INTRODUCTION

The telomeric region of eukaryotic chromosomes serves several biological functions. Telomeres are simple DNA that form the ends of chromosomes and are essential for chromosomal complete replication, protection from degradation and end-to-end fusion, and segregation at mitosis (2, 3). These events are vital to the cell and impact on genome stability, replicative capacity, senescence, immortalization, and neoplastic transformation (4). Telomeres also silence expression of genes in their proximity by telomeric position effect (5).

In Trypanosoma brucei, the causative agent of African sleeping sickness, telomeric regions also are recombinogenic and play an additional crucial role in the survival of the parasite. They are the site where copies of the ~1000 different variant surface glycoprotein (VSG)¹ genes are transposed by recombinational processes and individually activated, allowing the bloodstream form of the parasite to evade its host's immune system. Numerous telomeres can serve as VSG expression sites, but only one is active at a time, and the remaining telomeres are silent. When not actively transcribing VSG genes, T. brucei telomeres exert a silencing effect on subtelomeric genes, similar to the telomere position effect in yeast (6, 7, and references therein).

Telomeric regions have interesting structures that almost certainly impart their important biological functions. Chromosome termini consist of very simple sequence telomere repeats (e.g. CCCTAA:TTAGGG in T. brucei) with multiple duplex copies and a protruding 3 single strand overhang of the G-rich strand. Abutting the simple telomere sequences are frequently telomere-derived sequences. Such subtelomeric repeats have been identified in many organisms, including Chlamydomonas reinhardtii (8), Arabidopsis thaliana (9), and Ustilagos maydis (10), Plasmodium berghei (11), and T. brucei (12). In T. brucei they are tandem 29-bp elements that separate the telomere repeats from the VSG sequences on multiple chromosomes (1, 12). Moving further inward, eukaryotic chromosomes generally contain middle-repetitive sequences named telomere-associated sequences (13-15). Although the subtelomere repeats and telomere-associated sequences are generally less well understood than the telomere repeats, they also appear to be the sites of several important functions. In humans, subtelomeric regions are hot spots for recombination between non-homologous chromosomes and for meiotic pairing (16, 17), and in P. berghei they are responsible for chromosomal size polymorphisms through deletions and insertions of subtelomeric repeats (18). The 27-bp subtelomeric repeats of P. berghei were found to be the junction for the subtelomeric rearrangement events (11). In Saccharomyces cerevisiae, amplification and recombination of the Y subtelomeric repeats can rescue the cells from senescence and eventual death (19), and the X subtelomeric repeats were found to be involved in DNA segregation (20). A striking example of the recombinogenicity of telomere-associated regions is in the VSG genes of T. brucei that results in antigenic variation.

By virtue of their repetitive sequences and organization, the telomeric and subtelomeric repeats may adopt peculiar structures, including single strand regions and G-quadruplexes (also called G-tetraplexes) that, in combination with their associated proteins, may mediate their various functions. Proteins that bind the telomere G-rich repeats as a single strand or as a 3-overhang have been identified in several organisms including Stylonychia mytilis (21), Euplotes crassus (22), Tetrahymena (23), Xenopus laevis (24), C. reinhardtii (8) S. cerevisiae (25), and A. thaliana (26). In Oxytricha (27), the  subunit of the telosomal heterodimer binds to the 3-telomere-G-strand overhang and promotes formation of a G-quadruplex, which could regulate telomere length, since the quadruplex appears to be a poor substrate for telomerase, the enzyme that synthesizes telomere G-rich repeats (28). The biological importance of the G-tetraplex structure, at least in yeast, was indicated by mutation of the G-tetraplex-specific nuclease, KEM1, which caused a meiotic block, chromosome loss, and cellular senescence (29). Telomere G-rich strand binding proteins may also play a role in nuclear architecture, since their mutation in yeast affects chromosome organization at the nuclear periphery (30). Proteins that interact with the telomere repeats in their double-stranded form have also been identified in diverse organisms, including Physarum polycephalum (31), humans (32), and yeast (33). The abundant yeast protein RAP1 binds the double-stranded telomere DNA and evidently regulates both telomere length and position effect on genes in proximity (33, 34, and references therein). Interestingly, RAP1 was shown to distort the DNA duplex (35, 36) and, more recently, to induce DNA-quadruplex structures in the telomere G-strand as well (37-39).

Proteins have also been isolated that bind to the subtelomeric repetitive region. In yeast, TBF/1, an essential protein, binds the TTAGGG repeats found at the junction between the telomere DNA and the subtelomere repeats (X), and its gene mutation affects telomere structure and position silencing effect (40, 41). In T. brucei, the ST-1 protein preferentially binds to the subtelomere sequence but also to the telomere repeats (1). We previously reported that the 39-kilodalton ST-1 protein shows specific affinity for the double-stranded 29-bp subtelomere repeat and for the double-stranded telomere repeat, but stronger affinity for their C-rich single-strands, the first protein with such binding properties to be reported (1).

In addition to the intrinsic ability of the telomeres' G-rich strand to fold into a tetraplex (Ref. 42 and references therein), their C-rich strand was shown to self-associate into a four-stranded structure called the i-motif (42, 43). If such structures also occur in vivo, then the telomere DNA would be present in a non-B form. Telomere chromatin in vertebrates consists of unusual nucleosomes (44, 45), and in lower eukaryotes it is non-nucleosomal (46, 47). Taken altogether, these and other data suggest that the telomeric region DNA, in conjunction with the telomere-associated proteins, may be arranged in a non-canonical manner to fulfill its various functions.

Given the interesting telomeric events that occur in T. brucei, including the exclusive and developmentally regulated expression of the recombinogenic VSG genes and an unusual telomeric nucleotide modification (48, 49), we have endeavored to further characterize telomere-binding proteins in trypanosomes. In this report we describe ST-2, a novel telomeric region binding protein complex. ST-2 exhibits affinity for the duplex telomere and subtelomere sequences but stronger affinity for the G-rich strand of the telomere repeat and still greater affinity for the G-rich strand of the 29-bp subtelomere repeat. The specific affinity of both ST-2 and ST-1 for both the subtelomere and the telomere may reflect a common structure extending from the subtelomeric region to the chromosome ends. Moreover, the complementarity of ST-1 and ST-2 binding properties indicates that in the cell this binding may help keep the DNA in a partially single-stranded conformation. Such structure may facilitate the crucial recombinogenic function of trypanosome telomeres, their chromatin silencing, or both.

EXPERIMENTAL PROCEDURES

The radiolabeled duplex telomere (Tel-ds) and subtelomere (Subtel-ds) repeats of T. brucei were prepared as described previously (1); homologous telomere (Tel-C and Tel-G (see Table I)) and subtelomere (Subtel-C and Subtel-G (see Table I)) strands were annealed and then labeled using the Klenow fragment of DNA polymerase and [-³²P]dATP. The G-rich strand of the subtelomere repeat (Subtel-G) was radiolabeled using T4 polynucleotide kinase and [-³²P]ATP. For use as double-stranded unlabeled competitors, R-Tel-ds (a rearranged version of Tel-ds in which the 2nd and 5th residues of each repeat were exchanged), MycB, MycA were prepared by annealing the complementary strands (R-Tel-C to R-Tel-G, MycB_u to MycB_l, and MycA_u to MycA_l, respectively (see Table I)). These plus Tel-ds and Subtel-ds (see above) were end-filled using the Klenow fragment of DNA polymerase and non-radioactive triphosphates. Additionally examined double-stranded competitors were AluI-digested pBR322 plasmid DNA and BglII-linkers (Life Technologies, Inc.) ligated to form a population of DNA fragments from 0.5 to 1.0 kilobase pairs. The transcripts of both strands of the subtelomere repeat (Subtel-C and Subtel-G) were synthesized in vitro from a single copy of the repeat inserted in pBluescript II KS plasmid as described previously (1). All single- and double-stranded nucleic acids were gel-purified before use.

Table I. Inhibition of ST-2 binding to Subtel-G by single-stranded and double-stranded oligomers Various single-stranded DNAs and two RNAs corresponding to the G- and C-strand of the subtelomere repeat were tested for inhibition of ST-2 complex formation on Subtel-G. +, showed competition; , showed no competition. Competitor Sequence Inhibition Single-stranded   Subtel-G 5-AGCTTTCGGGTTAGGGTGTTTCGGGTTAGCGGG-3 +   Subtel-C 5-AGCTCCCGCTAACCCGAAACACCCTAACCCGAA-3     Tel-G 5-GATCTTAGGGTTAGGGTTAGGG-3 +   Tel-C 5-GATCCCCTAACCCTAACCCTAA-3     Tel-end 5-CGCGGACTGCCGATATTGTTAGGGTTAGGGTTAGGGTTAGGG-3 3-GCGCCTGACGGCTATAACAATCCC-5 +   R-Tel-C 5-GATCACTCACACTCACACTCAC-3     R-Tel-G 5-GATCGTGAGTGTGAGTGTGAGT-3     1.3/3 5-TGCTCTAGAGCAAGCCGCCGCCCACTTGTG-3     Neo 5-GTTCAATCATTGATCAGATC-3     NeoII 5-CCGGATCCGGCCCAAGCGGCCGGAGAACCTGC-3     CAL5A 5-CTCTAGACCGCAATCCCGGCCCGTGCG-3     CAL5B 5-CTCTAGAGCCCAAGTGTGAAGGAAAAAGG-3     CAL3 5-GGAGTTCCGCCTCGGTGGGG-3     MycBu 5-GCTCAGATCTCCATGGAGCAAAAGCTCATTTCTGAAGA-3     MycBl 5-ACGGAGATCTAATTCAAGTCCTCTTCAGAAATGAGCTT-3     OL-2 5-CATGGAGAGACGATTCATCAT-3     OL-3 5-GACCCTGTACCTGCAAATGAG-3     OL-5 5-AGACCTCAGACTGCTCATTTG-3     OL-6 5-AATTCCCTCATGACATGTCTT-3     OL-7 5-TTACCCGCAGGACATATCCAC-3     OL-8 5-ATGGCCGATCAACTCTCCAAC-3     OL-9 5-TAGAACAGTTTCTGTACTATA-3     MycAu 5-ATCTTTTTTTTTTCTTTAACAGATTTG-3     MycAl 5-AAAAACAATTCTTAAATACAAATCTGTT-3     Poly(dC)     Poly(dG)     Subtel-C (RNA) 5-GGGAACA4GCUGGGUACCGGGC7UCGAGGUCGAGGUAUCGAUAAGCUCCCGCUAACCCGAAACA- CCCUAACCCGAAAGCUUGAU-3     Subtel-G (RNA) 5-GGGCGAAUUGGAGCUCCACCGCGGUGGCGGCCGCUCUAGAACUAGUGGAUC5GGGCUGCAGGAA- UUCGAUACAAGCUUUCGGGUUAGGGUGUUUCGGGUUAGCGGGAGCUUAU-3   Double-stranded   MycA Annealed MycAu and MycAl     Subtel-ds Annealed Subtel-G and Subtel-C +   Tel-ds Annealed Tel-G and Tel-C +   R-Tel-ds Annealed R-Tel-G and R-Tel-C

Insect (TREU667) and bloodstream (ILTAT 1.3 strain) form T. brucei were, respectively, grown in vitro and propagated in rats from which they were purified on DE-52 columns, as described (1). S-100 extracts were prepared from both developmental stages, and binding assays were conducted as described (1). 20 fmol of radiolabeled probe (Tel-ds, Subtel-ds, or Subtel-G) were incubated with the indicated amounts of extracts (0.2, 1.0, or 5.0 µg), in 20 µl of binding buffer (25 mM HEPES, pH 7.5, 5 mM MgCl₂, 0.1 mM EDTA, 0.5 mM dithiothreitol, 10% glycerol) also containing 100 mM KCl and 100 ng of poly(dA-dT) for 20 min at room temperature, and the resultant complexes were resolved on native 4% polyacrylamide gels at room temperature (1). Binding at 0.1 or 1.0 M KCl yielded similar results.

For competition experiments, the extracts were preincubated with the various unlabeled oligomers at the indicated concentrations for 30 min at room temperature before addition of radiolabeled probe. In such prebinding competition assays, at some point as the amount of unlabeled competitor is increased, the amount of radiolabeled slowly migrating complex shows a fairly precipitous decrease (e.g. Fig. 2, lanes 14 and 20-22; Fig. 3B lanes 17-18 and 20-22). This is to be expected, for once a sufficient amount of a tightly binding competitor is pre-bound, it will almost completely suppress binding by the subsequently added radioactive probe. (If the competitor were instead added at the same time as the probe, the decrease in the amount of radiolabeled slowly migrating complex would be more gradual.) Also, depending on the saturation point for that extract and the amounts of competitor added, differences in binding affinities of two DNAs like Subtel-G and Tel-G result in different apparent locations of the precipitous decrease (Fig. 2) or can be observed as quantitative differences in residual binding (Fig. 3C, lanes 8 and 11).

S-100 extracts were fractionated on Affi-Gel blue columns equilibrated in binding buffer plus 10 mM KCl by either a continuous (0.1-2 M KCl) or step (0.1, 0.5, 1.0, 1.5, and 2.0 M KCl) gradient, as described (1). For add-back experiments, the indicated fractions were dialyzed against binding buffer plus 100 mM KCl, and then 1 or 5 µl was incubated with renatured ST-1 (0.5 µl) in a standard binding reaction in the presence of radiolabeled Subtel-ds. Renaturation of ST-1 was performed as described (1). Briefly, after fractionation of T. brucei S-100 extracts on Affi-Gel blue columns followed by Bio-Rex chromatography, fractions containing ST-1 activity were electrophoresed on polyacrylamide protein gels from which the single band containing the 39-kDa ST-1 was excised, eluted, and then renatured using guanidine hydrochloride (1).

Binding assays were as described above except for using 20 fmol of radiolabeled Subtel-G and 10-15 µl of Affi-Gel blue fraction containing peak ST-2 activity or, when indicated, 10 µl of renatured ST-1. After electrophoresis, the gel was covered with plastic wrap, exposed to UV light (500 mJ) for 10 min, and then subjected to autoradiography overnight at room temperature. Gel slices containing ST-2 or ST-1 complexes were crushed in an equal volume of 2 × SDS sample buffer, boiled for 5 min, then electrophoresed on a 12.5% polyacrylamide SDS protein gel (50). For competition analysis, ST-2-containing Affi-Gel blue fractions were preincubated with 50-fold molar excess of the specific or nonspecific competitors as described above.

Treatment of ST-2 with potato acid phosphatase was done as follows (51). The gel slice containing ST-2 from the UV cross-linked polyacrylamide gel was divided in two parts. Each half was washed with 100 mM MES, pH 6.0, and 1 mM phenylmethylsulfonyl fluoride and then incubated at 37 °C for 15 min in an equal volume of MES buffer either alone or containing 1.0 unit of potato acid phosphatase. The reaction buffer was then removed by aspiration and an equal volume of 2 × SDS sample buffer was added, in which the gel slices were crushed and boiled for 5 min. The mixture was electrophoresed on 12.5% polyacrylamide SDS protein gel.

RESULTS

ST-2 binding activity in T. brucei shows specificity for the telomere repeats but a higher affinity for the 29-bp subtelomere repeats.

We previously reported the isolation of ST-1 (1), a 39-kDa telomere binding protein in T. brucei, which exhibited specificity for the telomere repeats (CCCTAA) but showed a higher affinity for both the duplex form of the 29-bp subtelomere repeats and still higher affinity for their C-rich strand. These subtelomere sequences consist of telomere-derived 29-bp repeats usually found in tandem arrays juxtaposed to the telomere DNA on many chromosomes of the parasite (12). We also noted a complex of slower migration that formed on both the double-stranded telomere and subtelomere repeats (1) (see Fig. 1A and Fig. 2, lanes 2), which is the subject of this report.

When radiolabeled substrate Tel-ds (the annealed Tel-C and Tel-G strands (Table I) containing three trypanosome telomere repeats (CCCTAA)) is incubated in an S-100 extract derived from bloodstream or procyclic form trypanosomes, it forms the previously characterized complex with ST-1 (lower arrow in Fig. 1A, lanes 2 and 7) and a slower migrating complex (upper arrow in Fig. 1A). Like the ST-1 complex, the slower migrating complex exhibits specific binding, since its formation is inhibited by pre-binding of unlabeled Tel-ds (Fig. 1A, lanes 3 and 8) but is not inhibited by seven nonspecific double-stranded DNAs (lanes 5, 6, 9, and 10, and Table I). These nonspecific competitors include R-Tel-ds (the annealed R-Tel-C and R-Tel-G strands (Table I)), a rearranged version of Tel-ds in which the 2nd and 5th residues of each repeat were exchanged, thus maintaining the G- and C-strand bias. This specific complex formation was observed using multiple different S-100 preparations and a wide range of extract concentrations, although with more protein in the binding assays the relative intensity of this slowly migrating complex increased and an increased amount of unlabeled competitor DNA was needed for effective competition of binding by the subsequently added radiolabeled DNA (Fig. 1A, lanes 2 and 7 and data not shown; in Fig. 1A, lanes 1-6 were exposed for more time than lanes 7-10).

When instead using Subtel-ds (annealed Subtel-C and Subtel-G strands (Table I) containing the 29-bp subtelomere repeat) as the radiolabeled substrate in the binding assays, similar results were observed, with a complex of slower migration detected in addition to that of ST-1 (Fig. 1B, lanes 1 and 5). Again, formation of this slower complex was inhibited by prebinding of self, in this case unlabeled Subtel-ds (Fig. 1B, lane 3), and it was undiminished by six nonspecific double-stranded DNA competitors, including R-Tel-ds (Fig. 1B, lane 4 and Table I). The relative amount of this slower migrating Subtel-ds complex also increased with increasing amounts of S-100 extracts (Fig. 1B, lanes 1 and 5). (The Subtel-ds complex of intermediate migration (heavy arrow in Fig. 1B) was previously shown (1) to represent a dimer of the ST-1 protein on the 29-bp subtelomere repeat.)

To determine the relationship of the slowly migrating complexes formed on Tel-ds and Subtel-ds, we performed cross-competition experiments. Formation of the slowly migrating complex on both radiolabeled Tel-ds and radiolabeled Subtel-ds could be inhibited by prebinding unlabeled Tel-ds (Fig. 1A, lanes 2, 3, 7, and 8; Fig. 1B lanes 1 and 2), but in both cases was more efficiently inhibited by prebinding the same amount of unlabeled Subtel-ds (Fig. 1A, lanes 3 and 4; Fig. 1B lanes 2 and 3). This indicates that the slowly migrating complexes formed on both the telomere (Tel-ds) and Subtelomere (Subtel-ds) repeats contain common binding activities of the trypanosome extract. This is also the case with the ST-1 complex (1). We have tentatively named the activity that generates this slowly migrating complex ST-2 (for subtelomere and telomere binding activity 2).

Both ST-1 and ST-2 specifically bind the duplex telomere repeats but have a higher affinity than the duplex subtelomere repeats. This is concluded because Subtel-ds is a more efficient competitior than is Tel-ds (Fig. 1, A and B) and because when they are the same specific activity probes and when one uses the same extract preparation, Subtel-ds is seen to form the slowly migrating complex more efficiently than does Tel-ds (data not shown). Like the ST-1 complex, the ST-2 complex was stable at 2 M KCl and was detected in all examined extracts derived from both bloodstream and procyclic form trypanosomes (data not shown). Furthermore, the limiting ST-2 factor appears to be present in the cell extract in lower abundance or activity than ST-1, for when using saturating amounts of probe, considerably more ST-1 than ST-2 complex is formed (Fig. 1 and data not shown). Interestingly, however, as the amount of extract is increased and the amount of free probe becomes limiting, the relative amount of ST-1 complex decreases (Fig. 1 and data not shown), a result that could arise if the ST-2 complex were dominant or contained the ST-1 protein (see below).

The specificity of ST-2 for double-strand versus single-strand sequences was next characterized by additional competition analyses using radiolabeled Subtel-ds as a probe (Fig. 2 and data not shown). Several concentrations of various single-stranded DNA oligomers (see Table I) including the G- and C-rich strands of the telomere (Tel-G and Tel-C), subtelomere (Subtel-G and Subtel-C), and rearranged telomere (R-Tel-C and R-Tel-G) repeats and various nonspecific single-stranded DNAs, as well as Subtel-ds and Tel-ds, were examined for pre-binding competition ability. (The extract used for the experiment shown in Fig. 2 has a higher binding capacity than those used in Fig. 1, for relatively larger amounts of Tel-ds and Subtel-ds oligonucleotides are needed to effectively compete the binding (Fig. 2 and data not shown).) Strikingly, the G-rich strand of the subtelomere repeat, Subtel-G, competed for ST-2 binding very efficiently, showing over a 10-fold greater efficiency of binding than did Subtel-ds (Fig. 2, compare lanes 20-22 with lanes 8-10). The G-rich strand of the telomere repeat, Tel-G, also competed for formation of the ST-2 complex more efficiently than did Subtel-ds or Tel-ds (lane 14; see also Fig. 3B, lanes 17-18). This, however, does not reflect a general preference of ST-2 binding to single-stranded DNA, for ST-2 binding was relatively resistant to six examined nonspecific single-stranded competitor DNAs (Fig. 2, lanes 27-30; data not shown), as well as to six examined nonspecific double-strand competitor DNAs (Fig. 1). Thus, ST-2 binds most preferentially to Subtel-G, but it also binds with considerable affinity to Tel-G and with lower but still appreciable affinity to Subtel-ds and Tel-ds. This same order of preference in binding is seen in all trypanosome S-100 extracts examined, and no matter which radioactive probe is used to assess these competitors DNAs (Fig. 1, 2; see ahead to Fig. 3B; data not shown).

Formation of the ST-2 complex on Subtel-ds was also relatively resistant to pre-binding by Subtel-C, the C-rich strand of the subtelomere repeat, and by Tel-C, the C-rich strand of the telomere repeat (Fig. 2, lanes 15-18 and 23-26). Subtel-C competed ST-1 binding many fold more efficiently than it did ST-2 binding (lanes 23-26), whereas Subtel-G competed binding of ST-2 considerably more efficiently than it competed binding of ST-1 (lanes 19-22). Thus, ST-2 and ST-1 have complementary binding specificities on the subtelomere and telomere repeats of T. brucei, with ST-1 showing a strong preference for the C-rich strands and ST-2 strongly favoring the complementary G-rich strands (Ref. 1; Fig. 2; data not shown). Indeed, the small amount of competition of ST-2 complex formation caused by Subtel-C and Tel-C may be an indirect effect of their preferential binding of the ST-1 protein (1), which appears to be a component of the ST-2 complex (see below). This binding also suggests that ST-2 and ST-1 could partially denature the subtelomere and possibly the telomere DNA, each on one strand.

We next wanted to directly examine the binding of ST-2 to the DNA sequence for which it has shown the highest affinity, Subtel-G. Indeed, ST-2 complex formation is considerably more efficient on Subtel-G than on the duplex subtelomere repeat, Subtel-ds, when comparable amounts of the two oligomers were used as the radiolabeled substrate (Fig. 3A). It is notable that the ST-2 complex shows similar migration when using the Subtel-G and Subtel-ds probes. Although one could propose that the apparent binding of ST-2 to Subtel-ds is due to its totally denaturing the input duplex DNA and then shifting the G-rich single-strand, an additional experiment showed that this is not the case. We prepared Subtel-ds DNA in which only the C-rich strand, only the G-rich strand, or both strands were labeled to the same extent. The complexes that formed on these DNAs showed the same migration and the same relative intensity (data not shown), demonstrating that when ST-2 binds to Subtel-ds, it does not fully denature the oligomer. This result also indicates the ST-2 complex formed on Subtel-G and Subtel-ds DNA (Fig. 2A) migrate similarly because their migration is determined mainly by the ST-2 extract components and only little by the oligonucleotide that is bound.

Further examining the specificity of ST-2 complex formation on Subtel-G, we tested various single- and double-stranded competitors (Fig. 3B). ST-2 revealed a clear preference for binding: strongest for the G-rich strands of the subtelomere repeat (Fig. 3B, lanes 19-22), less for the G-rich strand of the telomere repeat (Fig. 3B, lanes 15-18), and still less but significant for Subtel-ds (Fig. 3B, lane 10). As noted above, this is the same order of preferential competition as was seen when using radioactive Subtel-ds DNA (Figs. 1 and 2). Furthermore, multiple examined nonspecific competitors all failed to inhibit ST-2 formation on radiolabeled Subtel-G (Fig. 3B, lanes 27-30, and data not shown). These 19 nonspecific single-stranded DNA oligomers (Table I) ranged in nucleotide composition to highly G-rich, such as poly(dG) and included the separated R-Tel strands; the nine other double-stranded competitors included R-Tel-ds and circular and linearized Bluescript DNA plasmids. On an equal weight basis, Bluescript plasmid containing 1 or 7 tandem 29-bp subtelomere repeats or 12 tandem telomere repeats also exhibited no competition in these analyses. Moreover, the C-rich strand of both the subtelomere (Fig. 3B, lanes 23-26) and telomere (Fig. 3B, lanes 11-14) repeats failed to compete the ST-2 binding.

Taken together, these results indicate that ST-2 is a DNA binding complex specific for the subtelomere and the telomere repeats, with greatest affinity for their single-stranded G-rich strands. We next wanted to examine whether ST-2 could potentially be equivalent to the telosome proteins of Oxytricha (27) and the recently described G-overhang protein of T. brucei (52) which cap chromosome ends by binding preferentially to the telomere G-strand overhang. To this end, we used competition experiments to compare the affinities of ST-2 for Subtel-G, Tel-G, and a pseudo-telomere, a structure consisting of two single-stranded trypanosome telomere repeats protruding from the 3-end of a double-stranded telomere repeat of a 24-bp duplex DNA (Tel-end; see Table I). There was no significant difference in the extent of competition of ST-2 complex formation by Tel-end and Tel-G (Fig. 3C, lanes 4, 5, 10 and 11), indicating that the duplex telomere DNA in Tel-end does not augment the binding of ST-2 to the G-strand telomere repeat. This property is unlike telomeric proteins believed to cap the chromosome ends in vivo which bind much more tightly to a telomere G-strand present as a 3-overhang on duplex DNA than to the telomere G-strand alone (24, 27).

Further analysis of ST-2 complex formation on radiolabeled Subtel-G showed that it is not competed by in vitro transcripts of either the C-rich or the G-rich strand of the subtelomere repeat (Table I; Fig. 3D, lanes 2-4), indicating that ST-2 is not an RNA binding complex. In addition, ST-2 proved to be proteinaceous, since it was sensitive to digestion by proteinase K and to temperature >50 °C. However, it was resistant to ribonuclease treatment (RNase A and T1, data not shown), indicating the absence of an available critical RNA component in the ST-2 complex.

The binding properties of ST-2 make it a unique telomeric complex, the first reported activity able to interact with both the telomere and the subtelomere G-strands. This could suggest a structural cross-talk between the telomeres and their neighboring subtelomere sequences.

Upon fractionation of trypanosome S-100 extracts on Affi-Gel blue columns (1), ST-1 peaked between 0.25 and 0.6 M KCl, and ST-2 peaked between 0.8 and 1.3 M KCl (Fig. 4A). (As noted above, the RNA gel shift complex of intermediate migration whose activity has the identical elution profile as ST-1 (Fig. 4A) represents a dimer of ST-1 on the Subtel-ds probe (1) which was used in this experiment to assess both ST-1 and ST-2 binding.) However, there is also a detectable trail of ST-1 complex forming activity across the entire ST-2 peak (Fig. 4A and data not shown). Such an ST-1 trail has been seen in all eight Affi-Gel blue columns, assayed using Subtel-ds, Tel-ds probes, or Subtel-G probes. This observation prompted examination of whether ST-1 may be part of the ST-2 complex. In that case, the ST-2 activity in the later fractions might be limited by the amount of ST-1 and might be augmented if additional ST-1 were provided in trans.

To test this possibility, we assayed the effect on ST-2 binding of adding homogeneously purified and renatured ST-1 protein (1) to the ST-2 containing fraction that eluted at 1.2 M KCl from Affi-Gel blue (Fig. 4B, lanes 4 and 6). Because assays performed using radiolabeled Subtel-ds could potentially show an indirect stimulation (for instance, ST-1 protein could bind to the C-rich strand, making the G-rich strand partly single-stranded, and thereby favor the binding of ST-2), we instead used a Subtel-G probe, which binds ST-2 very selectively (Ref 1; Fig. 2 and Fig. 3, A and B). Added ST-1 protein indeed markedly increased the amount of the ST-2 complex that the Affi-Gel blue ST-2 fraction is able to assemble (Fig. 4B, compare lanes 3-6). Notably, however, it does not affect the electrophoretic migration of the ST-2 complex (Fig. 4B).

Pursuing the stimulation of ST-2 complex formation by added purified ST-1 protein, we next examined the stimulation of various Affi-Gel blue fractions. In the early eluting, ST-2-containing fractions that also include considerable amounts of free ST-1, ST-2 complex formation is increased only a limited amount by addition of ST-1 (Fig. 4C, fractions 23-27), but in progressively higher salt eluting fractions, where ST-1 concentration decreases (1) and becomes more limiting, ST-2 complex formation is dramatically increased by addition of ST-1 (Fig. 4C, fraction 33-43). The highest salt eluting material showed virtually no ST-2 complex formation unless supplemented with ST-1 (fractions 47 of Fig. 4C and data not shown). Evidently the ST-2 shifting activity requires at least two trans-acting components, ST-1 and another activity whose elution from Affi-Gel blue partly overlaps with ST-1 but peaks at a higher KCl concentration.

The simplest explanation for the requirement of ST-1 for ST-2 complex formation is that ST-1 is an inherent component of the ST-2 complex. This would readily explain the observation that added ST-1 does not affect the electrophoretic mobility of the ST-2 complex on either Subtel-ds or Subtel-G DNA (Fig. 3, B and C), since ST-1 protein would always be part of the ST-2 DNA-binding complex. It would also provide a simple explanation for the observations that ST-1 can be detected trailing throughout the observed ST-2 peak (Fig. 4A), which still later eluting Affi-Gel blue fractions show ST-2 complex forming activity only when ST-1 is added (Fig. 4C) and that, although the ST-1 complex is extremely stable (1), its relative abundance decreases as more extract is added (Fig. 1), The concept that ST-1 is an inherent component of the ST-2 complex that forms on Subtel-ds is also supported by the observation that the amount of the apparent ST-1 complex on radiolabeled Subtel-ds probe actually increases with increasing amounts of Tel-G competitor (Fig. 2 and data not shown), which shows strong binding of ST-2 but no apparent binding of the ST-1 protein alone (Ref. 1; data not shown). Finally, an alternative possible explanation for the requirement of ST-1 in ST-2 complex formation (that ST-1 modifies the ST-2 proteins in some manner to allow their binding to the DNA but dissociates prior to the binding) is disfavored because ST-2 complex formation is enhanced by ST-1 addition at the time of DNA binding, even though the ST-2 factors had previously been in the presence of and then separated from excess ST-1 protein (Fig. 4, A and C).

To examine which components of the ST-2 complex directly contact Subtel-G DNA strand (that shows strong binding for the ST-2 complex but not the ST-1 complex alone) ST-2-rich Affi-Gel blue column fractions were incubated with radiolabeled Subtel-G, resolved by the gel shift assay and UV cross-linked in situ (Fig. 5A). The excised, gel-eluted complex was then resolved by SDS-PAGE and the radiolabeled polypeptides visualized by autoradiography (Fig. 5B). The Subtel-G strand in this ST-2 complex (Fig. 5A, lane 1, complex A) contacts three polypeptides in the range of 48 and 58 kDa and a doublet around 37 kDa (Fig. 5B, lane 1). (The uncomplexed size of these polypeptides should be a few kDa smaller, due to the contribution of the cross-linked Subtel-G oligomer.) The binding specificity of these polypeptides in the ST-2 complex (Fig. 5A, complex A) was verified by competition analysis. Specifically, the same radiolabeled polypeptides were observed (Fig. 5B, lane 2) using complex formed in the presence of a 50-fold excess of a nonspecific DNA oligomer, OL9 (Fig. 5A, lane 2), but they were virtually undetectable (Fig. 5B, lane 3) when the initial binding reaction instead contained a 50-fold excess of the specific competitor Subtel-G (Fig. 5A, lane 3). Furthermore, the same cross-linking profile as in Fig. 5B, lane 1 was obtained with ST-2 complexes formed on Subtel-G in 0.1 or 1 M KCl and with ST-2 complexes formed using extract from procyclic or bloodstream form trypanosomes (data not shown). As an additional control, a parallel analysis was also performed using renatured ST-1 (1) and Subtel-G DNA, which forms a low affinity complex when large amounts of the protein are introduced (complex C in Fig. 5A, lane 4). This complex, as well as the dimer-size one (complex B in Fig. 5A, lane 4), yielded a single band on the SDS-PAGE, with an apparent molecular mass of 42 kDa (Fig. 5B, lanes 4 and 5); this corresponds well with the known (1) 39-kDa molecular mass of ST-1 plus the cross-linked Subtel-G oligomer. Thus, all five of the cross-linked polypeptide bands of Fig. 5B, lanes 1 and 2 are due to a bona fide ST-2 complex.

It is striking that none of the cross-linked bands from the ST-2 complex (Fig. 5B, lanes 1 and 2) comigrated with that from ST-1 (Fig. 5B, lane 4). Thus, it appears that although ST-1 is part of the ST-2 complex (see above), it does not contact the G-rich strand of the subtelomere repeat directly. Instead, ST-1 may associate indirectly with the G-rich strand via other protein members of the ST-2 complex.

To examine whether some of the cross-linkable ST-2 polypeptides may be phosphorylated forms of one another, UV cross-linked ST-2 complex (as in Fig. 5A, lane 1) was treated with potato acid phosphatase (51). This treatment caused a faster migration of the largest UV cross-linkable polypeptide of the complex, indicating that it is a phosphorylated polypeptide (Fig. 5C). Because the phosphatase treatment did not decrease the total number of cross-linkable bands, this experiment also suggests that the various polypeptide bands of the ST-2 complex (Fig. 5B) are distinct and do not represent different phosphorylation states of a single polypeptide.

DISCUSSION

We have identified ST-2, a multisubunit activity from the parasite T. brucei that forms a specific complex on the duplex GGGTTA telomere repeats (Fig. 1A) and exhibits a higher affinity for the duplex 29-bp subtelomere repeats (Fig. 1B). Notably, however, ST-2 binds to the G-rich strands of the telomere and subtelomere repeats with at least 5-10-fold higher affinity than to the double-stranded forms of these repeats (Fig. 2 and Fig. 3A). The specificity and relative affinities of these bindings were demonstrated in pre-binding competition experiments where these trypanosome double- and single-stranded telomere and subtelomere sequences function as differentially efficient competitors of ST-2 binding (Figs. 1, 2, and 3B). In contrast, a large number of other examined double- and single-stranded DNAs did not compete for binding, including ones that are also highly G-rich (Table I).

Significantly, binding of ST-2 to the duplex form of the telomere and subtelomere repeats does not appear to be the consequence of its fully separating the two strands of the oligomer and then binding to the single-stranded G-rich single strand. This was shown by the presence of both DNA strands in the gel shifted complexes that ST-2 forms on duplex subtelomeric DNA (see "Results"). An analogous result was earlier observed for the ST-1 complex formed on duplex subtelomere DNA (1). We therefore conclude that although ST-2 may induce a partial denaturation of the substrate double-stranded DNA, it does not cause complete strand separation.

The properties of ST-2 are distinct from those of several G-strand telomere binding proteins reported from other species. Specifically, ST-2 binds with no greater affinity to the telomere G-strand present as a 3 protrusion on duplex DNA than when present as a free single strand (Fig. 3C), whereas known telomere capping proteins bind preferentially to the G-rich strand when present as a 3-overhang (24, 27). ST-2 is also unlike a recently reported gel shifting activity from Leptomonas that binds to the G-rich strand of the telomere repeats when present as a single strand but not when in a duplex structure (52).

We previously reported another telomere- and subtelomere-binding protein from T. brucei, ST-1 (1), a 39-kDa polypeptide whose binding properties are complementary to that of ST-2. Whereas ST-2 binds preferentially to the duplex version and more strongly to the G-rich strands of the subtelomere and telomere repeats, ST-1 binds preferentially to the duplex version and more strongly to the C-rich strands of both these sequences. Interestingly, ST-1 can exhibit two functions. In addition to binding the C-rich strand by itself (1), ST-1 was found to be a critical component for ST-2 complex formation, not only on duplex DNA but also on the G-rich subtelomere strand (Fig. 4). However, ST-1 does not appear to function by directly interacting with the G-rich strand of the DNA (Fig. 5B). Instead, five other polypeptides of the ST-2 complex, ranging between ~35 and ~55 kDa, directly interact with the G-rich strand of the DNA (Fig. 5B).

Taken together, these data imply that ST-2 is a telomeric/subtelomeric binding activity in T. brucei with unique features. Components of the multisubunit complex can evidently bridge between the two strands of subtelomeric and telomeric DNA, with five polypeptides binding to the G-rich strand and one polypeptide, ST-1, binding to the C-rich strand. Although the binding of ST-2 does not cause the two DNA strands to completely separate and move apart, the binding could cause the DNA to partially denature, with the two strands held in proximity by the bridging polypeptides. We speculate that also in vivo ST-2 holds subtelomeric DNA, and possibly telomere DNA, in a semi-denatured configuration. These binding properties could help to induce a state of open chromatin in the trypanosome telomeric region.

Unusual packaging of telomere DNA has been observed in all organisms studied so far. In Tetrahymena (46) and Saccharomyces (47), telomere chromatin is non-nucleosomal, and in several other invertebrates and vertebrates, telomere nucleosomes differ in length and spacing from bulk chromatin (44). This could be a reflection of the non-B form intrinsic structural properties of simple telomere repeats, such as the G-strand quartet or the C-strand i-motif. Such alternate structure could also be favored by the associated proteins. For example, the yeast protein RAP1, which is essential for telomere maintenance and the assembly of silencing telomeric chromatin (34), has recently been shown to distort telomeric DNA and to induce its folding into a four-stranded structure (35, 37). Moreover, the  subunit of the telosome in Oxytricha induces G-quartet folding of the telomere 3-overhang, which is speculated to function in telomere length regulation, meiotic interchromosomal associations, and chromosome stability. In addition, as noted above, our data indicate that the ST-2 protein complex of T. brucei could favor partial denaturation of the telomere repeats in this organism. How the peculiar conformations that telomeres adopt in vivo, due to their inherent properties and associated proteins, facilitate their different cellular functions remains to be discovered.

Less is known about subtelomere repeats, middle-repetitive elements found on several chromosomes separating the telomere DNA from internal genomic sequences (10, 13, 15, 17, 18), such as the telomere-derived 27-bp repeats in Plasmodium and 29-bp repeats in T. brucei (11, 12). These proximally abut the telomere repeats and can form the junction of the subtelomere repeat insertions, deletions, and amplifications which lead to chromosome size polymorphisms (11). The subtelomeric region is also the location of other chromosomal events, such as gene rearrangements and meiotic pairing. Until now, only one protein besides ST-1 had been found to bind to subtelomeric repeats, and this yeast TBF /1 protein evidently plays a role in telomere structure and position silencing effect. Our current data show that ST-2 may be another subtelomere binding protein complex that could affect the subtelomere DNA structure.

ST-1 was the first identified protein that binds subtelomeric repeats in T. brucei (1). Its affinity for the C-rich strands of the telomere and subtelomere repeats also gave a first indication of a T. brucei protein that binds to one of the two strands of telomeric/subtelomeric DNA. The characterization of a polypeptide complex, ST-2, with preferential affinity for the G-strands of the telomere and the subtelomere repeats now identifies a complementary activity. We speculate that, on duplex DNA, ST-2 binds to the G-rich strand and its component ST-1 binds to the C-rich strand, serving to place the subtelomeric/telomeric DNA in a partially denatured state in the cell. Although ST-2, with its component ST-1, does not fully denature the DNA it binds in vitro, it remains to be seen whether ST-2 may keep the telomeres and their neighboring sequences in an appreciably single-stranded configuration also in vivo, which could aid in the critical events that occur at the trypanosome telomeres, such as VSG gene diversification or silencing of telomeric genes (6, 7).