DNA breaks as triggers for antigenic variation in African trypanosomes.

The DNA repair machinery has been co-opted for antigenic variation in African trypanosomes. New work directly demonstrates that a double-strand break initiates a switch in the expressed variant surface coat.

The African trypanosome, Trypanosoma brucei, is a protozoan parasite of major medical and economic importance. The phenomenon of antigenic variation in T. brucei and some other pathogenic microbes enables them to evade the immune system and persist in the host [1]. In T. brucei, antigenic variation depends on expression of a single variant surface glycoprotein (VSG) from one among multiple expression sites [2] and an archive of more than 1,600 VSG genes [3]. Two principal mechanisms have emerged for the activation of a new VSG: duplicative gene conversion into the active expression site; and in situ activation of a silent expression site coupled with repression of the one previously active. The former is the dominant mechanism underlying antigenic variation [4], and is the focus of new work reported recently in Nature by Boothroyd et al. [5]. As double-strand breaks (DSBs) have long been known to initiate genetic recombination processes by facilitating strand invasion, it had been proposed that a DSB in the 70 bp repeats adjacent to each VSG gene (Figure 1) could initiate the process that copies a new VSG into the active expression site [6]. Boothroyd and colleagues now provide compelling evidence in support of this idea [5].

Figure 1

Model for homologous-recombination-mediated double-strand break repair and VSG switching. Center, a double-strand break (scissors) within the 70 bp repeats (yellow bar) at the active expression site initiates a search for a homologous repair template. Telomeric (T2AG3) repeats are indicated by the black arrowheads. Top, a silent telomeric VSG (up to 250 are available) can be copied into the active expression site via a gene-conversion cassette mechanism or by break-induced replication extending to the end of the chromosome. The longer stretches of 70 bp repeats found at these loci may increase the likelihood that these tracts are recognized and invaded by the broken strand at the active expression site. Bottom, replacement by a silent array VSG must use a cassette mechanism. The synthesis-dependent strand-annealing model [25] favors the type of recombination illustrated rather than crossover between silent VSG arrays and expression sites.

VSG genes are copied into the active expression site

The VSG genes are located at subtelomeres, and immune selection has had a major impact on the evolution of these chromosomal regions in T. brucei [7]. The expression sites are found on the megabase (1-6 Mbp) and intermediate-sized (150-500 kbp) chromosomes. They form long polycistronic transcription units that include a variable number of expression-site-associated genes (ESAGs) and a VSG located within a few kilobases of the telomere [2]. In addition, each of the 100 or so minichromosomes contains at least one telomeric VSG. Surprisingly, genome sequencing revealed that the remaining 1,000-2,000 archive VSGs were mostly pseudogenes (more than 85%) arranged in subtelomeric tandem arrays on the megabase chromosomes [3]. In total, about 30% of the genome is dedicated to VSG-based immune evasion.

It has been known for some time that 70 bp repeats form the boundaries of VSG gene conversion events (Figure 1), suggesting that they provide a substrate for recombination between the active expression site and a silent VSG locus [8]. Comparison of genetic maps and DNA sequencing before and after a switch suggested that VSG copying initiates 5' to the VSG gene, but that the termination point depends on the source of the copied gene; array VSGs often recombine such that their 3' end is altered [9], whereas recombination with a silent telomeric VSG results in a conversion event that extends beyond this point [10,11] (Figure 1). Genome sequencing further revealed the distribution and degeneracy of the 70 bp repeats, with most (92%) full-length array VSGs having at least one repeat 1-2 kbp upstream (one, 74%; two, 14%; or more, 5%) [3]. Importantly, the expression-site VSGs are typically flanked upstream by tens to maybe more than a hundred 70 bp repeats (Figure 1).

It has been challenging to determine how the variety of VSG templates available for recombination intersect with the host immune response and contribute to the course of a natural infection. Together with the sequence data, functional studies have begun to clarify some details. Silent VSGs fall into distinct activation-probability groups that correlate with the amount of flanking sequence shared with the active VSG [12]. As an infection progresses and immunity develops against high-probability VSGs, lower-probability VSGs emerge. Thus, VSG genes located adjacent to telomeres are more likely to be activated than those located in an array. The number of flanking 70 bp repeats may be particularly important if a DSB in this sequence triggers a homology search among the vast VSG archive. In fact, VSG gene conversion can occur in the absence of 70 bp repeats at the active expression site [13] and chimeric VSGs may be formed via VSG-VSG recombination [3], but these events are analogous to the low-probability switches in the scenario outlined above.

Double-strand breaks trigger VSG switching

The 70 bp repeats were first implicated as recombination 'hotspots' able to initiate the VSG copying process more than 25 years ago [6]. In order to test the hypothesis that a DSB adjacent to these repeats could initiate a switch, Boothroyd et al. [5] placed a yeast meganuclease (I-SceI) recognition sequence in the active expression site immediately adjacent to the 70 bp repeats; the trypanosome strain was also modified to express an inducible copy of the meganuclease [5]. Induction of I-SceI expression generated a DSB in around 1% of the cells, which increased the switching frequency by around 250-fold. In contrast, a DSB elsewhere in the expression site or after deletion of the 70 bp repeats did not increase switching frequency. These data indicate that both the location of the DSB next to the repeats and the repeats themselves are critical to increased switching.

Boothroyd et al. [5] then analyzed 18 of the switched progeny, revealing that they had all lost the active VSG, replacing it with donor VSGs from a silent expression site (15/18) or a mini-chromosome (3/18, top panel in Figure 1). To determine how the induced DSB was processed, the authors next sequenced across the repaired region in five of the progeny, all of which had switched by recombination with the same silent expression site. They were able to show that the DSB invaded 70 bp repeats proximal to the donor VSG, resulting in the replacement of some of the 70 bp repeats as well as the original VSG. Furthermore, they demonstrated that the subtelomeric region downstream of the original VSG had also been replaced. This is reminiscent of break-induced replication, where a broken chromosome invades a homologous template and initiates replication to the telomere [14] (Figure 1).

The results detailed above demonstrate that a meganuclease-generated DSB adjacent to the 70 bp repeats triggers antigenic variation, but is this what happens naturally? To investigate this the authors used ligation mediated-PCR (LM-PCR) with DNA templates from wild-type trypanosomes. More frequent, mostly staggered DSBs were detected in the 70 bp repeats at the active expression site compared with a 70 bp repeat array in a silent expression site or at a chromosomal internal locus, adding further support to the hypothesis that DSBs occur in vivo and trigger VSG switching. Interestingly, however, an I-SceI-induced break appears to be a far more potent trigger for a VSG-switch than these natural breaks.

Is there a rationale for why the 70 bp repeats might be more prone to DSBs, and why there should be a difference between an active and a silent expression site? The 70 bp repeats consist, in part, of a large number of TTA repeats. These have been shown to destabilize plasmid duplex DNA [15], leading Boothroyd et al. [5] to speculate that this region becomes destabilized during active expression site transcription, resulting in the formation of natural DSBs. Also, unusual structures formed by repetitive sequences may be prone to replication-fork stalling and collapse, or they may provide a substrate for a trypanosome endonuclease. Any combination of these factors could explain why breaks are frequently seen in the 70 bp repeats and specifically at the active expression site. In addition, base-J, a modified thymidine implicated in suppressing homologous recombination, is found throughout silent expression sites but not in the active one nor in insect-stage trypanosomes [16]. The replacement of a proportion of the thymidines in silent expression sites may somehow restrict the occurrence of DSBs.

One intriguing feature of antigenic variation in T. brucei is the rapid decline in switching frequency when fresh 'natural' isolates are adapted to laboratory culture; switching drops from about 10-2 per population doubling and by up to 10,000-fold during growth in culture [17]. Two models that could account for this substantial but reversible decrease include changes in telomere length and accumulation of base-J. Cross and colleagues [18] proposed a model whereby increases in telomere length during growth in culture reduces the frequency of subtelomeric breaks. Alternatively, and as described above, the accumulation of base-J [16] may stabilize the 70 bp repeats in silent expression sites, making them less prone to DSBs or to strand invasion and recombination.

Questions for the future

Some additional questions are worth considering in light of the results of Boothroyd et al. [5]. First, if breaks in the 70 bp repeats can also initiate recombination with array VSGs (bottom panel in Figure 1), then what prevents replication from continuing beyond the VSG 3'-end? It may be that strand invasion at ectopic 70 bp repeats prompts a second homology search that favors extensive local homology (long tracts of telomeric repeats, for example), but will use more restricted homology or even microhomology if necessary. Second, is replication directional and, if so, how is this controlled? That is, can sequences be copied on either side of the 70 bp repeats and could breaks in the 70 bp repeats also be responsible for recombination events that generate upstream ESAG diversity? Third, subtelomeric clusters have been implicated in recombination events in T. brucei [19] and in other protozoa but the active expression site is sequestered in the expression-site body, an extranucleolar focus of RNA polymerase I that drives monoallelic VSG transcription [20]. During a recombination-mediated switch, is the DNA break repaired inside or outside of the expression-site body or does the structure disassemble transiently?

The 70 bp repeats are likely to play a key role in antigenic variation and they are probably also important for VSG diversification beyond the active expression site [3,21]. Although the repair pathways available to T. brucei [22] have yet to be dissected genetically in any detail, it is known that RAD51-dependent homologous recombination is responsible for only a subset of VSG-copying events [23]. This is consistent with a predominant break-induced replication model (Figure 1), as RAD51-dependent and independent break-induced replication pathways have been described in yeast [14]. Fortunately, meganucleases such as I-SceI provide powerful tools to study the repair of unique DSBs in vivo [24]. It is exciting to see new opportunities for probing the molecular mechanisms underlying antigenic variation in trypanosomes and for genetically dissecting the pathways used to switch VSG gene expression.