Ku is important for telomere maintenance, but not for differential expression of telomeric VSG genes, in African trypanosomes.

Trypanosome antigenic variation, involving differential expression of variant surface glycoprotein (VSG) genes, has a strong association with telomeres and with DNA recombination. All expressed VSGs are telomeric, and differential activation involves recombination into the telomeric environment or silencing/activation of subtelomeric promoters. A number of pathogen contingency gene systems associated with immune evasion involve telomeric loci, which has prompted speculation that chromosome ends provide conditions conducive for the operation of rapid gene switching mechanisms. Ku is a protein associated with eukaryotic telomeres that is directly involved in DNA recombination and in gene silencing. We have tested the hypothesis that Ku in trypanosomes is centrally involved in differential VSG expression. We show, via the generation of null mutants, that trypanosome Ku is closely involved in telomere length maintenance, more so for a transcriptionally active than an inactive telomere, but exhibits no detectable influence on DNA double strand break repair. The absence of Ku and the consequent great shortening of telomeres had no detectable influence either on the rate of VSG switching or on the silencing of the telomeric promoters of the VSG subset that is expressed in the tsetse fly.

Trypanosome antigenic variation, involving differential expression of variant surface glycoprotein (VSG) genes, has a strong association with telomeres and with DNA recombination. All expressed VSGs are telomeric, and differential activation involves recombination into the telomeric environment or silencing/activation of subtelomeric promoters. A number of pathogen contingency gene systems associated with immune evasion involve telomeric loci, which has prompted speculation that chromosome ends provide conditions conducive for the operation of rapid gene switching mechanisms. Ku is a protein associated with eukaryotic telomeres that is directly involved in DNA recombination and in gene silencing. We have tested the hypothesis that Ku in trypanosomes is centrally involved in differential VSG expression. We show, via the generation of null mutants, that trypanosome Ku is closely involved in telomere length maintenance, more so for a transcriptionally active than an inactive telomere, but exhibits no detectable influence on DNA double strand break repair. The absence of Ku and the consequent great shortening of telomeres had no detectable influence either on the rate of VSG switching or on the silencing of the telomeric promoters of the VSG subset that is expressed in the tsetse fly.
The heterodimeric protein Ku, which consists of the subunits Ku70 and Ku80 (or Ku86), associates tightly in a sequenceindependent fashion with free ends of double strand DNA and has been associated with a range of nuclear functions in different eukaryotes (reviewed in Refs. [1][2][3], including DNA repair, retrotransposition (4), gene silencing, transcriptional reinitiation (5), chromosome maintenance, and chromosome localization. Recently, homologues possibly associated with DNA repair have been discovered also in prokaryotes (6,7). Ku is central to the non-homologous end joining (NHEJ) 1 pathway for DNA double strand break repair in which double strand breaks with dissimilar ends can be rejoined with minimal error (8). A main function of Ku is to bind to DNA and recruit proteins, such as the catalytic subunit of mammalian DNA-dependent protein kinase, that catalyze steps in the repair process, although a weak helicase activity has been ascribed by some to Ku itself (1). Ku also plays a major role in telomere length maintenance in yeasts such that the deletion of Ku genes causes shortening of the telomere tract (9 -11). This function of Ku again is thought to operate mainly through its interaction with other molecules (12,13). The NHEJ function of Ku is prevented from acting at telomeres by other telosome proteins such as Taz1 (14), and indeed the presence of Ku is required for the prevention of recombination at the free ends of telomeres (15,16).
In the phenomenon known as telomere position effect (TPE), telomere-proximal genes in yeast become transcriptionally silent in a reversible manner (17). The deletion of KU genes in Saccharomyces cerevisiae prevents the silencing (18), although the removal of Ku from Schizosaccharomyces pombe does not have this effect (11). Although TPE has been demonstrated where reporter genes have been inserted experimentally into telomeres (17,19), there is only one example of it occurring naturally. A screen of telomere-proximal genes on a number of S. cerevisiae chromosomes revealed that, although generally they are not subject to TPE, one does appear to be silenced in that way (20).
Pathogenic microorganisms commonly use contingency gene systems that, by undergoing random and rapid mutational events, create phenotypic diversity during infection and can enable evasion of host immune systems (21)(22)(23). Contingency genes often are members of large gene families of which only one is expressed at a time, and the mutational events can include a random switch between expression of individual members of the family. It is common for contingency genes, including the one that is expressed, to be located at telomeres (24 -26). Just why telomeres are so popular for such genes is unknown, but it has been speculated that the high rates of recombination between telomeres contribute to diversification within the family. It has also been proposed that telomeric location is more directly involved in phenotype switching by using effects such as TPE to enable transcriptional switching between members of the family (25,(27)(28)(29). The sleeping sickness parasite Trypanosoma brucei has a variant surface glycoprotein (VSG) protective coat that undergoes extensive antigenic variation during evasion of host immunity (reviewed in Refs. 23, 24, and 30 -32). When antibodies kill trypanosomes expressing a specific VSG, a minority of cells that have switched to expression of a distinct VSG survive and multiply. Each trypanosome has a family of probably hundreds of VSG genes, most of which are located in tandem arrays within chromosomes, but a sizeable minority of which are located at the many telomeres in this genome. To obtain exclusive expression of VSGs, trypanosomes transcribe them singly and only at telomeric sites known as bloodstream expression sites (BESs). Although there can be occasional transcriptional switches between BESs, the main route to VSG switching is the replacement of the VSG in the BES with a copy of a silent VSG (33), probably by homologous recombination (34). However, it is not known yet whether special, telomere-associated mechanisms enhance recombination into BESs, and a role for telomereassociated recombination-promoting proteins is a possibility. Ku, interacting with proteins other than those preventing recombination at the free end of the telomere, is a candidate.
It is also possible to invoke Ku, in its role as mediator of TPE, as a candidate for regulating differential expression and repression of the distinct set of VSGs employed when the VSG coat first appears in the metacyclic trypanosome stage in the salivary glands of the tsetse (Glossina) vector (reviewed in Refs. 29 and 35). In that stage, a subset of telomeric genes known as MVSGs becomes differentially activated at the transcriptional level, generating a population diverse for VSGs. As these are the only trypanosome protein-coding genes known to have their own promoters, and to be subject to absolute control by transcription initiation, it has been speculated that they have achieved such control through the operation of a TPE on the MVSG promoters, which are only ϳ5 kb from the telomere tract (29). Here, we examine the functions of Ku in trypanosomes and test the hypothesis that it is important for VSG switching in the bloodstream stage and that it is also important for the silencing of MVSG genes prior to the metacyclic stage.
Cloning and Sequencing T. brucei KU70 and KU80 Genes-Using detailed searching of the Sanger and TIGR trypanosome genome databases, we obtained entries recognized as putative KU70 and KU80 homologue short fragments (see "Results"). From these, we designed the following oligonucleotides: KU705Ј CCGAATTCGGGGAGGGTAC-GTCAGCGGG; KU703Ј CCAAGCTTCAAGCTCTCGATTTGGATAGC; KU805Ј GAATTCTTCCTCTGCGTGCCGTTGAC; KU803Ј GAATTCG-CACCTCTTCACCTACCGTC (EcoRI underlined, HindIII in bold). PCR was carried out with ϳ3 pmol of each primer with an initial 10-min denaturation at 95°C followed by 30 cycles of 95°C for 1 min, 65°C for 1 min, and 72°C for 1 min. Products of 248 bp (KU70) and 214 bp (KU80) were cloned into pBluescript (Stratagene). The sequence was determined on both strands using ABI-PRISM automated sequencing with custom-designed primers. An insert sequence was subsequently used to probe a T. brucei ILTat 1.2 genomic library (Sau3A partial digests in -GEM-12; gift of Nils Burman). This initially allowed the isolation of two DNA fragments for each gene believed to contain putative KU70 and KU80 sequence. Once again, these clones were sequenced on both strands by ABI-PRISM automated sequencing, yielding the open reading frames (ORFs) described under "Results." Generation and Analysis of T. brucei KU70 and KU80 Knockouts-Two constructs were used to delete each gene, using two rounds of transformation. Each construct contained ϳ400 bp of targeting sequence derived from sequences immediately upstream and downstream of either ORF, meaning that transformants delete the entire ORF. Targeting flanks for each gene were amplified by PCR using genomic DNA and Pfu DNA polymerase (Stratagene). ApaI and cloned into both pTBT and pTPT already containing the respective 5Ј-targeting flank. The final transformation constructs were digested with NotI and XhoI prior to being transformed into T. brucei, and transformants were selected with 10 g⅐ml Ϫ1 blasticidin or 1 g⅐ml Ϫ1 puromycin. Correct integration of cassettes into both KU loci was determined by Southern blot analysis. Genomic DNA from KU wild type and mutant cell lines was digested with EcoRI and separated on a 0.6% agarose gel. The gel was subsequently Southern-blotted onto Hybond XL (Amersham Biosciences) and probed with the same KU70 and KU80 ORFderived sequences used to probe the genomic library. Following hybridization, the blots were washed to a final stringency of 0.2ϫ SSC, 0.1% SDS at 65°C. A 12-kb fragment corresponding to intact KU70 was detected in both wild type and heterozygous cell lanes but was absent in the homozygous mutants. Two bands were present in KU80 wild type lanes due to allelic differences (ϳ9 and 15 kb). One of these bands was lost upon disruption of the first allele, whereas both were lost in ku80 homozygous mutants. Probing the blots with KU70 and KU80 5Ј-flanking sequence confirmed the correct integration of the BSR and PAC resistance cassettes. Again, the blots were washed to a final stringency of 0.2ϫ SSC, 0.1% SDS.
Re-expressing KU70 in the Homozygous Knockouts-Construct pCC101 was used to reintroduce KU70 to its original locus. The entire KU70 ORF, along with 620 bp of 5Ј-flanking sequence, was isolated as a 2.9-kb PCR-amplified fragment using the following primers: KU70RE-EXPFOR, 5Ј-GGCCAAGTTAACGCTTCAGCTGATGGTCGC-C-3Ј, KU70RE-EXPREV; 5Ј-GGATCGGTTAACGCGCAACCGAGGAG-GAAACC-3Ј (ApaI underlined). The fragment was amplified using a 10-min denaturation step followed by 30 cycles of 1 min at 95°C, 1 min at 55°C, and 6 min at 72°C. This was followed by 10 min at 72°C. This reaction was carried out using Pfu DNA polymerase, and the fragment was cloned into a unique HpaI restriction enzyme site in pNR101. This placed the KU70 sequence upstream of a construct with ϳ400 bp of actin intergenic sequence at the 5Ј flank of the bleomycin resistance gene and 330 bp of an ␣-␤-tubulin intergenic region at the 3Јend. In the homozygous mutants, the 620 bp of KU70 5Ј flank and ␣-␤ intergenic region act as targeting flanks, allowing KU70 to be reintroduced into the disrupted locus. The only potential difference in expression is that 3Ј processing is provided by the actin intergenic region rather than the natural KU70 3Ј signals. The pCC101 was linearized with NotI prior to electroporation, and transformants were selected on semisolid agarose plates containing 2.0 g⅐ml Ϫ1 bleomycin (Invitrogen). Integration of re-expression cassettes was tested by Southern blot analysis; genomic DNA was digested with EcoRI, separated on a 0.6% agarose gel, Southern-blotted, and probed with the KU70 ORF sequence. Each transformant had integrated the KU70 re-expression cassette, as expected. The control construct used in this analysis (pRM450) contains the bleomycin-resistant gene (BLE) surrounded by the same processing signals as in pTBT/pTPT and targets the tubulin array. This was cut with NotI and XhoI for transformation and targets the tubulin locus.
Generation and Analysis of Trypanosome VSG Switchers-The method used to measure frequencies at which the various cell lines switch their VSG coat has been described previously (34). Mice immunized against the VSG221 coat were injected with 5 ϫ 10 7 trypanosomes of the clones 3174.2 wild type (wt), 70Pac3.1ϩ/Ϫ, 70Pac4.2ϩ/Ϫ, 70Pac3.1BϪ/Ϫ, or 70Pac4.2BϪ/Ϫ. The mice were sacrificed by cardiac puncture 24 h later, and trypanosomes were isolated and cloned in 96-well plates as described previously (40). The number of switched clones that grew through was used to calculate the frequency of switching events and assumed an in vivo doubling time of 8 h.
Viability and Growth of ku70 Mutants in the Presence of DNA Damaging Agents-To test the ability of ku70 mutants to grow and divide in the presence of the DNA damaging agents methyl methanesulfonate (MMS; Sigma) or phleomycin (Cayla), a subclonal viability assay was adopted. Cell lines were grown to a maximum density of 2 ϫ 10 6 cells⅐ml Ϫ1 but more typically from 5 ϫ 10 5 to 1 ϫ 10 6 cells⅐ml Ϫ1 . Each cell line was then diluted to a concentration of 25 cells⅐ml Ϫ1 in HMI-9 with or without MMS or 50 cells⅐ml Ϫ1 for phleomycin. Each mutagen was titrated against wild type trypanosomes to determine the sublethal range. To assess the effect of MMS on growth, the diluted cells were spread over a 96-well culture dish (5 cells/well for MMS, 2 cells/well for phleomycin, minimum densities giving reproducible growth data), and the number of wells with growing trypanosome populations were counted after 7-9 days. In vitro growth rates were measured by diluting mid-log bloodstream trypanosomes to a concentration of 1 ϫ 10 5 ⅐ml Ϫ1 in 1.5 ml of HMI-9 medium in 24-well culture dishes. Cell concentrations were measured using a hemocytometer (Sigma).
Telomere Length Maintenance Assay-Wild type, heterozygotic, and homozygotic KU70 and KU80 cell lines were subcloned on semisolid agarose HMI-9 plates. Multiple clones for each were then grown to a density of 2 ϫ 10 6 trypanosomes⅐ml Ϫ1 , and genomic DNA was prepared and digested with EcoRI or AgeI depending on the expression site under analysis. EcoRI cuts 3 kb upstream of VSG221, which is located in the active expression site. AgeI cuts just within the 5Ј end of the coding sequence of the VO2 gene in the transcriptionally inactive expression site analyzed. Genomic DNA digests were separated on a 0.6% agarose gel, Southern-blotted onto Hybond XL (Amersham Biosciences), and probed with the amino terminus region of either VO2 or VSG221.
Analysis of Transcriptional Status of Metacyclic VSGs-Three previously characterized metacyclic VSG loci were used to analyze the transcriptional status of MVSGs in the procyclic stage. Total RNA from KU70 wild type, heterozygous, and homozygous mutant cell lines was prepared using TRIzol. cDNA was generated using random hexamers and Superscript II reverse transcriptase (following protocols from Invitrogen). Integrity of cDNA was tested using a control PCR on cDNA to amplify RNA polymerase I mRNA as described previously (41). Three characterized MVSG loci, MVSG1.22, 1.61, and 1.63, were examined by PCR for the presence of transcripts both proximal to the promoter and around the VSG. PCR was carried out on the cDNAs with ϳ3 pmol of upstream and downstream primers with an initial 10-min denaturation at 95°C followed by 30 cycles of 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min. These reactions were carried out using TaqDNA polymerase (Abgene Ltd.) for each locus except the MVSG1.61 promoter proximal region, where Pfu DNA polymerase was used with 2 mM MgSO 4 and a lower annealing temperature of 50°C. The PCR primers were designed as follows: MVSG1.22 promoter proximal, primer 5Ј ends were ϩ29 and ϩ214 from the transcription start site defined from metacyclic nascent transcripts, 2 and VSG coding sequence primer 5Ј ends were ϩ426 and ϩ666 from the ATG start codon, which is 2934 bp from the transcription start; MVSG1.61 proximal promoter ends were ϩ28 and ϩ287 from transcription start, and coding sequence ends were Ϫ5 and ϩ305 from ATG, which is 2783 bp from the transcription start; MVSG1.63 proximal promoter ends were ϩ59 and ϩ230 from transcription start, and coding sequence ends were ϩ621 and ϩ837 from ATG, which is 2988 bp from the transcription start.

Trypanosome Ku Homologues Are Not Colinear-As neither
Ku70 nor Ku80 has a high degree of interspecies similarity even at the polypeptide level (42), the T. brucei genome sequence databases were blast-searched individually with all known homologues. Although most hits were unique to each query sequence, some were shared by more than one. Detailed searching of databases, using Smith-Waterman algorithms, supported the view that these common hits did indeed correspond to Ku homologues. Using PCR-amplified fragments based on these sequences as probes, full-length trypanosome sequences were obtained by the isolation of overlapping genomic clones from our existing genomic library. Sequence analysis revealed complete open reading frames (trypanosomes are almost fully intronless) potentially encoding polypeptides of 81 kDa for the Ku70 homologue and 69 kDa for the Ku80 homologue. It was possible to align parts of those sequences with the primary homology regions (PHRs) of other Ku homologues by Clustal W analysis (Fig. 1). The predicted sizes of the trypanosome Ku proteins are the inverse of what occurs in other organisms, and their designation is complicated further by the general relatedness of Ku70 and Ku80 and the dispersion of homologous sequences throughout the proteins. Our belief that the larger protein corresponds to Ku70 relies on the outcome of blastp and PSI-blast searching, which routinely place it closer to Ku70 than Ku80 proteins, and the presence at its carboxyl-terminal of a partial putative DNA binding SAP domain (6,43) that has the secondary structure potential of other SAP sequences. Trypanosome Ku80 likewise detects mainly Ku80 sequences in blast searching. It does not have a recognizable DNA-dependent protein kinase binding sequence found at the end of vertebrate Ku80 proteins, but neither do the yeast Ku80 proteins.
Homology-based modeling using JPRED (jura.ebi.ac.uk:8888/) reinforced the similarity of the trypanosome sequences to known Ku proteins of other organisms and also suggested a likely reason for most of the extra length of the trypanosome Ku70. Essentially all ␣ helices and ␤ strands in the crystal structure of the human protein (44) are present in the predicted trypanosome Ku70 structure, except in the region of the ring that encircles DNA. Although in other species there are about 50 residues between the ␤G and ␤M strands, the trypanosome has about 120 amino acids. It is intriguing that trypanosome Ku80 has a typical 75 residues in the corresponding region, revealing an asymmetry between the two monomers in the ring region.
Null Mutants Have Normal Growth-To analyze the function of Ku, initially looking for general phenotypes described previously in other organisms and then questioning possible specific roles in VSG switching, we adopted the approach of deleting both alleles of either KU to generate null mutants. Genomic Southern analysis revealed that both KU70 and KU80 probably are single copy genes (data not shown), an interpretation subsequently confirmed during gene knockouts. Initially, we deleted genes in bloodstream stages of T. brucei S427 grown in vitro. In the 3174.2 strain, which is marked with the HYG and NEO antibiotic resistance genes in its transcriptionally active telomere, we deleted the first KU70 copy using the puromycin resistance cassette (PAC), obtaining the independent clones, 70P3.1(ϩ/Ϫ) and 70P4.2(ϩ/Ϫ). Both then had the second allele deleted with the blasticidin resistance cassette (BSR), generating the clones 70P3.1B(Ϫ/Ϫ) and 70P4.2B(Ϫ/Ϫ). A number of subclones of each of these were derived, enabling the study of events in individual sublineages. A similar approach was used for KU80, except that blasticidin resistance was used first, followed by puromycin. This generated the independent heterozygotic clones 80B1.1(ϩ/Ϫ) and 80B1.3(ϩ/Ϫ), from which were derived the homozygous mutant clones 80B1.1P(Ϫ/Ϫ) and 80B1.3P(Ϫ/Ϫ), respectively.
The first phenotype examined was growth. The population doubling time of wild type 3174 trypanosomes (9.

Ku Is Essential for Maintenance of a Transcriptionally Active
Telomere-Telomere length maintenance depends on Ku. In T. brucei, it is possible to monitor the length of individual telomeres because of the subtelomeric location of individual VSG genes. By digesting DNA with a restriction enzyme with a site upstream, but not downstream, of a VSG and then probing for that gene, a fragment containing the entire telomere tract can be detected. We have studied two distinct VSGs. The 221 gene is in the transcriptionally active telomere in the bloodstream trypanosomes under study. The VO2 gene has one copy in a silent telomere and a second copy that, being within a chromosome, is flanked on both sides by restriction sites. For KU70, the clones 70P3.1(ϩ/Ϫ) and 70P4.2(ϩ/Ϫ), as well as a number of subclones of 70P3.1B(Ϫ/Ϫ) and 70P4.2B(Ϫ/Ϫ), were examined. Similarly, for KU80, we studied the clones 80B1.1(ϩ/Ϫ) and 80B1.3(ϩ/Ϫ) and the homozygous deletion clones 80B1.1P(Ϫ/Ϫ) and 80B1.3P(Ϫ/Ϫ).
We initially looked at the effect the absence of Ku had on the telomere immediately downstream of the transcriptionally active expression site, 221 ( Fig. 2A). Subclones were derived for each independent homozygous and heterozygous KU mutant. One subclone from each independent KU80 mutant was used (Fig. 2C), whereas a more extensive analysis was carried out for KU70 (Fig. 2B). In this, five subclones were derived from each independent homozygous mutant and the progenitor wild type cell line, and a total of five heterozygous subclones were also analyzed. Terminal restriction fragment sizes were measured by digestion with EcoRI and Southern blot analysis. For homozygous mutants, these were generally smaller than those seen for heterozygous or wild type cell lines (Fig. 2), indicating an impairment of telomere maintenance or an increase in telomere degradation.
To quantify the extent of telomere loss in KU mutants, and to be certain that this phenotype was due to the loss of Ku, we reintroduced wild type Ku into its own locus using a co-transcribed bleomycin (BLE) resistance cassette (pCC101) and, as a control, we targeted BLE to the tubulin locus (pRM450). Four null mutant subclones (70P3.1BϪ/Ϫ5, 70P3.1BϪ/Ϫ4, 70P4.2BϪ/Ϫ3, 70P4.2BϪ/Ϫ5; Fig. 2B, lanes 20, 19, 13, and 15) were chosen for this experiment. Analysis of the active expression site (once again using EcoRI to release terminal restriction fragments) showed that in each case, re-expressors harbored longer telomere tracts than did the null mutants (Fig. 3). These were grown for an estimated 18 generations during selection for transformants and then 12 more generations in the absence of phleomycin, at which time they were analyzed for telomere tract length. Taking the fragment sizes for all these KU70 experiments, the transcriptionally active 221-telomere fragment was a mean 7.2 kb long (range 3-10 kb) in homozygous mutant trypanosomes and 10.3 kb long (range 9 -15 kb) in the re-expressor trypanosomes. This fragment contains about 3 kb of non-telomere tract sequence.
A Transcriptionally Inactive Telomere Is Less Prone to Length Changes in Ku Null Mutants-We next looked at the effect disrupting Ku had on a transcriptionally inactive telomere (Fig. 4). This was carried out using restriction digestion with AgeI and Southern blot analysis. In this instance, three of the four subclones used in the 221 experiment were analyzed. The VO2 gene has one copy in a silent telomere and a second copy that, being within a chromosome, is flanked on both sides by restriction sites. Fragment sizes for the telomere downstream of VO2, measured on the same DNA samples and therefore identical lineages as for the 221 analysis, are 9.4 kb (9.1-10 kb) in homozygous mutants and 9.8 kb (9.5-10.5 kb) in re-expressors. The telomere immediately downstream of the VO2 gene has not been characterized, so we cannot say how much non-telomere sequence is present in the AgeI terminal restriction fragments.  11.75 h at 0.0003%, 15 h at 0.0004%, and 25.25 h at 0.0005%. These results represent an average of two independently obtained data sets. Thus, for both heterozygous and homozygous mutants of both KU70 and KU80, the same dose-response effect occurred, indicating no increase in mutagen sensitivity in the mutants. Confirmation of the apparent lack of increased sensitivity in the mutants came from a clonal growth assay in which five trypanosomes were plated in each well of a 96-well plate and the percentage of wells displaying growth after 96 h was scored. This test is more sensitive but also more variable due to trypanosome founder effects. The assay showed, again, no increased sensitivity to MMS or to phleomycin, which we tested at two trypanosomes per well (Fig. 5).
Ku Null Mutants Are Not Impaired in Bloodstream VSG Switching-Having established some of the general phenotypes predicted for KU genes, we next asked whether there is a specific role in the differential expression of VSG genes. In the bloodstream stage of T. brucei S427, switching occurs at a background rate and is achieved by a number of routes, including transcriptional switching between telomeres and gene duplication into telomeres. The marked strain we have used allows most of these activation mechanisms to be detected (40),  providing a convenient means of testing for alteration in switch rate. This assay was applied to the KU70 mutant lines 70P3.1(ϩ/Ϫ), 70P4.2(ϩ/Ϫ), 70P3.1B(Ϫ/Ϫ), and 70P4.2B(Ϫ/Ϫ). All gave the rate of 0.1-0.6 ϫ 10 Ϫ6 switch/cell/generation (Fig.  6), which is typical of wild type cells, there being no evidence for a significant decrease in switch rate in the absence of Ku70. Examination of the switching events from a selection of the switched variants recovered indicated further that there was no change in the relative levels of transcriptional versus recombinational VSG switching (data not shown).
Ku Is Not Essential for Silencing of the Naturally Telomeric MVSG Genes-VSG genes are first expressed in the metacyclic stage of the trypanosome life cycle, in the tsetse. Differential expression of the subset of telomeric genes used there, known as MVSG genes, occurs by transcriptional activation. In most life cycle stages, MVSGs are not expressed, but individual ones are activated in individual metacyclic trypanosomes. Here, we tested the hypothesis that the silencing of MVSGs in the procyclic (tsetse midgut) stage is exerted by a Ku-mediated telomere position effect. We have tried repeatedly to transmit T. brucei S427 through tsetse, to no avail (data not shown), so we have been unable to identify MVSGs in this strain. Therefore, we instead used T. brucei EATRO 795, which routinely develops to the metacyclic stage, permitting characterization of the MVSGs encoding the ILTat 1.22, 1.61, and 1.63 VSGs. We deleted both KU70 alleles in procyclic stage T. brucei EATRO 795 grown in vitro, creating the independent heterozygous, puromycin-resistant knockout cloned lines 70ϩ/Ϫ1 and 70ϩ/Ϫ2 and then their homozygous, blasticidin-resistant doubly deleted clonal descendants called, respectively, 70Ϫ/Ϫ1.1 and 70Ϫ/Ϫ2.1. RT-PCR was used to search for transcripts from two regions of each MVSG locus. The first, downstream of the transcription start sites (Fig. 7), would detect primary transcripts. The second, including the amino-terminal VSG coding sequence, would detect primary transcripts and stable RNA. The scant availability of metacyclic stage RNA prevented its use as a positive control, so instead of metacyclic cDNA, we used the corresponding trypanosome genomic DNA. Although this control (Fig. 7, A and B, lane 11) and an RNA POLI positive control for integrity of cDNA were both positive (Fig. 7C), no RT-PCR products were detected in wild type, heterozygous, or homozygous mutants for any of the three MVSG loci immediately downstream of their promoters (Fig. 7A). Similarly, nothing was detected for the coding region (Fig. 7B). DISCUSSION We have cloned and characterized, by gene deletion, the trypanosome homologues of Ku70 and Ku80. Our analysis of telomere lengths revealed the telomere shortening phenotype that occurs in other eukaryotes, but the classical role in DNA damage repair by NHEJ was not detected. Our hypothesis that Ku is important in VSG recombinational or transcriptional control was not supported, revealing at least that the Kumediated TPE described for S. cerevisiae is not necessary for regulation of antigenic variation.
Due to the limitation that low sequence homology between Ku proteins from different species prevents construction of homology models based on the human Ku structure (44), our identification of these trypanosome sequences has relied on the five primary homology regions (42) and on our observation that most of the determined human ␣ helices and ␤ strands have counterparts, with corresponding spacing, in the trypanosome sequences. In this way, it became apparent that trypanosome Ku70 has an extra, ϳ70 residues in the ring that encircles DNA, immediately upstream of PHR3. This correlates well with the observed gaps between PHR2 and PHR3: 250 amino acids in human Ku70 (42) as compared with 315 residues in trypanosome Ku70. Although the human Ku80 PHR2-PHR3 gap (274 residues) is colinear with Ku70, that of trypanosome Ku80 is considerably shorter (273 residues) than in its presumed heterodimeric partner; perhaps trypanosome Ku70 has an extra loop available for interaction with other proteins. In common with other simple eukaryotes, trypanosome Ku80 lacks the carboxyl-terminal motif of higher eukaryotes for binding to the DNA-PK catalytic subunit (42), and we cannot find evidence for the catalytic subunit sequence in the (incomplete) trypanosome genome database. The trypanosome Ku70 contains what might be a SAP domain, in common with Ku70 from other species (6,43).
Our data show that Ku is central to telomere length maintenance in trypanosomes and reveal that, in the presence of Ku, there are marked differences between short and long telomeres and between the transcriptionally active and inactive telomeres examined in the kinetics of telomere lengthening. Although quantification of the extent of telomere loss was complicated by the large terminal restriction fragment sizes involved, regrowth of short, more accurately sized telomeres could be monitored following reintroduction of KU70. The EcoRI terminal fragments liberated from the active telomeric 221 locus in null mutants were, on average, 3.1 kb shorter than in the Ku70 re-expressors after 30 population doublings. This difference was substantially lower for the transcriptionally inactive VO2 telomere, where reintroduction of Ku70 caused an average increase of only 400 bp. Transcriptionally active trypanosome telomeres have previously been demonstrated to grow faster than inactive telomeres (46). There was also a growth difference between particularly short (ϳ1 kb) and much longer (Ͼ7 kb) telomeres upon re-expression of Ku70; the short telomeres extended by ϳ170 bp/population doubling, as compared with 103 bp/population doubling for the longer telomeres. Such rapid expansion resembles the growth of newly formed trypanosome telomeres (47). Our findings presumably reflect that telomere capping is less pronounced in transcriptionally active, or new, telomeres. Despite the great decrease in telomere length, no trypanosome clone analyzed appeared to have lost all telomeric sequence, and null mutant clones with terminal restriction fragments as short as 4 kb multiplied at the same rate as wild type cells even after 150 generations (data not shown). A length equilibrium was reached within such periods, indicating that the trypanosome has at least one alternative pathway for telomere maintenance, as has been observed in other organisms (reviewed in Ref. 48).
The main DNA repair role for Ku is in NHEJ reactions (49,50). We found no difference between wild type, heterozygotic, or homozygotic KU mutants in their sensitivity to MMS, in contrast to rad51 mutants, which display increased MMS sensitivity in the same range of drug concentrations (34). We see also that there is not an increased sensitivity to phleomycin. It appears, therefore, that if Ku plays a role in the repair of trypanosome DNA damage, it would be masked by the greater prevalence of homologous recombination. To some extent, this is what happens in yeast KU mutants, where no increased MMS sensitivity is detectable in diploids, and the background role played by Ku is seen only when the more central homolo- gous recombination gene RAD52 is also inactivated (9,11,18,51), although there is a significant increase in sensitivity to bleomycin in both haploid and diploid stages (52). Indeed, wild type trypanosomes and S. cerevisiae nearly exclusively use homologous recombination when transfected, linearized DNA is integrated into the genome. Nevertheless, the trypanosome appears to have more capacity for alternative repair, such as is seen in the persistent resistance to phleomycin, and there is evidence for at least one Rad51-independent homologous recombination pathway (34). We have been unable to find NHEJ in wild type trypanosomes using an assay for recircularization of electroporated, restriction-digested plasmids, 3 although chromosomal integration of linear DNA constructs with nonhomologous flanking sequence gives some success. 4 It remains possible that Ku-dependent NHEJ exists in trypanosomes but is not observed in these transformation experiments because DNA integration operates at only the S to M cell cycle phase, where homologous recombination may dominate in higher eukaryotes (53). Perhaps NHEJ is limited to the G 1 phase or to non-dividing trypanosome life cycle stages, when limited DNA replication occurs.
There has been much speculation as to why parasite contingency genes are found at telomeres. Besides the trypanosome VSGs, which are transcribed only from telomeric loci in the distinct metacyclic and bloodstream populations and have a large subset of silent genes at minichromosomal telomeres, there are examples in protozoa (25,54), the fungus Pneumocystis (26), and even on linear plasmids in the bacterium Borrelia hermsii (55). These systems rely on the existence of a wide set of silent genes (or segments) and the expression of only one gene at a time. There is a constant pressure for expansion of the silent information, which probably is best served by DNA recombination, generating novel sequence combinations. Telomeres may enhance this process as they have high recombination rates and are able to recombine with each other promiscuously, independent of homology elsewhere on the chromosome (56). Whether telomeric location is directly associated also with singular contingency gene expression is unclear. Where the switch between different genes proceeds by promoter activation and deactivation, such as in Plasmodium or in trypanosome transcriptional switches, TPE has been invoked (25,27,28). However, early hopes that bloodstream VSG transcriptional switching was controlled by a TPE were not well supported by further experimental analysis (57), although a modified TPE was not ruled out. A degree of TPE of a reporter gene in T. brucei has been described (58), although the effect extended only some of the way toward the VSG promoter, which can be more than 45 kb from the telomere tract. A case for TPE controlling the MVSGs was more compelling because their promoters are only about 5 kb from the telomere tract and because the trypanosome is committed globally to post-transcriptional control of gene expression, raising the possibility that exploitation of TPE might have been an easy way for the trypanosome to achieve the transcriptional control that operates on MVSGs. Within the same type of genomic environment, most bloodstream VSG switching occurs by the non-reciprocal duplication of silent VSGs into the final 5 kb of chromosomes. Given this involvement of MVSG transcriptional silencing and telomere recombination, we have asked whether Ku is involved as a global regulator in the VSG system.
It is clear from the bloodstream switching assay, which we have used previously to demonstrate an important role for Rad51, that the absence of Ku has no influence on either recombinational or transcriptional switch mechanisms. Of technical necessity, this high throughput assay is performed with the S427 strain, which switches at background rates and with a bias against the recombinational mechanism(s); study of more rapidly switching trypanosomes is being developed. Although we find no evidence for Ku function in MVSG repression in the procyclic stage, it is not ruled out that Ku does have a necessary role in the differential expression of MVSGs in the metacyclic stage, when global silencing is lifted, but experiments to test that are very difficult, involving tsetse transmissions. Taking our findings together, we believe that Ku in trypanosomes does not have a necessary role in the VSG system. Furthermore, the great shortening of telomeres also did not alter VSG regulation, which would be consistent with the telomere not playing a major regulatory role in antigenic variation. Functionally, Ku in trypanosomes may act as in S. pombe, where it is involved in telomere maintenance but is not involved in TPE (11), rather than as in S. cerevisiae, where it plays both roles. Two possible reasons for Ku not being involved in VSG regulation are that it is not localized preferentially to telomeres, as in S. pombe, or that the structure of the chromosome end in trypanosomes differs qualitatively from that in other organisms, keeping Ku physically separate from the VSG genes. There is evidence in favor of the latter possibility. In S. cerevisiae, the telomere is thought to fold back, creating the capped structure that results in the TPE. In organisms with longer telomere tracts, a t-loop is formed in which it is believed the end of the telomere tract base-pairs within the DNA duplex further up the tract. In human cells and the protozoan Oxytricha, t-loops are long (59,60), but recently it has been shown that, in T. brucei, they are shorter, being only ϳ1 kb long (61). We believe a consequence of this is that the complex of proteins associated with the chromosome end is held well away from VSGs, preventing the possibility of TPE. It may even be that evolution of the brief t-loop in trypanosomes has occurred as a means of divorcing telomere effects from the more exacting requirements of the VSG system.