ABSTRACT
We have previously shown that nuclear DNA of bloodstream form
Trypanosoma brucei
contains a novel base
[beta]
-glucosyl-hydroxymethyluracil, called J. Base J is enriched in minichromosome
fractions but not in the minichromosome internal repeats, suggesting the
association of J with telomeric DNA. To test whether J is present in the long
telomeric (GGGTTA)
n
repeat arrays, which are 2-26 kb in
T.brucei
, we have purified these arrays both by hybrid selection and by isolating 2-26 kb fragments from DNA digested with multiple restriction enzymes. We
find that in purified telomeric repeats
~13% of T is replaced by J, compared to 0.8% in total DNA, and we estimate that
~50% of the total J is in these repeats. Highly purified complementary strands of
the repeats were obtained by alkaline CsCl equilibrium centrifugation. In the
(TAACCC)
n
strand 14% of T was replaced by J. In the (GGGTTA)
n
strand ~36% of the second T was replaced by J; the first T was not detectably replaced.
Modified bases have not been found in telomeric repeats before. How the bulky
base J affects telomere function and structure in bloodstream form trypanosomes
remains to be determined.
African trypanosomes have remarkable chromosome ends and there are many of them.
Most of these telomeres are in some 100 minichromosomes, that vary in size
between 50 and 150 kb (
1
). In addition there are ~20 larger chromosomes varying in size between 200 kb and 5.7 Mb (
1
-
3
).
In most organisms telomeric DNA consists of a tandemly repeated simple sequence
with a strong bias towards G residues in one of the strands (reviewed in
4
). The trypanosome telomeres end with (GGGTTA)
n
repeats (
5
,
6
), which are also found in mammalian chromosomes (reviewed in
7
) and are accompanied by fairly standard subtelomeric repeats (
6
,
8
-
10
). What makes these telomeres remarkable is the length of the repeat arrays
compared with the chromosome size and their spectacular growth and contraction
(
11
). The length of individual repeat arrays varies between 2 and 26 kb (
12
,
13
). Most telomeres grow at an average rate of 6-10 bp/cell division (
6
,
12
). Shortening may occur by large deletions (
11
), but what brings them on is not clear.
Telomeres are also associated with two interesting phenomena observed in
T.brucei
, antigenic variation and subtelomeric DNA modification.
Trypanosoma brucei
can survive in the bloodstream of its mammalian host by regularly changing
expression of its major coat protein, the variant surface glycoprotein (VSG).
To do so, trypanosomes have a large repertoire of VSG genes which are
exclusively expressed in one of the telomeric VSG expression sites (ESs)
(reviewed in
14
,
15
). There are at least six and probably ~20 ESs in a trypanosome nucleus and each of these is located near a
telomere. Usually, only one of these ESs is active at a time, with
transcription of VSG genes directed towards the chromosome end. Gottschling
et al.
(
16
) have suggested that the inactivation of all ESs but one occurs by telomeric
silencing. Recent studies are compatible with this idea (
17
,
18
).
DNA modification could be one of the mechanisms involved in silencing of ESs.
Transcriptional inactivation of an ES is accompanied by DNA modifications in
and around the inactivated telomeric VSG gene (
13
,
19
). This modification affects
Pst
I and
Pvu
II sites, it is found in non-transcribed VSG genes near telomeres, but not in silent chromosome-internal VSG genes and it is only found in bloodstream form
trypanosomes and not in insect form trypanosomes, which do not transcribe VSG
genes. Modification of any
Pst
I and
Pvu
II site is only present in a fraction of the trypanosome population and this
fraction increases with the length of the associated telomeric repeat array (
13
), linking DNA modification to telomeres. A novel hypermodified base, [beta]-D-glucosyl-hydroxymethyluracil ([beta]-gluc-HOMeU) or J, recently identified in
trypanosome DNA (
20
), has the characteristics expected for the modified base postulated to be
present in
Pst
I and
Pvu
II sites. J is present in bloodstream form trypanosomes and absent from insect
form trypanosomes (
21
). J is 6-fold enriched in minichromosomes, suggesting that it is associated with
chromosome ends.
One discrepancy between J and the previously detected subtelomeric DNA
modification remained, however: whereas the modification of subtelomeric
Pst
I and
Pvu
II sites would only require 0.0001 mol% J, we find 0.2% (
14
). A bulky base, such as J, may be expected to block cleavage by nearly all
restriction endonucleases (
22
) and it is therefore unlikely that substantially modified sites would have been missed. This leaves simple repeats not cut by the available restriction enzymes as potential candidates for J
localization, telomeric repeats being the most obvious ones. We have therefore
purified telomeric repeats of bloodstream form
T.brucei
and we show that on average as much as 16% of T residues in these repeats are
replaced by J.
221a bloodstream trypanosomes (MiTat1.2a) of
T.brucei
strain 427 (
23
) were grown and isolated as described (
21
). Procyclic (insect form) trypanosomes were grown in the semi-defined medium as described (
24
).
The fully protected J derivative 3'-
O
-[5-(2,3,4,6-tetra-
O
-benzoyl-[beta]- D-glucopyranosyloxymethyl)]-5'-
O
-dimethoxytrityl-2'-deoxyuridine 2-cyanoethyl-(
N,N
-diisopropyl)phosphoramidite (
25
) was used as a building unit in automated DNA synthesis (Pharmacia Gene
Assembler) of the J-containing oligonucleotides (GGGTJA)
4
, (GGGJTA)
4
, (GGGJJA)
4
and (ACCCJA)
4
. The synthetic oligomers were used as standards in nucleotide postlabelings to
determine the labeling efficiency of J.
Total genomic DNA was isolated as described (
26
) and resuspended in 10 mM Tris-HCl, 1 mM EDTA, pH 7.4. Digested DNA was transferred to nitrocellulose or
Hybond-N (Amersham) by standard procedures (
27
). For dot blots, DNA was denatured for 20 min on ice in 0.4 N NaOH, neutralized
by adding 1 vol. ice-cold 2 M ammonium acetate and blotted onto Hybond-N. Probes were labeled with [[alpha]-
32
P]dATP by random priming. Probes for telomeric repeats were either derived from
plasmid pT6, which contains a 330 bp stretch of duplex GGGTTA repeats (
6
), or were made by 5'
32
P-labeling of oligomers consisting of five telomeric GGGTTA or CCCTAA single-stranded repeats. Other
T.brucei
probes used were a subtelomeric sequence (
6
,
8
), the conserved 3'-half of VSG genes, 177, 70 (
28
) and 50 bp repeats (
29
), a 570 bp tubulin
Hin
dIII-
Bam
HI fragment from the [beta]-tubulin gene (
30
), ribosomal DNA (
31
) and kinetoplast DNA (
32
). Dot blots were scanned and quantitated on a phosphorimager (Fujix BAS 2000,
TINA 2.08b).
Genomic trypanosome DNA was sonicated to fragments of an average size of 250-500 bp, denatured by heating and incubated with biotinylated
oligonucleotides (CCCTAA)
4
or (GGGTTA)
4
, which had been coupled to magnetic Dynabeads as described by the manufacturer
(Dynal). DNA was selected in 50 mM sodium phosphate, pH 7.4, 0.9 M NaCl, 5 mM
EDTA, 0.1% SDS, 0.1% BSA and 0.2 mg/ml glycogen for 2 h at 50oC. The beads were then washed three times for 5 min at 50oC with 0.9 M NaCl, twice with 0.5 M NaCl and once with 0.1 M NaCl, all
in the presence of 50 mM sodium phosphate, pH 7.4, 0.1% SDS and 5 mM EDTA, and
finally incubated in 0.1 N NaOH for 10 min at room temperature. The
supernatant, which contained the selected DNA, was neutralized by adding 1 vol.
of a mixture of 0.1 N HCl, 0.2 M Tris-HCl, pH 7.4.
An aliquot of 500 [mu]g bloodstream trypanosome DNA was digested for 16 h with 1000 U
Alu
I,
Cfo
I,
Hin
fI,
Rsa
I,
Ssp
I and 500 U
Ava
II, deproteinized, phenol extracted and run on a preparative 1% LMP agarose gel.
DNA was eluted from the agarose according to Sambrook
et al.
(
27
).
DNA fragments were run in alkaline cesium chloride gradients [8 ml CsCl
(Suprapur; Merck) with a final density of 1.76 g/cm
3
in 0.1 N KOH, 1 mM EDTA] in polyallomer tubes (13 * 51 mm; Beckman) at 40 000 r.p.m. in a Beckman Ti50 rotor (107 000
g
) for 72 h at 20oC. Forty fractions of 200 [mu]l were collected from the bottom of the tube and 10 [mu]l of each fraction was used to measure the refractive index.
Samples of 2 [mu]l were analyzed by dot blot filter hybridization as described above, but
without adding NaOH. DNA from pooled fractions was desalted by ethanol
precipitation and analyzed by postlabeling for nucleotide composition.
32
P-Postlabeling combined with two-dimensional thin layer chromatography was done as described (
21
).
32
P-Labeled nucleotide 5'-monophosphates were quantitated by scintillation counting of
the separate spots or with a phosphorimager (Fujix). Gommers-Ampt
et al.
(
21
) have noted already that the labeling efficiency of J seemed to vary. This is
not surprising, as nucleotides with bulky adducts are known to be labeled
inefficiently (
33
). The chemical synthesis of J (
25
) made it possible to generate oligonucleotides with known amounts of J and test
the labeling efficiency of J-containing nucleotides directly. We found that only 36 +- 9% of J is recovered in the postlabeling assay. The
underestimation is probably due to a combination of inefficient labeling by
polynucleotide kinase and inefficient dephosphorylation by nuclease P1. We have
not succeeded in improving the labeling efficiency of J without severely
affecting the labeling of other nucleotides. Fortunately, the labeling
efficiency is not affected by excess DNA without J. This allowed us to correct
all data presented in this paper to 100% recovery by including samples with
known amounts of J in each series of post-labeling analyses. This correction resulted in a level of J in
T.brucei
DNA of 0.23 +- 0.06%, rather than the 0.04-0.10% of total DNA bases reported previously (
21
).
Depurination of DNA was performed by incubating DNA in 2% (w/v) diphenylamine
and 66% (v/v) aqueous formic acid for 17 h at 30oC as described (
34
). Pyrimidine tracts were purified by three extractions with 6 vol.
diethylether, dried by rotary evaporation and resuspended in H
2
O. The 5' pyrimidine was labeled by three consecutive reactions. First, the 5' terminal phosphate was removed by incubation with 1 U shrimp
alkaline phosphatase (USB) in the supplied reaction buffer for 30 min at 37oC in 10 [mu]l, after which the enzyme was inactivated by a 20 min incubation at 68oC. Second, 2 [mu]l of the dephosphorylation reaction was used to label the first pyrimidine with [[gamma]-
32
P]ATP and polynucleotide kinase (PNK) at the created free 5' hydroxyl as described (
21
), but in the absence of cold ATP. Third, the labeled 5' pyrimidine was released from the dimer by nuclease P1 (NP1), as
described for postlabeling (
21
). Labeling of the 3' pyrimidine was achieved by the standard post-labeling reactions for small amounts of DNA. Control experiments
with (GGGTJA)
4
and (GGGTTA)
5
showed that J at the 3'-T position of the dipyrimidine tract was labeled very inefficiently
compared to non-depurinated DNA and was mainly recovered as HOMeU. The degree of J
replacement (%J) in the standard curves and in the purified telomeric repeats
(Fig.
4
B) was therefore taken as the sum of HOMeU and J.
Trypanosoma brucei
has a small genome (8 * 10
7
bp/diploid nucleus) distributed over more than 100 chromosomes, with long
telomeric repeat segments of ~2-26 kb making up ~3% of the DNA. To rapidly check for DNA modifications in
telomeric repeats hybrid selection was employed, using the Dynal magnetic
separation system with two biotinylated oligonucleotides that consist of four
telomeric repeat units of either the (GGGTTA)
n
or the (TAACCC)
n
strand. Complementary DNA strands were selected from denatured bloodstream
trypanosome DNA that had been sonicated to small fragments of 250-500 bp. These small fragments can be efficiently selected and contain
minimal adjacent subtelomeric sequences that can be co-selected.
To analyze DNA modifications in genomic DNA we used nucleotide
32
P-postlabeling combined with separation by two-dimensional thin layer chromatography, a method developed by Reddy
et al.
(
35
) and applied to trypanosome DNA by Gommers-Ampt
et al.
(
21
). This assay allows separation of the standard nucleotides (indicated by their
base) A, T, C and G from hydroxymethyluracil (HOMeU) and [beta]-D-glucosyl-HOMeU (J). While 5-methylcytosine (MeC) is also efficiently labeled (
35
), this base is not detectable in total genomic DNA of trypanosomes (Fig.
1
C). As shown in Figure
1
D and E, the telomere-enriched DNA selected with both oligonucleotides contained high levels of
J. Quantitation of the nucleotide spots (see Materials and Methods for details)
showed that J was 10-fold enriched compared to total DNA. However, both biotin-(GGGTTA)
4
and biotin-(CCCTAA)
4
selected DNA in which A, T, C and G were present at equal levels (Fig.
1
F). As no release of oligonucleotides from the magnetic beads was detected, we
conclude that during hybrid selection of the (GGGTTA)
n
strand the complementary (TAACCC)
n
strand efficiently re-annealed with its partner (and vice versa) because of the ultrashort
repeats. Although it should in principle be possible to avoid re-annealing by using very dilute DNA solutions, this proved impractical. We
therefore turned to a combination of classical techniques to separately purify
both strands of intact telomeres.
Telomeric repeats lack recognition sequences for most type II restriction endonucleases. Digestion of DNA with frequent cutters and subsequent isolation of the high molecular weight DNA results in
DNA that contains most of the telomeric repeats, whereas the bulk of the DNA is
digested to small fragments (
9
). We used a combination of restriction enzymes that cut frequently and/or cut in
repetitive sequences known to be present in the
T.brucei
genome, such as subtelomeric repeats (Fig.
2
A). Bloodstream form trypanosome DNA was digested with
Alu
I,
Ava
II,
Cfo
I,
Hin
fI,
Rsa
I and
Ssp
I and run through a preparative agarose gel. The ethidium stain of the gel
confirmed that most of the DNA was in the bottom part, whereas telomeric
sequences, detected by Southern blotting, ran in the top part of the gel (Fig.
2
B). The gel was split into five fractions: low molecular weight DNA (LMW),
intermediate molecular weight DNA (IMW), high molecular weight DNA (HMW), slot
DNA and, as a control, DNA of one lane containing all four fractions (Fig.
2
B). DNA eluted from the agarose was analyzed by dot blot hybridization and
postlabeling.
In neutral CsCl gradients an almost linear relationship exists between the
percentage of (C+G) and the buoyant density of DNA unless particular unusual
bases are present (reviewed in
36
). As we found substantial amounts of base J in telomeric repeats, we tested
whether this affected the density of these repeats in neutral CsCl, as this
would allow separation of modified from normal DNA. However, telomeric repeats
of sonicated DNA from bloodstream trypanosomes and from insect form
trypanosomes (which does not contain J) both banded at 1.71 g/cm
3
, the density expected for DNA with 50 mol% G+C (data not shown).
In alkaline CsCl solutions mainly G and T residues are titrated and are thought
to acquire a cesium ion that increases the density of the DNA (
37
). This proved to be useful for further purification of telomeric repeats
because of their strong GT strand bias. In alkaline CsCl the (GGGTTA)
n
strand banded at 1.82 g/cm
3
and the (TAACCC)
n
strand at 1.72 g/cm
3
(Fig.
3
A). The complete separation of the strands allowed us to study each strand
separately, but also allowed separation of telomeric repeat DNA from sequences
without a strong GT strand bias, such as 70 bp repeats, kDNA, rDNA and tubulin
genes, which banded at a density of ~1.77 g/cm
3
. We therefore ran the HMW DNA fraction of Figure
2
A in an alkaline CsCl gradient (Fig.
3
B), pooled the peak fractions and analyzed them by postlabeling. Analysis of J
showed that the modified base is stable in alkaline CsCl and is present at high
levels in both strands (Fig.
3
C). The degree of modification is 4.9% in the (GGGTTA)
n
strand and 2.2% in the (TAACCC)
n
strand, which corresponds to 18 and 14% replacement of T respectively. The base
compositions of the separated strands were in close agreement with the
predicted sequences (Fig.
3
C), confirming the high degree of purity of each telomeric strand.
To analyze the position of J in the (GGGTTA)
n
strand, DNA was chemically depurinated (
34
) and the resulting dinucleotide pyrimidine tract was purified for differential
analysis of the 5'- and 3'-T positions, as outlined in Figure
4
A. This method was tested on synthetic oligonucleotides consisting of four
telomeric GGGTTA repeat units with J replacing T at either position. To mimic
the partial replacement of T by J found in telomeric repeats of
T.brucei
, we made dilution series of the J-containing oligomers in oligomers without J (Fig.
4
B). After depurination, a 5'-T, a 5'-J or a 3'-T in the dipyrimidine tract were labeled
with 99% efficiency, but a 3'-J was only partially labeled (Fig.
4
B). This partial labeling was only seen after chemical depurination. We used the
dilution series of (GGGTJA)
4
in (GGGTTA)
5
as a standard curve to determine the approximate degree of J replacement at the
3'-T position in the purified telomeric (GGGTTA)
n
strand of
T.brucei
(see Materials and Methods). We found that the measured 3% J corresponds to ~30% replacement of the 3'-T (Fig.
4
B). This high degree of replacement is in agreement with the 0.3% replacement
measured for the 5'-T, resulting in an average replacement of ~15% of T, close to the 18% replacement of T found in the non-depurinated GGGTTA strand (Fig.
3
C). The low level of J at the 5' position is probably the 1% non-specific labeling of the other position and may be caused by breaks
in the strand that have occurred during the experiment. These data clearly show
that the distribution of J in telomeric DNA is not random but that J replaced
only two of the three T residues, GGGT
T
A and
T
AACCC, with ~30-36 and 14% replacement of those T residues respectively.
Our results show that the telomeric repeats of
T.brucei
contain substantial amounts of J. This novel base replaces 14% of the T
residues in the (TAACCC)
n
strand and 18% in the (GGGTTA)
n
strand. Replacement of T in the G-rich strand is highly asymmetric, as the 5'-T was not significantly replaced by J. Even with this high
degree of replacement, not every telomeric repeat can contain J. Whether
telomeres without J exist and how J is distributed within telomeres remains to
be determined. We also found increased levels of HOMeU in telomeric DNA. This
could either be due to loss of glucose from J
in vivo
or during DNA isolation, or due to a lag in glycosylation of HOMeU during
synthesis of J. Indirect evidence suggests that J is made by modification of T
in DNA and that HOMeU is an intermediate in this process (
14
,
38
).
Trypanosomes have long telomeric repeat arrays, encompassing ~3% of their genome. An average replacement of 16% of T in telomeric repeats
versus 0.8% in total DNA implies that ~50% of the trypanosomal J is in telomeric repeats. The other half is mainly
in subtelomeric DNA, as suggested by the distribution of blocked
Pst
I and
Pvu
II sites (
13
,
19
,
21
) and of restriction fragments reacting with antibodies against J (unpublished results).
Exhaustive digestion of non-repetitive DNA with restriction endonucleases has been used before to
purify telomeric repeat arrays from mammals (
9
). Although this method resulted in highly purified trypanosomal telomeric
repeat DNA, this DNA still contained significant amounts of other simple
repeats, such as 70 bp repeats (Fig.
2
C). We therefore tried two other purification methods, hybrid selection and CsCl
equilibrium centrifugation. Hybrid selection is rapid, but we were unable to
get pure telomeric repeats by this procedure, even if the procedure was
successively repeated several times (results not shown). Isopycnic
centrifugation in Cs
2
SO
4
containing Ag
+
has been used before to obtain a limited enrichment of native human telomeric
DNA (
39
). Instead of this complex procedure, we have purified single-stranded DNA in alkaline CsCl. This approach is based on the extreme
difference in base composition of the complementary strands of the telomeric
repeats, allowing separation of these strands from each other and from bulk
DNA. The GT-rich strand bands at 1.82 g/cm
3
, close to the density of pure poly(dT[middot]dG), 1.83 g/cm
3
; the AC-rich strand at 1.72 g/cm
3
has an equilibrium density substantially above the 1.69 g/cm
3
reported for pure poly(dC[middot]dA) (
36
,
37
), but still comfortably away from bulk trypanosome DNA at 1.77 g/cm
3
(Fig.
3
A). This purification method may also be useful for telomeric repeats from other
sources.
In neutral CsCl gradients unusual bases like hydroxymethylcytosine (HOMeC) or
glucosylated HOMeC can affect the density of DNA (
35
,
36
). Despite the replacement of 16% of T by J ([beta]-gluc-HOMeU) in bloodstream form trypanosome telomeric repeats, we
found no detectable difference in density between bloodstream form and insect
form telomeric repeat arrays (data not shown). This suggests that either the
degree of modification is too low to give a change or that the expected
increase in density due to HOMeU (0.039 g/cm
3
if HOMeU replaced every T) is counteracted by a decrease caused by the glucose
moiety (the density of glucose is 1.6 g/cm
3
; the presence of glucose on part of HOMeC in T-even phages results in a decrease in density of 0.005 g/cm
3
;
35
).
Base J is not an essential structural component of telomeres in trypanosomes
because it is absent from insect stage trypanosomes (
20
,
21
). J must therefore be involved in a life cycle-specific telomeric function. The proposed role in shutting down (or
tightening the shut-down) of telomeric VSG gene expression sites (
13
,
14
,
18
) remains a plausible possibility. Nevertheless, it remains possible that J is
used for telomere-related functions in organisms other than African trypanosomes. Base
modification has never been found before in telomeric repeats, but in most
organisms the repeats represent such a small fraction of total DNA that the
presence of unusual nucleotides may have been overlooked. Our recent isolation
of J-specific antibodies (unpublished results) now provides the tools to check
whether J occurs elsewhere in nature.
We thank Janet Gommers-Ampt, Martin Taylor and Wilbert Bitter for helpful discussions, Francesca
Fase-Fowler for advice with the CsCl gradients, Pat Blundell, Mike Cross, Anita
Dirks-Mulder, Herlinde Gerrits, Henri van Luenen, Richard McCulloch, Rudo Kieft
and Gloria Rudenko for critical reading of the manuscript and Titia de Lange
(Rockefeller University, New York, NY) for many helpful comments. This work was
supported by grants from the Netherlands Foundation for Chemical Research, with
financial support of the Netherlands Organisation for Scientific Research
(NWO).
REFERENCES
Return