Monoallelic transcription of a single member of a multigene family is a phenomenon seen in several eukaryotes. In mammals, for example, a single olfactory receptor gene is selected from a family of ~1,000 [25
]. For several pathogenic protozoa, this mechanism is the basis for antigenic variation (reviewed in [26
]). For example, the human malaria parasite Plasmodium falciparum
transcribes only one of ~60 var
genes on the surface of infected red blood cells [27
]. To switch from one gene to another, most organisms use epigenetic transcriptional control mechanisms that are poorly characterized. In this study, we provide the first evidence that a well-characterized histone methyltransferase, DOT1B, affects antigenic variation in T. brucei
on three levels: maintenance of complete BES silencing, rapid transcriptional switching between BESs, and, as a consequence, VSG
We previously reported that RT-PCR detected no VSG derepression in ΔDOT1B cells [20
]. Here, we used a more sensitive approach, quantitative RT-PCR, to quantify the transcript levels of silent VSG
s. In the absence of DOT1B, silent BESs were 10-fold derepressed in single- and double-expressers, indicating that DOT1B is required for maximum repression of silent BESs. Importantly, disruption of DOT1B does not seem to affect the expression of all genes: no changes were detected in the transcript levels of non-BES genes β-tubulin
, which are transcribed by RNA polymerase II, or PAG3
, whose transcription by RNA polymerase I is partially repressed at this stage of the life cycle [22
]. This selectivity is consistent with what has been described in Saccharomyces cerevisiae
, in which loss of Dot1 only derepresses telomeric marker genes and—to a lesser extent—the HM loci [28
]. In yeast, this derepression is most likely an indirect consequence of the redistribution of Sir proteins throughout the genome [2
]. Although telomeric silencing has been described in T. brucei
], no causal proteins have been identified. Nevertheless, it is tempting to speculate that disruption of DOT1B causes a similar cascade of events in T. brucei
, which results in an indirect derepression of telomeric BESs but not of other loci.
Although silent BESs are partially derepressed in ΔDOT1B cells, the number of G418-resistant clones arising is similar to that of wild-type cells, indicating that partial derepression of a silent BES does not affect the BES switching frequency. This is consistent with previous observations showing that the rate-limiting step of a BES switch is silencing of the previously active BES rather than activation of a new one [30
]. After 1 wk of selection, all parental G418-resistant clones were puromycin-sensitive and exclusively expressed VSG13, indicating that BES1 was completely and rapidly silenced and BES17 was fully activated. In contrast, after the same time, most ΔDOT1B G418-resistant clones were resistant to both puromycin and G418 and expressed maximum levels of VSG221 and variable levels of VSG13, indicating that two BESs were simultaneously transcribed. To our knowledge, so far, this is the only gene whose disruption leads to the loss of monoallelic VSG
A previous study showed, by using a double-BES-tagged cell line similar to ours, that double-resistant clones could be obtained in which ~65% of the cells simultaneously expressed two VSGs at the surface [12
]. There are two important features that distinguish these cells from ΔDOT1B double-resistant clones. First, ΔDOT1B double-resistant clones are obtained by selecting for resistance to only one antibiotic (G418), whereas double-resistant wild-type clones could only be obtained by simultaneously selecting with two antibiotics. Second, ΔDOT1B double-resistant cells are stable and inheritable for at least 30 d in the absence of drug selection, whereas wild-type double-resistant clones rapidly lose double resistance and revert to a 50:50 mixture of trypanosomes with only one of the two marked BESs active. We propose, however, that both studies isolated a comparable switching intermediate, but whereas this is a transient state in wild-type cells, it is stable and inheritable in the absence of DOT1B.
On the basis of FACS and quantitative RT-PCR, we were able to quantify the activities of BES1 and BES17 in switching intermediates. Whereas BES1 remained 100% active, BES17 showed variable expression levels. Interestingly, we never found ΔDOT1B double-expressers that expressed wild-type levels of VSG13, indicating that, although the entire BES17 is transcribed, the level of transcription is not maximal. These results suggest an important VSG
regulation mechanism that strictly prevents two BESs from being fully transcribed but that is permissive to one BES being fully active and another BES being partially active. We do not know how such a limitation is imposed, but it is possible that it relies on the competition for a limiting transcription factor present in the ESB [10
]. This nuclear compartmentalization hypothesis is consistent with our RNA polymerase I localization studies. We observed no difference in the number of RNA Pol I sites between wild-type and ΔDOT1B cells, suggesting that there is a spatial restriction of Pol I transcription sites in the nucleus. We speculate that BES17 is expressed either in the nucleolus or in the unique ESB, close to BES1. This model is consistent with what was observed in double-resistant wild-type clones, in which the two rapidly switching active BESs often were found in close proximity in the nucleus [12
], and with the localization of a partially active BES at the periphery of the nucleolus [32
]. Despite the localization in a Pol I competent compartment, BES17 is only partially activated, which suggests that another limiting factor is missing. We cannot exclude the possibility that the levels of expression of BES17 are too low to allow the detection of an additional RPB6z site, but this seems unlikely because, for these experiments, we chose ΔDOT1B double-expressers where at least 40% of the cells were displaying 40–50% of maximal VSG13 expression. If BES17 is transcribed by half the amount of Pol I complexes than BES1, then we should still be able to detect a bright spot.
After 8–24 d of continuous G418 selection, switching intermediates eventually express only VSG13, suggesting that the competition for a limiting factor is resolved in some double-resistant cells. Consequently, this leads to the rapid silencing of BES1 and full activation of BES17. BES switching is most likely a complex mechanism that involves several events in a precise order. Nothing is known about the intermediates of a BES switch, which means that we cannot exclude the possibility that double-resistant cells are not a switching intermediate but simply a fraction of the population selected by drug pressure. If the latter was true, then one would predict that the number of G418-resistant clones would be higher in ΔDOT1B because the drug would select not only for VSG13 expressers but also for the fraction of cells with higher transcription levels in BES17. Also, if drug double-expressers were a product of drug selection, then, when pressure is removed, one would predict that the double-resistance phenotype should be lost, as demonstrated by [12
] and discussed above. None of these predictions is consistent with our data, suggesting that double-expressers are more likely switching intermediates and, consequently, DOT1B is involved in the kinetics of switching.
The VSG profile detected by FACS indicates that ΔDOT1B double-expressers have the same amount of VSG221 molecules at the surface as a VSG221 expresser and an extra 5–90% of the total VSG13 usually present in a VSG13 expresser. As VSG molecules form a tightly packed surface coat, we speculate that the volume of the cells might be slightly larger to accommodate the excess of VSG molecules. A similar observation was made previously, when two VSGs were simultaneously expressed at the same levels as one normally is [33
]. By means that we can only speculate upon, the cells seem able to accommodate up to twice as much VSG expression.
Several studies suggest that silencing of a previously active BES and activation of a new BES are tightly coupled (reviewed in [34
]), which predicts the existence of a sensing mechanism that facilitates cross-talk between BESs. It remains unclear what step of switching is affected in ΔDOT1B: activation of BES17 or silencing of BES1. If DOT1B is a transcription activator, as shown in cancer cells [7
], then its absence may interfere with BES17 activation. In this scenario, double-resistant cells are a consequence of a defective BES17 activation that leads, due to incomplete cross-talk between BESs, to BES1 remaining fully active. Alternatively, the effects of DOT1B depletion may be an indirect consequence of a disruption of telomeric silencing. In this case, the absence of DOT1B may interfere with silencing of BES1, and by cross-talk, a second BES is temporarily prevented from being fully activated. Interestingly, no population of ΔDOT1B double-expressers ever was found with intermediate levels of VSG221, indicating that, even in the absence of DOT1B, BES1 either remains fully active or becomes fully silenced. These results indicate that DOT1B is not implicated in the all-or-nothing property of the machinery that silences a previously active BES.
BES switching probably requires the ordered recruitment of complex machinery that recognizes a specific chromatin configuration and remodels it to establish a new chromatin structure in the two involved BESs. In this study, we showed that a histone-modifying enzyme is part of this machinery, which unveils the importance of chromatin modifications in antigenic variation of African trypanosomes.