The instability problem in Streptomyces strain development The genus Streptomyces contains a large group of filamentous, differentiating, gram-positive, soil bacteria. These bacteria qualify as the most important group of industrial microorganisms, with a repertoire of about 4,000 varieties of antibiotics (approximately two thirds of all known), and many other useful secondary metabolites and extracellular enzymes. Boosting the fermentation productivity of useful biochemicals by Streptomyces has traditionally been achieved by empirical mutation-screening schemes with good success rates. Because of the random nature of mutations in classical genetic programmes, however, the high-yielding cultures inevitably accumulate mutations not contributing to the increased productivity. A number of these mutations are surely deleterious to the health of the mutants, as evident from the general trend of growth deterioration along the genealogy lines as the programme progresses. Another frequent sign of deterioration is the elevated genetic instability that is almost universally present in Streptomyces. The instability of certain genetic traits in Streptomyces was first reported by Beijerinick1 more than 80 years ago. Since then it has been discovered in essentially every species investigated (For review see references 2-5). The rates of spontaneous mutations for the unstable genetic traits are in the order of 10-4 to 10-2, or about 3 to 4 orders of magnitude higher than those in typical genes. The mutations frequently exhibit pleiotropy, i.e., several traits being altered at the same time. Most of the traits affect, in a broad sense, secondary metabolism, e.g., aerial mycelium formation and sporulation, production of pigments and extracellular enzymes, and biosynthesis of and resistance to antibiotics. The only common unstable trait concerning a primary metabolite is that of arginine biosynthesis. Arginine auxotrophy arises at high frequencies usually, but not always, as the result of the loss of the argininosuccinate synthetase (argG) gene. Molecular studies have shown that the trait losses are the results of large deletions in the chromosomal DNA. The sizes of deletions are tens to thousands of kb long, removing the structural genes of particular enzymes as well as regulatory genes. The deletions are frequently accompanied by tandem amplifications of up to several hundred copies of specific DNA sequences (named `amplifiable units of DNA' or AUD6) of about 3 to 100 kb. Two types of AUD have been found -- Type I AUD exist in single copies flanked by short repeats in the progenitor chromosome, and their amplifications are non-reproducible; Type II AUD are either flanked by long direct repeats or already present in duplications in the chromosome, and their amplifications are reproducible. The instability poses difficulties to industrial geneticists and fermentation engineers. The instability frequently results in the total or almost total loss of productivity, or the loss of sporulation capability, or both. The loss of sporulation creates difficulties in strain preservation, seed culture preparation in fermentation, as well as genetic manipulations (which, for lack of haploid spores, must resort to the multi-nuclei hyphal fragments). In the early phase of a strain improvement program the frequencies of instability are usually not high enough to encumber the genetic or fermentation processes significantly. However, as the program progresses, the instability problems tend to worsen. In the most extreme case I have witnessed the highest-yielding industrial strains of Streptomyces throw off non-sporulating and non-producing mutants at frequencies reaching 50%. Hypervariable mutants like these may in principle be eliminated during the screening. This, however, may not always be easily executed, because these hypervariable strains are sometimes the best producers. This is witnessed by another industrial case: an oxytetracycline high producer of Streptomyces rimosus gave off 80% of variants, most of which were asporulant and produced little or no oxytetracycline7. There are also numerous other less well-studied (many undocumented) examples of spontaneous high-frequency loss of antibiotic productivity in Streptomyces cultures.
The unstable regions are at the chromosomal ends The unstable regions of the Streptomyces chromosomes have been located in one or two contiguous segments, ranging from tens to thousands of kb. Recently, Lin et al.8 discovered that the chromosomes of many Streptomyces spp. are linear DNA molecules. These large chromosomal DNAs (most about 8 Mb) have terminal inverted repeats. Covalently attached to the 5' end of the chromosomal DNAs are proteins which presumably act as the primers of replication (as in adenoviruses, bacteriophage [phi]29, and other linear replicons containing terminal proteins9). Comparison of the results of Lin et al. 8 with those of Kieser et al.10 and Redenbach et al,11 revealed that the unstable regions are at both termini of the chromosomes of S. lividans 66 and S. coelicolor A3(2). This feature is probably true also for other Streptomyces species as well. In the best-studied S. lividans case, essentially all deletions involve one or both telomeres11. The association of the telomeres with the instability may shed some light on the directionality of sequential deletions observed in some Streptomyces species. In S. lividans the telomere-proximal chloramphenicol resistance (cmr) gene tends to be deleted before the telomere-distal argG gene11-13. In Streptomyces glaucescens, the directional and step-wise deletions were observed for three markers (in order of decreasing instability), strS (streptomycin resistance), melC (melanin), and a 100-kb AUD2. The three markers were indeed physically mapped in that order separated by 350 and 150 kb, respectively. Directionality of deletional events has also been observed in other less well-characterized species. The apparent sequential deletions suggested a multi-step mechanism, whereby a primary deletional event involving the telomere leads to an unstable intermediate stage which provokes a subsequent rearrangement.
Circularization of the linear chromosome An important question is: what makes the terminal sequences highly susceptible to deletions? Specifically, what biochemical machinery is responsible, and which is the primary determinant there -- the presence of the ends or specific sequences? There is no a prior knowledge applicable to these questions. To screen for mutations blocked in structural instability directly and randomly would be a very tedious operation. On the other hand, if we assume that instability is an inherent property of linear chromosomes, perhaps they may be stabilized by circularization. Circularization does indeed occur both spontaneously and artificially. Redenbach et al.11 isolated several spontaneous cmr mutants of S. lividans ZX7 with deletions of hundreds of kb (including the cmr gene). When put into the context of the linear structure of the S. lividans ZX7 chromosome, some deletions appeared to span both termini. If these deletions were produced by recombinational events between sequences on the two arms, the two truncated arms would have been joined together and the chromosome circularized. Lin et al.8 showed that these were indeed the case. Lin et al.8 also showed that circularization could be achieved artificially with relative ease using a targeted recombination process (Fig. 1). In this case, the telomeres were removed and replaced by a continuous piece of DNA - a kanamycin resistance gene (aphII). The circularized chromosomes, maintained in the presence of kanamycin, appeared to be much more stable than the linear chromosomes as far as the cmr gene was concerned, but the cultures harbouring the circularized chromosomes grew and sporulated poorly.
FIGURE 1. Targeted circularization of the S. lividans chromosome8. The linear chromosome of S. lividans ZX7 is shown as an arc with the indicated origin of replication (oriC), the unstable regions (in red), terminal proteins (red solid circles), and some markers in the unstable regions -- cmr (chloramphenicol resistance), argG (argininosuccinate synthetase), AUD1 (Type II) and AUD2 (Type I). Typical spontaneous deletions resulting in circularization are exemplified by the three red arcs. The terminal regions are enlarged to show the 30-kb terminal inverted repeats (thick converging arrows). Artificial circularization was achieved by targeted recombination using a suicide (non-replicating) vector that contained a DNA segment (solid red box) from each arm of the chromosome flanking the kanamycin resistance (aphII) gene (solid blue box). Transformants were isolated that had undergone double crossovers (dashed lines) that deleted the telomeres and circularized the chromosome. However, the circularized chromosome contained, instead of the simple replacement (bracketed), tandem amplifications (bracket with the subscript n) of the aphII gene together with one of the flanking DNAs. On release from kanamycin selection, all the amplified DNA was deleted together with long stretches of neighbourin
Structural complications in circularized chromosomes At this point it may seem that stabilizing the chromosomes by planned circularization is in grasp. However, things turned out to be more complicated. First of all, at the junction of the two arms of the chromosomes a 22-kb DNA segment including the aphII gene plus the neighboring inverted repeat was tandemly amplified (Fig. 1). This did not seem to be a problem, and the amplification was stably maintained under kanamycin selection. However, on removal of kanamycin further rearrangements occurred: the amplifications at the junction were deleted together with large stretches of neighbouring sequence. Interestingly, the resulting cultures appeared to regain growth vigour. Hence, the simple idea of stabilizing a Streptomyces chromosome by circularization, although easily carried out in a controlled manner, is complicated in outcome. Circularization alone appears to be insufficient for stability. If anything, circularization per se may cause poor growth and induce instability. This is probably not surprising, considering that the Streptomyces chromosomes have presumably adapted to linearity through a long period of evolution. The sizes and sequences of the chromosomal DNA deleted during circularization may have a direct bearing on the well-being of the cells that harbour the circularized chromosomes, and to the stability of the latter. It is likely that the presence or absence of some particular sequences is crucial. In particular, the telomeres of the Streptomyces chromosomes, like their eukaryotic counterparts, may also play important roles in anchoring, motility, and sorting during the cell cycle. Further studies are necessary to determine the sequences that play important structural and/or functional roles in the stability of the termini of the Streptomyces chromosomes. This knowledge is important for a possible rational design of circularization, in which particular terminal sequences are to be removed to achieve stable structures. For examples, perhaps one or both of the inverted repeats must be eliminated. The linear chromosomes of Streptomyces spp. contain an oriC in the centre8, from which replication is initiated and travels towards to the telomeres14. The termini probably are patched up by replication primed by the terminal proteins15. There may perhaps be terminators on both arms of the chromosomes where the replication forks from oriC and the telomeres meet. Replication terminators in bacteria are known to be prone to high frequencies of recombination16. The presence of the putative terminators in the terminal regions of Streptomyces chromosomes may be responsible for the instability. It is also possible that the terminators on the linear chromosomes of Streptomyces are not equipped to deal with the replication of a circular chromosome, and thus further rearrangements are necessary for improved growth. In planned circularization schemes, one may deliberately remove one or both of the terminators, and perhaps a relatively well-characterized terminator system from a circular chromosome (such as that of Escherichia coli or Bacillus subtilis) may be `borrowed' for the circularized chromosome. Genetic complications in circularization In the circularization of the linear chromosome, whether spontaneous or artificial, a stretch of DNA from each chromosomal arm must be removed. Many undesirable spontaneous circularization events remove genes necessary for normal growth, differentiation, and production of certain antibiotics. In designing an artificial circularization, one would like to choose the deletion end points so that no such genes are removed. This adds a constrain to the designing of artificial circularization. It is possible that, to achieve a final stable structure, one or more of beneficial genes must be removed. In this case, perhaps the gene(s) in question may be moved to another locations of the chromosome or a plasmid. Studies on the instability of Streptomyces ambofaciens RP181100, however, revealed a probably different picture17. All variants containing amplifications of a particular AUD in the unstable regions of S. ambofaciens lose spiramycin productivity. The spiramycin-less mutations were not stable. Spontaneous spiramycin-producing revertants were readily isolated, in which the amplifications had disappeared. These observations suggest that the AUD contains a repressor gene for the production of spiramycin. This conjecture, if proven, would represent a reverse scenario where deletions become desirable. Other chromosome stabilization possibilities The aforementioned approach aims at the removal of the unstable sequences. An alternative would be to remove the enzymatic machinery responsible for rearrangements. Unfortunately, we know nothing about the biochemical mechanism of the instability. Homologous recombination is an obvious suspect. However, there have been no proven homologous recombination mutants available in Streptomyces, except for JT46 of S. lividans18. Unfortunately, JT46 while defective in recombination within a circular plasmid, was not impaired of general chromosomal recombination, thus resembling mutants in the RecF pathway in E. coli19. The chromosome of JT46 is as unstable as that of its recombination-proficient parent. Since the structural instability may be mediated by a pathway corresponding to the RecBCD pathway of E. coli, one would like to test mutants in this pathway, preferably in the recA gene. A classical approach to screening for recA mutants among ultraviolet light-sensitive mutants20 and a more direct screening of mutants defective in intraplasmid recombination18 both failed to find recA mutants. Recently, recA genes from Streptomyces species are now available21, 22, and attempts are being made to generate recA mutations using these sequences in targeted mutagenesis. The recA mutants may stabilize the chromosomes, but they would also be barred from genetic exchanges, and would most likely acquire undesirable characteristics known to associated with the mutation, such as poor growth and hypersensitivity to DNA damaging agents. It has also been proposed that essential genes might be moved into the deletion-prone region to kill the deletion mutants23. This would solve the instability problem in genetic programmes where haploid spores are manipulated. However, Streptomyces cultures in liquid fermentation in general do not sporulate, and the nucleoids in the mycelium are not completely isolated from one another. This multi-chromosomal nature of the Streptomyces mycelium could allow genomes with deletions of an essential gene to persist, being complemented by intact genomes sharing the same cytoplasm. Alternatively, one may move important genes from the unstable terminal regions to a stable location on the chromosome or a stable plasmid. The frequent loss of some antibiotics biosynthesis gene clusters, such as those for oxytetracycline in S. rimosus, is probably due to their terminal locations (In contrast, the internally located actinorhodin biosynthesis gene cluster is very stable in S. coelicolor). Frequent deletions of these production genes may possibly be avoided in the new locations. Placing the gene cluster on a multicopy plasmid has the potential bonus of increased production resulting from higher gene dosage. It seems, therefore, that the instability problem is not to be solved in a straightforward fashion. Better understanding of the structure, replication, and recombination of the Streptomyces chromosomes, and the nature of the instability of the chromosomal termini, are called for. These studies may answer the intriguing question whether structural instability is so much a part of the linear chromosomes of Streptomyces that any complete cure is impossible without some deleterious effects on the chromosomes. It is hoped that at least some compromises might be engineered. Acknowledgments Studies done at Yang-Ming University are supported by research grants (NSC83-0418-B010-017-BC and NSC84-0418-B010-033-BC ) from National Science Council, Taiwan, ROC. I thank Dr. David A. Hopwood for helpful comments on the manuscript. References 1. Beijerinick, M.W. (1913) Folia Microbiol. 2, 185-200 |
