|Home | About | Journals | Submit | Contact Us | Français|
Trypanosoma brucei relies on antigenic variation of its Variant Surface Glycoprotein (VSG) coat for survival. We show that VSG switching can be efficiently studied in vitro using VSG RNAi in place of an immune system to select for switch variants. Contrary to models predicting an instant switch after inhibition of VSG synthesis, switching was not induced by VSG RNAi and occurred at a rate of 10−4 per division. We find a highly reproducible hierarchy of VSG activation which appears to be capable of resetting, whereby more than half of the switch events over 12 experiments were to one of two VSGs. We characterised switched clones according to switch mechanism using marker genes in the active VSG expression site (ES). Transcriptional switches between ESs were the preferred switching mechanism, whereby at least 10 of the 17 ESs identified in T. brucei 427 can be functionally active in vitro. We could specifically select for switches mediated by DNA rearrangements by inducing VSG RNAi in the presence of drug selection for the active ES. Most of the preferentially activated VSGs could be activated by multiple mechanisms. This VSG RNAi based procedure provides a rapid and powerful means for analysing VSG switching in African trypanosomes entirely in vitro.
The African trypanosome Trypanosoma brucei can successfully proliferate in the mammalian bloodstream due to a sophisticated strategy of antigenic variation of a homogeneous Variant Surface Glycoprotein (VSG) coat. An infected host can effectively clear a given VSG variant via antibody mediated lysis once it has mounted the appropriate antibody response. However, as VSG switch variants are continuously being generated, these temporarily escape destruction. As individual trypanosomes have many hundreds of VSG genes, cyclical waves of parasitaemia make up a chronic infection which can persist for years. Antigenic variation in African trypanosomes is reviewed in: (Barry and McCulloch, 2001; Donelson, 2003; Pays et al., 2004; Vanhamme et al., 2001).
The active VSG is transcribed in a mutually exclusive fashion from one of about twenty telomeric bloodstream form VSG expression sites (ES) (Chaves et al., 1999; Pays et al., 1989). The active ES is located in a discrete extranucleolar body (ESB), which appears to contain the transcription and RNA processing machinery necessary for high level expression of VSG (Navarro and Gull, 2001). Switching can involve transcriptional control, as the cell switches between ESs [reviewed in: (Borst and Ulbert, 2001)]. Alternatively, DNA rearrangements can slot a previously inactive VSG into the active ES transcription unit via gene conversion or telomere exchange (Pays et al., 1983; Pays et al., 1985). Gene conversion is the most important mechanism during the course of an infection, as it allows access to a much larger pool of silent VSGs rather than just those at telomeres (Robinson et al., 1999). Segmental gene conversion reactions appear to play a predominant role later in an infection, allowing new mosaic VSGs to be formed (Roth et al., 1989; Thon et al., 1989; Thon et al., 1990).
Traditionally VSG switching has been investigated using animals. Chronic infections can be established in rabbits, goats or cattle (Barry, 1986; Capbern et al., 1977; Gray, 1965). However, these experiments are very laborious to perform, as peaks of parasitaemia are typically too low to detect reliably by microscopy. Subsequent amplification steps in mice are normally necessary before variants can be characterised (Michels et al., 1983). In addition, experiments using chronic infections have the disadvantage that it is normally impossible to establish if a switch variant has arisen from the variant preceding it, or was already present as a minor variant in the infection. For this reason, ‘single relapse’ experiments in mice or rats are performed, whereby one defined switch away from a given VSG is analysed (Miller and Turner, 1981). However, it requires a large number of rodents to generate a panel of independent clonal switch variants. Multiple switch variants can be isolated from a single mouse (McCulloch et al., 1997; Rudenko et al., 1998), but it can be impossible to determine if superficially identical switch variants arose independently. All of these approaches to study VSG switching generate relatively few clonal T. brucei variants with a known pedigree.
We show here that VSG RNAi can be used in place of an immune system to rapidly select for large and statistically significant numbers of clonal T. brucei antigenic variants with a known parentage in vitro. We do not find any evidence that blocking VSG synthesis triggers a switch, as fluctuation test analysis showed that VSG switching occurred spontaneously at a rate of about 10−4 per cell division, and was not induced by the VSG RNAi. We identified the activated VSG expressed in a total of 168 switched clones by sequencing VSG cDNA amplified by RT-PCR, and show a highly reproducible preferential hierarchy of VSG activation. Most of the preferentially activated subset of VSGs were found to be activated by different switching mechanisms.
Induction of VSG221 RNAi in the T. brucei 221VG1.1 and 221VG2.1 strains results in the ablation of VSG221 transcript down to 1–2% normal levels, and the cell-cycle arrest and subsequent death of all trypanosomes expressing VSG221 (Sheader et al., 2005). In these strains a puromycin resistance gene integrated behind the promoter of the active VSG221 ES allows maintenance of homogeneous VSG221 expression in the presence of puromycin selection. However, after induction of VSG221 RNAi for about four days, resistant trypanosomes invariably grow out. Immunofluorescence analysis (result not shown) and protein analysis of these revertants show that these resistant cells had switched their VSG (Fig. 1). VSG is normally the major band in Coomassie stained gels of protein lysates from bloodstream form T. brucei. After over four days of induction of VSG221 RNAi, cells expressing VSG221 had almost completely disappeared from the population, and other abundant proteins appeared, of which one of the identified bands corresponded to VSG1.8 (Fig. 1). This indicated that VSG221 RNAi could be used as a novel means for selecting VSG switch variants in the absence of an immune system.
In order to investigate VSG switching in more detail, we designed an experimental procedure allowing us to select for very large numbers of independent VSG switch events in vitro using microtitre dishes (Fig. 2). Pilot experiments indicated that VSG221 RNAi resistant trypanosomes were generated at a frequency of approximately 10−4 per division, and that these had indeed switched their VSG. In order to ensure that we were generating independent VSG switch events and not reisolating VSG switch events already present in the parental population, we repeated the switching experiments with parallel cultures of trypanosomes too small to contain a switch variant. A range of dilutions were distributed over the wells of 96-well microtitre dishes (ranging from 15–120 cells per well). These plates were then incubated for typically three generations (7–8 hours per generation) to allow independent switches to occur in the different wells. Tetracycline was then added to induce the VSG221 RNAi, and plates were incubated for a further 6–8 days before scoring positive wells.
First, immunofluorescence microscopy was used to establish that the VSG221 RNAi resistant clones had indeed switched their VSG coat away from VSG221. T. brucei can switch its VSG using different mechanisms based on DNA rearrangement (gene conversion or telomere exchange) or transcriptional control (in situ switch). Our parental T. brucei 221VG1.1 and 221VG2.1 cell lines contain the single copy VSG221 in the active ES, in addition to the puromycin resistance gene and GFP. Comparing the phenotypes and genotypes of the switched trypanosomes allows us to extrapolate which switching mechanism was used (Fig. 3). If the switched clone was still expressing GFP, it must have switched its VSG using DNA rearrangements (gene conversion or telomere exchange). PCR analysis allows us to differentiate between these mechanisms using the single copy genes located in the active VSG221 ES. If the switched GFP expressing clones had lost VSG221 sequences, they must have switched via gene conversion. If they had retained VSG221, they presumably switched via telomere exchange. Trypanosomes that did not express GFP, but contained all marker genes must have switched via an in situ switch. We have regularly found that trypanosomes that have undergone an in situ switch have lost the old ES (Cross et al., 1998; Rudenko et al., 1998). In this case the switched trypanosomes have lost all sequences present in the VSG221 ES.
Using this strategy we analysed a total of 127 VSG221 RNAi resistant trypanosome clones (Fig. 4A). 3% of the VSG221 RNAi resistant clones had not switched away from VSG221, and had instead presumably inactivated the inducible RNAi machinery. This reversion through mutation has been documented before (Chen et al., 2003). However, 97% of clones had switched away from VSG221. Further analysis showed that these had used all expected VSG switching mechanisms. The most frequent switch event was an in situ switch to another ES (77% of total switch events). Switches mediated by gene conversion or telomere exchange were found relatively infrequently, and constituted respectively 4% or 2% of total switch events. During this experimental procedure VSG221 RNAi was performed without drug selection pressure to maintain activity of the VSG221 ES.
In order to select for VSG switches mediated by DNA rearrangements, we next performed VSG221 RNAi in the presence of puromycin selection on the active VSG221 ES (Fig. 4B). This would be expected to select against in situ switch events and leave those mediated by DNA rearrangements. This was indeed the case. A total of 41 VSG221 RNAi resistant clones were analysed. The majority of these clones appeared to have switched via gene conversion (73%) or telomere exchange (7%), indicating that maintaining drug selection on the active ES was indeed a means for selectively isolating VSG switch events mediated by DNA rearrangements. As expected, this modification also resulted in the relative enrichment for clones that had become VSG221 RNAi resistant through inactivation of the RNAi machinery (20 %).
We next determined the switch rate in our T. brucei 427 line with this VSG RNAi mediated procedure using fluctuation test analysis (Luria and Delbruck, 1943). In addition to a determination of the switch rate, this analysis also allowed us to establish if the VSG switch events we were observing were occurring spontaneously, or were being induced by the VSG RNAi. Fluctuation tests rely on inoculation of many parallel cultures containing dilutions too low to contain a mutant, or in our case switch variant. These cultures are then expanded to a predetermined concentration before the selection (VSG RNAi) is induced. These induced cultures are spread over 96-well plates to allow the scoring of individual clones (see Experimental Procedures for details). The more generations a culture is expanded for, the more fluctuation will be seen in the percentage of clones that are resistant to the VSG RNAi depending on exactly when the switch occurred during the amplification step (Luria and Delbruck, 1943).
In this way, we determined the switching frequency in T. brucei 427 using VSG221 RNAi to be approximately 1 × 10−4 per generation (Table 1). If a mutation or switch event is induced by the selection procedure, one would expect the variance in number of clones generated per culture divided by the mean to have a value of one. In contrast, in our experiments this value was consistently higher, arguing that we are indeed analysing spontaneous switch events. In a similar fashion, we performed fluctuation test analysis on VSG switching experiments performed in the presence of puromycin selection pressure for the active ES (Table 1). The frequency of generation of RNAi resistant clones after screening away in situ switches was approximately 7 × 10−6 per division. Similarly, there was no evidence that these were induced by the VSG RNAi, and we therefore appear to be isolating spontaneous switch events. In the experiments where in situ switches were screened away, the difference between the variance and the mean was greater, presumably as a consequence of the larger number of generations that the cultures were expanded for. An increased number of generations would be expected to increase the degree of fluctuation (Luria and Delbruck, 1943).
In order to determine whether or not preferential hierarchies of VSG activation were present, we identified the activated VSG in the switched clones. VSGs have a conserved carboxy terminus (Borst and Cross, 1982). We made cDNA for the new VSG variants using a primer against the conserved 3′ end of VSG. This cDNA was then amplified by PCR using this VSG 3′ primer, and a primer targeted to the spliced leader present on all T. brucei transcripts. Both ends of the VSG cDNAs were then sequenced to determine their identity.
We found that there was a clear and reproducible preferential hierarchy in the VSGs that were activated in our switching experiments (Fig. 5A). VSGbR2 and VSG121 were highly preferentially activated, making up more than half of the total switch events. We have previously cloned 17 T. brucei 427 ESs in yeast, which presumably comprises the entire ES repertoire (Becker et al., 2004). Nine of our activated VSGs are present in our cloned set of ESs (VSGbR2, VSG121, VSG1.8, VSGJS1, VSGNA1, VSG224, VSG800, VSGT3, VSGV02) (see Experimental Procedures for sequence accession numbers). As these VSGs were all activated by in situ switches, our experiments show that at least 10 of the previously characterised 17 ESs in T. brucei 427 can be functionally active in vitro. As expected, switching experiments selecting for DNA rearrangements also resulted in the activation of VSGs for which there is no evidence that they are linked to ESs (for example VSGST1 and VSGST2). Both of these VSGs are located on T. brucei minichromosomes (result not shown), which could be a preferred location for VSGs activated by gene conversion early in an infection (Robinson et al., 1999).
In general, VSGs that were preferentially activated (like VSGbR2 or VSG1.8) were in the subset of VSGs preferentially activated by DNA rearrangements as well as transcriptional control (Fig. 5A). The striking exception to this was VSG121, which was frequently activated by in situ switches but infrequently activated by DNA rearrangements (Fig. 5A). The T. brucei 427 ESs are currently being sequenced, which could provide us with an explanation for the aberrant behaviour of this one ES. This preferential hierarchy of activation of a subset of T. brucei 427 VSGs was highly reproducible over a series of 12 experiments (Fig. 5B). If potential “resetting” of the hierarchy is possible, this did not occur during the six to eight weeks that the T. brucei parental strains were in culture after their rederivation by cloning (see Experimental Procedures for details). Not all activated VSGs could be identified using our RT-PCR based approach (indicated with ? in Fig. 5), however this was a minor subset of total. Possibly some VSGs are more divergent at their 3′ ends, and are not recognised by our anti-sense VSG primer. These would have to be identified by other methods.
RNA analysis of a representative set of different T. brucei clones which had switched their VSG using different mechanisms, confirmed that GFP was only expressed in clones which had undergone DNA rearrangements, and in virtually all cases only the major VSG variant was detectable indicating clonality (Fig. 6). Pulsed Field gel analysis of a representative set of clones confirmed the absence of DNA rearrangements in clones which had undergone in situ switches (Fig. 7). In contrast, clones which had undergone a telomere exchange showed an exchange of VSG221 with another telomeric VSG (Fig. 7). As expected, after having undergone a switch mediated by gene conversion, VSG221 was lost from the active VSG221 ES, and another VSG was inserted in its place.
Switches mediated via gene conversion led to a duplication of the new VSG into the active VSG221 ES. However, in some cases (for example in the case of gene conversion of VSGbR2) there was evidence for loss of copies of VSGbR2, as well as the duplication of VSGbR2 into the active ES (Fig. 8). Multiple rearrangement events during switching in T. brucei that are not immediately related to the switch have been documented before (Aline et al., 1989; Myler et al., 1988; Navarro and Cross, 1996; Van der Ploeg and Cornelissen, 1984). This could indicate the trypanosome undergoes a hyper-recombinogenic state during switching. Alternatively these events could indicate high background rates of recombination between VSG genes.
We show that VSG RNAi can be used instead of an immune system to rapidly generate clonal VSG switch variants completely in vitro. This has allowed us to investigate mechanisms of VSG switching. One possible model for how VSG switching operates, entails a decrease in VSG mRNA or protein synthesis which causes a “stress” response triggering an immediate VSG switch (Borst and Ulbert, 2001). We do not find evidence in support of this, and show using Luria-Delbruck fluctuation tests that VSG switch variants are not induced by the VSG RNAi but arise spontaneously at a frequency of about 10−4 per division. The most frequent switch event is a transcriptional in situ switch away from the VSG221 ES, although we could selectively isolate switches mediated by DNA rearrangements by performing the switch experiments in the presence of drug selection for the active VSG221 ES. More than half of the T. brucei 427 ESs can be functionally active in vitro, as 10 of the 17 T. brucei 427 VSG ESs cloned (Becker et al., 2004) were active in our experiments. We show a highly preferential hierarchy of VSG activation, which was reproducible over a series of 12 experiments. With the exception of VSG121, VSGs that were preferentially activated by transcriptional control also appeared in the set of VSGs preferentially activated by DNA rearrangements.
Is our VSG RNAi based system comparable to VSG switching experiments relying on immunised animals for selection against a given VSG? Both experimental approaches rely on effectively killing trypanosome variants expressing the old VSG ( in our case VSG221). An antibody based approach results in the lysis of all trypanosomes expressing a VSG with VSG221-like epitopes, while an RNAi based method relies on sequence identity. RNAi mediated ablation of transcript requires stretches of identity of between 21–23 nucleotides between the RNAi target fragment and the transcript to be targeted (Scherer and Rossi 2003). Regions of identity between the VSG221 RNAi target fragment and other related VSGs in T. brucei 427 could select against their activation. We do not think that this concern applies to our experiments, as VSG221 is single copy in T. brucei 427, and does not appear to be a member of an extensive family of related genes (Frasch et al., 1982). Our VSG221 RNAi target fragment is from the divergent 5′ end of the VSG221 gene, and does not include the conserved 3′ end common to all VSGs (Borst and Cross, 1982). Using a 5′ probe which includes our VSG221 RNAi fragment, only one very faintly hybridising band was seen in T. brucei 427 under low stringency washing conditions (3 × SSC at 65°C) (Frasch et al., 1982). We therefore find it highly unlikely that cross-hybridisation with other VSGs is introducing bias into our results.
VSG switching has previously been studied using chronic infections typically in rabbits, or single relapses in mice or rats. In these approaches it can be impossible to establish the exact parentage of different switch variants in mixed infections. In contrast, we show that VSG RNAi can provide a much more manipulable means of studying VSG switching, as large and statistically significant numbers of clonal independent switch events are easily and rapidly generated. The manipulability of this system has allowed us to accurately determine rates of switching using Luria-Delbruck fluctuation tests, and establish that switching is not induced by the RNAi selection procedure. In addition, this approach can be used to investigate the effect of perturbations including gene knock-outs, for their effect on VSG switching.
Various genetic modifications disrupting the active ES have resulted in increased rates of ES switching. These have included manipulation of the VSG221 co-transposed region, which resulted in activation of a new ES within a few cell divisions (Davies et al., 1997). Alternatively, replacement of the ES promoter by a T7 promoter in bloodstream form T. brucei prompted an ES switch (Navarro et al., 1999). One model explaining these results, postulated that these ES modifications resulted in decreased synthesis of VSG mRNA or protein. This decrease could lead to a “stress” response resulting in a switch (Borst and Ulbert, 2001). Our results disprove this model, as VSG RNAi did not trigger an instant switch. In contrast, there was no evidence that switching was induced by ablation of VSG transcript by VSG RNAi. This makes it unlikely that switching is coupled to a “sensing” mechanism for either VSG protein or transcript.
Using immunised mice, the rate of switching in T. brucei 427 has been determined as being around 10−6–10−7 per division (Lamont et al., 1986), in contrast to rates of 10−2–10−3 in pleomorphic strains of T. brucei (Turner and Barry, 1989). Using T. brucei 427 with a negative selectable marker inserted in the VSG221 ES, the switch off rate in vitro has been calculated as being between 1–3 × 10−5 per generation (Cross et al., 1998). Our rate of 10−4 per division is higher, which could be a consequence of the experimental procedure. Selection using VSG RNAi could be a more permissive procedure, whereby relatively recent switch events containing transcripts for both the old and the new VSG are still recovered. In contrast, in experiments using the thymidine kinase (TK) gene as a negative selectable marker (Cross et al., 1998), possibly only relatively older switch variants are rescued where TK transcript has been depleted down to low enough levels to allow resistance to the selective agent FIAU. However, an alternative explanation for our relatively high rates of switching, is that very low levels of VSGRNAi, due to “leakiness” in the system, have given rise to a relative increase in VSG switching frequencies.
The majority of our switch events are a consequence of transcriptional activation of other ESs. This does not appear to be the consequence of VSG RNAi resulting in the disruption of a polycistronic precursor transcript derived from the VSG221 ES transcription unit. RNAi against other genes present within the VSG221 ES (either GFP or various Expression Site Associated Genes) does not result in growth arrest (Sheader et al., 2005). This bias for switching by transcriptional activation of other ESs has been seen before in T. brucei 427, and could be a feature of monomorphic instead of pleomorphic T. brucei strains (Liu et al., 1985; McCulloch et al., 1997; Robinson et al., 1999). The extensive nature of our switching series has allowed us to establish that ten different VSG ESs in T. brucei 427 can be individually functionally activated in vitro. Not all of these ES switch variants grow equally well under our in vitro culture conditions, highlighting the importance of generating and amplifying VSG switch events in independent parallel cultures.
The relative frequency of switches mediated by gene conversion or telomere exchange within our experiments was low (6%), which is lower than the relative frequency of telomere exchange or gene conversion from the 70 bp repeats reported for T. brucei 427 (12%) (McCulloch et al., 1997). However, a direct comparison between these experiments is not possible as McCulloch et al documented a high rate of loss of all telomeric markers (65%). This could be explained by either telomere conversion, or loss of the old ES through telomere deletion events. ES deletion events are found frequently in T. brucei 427 (Cross et al., 1998; Rudenko et al., 1998). In our experiments, the majority of switch events isolated in the presence of drug selection for the VSG221 ES, were gene conversions with VSGs presumably located in other ESs. This bias could be a consequence of the unusually short 70 bp repeat arrays in the VSG221 ES, which would favour recombination on upstream ESAG sequences present in other ESs (Kooter et al., 1988; Liu et al., 1985). The 70 bp repeats provide upstream sequence homology facilitating the gene conversion of donor VSGs not located in ESs (Campbell et al., 1984). VSGbR2 and VSG1.8 are present in multiple copies within the cell, and we do not have proof that the donor VSG was the ES-located copy. However, this is highly likely due to the predisposition of the VSG221 ES for undergoing telomere conversions with other ESs (Kooter et al., 1988; McCulloch et al., 1997). It is likely that the relatively low rates of DNA rearrangement we observe, are a consequence of the VSG221 ES sequence, as well as being a feature of monomorphic lines. Establishing these VSG RNAi switching experiments in a pleomorphic T. brucei line, would provide a useful comparison point.
A striking and highly preferential hierarchy of VSG activation could be found reproducibly over a series of 12 switching experiments, whereby over half of the variants generated had switched to either VSGbR2 or VSG121. Earlier switching experiments with T. brucei 427 have documented preferred activation of specific VSGs. In the earliest T. brucei 427 switching experiments, up to half of the variants generated were VSG1.8 (Michels et al., 1984). Later experiments showed preferential activation of VSGV02 (Rudenko et al., 1995). This highly preferential activation of one or two ES located VSGs indicates that at any point in time one or two ESs are in a “preactive” state (Chaves et al., 1999; Ulbert et al., 2002). As VSG1.8 and VSGV02 were relatively infrequently activated in our experiments, this could indicate that resetting of the preferential hierarchy within T. brucei 427 can take place.
Our experiments were performed so as to minimise potential “resetting” of this hierarchy within the experimental period. The parental T. brucei transformants used for switching were recloned to minimise potential heterogeneity, and thawed stabilates were not maintained in culture longer than one month (see Experimental Procedures for details). At the moment we can not exclude that this apparent shift in the preferential hierarchy of activated VSGs is a direct consequence of the VSG RNAi technique itself. However, we find this scenario very unlikely, as VSG221 does not appear to be part of an extensive family of related genes which would be affected by VSG221 RNAi (Frasch et al., 1982). Excluding this possibility will require further experiments in order to determine if “resetting” of the preferentially activated subset of VSGs can occur within our experimental system.
Surprisingly, with the striking exception of VSG121, many VSGs that were preferentially activated by an in situ switch (for example VSGbR2 or VSG1.8) were also present in the set of VSGs that were preferentially activated by DNA rearrangement. Some of these activated VSGs are present as multiple copies, and we can not be sure which was the donor gene. However, as the VSG221 ES frequently undergoes telomere conversions with other ESs (Kooter et al., 1988; McCulloch et al., 1997), we find it likely that the donor VSG was ES-located. The active ES has been shown to be present within the nucleus in a discrete extranucleolar body (ESB) presumably containing both transcription and RNA processing factories (Navarro and Gull, 2001). Possibly the “preactive” state leading to preferential activation of an ES entails physical proximity to the active ES, whereby the “preactive” ES hijacks the transcription and RNA processing machinery within the ESB. However in addition, possibly proximity of these telomeres within the nucleus facilitates DNA recombination. Fluorescent In Situ Hybridisation experiments (FISH) with ES specific sequences in T. brucei could allow us to test if ESs thought to be preferentially activated are indeed in physical proximity to the active ES within the ESB.
The striking exception to this observation is VSG121, which was frequently transcriptionally activated, but rarely moved into the active ES via DNA rearrangement. VSG121 is present within an ES referred to as the Dominant Expression Site (DES), as this was a preferred site for VSGs to move via duplicative gene conversion in T. brucei 427 (Liu et al., 1985; Michels et al., 1983; Timmers et al., 1987). We do not know why none of the VSG121 genes were preferred VSG donors for the VSG221 ES. The basic organisation of both the VSG221 ES and DES appears similar, although the DES lacks the large duplications present in the VSG221 ES (Crozatier et al., 1990). The repertoire of T. brucei 427 ESs is currently being sequenced at the Sanger Institute, which should give us insight into what is causing this aberrant behaviour.
If the reason that some VSGs are preferentially activated is a consequence of subnuclear positioning, this would imply that chromosome ends within the nucleus of T. brucei could form relatively stable associations with each other. Telomeres in T. brucei have been shown to associate in clusters within the nuclei (Chung et al., 1990). It has been argued that these chromosome ends are predominantly in the central zone of the nucleus in long slender bloodstream form T. brucei (Perez-Morga et al., 2001), but it is not known if there is preferred clustering of specific ends. In Plasmodium the VAR genes are frequently located at telomeres, which form clusters within the nucleus presumably facilitating interchromosomal recombination (Freitas-Junior et al., 2000). However, in Plasmodium there is no evidence that specific ends form stable interactions with each other (Freitas-Junior et al., 2000). Activation of particular VAR loci appears to be associated with movement to particular regions of the nuclear periphery (Duraisingh et al., 2005; Freitas-Junior et al., 2005; Ralph et al., 2005). Possibly, in contrast in T. brucei, “preactive” ESs are in a more stable subnuclear structure within the cell whereby they enjoy privileged access to the active ES within the ESB.
The trypanosomes used were modifications of Trypanosoma brucei 90–13, a T. brucei 427 221a line (MiTat 1.2) containing genes encoding T7 RNA polymerase and tetracycline repressor [described in (Wirtz et al., 1999)]. The VSG RNAi cell lines T. brucei 221VG1.1 and 221VG2.1 are transformants of T. brucei 90–13 containing the MC177VSG221 RNAi construct targeted into minichromosomes (result not shown). This construct contains an 803 bp fragment of VSG221 (MiTat 1.2) (accession number X56762, positions 122–925) cloned between the opposing T7 promoters of construct p2T7Ti-177(Wickstead et al., 2002). Subsequently, the 221GP1 construct containing eGFP and the puromycin resistance gene (Sheader et al., 2004) was inserted immediately behind the VSG221 ES promoter in these cells. Maintaining these transformants on puromycin selection allowed us to select for activity of the VSG221 ES, and maintain the cultures as homogeneous VSG221 expressors. Immunofluorescence analysis of these trypanosomes indicated that non VSG221 variants were not present at detectable levels (less than 10−3).
T. brucei 427 lines were cultured in HMI-9 medium (Sheader et al., 2004). The parental T. brucei 221VG1.1 and 221VG2.1 lines were maintained on 5 μg ml−1 hygromycin (Roche), 2 μg ml−1 G418 (Gibco), 2.5 μg ml−1 phleomycin (Sigma), and 0.2 μg ml−1 puromycin unless otherwise indicated. During normal VSG switching experiments puromycin selection for the VSG221 ES was removed. Puromycin selection was maintained when VSG switches mediated by DNA rearrangements were specifically selected for.
Selection for independent VSG switch events using RNAi was carried out by diluting a logarithmic growth culture (about 106 cells ml−1) of T. brucei VG1.1 or T. brucei VG2.1 down to 100, 200, 400 or 800 cells ml−1 (15–120 cells per well) in medium. Each dilution was aliquoted over 96-well plates. As the switching frequency in our cell lines using this method in vitro was about 10−4 per cell division, cultures were so dilute that it was highly unlikely for individual wells to already contain a VSG switch event. The microtitre dishes were incubated for three generations resulting in approximately 800, 1600, 3200 or 6400 cells ml−1 (120–960 cells per well). This amplification step resulted in the generation of independent VSG switch events in individual wells. VSG221 RNAi was then induced by the addition of tetracycline to a final concentration of 750 ng ml−1 and trypanosomes were not removed from tetracycline during the expansion of the VSG221 RNAi resistant variants. VSG221 RNAi resistant clones emerged 6–8 days following induction of VSG221 RNAi. When VSG switch events mediated by DNA rearrangements were specifically selected for, cells were plated at the same density in the presence of puromycin selection. However, these cultures were allowed to grow for typically 7 generations before induction of RNAi, to compensate for the lower frequency of these switch events.
Fluctuation tests allow the determination of mutation or switching rates. An analysis of the variance compared to mean number of mutants or switch events generated during amplification of independent cultures also allows one to establish if generation of these variants has been induced by the experimental procedure (Luria and Delbruck, 1943; Rosche and Foster, 2000). We typically inoculated either 16 or 24 independent 1 ml cultures of T. brucei 221VG1.1 or 221VG2.1 at concentrations ranging from 5 to 50 cells ml−1. A control culture was set in as a control for cell growth. Cells were amplified for typically 8 or 9 generations, and then VSG221 RNAi was induced by adding tetracycline to an end concentration of 750 ng ml−1. Each culture was spread over 12 wells of a 96-well plate and wells were allowed to grow out over 6–8 days before scoring.
The P0 method was used to calculate the rates of VSG switching, whereby P0 is the fraction of wells from a given culture that do not contain VSG221 RNAi resistant cells (Luria and Delbruck, 1943; Rosche and Foster, 2000). The number of mutants per culture (m) was calculated from −ln P0. The swiching rate was calculated as m divided by the number of cells per culture at the point of induction of VSG RNAi (Nt). The mean number of positive wells per culture, and the variance in the number of switchers per culture were also calculated. A value for variance divided by mean that is close to 1.0 is an indication that VSG switching has been induced by the procedure.
All VSG221 RNAi resistant clones were first analysed by immunofluorescence microscopy using rabbit anti-VSG221 specific antibodies to determine if they had indeed switched their VSG221 coat. If so, GFP expression was monitored by microscopy. Diagnostic PCR reactions were performed using specific primers for VSG221, eGFP or the large subunit of RNA polymerase I (Pol I). Primer sequences available on request. RNA was isolated using RNeasy mini kits (Qiagen). To identify the newly expressed VSG, VSG cDNA was cloned using RT-PCR. VSG cDNA was generated using the Omniscript RT kit (Qiagen) and the primer VSG14XbaI-as (5′-CCCGCCTCTAGACGTGTTAAAATATATCAG-3′). PCR was performed using the same anti-sense primer and a primer against the spliced leader: s-miniexRI (5′-CCGGAATTCGGCTATTATTAGAACAGTTTCTG-3′). Both ends of the VSG cDNA were sequenced to determine its identity. Alternative names for the T. brucei 427 genes listed plus their GenBank accession numbers are as follows: VSGbR2 or MiTat 1.11 (AY935577), VSG121 or MiTat 1.6 (X56764), VSG1.8 or MiTat 1.8 (AY935574), VSGNA1 or MiTat 1.13 (AY935576), VSG224 or MiTat 1.3 (AY935575), VSGT3 or MiTat 1.21 (AY935572), VSGV02 or MiTat 1.9 (AY935573).
For protein analysis T. brucei 221VG2.1 was induced with tetracycline (750 ng ml−1) in the absence of puromycin selection at a density of 8 × 105 cells ml−1. Cells were washed twice, and then resuspended in cold lysis buffer (50 mM HEPES pH 7.5, 1.5 mM MgCl2, 1 mM EDTA, 10% glycerol, 1% Triton X-100 and one mini protease inhibitor cocktail tablet [Roche] per 10 ml mix) at 109 cells ml−1. After microfuge centrifugation, supernatant from the equivalent of 5 × 106 cells was loaded per lane of an 8% SDS polyacrylamide gel. The gel was blotted to Hybond-P membrane (Amersham) and subsequently reacted with rabbit polyclonal antibodies for VSG221 and VSG1.8 (gift of P. Borst) or BiP (gift of Jay Bangs) (Bangs et al., 1993). Signal was visualised using an ECL Plus Western blotting detection system (Amersham).
Total T. brucei RNA was isolated using RNeasy RNA isolation kits (Qiagen). RNA was loaded on formaldehyde agarose gels, blotted onto Hybond-XL (Amersham) and probed using radiolabeled probes using standard procedures (Sambrook and Russell, 2001). Equivalent loading was determined using ethidium stained gels.
Pulsed Field gel analysis was performed using a CHEF-DRIII system (BioRad). DNA from approximately 2.5 × 107 cells was embedded in 1% low melting point agarose blocks. Resolution of megabase chromosomes was performed at 2.5V/cm for 144 hours, 16°C, 120° angle, 1400–700 second switching time on a 1.2% high strength agarose gel (Helena Biosciences) in 1× TBE0.1 (Melville et al., 2000). Intermediate chromosomes (100–500 kb) were separated at 6V/cm, 14°C, 120° angle, 25 second switching time, 20 hours, on a 1% high strength agarose gel in 0.5XTBE. Chromosomal DNA was blotted to Hybond-XL membrane (Amersham) and probed with radiolabeled probes.
We are very grateful to Peter Warne for performing pilot experiments, and to Waleed Mohammed for generating and analysing VSG switch variants. We thank George Cross for communicating information on VSGs. We are grateful to Jay Bangs and Piet Borst for generous amounts of antibodies. We thank Bill Wickstead, Michael Ginger, Keith Gull, James Minchin and Stephen Terry for stimulating discussions and very useful comments on the manuscript. J.S. was supported by a Natural Sciences and Engineering Research Council of Canada Post-Graduate Scholarship. K.S. was funded by a Wellcome Prize studentship. G.R. is a Wellcome Senior Fellow in the Basic Biomedical Sciences. This work was funded by the Wellcome Trust.