Possibility of Bacterial Recruitment of Plant Genes Associated with the Biosynthesis of Secondary Metabolites

Microorganisms and plants synthesize an immense variety of metabolites, which are generally classified into two major groups based on their function. Primary metabolites are essential for growth and universally used, whereas secondary metabolism is highly diverse and variable and plays a role for the survival of the producing organism within its natural habitat (Demain, 1989; Harborne, 1993; Hartmann, 1996; Croteau et al., 2000; Kutchan, 2001). The study of the biochemistry of the formation of all these compounds is the study of the molecular basis of life (Stryer, 1995), which explains why there is so much ongoing interest in the field. Today, the chemical mechanisms of many central processes are well understood (Michal, 1999), and it is clear that common molecular patterns and principles underlie the diverse forms of life (e.g. the use of the same building blocks to construct macromolecules and the flow of genetic information from DNA to RNA; Stryer, 1995). Nevertheless, distinct differences in the basic biosynthetic pathways exist. For example, plants and microorganisms can synthesize their amino acids from simple building blocks, whereas most animals lost this ability in the course of their adaptation to heterotrophic life. Thus, many biosynthetic pathways are regarded as specific for certain groups of organisms. The ability to form thousands of structurally diverse natural products due to secondary metabolism is considered a typical feature of plants and microbes. These compounds are usually classified into major groups depending on the basic building blocks of the final structure, e.g. the terpenes that are formed from isoprenoid moieties, or the polyketides, which are assembled from short chain carboxylic acids. These products not only play an important role for their producers in the natural habitat, but also for human health (as opposed to their name indicating “of secondary interest”), because almost 50% of the most important medications are derived from these so-called natural products (Demain, 1999). These encompass not only antibacterials, antivirals, and antitumor compounds, but also products with immunosuppressant, antihypertensive, antidiabetic, antimalarial, and antihypercholesterolemic properties (Strohl, 1997; Grabley and Thiericke, 1999). Although approximately 170,000 secondary metabolites are known according to the Chapman & Hall dictionary of natural products (see http://www.chemnetbase.com/scripts/dnpweb.exe), there is a clear trend as to which group of organisms produces what type of compounds. For example, alkaloids (e.g. morphine) and phenylpropanoids are considered typical plant metabolites, whereas nonribosomally biosynthesized peptides (e.g. cyclosporin) are regarded as microorganism specific. Thus, a variety of structures, biosynthetic pathways, proteins, and genes are believed to be only found in plants, raising a question of the origin of biosynthetic pathways. Did plants “invent” all of these pathways without using and adapting genetic information from prokaryotes? In contrast to this generally accepted hypothesis, this update will show that a variety of metabolic pathways that were regarded to be present only in plants have in the recent years also been found in prokaryotes. Microorganisms with a complex life cycle and large genomes (e.g. myxobacteria [Shimkets, 1993; Pradella et al., 2002] and actinomycetes [Bentley et al., 2002]), which can frequently be isolated as soil inhabitants, seem to be particularly rich in such pathways, indicating that some “typical plant pathways” may actually originate from microorganisms. Alternatively, biosynthetic pathway genes might have been taken up by the microbes from plant material, which might be a consequence of their saprophytic lifestyle. One example is the myxobacterium Sorangium cellulosum, which actively degrades cellulose and, therefore, can be isolated from rotting plant material (Reichenbach and Dworkin, 1992). Such a close organismic association could explain the uptake of plant DNA eventually resulting in horizontal gene transfer. Here, we provide information about recent findings, mainly from actinomycetes and myxobacteria, both of which are very potent sources of secondary metabolites among the microorganisms. Because of the immense emerging body of knowledge from genome sequencing, this update is restricted to selected examples of microbial taxa and does not aim at reviewing an all-inclusive list. PAL is a ubiquitous enzyme in higher plants; it catalyzes the deamination of l-Phe to trans-cinnamic acid (Fig. 1). Without PAL, our world would be less colorful due to the lack of flavonoids and anthocyanins. Moreover, plants would most likely not exist in their current form due to the lack of lignin. PAL is the first enzyme in the conversion of l-Phe to benzoic acid (Fig. 1), an important structural element in a variety of natural products from plants (e.g. cocaine [Leete, 1990], paclitaxel [Walker and Croteau, 2000], and fungi, e.g. zaragozic acid [Bergstrom et al., 1995]). Biosynthesis of several primary and secondary metabolites from plants and/or bacteria. Broken arrows indicate multistep reactions. ADIC, 2-amino-2-desoxy-isochorismate; DAHP, 3-desoxyarabinoheptulosonate-7-phosphate; DAHPS, DAHP synthase; DHBA, 2,3-dihydroxybenzoic acid; E4P, erythrose-4-phosphate; PAL, phenylalanium ammonia; PEP, phosphoenolpyruvate; TAL, tyrosine ammonia. In prokaryotes, only very few cinnamic and benzoic acid derived metabolites have been described, presumably due to the rarity of PAL in these organisms. However, there is evidence for two PAL independent pathways to benzoate: the anaerobic degradation of l-Phe (Schneider et al., 1997; Breese et al., 1998) and, similar to plants, a benzoate biosynthesis directly from shikimate (Grond et al., 2000). PAL is highly homologous to His ammonia lyase (HAL), the enzyme catalyzing an analogous deamination of His to trans-urocanate in pro- and eukaryotes and, therefore, initiates His degradation in almost all organisms (Michal, 1999). It is also similar to Tyr ammonia lyase (TAL), a common enzyme of various monocotylic plants that is rarely found in prokaryotes (Kyndt et al., 2002). All three enzymes have the unique prosthetic group 4-methylidene imidazol-5-one (Schwede et al., 1999; Rother et al., 2001), indicating a very similar catalytic mechanism. However, because PAL is the first enzyme in plant phenylpropanoid secondary metabolism, it is the best studied member of this enzyme family (Schuster and Retey, 1995). The presence of a PAL activity in prokaryotes was first described in 1970 for Streptomyces verticillatus, which produces cinnamide (Emes and Vining, 1970). Indirectly, PAL activity has been shown by the incorporation of l-Phe and benzoate into the myxobacterial metabolite soraphen (Reichenbach and Höfle, 1994; Hill et al., 1998) and into the wailupemycins and enterocin in “Streptomyces maritimus” (Hertweck and Moore, 2000; Fig. 2). Additional prokaryotic secondary metabolites have been described with a benzoic acid moiety that is used as the polyketide synthase (PKS) starting unit. Two examples from myxobacteria are the crocacins (Jansen et al., 1999) from S. cellulosum and the phenalamids from Myxococcus stipitatus (Trowitzsch-Kienast et al., 1992), which closely resemble the myxalamids (Gerth et al., 1983), but use benzoate instead of short branched-chain carboxylic acids as starter molecules (Fig. 2). Sequencing of the myxalamid biosynthetic gene cluster revealed the presence of a HAL-like gene adjacent to the biosynthetic gene cluster (Silakowski et al., 2001) that might be inactive in the myxalamid producer Stigmatella aurantiaca Sg a15 because no phenalamide is produced. However, assuming a highly similar gene cluster for phenalamide production in M. stipitatus due to the similar chemical structures of both compounds, one would postulate the HAL-like gene to be active in this strain. Structures of selected secondary metabolites from myxobacteria and streptomycetes. The only case in which prokaryotic PAL activity has been clearly correlated to a biosynthetic protein is EncP from S. maritimus, which is needed for enterocin biosynthesis (Xiang and Moore, 2002). Although this enzyme shows higher similarities in size and sequence to prokaryotic HALs than to plant PALs (which is also the case for the above mentioned gene in S. aurantiaca), Phe was unambiguously proven as the substrate of EncP by heterologous expression of encP and production of cinammic acid in Streptomyces coelicolor. Nevertheless, the enzyme awaits a detailed characterization for a comparison of PALs from eukaryotic and prokaryotic sources. The absence of steroids in prokaryotes has been close to a dogma for many years. Until recently, only two bacterial species were known to have the triterpenes typical for eukaryotes: 4,4-dimethyl, 4-methyl, and 4-desmethylcholesterols in the methylotrophic bacterium Methylococcus capsulatus (Bird et al., 1971) and 8(9)-cholesten-3β-ol in the myxobacterium Nannocystis excedens (Kohl et al., 1983). Instead of steroids, bacteria often employ squalene-derived pentacyclic terpenes of the hopanoid type to build their membranes (Kannenberg and Poralla, 1999). Very recently, a detailed analysis of many myxobacterial strains from different genera revealed the presence of several steroids, which were previously only known from eukaryotes (Bode et al., 2003; Fig. 3). Steroid and hopanoid biosynthesis in eukaryotes and prokaryotes. Cholesterol could not be isolated from myxobacteria. Strains of the myxobacterium N. excedens are potent prokaryotic steroid producers that contain more than 2% steroids in their cellular dry weight and 10 distinct steroids. Almost all intermediates and side products of the biosynthesis of the major steroid in mammals, 5-cholesten-3β-ol (cholesterol; see Fig. 3), except for the final product itself, could be isolated from different N. excedens strains. In addition, the well known plant or phytosterol cycloartenol could be isolated from different species of Stigmatella and Cystobacter. The latter strains produce both lanosterol and cycloartenol, which are the first cyclization products of (S)-2,3-oxidosqualene, giving rise to the huge variety of zoo-/mycosterols and phytosterol, respectively (Fig. 3; Michal, 1999). Phylogenetic analysis of the cycloartenol synthase, whose corresponding gene was described as the first prokaryotic steroid biosynthesis gene from the myxobacterium S. aurantiaca Sg a15, revealed a higher similarity of up to 59% to eukaryotic oxidosqualene cyclases (e.g. cycloartenol synthase, lanosterol synthase, or lupeol synthase) than to prokaryotic squalene/hopene cyclases (42% similarity; Bode et al., 2003). However, almost no similarity could be detected on the DNA level. Studies with different types of steroid biosynthesis inhibitors underline the differences between the prokaryotic and the eukaryotic enzymes. No inhibition was observed in vivo in myxobacteria by terbinafin, tolnaftat, or AMO1618, well-known inhibitors of the eukaryotic squalene epoxidase. No inhibition of hydroxymethyl glutaryl-CoA reductase with fluvastatin was observed (Bode et al., 2003), indicating structural differences between eukaryotic and bacterial enzymes in mevalonate biosynthesis. Only miconazol, an inhibitor of the eukaryotic P450-dependent steroid C14-demethylase, showed the mode of action known from eukaryotes. Although Nannocystis strains produce quantities of steroids equal to that in eukaryotes, their function in myxobacteria remains a mystery. No difference in reproduction time, ethanol tolerance, fatty acid composition, or swarming could be observed upon comparing a wild-type strain of S. aurantiaca with a cycloartenol knockout (Bode et al., 2003). This contrasts with the knowledge of hopanoids in bacteria, which are assumed to serve as steroid surrogates. Nevertheless, the physiological function of hopanoids is only poorly understood (Kannenberg and Poralla, 1999), although it is known that they are formed during transition from substrate to aerial hyphae in S. coelicolor A3(2) (Poralla et al., 2000). The strong hydrophobicity of both classes of triterpenoids suggests they are located in the membranes and might be involved in controlling membrane fluidity and rigidity (Ourisson and Nakatani, 1994). Even though myxobacteria contain a sufficiently wide variety of different fatty acids to fine tune their membrane fluidity (Ware and Dworkin, 1973; Mahmud et al., 2002), steroids may also play a role. In addition, steroids could themselves serve as signaling compounds (e.g. like hormones) or as part of signaling molecules, which has been described for the hedgehog family of secreted proteins that are characterized by an essential cholesterol residue (Porter et al., 1996). Additional strains have to be investigated to support this hypothesis. The CHS and stilbene synthase (STS) superfamily of PKSs appears to be ubiquitous in higher plants. STSs, e.g. resveratrol synthase, produce the stilbene backbone as a key reaction in the biosynthesis of stilbene type phytoalexins (Fig. 1). CHS is a key enzyme in the biosynthesis of flavonoids, which exhibit a wide range of biochemical, physiological, and ecological activities. Resveratrol and CHSs condense 4-coumaroyl-CoA with three molecules of malonyl-CoA, which results in products differing in the newly formed ring systems (resveratrol and naringenin chalcone; see Figs. 1 and 4). The same type of condensing reaction is utilized by the 2-ketoacyl-ACP synthases of fatty acid biosynthesis. However, the available data show that these enzymes share little overall homology with either resveratrol synthase or CHS (Schröder, 1999). Over the last couple of years, a series of new additions to the CHS/STS superfamily has been reported from plants, which differ from the enzymes described above by utilizing non-phenylpropanoid starter units, varying numbers of condensation reactions, and different cyclization patterns (e.g. acridone synthase [Junghanns et al., 1995] and 2-pyrone synthases [Eckermann et al., 1998; compare with Fig. 5]). Biosynthesis of chalcones and stilbenes indicating their different cyclization mechanisms. Biosynthetic building blocks (coumaryl-CoA and malonyl-CoA) are shown in bold. A to C show differently folded forms of the same tetraketide. R-S, Enzyme-bound thioester. Biosynthesis of flaviolin, (S)-3,5-dihydroxyphenyl-Gly (DPG), and 1,3-dihydroxy-N-methyl-acridone. R-S, Enzyme-bound thioester. RppA and DpgA are CHSs involved in the formation of flaviolin and DPG, respectively. Unexpectedly, in 1999, “a new type of polyketide biosynthetic pathway” was reported in bacteria. This pathway is used for the assembly of flaviolin (Fig. 5), a small aromatic metabolite in streptomycetes (Funa et al., 1999). At the same time, a similar pathway was shown to be present in pseudomonads. The PKS involved was called “type III PKS” (Bangera and Thomashow, 1999) to distinguish it from the better studied types I and II PKS from microorganisms. The latter enzymes are responsible for the biosyntheses of a large family of structurally diverse natural products with broad ranges of biological activities, reminiscent of fatty acid biosynthesis (Staunton and Weissman, 2001). Phylogenetic analyses of more bacterial type III PKS from whole-genome sequencing (e.g. S. coelicolor, Bacillus subtilis, and Mycobacterium tuberculosis) have made it clear that these enzymes belong to the CHS/STS superfamily of PKSs previously only found in plants (Schröder, 1999; Moore and Höpke, 2001). Some unusual mechanisms of PKS biochemistry have been reported for bacterial type III PKS, e.g. for the enzyme involved in the formation of the non-proteinogenic amino acid DPG (Fig. 5), which is essential for glycopeptide biosynthesis in different actinomycetes (Li et al., 2001; Pfeifer et al., 2001). Thus far, none of the few characterized bacterial type III PKSs employs the typical plant starter molecule, 4-coumaroyl-CoA (see Figs. 4 and 5), suggesting a distinction between bacterial and plant type III PKS enzymes. This suggestion might in fact be wrong because several bacterial gene products might employ 4-coumaroyl-CoA as a starter molecule. For example, the purple photosynthetic bacterium Rhodospirillum centenum has a photoactive yellow protein the that is part of a bacterial with similarity to plant et al., 1999; 2002). the expression of a type III PKS in the genome of which suggests that this PKS may be involved in the formation of In this 4-coumaroyl-CoA is as the starter because centenum this the characterization of a (see Fig. 1 and the about has been reported from organism a In this the genes and are located adjacent to indicating that the protein is used for the production of the 4-coumaroyl-CoA of (Kyndt et al., 2002). Two more examples of bacterial type III PKS with close homology to CHS from plants are found in myxobacteria. In the genome of Myxococcus (see and S. cellulosum (see one and two type III PKS genes can be respectively. The M. protein sequence is most similar to a few bacterial proteins of function to one of the is the centenum protein mentioned above However, some plant naringenin synthases show similarities to the M. protein The first type III PKS of S. cellulosum to a typical bacterial protein to with RppA from different whereas the gene product is most similar to few bacterial proteins and plant resveratrol and to S. cellulosum is to Phe into the (see Fig. 1), which is used as a of polyketide biosynthesis in some species (Reichenbach and Höfle, 1999). This that the type III PKS could employ 4-coumaroyl-CoA or as a starter molecule. Thus, the latter examples of bacterial type III PKS genes from two myxobacterial species and from centenum can be to CHS enzymes. an enzyme that was considered to be plant and can also be found in myxobacteria. It is closely to plant enzymes and does not group with bacterial amino acid belong to the large group of which contain a of the are for the formation of with important physiological properties in animals and 1995). In plants, the catalyzing and from primary into secondary and serve as for the formation of a variety of alkaloids such as alkaloids (see Fig. e.g. and and 1993; et al., 2000). Phylogenetic analysis of different types of et al., the enzymes into I is by group II is by and group III is by prokaryotic forms of and and the prokaryotic type of and group is by eukaryotic forms of and and the prokaryotic biosynthetic type of and II prokaryotic and eukaryotic of different function only and been reported from prokaryotes the study was which the to that the between these enzymes the of their catalytic Thus, one would a of prokaryotic origin to show the similarities to group II of prokaryotic differences in substrate et al., 2000). A in the of the myxobacterium S. cellulosum was found and characterized expression in Although the function of the corresponding protein in the bacterium is it could be shown that the more properties with the of plant than with of origin et al., 2000). enzymes exhibit a for but also the aromatic amino acids and, to a et al., In plant enzymes usually show a more distinct substrate et al., of the of S. cellulosum only to In contrast to the mentioned in the last amino acid sequence analyses revealed of from S. cellulosum with plant and but not with bacterial aromatic amino acid Phylogenetic analysis of the amino acid suggests a to the plant of as which that in S. cellulosum is the of a horizontal gene et al., 2000). with myxobacteria and data show that are not found in bacteria. with the of and bacterial genomes at the for from S. cellulosum shows and only to two which are in the genomes of the plant and the to their these two microorganisms could have their corresponding genes in horizontal gene processes as All within the bacterial genomes were to show the presence of what was regarded as biosyntheses and pathways in prokaryotes, in actinomycetes and myxobacteria. The question as to which of organisms these and pathways be because we are only to these common to prokaryotes and eukaryotes. However, there is a that at some of the genes that were regarded as typical for plants might originate from prokaryotes. is the function of these pathways in prokaryotes, and why were they only The to the part of the question is the of information only available from various sequencing of prokaryotic organisms. The first part of the question is more to the of typical plant biosynthetic pathways in myxobacteria and actinomycetes might be due to their saprophytic life and close to plants in their soil as mentioned in the However, several bacteria have been (e.g. that not show the observed of myxobacteria and actinomycetes have the genomes of all bacteria known so with for S. cellulosum et al., and for S. coelicolor et al., 2002). These genomes might have from from various sources during their However, and genes in some for the organism because a large genome more to are there not more genera of prokaryotes with large the is that we only a small of the microorganisms present in the soil This is due to our to and most of et al., 2002). The of myxobacteria and streptomycetes might be an because they can be regarded as two of the most complex prokaryotes. build 1993; Fig. and streptomycetes form aerial 1999) and their in the formation of for survival of the myxobacteria S. cellulosum by S. aurantiaca and by These are to study metabolism in plants and microbes because we are a into the processes and In the we will be to our on genome and which will of the molecular basis of complex processes and the function of genes in microorganisms. would like to and for of the and and for