial cells from apoptosis . In addition, there is evidence to suggest that MBL, SP-A and SP-D stimulate apoptotic cell clearance by alveolar macrophages [144,145].
Interactions with microorganisms and their carbohydrate surface epitopes
Numerous studies have demonstrated binding of collectins to the whole range of microbes, from viruses to metazoa. Microbial targets for SP-A have been listed in references [118,146–149]; those for SP-D in references [146,147, 149,150] and those for MBL, conglutinin, CL-43 and CL-P1 in . Interestingly, in many cases, binding was found to be dependent on the growth conditions of the particular microbe, suggesting a complex interplay between host and microorganism. Most microorganisms display a diverse array of complex glycoconjugates on their outer surface, which represent possible ligands for the collectins. As most data are available for MBL and the surfactant proteins A and D, we will focus on these proteins.
Bacteria display on their outer surface an array of complex glycoconjugates, many of which are highly abundant or contain repeating saccharide units, thereby representing ligands for collectin binding. In several studies, the inability of collectins to bind to certain bacterial strains, correlated with increased pathogenicity . In addition, capsule production by bacteria is often accompanied by decreased collectin binding and a subsequent increase in pathogenicity . These findings clearly point to the importance of collectins in the early phase of host defense against bacteria.
In Gram-negative bacteria, LPS has been found to represent the most important ligand for collectin-mediated elimination. Initially it was thought that only bacteria displaying rough and not smooth LPS are bound by collectins. However, recently it was found that SP-A and SP-D bound both smooth and rough forms of Pseudomonas aeruginosa, suggesting that smooth LPS is recognized by both proteins on this type of bacteria . In addition, SP-D does selectively bind to smooth forms of LPS expressed by O-serotypes of Klebsiella pneumoniae with mannose-rich repeating units in their O-polysaccharides . In contrast, K. pneumoniae strains containing galactose-rich repeats in their O-polysaccharides were not bound, in agreement with the known low affinity of SP-D for the monosaccharide galactose . Rough forms of LPS act as a ligand for most collectins, although the latter bind to different sites on the LPS molecule: SP-A is thought to interact with the lipid-A moiety of LPS , whereas SP-D binds to LPS core saccharides . On the other hand, it still needs to be elucidated which parts of the LPS molecule are involved in MBL binding. There are indications that besides the type of terminal sugar residue, also the folding of the LPS molecule is important . Furthermore, it was found that the presence of glucose residues at the terminal LPS structure correlated with MBL binding, and more interestingly, a higher level of binding occurred to mutant forms of LPS terminating with heptose sugars . However, monosaccharide inhibition studies using heptose sugars have not been performed so far, so it is still unclear whether these heptose sugars represent real MBL binding sites, or whether the observed correlation is coincidental. It is interesting to note that SP-A binds to Haemophilus influenzae not via its LPS, but instead via its glycosylated major outer membrane protein P2 .
There are also numerous Gram-positive bacteria that are bound by the collectins. The amount of data concerning ligands for the collectins on this type of bacteria is still very limited. However, we recently found that lipoteichoic acid (LTA) of Bacillus subtilis and peptidoglycan of Staphylococcus aureus represent ligands on Gram-positive bacteria for SP-D, but not for SP-A . The structure of LTA varies among different strains of Gram-positive bacteria, whereas the structure of peptidoglycan in these bacteria is practically constant. Therefore, peptidoglycan may represent a universal ligand for SP-D. Although SP-A has been shown to bind to several Gram-positive bacteria , the surface structures that account for these interactions are as yet not known. In contrast, MBL has been shown to interact with a wide variety of Gram-positive bacteria [156,157], and various types of LTA were identified as MBL ligands .
The important lung pathogen Mycobacterium tuberculosis is bound by both SP-A and SP-D. To sustain a chronic infection and cause disease, M. tuberculosis needs to enter mononuclear phagocytic cells, where this pathogen survives by subverting cellular antimicrobial defense mechanisms . While the interaction of M. tuberculosis with SP-D reduces the uptake of bacilli by macrophages , SP-A promotes this uptake . Both proteins seem to interact with M. tuberculosis via lipoarabinomannan (LAM) molecules on their surface [89,159]. SP-A also binds to lipomannan (LM). Besides the presence of mannose residues on LAM and LM, fatty acids are an absolute requirement for SP-A binding . SP-D interacted with M. tuberculosis via mannose residues of the LAM moiety of M. tuberculosis.
SP-D binds to Mycoplasma pneumoniae via interactions with its membrane glycolipids .
Binding of collectins to viruses is especially interesting because viruses make use of the host cell machinery for the synthesis, folding and transport of proteins to the site of virus assembly at the cell surface. This machinery includes the array of biosynthetic and trimming enzymes responsible for attachment and processing of the oligosaccharides on their glycoproteins. No virus has been found to encode enzymes which can affect the glycosylation of its proteins by controlling commitment to particular processing pathways . The dependence on host cell glycosylation machinery is demonstrated by the fact that infection of different cell types with, for instance, the respiratory viruses influenza A virus (IAV) or human respiratory syncytial virus (RSV) results in different oligosaccharide side-chains on their glycoproteins [162,163]. Most studies concerning the binding of collectins to IAV have used virus grown in embryonated hen eggs, which results in the expression of different oligosaccharide side-chains on the viral surface glycoproteins compared to IAVs grown in mammalian cells [163–165]. Moreover, most glycans of the hemagglutinin (HA)1 subunit have been identified as complex-type oligosaccharides, similar to that found on membrane-bound glycoproteins in mammalian systems . It is therefore tempting to speculate that the acquisition of oligosaccharides antigenetically identical to those of the host helps the virus to escape the collectin-based immune defenses of the host organism, and is thus one of the mechanisms underlying antigenic drift .
Another interesting issue concerning the ‘self’ oligosaccharides exposed by many enveloped viruses, is how the collectins discriminate between ‘self’ oligosaccharides presented as part of the glycoproteins of the plasma membrane of the host cells and, the same oligosaccharides exposed on viral glycoproteins. One explanation might be that this discrimination is caused by a greater density of these epitopes in the latter situation. Furthermore, incomplete processing of the attached oligosaccharides, which increases the presence of oligosaccharides of the high-mannose type, might contribute to collectin binding to viruses. The presentation of oligosaccharides in a particular glycoprotein might further influence collectin binding. Although the carbohydrate structures present on viruses are of host origin, several lines of evidence suggest that collectins may play an important role in host defense against viral infections. These proteins bind to the enveloped viruses like IAV, herpes simplex virus type 1 (HSV-1), RSV, HIV, cytomegalovirus and the nonenveloped rotaviruses. Generally, collectins are thought to bind viruses or virus-infected cells in a manner that involves an interaction between the CRD of the collectin and surface-exposed glycoproteins containing oligosaccharides of the high-mannose type. In contrast, the binding of SP-A to IAV- and HSV-1-infected cells is mediated by interaction between the sugar binding activity of the virus and a carbohydrate moiety attached to SP-A [166,167]. The binding of SP-A to HSV-1 viral particles results in their enhanced uptake by alveolar macrophages . Collectin binding to IAV has been extensively studied. MBL, SP-D, SP-A and conglutinin all display anti-IAV activity in vitro, although their method of action differs . Although all collectins show inhibition of viral hemagglutination activity, SP-A was substantially less potent . This lesser potency of SP-A might be caused by the different manner of interaction with the HA moiety of IAV. SP-D and conglutinin are thought to inhibit mainly viral replication by forming large viral aggregates. These aggregates could then be removed via mucociliary clearance or by increased uptake by phagocytic cells. In addition, SP-D , MBL , conglutinin , but not SP-A , can prevent the IAV-induced inhibition of the superoxide production by neutrophils in response to the chemotactic peptide formylmethionylleucylphenylalanine, while SP-D , MBL  and conglutinin  enhance the IAV-induced H2O2 production by neutrophils. SP-D also increases the internalization of IAV by neutrophils [37, 168]. In contrast, MBL binding to IAV does not result in enhanced phagocytosis by neutrophils, but MBL dependent complement activation of IAV-infected cells  might contribute to the defense against IAV.
While SP-A binds to IAV through interaction between sialic acid residues on the carbohydrate moiety located in its CRD and (presumably) the sialic acid receptor present on the HA of IAV , SP-D from various species binds to IAV through interaction between the CRD of SP-D and oligosaccharide moieties located on the HA of IAV. Recently however, it was found that, like SP-A but in contrast to SP-D from all other animal species studied thus far, porcine SP-D contains a sialylated oligosaccharide moiety in its CRD [172,173]. This gives porcine SP-D an additional way of interacting with IAV: beside binding the carbohydrate moieties on HA of IAV, porcine SP-D can also bind IAV through interactions between the sialic acid residues on the carbohydrate moiety located in its CRD and the sialic acid receptor present on the HA of IAV. The presence of the sialylated oligosaccharide moiety enhances the anti-influenza activity of porcine SP-D, as demonstrated by assays of viral aggregation, inhibition of infectivity, and neutrophil response to IAV . Hemagglutination inhibition assays revealed that porcine SP-D displays substantially greater inhibitory activity against various IAV strains than SP-D from other animal species . The CRD carbohydrate of porcine SP-D is exclusively sialylated with α(2,6)-linked sialic acid residues . Studies of the enzymatic modification of the sialic acid linkages present on porcine SP-D demonstrated that the type of linkage is important for hemagglutination inhibitory activity . The more effective interaction between IAV and SP-D in the pig could result in a more effective clearance of IAV. Alternatively, however, it is conceivable that the more effective nonspecific immune response through SP-D in the pig could inhibit the induction of specific acquired immune responses which are elemental for the ultimate elimination of IAV. Evasion of IAV-induced immunity could thus give rise to conditions where IAV infection can persist. It is thought that pigs may act as ‘mixing vessels’ in which reassortment of IAV may occur upon coinfection with human and avian IAV strains . The presence of the sialylated oligosaccharide in the CRD of porcine SP-D may therefore play a role in providing conditions by which pigs can act as ‘mixing vessel’ hosts that can lead to the production of reassortant, pandemic strains of IAV.
Ghildyal et al.  described that SP-A, but not SP-D and MBL, bound to respiratory syncytial virus (RSV). In vivo, SP-A was found to play an important role in the clearance of this virus . In contrast to Ghildyal et al. , Hickling et al.  showed that SP-D did bind to RSV, and that the membrane envelope G-glycoprotein was involved in this interaction. Moreover, it was found in the same study that the trimeric recombinant head-neck fragments of SP-D had a protective effect on RSV infection in vivo, suggesting that multimerization of SP-D is not required for its protective role against RSV. It might be that carbohydrate moieties on the viral surface that are involved in receptor-mediated viral uptake by host cells, are bound by SP-D, thereby blocking viral entry into the host cell and subsequent infection . Furthermore, it cannot be excluded that direct influences upon host cells are involved in the protective role of SP-D against viruses, e.g. by altering production of certain cytokines. The cause of the discrepancy between the data by Ghildyal et al.  and those of Hickling et al.  concerning SP-D binding to RSV is unclear.
MBL binds to HIV-1 and HIV-2 via gp120 and gp110, respectively. Both viral glycoproteins were found to contain oligosaccharide side-chains of the high-mannose type of 7, 8 or 9 mannose residues. The consequences of MBL binding to HIV are not known, but it could lead to neutralization of the virus via complement activation, or lead to enhanced uptake by phagocytic cells, and thereby, depending on whether the phagocytes are able to kill the virus after stimulated uptake, either enhance or diminish infection of the whole organism. Interestingly, the presence of sialic acid residues on the carbohydrate moiety of gp120 has been shown to decrease MBL binding, indicating that modification of the high-mannose oligosaccharides in the Golgi system may lead to modification of collectin-mediated defense against the virus [178,179].
Most fungi are considered to be opportunistic pathogens, only causing disease in the absence of an adequate host immune response. An important site of entry for fungal infections is the lung. Therefore, most studies have focused on the effects and binding of the pulmonary collectins SP-A and SP-D to fungal pathogens. Possible binding sites for collectins on the surface of fungi can be divided into two groups. Firstly, structural polysaccharides consisting of repetitions of the same oligosaccharide elements can act as sites for collectin binding. In addition, many fungi express highly glycosylated proteins on their surface, which can also function as ligands for collectin binding. Some fungi produce a capsule, which is thought to represent a major virulence factor. Capsule production often leads to decreased collectin binding compared to acapsular fungal variants .
One of the first carbohydrate structures that was found to interact with the collectins was mannan, a structural component of the cell wall of the bakers yeast, Saccharomyces cerevisiae. Mannan is a branched homopolymer of mannose-residues that are coupled to each other via varying glycosidic linkages. SP-D binds and subsequently aggregates S. cerevisiae via binding with its C-type lectin domain . Mannan and β(1–6)linked glucan represent major ligands for SP-D on the cell wall of S. serevisiae. Other structures involved could include mannoproteins. Interestingly, SP-A does not bind to S. cerevisiae, although SP-A does bind to its isolated cell wall component mannan . The explanation for this apparent discrepancy may be that the specific mannan conformation on the yeast cell surface does not allow SP-A binding .
In addition to causing disease in immunocompromised individuals, Aspergillus fumigatus can cause allergen-induced allergic bronchopulmonary aspergillosis . A. fumigatus conidia are bound by both SP-A and SP-D, and binding results in enhanced aggregation and killing by phagocytic cells . Moreover, both pulmonary collectins interact with the glycosylated cell wall proteins gp55 and gp45 of A. fumigatus, inhibit specific IgE binding to these allergens and block histamine release from sensitized basophils . The trimeric head-neck domain of SP-D was found to be enough to protect mice against fungal hypersensitivity [43,183]. Although the mechanisms are still unknown, these results clearly implicate pulmonary SP-A and SP-D in the modulation of allergic responses. In addition, after allergic airway inflammation caused by A. fumigatus, SP-D levels in bronchoalveolar lavage fluid were found to be increased [139,184].
SP-D binding to Pneumocystis carinii is mediated via interaction with the mannose-rich cell wall glycoprotein, gpA. Interestingly, pulmonary infection with this fungus leads to increased amounts of SP-D protein in bronchoalveolar lavage fluid , and the binding capacity of SP-D recovered from the bronchoalveolar lavage fluid of infected lungs is higher than that of recombinant SP-D, possibly due to its higher oligomerization state . Although coating of P. carinii with SP-D was shown to increase the adhesion of fungal cells to macrophages , SP-D-induced aggregation seems to impair subsequent phagocytosis by alveolar macrophages . The net effect of SP-D on P. carinii clearance in vivo is still unknown. SP-A-deficient mice show increased susceptibility to P. carinii infection , implying a role for SP-A in the host defense against this fungus in vivo. SP-A binds via its CRD to gpA on the surface of P. carinii, and in three studies, SP-A coating of P. carinii was shown to stimulate binding to alveolar macrophages, which supports the idea that SP-A functions as a nonimmune opsonin . However, in another report, data indicated that SP-A decreased P. carinii attachment to alveolar macrophages and subsequent phagocytosis .
Binding of SP-D to Candida albicans not only induces aggregation of this organism, but more interestingly, coincubation of SP-D and C. albicans results in fungal growth inhibition, and decreased hyphal outgrowth, suggesting a direct effect of SP-D on fungal metabolism . Furthermore, binding of SP-D to C. albicans inhibits phagocytosis of this fungus by alveolar macrophages , probably due to the large size of the formed C. albicans complexes, which are several times larger than alveolar macrophages. SP-A also binds to C. albicans, but phagocytosis of viable C. albicans by alveolar macrophages was not augmented . In contrast, SP-A was found to inhibit increased phagocytosis induced by serum opsonization of C. albicans.
Both SP-A and SP-D bind to the yeast-like fungus Cryptococcus neoformans, although more binding was detected to the acapsular form [90,189,190]. In addition, mannoproteins of acapsular yeast cells and the major capsular component glucuronoxylomannan were identified as ligands for SP-D . Binding of SP-D to C. neoformans leads to a massive aggregation of acapsular but not of encapsulated C. neoformans. Moreover, secreted glucoronoxylomannan can inhibit the SP-D induced aggregation . Binding of SP-A to C. neoformans does not result in the increased uptake by phagocytic cells , in analogy with the effect of SP-A binding to C. albicans. MBL was found to bind to C. albicans and acapsular C. neoformans. Unfortunately, possible effects of these interactions were not studied.
MBL binds to a number of blood stage protozoa, including Plasmodium falciparum, Trypanosoma cruzi, and several Leishmania species. Glycolipids and N-linked glycans of the high-mannose type were identified as potential ligands on their surface [192–194]. Leishmania species are intracellular pathogens, mainly infecting macrophages. Several lines of evidence indicate that this parasite uses the lectin pathway of complement activation to its advantage. To enter macrophages, it uses the coating of its surface with complement, which stimulates its uptake via complement receptors on the surface of macrophages [195,196]. Therefore, MBL potentially provides a mechanism for cell entry via activation of the lectin-pathway of complement activation. This hypothesis is supported by the observation that there is a correlation between the plasma MBL concentration and the susceptibility to visceral leishmaniasis . Interestingly, intracellular Leishmania mexicana amastigotes secrete a structure called proteophosphoglycan, which is bound by MBL, resulting in turn in the activation of the complement cascade . As activation of the complement cascade results in the release of several pro-inflammatory peptides, it is thought that this is a mechanism used by the parasite to attract infectable monocytes to the site of infection . MBL also binds to several developmental stages of the multicellular blood fluke, Schistosoma mansoni, and at least in vitro, this binding results in the activation of the complement cascade . In addition, we recently demonstrated SP-D binding to specific larval stages of S. mansoni that are known to migrate through the lung .
Interactions of collectins with host cells
Collectins also display specific interactions with host cells. Immune cells are the most frequently studied cells in this respect, although for SP–A interactions with type II alveolar cells have also been studied in great detail. An important function of collectins is their ability to enhance phagocytosis of microorganisms. The mechanisms by which they stimulate the uptake of specific pathogens include opsonization of microorganisms [96–105], as well as direct interactions with phagocytic cells [40,97,108]. Stimulation of phagocytosis through opsonization by collectins is in most cases mediated via their CRD-dependent binding to microorganisms, after which specific cellular receptors are involved in the internalization of the collectin-coated microorganisms. Although an increasing number of receptors for collectins have been identified on host immune cells over the last decade (Table 2), the picture is far from complete.
Table 2. Binding of collectins to cell surface receptors and the biological consequences. Yes, direct binding detected; No, direct binding studied, but not (yet) detected; ?, no information available in literature about direct binding. LPS, lipopolysaccharide; TNF-α, tumor necrosis factor α; cC1qR, C1q recepor, also known as cell surface calreticulin; C1qRp, C1q receptor stimulating phagocytosis; CR1, complement receptor 1; SIRPα, signal regulating protein α; SP-R210, surfactant protein receptor of 210 kDa; gp-340, glycoprotein 340; TLR2, toll-like receptor 2; TLR4, toll-like receptor 4.
Phagocytosis of microorganisms  Inhibition phospholipid secretion by alveolar type-II cells  Enhancement of nitric oxide production  Enhancement of TNF-α production  Inhibition of T-lymphocyte proliferation 
Because of the structural similarity between C1q and the collectins MBL and SP-A, one of the first receptor types identified as a general collectin receptor involved in the collectin-mediated stimulation of phagocytosis was the C1q receptor, later identified as calreticulin . It was found that MBL, SP-A and conglutinin interact with this receptor, while binding could be inhibited using C1q and C1q collagen stalks, demonstrating that the collectin collagen domain is involved in receptor binding [56,213]. SP-A-induced phagocytosis of S. aureus by monocytes was shown to be dependent on the presence of the C1q receptor on the cell surface . In addition, SP-A-mediated attachment of M. tuberculosis to alveolar macrophages was shown to be inhibited by type V collagen, suggesting that the C1q receptor was involved [214,215]. SP-D can bind in a lectin-independent manner to alveolar macrophages , but in contrast to the binding of other members of the collectin family, the C1q receptor appears not to be involved in this interaction [216,217]. More recently, it was demonstrated that collectins stimulate the phagocytosis of apoptotic neutrophils  and jurkat cells [145,201] by (alveolar) macrophages. Collectins are thought to bind to apoptotic cells via CRD dependent mechanisms, whereas they probably interact with their collagen domain to the cell surface calreticulin/CD91 complex on macrophages, after which ingestion starts . As calreticulin lacks a transmembrane domain the endocytic receptor protein CD91 is thought to be involved in the transduction of the signals initiating engulfment after the calreticulin/collectin complex has been formed [144,201]. In vitro, SP-A was found to be more potent than SP-D in stimulating engulfment of apoptotic cells by alveolar macrophages , whereas using knockout (SP-A or SP-D) and overexpressing (SP-D) mice, only SP-D was found to alter apoptotic cell clearance from naive murine lung, suggesting that SP-D plays a particularly important role in vivo. In support of the suggested role of SP-D in the clearance of apoptotic cells, is the finding that in SP-D –/– mice an increased number of apoptotic alveolar macrophages is present, whose number is reduced by the intrapulmonary administration of a head and neck fragment of SP-D produced by recombinant techniques .
The structure of an additional C1q receptor was more recently elucidated, and demonstrated to be a highly glycosylated protein of 126 kDa. Due to its demonstrated involvement in the enhancement of phagocytosis of C1q- and collectin-opsonized microorganisms, it was named C1qRp (C1q receptor stimulating phagocytosis). Although direct binding of the collectins to this receptor has never been demonstrated directly, sequestration of the receptors using specific antibodies directed to this receptor, decreased collectin-mediated phagocytosis . C1qRp is expressed in cells of myeloid origin, platelets and on endothelial cells . It should be noted that Tino and Wright  demonstrated that stimulation of phagocytosis of specific pathogens by SP-A is inhibited in monocytes adhering to surface-bound C1q, but not to similarly treated alveolar macrophages. This suggests that depending on the cell type, receptors other than C1qR and C1qRp are also involved in collectin-based stimulation of phagocytosis.
MBL has been shown to interact with the complement receptor 1 (CR1/CD35) in a CRD-independent manner that was inhibitable by C1q . Furthermore, it was shown that CR1 was involved in the MBL-enhanced phagocytosis by neutrophils of Salmonella montevideo suboptimally opsonized with IgG .
MBL, SP-A and SP-D have all been shown to bind directly to the cell surface LPS receptor CD14. However, the specific domains, on both the receptor and the collectin, involved in these interactions differ for the various collectins: MBL and SP-A bind to the peptide portion of CD14, whereas SP-D binds to the N-linked glycan moiety of CD14 [126,204]. Moreover, the neck domain of SP-A was shown to be involved in SP-A binding to CD14 , whereas for SP-D its CRD mediated binding to this receptor . The domain of MBL involved in CD14 binding has not yet been identified, but is probably not its CRD, as binding of MBL to CD14 could not be inhibited by competing sugars and the presence of EDTA . The modulation of cellular effects elicited by the above mentioned collectins also varies upon stimulation with different bacterial membrane products. The fact that the type of LPS influences SP-A binding and subsequent receptor stimulation was clearly demonstrated by Sano et al. , who showed for alveolar macrophages, that SP-A inhibited TNF-α secretion elicited by smooth LPS, probably via binding of SP-A to CD14, thereby preventing smooth LPS from interaction with this receptor. However, SP-A binding to rough LPS enhanced the interaction of the LPS with CD14 and subsequent TNF-α release. In contrast to this study, Stamme et al.  found that SP-A could prevent the rough LPS-induced translocation of the transcription factor NF-κB, which is known to stimulate the secretion of TNF-α. This inhibitory effect was most probably caused by preventing the formation of LPS/LPS-binding protein complexes. The cause of this contrast in observations is not exactly known, but may include the use of different types of rough LPS and differences in the percentage of serum (containing LPS binding protein) that were used in both studies. For SP-D the picture was less complex, as this protein was shown to inhibit the binding of both smooth and rough LPS to CD14 . Although MBL has been shown to bind to CD14 as well as to streptococcal rhamnose/glucose polymers, this protein inhibited the interaction of these bacterial membrane products with CD14 on human monocytes, thereby preventing the subsequent release of TNF-α.
It was reported recently  that SP-A and SP-D inhibit LPS-induced macrophage cytokine release by interacting via their CRD with signal regulating protein α (SIRPα). This is a transmembrane protein involved in signal transduction that contains a glycosylated extracellular region . How the effect of SP-A via SIRPα is related to the observed effects brought about via binding of SP-A to CD14 and LPS [205,222] remains to be determined.
Besides the function of SP-A in innate immunity, in vitro studies have demonstrated that this protein can inhibit phospholipid secretion by lung alveolar type II cells, and that a specific SP-A receptor with a molecular mass of 210 kDa, designated SP-R210 is involved in this inhibition. Although binding of SP-A to SP-R210 was shown to require the presence of calcium, binding could not be inhibited by mannan, suggesting that the lectin activity of SP-A is not responsible for the observed interaction . In addition to its presence on alveolar type II cells, SP-R210 has been detected on macrophages and the macrophage cell-line U937 . Moreover, for alveolar macrophages it was shown that phagocytosis and subsequent killing of SP-A-opsonized M. tuberculosis was dependent on the presence of this receptor on their cell surface . SP-R210 is also involved in inhibition of T cell proliferation by SP-A, via an interaction with the SP-A collagen domain, most probably involving a highly charged RGD motif .
On bovine alveolar macrophages, an additional SP-A-specific receptor was shown to be present. SP-A bound to this 40-kDa protein in a calcium-dependent and mannose-inhibitable manner, indicating that the CRD of SP-A is involved in interactions with this receptor . Unfortunately until now, no data are available about possible functions of this 40 kDa SP-A receptor.
Glycoprotein-340 (gp-340) has been shown to interact with both SP-A and SP-D. Although this interaction was calcium-dependent, it did not involve the lectin activity of SP-A or SP-D [209,210], suggesting a protein–protein interaction. Furthermore, the expression of gp-340 and SP-D colocalize throughout the body, suggesting that SP-D has a role as an opsonin receptor .
As suggested by Kuan et al. , the fact that SP-D binding to macrophages is, for a large part, mediated via its CRD might indicate that besides the above mentioned receptors, glycolipids are also involved in SP-D binding to these cells.
SP-A has recently been shown to bind via its CRD to toll-like receptor (TLR)2, thereby preventing the induction by peptidoglycan (a cell wall component of Gram-positive bacteria) of TNF-α secretion by U937 and alveolar macrophages . It was demonstrated recently that SP-A-induced cytokine synthesis in mouse macrophages is critically dependent on functional TLR4 , but it remains to be determined whether a direct interaction between SP-A and TLR4 is involved in this effect.
SP-A and SP-D have been shown to stimulate chemotaxis of alveolar macrophages [122,227], but not of peripheral blood monocytes , suggesting the presence of specific cellular receptors on the former cell type. The effect of fully assembled SP-A and SP-D on chemotaxis correlated with their ability to stimulate directional actin polymerization . Interestingly, the ability of both SP-A and SP-D to stimulate chemotaxis of alveolar macrophages did not seem to be mediated by the lectin activity of both proteins, as it was not inhibitable by sugars, and for SP-A at least partly mediated via its collagen domain . Cai and coworkers  demonstrated that the trimeric head-neck domain of SP-D was sufficient to stimulate chemotaxis of peripheral neutrophils, and that this effect could be blocked by maltose, strongly suggesting that SP-D-induced chemotaxis of this cell type is stimulated by CRD-dependent interactions. These conflicting results concerning the involvement of CRD in the mediation of chemotaxis might be explained by the assumption that the chemotactic effect in both cell types is exerted by different cellular receptors. Although SP-A does not directly stimulate neutrophil chemotaxis, this protein does alter neutrophil responsiveness to chemoattractants: SP-A was found to enhance chemotaxis of inflammatory alveolar neutrophils, whereas it has the opposite effect on peripheral neutrophils . In agreement with a function of SP-A in chemotaxis within the lung is a recent report demonstrating that SP-A stimulates the recruitment of neutrophils in the lungs of preterm lambs . In SP-A –/– mice, there is an increased influx of immune cells into the lung upon infection, suggesting that SP-A decreases the influx of these cells into the lung. However, this increased influx of immune cells might be caused by secondary changes in the SP-A –/– mice, e.g. altered cytokine levels. In contrast to the chemotactic effect of both lung collectins, MBL does not seem to directly stimulate chemotaxis .
Conclusions and future directions
Collectins play an important role in innate immunity. Although most functional information regarding collectins is based on in vitro experiments, more recently, knock-out mice have become available for SP-A , SP-D [231,232] and MBL-A  that support the generally accepted concept that the main function of these proteins lies in the innate immunity against microorganisms [131,133,134, 185,233]. In addition, the fact that in humans mutations within the MBL-gene influence MBL serum levels, and at the same time susceptibility to certain pathogens (as reviewed by Turner et al. ), supplies additional evidence that this molecule is important in innate immunity. Collectins exert their role via a diversity of mechanisms, and they interact with surface structures present on host cells as well as on microorganisms. Binding of collectins to microorganisms often occurs via CRD-dependent interactions with glycoconjugates on their surface. On the other hand, binding to host cells is more complex, including both CRD-dependent interactions with surface glycoconjugates as well as protein–protein interactions that involve, for instance, the collectin collagen or neck domain.
Monosaccharide inhibition studies have revealed that collectins preferentially bind to monosaccharide units of the mannose-type. At present a great amount of data is available at the molecular level regarding the manner in which collectins bind their target monosaccharides. This not only increases our understanding why some pathogens are bound by the collectins while others are not, but may in the future allow for the development of recombinant proteins which have altered carbohydrate specificity. It is known, for instance, that a large part of the cases of pneumonia are caused by K. pneumoniae serotypes containing galactose-rich O-antigens on their LPS . These bacteria have been found to be relatively resistant to SP-D-mediated inactivation in vitro. Therefore, recombinantly produced SP-D, in which the carbohydrate specificity is altered from mannose-type to galactose-type, might have a therapeutic value for patients suffering from pneumonia caused by these types of bacteria. This hypothesis might be broadened to other collectins. As a first step to evaluate this hypothesis, it could be considered if recombinant collectins with galactose specificity can inactivate these high-galactose containing serotypes of K. pneumoniae by aggregation or stimulation of their phagocytotic uptake. Concerning the inactivation of microorganisms normally resistant to collectin mediated inactivation, additional experiments in which, for instance, SP-D with altered carbohydrate specificity is expressed in the SP-D knockout background could elucidate the effects of galactose-specific collectins in vivo. Furthermore, expressing SP-D with galactose specificity in this SP-D –/– background could provide information on whether the mannose specificity is required to restore normal surfactant homeostasis, and prevent the development of lung emphysema, alterations normally seen in the SP-D –/– phenotype. Many glycoconjugates present on the surface of host cells contain galactose units. Therefore the effects of collectins with altered carbohydrate specificity on these cells need to be investigated as well.
Although the list of microorganisms and their respective carbohydrate surface ligands that are bound by the collectins steadily increases, the exact pathways that result in the distinct effects of the different collectins upon binding to various microorganisms, are as yet not clear. It is for instance intriguing that binding of SP-D to K. pneumoniae results in their increased phagocytosis by alveolar macrophages , while binding of SP-D to M. tuberculosis has the opposite effect [41,89]. Therefore, a challenging task for the future will be to elucidate the mechanisms by which collectin binding to one type of pathogen leads to its removal, while binding to another increases its infectivity.
Recently, a number of receptors have been identified on the surface of host cells. However, as yet, not all collectin-mediated effects upon host cells can be explained by interactions with these receptors. A large amount of work still needs to occur in order to link collectin-mediated effects upon host cells to already identified receptors, as well as to identify additional collectin receptors. A better understanding of collectin-mediated immunity may in the future allow the identification of disease states in which the therapeutic administration of collectins may be beneficial.
The authors received financial support from the European Commission (contract QLK2-CT-2000–00325).