Genetic improvement of processes yielding microbial products
Editor: Alexander Boronin
Jose L. Adrio, Department of Biotechnology, Puleva Biotech, S.A., Camino de Purchil, 66, 18004 Granada, Spain. Tel.:+34 958 24 02 27; fax:+34 958 24 01 60; e-mail: firstname.lastname@example.org
Although microorganisms are extremely good in presenting us with an amazing array of valuable products, they usually produce them only in amounts that they need for their own benefit; thus, they tend not to overproduce their metabolites. In strain improvement programs, a strain producing a high titer is usually the desired goal. Genetics has had a long history of contributing to the production of microbial products. The tremendous increases in fermentation productivity and the resulting decreases in costs have come about mainly by mutagenesis and screening/selection for higher producing microbial strains and the application of recombinant DNA technology.
Microorganisms can generate new genetic characters (‘genotypes’) by two means: mutation and genetic recombination. In mutation, a gene is modified either unintentially (‘spontaneous mutation’) or intentially (‘induced mutation’). Although the change is usually detrimental and eliminated by selection, some mutations are beneficial to the microorganism. Even if it is not beneficial to the organism, but beneficial to humans, the mutation can be detected by screening and can be preserved indefinitely. This is indeed what the fermentation microbiologists did in the strain development programs that led to the great expansion of the fermentation industry in the second half of the twentieth century.
It was fortunate that the intensive development of microbial genetics began in the 1940s when the fermentative production of penicillin became an international necessity. The early studies in basic genetics concentrated on the production of mutants and their properties. The ease with which ‘permanent’ characteristics of microorganisms could be changed by mutation and the simplicity of the mutation techniques had tremendous appeal to microbiologists. Thus began the cooperative ‘strain-selection’ program among workers at the U.S. Department of Agriculture Laboratories in Peoria, the Carnegie Institution, Stanford University and the University of Wisconsin, followed by the extensive individual programs that still exist today in industrial laboratories throughout the world. It is clear that mutation has been the major factor involved in the hundred- to thousand-fold increases obtained in production of microbial metabolites and that the ability to modify genetically a microbial culture to higher productivity has been the most important factor in keeping the fermentation industry in its viable, healthy state.
Applications of mutation
Mutation has improved the productivity of industrial cultures (Vinci & Byng, 1999; Parekh et al., 2000). It has also been used to shift the proportion of metabolites produced in a fermentation broth to a more favorable distribution, elucidate the pathways of secondary metabolism, yield new compounds, and for other functions. The most common method used to obtain high yielding mutants is to treat a population with a mutagenic agent until a certain ‘desired’ kill is obtained, plate out the survivors and test each resulting colony or a randomly selected group of colonies for product formation in flasks. The most useful mutagens include nitrosoguanidine (NTG), 4-nitroquinolone-1-oxide, methylmethane sulfonate (MMS), ethylmethane sulfonate (EMS), hydroxylamine (HA) and ultraviolet light (UV). The optimum level of kill for increased production of antibiotics is thought to be in the range 70–95% (Simpson & Caten, 1979), although some industrial programs use much higher levels, e.g. up to 99.99%. It is incorrect to condemn a mutation and screening procedure because, on average, it decreases production ability; indeed, this is the case for successful mutagenesis. One should only be interested in the frequency of improved mutants.
Although single cells or spores are preferred for mutagenesis, non-spore-forming filamentous organisms have been mutated successfully by mutagenizing mycelia, preparing protoplasts and regenerating on solid medium (Keller, 1983). Sonication is sometimes used to break up Streptomyces mycelia after mutagenesis and before screening for improved mutants (Takebe et al., 1989). Poorly sporulating filamentous organisms can be mutagenized after fragmentation or formation of protoplasts (Kim et al., 1983; Kurzatkowski et al., 1986).
More detailed information can be found in several authorative reviews on genetics and especially on mutation in actinomycetes (Baltz, 1986, 1995, 1998, 1999; Hopwood, 1999).
Mutants producing increased quantites of metabolites
Genetics has had a long history of contributing to the production of microbial products. The tremendous increases in fermentation productivity and the resulting decreases in costs have come about mainly by mutagenesis and screening for higher producing microbial strains. At least five different classes of genes control metabolite production (Malik, 1979): (i) structural genes coding for product synthases, (ii) regulatory genes determining the onset and expression of structural genes, (iii) resistance genes determining the resistance of the producer to its own antibiotic, (iv) permeability genes regulating entry, exclusion and excretion of the product, and (v) regulatory genes controlling pathways providing precursors and cofactors. Overproduction of microbial metabolites is effected by (i) increasing precursor pools, (ii) adding, modifying or deleting regulatory genes, (iii) altering promoter, terminator and/or regulatory sequences, (iv) increasing copy number of genes encoding enzymes catalyzing bottleneck reactions, and (v) removing competing unnecessary pathways (Strohl, 2001).
It is now over 60 years since the first superior penicillin-producing mutant, Penicillium chrysogenum X-1612, was isolated afer X-ray mutagenesis. This heralded the beginning of a long and successful relationship between mutational genetics and industrial microbiology (Hersbach et al., 1984). The improvement of penicillin production by conventional strain improvement resulted both from enhanced gene expression and from gene amplification (Barredo et al., 1989; Smith et al., 1989). Increased levels of mRNA corresponding to the three enzymes of penicillin G biosynthesis were found in high-penicillin producing strains of P. chrysogenum as compared to wild-type strains (Smith et al., 1990). High-producing strains contained an amplified region; a 106-kb region amplified five to six times as tandem repeats was detected in a high-producing strain, whereas wild-type P. chrysogenum and Fleming's original strain of P. notatum contained only a single copy (Fierro et al., 1995).
Strain improvement has been the main factor involved in the achievement of impressive titers of industrial metabolites. The production titer of tetracycline as far back as 1979 was reported to be over 20 g L−1 (Podojil et al., 1984), mainly due to strain improvement. Later, titers of 30–35 g L−1 were reached for chlortetracycline and tetracycline. The production titer of penicillin is 70 g L−1 and that of cephalosporin C over 30 g L−1 (Elander, 2003). The production titer of tylosin has been reported to be over 15 g L−1 (Chen et al., 2004) and that of salinomycin is 60 g L−1 (Liu, 1982).
Morphological and pigment mutants
Although almost nothing is known about the mechanisms causing higher production in superior random or morphological mutants, it is likely that many of these mutations involve regulatory genes, especially as regulatory mutants obtained in basic genetic studies are sometimes found to be altered in colonial morphology. Thus, morphological mutants have been very important in strain improvement. These include mutants affected in mycelia formation, which produce colonies with a modified appearance or a new color. Color changes have also been important for pigment producers (Table 1).
Very early in the development of the concepts of regulation, geneticists realized that the end product of a biosynthetic pathway to a primary metabolite excercises strict control over the amount of an intermediate accumulated by an auxotrophic mutant of that pathway. Only at a growth-limiting concentration of the end product would a large accumulation of the substrate of the deficient enzyme occur. This principle of decreasing the concentration of an inhibitory or repressive end product to bypass feedback inhibition or repression was best accomplished by the use of auxotrophic mutants. Indeed, auxotrophic mutation has been a major factor in the industrial production of primary products such as amino acids and nucleotides (Table 1). The production of secondary products such as antibiotics is also markedly affected by auxotrophic mutation, even when auxotrophs are grown in nutritionally complete and even complex media. Although the change in product formation is usually in the negative direction, higher-producing auxotrophs are obtained from producers of antibiotics.
When several primary metabolites are produced by a single branched pathway, mutation in one branch of the pathway often leads to overproduction of the product of the other branch. Examples include the overproduction of phenylalanine by tyrosine auxotrophs and vice versa, and the overproduction of lysine by auxotrophs requiring threonine and methionine. In the case of branched pathways leading to a primary metabolite and a secondary metabolite, auxotrophic mutants requiring the primary metabolite sometimes overproduce the secondary metabolite (Table 1).
Reversion of an auxotroph to prototrophy sometimes leads to new prototrophs possessing higher enzyme activity than present in the original ‘grandparent’ prototroph. Such increased enzyme activity was probably the result of a structural gene mutation producing a more active enzyme or an enzyme less subject to feedback inhibition (Table 1).
Revertants of non-producing mutants
A high proportion of such mutants has been found to produce increased amounts of antibiotics (Table 1).
Basic studies on regulation have shown that it is possible to select regulatory mutants, which overproduce the end products of primary pathways, using toxic metabolite analogues. Such antimetabolite-resistant mutants often possess enzymes that are insensitive to feedback inhibition, or enzyme-forming systems resistant to feedback repression. The selection of mutants resistant to toxic analogues of primary metabolites has been widely employed by industrial microbiologists (Table 1).
A variation of the antimetabolite selection techniques is possible when a precursor is toxic to the producing organism. The principle here is that the mutant most capable of detoxifying the precursor by incorporating it into the antibiotic will be the best grower in the presence of the precursor (Table 1). When the secondary metabolite produced is itself a growth inhibitor of the producing culture, as in the case of certain antibiotics, the metabolite can sometimes be used to select resistant mutants that are improved producers.
Certain streptomycin resistance mutations cause increased production of unrelated antibiotics. In addition to improvement in secondary metabolite formation by mutation to streptomycin resistance, resistance to gentamicin or to paromomycin is even more effective. Furthermore, triple mutation to resistance to streptomycin, gentamicin and rifampicin, each of which individually increased actinorhodin formation, was found to be the most effective (Table 1).
Mutants resistant to nutritional repression
Nutritional repression can also be decreased by mutation to antimetabolite resistance. Examples of selection agents are 2-deoxyglucose (2-DOG) for enzymes and pathways controlled by carbon source regulation (Table 1), methylammonium for those regulated by nitrogen source repression, and arsenate for phosphate regulation.
Mutants that use phosphate less efficiently for growth sometimes show improved antibiotic production. Thus, screening for small colonies on phosphate-limiting media could be a useful strain improvement technique for phosphate-regulated products (Table 1).
Improved production on agar
In many cases, fermentation performance on an agar plate is related to production in submerged liquid culture, and the method has application as a means of detecting superior mutants. So-called ‘zone mutants’ have proven useful for improving several different processes (Table 1).
A widely used modification involves the production of antibiotics by colonies on separate plugs of agar followed by placement of these plugs on a seeded assay plate and measurement of the resultant clear zones. The use of this ‘agar piece method’ resulted in improvement of antibiotic production (Table 1). Agar-piece screening of antibiotic production in the presence of inhibitory levels of phosphate (15 mM) led to the isolation of six markedly improved and stable Streptomyces hygroscopicus strains producing the macrolide antifungal complex ‘165’ (Gesheva, 1994).
Product excretion in overproducing strains often occurs when uptake and/or catabolism is impaired. Thus, genetic lesions eliminating active uptake can be used to specifically enhance excretion of metabolites (Table 1). It is often of benefit to isolate mutants unable to grow on the desired product as sole carbon or energy source. Such mutants are often impaired in their ability to takeup the product and they contain lower intracellular levels of the product, thus lessening feedback regulation. In certain improved mutants, there is an increase in sensitivity to deoxycholate and lysozyme, indicating a change in permeability.
Mutants showing qualitative changes in the mix of fermentation products
As many organisms produce secondary metabolites as mixtures of a chemical family or of several chemical families, mutation has been used to eliminate undesirable products in such fermentations. An example is that of Nakatsukasa and Mabe (Nakatsukasa & Mabe, 1978), who found that streaking out a natural single colony isolate from Streptomyces aureofaciens (producing the polyether narasin and the broad-spectrum antibiotic enteromycin) on galactose led to yellow and white sectoring. The effect was specific for galactose. Of the four colony types obtained, one produced only narasin and two produced only enteromycin.
Streptomyces griseus ssp. cryophilus makes four R3− sulfated and four R3− unsulfated carbapenems. The sulfated forms are less active than the unsulfated forms. To completely eliminate the R3 sulfated forms, sulfate transport mutants were obtained. These were of two types: (i) auxotrophs for thiosulfate or cysteine; and (ii) selenate-resistant mutants. Each type produced completely unsulfated forms and titers were equivalent to the total titer of the parent (Kitano et al., 1985).
Eight avermectins are produced by Streptomyces avermitilis, of which only a small number are desirable. A non-methylating mutant produced only four of the compounds and a mutant that failed to make the 25-isopropyl substituent (from valine) produced a different mixture of components. By protoplast fusion, a hybrid strain was obtained which made only two components, B2a and B1a (Omura et al., 1991). Stutzman-Engwall and colleagues (Stutzman-Engwall et al., 2003) introduced random mutations by PCR into gene aveC and obtained a mutant that produced an avermectin B1 : B2 ratio of 2.5, much improved over the 0.6 ratio of the parent S. avermitilis strain.
Mutation was used to eliminate the undesirable polyketides sulochrin and asterric acid from broths of the lovastatin producer, Aspergillus terreus (Vinci et al., 1991). Mutants have also been employed to eliminate undesirable coproducts from the monensin fermentation (Pospisil et al., 1984).
Mutants producing new antibiotics
Mutant methodology has been used to produce new molecules. The medically useful products demethyltetracycline and doxorubicin were discovered by simple mutation of the cultures producing tetracycline and daunorubicin, respectively. Later, the technique of ‘mutational biosynthesis’ (=mutasynthesis) was devised (Shier et al., 1969). In this process, a mutant blocked in secondary metabolism is fed analogs of the moiety whose biosynthesis is blocked. If successful, the mutant (called an ‘idiotroph’) produces a new antibiotic derivative (Nagaoka & Demain, 1975). The hybramycins were the first compounds to be made this way (Shier et al., 1969). Since then, mutational biosynthesis has been used for the discovery of many new secondary metabolites (Lemke & Demain, 1976; Daum & Lemke, 1979; Kitamura et al., 1982). The most well-known is the commercial antihelmintic agent doramectin, the production of which employed a mutant of the avermectin producer S. avermitilis (Cropp et al., 2000).
New anthracyclines and aglycones have been isolated from blocked mutants of the daunorubicin and doxorubicin producers (Cassinelli et al., 1980; McGuire et al., 1981). By adding carminomycinone or 13-dihydrocarminomycinone to an idiotroph of Streptomyces galilaeus (the producer of aclacinomycin), the aglycones were glycosylated to form a new trisaccharide anthracycline, trisarubicionol (Yoshimoto et al., 1981).
New macrolide antibiotics have been produced from blocked mutants of the tylosin-producer, Streptomyces fradiae (Kirst et al., 1983). Four new hybrid macrolide antibiotics were obtained by feeding erythronolide B to a blocked mutant of the oleandomycin producer, Streptomyces antibioticus (Spagnoli et al., 1983). A blocked-mutant of the mycinamicin producer, Micromonospora polytrota, was fed various rosaramicin precursors and converted them into new rosaramicins (Lee et al., 1983).
Use of mutants to elucidate biosynthetic pathways
A further use of mutants has been the elucidation of metabolic pathways. This has been exploited for the biosynthesis of tetracyclines (McCormick, 1965), novobiocin (Kominek, 1972), erythromycin (Martin et al., 1966; Martin & Rosenbrook, 1967), neomycin (Pearce et al., 1978), tylosin (Baltz et al., 1983), other aminoglycosides (Penzikova & Levitov M, 1970; Takeda et al., 1978; Fujiwara et al., 1980; Kase et al., 1982; Hanssen & Kirby, 1983), rosaramicin (Vaughn et al., 1982), daunorubicin (McGuire et al., 1981), other anthracyclines (Motamedi et al., 1986; Yue et al., 1986), actinomycin (Troost & Katz, 1979), carbapenems (Nozaki et al., 1984; Kojima et al., 1988), ansamycins (Kibby et al., 1980; Ghisalba et al., 1981), patulin (Gaucher et al., 1981) and phenazines (Byng et al., 1979).
In contrast to the extensive use of mutation in industry, genetic recombination was not much used at first, despite early claims of success (Jarai, 1961; Mindlin, 1969), mainly due to the absence or the extremely low frequency of genetic recombination in industrial microorganisms (in streptomycetes, it was usually 10−6 or even less). Other problems were evident with the β-lactam-producing fungi. Although Aspergillus exhibited sexual and parasexual reproduction, the most commercially interesting genera, Cephalosporium and Penicillium, were the most difficult to work with as they only reproduced parasexually, which rarely resulted in recombination.
Recombination was erroneously looked upon as an alternative to mutation instead of a method that would complement mutagenesis programs. The most balanced and efficient strain development strategy would not emphasize one to the exclusion of the other; it would contain both mutagenesis-screening and recombination-screening components. In such a program, strains at different stages of a mutational line, or from lines developed from different ancestors, would be recombined. Such strains would no doubt differ in many genes and by crossing them, genotypes could be generated which would never occur as strictly mutational descendants of either parent. Recombination was also of importance in the mapping of production genes. Studies on the genetic maps of overproducing organisms such as actinomycetes are relatively recent. The model for such investigations was the genetic map of Streptomyces coelicolor (Kieser et al., 1992), which was found to be very similar to those of other Streptomyces species, such as S. bikiniensis, S. olivaceous, S. glaucescens and S. rimosus.
As mentioned above, genetic recombination was virtually ignored in industry, mainly due to the low frequency of recombination. However, use of protoplast fusion changed the situation markedly. After 1980, there was a heightened interest in the application of genetic recombination to the production of important microbial products such as antibiotics. Today, frequencies of recombination have increased to even greater than 10−1 in some cases (Ryu et al., 1983), and strain improvement programs routinely include protoplast fusion between different mutant lines. The power of recombination was demonstrated by Yoneda (Yoneda, 1980), who recombined individual mutations, each of which increased α-amylase production by two- to seven-fold in Bacillus subtilis. A strain constructed by genetic transformation, which contained all five mutations, produced 250-fold more α-amylase.
Recombination is especially useful when combined with conventional mutation programs to solve the problem of ‘sickly’ organisms produced as a result of accumulated genetic damage over a series of mutagenized generations. For example, a cross via protoplast fusion was carried out with strains of Cephalosporium acremonium from a commercial strain improvement program. A low-titer, rapidly-growing, spore-forming strain which required methionine to optimally produce cephalosporin C was crossed with a high-titer, slow-growing, asporogenous strain which could use the less expensive inorganic sulfate. The progeny included a recombinant which grew rapidly, sporulated, produced cephalosporin C from sulfate and made 40% more antibiotic than the high-titer parent (Hamlyn & Ball, 1979).
Protoplast fusion was used to modify the characteristics of an improved penicillin-producing strain of P. chrysogenum which showed poor sporulation and poor seed growth. Backcrossing with a low-producing (12 g L−1) strain yielded a high-producing (18 g L−1) strain with better sporulation and better growth in seed medium (Lein, 1986). Interspecific protoplast fusion between the osmotolerant Saccharomyces mellis and the highly fermentative S. cerevisiae yielded stable hybrids fermenting high concentrations of glucose (49% w w−1) (Legmann & Margalith, 1983).
Another application of protoplast fusion is the recombination of improved producers from a single mutagenesis treatment. By recombination, one could combine the yield-increase mutations and obtain an even more superior producer before carrying out further mutagenesis. Two improved cephamycin-C producing strains from Nocardia were fused and among the recombinants were two cultures that produced 10–15% more antibiotic than the best parent (Wesseling & Lago, 1981). Genetic recombination allows the discovery of new antibiotics by fusing producers of different or even the same antibiotics. A recombinant obtained from two different rifamycin-producing strains of Nocardia mediterranei produced two new rifamycins (16,17-dihydrorifamycin S and 16,17-dihydro-17-hydroxy-rifamycin S) (Traxler et al., 1982). However, according to Hopwood (Hopwood, 1983), these examples may reflect the different expression of genes from parent A in the cytoplasm of parent B, rather than the formation of hybrid antibiotics. Interspecific protoplast fusion between S. griseus and five other species (Streptomyces cyaneus, Streptomyces exfoliatus, Streptomyces griseoruber, Streptomyces purpureus and Streptomyces rochei) yielded recombinants of which 60% produced no antibiotics and 24% produced antibiotics different from the parent strains (Okanishi et al., 1996). New antibiotics can also be created by changing the order of the genes of an individual pathway in its native host (Hershberger, 1996).
A new antibiotic, indolizomycin, was produced by protoplast fusion between non-antibiotic producing mutants of Streptomyces griseus and Streptomyces tenjimariensis (Gomi et al., 1984). Osmotolerance of food yeasts such as Saccharomyces cerevisiae and S. diastaticus was increased by protoplast fusion with osmotolerant yeasts. Other traits transferred between yeasts by protoplast fusion include flocculation (Panchal et al., 1982), lactose utilization (Farahnak et al., 1986), the killer character (Bortol et al., 1986; Farris et al., 1992), cellobiose fermentation (Pina et al., 1986) and methionine overproduction (Brigidi et al., 1988).
Plasmids, transposons, cosmids and phage
Plasmid DNA has been detected in virtually all antibiotic-producing species of Streptomyces. Some are sex plasmids and constitute an essential part of the sexual recombination process and others contain either structural genes or genes somehow influencing the expression of the chromosomal structural genes of antibiotic biosynthesis.
Most plasmids have no apparent effect on metabolite production and only very few antibiotic biosynthesis processes are encoded by plasmid-borne genes. However, the production of methylenomycin A is encoded by genes present on plasmid SCP1 in Streptomyces coelicolor. When the plasmid was transferred to other streptomycetes, the recipients produced the antibiotic. For many years, plasmid SCP1 was never observed or isolated as a circular DNA molecule, because it was a giant linear plasmid. It was initially difficult to separate such giant linear plasmids from chromosomal DNA but this was later accomplished by pulsed field gel electrophoresis or orthogonal field alteration gel electrophoresis (OFAGE) (Kinashi & Shimaji, 1987).
Some products of unicellular bacteria are plasmid-encoded. These include aerobactin, a hydroxamate siderophore and virulence factor produced by Escherichia coli (McDougall & Neilands, 1984) and other Gram-negative bacteria (Enterobacter aerogenes, Enterobacter cloacae, Vibrio mimicus, and species of Klebsiella, Salmonella and Shigella). Aerobactin is synthesized by a plasmid-borne five-gene cluster, which is negatively regulated by iron (Roberts et al., 1986); it can also be produced via chromosomal genes (Moon et al., 2004). It also appears that siderophore production by Arizona hinshawii is plasmid-encoded. A microcin, an antimetabolite of methionine, which is produced by E. coli and acts as a competitive inhibitor of homoserine-O-transuccinylase, is encoded by a plasmid that occurs at 20 copies per genome equivalent (Perez-Diaz & Clowes, 1980). The gene coding for the parasporal crystal body (δ-endotoxin) of Bacillus thuringiensis is plasmid-borne (Whiteley & Schnepf, 1986; De Maagd et al., 2003) in most species but is on the chromosome in a few other species.
Instability in Streptomyces is brought about by environmentally stimulated macrolesions, e.g. deletions, transpositions, rearrangements and DNA amplification. They occur spontaneously or are induced by environmental stresses such as intercalating dyes, protoplast formation and regeneration, and interspecific protoplast fusion. Streptomycetes are the only prokaryotes known to be subject to spontaneous DNA amplification, sometimes amounting to several hundred tandem copies, accounting for over 10% of total DNA, in the absence of selection. Amplification seems to be coupled to DNA deletion and may involve insertion sequence (IS)-like elements (Baltz, 1986). Ethidium bromide cures plasmids in streptomycetes but also increases the frequency of deletion mutations, especially in areas of the chromosome that are already unstable (Crameri et al., 1986).
Transposable elements, DNA sequences encoding a transposase enzyme (Berg & Berg, 1983) that move from one replicon to another without host recombination functions or extensive homology with the site of integration, have been extremely useful for the following reasons: (i) they usually provide stable, nonreverting mutants; (ii) they can be used to determine the order of genes in an operon; (iii) it is easy to select for mutants because transposons contain antibiotic- or mercury-resistance markers; (iv) they provide portable regions of homology for chromosomal mobilization; (v) they provide markers for non-selectable genes and allow the cloning of such genes which can then be used as hybridization probes to fish out the wild-type gene from a genomic library; and (vi) they often have unique restriction sites, and thus are good markers for isolating defined deletion derivatives or locating the precise position of a gene by heteroduplex mapping.
In the daptomycin producer Streptomyces roseosporus, some Tn 5099 transposition mutants produced 57–66% more daptamycin than the parent, whereas others produced less or the same (McHenney & Baltz, 1996; Baltz et al., 1997). Transposition increased the rate-limiting step of tylosin biosynthesis in Streptomyces fradiae, i.e. the conversion of macrocin to tylosin. Transposing a second copy of tylF into a neutral site on the S. fradiae chromosome increased its gene product, macrocin O-methyltransferase, and tylosin production, while decreasing the concentration of the final intermediate (macrocin). Tylosin production was increased by up to 60% and the total macrolide titer was unchanged (Solenberg et al., 1996). Transposon mutagenesis eliminated the production of the toxic oligomycin by the avermectin-producing Streptomyces avermitilis (Ikeda et al., 1993).
Cloning a 34-kb fragment from Streptomyces rimosus via a cosmid into Streptomyces lividans and Streptomyces albus resulted in oxytetracycline production by the recipients (Binnie et al., 1989). Contrary to earlier reports, all the oxytetracycline genes were clustered together on the S. rimosus chromosomal map (Butler et al., 1989).
Improvement of microbial processes by genetic engineering
New processes for the production of amino acids and vitamins have been developed by recombinant DNA technology. Escherichia coli strains were constructed with plasmids bearing amino acid biosynthetic operons. Plasmid transformation was accomplished in Corynebacterium, Brevibacterium and Serratia and, as a result, recombinant DNA technology has been used routinely to improve such commercial amino acid-producing strains (Sahm et al., 2000).
A recombinant strain of E. coli (made by mutating to isoleucine auxotrophy, cloning in extra copies of the thrABC operon, inactivating the threonine-degrading gene tdh, and mutating to resistance to high concentrations of l-threonine and l-homoserine) produced 80 g L−1l-threonine in 1.5 days at a yield of 50% (Eggeling & Sahm, 1999). Cloning extra copies of threonine export genes into E. coli led to increased threonine production (Kruse et al., 2002).
The introduction of the proline 4-hydroxylase gene from Dactylosporangium sp. into a recombinant strain of E. coli producing l-proline at 1.2 g L−1 lead to a new strain producing 25 g L−1 of hydroxyproline (trans-4-hydroxy-l-proline) (Shibasaki et al., 2000). When proline was added, hydroxyproline reached 41 g L−1, with a yield of 87% from proline.
An engineered strain of Corynebacteriumglutamicum producing 50 g L−1 of l-tryptophan was further modified by cloning in additional copies of its own transketolase gene to increase the level of the erythrose-4-phosphate precursor of aromatic biosynthesis (Ikeda & Katsumata, 1999). A low copy number plasmid increased production to 58 g L−1, whereas a high copy number plasmid decreased production.
l-Valine production by mutant strain VAL1 of C. glutamicum amounted to 105 g L−1 (Radmacher et al., 2002; Lange et al., 2003). The mutant was constructed by overexpressing biosynthetic enzymes via a plasmid, eliminating ilvA encoding threonine dehydratase, and deleting two genes encoding enzymes of pantothenate biosynthesis. The culture was grown with limitation of isoleucine and pantothenate.
By introduction of feedback-resistant threonine dehydratases and additional copies of genes encoding branched amino and biosynthetic enzymes, lysine- or threonine-producing strains were converted into L-isoleucine producers with titers up to 10 g L−1 (Morbach et al., 1996; Guillouet et al., 1999; Hashiguchi et al., 1999). Amplification of the wild-type threonine dehydratase gene ilvA in a threonine-producing strain of Corynebacterium lactofermentum led to 15 g L−1 of isoleucine overproduction (Colon et al., 1995).
Biotin has been made traditionally by chemical synthesis but recombinant microbes have approached a competitive economic position. The cloning of a biotin operon (bioABFCD) on a multicopy plasmid allowed E. coli to produce 10 000 times more biotin than did the wild-type strain (Levy-Schil et al., 1993). Sequential mutation of Serratia marcescens to resistance to the biotin antimetabolite acidomycin (=actithiazic acid) led to mutant strain SB412, which produced 20 mg L−1 biotin (Sakurai et al., 1994). Further improvements were made by mutating selected strains to ethionine-resistance (strain ET2, 25 mg L−1), then mutating ET2 to S-2-aminoethylcysteine resistance (strain ETA23, 33 mg L−1) and finally cloning in the resistant bio operon (Sakurai et al., 1994) yielding a strain able to produce 500 mg L−1 in fed-batch fermentor culture along with 600 mg L−1 of biotin vitamers. Later advances led to production by recombinant S. marcescens of 600 mg L−1 of biotin (Masuda et al., 1995).
A process for riboflavin production in Corynebacterium ammoniagenes (previously Brevibacterium ammoniagenes) was developed by cloning and overexpressing the organism's own riboflavin biosynthesis genes (Koizumi et al., 2000) and its own promoter sequences. The resulting culture produced 15.3 g L−1 riboflavin in 3 days. Genetic engineering of a Bacillus subtilis strain already containing purine analog-resistance mutations led to the improved production of riboflavin (Perkins & Pero, 1993). An industrial strain of B. subtilis was produced by inserting multiple copies of the rib operon at two different sites in the chromosome, making purine analog-resistance mutations to increase guanosine triphosphate (GTP; a precursor) production and a riboflavin analog (roseflavin)-resistance mutation in ribC that deregulated the entire pathway (Perkins et al., 1999).
Vitamin C (ascorbic acid) has traditionally been made in a five-step predominantlychemical process by first converting glucose to 2-keto-l-gulonic acid (2-KGA) with a yield of 50% and then converting the 2-KGA by acid or base to ascorbic acid. A novel process for vitamin C synthesis involved the use of a genetically engineered Erwinia herbicola strain containing a gene from Corynebacterium sp. The engineered organism converted glucose into 1 g L−1 of 2-KGA (Anderson et al., 1985; Pramik, 1986). A better process was devised independently, which converted 40 g L−1 glucose into 20 g L−1 2-KGA (Grindley et al., 1988). This process involved cloning and expressing the gene encoding 2,5-diketo-d-gluconate reductase from Corynebacterium sp. into Erwinia citreus. Another process uses a recombinant strain of Gluconobacter oxydans containing genes encoding l-sorbose dehydrogenase and l-sorbosone dehydrogenase from G. oxydans T-100. The new strain was an improved producer of 2-KGA (Saito et al., 1997). Further mutation to suppress the l-idonate pathway and to improve the promoter led to the production of 130 g L−1 of 2-KGA from 150 g L−1 sorbitol.
Carotenoids were overproduced by introducing carotenoid gene clusters from Erwinia uredovora into E. coli and overexpressing E. coli deoxyxylulose phosphate synthase, the key enzyme of the non-mevalonate isoprenoid biosynthetic pathway (Matthews & Wurtzel, 2000). Lycopene accumulated to 1.3 mg g−1 dry cell weight and zeaxanthin to 0.6 mg g−1.
Cloning of aldehyde dehydrogenase of Acetobacter polyoxogenes on a plasmid vector into Acetobacter aceti ssp. xylinum increased the rate of acetic acid production by over 100% (1.8 g L−1 h−1 to 4 g L−1 h−1) and titer by 40% (68 g L−1 to 97 g L−1) (Fukaya et al., 1989).
Genetic engineering of the inosine monophosphate (IMP) dehydrogenase gene in a B. subtilis strain producing 7 g L−1 of the desirable guanosine and 19 g L−1 of the undesirable inosine changed production to 20 g L−1 guanosine and 5 g L−1 inosine (Miyagawa et al., 1986).
A recombinant E. coli strain was constructed that produced optically active pure d-lactic acid from glucose at virtually the theoretical maximum yield, e.g. two molecules from one molecule of glucose (Zhou et al., 2003). The organism was engineered by eliminating genes of competing pathways encoding fumarate reductase, alcohol/aldehyde dehydrogenase and pyruvate formate lyase and by a mutation in the acetate kinase gene.
New technologies that have proven to be very useful for increasing production of primary metabolites include genome-based strain reconstruction, metabolic engineering, and whole genome shuffling (see section on Novel genetic technologies).
The application of recombinant DNA technology to the production of secondary metabolites has been of great interest (Baltz & Hosted, 1996; Diez et al., 1997). The tools of the recombinant geneticist for increasing the titers of secondary metabolites have included: (i) transposition mutagenesis, (ii) targeted deletions and duplications by genetic engineering and (iii) genetic recombination by protoplast fusion (Baltz, 2003). Recent additions to these techniques include genomics, transcriptome analysis, proteomics, metabolic engineering, and whole genome shuffling (see section on Novel gene technologies).
One of the first indications that rDNA technology could be applied to antibiotics and other secondary metabolites was that it could be carried out in streptomycetes (Thompson et al., 1982). Plasmids were constructed from plasmid SLP 1.2 of Streptomyces lividans and plasmid SCP2* from Streptomyces coelicolor. In mating of plasmid-negative S. lividans, ‘pocks’ (circular zones of sporulation inhibition associated with plasmid transfer in the lawn of streptomycete growth arising from a regenerated protoplast population) were seen. This was due to looping out of a piece of S. coelicolor DNA, which became a series of small S. lividans plasmids (SLP 1.1 to 1.6) that were good cloning vehicles.
The genetic engineering of actinomycetes was limited for a number of years by restriction barriers hindering DNA introduction and by the inhibition of secondary metabolism by self-replicating plasmid-cloning vectors (Baltz & Hosted, 1996), but these problems were mainly overcome. Early reviews on cloning and expressing antibiotic production genes in Streptomyces were by Martin and Gil (Martin & Gil, 1984) and Liras (Liras, 1988).
An interesting possibility was the transfer of operons from one streptomycete to another in the hope that the structural genes might be better able to express themselves in another species. Clustering facilitated the transfer of an entire pathway in a single manipulation. Studies revealed that many antibiotic biosynthesis genes were arranged in clusters including undecylprodigiosin, actinorhodin, chloramphenicol, rifamycin, cephamycin, erythromycin, tetracyclines and tylosin among others. Thus, the entire undecylprodigiosin pathway (‘red’ pathway) of S. coelicolor was transferred on a 37-kb fragment into Streptomyces parvulus and the antibiotic was produced (Coco et al., 1991). Similarly, the entire cephamycin C pathway was cloned and expressed from a cephamycin-producing strain of Streptomyces cattleya. When the 29-kb DNA fragment was cloned into the non-β-lactam producer, S. lividans, one transformant (out of 30 000) made cephamycin (Chen et al., 1988). When the fragment was introduced into another cephamycin producer, Streptomyces lactamgens, a two- to three-fold improvement was obtained.
In fungi making penicillin G, the three structural genes (ACVS, cyclase and penicillin acyltransferase) are clustered on a single chromosome of Penicillium chrysogenum (Smith et al., 1990) and of Aspergillus nidulans (MacCabe et al., 1990). In these fungi, the genes of the cluster are separately transcribed. By contrast, fungal genes coding for cephalosporin biosynthesis are distributed among different chromosomes. The deacetylcephalosporin C acetyltransferase gene from Cephalosporium acremonium (cefG) is closely linked to the expandase (cefEF) gene (Gutierrez et al., 1992; Matsuda et al., 1992) and both are on chromosome II, whereas the early genes of the pathway (pcbAB, pcbC) are located on chromosome VI.
Cloning has been very important in understanding the biosynthesis of β-lactam antibiotics (Demain & Elander, 1999), its genetics and improving the production processes.
Early common pathway
All producers of penicillins and cephalosporins, including cephamycins, use the same two enzymes to start the biosynthetic process. The steps involve the condensation of l-α-aminoadipic acid, l-cysteine and l-valine to form the tripeptide, δ-(α-aminoadipyl)-l-cysteinyl-d-valine (ACV) by ACV synthetase (ACVS), encoded by gene pcbAB (also known as acvA in A. nidulans). This is followed by cyclization of ACV into isopenicillin N (IPN) by IPN synthase (cyclase; encoded by pcbB). The cloning of the gene encoding ACVS from P. chrysogenum (Diez et al., 1990), C. acremonium (Gutierrez et al., 1991) and Nocardia lactamdurans (Castro et al., 1988) contributed greatly to the elucidation of the biosynthetic pathway. Overexpression of acvA in A. nidulans, by replacing the normal promoter with the ethanol dehydrogenase promoter (Kennedy & Turner, 1996), increased penicillin production up to 30-fold. The cyclase genes from different microorganisms were all cloned (Aharonowitz et al., 1992; Martin et al., 1997) and provided pure enzyme for structural studies. Cloning multiple copies of cyclase into C. acremonium yielded an improved cephalosporin C-producing strain (Skatrud et al., 1987).
The hydrophobic branch
Producers of penicillin use a single step branch involving penicillin acyltransferase acting on IPN. Its gene penDE (also known as iat, aat and acyA in A. nidulans) was cloned from P. chrysogenum into C. acremonium, which led to the production of penicillin G (in the presence of exogenous phenylacetic acid) along with cephalosporin C (Gutierrez et al., 1991). Without cloning, C. acremonium cannot produce penicillin G.
The hydrophilic branch
All producers of cephalosporins and cephamycins employ a series of enzymes leading from IPN. First, IPN is epimerized to penicillin N by IPN epimerase (encoded by cefD). The next steps include ring expansion of penicillin N by deacetoxycephalosporin C (DAOC) synthase (expandase, encoded by cefE) and hydroxylation by DAOC 3′-hydroxylase (encoded by cefF) to deacetylcephalosporin C (DAC). Although expandase and hydroxylase are separate enzymes encoded by separate genes in bacteria, these two activities are found on the same protein in fungi, which is encoded by one gene cefEF. At the DAC stage, the overall pathway again splits into two branches. In C. acremonium, DAC is acetylated to cephalosporin C by DAC acetyltransferase encoded by cefG. This step is the terminal reaction in cephalosporin-producing fungi. By contrast, actinomycetes carbamoylate DAC using three enzymes, encoded by cmcH, cmcI and cmcJ genes to yield cephamycin C (Brewer et al., 1980).
When an industrial production strain of C. acremonium 394-4 was transformed with a plasmid containing the pcbC and the cefEF gene from an early strain of the C. acremonium mutant line, a transformant producing 50% more cephalosporin C than the production strain, as well as less penicillin N, was obtained. Production in pilot plant (150 L) fermentors was further improved by 15% (Skatrud et al., 1989). One copy of the cefEF had been integrated into chromosome III, whereas the native gene is on chromosome II.
Transformation of P. chrysogenum with the Streptomyces lipmanii cefD and Streptomyces clavuligeruscefE genes allowed the production of the intermediate DAOC (Cantwell et al., 1992) at titers of 2.5 g L−1. DAOC is a valuable intermediate in the commercial production of semi-synthetic cephalosporins. Also, cloning of cefE from S. clavuligerus or cefEF and cefG (see next paragraph) from C. acremonium into P. chrysogenum grown with adipic acid as side-chain precursor (Crawford et al., 1995) resulted in formation of adipyl-6-aminopenicillanic acid (adipyl-6-APA) and adipyl-7-aminodeoxycephalosporanic acid (adipyl-7-ADCA) in the case of cefE and adipyl-6APA, adipyl-7ADCA, adipyl-7-DAC and adipyl-7-aminocephalosporanic acid (7-ACA) in the case of cefEF and cefG.
Disruption and one-step replacement of the cefEF gene of an industrial cephalosporin C production strain of A. chrysogenum yielded strains accumulating up to 20 g L−1 of penicillin N. Cloning and expression of the cefE gene from S. clavuligerus into those high-producing strains yielded recombinant strains producing high titers of DAOC (Velasco et al., 2000). Production levels were nearly equivalent (80%) to the total β-lactams biosynthesized by the parental strain.
Weak acetyltransferase promoter activity appears to be the cause of DAC accumulation in broths of C. acremonium. Cloning of cefG increased its copy number and cefG mRNA, tripled acetyltransferase activity, and increased cephalosporin C titers in a dose-dependent manner (Matsuda et al., 1992; Mathison et al., 1993). Cloning of the gene with its own promoter had no effect on the low level of DAC acetyltransferase normally observed in C. acremonium (Gutierrez et al., 1997). However, the use of foreign promoters (the gpd promoter from A. nidulans, the bla promoter from A. niger or the pbcC promoter from P. chrysogenum) had a major effect on the level of cefG transcripts, DAC acetyltransferase protein level and activity, and antibiotic production; cephalosporin C production rose by two- to three-fold. Of the cephalosporins produced, the undesirable DAC decreased from 80% of the total to 30–39%, whereas cephalosporin C increased by a similar amount.
Transformation of early strain P. chrysogenum Wis54-1255 with individual genes, pairs of genes, and all three genes of the penicillin pathway showed that the major increases occurred when all three genes were overexpressed (Theilgaard et al., 2001). The best transformant contained three extra copies of pcbAB, one extra copy of pcbC and two extra copies of penDE and produced 299% of control shake flask production and 276% of control productivity in continuous culture.
Genes encoding many microbial enzymes have been cloned and the enzymes expressed at levels hundreds of times higher than those naturally produced. Recombinant DNA technology has been used (Falch, 1991): (i) to produce in industrial organisms enzymes obtained from microbes that are difficult to grow or handle genetically; (ii) to increase enzyme productivity by use of multiple gene copies, strong promoters, and efficient signal sequences; (iii) to produce in a safe host useful enzymes obtained from a pathogenic or toxin-producing microorganism; and (iv) to improve the stability, activity or specificity of an enzyme by protein engineering. The industrial enzyme business adopted rDNA methods to increase production levels and to produce enzymes from industrially-unknown microorganisms in industrial organisms such as species of Aspergillus and Trichoderma, as well as Kluyveromyces lactis, S. cerevisiae, Yarrowia lipolytica and Bacillus licheniformis. Virtually all laundry detergents contain genetically-engineered enzymes and much cheese is made with genetically-engineered enzymes. Indeed, over 60% of the enzymes used in the detergent, food and starch processing industries are recombinant products (Cowan, 1996).
Scientists at Novo Nordisk isolated a very desirable lipase for use in detergents from a species of Humicola. For production purposes, the gene was cloned into Aspergillus oryzae, where it produced 1000-fold more enzyme (Carlsen, 1990) and is now a commercial product. Such lipases are used for laundry cleaning, interesterification of lipids, and esterification of glucosides producing glycolipids which have applications as biodegradable non-ionic surfactants for detergents, skin care products, contact lens cleaners and as food emulsifiers.
The α-amylase gene from Bacillus amyloliquefaciens was cloned using multicopy plasmid pUB110 in B. subtilis (Palva, 1982). Production was 2500-fold that in wild-type B. subtilis and five-fold that of the B. amyloliquefaciens donor. An exoglucanase from the cellulolytic Cellulomonas fimi was overproduced after cloning in E. coli to a level of over 20% of cell protein (O'Neill et al., 1986). The endo-β-glucanase components of the cellulase complexes from Thermomonospora and Clostridium thermocellum were cloned in E. coli as was the cellobiohydrolase I gene of Trichoderma reesei (Shoemaker et al., 1983; Teeri et al., 1983). Pichia pastoris, a methanol-utilizing yeast, was engineered to produce S. cerevisiae invertase and to excrete it into the medium at 100 mg L−1 (Van Brunt, 1986). Interestingly, in S. cerevisiae, the invertase is periplasmic. Self-cloning of the xylanase gene in S. lividans resulted in six-fold overproduction of the enzyme (Mondou et al., 1986).
Many enzymes are made by filamentous organisms, which are slow-growing and difficult to handle in fermentors. The transfer of these genes to rapidly-growing unicellular bacteria means that rapid growth and more reproducible production can be achieved. Other advantages are more rapid nutrient uptake due to a greater surface/volume ratio, better oxygen transfer, better mixing and thus more reliable control of pO2, pCO2 and pH, and a better organism for mutagenesis.
Aspartase production was increased by 30-fold by cloning in E. coli (Komatsubara et al., 1986). Captopril esterase of Pseudomonas putida, used in preparing the chiral captopril sidechain, was cloned in E. coli with a 38-fold increase in activity (Elander, 1995). A 1000-fold increase in phytase production was achieved in A. niger using recombinant technology (Van Hartinsveldt et al., 1993). Cloning of the benzylpenicillin acylase gene of E. coli on multicopy (50) plasmids resulted in a 45-fold increase as compared to uninduced wild-type production. Interestingly, the cloned enzyme is constitutive (Mayer et al., 1980). Cloning additional penicillin V amidase genes into wild-type Fusarium oxysporium increased enzyme titer by 130-fold (Komatsubara et al., 1986).
The properties of many enzymes have been altered by genetic means. ‘Brute force’ mutagenesis and random screening of microorganisms over the years led to changes in pH optimum, thermostability, feedback inhibition, carbon source inhibition, substrate specificity, Vmax, Km and Ki. This information was later exploited by the more rational techniques of protein engineering. Single changes in amino acid sequences have yielded similar types of changes in a large variety of enzymes. Today, it is no longer necessary to settle for the natural properties of an enzyme; these can be altered to suit the needs of the investigator or the process. For example, a protease from Bacillus stearothermophilus was increased in heat tolerance from 86°C to 100°C, being made resistant to boiling. The enzyme was developed by site-directed mutagenesis (Van den Burg et al., 1998). Only eight amino acids had to be modified. Temperature stability at 100°C was increased 340-fold and activity at lower temperature was not decreased. All eight mutations were far from the enzyme's active site. Washing powders have been improved in activity and low temperature operation by the application of recombinant DNA technology and site-directed mutagenesis to proteases and lipases (Falch, 1991; Wackett, 1997).
Polymers, fuels, foods and beverages
Microbially-produced xanthan gum is not only an acceptable food-thickener but is one of the most promising agents for enhanced oil recovery in the petroleum industry. Recombinant DNA manipulation of Xanthomonas campestris increased titers of xanthan by two-fold and increased pyruvate content by over 45% (Bigelas, 1989; Tseng et al., 1992). The yield was 0.6 g g−1 of sucrose utilized (Letisse et al., 2001). Ten to twenty thousand tons of xanthan are produced annually for use in the oil, pharmaceutical, cosmetic, paper, paint and textile industries (Becker et al., 1998).
Escherichia coli was converted into a good ethanol producer (4.3%, v v−1) using recombinant DNA technology (Ingram et al., 1987). Alcohol dehydrogenase II and pyruvate decarboxylase genes from Zymomonas mobilis were inserted in E. coli and became the dominant system for NAD regeneration. Ethanol represented over 95% of the fermentation products in the genetically-engineered strain. By cloning and expressing the same two genes into Klebsiella oxytoca, the recombinant was able to convert crystalline cellulose to ethanol in high yield when fungal cellulase was added (Doran & Ingram, 1993). The maximum theoretical yield was 81–86% and titers as high as 47 g L−1 of ethanol were produced from 100 g L−1 of cellulose.
Cloning of its ace (acetone) operon gene adc (encoding acetoacetate decarboxylase), ctfA and ctfB (two genes encoding coenzyme A transferase) on a plasmid containing the adc promoter into Clostridium acetobutylicum resulted in a 95% increase in production of acetone, a 37% increase in butanol, a 90% increase in ethanol, a 50% increase in solvent yield from glucose and a 22-fold lower production of undesirable acids (Mermelstein et al., 1993). The introduction of the acetone operon from C. acetobutylicum into E. coli led to high acetone production by the latter (Bermejo et al., 1998).
Beer wort contains barley β-glucans which reduce the filtrability of beer and lead to precipitates and haze in the final product. The gene coding for endoglucanase was transferred from T. reesei to brewer's yeast and the engineered yeast strain efficiently hydrolyzed the β-glucans (Penttiläet al., 1987). Similiar technology created starch-utilizing S. cerevisiae strains and wine yeast strains producing lower acidity and enhanced flavor. Brewing yeasts were modified using recombinant DNA technology so that they could produce A. niger amyloglucosidase and break down unfermentable dextrins for light beer production (Van Brunt, 1986; Hammond, 1988). The glucoamylase gene from Aspergillus awamori was cloned and expressed stably in polyploid distiller's yeast. A high level of glucoamylase was secreted. Almost all (95%) of the carbohydrates in the 25% starch substrate were utilized and high levels of ethanol were produced. The engineered strain outperformed S. diastaticus (Cole et al., 1988).
Brewing yeasts were engineered to produce acetolactate decarboxylase from Enterobacter aerogenes or A. aceti. This enzyme eliminated diacetyl and the requirement for the three- to five-week flavor maturation period which normally follows a one-week fermentation stage (Sone et al., 1988). The resulting beer suffered no loss of quality or flavor (Holzman, 1994).
Recombinant DNA techniques have been useful in developing new bioconversions and improving old ones. Using a plasmid containing tryptophan synthase plus induction with 3-indole acrylate, recombinant E. coli was able to produce 180 g L−1 of l-tryptophan from indole plus l-serine in 8 h (Yukawa et al., 1988). Whereas S. cerevisiae normally produces 2 g L−1 of malic acid from fumaric acid, a recombinant strain containing a cloned fumarase gene was able to produce 125 g L−1 with a yield of almost 90% (Neufeld et al., 1991).
An oxidative bioconversion of saturated and unsaturated linear aliphatic 12–22 carbon substrates to their terminal dicarboxylic acids was developed by gene disruption and gene amplification (Picataggio et al., 1992). Product concentrations reached 200 g L−1 and problematic side-reactions such as unsaturation, hydroxylation and chain-shortening did not occur.
3-0-Acetyl-4′′-0-isovaleryltylosin (AIV) is useful in veterinary medicine against tylosin-resistant Staphylococcus aureus. It is made by first producing tylosin with Streptomyces fradiae and then using Streptomyces thermotolerans (producer of carbomycin) to bioconvert tylosin into AIV. A new direct fermentation organism was constructed by transforming S. fradiae with S. thermotolerans plasmids containing acyl transferase genes (Arisawa et al., 1996).
Recombinant Candida pasteurianum can carry out the conversion of glycerol to 1,3-propanediol (Luers et al., 1997). A more economical process involving conversion of the less expensive glucose to 1,3-propanediol has been achieved with a recombinant E. coli strain (Nakamura & Whited, 2003). The project is a collaborative effort by Genencor International and DuPont (Potera, 1997). The recombinant strain contains two metabolic pathways, one for conversion of glucose to glycerol and the other for conversion of glycerol to 1,3-propanediol (Tong et al., 1991; Laffend et al., 1996). The 1,3-propanediol (also known as trimethylene glycol or 3G) is used as the building block to produce a new biodegradable polyester (3G+).
Novel genetic technologies
A new genomic technique called ‘genome-based strain reconstruction’ allows one to construct a strain superior to the production strain because it only contains mutations crucial to hyperproduction, but not other unknown mutations which accumulate by brute-force mutagenesis and screening (Ohnishi et al., 2002). This approach was used to improve the lysine production rate of Corynebacterium glutamicum by comparing high producing strain B-6 developed by Hirao and coworkers (Hirao et al., 1989) (production rate slightly less than 2 g L−1 h−1) and a wild-type strain. Comparison of 16 genes from strain B-6, encoding enzymes of the pathway from glucose to lysine, revealed mutations in five of the genes. Introduction of three of these mutations into the wild-type created a new strain which produced 80 g L−1 in 27 h, at a rate of 3 g L−1 h−1, the highest rate ever reported for a lysine fermentation.
‘Metabolic engineering’ is the directed improvement of product formation or cellular properties through the modification of specific biochemical reactions or introduction of new ones using recombinant DNA technology (Stephanopoulos, 1999; Nielsen, 2001). Its essence is the combination of analytical methods to quantify fluxes and the control of fluxes with molecular biological techniques to implement suggested genetic modifications. Flux is the focal point of metabolic engineering. Different means of analyzing flux are: (i) kinetic based models; (ii) control theories; (iii) tracer experiments; (iv) magnetization transfer; (v) metabolite balancing; (vi) enzyme analysis and (vii) genetic analysis (Eggeling et al., 1996). Metabolic control analysis revealed that the overall flux through a metabolic pathway depends on several steps, not just a single rate-limiting reaction (Kacser & Acerenza, 1993).
Metabolic engineering has been applied to antibiotic production (Khetan & Hu, 1999, 1999; Thykaer & Nielsen, 2003). The increases in metabolic flux were carried out by enhancing enzymatic activity, manipulating regulatory genes, enhancing antibiotic resistance and heterologous expression of novel genes. Table 2 summarizes several examples of progress on the production of those secondary metabolites.
The production of amino acids shows many examples of this approach. A useful review of metabolic engineering in C. glutamicum, especially in relation to l-lysine production, was published by Sahm and colleagues (Sahm et al., 2000). Metabolic flux studies of wild-type C. glutamicum and four improved lysine-producing mutants available from the ATCC showed that yield increased from 1.2% to 24.9% relative to the glucose flux. Other recent examples are on overproduction of aromatic amino acids and derivatives (Bongaerts et al., 2001), l-lysine (Wittmann & Heinzle, 2002) and glutamate (Kimura, 2003).
There are many other successful applications of metabolic engineering for products such as 1,3-propanediol (Nakamura & Whited, 2003), carotenoids (Rohlin et al., 2001; Visser et al., 2003; Wang & Keasling, 2003), organic acids (Kramer et al., 2003), ethanol (Nissen et al., 2000), vitamins (Zamboni et al., 2003; Sybesma et al., 2004) and complex polyketides in bacteria (Pfeifer et al., 2001; Khosla & Keasling, 2003).
During the last few years, an expanded view of the cell has been possible due to impressive advances in all the ‘omics’ techniques (genomics, proteomics, metabolomics, etc.) and high-throughput technologies for measuring different classes of key intracellular molecules. ‘Systems biology’ has recently emerged as a term to describe an approach that considers genome-scale and cell-wide measurements in elucidating processes and mechanisms (Stephanopoulos et al., 2004). Progress in strain development will depend, not only on all the technologies mentioned above, but also on the development of mathematical methods that facilitate the elucidation of mechanisms and identification of genetic targets for modification.
Integrating transcriptional and metabolite profiles from 21 strains of A. terreus producing different levels of lovastatin and another 19 strains with altered (+)-geodin levels led to an improvement in lovastatin production of over 50% (Askenazi et al., 2003). The approach, named ‘association analysis’, was used to reduce the complexity of profiling data sets to identify those genes in which expression was most tightly linked to metabolite production. Such an application is suitable to all industrially useful organisms for which genome data are limited.
A genome-wide transcript expression analysis called ‘massive parallel signature sequencing’ (Brenner et al., 2000) was used successfully to discover new targets for further improvement of riboflavin production by the fungus A. gossypii (Karos et al., 2004). The authors identified 53 genes of known function, some of which could clearly be related to riboflavin production. This approach also allowed the finding of sites within the genome with high transcriptional activity during riboflavin biosynthesis that are suitable integration loci for the target genes found.
Gene expression analysis of wild-type and improved production strains of Saccharopolyspora erythraea and S. fradiae using microarrays of S. coelicolor revealed that regulation of antibiotic biosynthetic enzymes as well as enzymes involved in precursor metabolism were altered in those mutated strains (Lum et al., 2004). The S. erythraea overproducer expressed the entire erythromycin gene cluster for several more days than the wild-type. It seemed that the eryA gene and protein expression differences observed for the overproducer could account for over 50% of the total erythromycin titer increase. The S. fradiae mutant expressed the tylosin biosynthetic genes in a similar way to the wild-type strain; however, two genes, aco (encoding acyl-CoA dehydrogenase) and icmA (encoding isobutyryl-CoA mutase), were expressed more highly than in the wild-type strain. The induction of these two genes could increase the flux of metabolites from fatty acids to tylosin precursors in the overproducer.
These recent technologies and mathematical approaches will all contribute to the generation and characterization of microorganisms able to synthesize large quantities of commercially important metabolites. The ongoing sequencing projects involving hundreds of genomes, the availability of sequences corresponding to model organisms, new DNA microarray and proteomics tools, as well as the new techniques for mutagenesis and recombination described above will accelerate strain improvement programs. The development and combined application of these technologies will help to develop what was already succinctly described several years ago as ‘Inverse netabolic engineering’ (Bailey et al., 1996), which means to identify, construct or calculate a desired phenotype, identify the molecular basis of that desirable property, and incorporate that phenotype into another strain or other species by genetic and environmental manipulations.
‘Directed evolution’ (=applied molecular evolution=directed molecular evolution) is a rapid and inexpensive way of finding variants of existing enzymes that work better than naturally occurring enzymes under specific conditions (Kuchner & Arnold, 1997; Skandalis et al., 1997; Arnold, 1998). The process involves evolutionary design methods using random mutagenesis, gene recombination and high-throughput screening (Arnold, 2001). Diversity is initially created by in vitro mutagenesis of the parent gene using repeated cycles of mutagenic polymerase chain reaction (error-prone PCR) (Leung et al., 1989), repeated oligonucleotide-directed mutagenesis (Reidhaar-Olson et al., 1991), mutator strains (Bornscheuer et al., 1998) or chemical agents (Taguchi et al., 1998). A key limitation of these strategies is that they introduce random ‘noise’ mutations into the gene at every cycle and hence improvements are limited to small steps. This strategy has been used successfully in different applications (Zhao et al., 2002).
‘Molecular breeding techniques’ (DNA shuffling, Molecular Breeding™) come closer to mimicking natural recombination by allowing in vitro homologous recombination (Ness et al., 2000). These techniques not only recombine DNA fragments but also introduce point mutations at a very low controlled rate (Stemmer, 1994; Zhao & Arnold, 1997). Unlike site-directed mutagenesis, this method of pooling and recombining parts of similar genes from different species or strains has yielded remarkable improvements in enzymes in a very short amount of time (Patten et al., 1997). A step forward in this technique was breeding a population with high genetic variability as a starting point to generate diversity (DNA Family Shuffling). This approach led to a 240- to 540-fold improvement in cephalosporinase activity when four cephalosporinase genes were shuffled as a starting point (Crameri et al., 1998). When each of these genes was shuffled independently, only eight-fold improvements were obtained. Innovations that expand the formats for generating diversity by recombination include formats similar to DNA shuffling and others with few or no requirements for parental gene homology (Kurtzman et al., 2001; Lutz et al., 2001).
Random redesign techniques are currently being used to generate enzymes with improved properties such as: activity and stability at different pH values and temperatures (Ness et al., 1999), increased or modified enantioselectivity (Jaeger & Reetz, 2000), altered substrate specificity (Suenaga et al., 2001), stability in organic solvents (Song & Rhee, 2001), novel substrate specificity and activity (Raillard et al., 2001), increased biological activity of protein pharmaceuticals and biological molecules (Patten et al., 1997; Kurtzman et al., 2001) as well as novel vaccines (Marshall, 2002; Locher et al., 2004). Proteins from directed evolution work were already on the market by 2000 (Tobin et al., 2000). These were green fluorescent protein of Clontech (Crameri et al., 1996) and Novo Nordisk's LipoPrime® lipase.
‘Whole genome shuffling (WGS)’ is a novel technique for strain improvement combining the advantage of multi-parental crossing allowed by DNA shuffling with the recombination of entire genomes. This method was applied successfully to improved tylosin production in S. fradiae (Zhang et al., 2002). Historically, 20 cycles of classical strain improvement at Eli Lilly and Co. carried out over 20 years employing about one million assays improved production six-fold. In contrast, two rounds of WGS with seven early strains each were sufficient to achieve similar results in one year and involved only 24 000 assays. Such recursive genomic recombination has also been used to improve the acid-tolerance of a commercial lactic acid-producing Lactobacillus sp. (Patnaik et al., 2002).
‘Combinatorial biosynthesis’ is being used for the discovery of new and modified drugs (Hutchinson, 1998; Reeves, 2003). In this technique, recombinant DNA techniques are utilized to introduce genes coding for antibiotic synthases into producers of other antibiotics or into non-producing strains to obtain modified or hybrid antibiotics. The first demonstration of this technology involved gene transfer from a streptomycete strain producing the isochromanequinone antibiotic actinorhodin into strains producing granaticin, dihydrogranaticin and mederomycin (which are also isochromanequinones). This led to the discovery of two new antibiotic derivatives, mederrhodin A and dihydrogranatirhodin. Since this breakthrough paper by Hopwood and coworkers (Hopwood et al., 1985), many hybrid antibiotics have been produced by recombinant DNA technology.
Hundreds of new polyketides have been made by combinatorial biosynthesis (Rodriguez & McDaniel, 2001; Donadio & Sosio, 2003; Kantola et al., 2003). Manipulations include: (i) deletion of one of the domains of a particular module; (ii) addition of a copy of the thioesterase domain to the end of an earlier module resulting in a shortened polyketide; (iii) replacement of an AT domain of a polyketide synthase (PKS) with an AT domain from another PKS, resulting in addition of a methyl group at a particular site or removal of a methyl group; (iv) addition of a reductive domain(s) to a particular module, thus changing a keto group to a double bond or to a methylene group; (v) use of synthetic diketides delivered as N-acetylcysteamine thioesters to load onto the active site of the ketosynthase (KS) in module 2 and to be taken all the way to a novel final product; (vi) replacement of the loading module of one PKS with the loading module of another PKS, thus changing the starter unit from propionate to acetate, for example; and (vii) replacement of the hydroxylase or glycosylase enzymes from one pathway to another, thus modifying the ring structure with respect to OH groups and/or sugars (Staunton, 1998).
As mentioned above, there are many examples of new polyketides been made by combining polyketide biosynthetic genes from different producers (McAlpine et al., 1987; Epp et al., 1989; Donadio et al., 1991, 1993; Weber et al., 1991; Hara & Hutchinson, 1992; Decker & Hutchinson, 1993; Hopwood, 1993; Katz & Donadio, 1993; Khosla et al., 1993; McDaniel et al., 1993a, b, 1999; Hutchinson & Fujii, 1995; Kao et al., 1995; Tsoi & Khosla, 1995; Pacey et al., 1998; Wohlert et al., 1998; Xue et al., 1999; Pfeifer & Khosla, 2001). Some of these novel polyketides contain sugars at normally unglycosylated positions (Trefzer et al., 2002) or as new sugar moieties (Zhao et al., 1999; Mendez & Salas, 2001). New anthracyclines (Bartel et al., 1990; Strohl et al., 1991; Hwang et al., 1995; Niemi & Mantsala, 1995; Kim et al., 1996; Ylihonko et al., 1996) and peptide antibiotics (Stachelhaus et al., 1995) have also been made by combinatorial biosynthesis.
Microorganisms produce many compounds of industrial interest. These may be very large materials such as proteins, nucleic acids, carbohydrate polymers, or even cells, or they can be smaller molecules that can be essential for vegetative growth or inessential, i.e. primary and secondary metabolites, respectively. The power of the microbial culture in the competitive world of commercial synthesis can be appreciated by the fact that even simple molecules are made by fermentation rather than by chemical synthesis. Most natural products are so complex that they probably will never be made commercially by chemical synthesis. Strains isolated from nature produce only tiny amounts of product. This is because they need these secondary metabolites for their own competitive benefit, and they do not overproduce these metabolites. Regulatory mechanisms have evolved in microorganisms which enable a strain to avoid excessive production of its metabolites, thus, strain improvement programs are required for commercial application. The goal is to isolate cultures exhibiting desired phenotypes. Most commonly, the ability of a strain to improve titer is what is desired, although the other traits may also be improved on. The tremendous increases in fermentation productivity and the resulting decreases in costs have come about mainly by using mutagenesis. In recent years, recombinant DNA technology has also been applied. The promise of the future is via extensive use of new genetic techniques such as: (i) metabolic engineering accomplishing quantification and control of metabolic fluxes and including inverse metabolic engineering and transcript expression analyses such as association analysis and massive parallel signature sequencing; (ii) directed evolution; (iii) molecular breeding including DNA shuffling and whole genome shuffling; and (iv) combinatorial biosynthesis. These efforts will facilitate not only the isolation of improved strains but also the elucidation and identification of new genetic targets to be used in strain improvement programs.
The authors thank the following colleagues for supplying information: Richard H. Baltz, Graham S. Byng, Richard P. Elander, David A. Hopwood, Daslav Hranueli, Krishna Madduri, Jaraslav Spizek, William R. Strohl and J. Mark Weber.