Sequential gene disruption in Candida albicans by FLP-mediated site-specific recombination

The genetic manipulation of the human fungal pathogen Candida albicans is difficult because of its diploid genome, the lack of a known sexual phase and its unusual codon usage. We devised a new method for sequential gene disruption in C. albicans that is based on the repeated use of the URA3 marker for selection of transformants and its subsequent deletion by FLP-mediated, site-specific recombination. A cassette was constructed that, in addition to the URA3 selection marker, contained an inducible SAP2P-FLP fusion and was flanked by direct repeats of the minimal FLP recognition site (FRT). This URA3 flipper cassette was used to generate homozygous C. albicans mutants disrupted for both alleles of either the CDR4 gene, encoding an ABC transporter, or the MDR1 gene, encoding a membrane transport protein of the major facilitator superfamily. After insertion of the URA3 flipper into the first copy of the target gene, the whole cassette could be efficiently excised by induced FLP-mediated recombination, leaving one FRT site in the disrupted allele of the target gene. The URA3 flipper was then used for another round of mutagenesis to disrupt the second allele. Deletion of the cassette from primary and secondary transformants occurred exclusively by intrachromosomal recombination of the FRT sites flanking the URA3 flipper, whereas interchromosomal recombination between FRT sites on the homologous chromosomes was never observed. This new gene disruption strategy facilitates the generation of specific, homozygous C. albicans mutants as it eliminates the need for a negative selection scheme for marker deletion and minimizes the risk of mitotic recombination in sequential disruption experiments.     

Development of an efficient genetic manipulation strategy for sequential gene disruption and expression of different heterologous <Emphasis Type="Italic">GFP</Emphasis> genes in <Emphasis Type="Italic">Candida tropicalis</Emphasis>

The diploid yeast Candida tropicalis, which can utilize n-alkane as a carbon and energy source, is an attractive strain for both physiological studies and practical applications. However, it presents some characteristics, such as rare codon usage, difficulty in sequential gene disruption, and inefficiency in foreign gene expression, that hamper strain improvement through genetic engineering. In this work, we present a simple and effective method for sequential gene disruption in C. tropicalis based on the use of an auxotrophic mutant host defective in orotidine monophosphate decarboxylase (URA3). The disruption cassette, which consists of a functional yeast URA3 gene flanked by a 0.3 kb gene disruption auxiliary sequence (gda) direct repeat derived from downstream or upstream of the URA3 gene and of homologous arms of the target gene, was constructed and introduced into the yeast genome by integrative transformation. Stable integrants were isolated by selection for Ura⁺ and identified by PCR and sequencing. The important feature of this construct, which makes it very attractive, is that recombination between the flanking direct gda repeats occurs at a high frequency (10^?8) during mitosis. After excision of the URA3 marker, only one copy of the gda sequence remains at the recombinant locus. Thus, the resulting ura3 strain can be used again to disrupt a second allelic gene in a similar manner. In addition to this effective sequential gene disruption method, a codon-optimized green fluorescent protein-encoding gene (GFP) was functionally expressed in C. tropicalis. Thus, we propose a simple and reliable method to improve C. tropicalis by genetic manipulation.     

Mutagenic inverted repeat assisted genome engineering (MIRAGE)

Here we describe a one-step method to create precise modifications in the genome of Saccharomyces cerevisiae as a tool for synthetic biology, metabolic engineering, systems biology and genetic studies. Through homologous recombination, a mutagenesis cassette containing an inverted repeat of selection marker(s) is integrated into the genome. Due to its inherent instability in genomic DNA, the inverted repeat catalyzes spontaneous self-excision, resulting in precise genome modification. Since this excision occurs at very high frequencies, selection for the integration event can be followed immediately by counterselection, without the need for growth in permissive conditions. This is the first time a truly one-step method has been described for genome modification in any organism.     

A Method for Gene Disruption That Allows Repeated Use of USR3 Selection in the Construction of Multiply Disrupted Yeast Strains

In this paper, we describe a 3.8-kb molecular construct that we have used to disrupt yeast genes. The construct consists of a functional yeast URA3 gene flanked by 1.1-kb direct repeats of a bacterial sequence. It is straightforward to insert the 3.8-kb segment into a cloned target gene of interest and then introduce the resulting disruption into the yeast genome by integrative transformation. An appropriate DNA fragment containing the disruption plus flanking homology can be obtained by restriction enzyme digestion. After introducing such fragments into yeast by transformation, stable integrants can be isolated by selection for Ura+. The important feature of this construct that makes it especially useful is that recombination between the flanking direct repeats occurs at a high frequency (10(-4)) in vegetatively grown cultures. After excision, only one copy of the repeat sequence remains behind. Thus in the resulting strain, the Ura+ selection can be used again, either to disrupt a second gene in similar fashion or for another purpose.     

Large-scale mutagenesis and functional genomics in yeast

One of the major goals for the post-genome era is determining of the function of proteins predicted in the genome sequence. In many organisms functional assignments have been the results of comparative sequencing, proteomics or expression profiling. In the yeast, Saccharomyces cerevisiae, however, the functional role of a gene can be tested directly by disrupting the gene and examining the phenotype of the mutant. Because precise targeted deletions can be easily constructed, it is also possible to systematically delete every gene in the genome. Here we describe recent progress in yeast genome-wide mutagenesis programs and the results produced from analyzing the mutants created by them. Electronic Publication     

Genetic transformation of the yeast Rhodotorula gracilis ATCC 26217 by electroporation

We examined the genetic transformation of the biotechnologically relevant yeast Rhodotorula gracilis ATCC 26217 by electroporation. To evaluate the yeast transformation, we created a genomic integration cassette that was targeted to the yeast orotidine-5′-phosphate decarboxylase gene (URA3 gene) locus and composed of the zeocin-resistant gene (Sh ble gene) with the URA3 promoter and terminator of the yeast. The yeast was unable to grow on medium containing 2.0 μg/mL zeocin, even with the inoculation of a large number of cells (approximately 1.0 × 10⁸ cells/plate). Using the integrative cassette and zeocin-containing medium, we successfully obtained yeast transformants by electroporation, and the highest transformation efficiency of approximately 40 colony-forming units/μg DNA was obtained with a 0.6-kV electrical pulse. No homologous integration of the cassette at the URA3 gene locus was detected by the analyses of uracil auxotrophy and genomic PCR of transformants, suggesting that this method is a useful tool for randomly mutating the yeast genome.     

Novel Method for Genomic Promoter Shuffling by Using Recyclable Cassettes

Genetic elements of interest can be introduced into the Saccharomyces cerevisiae genome via homologous recombination. The current method is to link such an element to a selectable marker gene to be integrated into the target locus. However, the marker gene in this method cannot be reused, which limits repeated manipulation of the yeast genome. An alternative method is to utilize a counterselectable gene, such as URA3, with flanking tandem repeats. After integration, URA3 along with one copy of the repeat can be popped out via internal recombination, leaving behind one copy of the unwanted repeat. Here we describe a novel concept of genetic element shuffling in which the tandem repeats are made of the desired genetic element, so that after integration and popping out, only one copy of the element remains at the desired locus to function. As a proof of principle, we constructed three recyclable cassettes (P_PGK1-URA3-P_PGK1, P_GAL1-URA3-P_GAL1, and P_tetO7-URA3-P_tetO7) and integrated them upstream of an engineered chromosomal P_HIS3-mCherry-Myc locus. After promoter shuffling, the mCherry-Myc gene was regulated precisely as anticipated.     

The Green Monster Process for the Generation of Yeast Strains Carrying Multiple Gene Deletions

Phenotypes for a gene deletion are often revealed only when the mutation is tested in a particular genetic background or environmental condition^1,2. There are examples where many genes need to be deleted to unmask hidden gene functions^3,4. Despite the potential for important discoveries, genetic interactions involving three or more genes are largely unexplored. Exhaustive searches of multi-mutant interactions would be impractical due to the sheer number of possible combinations of deletions. However, studies of selected sets of genes, such as sets of paralogs with a greater a priori chance of sharing a common function, would be informative.In the yeast Saccharomyces cerevisiae, gene knockout is accomplished by replacing a gene with a selectable marker via homologous recombination. Because the number of markers is limited, methods have been developed for removing and reusing the same marker^5,6,7,8,9,10. However, sequentially engineering multiple mutations using these methods is time-consuming because the time required scales linearly with the number of deletions to be generated.Here we describe the Green Monster method for routinely engineering multiple deletions in yeast¹¹. In this method, a green fluorescent protein (GFP) reporter integrated into deletions is used to quantitatively label strains according to the number of deletions contained in each strain (Figure 1). Repeated rounds of assortment of GFP-marked deletions via yeast mating and meiosis coupled with flow-cytometric enrichment of strains carrying more of these deletions lead to the accumulation of deletions in strains (Figure 2). Performing multiple processes in parallel, with each process incorporating one or more deletions per round, reduces the time required for strain construction.The first step is to prepare haploid single-mutants termed ''ProMonsters,'' each of which carries a GFP reporter in a deleted locus and one of the ''toolkit'' loci—either Green Monster GMToolkit-a or GMToolkit-α at the can1Δ locus (Figure 3). Using strains from the yeast deletion collection¹², GFP-marked deletions can be conveniently generated by replacing the common KanMX4 cassette existing in these strains with a universal GFP-URA3 fragment. Each GMToolkit contains: either the a- or α-mating-type-specific haploid selection marker¹ and exactly one of the two markers that, when both GMToolkits are present, collectively allow for selection of diploids.The second step is to carry out the sexual cycling through which deletion loci can be combined within a single cell by the random assortment and/or meiotic recombination that accompanies each cycle of mating and sporulation.     

Genome-wide Gene Deletions in Streptococcus sanguinis by High Throughput PCR

Transposon mutagenesis and single-gene deletion are two methods applied in genome-wide gene knockout in bacteria ^1,2. Although transposon mutagenesis is less time consuming, less costly, and does not require completed genome information, there are two weaknesses in this method: (1) the possibility of a disparate mutants in the mixed mutant library that counter-selects mutants with decreased competition; and (2) the possibility of partial gene inactivation whereby genes do not entirely lose their function following the insertion of a transposon. Single-gene deletion analysis may compensate for the drawbacks associated with transposon mutagenesis. To improve the efficiency of genome-wide single gene deletion, we attempt to establish a high-throughput technique for genome-wide single gene deletion using Streptococcus sanguinis as a model organism. Each gene deletion construct in S. sanguinis genome is designed to comprise 1-kb upstream of the targeted gene, the aphA-3 gene, encoding kanamycin resistance protein, and 1-kb downstream of the targeted gene. Three sets of primers F1/R1, F2/R2, and F3/R3, respectively, are designed and synthesized in a 96-well plate format for PCR-amplifications of those three components of each deletion construct. Primers R1 and F3 contain 25-bp sequences that are complementary to regions of the aphA-3 gene at their 5'' end. A large scale PCR amplification of the aphA-3 gene is performed once for creating all single-gene deletion constructs. The promoter of aphA-3 gene is initially excluded to minimize the potential polar effect of kanamycin cassette. To create the gene deletion constructs, high-throughput PCR amplification and purification are performed in a 96-well plate format. A linear recombinant PCR amplicon for each gene deletion will be made up through four PCR reactions using high-fidelity DNA polymerase. The initial exponential growth phase of S. sanguinis cultured in Todd Hewitt broth supplemented with 2.5% inactivated horse serum is used to increase competence for the transformation of PCR-recombinant constructs. Under this condition, up to 20% of S. sanguinis cells can be transformed using ~50 ng of DNA. Based on this approach, 2,048 mutants with single-gene deletion were ultimately obtained from the 2,270 genes in S. sanguinis excluding four gene ORFs contained entirely within other ORFs in S. sanguinis SK36 and 218 potential essential genes. The technique on creating gene deletion constructs is high throughput and could be easy to use in genome-wide single gene deletions for any transformable bacteria.     

Use of site-specific recombination to regenerate selectable markers

Summary A method which allows the repeated use of a single selectable marker in DNA transformations was demonstrated. This marker regeneration method employed portions of the Saccharomyces cerevisiae 2 m circle plasmid: the inverted repeat sequences (FRTs), and the FLP gene whose product, a site-specific recombinase, catalyzes recombination events between FRTs. When FRTs were oriented as direct repeats and integrated into the genome of the yeast Pichia pastoris, FLP-mediated recombination resulted in the efficient and precise deletion of DNA located between the repeats. In the example described, the S. cerevisiae ARG4 gene, placed between a set of FRTs and integrated into Pichia in a prior transformation, was deleted by FLP, thereby regenerating an arginine-requiring phenotype in the P. pastoris strain.     

Isogenic Strain Construction and Gene Mapping in Candida Albicans

Genetic manipulation of Candida albicans is constrained by its diploid genome and asexual life cycle. Recessive mutations are not expressed when heterozygous and undesired mutations introduced in the course of random mutagenesis cannot be removed by genetic back-crossing. To circumvent these problems, we developed a genotypic screen that permitted identification of a heterozygous recessive mutation at the URA3 locus. The mutation was introduced by targeted mutagenesis, homologous integration of transforming DNA, to avoid introduction of extraneous mutations. The ura3 mutation was rendered homozygous by a second round of transformation resulting in a Ura(-) strain otherwise isogenic with the parental clinical isolate. Subsequent mutation of the Ura(-) strain was achieved by targeted mutagenesis using the URA3 gene as a selectable marker. URA3 selection was used repeatedly for the sequential introduction of mutations by flanking the URA3 gene with direct repeats of the Salmonella typhimurium hisG gene. Spontaneous intrachromosomal recombination between the flanking repeats excised the URA3 gene restoring a Ura(-) phenotype. These Ura(-) segregants were selected on 5-fluoroorotic acid-containing medium and used in the next round of mutagenesis. To permit the physical mapping of disrupted genes, the 18-bp recognition sequence of the endonuclease I-SceI was incorporated into the hisG repeats. Site-specific cleavage of the chromosome with I-SceI revealed the position of the integrated sequences.     

Selectable cassettes for simplified construction of yeast gene disruption vectors

Cassettes based on a hisG-URA3-hisG insert have been modified by the addition of a Km^R-encoding gene and flanking polylinker sites, greatly simplifying construction of gene disruption vectors in Escherichia coli. After gene disruption in yeast, URA3 can then be excised by recombination between the hisG repeats flanking the gene, permitting reuse of the URA3 marker     

Molecular Analysis of a Deletion Hotspot in the NRXN1 Region Reveals the Involvement of Short Inverted Repeats in Deletion CNVs

NRXN1 microdeletions occur at a relatively high frequency and confer increased risk for neurodevelopmental and neurobehavioral abnormalities. The mechanism that makes NRXN1 a deletion hotspot is unknown. Here, we identified deletions of the NRXN1 region in affected cohorts, confirming a strong association with the autism spectrum and other neurodevelopmental disorders. Interestingly, deletions in both affected and control individuals were clustered in the 5′ portion of NRXN1 and its immediate upstream region. To explore the mechanism of deletion, we mapped and analyzed the breakpoints of 32 deletions. At the deletion breakpoints, frequent microhomology (68.8%, 2–19 bp) suggested predominant mechanisms of DNA replication error and/or microhomology-mediated end-joining. Long terminal repeat (LTR) elements, unique non-B-DNA structures, and MEME-defined sequence motifs were significantly enriched, but Alu and LINE sequences were not. Importantly, small-size inverted repeats (minus self chains, minus sequence motifs, and partial complementary sequences) were significantly overrepresented in the vicinity of NRXN1 region deletion breakpoints, suggesting that, although they are not interrupted by the deletion process, such inverted repeats can predispose a region to genomic instability by mediating single-strand DNA looping via the annealing of partially reverse complementary strands and the promoting of DNA replication fork stalling and DNA replication error. Our observations highlight the potential importance of inverted repeats of variable sizes in generating a rearrangement hotspot in which individual breakpoints are not recurrent. Mechanisms that involve short inverted repeats in initiating deletion may also apply to other deletion hotspots in the human genome.     

Long inverted repeats are an at-risk motif for recombination in mammalian cells

Certain DNA sequence motifs and structures can promote genomic instability. We have explored instability induced in mouse cells by long inverted repeats (LIRs). A cassette was constructed containing a herpes simplex virus thymidine kinase (tk) gene into which was inserted an LIR composed of two inverted copies of a 1.1-kb yeast URA3 gene sequence separated by a 200-bp spacer sequence. The tk gene was introduced into the genome of mouse Ltk(-) fibroblasts either by itself or in conjunction with a closely linked tk gene that was disrupted by an 8-bp XhoI linker insertion; rates of intrachromosomal homologous recombination between the markers were determined. Recombination between the two tk alleles was stimulated 5-fold by the LIR, as compared to a long direct repeat (LDR) insert, resulting in nearly 10(-5) events per cell per generation. Of the tk(+) segregants recovered from LIR-containing cell lines, 14% arose from gene conversions that eliminated the LIR, as compared to 3% of the tk(+) segregants from LDR cell lines, corresponding to a >20-fold increase in deletions at the LIR hotspot. Thus, an LIR, which is a common motif in mammalian genomes, is at risk for the stimulation of homologous recombination and possibly other genetic rearrangements.     

Replication slippage between distant short repeats in Saccharomyces cerevisiae depends on the direction of replication and the RAD50 and RAD52 genes.

Small direct repeats, which are frequent in all genomes, are a potential source of genome instability. To study the occurrence and genetic control of repeat-associated deletions, we developed a system in the yeast Saccharomyces cerevisiae that was based on small direct repeats separated by either random sequences or inverted repeats. Deletions were examined in the LYS2 gene, using a set of 31- to 156-bp inserts that included inserts with no apparent potential for secondary structure as well as two quasipalindromes. All inserts were flanked by 6- to 9-bp direct repeats of LYS2 sequence, providing an opportunity for Lys+ reversion via precise excision. Reversions could arise by extended deletions involving either direct repeats or random sequences and by -1-or +2-bp frameshift mutations. The deletion breakpoints were always associated with short (3- to 9-bp) perfect or imperfect direct repeats. Compared with the POL+ strain, deletions between small direct repeats were increased as much as 100-fold, and the spectrum was changed in a temperature-sensitive DNA polymerase delta pol3-t mutant, suggesting a role for replication. The type of deletion depended on orientation relative to the origin of replication. On the basis of these results, we propose (i) that extended deletions between small repeats arise by replication slippage and (ii) that the deletions occur primarily in either the leading or lagging strand. The RAD50 and RAD52 genes, which are required for the recombinational repair of many kinds of DNA double-strand breaks, appeared to be required also for the production of up to 90% of the deletions arising between separated repeats in the pol3-t mutant, suggesting a newly identified role for these genes in genome stability and possibly replication.     

Gene deletions causing human genetic disease: mechanisms of mutagenesis and the role of the local DNA sequence environment

Summary Reports describing short (< 20 bp) gene deletions causing human genetic disease were collated in order to study underlying causative mechanisms. Deletion break-point junction regions were found to be non-random both at the nucleotide and dinucleotide sequence levels, an observation consistent with an endogenous sequencedirected mechanism of mutagenesis. Direct repeats of between 2 bp and 8 bp were found in the immediate vicinity of all but one of the 60 deletions analysed. Direct repeats are a feature of a number of recombination, replication or repair-based models of deletion mutagenesis and the possible contribution of each to the spectrum of mutations examined was assessed. The influence of parameters such as repeat length and lenght of DNA between repeats was studied in relation to the frequency, location and extent of these deletions. Findings were broadly consistent with a slipped mispairing model but the predicted deletion of one whole repeat copy was found only rarely. A modified version of the slipped mispairing hypothesis was therefore proposed and was shown to possess considerable explanatory value for 25% of deletions examined. Whereas the frequency of inverted repeats in the vicinity of gene deletions was not significantly elevated, these elements may nevertheless promote instability by facilitating the formation of secondary structure intermediates. A significant excess of symmetrical sequence elements was however found at sites of single base deletions. A new model to explain the involvement of symmetric elements in frameshift mutagenesis was devised, which successfully accounted for a majority of the single base deletions examined. In general, the loss of one or a few base pairs of DNA was found to be more compatible with a replication-based model of mutagenesis than with a recombination or repair hypothesis. Seven hitherto unrecognized hotspots for deletion were noted in five genes (AT3, F8, HBA, HBB and HPRT). Considerable sequence homology was found between these different sites, and a consensus sequence (TGA/GA/GG/ TA/C) was drawn up. Sequences fitting this consensus (i) were noted in the immediate vicinity of 41% of the other (sporadic) gene deletions, (ii) were found frequently at sites of spontaneous deletion in the hamster APRT gene, (iii) were found to be associated with many larger human gene deletions/translocations, (iv) act as arrest sites for human polymerase a during DNA replication and (v) have been shown by in vitro studies of human polymerase a to be especially prone to frameshift mutation. It is proposed that dissociation of polymerase a at arrest sites may, by providing a stable single stranded substrate, lead to deletion of a DNA sequence either by slipped mispairing via a number of different secondary structure intermediates, or by strand-switching or base misincorporation. Human gene deletions thus appear to be caused by multiple mechanisms whose relative importance is probably governed by local primary and secondary DNA structure. Our ability to predict precisely the location and extent of a gene deletion is however hampered both by this complexity and by the possibility that these mechanisms may often act in combination.     

Stability of an inverted repeat in a human fibrosarcoma cell.

Deletions and rearrangements of DNA sequences within the genome of human cells result in mutations associated with human disease. We have developed a selection system involving a neo gene containing a DNA sequence inserted into the NcoI site that can be used to quantitatively assay deletion of this sequence from the chromosome. The spontaneous deletion from the neo gene of a 122 bp inverted repeat occurred at a rate of 2.1 x 10(-8) to <3.1 x 10(-9) revertants/cell/generation in three different cell lines. Deletion of the 122 bp inverted repeat occurred between 6 bp flanking direct repeats. Spontaneous deletion of a 122 bp non-palindromic DNA sequence flanked by direct repeats was not observed, indicating a rate of deletion of <3.1 x 10(-9) revertants/cell/generation. This result demonstrates that a 122 bp inverted repeat can exhibit a low level of instability in some locations in the chromosome of a human cell line.     

Sequence analysis of the insertion element ISH1.8 and of associated structural changes in the genome of phage PhiH of the archaebacterium Halobacterium halobium

We have sequenced the insertion element ISH1.8 which can be present in one or two copies in the genome of phage ΦH of Halobacterium halobium. ISH1.8 is 1895 bp long, has no inverted repeat at its ends, and one only of the two copies is flanked by two 5-bp duplications. An 8-bp sequence composed of 4 bp from each end of ISH1.8 is present in both sites lacking the element. This 8-bp sequence could either be a specific insertion sequence or a part of the element that is left behind upon deletion. The plasmid pΦHL, consisting of the invertible L segment of the phage genome which is, in ΦH2 and ΦH5, flanked by two copies of ISH1.8, contains 112 bp of ISH1.8 and is released from the phage genome by recombination within a direct repeat of 9 bp. This 9-bp sequence (TCCCGCCCT) exists as an inverted repeat in ISH1.8 and therefore as two distinct repeats in phage genomes containing two copies of ISH1.8 in inverted orientation.     

Analysis of repeat-mediated deletions in the mitochondrial genome of Saccharomyces cerevisiae

Mitochondrial DNA deletions and point mutations accumulate in an age-dependent manner in mammals. The mitochondrial genome in aging humans often displays a 4977-bp deletion flanked by short direct repeats. Additionally, direct repeats flank two-thirds of the reported mitochondrial DNA deletions. The mechanism by which these deletions arise is unknown, but direct-repeat-mediated deletions involving polymerase slippage, homologous recombination, and nonhomologous end joining have been proposed. We have developed a genetic reporter to measure the rate at which direct-repeat-mediated deletions arise in the mitochondrial genome of Saccharomyces cerevisiae. Here we analyze the effect of repeat size and heterology between repeats on the rate of deletions. We find that the dependence on homology for repeat-mediated deletions is linear down to 33 bp. Heterology between repeats does not affect the deletion rate substantially. Analysis of recombination products suggests that the deletions are produced by at least two different pathways, one that generates only deletions and one that appears to generate both deletions and reciprocal products of recombination. We discuss how this reporter may be used to identify the proteins in yeast that have an impact on the generation of direct-repeat-mediated deletions.     

Instability of artificially circularized chromosomes of Streptomyces lividans

The chromosomes of Streptomyces species are linear molecules, containing long terminal inverted repeats and covalently bound terminal proteins. These chromosomes undergo spontaneous deletions of the terminal sequences at high frequencies and become circularized in several cases examined. Artificial circularization of the Streptomyces lividans chromosome was also achieved by targeted recombination in vivo, in which the terminal inverted repeats of the chromosome were connected by a kanamycin resistance gene (aphII ). Under kanamycin selection, the circularized chromosomes harboured tandem amplifications of a 20.2 kb sequence that included the aphII gene flanked by direct repeats and deletions nearby. On release from kanamycin selection, the aphII amplifications and the neighbouring sequences were deleted from the chromosomes, rendering all the cultures kanamycin sensitive. The chloramphenicol resistance gene, which was prone to deletion in wild-type S. lividans, became much more stable in the kanamycin-sensitive derivatives. These results indicate that the telomeres and/or certain terminal sequences may be involved in the structural instability of Streptomyces chromosomes.