Nucleic Acids Research

Formation of linear amplicons

As previously reported (12), linear amplicons derived from the H locus of L.tarentolae were generated during the first step of selection with MTX (Fig. 2A). At higher drug concentrations the copy number of linear amplicons either increased (e.g. Fig. 2A and D, lane 13) or were lost when extrachromosomal circles were formed (Fig. 2D, lane 20). An unexpected result was the presence of mixed populations of neo and hyg linear amplicons in several of the mutants (Fig. 2A-C). In previous experiments, only one allele was tagged (12), which, with the probes used, did not permit distinction of whether a mixed population of neo and non-tagged linear amplicons was present. In this study, linear amplicons created from both alleles were shown to occur not only in total populations, but also in individual clones isolated from a mutant containing mixed neo and hyg linear amplicons (Fig. 3).

Gene amplification in Leishmania was thought to be mainly conservative (i.e. chromosomal copies remain intact), although clear examples of non-conservative amplification have been described, which in the cases studied were shown to occur by intrachromosomal recombination between direct repeats (1,10,24). The availability of two tagged alleles has permitted the observation that chromosomal deletions are frequent (Fig. 2E, lanes 8, 10, 15, 20, Fig. 2F, lane 2, Fig. 3B, lane 2 and Fig. 3C, lane 5). The extent (in size) of these deletions is unknown. Although deletions can clearly occur during the first selection steps (see Fig. 3), in other instances gene deletion is observable only in cells highly resistant to MTX (compare Fig. 2B and E, lanes 8, 10 and 20). Even if in several cases the allele deleted is that found to be amplified (Fig. 2E, lanes 8, 10 and 20 and Fig. 2F, lane 2), there are examples to the contrary (e.g. Fig. 3, lanes 2 and 5), where the deleted chromosomal locus is neither amplified nor present in the cell. Loss of the deleted fragment (most likely present initially as an unstable amplicon) may occur by unequal segregation in daughter cells. This loss may be accelerated by the appearance of more effective MTX resistance mechanisms, such as transport mutations (19,30), rendering the amplicon superfluous. Deletions may be even more frequent, but would remain unnoticed if they occur during normal replication, as the daughter with the amplicon has an equal chance of getting the intact or deleted chromosome depending on segregation (see also 10). As these deletions may be large, the intact allele may be selected more frequently. Gene deletion may be more easily observable in highly resistant mutants as the high concentration of MTX may unbalance the deoxynucleotide pools, which could contribute to prolonged replication pausing, hence facilitating segregation of shorter allelic versions. This suggests that gene amplification is a continuing process in Leishmania, with several non-productive events (see also the analysis of several clones of mutant 5MTX50 presented in Fig. 3) until one event is selected that is advantageous to the cell.

Formation of extrachromosomal circular amplicons

This work was initiated in part to test whether linear amplicons were precursors of circles in MTX-selected L.tarentolae. The unexpected frequent amplification of both alleles as part of linear amplicons in the same cell (Fig. 2) has complicated our correlative analysis between linear and circular amplicons, as circles could as well be generated from the chromosomal copy instead of the linear amplicons. To circumvent that complication, we integrated a hyg tag into a linear amplicon (Fig. 4). Ten independent clones with an hyg marker present in the linear amplicon but absent from the chromosomal locus were selected for MTX resistance. Extrachromosomal circles were present in five mutants and three of them hybridized to the hyg probe (Fig. 5). These results unambiguously demonstrate that linear amplicons can undergo secondary rearrangements to generate circular molecules. For the two circles not hybridizing with hyg (Fig. 5) the circles could still be formed from linear amplicons, as only half of the linear amplicons were shown to be tagged with hyg (Fig. 4C).

In MTX-selected L.tarentolae, the formation of extrachromosomal linear molecules seems to be an initial event. The event leading to formation of linear amplicons is not defined with precision, but in the few instances where the amplicons were mapped, they were shown to contain large inverted repeats (9,11,13,31-33) and a fold-back mechanism was proposed to explain the appearance of such an inverted structure. This may lead to chromosomal deletions (see above). With increasing concentrations of the selective pressure, circular amplicons are formed and, most likely, linear amplicons are the source of the circles (Fig. 5). As it is apparently easier to obtain a higher copy number of circles than linear amplicons (Fig. 2), it may explain why circles are formed at higher selection pressure. Interestingly, one of the mutants selected with 50 µM MTX and containing only the 450 kb linear amplicon was kept in culture in the presence of 50 µM MTX for several passages. Analysis of this mutant after 6 months of culture in the presence of 50 µM MTX indicated that circular amplicons were formed (not shown), suggesting that circles are more easily maintained and are more stable than linear amplicons. This may provide an explanation for the presence of H circles in wild-type Leishmania cells (22,34,35).

Although linear amplicons seem to be the likely initial event and circles are possibly derived from these, our work did not address why cells need to go through a linear intermediate. It is possible that specialized sequences are present on the 800 kb chromosome that may facilitate the formation of linear amplicons. It is quite remarkable that most linear amplicons have the same size (Fig. 2), an observation consistent with the presence of those putative specialized sequences. Mapping of the rearranged junction in linear amplicons may shed light on the mechanism of formation of the linear amplicons. With an increased copy number of ptr1-containing sequences, circles may be more easily formed by the recombination mechanisms already described (10,12,24). Indeed, recombination frequency was recently found to be proportional to the copy number of the target sequence in Leishmania (36).

Finally, these studies may also have implications in the study of drug resistance in clinical isolates. Individual clones of a resistant population, even if selected under controlled laboratory conditions, are remarkably different at the genome level (Figs 3 and 4B). This pinpoints the challenge of determining resistance mechanisms in field isolates, where heterogeneity is expected to be large (20). The growing of the organism necessary to obtain sufficient material to study resistance may select for subpopulations that are not necessarily a reflection of the initial resistant isolates.

ACKNOWLEDGEMENTS

We thank B.Papadopoulou (CHUL) for critical reading of the manuscript. This work was supported in part by the a grant from Natural Science and Engineering Research Council of Canada (NSERC) to M.O. K.G. was the recipient of NSERC and FCAR studentships, C.K. is a post-doctoral fellow of the Schweizerischer Nationalfonds and M.O. was a research fellow of the Fonds de Recherche en Santé du Québec and is now an MRC Scientist and is the recipient of a Burroughs Wellcome Fund New Investigator Award in Molecular Parasitology.