Bloodstream Form-specific Up-regulation of Silent VSG Expression Sites and Procyclin in Trypanosoma brucei after Inhibition of DNA Synthesis or DNA Damage*

The African trypanosome Trypanosoma brucei transcribes the active variant surface glycoprotein (VSG) gene from one of about 20 VSG expression sites (ESs). In order to study ES control, we made reporter lines with a green fluorescent protein gene inserted behind the promoter of different ESs. We attempted to disrupt the silencing machinery, and we used fluorescence-activated cell sorter analysis for the rapid and sensitive detection of ES up-regulation. We find that a range of treatments that either block nuclear DNA synthesis, like aphidicolin, or modify DNA-like cisplatin and 1-methyl-3-nitro-1nitrosoguanidine results in up-regulation of silent ESs. Aphidicolin treatment was the most effective, with almost 80% of the cells expressing green fluorescent protein from a silent ES. All of these treatments blocked the cells in S phase. In contrast, a range of toxic chemicals had little or no effect on expression. These included berenil and pentamidine, which selectively cleave the mitochondrial kinetoplast DNA, the metabolic inhibitors suramin and difluoromethylornithine, and the mitotic inhibitor rhizoxin. Up-regulation also affected other RNA polymerase I (pol I) transcription units, as procyclin genes were also up-regulated after cells were treated with either aphidicolin or DNA-modifying agents. Strikingly, this up-regulation of silent pol I transcription units was bloodstream form-specific and was not observed in insect form T. brucei. We postulate that the redistribution of a limiting bloodstream form-specific factor involved in both silencing and DNA repair results in the derepression of normally silenced pol I transcription units after DNA damage.

African trypanosomes including Trypanosoma brucei evade immune attack during chronic infections by periodically switching a variant surface glycoprotein (VSG) 1 coat (1)(2)(3)(4). The predominant VSG is encoded by a gene transcribed from 1 of about 20 telomeric VSG expression sites (ES). Switching VSG coats is mediated by DNA rearrangements moving a new VSG into the active ES from a repertoire of hundreds of silent VSG genes and pseudogenes in chromosome internal tandem arrays or at telomeres. Alternatively, a switch can be mediated by a transcriptional switch between ESs.
The large polycistronic ES transcription units contain an assortment of expression site-associated genes (ESAGs) in addition to the telomeric VSG (reviewed in Ref. 5). Although the basic ES structure is conserved, there is variation in exactly which ESAGs are present, as well as their number and order (6). Switching between the polymorphic ESAGs present in different ESs can allow the trypanosome to adapt to life in different hosts. This has been best investigated for the polymorphic ESAG6 and -7 genes encoding transferrin receptor subunits (7)(8)(9) and the serum resistance-associated gene conferring human serum resistance (10,11). The polycistronic ESs are regulated as domains flanked upstream by extensive arrays of 50-bp repeats (12). Exogenous promoters integrated upstream of the 50-bp repeat arrays escape the transcriptional control operating on the downstream ES (13). In bloodstream form T. brucei, ES control is not sequence-specific (12,14), does not appear to involve a repressed chromatin state (15,16), or require homologues of genes involved in yeast telomere position effect, despite having a superficial resemblance to this phenomenon (17).
The multiple ESs are transcribed in a mutually exclusive fashion (reviewed in Ref. 18). Stable maximal activation of two ESs does not appear possible (19). Selection for double ES expressors using selectable markers inserted immediately downstream of different ES promoters results in trypanosomes that have one ES maximally active and another ES partially up-regulated, or that rapidly switch between two different ESs (19,20). Possibly these rapidly switching trypanosomes alternate between a privileged subnuclear location that has been called an expression site body (ESB), a pol I transcriptional body specific to bloodstream form T. brucei (21). Location in an ESB transcription/RNA processing factory may give the active ES access to limiting factors necessary for transcription elongation and polyadenylation (22).
In contrast, in insect form T. brucei all ESs appear to be down-regulated to a great extent, although not as tightly as in the bloodstream form (23). Silencing is mechanistically different, as it is promoter sequence-specific and appears to involve repressed chromatin (15). However, it is unlikely that the telomeric location of ESs is critical for silencing in this life cycle stage, as ES promoters on circular bacterial artificial chromosomes (BACs) are as effectively silenced as those in the genome (24).
In order to investigate the machinery mediating ES downregulation in bloodstream form T. brucei, we constructed reporter T. brucei strains with GFP inserted immediately behind the promoter of silenced ESs. We attempted to disrupt the silencing machinery using various chemical treatments. Fluorescence-activated cell sorting (FACS) analysis allowed rapid and sensitive detection of ES derepression. We find that treatments inhibiting DNA synthesis or causing DNA damage result in a block in S phase and a concurrent up-regulation of silent ESs. In addition, up-regulation of procyclin transcript was also observed. Procyclin is transcribed from RNA polymerase I transcription units that are normally down-regulated in the bloodstream form. Surprisingly, this up-regulation of transcripts from normally silent sites is life cycle-specific, as there was no evidence for up-regulation of silenced ESs in insect form T. brucei after comparable treatments. We postulate the presence of a limiting bloodstream form-specific factor involved in both silencing and DNA repair. Redistribution after DNA damage could result in the derepression of normally silenced pol I transcription units.

EXPERIMENTAL PROCEDURES
T. brucei Transformants and Culturing Conditions-All trypanosomes used were T. brucei 427, and bloodstream form variants were all derived from T. brucei 427 VSG221a (25). Bloodstream form T. brucei was maintained at 37°C in HMI-9 medium with the addition of 10% fetal calf serum and 10% Serum Plus (JRH Biosciences) (26). In order to ensure population homogeneity of the VSG coat expressed, trypanosome transformants with drug markers in the active VSG ES were continuously maintained on the appropriate drug selection pressure to prevent ES switching. The T. brucei RP2X-1 transformant was described previously (14). T. brucei HNI has a hygromycin gene inserted immediately behind the promoter of the 221 ES and a neomycin gene inserted behind the promoter of the VO2 ES and was described previously (27).
T. brucei 221GP1(221ϩ) has a GFP gene in the active 221 ES and was made by replacing the hygromycin gene in the active 221 ES of T. brucei HNI(221ϩ) with the 221GP1 construct containing GFP and the puromycin resistance gene. These trypanosomes were subsequently selected on G418 to select for trypanosomes that had activated the neomycin gene marked VO2 ES and silenced the GFP, resulting in T. brucei 221GP1(VO2ϩ).
T. brucei VO2GP1(VO2ϩ) has GFP in the active VO2 ES and was made by modifying T. brucei HNI(VO2ϩ) by replacing the neomycin gene in the active VO2 ES with the VO2GP1 construct containing the GFP and blasticidin genes. These trypanosomes were subsequently selected on hygromycin to select for trypanosomes that had activated the hygromycin gene marked 221 ES and silenced the VO2 ES producing T. brucei VO2GP1(221ϩ). The hygromycin resistance gene was subsequently replaced by a gene encoding puromycin resistance to create T. brucei VO2GP2(221ϩ).
The VO2GP1 construct was analogous to the 221GP1 construct, and only the puromycin resistance gene was replaced by a blasticidin resistance gene from construct tubBSRtub (gift of P. Borst, Netherlands Cancer Institute, Amsterdam), and the target fragments were amplified by PCR from the N19B2 BAC containing part of the VO2 ES (Gen-Bank TM account number AL671256) (6). The 5Ј target fragment was amplified using the HNES481s and VO228551as primers listed above. The 3Ј target fragment was amplified using primers VO2-118336s 5Ј-CTCTAGTGAGCGTATTTTAGAGG-3Ј and HNES 4244as. T. brucei Transformation-Transfection was performed using midlog phase bloodstream form trypanosomes, which were washed and then resuspended at 5 ϫ 10 7 ml Ϫ1 in cytomix without glutathione or ATP (30). 2.5 ϫ 10 7 cells were electroporated with 10 g of linearized DNA using a Bio-Rad Gene Pulser II with a single pulse of 1.5-kV and 25-microfarad capacitance in 0.2-cm cuvettes (Bio-Rad). After ϳ6 h recovery in HMI-9 medium, cells were distributed over 24-wells plates at densities between 7 ϫ 10 4 and 3 ϫ 10 5 cells ml Ϫ1 . Selection was with 0.2 g ml Ϫ1 puromycin (Sigma) or 2 g ml Ϫ1 blasticidin (Invitrogen). After 5-7 days on selection transformants were expanded for further analysis.
ES Switching-ES switching was performed using drug selection on agarose plates (31). T. brucei 221GP1(221ϩ) with the puromycin gene in the active 221 ES was maintained in the presence of 0.2 g of puromycin ml Ϫ1 (Sigma). To switch to the VO2 ES, 1 ϫ 10 5 to 1 ϫ 10 7 trypanosomes were spread on agarose plates containing 5 g of G418 (Invitrogen) ml Ϫ1 . Colonies were counted after 9 -10 days to determine the switching frequency (average of 4.4 ϫ 10 Ϫ6 determined on three plates) and analyzed by PCR to check for retention of the previously active 221 ES (27). T. brucei VO2GP1(VO2ϩ) was switched in an analogous fashion; only these trypanosomes were maintained on 2 g of blasticidin ml Ϫ1 and selected on 5 g of hygromycin ml Ϫ1 to select for activation of the 221 ES. The frequency of 221 ES activation was 3 ϫ 10 Ϫ6 (average of four plates). Again PCR was used to monitor for retention of the VO2 ES.
T. brucei Chemical or DNA Damage Treatments-Mid-log phase cultures were pelleted and resuspended in fresh medium to remove drug selection pressure. Bloodstream form T. brucei was resuspended at 5 ϫ 10 5 cells ml Ϫ1 , and insect form T. brucei was resuspended at 5 ϫ 10 6 cells ml Ϫ1 . Cultures were treated with the appropriate chemical for 24 h for bloodstream form T. brucei unless stated otherwise in the figure legends or 48 h for insect form T. brucei. Alternatively, bloodstream form T. brucei was exposed to UV or ␥-radiation and then recovered for 24 h. Aphidicolin, MNNG, berenil, pentamidine, suramin, trichostatin A, difluoromethylornithine (DFMO), and pCPT-cAMP were all from Sigma, and ethidium bromide from BDH. Cisplatin was a gift of Adrian Begg and Ben Floot (Netherlands Cancer Institute, Amsterdam) and rhizoxin a gift of Keith Gull.
Growth curves were initially performed with each of the different substances, and concentrations were chosen that flattened cell growth in a comparable fashion. Concentrations used are as follows: aphidicolin UV irradiation was performed with 3840 J/m 2 UV light (254 nm) using a UV cross-linker (Stratagene). Cells were in 25 ml of tissue culture dishes with the lid removed during exposure (adapted from Ref. 39). ␥-Irradiation was performed using a 137 Cs source (Gravitron RX30/ 55M, Graviner Manufacturing, Gosport, Hampshire, UK). Trypanosomes in HMI-9 medium were placed at a proximity to the cesium source so that the dose rate was 4.4 gray/min for a total dose of 160 gray. After removal from the exposure to UV or ␥-radiation, cells were recovered for 24 h. The number of double-strand DNA breaks introduced into the T. brucei genome was estimated using agarose blocks containing T. brucei transfected with BACs of ϳ150 kb (24). Introduction of a single double-strand break into these large circles results in linearization. We estimated that 160 gray irradiation resulted in the introduction of ϳ370 double-strand breaks into the T. brucei genome.
FACS Analysis-Derepression of GFP was analyzed by FACS using a BD Biosciences FACSCalibur with an excitation wavelength of 488 nm. After treatment and recovery, cells were washed once in PSG (60 mM Na 2 HPO 4 , 3 mM NaH 2 PO 4 , 44 mM NaCl, 56 mM glucose) and resuspended at 10 6 cells ml Ϫ1 for analysis. Data were analyzed using CellQuest software version 3.3. To correct for fluorescence of the chemical used, T. brucei HNI, which lacks GFP, was also treated and analyzed by FACS, and any fluorescence in the FL-1 channel was subtracted from the final values. In addition, values were corrected by subtracting background fluorescence of untreated cells containing silenced GFP. 2 T. Isobe and G. Rudenko, unpublished data.
For cell cycle analysis cells were stained with propidium iodide as described previously (40). Briefly, after treatment and recovery, cells were fixed in 70% methanol (10 6 cells ml Ϫ1 ) and incubated at 4°C overnight. Cells were washed in cold phosphate-buffered saline, resuspended in 1 ml of phosphate-buffered saline containing 10 g ml Ϫ1 propidium iodide and 10 g ml Ϫ1 RNase A, and incubated at 37°C for 45 min. Detector FL2-A and an AmpGain value of 1.75 were used.
RNA Analysis-Total RNA was isolated from bloodstream form T. brucei treated for 24 (bloodstream form) or 48 h (procyclic form) with various chemicals in the absence of drug selection pressure. RNA was isolated from ϳ3 to 5 ϫ 10 7 cells using an RNeasy RNA isolation kit (Qiagen). 5 g of total RNA was electrophoresed in formaldehyde gels and blotted according to Ref. 41. Northern blots were hybridized with probes radiolabeled by random priming using the Megaprime DNA labeling system (Amersham Biosciences). The probe for eGFP is a 719-bp SalI/NotI fragment from the eGFP plasmid (Clontech). The probe for the 221 VSG is an 800-bp fragment of the 221 VSG corre-sponding to positions 122-925 in the sequence (GenBank TM account number X56762). The probe for the VO2 VSG is a 600-bp EcoRI/HindIII fragment from the VO2 VSG cDNA. 3 The tubulin probe is a 700-bp HindIII/EcoRI fragment. The procyclin probe is the entire CPT4 cDNA (42). The procyclin probe can be expected to hybridize with transcript from both the EP and GPEET procyclin variants. Quantitation was performed with a Bio-Rad PhosphorImager.

DNA Damage Results in the Up-regulation of GFP Marked
Silent ESs-We constructed T. brucei strains with GFP immediately downstream of the promoter of the inactivated 221 or VO2 VSG ESs. As the ES promoters were silent, these strains allowed us to assay for disruption of the ES silencing machin-3 K. Sheader and G. Rudenko, unpublished data.  neomycin gene located downstream of the silent VO2 VSG ES promoter, selection with G418 allowed us to screen for reactivation of the VO2 VSG ES, producing the reporter strain T. brucei 221GP1(VO2ϩ) (Fig. 1B). T. brucei expressing GFP from an active VSG ES is more than 1000-fold brighter than background, allowing it to be very easily detected (Fig. 1C). FACS analysis of the T. brucei 221GP1(VO2ϩ) trypanosomes with a silenced GFP provided a very sensitive and rapid assay system for measuring a change in expression of the silent VSG ES.
We incubated the T. brucei 221GP1(VO2ϩ) reporter strain with a range of chemicals in order to screen for those resulting in the up-regulation of silent VSG ESs. First, we tested the histone deacetylase inhibitor trichostatin A, which generally causes derepression of silenced transcription units in a wide range of experimental systems due to disruption of repressed chromatin (reviewed in Ref. 43). Although trichostatin A suppressed growth of bloodstream form T. brucei, it had no detectable effect on derepression of the silent 221 VSG ES (result not shown). As there is no evidence that silenced ESs in blood-stream form T. brucei have a more repressed chromatin structure than active ESs (15,16), this result is not surprising. Although trichostatin A did not cause up-regulation of silent ESs, a range of treatments resulting in DNA modification or damage did (Fig. 2).
Effective treatments that resulted in high levels of up-regulation of normally silent VSG ESs included DNA-modifying agents. Cisplatin treatment results in the introduction of a bulky cisplatin adduct into DNA (44), and MNNG methylates DNA (45). Treatment of bloodstream form T. brucei with both of these chemicals resulted in up to 40% of the cells expressing GFP (Fig. 2). Treatment with UV light also resulted in DNA modification in the form of thymidine dimers. This treatment was also effective at producing the up-regulation of GFP, as was the intercalating agent ethidium bromide. DNA damage in the absence of DNA modification was also effective in inducing up-regulation of GFP transcript. The introduction of doublestrand DNA breaks with a cesium source also resulted in very high levels of silent VSG ES up-regulation, with more than 50% of the cells expressing GFP (Fig. 2).
In contrast, berenil and pentamidine selectively induced double-strand breaks in the kinetoplast DNA rather than nuclear DNA (34). Treatment with these drugs did not result in significant ES derepression. Treatment with the metabolic inhibitor suramin (46) or the ornithine decarboxylase inhibitor DFMO (35, 47) also had no detectable effect, despite inhibiting T. brucei cell growth in a comparable fashion to the other treatments (see "Experimental Procedures" for details). It is therefore likely that modification or damage of nuclear DNA is the key factor in the up-regulation of silent ESs that we observed.  In order to block nuclear DNA synthesis without introducing DNA damage, we incubated bloodstream form T. brucei with aphidicolin, which selectively binds and inhibits nuclear DNA polymerase ␣ (49, 50). Aphidicolin has been used previously to stall trypanosomes at the G 1 /S phase transition (32,51). Treatment of bloodstream form T. brucei 221GP1(VO2ϩ) with 30 M aphidicolin resulted in very striking up-regulation of the silenced GFP (Fig. 2). Almost 80% of the cells expressed GFP, with almost 20% of the cells expressing GFP at maximal levels (dark bar in Fig. 2) (see Fig. 1C for description of the gates used).

Up-regulation of Silent ESs Correlates with a Block in S
As up-regulation of silent ESs was observed after treatments that resulted in a block in nuclear DNA synthesis, we investigated if this could also be observed if T. brucei was stalled in the G 1 /G 0 phase of the cell cycle. DNA synthesis is inhibited in stumpy form T. brucei, which is a nonreplicating form stalled in G 1 /G 0 . pCPT-cAMP can induce this cell cycle block, also in the monomorphic T. brucei 427 strain used in this study (36,52). Although incubation of T. brucei 221GP1(VO2ϩ) in 1 mM pCPT-cAMP completely blocked cell growth and resulted in a block in DNA synthesis as measured by monitoring BrdUrd incorporation (result not shown), there was no significant ES derepression (Fig. 2). This indicates that stalling in S phase could be a critical factor, rather than simply blocking DNA synthesis.
Finally, we tested the effect of the mitotic inhibitor rhizoxin. Silenced telomeres in yeast are present in clusters at the nuclear periphery, and activation involves movement away from the perinuclear region (53,54). The microtubule polymerization inhibitor rhizoxin inhibits mitosis in trypanosomes (32). Treatment of procyclic T. brucei with rhizoxin has been shown to result in the disruption of telomere clusters at the nuclear periphery (55). However, treatment of our T. brucei GFP reporter strain with rhizoxin did not result in significant upregulation of the ES (Fig. 2).
This up-regulation of a normally silent ES was not specific to the 221 VSG ES. We constructed the reporter line T. brucei VO2GP2(221ϩ), where a construct containing the blasticidin gene linked to GFP was downstream of a silenced VO2 VSG ES promoter (Fig. 3A). As expected, treatment of this cell line with either aphidicolin or cisplatin resulted in significant up-regulation of this ES, which was not observed after treatment with pCPT-cAMP (Fig. 3B). Up-regulation of the 221 VSG ES in T. brucei 221GP1(VO2ϩ) increased as the amount of aphidicolin used for treatment was raised between 0 and 30 M (Fig. 4, A and B). In a time course using 30 M aphidicolin, GFP expression was extensive at 12 h (60% of the cells GFP-positive) and essentially maximal at 24 h (about 80% of the cells GFP-positive) (Fig. 4C). As the doubling time of the T. brucei 221GP1(VO2ϩ) cells is about 7 h, the extensive up-regulation observed at 12 h is compatible with most of the ES derepression occurring during one cell division. The percentage of cells scoring as maximally GFP-positive increased between 12 and 24 h, presumably due to accumulation of GFP protein. We attempted washing aphidicolin-treated cells to see if treatment was reversible, and if high levels of up-regulation of silent ESs resulted in increased ES switching. However, we were not able to recover viable cell lines from cells treated with aphidicolin at concentrations that resulted in significant up-regulation.
As expected for an inhibitor of nuclear DNA synthesis, incubating T. brucei 221GP1(VO2ϩ) in increasing amounts of aphidicolin resulted in a block in S phase (Fig. 5A). An accumulation of cells in S phase was detectable using even small amounts of aphidicolin (0.5-1 M) (Fig. 5, A and B). As the amount of aphidicolin used for treatment increased, the block became increasingly tighter, resulting in cells that don't enter S phase but stall at the G 1 /S phase transition as observed previously (32,51). The G 2 /M peak disappeared at these increasing concentrations, presumably as cells that had passed S phase progressed further through the cell cycle, accumulating in G 1 . Treating T. brucei with DNA-modifying agents like cis-FIG. 6. Cell cycle arrest in S phase or in the G 1 /S phase transition in T. brucei exposed to treatments that damage DNA or block DNA synthesis. T. brucei HNI was treated for 24 h with the substances indicated with the exception of ␥-irradiation, where cells were treated with 160 gray and allowed to recover for 24 h. Cells were stained with propidium iodide to measure DNA content. 20,000 cells were analyzed in each experiment. The figure is further labeled according to Fig. 5A. WT, wild type. platin and MNNG also resulted in an accumulation of cells in S phase (Fig. 6). This is presumably because the T. brucei DNA damage repair machinery recruited to sites of DNA damage triggers an S phase cell cycle checkpoint, as has been described in other organisms (reviewed in Refs. 56 and 57).
Bloodstream Form-specific Partial ES Transcription-Upregulation of the silenced 221 VSG ES resulted in transcript levels of about 20 -30% of an active ES, but transcripts did not extend as far as the 221 VSG itself (Fig. 7). Possibly, stalled replication forks accumulating in the treated cells interfere with transcription of the large (ϳ60 kb) 221 VSG ES (6,39,58). Down-regulation of VO2 VSG transcript from the active ES itself was also observed (Fig. 7). Unexpectedly, there also appeared to be general up-regulation of other repressed RNA polymerase I transcription units. Not only VSG ESs but also procyclin, which is down-regulated in bloodstream form T. brucei, was greatly up-regulated (Fig. 7). Procyclin mRNA levels were up-regulated to about 30 -50% of the normal level in procyclic trypanosomes (42, 59 -60). As our probe recognizes both the EP and GPEET procyclin variants (see "Experimental Procedures"), we do not know if one of these procyclin variants was preferentially up-regulated. Amounts of tubulin transcript were slightly decreased after treatment (Fig. 7). Although VSG ESs are polymorphic, their global architecture is roughly similar (6). In Northern blot hybridizations using probes extending along the VSG ES transcription unit, transcripts appeared to be particularly up-regulated near the ES promoter. For example, the transcript for ESAG6/7 and ESAG5 was up-regulated but not for ESAG3 (Fig. 8). Because the ESAG probes used were not specific for the ESAGs in the 221 ES, we were not able to determine from which up-regulated ESs these transcripts were derived.
Surprisingly, this up-regulation of repressed RNA polymerase I transcription units after DNA modification or inhibition of DNA synthesis was specific to bloodstream form T. brucei and was not observed in procyclic trypanosomes. ESs are all downregulated to a great extent in insect form trypanosomes, although repression is mechanistically different to the bloodstream form (15,23,61). Procyclic T. brucei transformants containing a hygromycin gene behind an ES promoter (ESX1-1 in Ref. 23) or an rDNA promoter in an ES (RPX1-1 in Ref. 23) were treated with aphidicolin, cisplatin, and MNNG. No upregulation of these marker genes was observed after these chemical treatments (Fig. 9A). Similarly, when these T. brucei transformants plus a transformant containing an ES promoter in the rDNA spacer (rDES1-1 in Ref. 23) were incubated in 30 M aphidicolin, no up-regulation of these marked promoters was observed (Fig. 9B) despite evidence for the induction of a stringent block in G 1 /S as has been observed previously (32) (Fig. 9C). DISCUSSION We show that treatment of bloodstream form T. brucei with agents that cause a block in nuclear DNA synthesis or DNA damage results in the up-regulation of silenced VSG ESs. The extent of up-regulation increased with the amount of aphidicolin used and correlated with the tightness of the S phase block. Up-regulated silent ESs were only partially transcribed, with transcription extending past ESAG5 but terminating before the telomeric VSG. In addition to inactive ESs, other RNA polymerase I transcription units like procyclin, which are normally silenced in bloodstream form T. brucei, were also upregulated. In contrast, no up-regulation of silent ESs was observed in procyclic T. brucei after treatments resulting in a comparable block in S phase.
In our experiments, we find the most likely scenario is that normally repressed pol I transcription units have become derepressed, resulting in increased pol I transcription of normally silent locations. Formally, we have not shown that the upregulated VSG ESs and procyclin loci are transcribed by pol I rather than pol II. However, we find it implausible that the promiscuity of pol II initiation in T. brucei would necessarily increase after treatments resulting in a block in S phase.
One initial explanation for the general up-regulation of normally silenced ESs was that the treatments used had severely perturbed the bloodstream form nuclear architecture. In bloodstream form T. brucei, RNA polymerase I is normally present in high concentrations in the nucleoli and the ESB (21). Blocking DNA synthesis could have resulted in the disintegration of these subnuclear compartments, resulting in the transcription of sequences that might normally not have access to pol I. There was some disruption of the normal nuclear architecture after aphidicolin treatment, in in vitro transcription experiments with BrUTP, and ␣-amanitin in permeabilized cells (re- sults not shown) (21,62,63). However, ␣-amanitin-resistant transcription foci corresponding to the nucleolus and the ESB could still be detected in bloodstream form T. brucei treated with aphidicolin (result not shown). Although transcription of up-regulated ESs was likely to be taking place outside of the ESB, experimental sensitivity was not high enough to establish this unequivocally.
Partial activation of ESs in bloodstream form T. brucei has been achieved before in experiments targeting drug resistance genes into "silent" ESs, although the levels of transcription were much lower than we show here (19,64). These low levels of transcription were "erased" after activation and subsequent inactivation of the partially active ES, implying the resetting of an epigenetic state (19). In addition, Ulbert et al. (20) investigated ES control in T. brucei strains with three ESs marked with drug resistance genes. Different combinations of drug selection pressure resulted in either of the following scenarios: an unstable rapidly switching state existing between any two of three marked ESs. Alternatively, clones arose with one maximally active ES and other ESs partially active at a very low rate. However, by using MNNG treatment, novel T. brucei lines were isolated that had higher partial activation of two different marked ESs in addition to the active ES. Levels of derepression were about 10% compared with the active state.
Finally, derepression of silent ESs was shown in bloodstream form T. brucei treated with BrdUrd, a thymidine analogue that cannot be converted to the modified nucleotide J (65). ES derepression was interpreted as a consequence of reduction in J, as incubation with BrdUrd resulted in a 12-fold decrease in the levels of J. However, overproduction of J by incubation with hydroxymethyldeoxyuridine also resulted in derepression of silent sites, making it possible that perturbance of the DNA structure was the critical factor rather than the amount of J itself (65). Although incubation of our T. brucei GFP reporter strain with up to 400 M BrdUrd resulted in silent ES upregulation, levels were not as striking as with the other treatments shown here (results not shown).
Silencing in different experimental organisms has been shown to require transition through S phase of the cell cycle for the establishment of silenced chromatin (66 -69). However, although this silencing is S phase-dependent, it does not always require DNA replication itself (70 -72). The common feature of all of the treatments we performed that resulted in up-regulation of silent ESs is that they caused either a block in nuclear DNA synthesis by inhibiting DNA polymerase ␣ like aphidicolin, or resulted in DNA damage, which typically triggers an S phase checkpoint, as has been extensively documented in other organisms (56,57). In T. brucei there are life cycle-specific differences in cell cycle checkpoints. Procyclic trypanosomes can undergo cytokinesis despite a block in S phase or mitosis, but in bloodstream form T. brucei inhibition of mitosis inhibits cytokinesis but not further rounds of DNA replication (32,40). No cyclin homologue has yet been identified in T. brucei that is involved in S phase transition in the bloodstream form. We find that incubation with even very low amounts of aphidicolin resulted in cells starting to accumulate in S phase. Stalled replication forks are a potent signal for S phase checkpoint activation, which blocks progression through the cell cycle allowing DNA damage to be repaired (discussed in Ref. 56). However, the amount of ES up-regulation that we observed increased with the concentration of the DNA polymerase inhibitor aphidicolin used for treatment. We therefore find it unlikely that the upregulation we observe is a direct consequence of stalling in S phase, and is more likely to be correlated to the number of stalled replication forks, which would be expected to increase as the concentration of aphidicolin increases. We cannot exclude that the stalled replication forks impact on RNA polymerase processivity. It is unknown if T. brucei RNA polymerases fall off at stalled forks, making themselves subsequently available for reinitiation at normally silent promoters.
In contrast, we prefer a model whereby a limiting factor involved in both silencing and DNA repair has been redistributed in our treated bloodstream form T. brucei, resulting in derepression. There is precedent for chromatin proteins involved in transcription repression and present in limiting con- centrations being relocalized to sites of DNA damage in the cell, resulting in derepression of normally silent transcription units. For example, the Ku and SIR silencing proteins are normally sequestered at telomeres in Saccharomyces cerevisiae (54,73,74). However, after the introduction of DNA damage, these proteins are relocalized from the telomeres to DNA repair sites (75,76), resulting in derepression of genes normally silenced by telomere position effect (77). A SIR2 homologue has been described recently in T. brucei (78) that plays a role in DNA repair and is enriched at telomeres. Although it appears unlikely that the telomere position effect plays a central role in ES down-regulation in bloodstream form T. brucei, one scenario is that the accumulation of stalled replication forks or DNA damage results in the relocalization of limiting silencing factors, resulting in derepression. This scenario would predict the presence of a bloodstream form-specific silencing factor, which is involved in both repression of silent ESs and procyclin as well as DNA repair and is present in limiting concentrations in the cell. This model would be compatible with the correlation observed between the degree of ES up-regulation and the amount of aphidicolin used for treatment.
The recruitment of DNA repair machinery to sites of DNA damage in repressed genes might have a secondary consequence of making these areas more accessible for transcription after the damage has been repaired. For example, treatments resulting in DNA damage or a block in DNA synthesis could result in the recruitment of DNA damage repair machinery to silenced sites, resulting in the displacement of factors blocking elongation of transcription. Blocking DNA synthesis with aphidicolin produces an accumulation of stalled replication forks and single strand DNA, which might also recruit proteins that are part of the DNA damage repair machinery.
A less likely scenario is that factors recruited to sites of DNA damage result in the stimulation of transcription. Some core factors have been shown to be involved in both transcription and DNA repair. Transcription factor TFIIH is a multisubunit complex that has been shown to be involved in RNA polymerase I and II transcription, DNA repair, and cell cycle control (reviewed in Ref. 79). TFIIH is not required for RNA pol I-mediated transcription initiation, but for a subsequent step like promoter clearance, transcription elongation, or reinitiation (80). Studies using GFP-tagged TFIIH showed that although TFIIH is normally involved in transcription, TFIIH rapidly relocates to repair sites after DNA damage (81). However, the argument that a factor involved in DNA repair stimulates transcription might be a bit farfetched, as it would presumably be more advantageous for a cell to induce silencing around the area of a DNA lesion rather than transcription. For example, the Ku and SIR proteins that are targeted to double-strand breaks are involved in transcription repression in yeast (75).
A striking feature of our results is that the up-regulation of normally silenced RNA polymerase I transcription units that we observe is bloodstream form-specific. We saw no evidence for increased transcription from down-regulated ESs in treated procyclic T. brucei. There are significant mechanistic differences in ES down-regulation in procyclic compared with bloodstream form T. brucei. In procyclic T. brucei there is evidence for a repressed chromatin state down-regulating ESs in a promoter sequence-specific fashion, which is absent in bloodstream form T. brucei (15,23). The bloodstream form-specific nature of the up-regulation that we observe makes it plausible that bloodstream form-specific factors involved in both DNA repair and transcription play a role in the ES control necessary to mediate antigenic variation.