MMBL proteins: from lectin to bacteriocin

Maarten G. K. Ghequire1, Remy Loris2,3, and René De Mot1#

1 Centre of Microbial and Plant Genetics, KU Leuven, Kasteelpark Arenberg 20, 3001 Heverlee, Belgium

2 Molecular Recognition Unit, Department of Structural Biology, Vlaams Instituut voor Biotechnologie, Pleinlaan 2, 1050 Brussel, Belgium

3 Structural Biology Brussels, Department of Biotechnology (DBIT), Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussel, Belgium

#: author of correspondence, , Tel. +32 (0) 16 329681, Fax: +32 (0) 16 321963

Running title: MMBL lectins and bacteriocins

Abbreviations used: GNA, Galanthus nivalis agglutinin; MACPF, membrane attack complex component/perforin ; MMBL, monocot mannose-binding lectin

Keywords: LlpA, antagonism, chimeric lectin, MMBL, bacteriocin, phylogeny


Abstract

Arguably, bacteriocins deployed in warfare among related bacteria, are among the most diverse proteinacous compounds with respect to structure and mode of action. Identification of the first prokaryotic member of the so-called “monocot mannose-binding lectins” (MMBLs or GNA lectin family) and discovery of its genus-specific killer activity in the Gram-negative bacteria Pseudomonas and Xanthomonas has added yet another kind of toxin to this group of allelopathic molecules. This novel feature is reminiscent of the protective function, based on antifungal, insecticidal, nematicidal or antiviral activity, assigned to or proposed for several of the eukaryotic MMBL proteins that are ubiquitously distributed among monocot plants but also occur in some other plants, fishes, sponges, amoebas and fungi. Direct bactericidal activity can also be effected by a C-type lectin but this is a mammalian protein that limits mucosal colonization by Gram-positive bacteria. The presence of two divergent MMBL domains in the novel bacteriocins raises questions about task distribution between modules and the possible role of carbohydrate binding in specificity of target strain recognition and killing. Notably, bacteriocin activity was also demonstrated for a hybrid MMBL protein with an accessory protease-like domain. This association with one or more additional modules, often with predicted peptide-hydrolyzing or –binding activity, suggests that additional bacteriotoxic proteins may be found among the diverse chimeric MMBL proteins encoded in prokaryotic genomes. A phylogenetic survey of the bacterial MMBL modules reveals a mosaic pattern of strongly diverged sequences, mainly occurring in soil-dwelling and rhizosphere bacteria, which may reflect a trans-kingdom acquisition of the ancestral genes.

MMBL-type lectins: what’s in a name?

Mannose-binding B-lectins have been purified from numerous monocot plants. The hallmark of these so-called MMBLs (monocot mannose-binding lectins) is the presence of a domain with three potential carbohydrate-binding pockets, each generated by a QxDxNxVxY motif. Due to variable degeneracy of this signature sequence, some of these binding sites may not be active [1-3]. Most plant MMBLs are built from two to four identical or homologous protomers, though some are monomeric [4,5] or have a tandem domain structure [6]. Crystal structures of several of them, some with complexed mannose/mannoside oligomers, have been determined, revealing a common β-prism fold (Figure 1) [3,5-12]. Typically these plant lectins bind mannose only weakly and display a somewhat higher affinity for oligomannosides or high-mannose N-glycans [12,13].

It has been proposed that these lectins serve a defensive role, providing protection against plant predators or phytopathogens [4]. Indeed, some members of this family possess antifungal activity (e.g. gastrodianin from the orchid Gastrodia elata) and several others display insect-killing capacity (e.g. ASAL from Allium sativum) or are nematicidal (e.g. RVL from Remusatia vivipara). These protective potentials have been demonstrated in transgenic crop plants [14,15]. Notably, by converting the homodimeric insecticidal ASAL into a monomeric form, antifungal properties were acquired [16]. Additional interest in the plant MMBLs stems from their therapeutic potential for suppressing enveloped viruses [17] and triggering apoptosis in cancer cells [18]. Though lacking any detectable lectin activity, neoculin, a heterodimeric MMBL protein from Molineria (Curculigo) latifolia fruit, possesses sweet-tasting and taste-modifying properties by interacting with the human taste receptor T1R2-T1R3 [19,20].

Eukaryotic MMBLs revisited

The ubiquitous occurrence in monocots, lending the original family name, contrasts with a rare distribution across other plants, as inferred from isolated reports on MMBL-like proteins found in the liverwort Marchantia polymorpha [21], the dicot Hernandia moerenhoutiana [22], and the gymnosperm Taxus media [23]. Plant-like MMBLs were also identified in the fresh-water sponge Lubomirskia baicalensis [24], in the ascomycetous fungus Fusarium verticillioides and basidiomycete Marasmius oreades [25,26], and in the slime mold Dictyostelium discoideum (comitin; [27]). A comitin-deficient amoeba mutant appeared more susceptible to infections by the intracellular bacterial pathogen Legionella [28]. A protective function has also been proposed for some of the MMBL lectins that were identified in fishes. Pufflectin-s from skin mucus of Takifugu rubripes, a homodimeric lectin containing one functional mannose-binding site [29,30], was found to bind the parasitic trematode Heterobothrium okamotoi, suggesting that it contributes to the parasite-defense system in fugu [29]. More recently, the homotetrameric lectin plumieribetin was isolated from skin mucus and fin stings of Scorpaena plumieri. An integrin-inhibiting effect was demonstrated and thought to contribute to some of the local and systemic effects of envenomation by scorpionfish [31]. Another MMBL family member was also identified in skin mucus of Atlantic cod (Gadus morhua) by proteomic analysis [32].

The MMBL domain is also found in several types of multi-domain proteins from both monocot and dicot plants, in particular S-locus glycoproteins and S-locus receptor kinases, involved in self-incompatibility [22]. To reflect the wider distribution beyond monocots, an alternative family name, GNA, referring to the first described characterized member (Galanthus nivalis agglutinin) has been proposed [22].

Prokaryotic MMBLs: in search of a function

While not yet detected in Archaea, genes encoding MMBL-like (hypothetical) proteins have been identified in several bacterial genomes. Different domain architectures can be distinguished in these prokaryotic proteins: some are built from a single or tandem MMBL domain only, whereas others carry carboxy- or aminoterminally fused polypeptides with one or more additional domains. Figure 2 depicts the phylogenetic diversity of MMBL modules extracted from bacterial proteins with representative architectures, in comparison with those present in eukaryotic proteins from the major lineages. Separate clusters are apparent for the plant, fish and fungal lectins. In contrast to these well-delineated eukaryotic clades, the bacterial MMBL domains are found on several separate small or bigger branches indicating increased sequence divergence. A highly patchy taxonomic distribution with extensive sequence divergence, even among strains from the same species, is apparent. However, some genera of Proteobacteria (Burkholderia, Pseudomonas) and actinomycetes, with larger-than-average genome sizes, seem to be relatively enriched in MMBL-containing proteins, which may reflect acquisition of MMBL-encoding genes by horizontal gene transfer.

Bacterial killer MMBLs

LlpA (‘lectin-like putidacin A’) from banana rhizosphere isolate Pseudomonas putida BW11M1 was the first bacterial MMBL protein to be characterized. Built from an MMBL tandem, it was shown to function as a bacteriocin with a genus-specific target spectrum, inhibiting growth of several phytopathogenic P. syringae strains. LlpA does not require a cleavable signal sequence for secretion nor an immunity protein, the latter characteristic being often observed for proteins exhibiting similar bacteriotoxic activities [33]. Subsequently, narrow-spectrum bacteriocin activity was also assigned to two proteins with a similar tandem MMBL architecture in biocontrol strain P. fluorescens Pf-5, called LlpA1 and LlpA2, with near identical amino acid sequences and indistinguishable target strain spectrum [34]. More recently, antibacterial activity of two tandem MMBL bacteriocins from phytopathogenic Pseudomonas syringae and Xanthomonas citri has been demonstrated. In the latter case, activity against several xanthomonads was observed, but not against Pseudomonas and vice versa, confirming lectin-like bacteriocins to represent genus-specific killer proteins but acting across species borders [35]. We are currently investigating whether this concept can be extended beyond g-Proteobacteria, using the equivalent proteins encoded by Burkholderia cenocepacia and Burkholderia ambifaria (β-Proteobacteria) as test cases.

The phylogenetic analysis of individual MMBL modules in these bacteriocins reveals a clustering of N-terminal domains disparate from branches with C-terminal domains (Figure 2). Such independent evolution of MMBL domains within these tandems probably points towards a yet unresolved dedicated function contributing to their antibacterial activity. A notable exception to this within-tandem module divergence is found for a predicted protein from the actinomycete Arthrobacter sp. FB24, currently the sole representative of this type from a Gram-positive bacterium, for which biological activity also remains to be assessed.

In this context it will also be of interest to functionally characterize those ‘minimal’ bacterial MMBL family members containing a single MMBL module without apparent additional domain. Genes encoding such mono-MMBLs are common in fishes, fungi and some monocot plants, but the bacterial representatives of bacilli (Firmicutes) and pseudomonads are assigned to two well-resolved branches in the phylotree, separate from the eukaryotic sequences (Figure 1). In the case of mono-MMBLs from Pseudomonas, they show obvious sequence relationship with the N-terminal domains from the tandem MMBLs encoded by other strains of the genus.

Bacterial chimeric MMBLs: toxic proteins as well?

The MMBL module has been integrated, individually or as a tandem of closely related domains, in several bacterial multi-domain proteins representing various domain topologies (Figure 2). However, so far only one such hybrid MMBL protein has been functionally characterized. Albusin B, secreted by the Firmicutes member Ruminococcus albus 7, consists of a single aminoterminal MMBL module fused to a putative peptidase domain and inhibits growth of another ruminal bacterium, Ruminococcus flavefaciens [36]. The contribution of individual domains to this bacteriocin-like activity has not been further investigated.

From inspection of the various hybrid architectures, the association with domains potentially conferring hydrolytic activities emerges as a recurrent theme (Figure 2). Sequence-based clustering of several of the corresponding fused MMBL domains indicates evolutionary relatedness, despite their occurrence in taxonomically unrelated prokaryotic genera: associated domains include subtilase (subtilisin-like serine protease) in Burkholderia and Stigmatella (β- and δ-Proteobacteria, respectively), unspecified hydrolase (GDSL-like lipase/acylhydrolase) in Granulicella and Terriglobus (Acidobacteria), trypsin-like protease and NLPC/P60-type cysteine peptidase in Streptomyces and Cellulomonas, respectively (both Actinobacteria). In the Burkholderia ambifaria and Stigmatella aurantiaca polypeptides, the presence of a propeptide domain preceding the actual protease domain suggests involvement of proteolytic processing for biological activity [37,38]. The trypsin-related MMBL proteins are equipped with an additional carboxyterminal module potentially involved in adhesion (β-propeller domain VCBS, [39]) or sugar-binding (ricin-type β-trefoil lectin).

In several nocardioform actinomycetes (Mycobacterium, Nocardia, Rhodococcus, Segniliparus, Tsukamurella), a fusion with a carboxyterminal peptidoglycan-binding LysM domain [40] is prominent. The MMBL module of the corresponding Mycobacterium smegmatis protein, devoid of the C-terminal domain, has been crystallized, awaiting further functional characterization [41]. It has been shown that the mammalian peptidoglycan recognition proteins (PGRPs) upon binding to the bacterial cell wall can trigger lethal activation of stress-responsive two-component systems [42].

In a Mucilaginibacter paludis strain (Bacteroidetes), two hybrid MMBL proteins are found: one module is joined to a papain family cysteine protease domain, while a quite similar module occurs in combination with a MACPF domain. Among prokaryotes, the MACPF domain is particularly abundant among Chlamydiae and Bacteroidetes. In the latter group, MACPF also occurs in combination with the carbohydrate-binding module BACON [43]. Originally, the MACPF designation refers to its occurrence in mammalian components of the complement cascade (MAC), targeting Gram-negative bacteria, and in perforin (PF), killing virus-infected cells [44]. These protective functions rely on the ability to form large membrane pores [45]. However, no such lytic activity on eukaryotic cells could be demonstrated for the MACPF protein of the insect pathogen Photorhabdus luminescens (g-Proteobacteria) that contains an additional β-prism domain [46]. No pathogenicity has been attributed to members of the Mucilaginibacter genus that was described recently [47] and now accommodates several new species isolated from soil and rhizosphere. Possibly, a MACPF-MMBL hybrid may have evolved to serve as an antagonistic factor in competition with other (micro)organisms residing in these environments.

Bacteriocins with a novel mode of action?

The identification of MMBL proteins with a role in warfare among closely related bacteria has assigned another type of defense-related function in addition to the antifungal, insecticidal, nematicidal and antiviral activities mainly associated with its plant members. The narrow target spectrum, with allelopathic activity confined within genus borders, is reminiscent of the killing range of a bacteriocin. Another lectin fold (C-type) has also been recruited to serve a bactericidal function and enables mammalian RegIII proteins (such as mouse RegIIIγ and human HIP/HAP) to bind to the surface-exposed peptidoglycan layer of Gram-positive bacteria [48] and restrict colonization of the small intestinal mucosal surface [49].

Antibacterial activity has not yet been reported for a prokaryotic mono-domain MMBL protein but has been described for tandem-MMBL proteins (demonstrated in the Gram-negative bacteria Pseudomonas and Xanthomonas), as well as for a hybrid architecture with an adjacent protease domain (as found in the Gram-positive bacterium Ruminococcus). In the uncharacterized prokaryotic hetero-domain MMBL proteins, different types of peptide-hydrolytic modules are frequently found as an accessory domain. These are good candidates for novel MMBL members with antibacterial function. Characterization of such new bactericidal members will assist in elucidation of the contribution of individual domains to the killing process. Conceivably, separate modules may be involved in target recognition and binding, and in subsequent lethal action. How carbohydrate-binding to a mannose-containing ligand or N-glycan would be involved in this process is currently unclear. The export route followed by some the (candidate) bacteriocins is also not known. Some are equipped with a cleavable N-terminal signal peptide for Type II secretion, whereas others lack an identifiable export motif. These features seem not to be linked with phylogenetic affiliation.

The MMBL family is also intriguing from an evolutionary viewpoint. These proteins are particularly abundant in monocot plants but display a mosaic distribution among other organisms, including bacteria. As the prokaryotic family members are predominantly found in soil-dwelling and plant-associated bacteria, it may be hypothesized that these proteins evolved from genes originally acquired from plants. Compared to the eukaryotic proteins, the bacterial MMBL domains have diverged more extensively, even within some tandemly organized members, and they cannot be readily traced back to a specific origin.