A yeast protein that binds to vertebrate telomeres and conserved yeast telomeric junctions Zhiping Liu and Bik-Kwoon Tye Section of Biochemistry, Molecular and Cell Biology, Cornell University, Ithaca, New York 14853 USA
We have identified three yeast proteins that bind to poly(C • A]/poly(T • G] repeats characteristic of telomeric sequences from yeast to human. TBFa binds to the telomeric sequences of yeast, Jbtrahymena, and vertebrates. In contrast, TBFp binds only to yeast telomeric sequences. Also identified was RAPl, the transcriptional silencer protein, which binds to a sequence motif found in upstream activating sequences (UASs) of a number of genes; the sequence motif also occurs frequently in yeast telomeric sequences. Because poly(C • A)/poly(T • G) sequences from a wide range of organisms will serve as the primer for the in vivo extension of telomeres in yeast, TBFa is of particular interest. DNase I footprinting analysis indicated that TBFa binds to the junction between the subtelomeric X sequence and poly(Ci_3A) in a cloned yeast telomere. Examination of the junctions of known X sequences indicated that they all contain one or more repeats of CCCTAA, a sequence that is repeated in vertebrate telomeres. Earlier, Murray et al. (1988) reported that heterologous telomeric sequences positioned as far as several hundred base pairs from the termini of linear molecules can allow the addition of yeast telomeric sequences from nontelomeric termini in vivo. A possible function for TBFa might be to serve as an anchoring protein for the yeast telomerase by binding to the conserved junction sequence at a distance from the terminus to allow addition of an irregular repeating sequence at the chromosome end. [Key Words: Telomere-binding factor; Saccharomyces cerevisiae; TBFa; TBFp telomere extension] Received September 13, 1990; revised version accepted November 13, 1990.
Telomere structure had been described mostly in theoretical terms (Cavalier-Smith 1974; Bateman 1975; Dancis and Holmquist 1979) on the basis of our perception of the functions of telomeres (McClintock 1941; Watson 1972), until the work of Szostak and Blackburn (1982), who showed that termini from ribosomal DNA of Tetrahymena can serve as telomeres in the propagation of linear plasmids in yeast. Their work provides a means to physically isolate functional telomeres not only from yeast but also from heterologous systems using yeast as a host (Schechtman 1987; Richards and Ausubel 1988; Brown 1989; Cross et al. 1989). Telomeres from a number of organisms have been cloned and characterized. They contain a tandem array of simple repeat sequences conforming to the general consensus (Ci_8)(T/A)i_4, with a conserved sequence polarity of the G-rich strand running 5' to 3' toward the terminus. This simple repeat sequence is Ci_3A in the budding yeast Saccharomyces cerevisiae, C4A2 in the ciliated protozoan Tetrahymena, and C3TA2 in human and all vertebrates examined so far. Sequence analysis of yeast-propagated Tetrahymena telomeres showed that these telomeres are chimeric products of Tetrahymena and yeast telomeric sequences with the yeast sequence added on to the Tetrahymena telomere (Shampay et al. 1984; Walmsley et al. 1984). Models invoking terminal exten-
sion by terminal transferase (Shampay et al. 1984; Murray et al. 1988) or recombination (Walmsley et al. 1984; Pluta and Zakian 1989) were proposed to explain the addition of yeast telomeric sequences to heterologous telomeres. Yeast telomeres are heterogeneous in length, ranging from 300 to 600 bp (Chan and Tye 1983a). Work from this laboratory showed that Ci_3A repeat sequences of heterogeneous lengths are not only found in the extreme termini of chromosomes but also at the junctions of moderately repetitive and highly conserved subtelomeric sequences {X and Y') next to autonomously replicating sequences (ARSs) (Fig. 1); (Walmsley et al. 1984; Chan 1985). No functions have yet been identified for the subtelomeric repetitive sequences. Because of the variation in their copy number (0-4) from telomere to telomere and complete absence in certain yeast strains, the Y' sequenoes-^were suggested to be derived from mobile elements that had high sequence specificity for integration at chromosome ends (Horowitz and Haber 1985; Jager and Philippsen 1989). The internal (Ci_3A)„ sequences were suggested to be the reservoir for telomeric sequences in the templated extension and unequal exchange models (Walmsley et al. 1984). Although numerous reports indicate that yeast telomeres undergo active recombinational exchanges (Dunn et al. 1984;
GENES & DEVELOPMENT 5:49-59 © 1991 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/91 $1.00
49
Liu and Tye
X
(a)
(6|-3T)r,
(C,_-,A)„
v^^^^^^i
-3'
(C,_^A)„
(b)
c
Figure 1. Structure of yeast telomeres. Yeast telomeres exist in either of two forms, a or b. Y' sequences are highly conserved repetitive sequences —6.7 kb in length. X sequences range from 0.5 to 4 kb in length, sharing varying degrees of homology to one another but no homology to Y' sequences.
Pluta and Zakian 1989) allovs^ing regeneration of telomeres (Surosky et al. 1986), the discovery of telomerase not only in Tetrahymena (Greider and Blackburn 1985) but also in human (Morin 1989), suggests that telomerase probably plays a major role in the replication of yeast telomeres as v^rell. The first telomere terminal transferase or telomerase activity was identified in Tetrahymena (Greider and Blackburn 1985). This activity is most abundant during macronuclear development of Tetrahymena when a large number of telomeres are formed and replicated. It is capable of adding tandem repeats of G4T2 to singlestranded DNA primers corresponding to telomeric repeats of Tetrahymena (G4T2), independent of endogenous Tetrahymena DNA. Telomerase is a ribonucleoprotein enzyme containing multiple protein components and at least one RNA component (Greider and Blackburn 1989). In Tetrahymena, this RNA component is 159 nucleotides in length and contains the sequence 5'-CAACCCCAA-3' believed to serve as the template for the synthesis of GGGGTT repeats. Repeated translocation of the enzyme results in repeated synthesis of the same sequence. Thus, telomerase is a specialized reverse transcriptase carrying its own RNA template, which specifies the signature sequence polymerized at telomeres in different organisms. Attempts to identify yeast telomerase activity in this and several other laboratories have been unsuccessful. Recently, Lundblad and Szostak (1989) reported the isolation of a mutant estl, which shows a progressive decrease in telomere length and a senescence phenotype. Although ESTl has been cited as a possible candidate for a subunit of the telomerase complex, based on homology of the ESTl open reading frame to motifs foimd in RNA-dependent polymerases (Limdblad and Blackburn 1990), biochemical fimction for the ESTl gene product has yet to be identified. If replication of yeast telomeres is carried out by an enzyme similar to telomerase, this enzyme could have an imprecise translocation or copying mechanism, or there could be multiple telomerases carrying different RNA templates. 50
GENES & DEVELOPMENT
Telomeres undergo dynamic interactions between lengthening and shortening processes. Activities such as synthesis of repeat sequences and recombination by unequal exchanges can contribute to the lengthening of telomeres. Incomplete replication of the lagging daughter strand and exonucleolytic degradation result in shortening of telomeres. To maintain an optimal balance between these many metabolic processes that take place at telomeres, many proteins may be expected to interact directly with telomeric sequences. Work on Oxytricha showed that a heterodimeric protein binding tightly to the single-stranded 3'-terminal (G4T4)2 tail may provide a telomeric capping function (Gottschling and Cech 1984; Gottschling and Zakian 1986). Previously, we have identified a telomere-binding activity from yeast extracts (Berman et al. 1986). This activity turned out to be the same as RAP I (Longtine et al. 1989), an abundant protein of —120 kD (Shore and Nasmyth 1987). RAPl binds to a consensus sequence that occurs in telomeric sequences, as well as in the silent mating type loci and in a number of gene promoters including TEF2 (Shore et al. 1987; Buchman et al. 1988a). On the basis of its binding specificities, RAPl has been suggested to play multiple roles in silencer function, transcriptional activation, and perhaps telomere maintenance. Re-examination of the properties of RAPl led us to search for additional telomere-binding proteins. In this paper we describe our identification of two novel proteins, telomere-binding factors, TBFa and TBF^, and our studies on some of their properties. We discuss the possible function for one of these proteins, TBFa.
Results The binding of RAPl to polyfCj^sA) is not sufficient for the mitotic segregation of plasmids RAPl binds to multiple sites on a 300-bp poly(Ci_3A) yeast telomeric sequence (Longtine et al. 1989). These binding sites conform to the consensus binding sequence 5'-(A/G)(A/C)ACCCANNCA(T/C)(T/C)-3' determined previously for the RAPl protein at the regulatory regions of the silent mating-type loci, HMLa and HMRa, and other promoter sequences (Buchman et al. 1988a). Recent reports suggest that RAPl may be associated with the nuclear scaffold (Hofmann et al. 1989). This scaffold association may explain the stability of plasmids containing the HMRE RAPl-binding site, which contributes a centromere-like segregation function (Brand et al. 1987; Kimmerly and Rine 1987). A plausible unified function served by RAPl at the many seemingly unrelated locations is that it mediates scaffold associations at these sites. If telomeric sequences are also scaffold bound via the action of RAPl, they should exhibit the same centromere-like segregation behavior as HMRE. Thus, one would expect that poly(Ci_ 3A) would be antagonistic to the presence of "another" centromere, rendering CEN plasmids unstable, but would stabilize ARS plasmids without a centromere. We cloned a poly(Ci_3A) sequence of —300 bp con-
Telomere binding proteins taining multiple RAP 1-binding sites from the left telomere of chromosome III into two almost identical plasmids: YRP74, which contains ARSl URA3, and YCP86, which contains CEN3 ARSl URA3 (plasmids were gifts from K. Struhl). As a control, we also cloned the HMRE sequence into these same plasmids. The stabilities of the resulting plasmids were examined in transformed yeast strains. We found that poly(Ci_3A) does not possess centromere-like functions as observed in HMRE (Table 1, see YRP74 vs. pRHE3 and YCP86 vs. pCHE3). It neither stabihzes ARS plasmids (cf. YRP74 with pRYT3) nor destabilizes CEN plasmids (cf. YCP86 with pCYT3). These results suggest that either the binding of RAPl alone has no effect on plasmid segregation or that RAPl does not bind to telomeric sequences in vivo. These results caution us not to overinterpret the role of RAPl in telomere function based solely on its binding to poly(Ci_3A) sequences in vitro. RAPl binds to yeast but not to other telomeric sequences
heterologous
Previous experiments showed that telomeric sequences from Tetrahymena (Szostak and Blackburn 1982), Oxytricha (Pluta et al. 1984), and human (Cross et al. 1989) can serve as primers for the extension of telomeres in yeast. We argue that proteins involved in the extension mechanism, whether as part of a telomerase-like machinery or a recombination machinery, should be able to bind to heterologous as well as yeast telomeric sequences. To further assess the role of RAPl in telomere functon, we examined the binding specificity of RAPl and found that among all telomeric sequences tested, it binds only to the yeast sequence (Fig. 2a). This result suggested to us that there must be other yeast telomerebinding proteins with DNA-binding properties consistent with the in vivo data on telomere propagation in yeast. Carrying out the same experiment using crude extracts indicated that there are proteins that bind specifically to the human (lanes 2 - 4 ) and Tetrahymena (lanes 6-8) telomeric sequences in yeast (Fig. 2b). Furthermore, cross competition of these substrates (lane 9) suggested that the same protein or protein complex is re-
Table 1. sites
Mitotic
stability of plasmids
bearing
RAPl-binding Stability
Plasmid
Insert
CEN3
YRp74 YCp86 pRHE3 pCHE3 pRYT3 pCYT3
-
+ -
HMRE HMRE (Ci_3A)„ (Ci_3A)n
+
-t-
(%)" 16.7 91.6 85.7 19.6 19.3 71.6
± ± ± ± ± ±
2.3 1.6 1.8 0.7 1.1 2.6
'Percent of cells with plasmid after 10 generations of nonselective growth. Plasmids were transformed into the yeast strain TDl {MATa his4-32 trpl-289 ma3-52].
Lane 1 2 3 Protein O 3 30
4 5 6 7 8 9 1 0 1 1 '3 ?. 30 300 O 3 30 30Q
tic Probe poly(q.3A) (C3TA2}27
(C3TA2)27
(04^2)50
(C4A2)5o
Figure 2. [a] Binding of RAPl protein to different telomere sequences. The binding reaction includes 3 fmoles of specific telomeric DNA probes, 2 jig of sonicated E. coli DNA, and partially purified £. coii-expressed RAPl protein, [b] Telomerebinding activities in crude yeast extracts. The binding reaction includes 3 fmoles of telomeric DNA and 8 |jLg (lanes 2-4] or 16 |jLg (lanes 6-9) of yeast extracts. Competitor DNA used are 2 \xg of sonicated E. coli DNAs (lanes 3 and 7), 100 fmoles of (C3TA2)27 (lanes 4 and 9), or 100 fmoles of (C4A2)5o (lane 8). (v) Vector sequence; (c) protein-DNA complex; (p) probe.
sponsible for the observed DNA-binding activity with specificity for telomeres in general. Other telomere-binding
proteins
Using different telomeric sequences as binding substrates in an agarose gel mobility shift assay (Berman et al. 1987), we were able to identify three activities from fractionated yeast extracts that bind to one or more of the telomeric sequences tested (Fig. 3a). Figure 3b shows the protein profile and the DNA-binding activity profile of eluates from phosphocellulose chromatography. The first activity peak (Fig. 3a, lanes 4-11) binds to the yeast telomeric sequence [138 bp of poly(Ci_3A) and 27 bp of X sequence; Fig. 3a(D)], as well as a restriction fragment containing a RAPl-binding site within the UAS of the TEF2 gene [Fig. 3a(A)]. This activity peak has been confirmed as that of the RAPl protein by antisera to RAPl protein expressed in Escherichia coli (Z. Liu, unpubl.). The second activity peak, which we call TBFa (Fig. 3a, lanes 10-14) binds to telomeric sequences of yeast [poly(Ci_3A) • X; Fig. 3a (D)], Tetrahymena [poly(C4A2); Fig. 3a(C)], and human [poly(C3TA2); Fig. 3a(B)]. Like RAPl, the third activity peak, TBFp (lanes 17-20), binds only to the yeast telomeric sequence, but unlike RAPl, it does not hiad to TEF2. Immunoprecipitation of fraction II incubated with yeast and vertebrate telomeric sequences using RAPl antibody enriches for the poly(Ci_ 3A)-binding activity but not the poly(C3TA2)-binding activity (data not shown). Thus, all three proteins, RAPl, TBFa, and TBFp, bind to poly(Ci_3A) • X but can be dif-
GENES & DEVELOPMENT
51
Liu and Tye
A.
probe: TEF-2
U •••KW* —P
!!»•»'w
B
r 0.10
orobe. i C ^ T A j l j y
10
C. probe- (C4 Ag) 50
15
2C
25
30
Fractiori
35
40
45
Number
|tff«r#fiPi||g!!;«ifiif*'-»
Figure 3. Phosphocellulose column chromatography of yeast telomere-binding proteins. Dialyzed L i it< . •' ;''.i i i ; i!i» *> '-i.• .oi JLii& j s ijs ifraction II [25-50% (NH4)2S04 cut] was loaded on QL £ sS O a S o . a 100-ml PC colurrm and developed with a o m ^ °in e 2 + probe: poly(C^ ,jA).X 150-600 mM linear KCl gradient in buffer A. [a] 2 H- o CL a . 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 18 19 20 Telomere-binding activities in the column frac12 3 4 5 6 mmmtmUmmmtim tions were assayed as described in Fig. 2 using one -V •f - f - l S f - f - l I—V ic of the following substrates: TEF2 [A], (CaTA^lj/ {Bl (C4A2)5o (C), poly(Ci_3A) • X (D). (Lane 1] No 4a( « • « • • • • - ' -p protein added; (lane 2) onput material (fraction II); (lane 3) flowthrough from PC column; (lanes 4-20] even-numbered fractions from 8 to 40. A 10-|xl aliquot of the fractions was added to each binding reaction. The same lane in each panel corresponds to the same fraction assayed, (v) Vector sequence; (c) complex; (p) probe, [b] Elution profiles of the different telomere-binding activities in PC chromatography. All column fractions were assayed. RAPl was determined by binding activity to the TEF2 promoter. TBFa was determined by binding activity to (C3TA2)4, which coincides with that for (C3TA2)27 and (C4A2)5o in a. TBFp was determined as binding activity that binds only to poly(Ci_3A) • X. (D) The protein concentration in each fraction, (c) Inactivation of TBFa-binding activity. Agarose gel band-shift assay was carried out as in a using (C,TA,) 27 as substrate. (Lane 1) No protein; (lanes 2-6] TBFa (1 |xg fraction III). Inactivation by 0.1% SDS (lane 3); 65°C, 10 min (lane 4); 1 )jLg proteinase K, 26°C 10 min (lane 5); 1 \xg proteinase K -I- 1 mM PMSF (lane 6). p
*•• »y*
•
••: ;•
::•;. p^ SB »
':^
:• •
.-:•:••
j c
' c S
ferentiated from one another based on their DNAbinding specificities for other substrates, their antigenicity, and their distinct chromatographic behavior on phosphocellulose chromatography (Fig. 3b). We pooled the peak fractions of TBFa and show^ed that this binding activity to poly(C3TA2) is sensitive to SDS, heat, and proteinase K treatment (Fig. 3c, lanes 3 - 5 ) . The effect of proteinase K on the binding activity was specific, because the presence of a serine protease inhibitor blocked the inactivation (lane 6). These results suggest that TBFa is a protein or at least contains a protein component. TBFa is distinct from the RAPl
protein
Because of the abundance of RAPl, the TBFa activity peak (fraction III) from the phosphocellulose column is still rich in the RAPl protein. To purify TBFa from RAPl, vv^e carried out D N A column chromatography (Alberts and Flerrick 1971) by passing fraction III through a DNA cellulose column coupled v\rith plasmid DNA containing (C3TA2)27. Because RAPl does not bind specifically to poly(C3TA2), it w^as eluted at low^ salt, whereas TBFa w^as eluted at higher salt as a sequence52
GENES & DEVELOPMENT
CO ^ Q ^
05 ) ^ 9 ?0 21 22 23
Figure 4. DNA column chromatography of TBFa. Fraction III was loaded onto a plasmid DNA column containing (C3TA2)27After an extensive wash at 0.1 M KCI, the column was developed using two linear KCI gradients, 0.1-0.3 M and 0.3-0.6 M, with a three-column wash at 0.3 M KCI in between. The elution of RAP I and TBFa-binding activities was monitored by agarose gel band-shift assay with four different DNA probes (3 fmoles DNA per binding reaction), shown as four different panels. A 10-|JL1 aliquot of the fractions was added to each binding reaction. (Lane 1) No protein added; (lane 2) onput material (fraction III); (lanes 3-8] even-numbered fractions from the low salt gradient; (lanes 9 and 10) fractions from 0.3 M KCI step wash; (lanes 11-20) even-numbered fractions from the high salt gradient elution. (v) Vector DNA; (p) probe DNA; (c) proteinDNA complex.
205 Kd —
116 Kd — 97 Kd — 66 Kd —
45 Kd — 29 Kd —
FRACTION PC S 0
to proteinase K treatment, we wanted to identify the polypeptide that copurifies with the binding activity by examining column fractions using SDS-polyacrylamide gel electrophoresis. To remove DNA released from the DNA column as a result of nuclease contamination in fraction III, we pooled the activity peak (fraction IV) of TBFa resolved by D N A chromatography and once again passed it through a phosphocellulose column. The TBFa activity, assayed by gel shift (Fig. 5b, fractions 15-20), copurifies with a protein with an apparent molecular mass of 63 kD (Fig. 5a, fractions 15-20), which is considerably lower than that of RAPl (120 kD; Shore and Nasmyth 1987). This 63-kD protein also copurifies with the TBFa-binding activity on the DNA column (data not shown). There are other minor proteins that seem to copurify with this activity. At this time we are unable to discern whether TBFa contains multiple subunits or
(CaTAg)^.
11:-^
^^^.0f, ^ ^ mmmmm^mim -p
Figure 5. A 63-kD protein copurified with TBFa-binding activity. The DNA column fractions containing TBFa-binding activity were pooled (fraction IV) and concentrated on a 1.0-ml phosphocellulose column. TBFa was eluted with a 0.15-0.6 M linear KCI gradient, [a] Silver staining of the elution profile. Aliquots of 70-|xI of each fraction, from fraction 8 to 23, were precipitated with 15% TCA, resuspended in cracking buffer, and loaded on a 8% SDS-polyacrylamide gel. [Lane 1 (PC)] 6 |xg protein from fraction III (DNA column onput) was loaded. Arrowhead indicates position of 63-kD protein, {b) Agarose gel band-shift assay for the column fraction. A 5-^,1 aliquot of each fraction, from fractions 8 to 23, was added to the binding reaction containing 3 fmoles of (C3TA2)4 oligonucleotide, and proceeded as described in Materials and methods. (Lane 1) The binding reaction contained 1.2 |xg proteins from fraction III and 3 fmoles of (C3TA2)4 oligonucleotide. GENES & DEVELOPMENT
53
Liu and Tye
C-rich (Fig. 6a, left) and the G-rich (Fig. 6a, right) strands is shown. A region of —34 bp located at the junction of X and poly(Ci_3A) sequence is protected from DNase I digestion by TBFa. There are two copies of the sequence ACCCTA(A/C) located in this protected region at the junction (Fig. 6b). Although gel shift assay indicates that there is only a single retarded complex formed under the condition in which the footprinting was carried out, we believe that this region consists of two tandem footprints of TBFa, one (I) spanning the junction sequence (J) and another (II) lying within the X sequence, each pro-
a c-rich Strand
G-rich Strand
< I-QI-ECJP3£ Q i - a :
We have identified three yeast proteins that bind to poly(C • A]/poly(T • G] repeats characteristic of telomeric sequences from yeast to human. TBFa binds to the telomeric sequences of yeast, Jbtrahymena, and vertebrates. In contrast, TBFp binds only to yeast telomeric sequences. Also identified was RAPl, the transcriptional silencer protein, which binds to a sequence motif found in upstream activating sequences (UASs) of a number of genes; the sequence motif also occurs frequently in yeast telomeric sequences. Because poly(C • A)/poly(T • G) sequences from a wide range of organisms will serve as the primer for the in vivo extension of telomeres in yeast, TBFa is of particular interest. DNase I footprinting analysis indicated that TBFa binds to the junction between the subtelomeric X sequence and poly(Ci_3A) in a cloned yeast telomere. Examination of the junctions of known X sequences indicated that they all contain one or more repeats of CCCTAA, a sequence that is repeated in vertebrate telomeres. Earlier, Murray et al. (1988) reported that heterologous telomeric sequences positioned as far as several hundred base pairs from the termini of linear molecules can allow the addition of yeast telomeric sequences from nontelomeric termini in vivo. A possible function for TBFa might be to serve as an anchoring protein for the yeast telomerase by binding to the conserved junction sequence at a distance from the terminus to allow addition of an irregular repeating sequence at the chromosome end. [Key Words: Telomere-binding factor; Saccharomyces cerevisiae; TBFa; TBFp telomere extension] Received September 13, 1990; revised version accepted November 13, 1990.
Telomere structure had been described mostly in theoretical terms (Cavalier-Smith 1974; Bateman 1975; Dancis and Holmquist 1979) on the basis of our perception of the functions of telomeres (McClintock 1941; Watson 1972), until the work of Szostak and Blackburn (1982), who showed that termini from ribosomal DNA of Tetrahymena can serve as telomeres in the propagation of linear plasmids in yeast. Their work provides a means to physically isolate functional telomeres not only from yeast but also from heterologous systems using yeast as a host (Schechtman 1987; Richards and Ausubel 1988; Brown 1989; Cross et al. 1989). Telomeres from a number of organisms have been cloned and characterized. They contain a tandem array of simple repeat sequences conforming to the general consensus (Ci_8)(T/A)i_4, with a conserved sequence polarity of the G-rich strand running 5' to 3' toward the terminus. This simple repeat sequence is Ci_3A in the budding yeast Saccharomyces cerevisiae, C4A2 in the ciliated protozoan Tetrahymena, and C3TA2 in human and all vertebrates examined so far. Sequence analysis of yeast-propagated Tetrahymena telomeres showed that these telomeres are chimeric products of Tetrahymena and yeast telomeric sequences with the yeast sequence added on to the Tetrahymena telomere (Shampay et al. 1984; Walmsley et al. 1984). Models invoking terminal exten-
sion by terminal transferase (Shampay et al. 1984; Murray et al. 1988) or recombination (Walmsley et al. 1984; Pluta and Zakian 1989) were proposed to explain the addition of yeast telomeric sequences to heterologous telomeres. Yeast telomeres are heterogeneous in length, ranging from 300 to 600 bp (Chan and Tye 1983a). Work from this laboratory showed that Ci_3A repeat sequences of heterogeneous lengths are not only found in the extreme termini of chromosomes but also at the junctions of moderately repetitive and highly conserved subtelomeric sequences {X and Y') next to autonomously replicating sequences (ARSs) (Fig. 1); (Walmsley et al. 1984; Chan 1985). No functions have yet been identified for the subtelomeric repetitive sequences. Because of the variation in their copy number (0-4) from telomere to telomere and complete absence in certain yeast strains, the Y' sequenoes-^were suggested to be derived from mobile elements that had high sequence specificity for integration at chromosome ends (Horowitz and Haber 1985; Jager and Philippsen 1989). The internal (Ci_3A)„ sequences were suggested to be the reservoir for telomeric sequences in the templated extension and unequal exchange models (Walmsley et al. 1984). Although numerous reports indicate that yeast telomeres undergo active recombinational exchanges (Dunn et al. 1984;
GENES & DEVELOPMENT 5:49-59 © 1991 by Cold Spring Harbor Laboratory Press ISSN 0890-9369/91 $1.00
49
Liu and Tye
X
(a)
(6|-3T)r,
(C,_-,A)„
v^^^^^^i
-3'
(C,_^A)„
(b)
c
Figure 1. Structure of yeast telomeres. Yeast telomeres exist in either of two forms, a or b. Y' sequences are highly conserved repetitive sequences —6.7 kb in length. X sequences range from 0.5 to 4 kb in length, sharing varying degrees of homology to one another but no homology to Y' sequences.
Pluta and Zakian 1989) allovs^ing regeneration of telomeres (Surosky et al. 1986), the discovery of telomerase not only in Tetrahymena (Greider and Blackburn 1985) but also in human (Morin 1989), suggests that telomerase probably plays a major role in the replication of yeast telomeres as v^rell. The first telomere terminal transferase or telomerase activity was identified in Tetrahymena (Greider and Blackburn 1985). This activity is most abundant during macronuclear development of Tetrahymena when a large number of telomeres are formed and replicated. It is capable of adding tandem repeats of G4T2 to singlestranded DNA primers corresponding to telomeric repeats of Tetrahymena (G4T2), independent of endogenous Tetrahymena DNA. Telomerase is a ribonucleoprotein enzyme containing multiple protein components and at least one RNA component (Greider and Blackburn 1989). In Tetrahymena, this RNA component is 159 nucleotides in length and contains the sequence 5'-CAACCCCAA-3' believed to serve as the template for the synthesis of GGGGTT repeats. Repeated translocation of the enzyme results in repeated synthesis of the same sequence. Thus, telomerase is a specialized reverse transcriptase carrying its own RNA template, which specifies the signature sequence polymerized at telomeres in different organisms. Attempts to identify yeast telomerase activity in this and several other laboratories have been unsuccessful. Recently, Lundblad and Szostak (1989) reported the isolation of a mutant estl, which shows a progressive decrease in telomere length and a senescence phenotype. Although ESTl has been cited as a possible candidate for a subunit of the telomerase complex, based on homology of the ESTl open reading frame to motifs foimd in RNA-dependent polymerases (Limdblad and Blackburn 1990), biochemical fimction for the ESTl gene product has yet to be identified. If replication of yeast telomeres is carried out by an enzyme similar to telomerase, this enzyme could have an imprecise translocation or copying mechanism, or there could be multiple telomerases carrying different RNA templates. 50
GENES & DEVELOPMENT
Telomeres undergo dynamic interactions between lengthening and shortening processes. Activities such as synthesis of repeat sequences and recombination by unequal exchanges can contribute to the lengthening of telomeres. Incomplete replication of the lagging daughter strand and exonucleolytic degradation result in shortening of telomeres. To maintain an optimal balance between these many metabolic processes that take place at telomeres, many proteins may be expected to interact directly with telomeric sequences. Work on Oxytricha showed that a heterodimeric protein binding tightly to the single-stranded 3'-terminal (G4T4)2 tail may provide a telomeric capping function (Gottschling and Cech 1984; Gottschling and Zakian 1986). Previously, we have identified a telomere-binding activity from yeast extracts (Berman et al. 1986). This activity turned out to be the same as RAP I (Longtine et al. 1989), an abundant protein of —120 kD (Shore and Nasmyth 1987). RAPl binds to a consensus sequence that occurs in telomeric sequences, as well as in the silent mating type loci and in a number of gene promoters including TEF2 (Shore et al. 1987; Buchman et al. 1988a). On the basis of its binding specificities, RAPl has been suggested to play multiple roles in silencer function, transcriptional activation, and perhaps telomere maintenance. Re-examination of the properties of RAPl led us to search for additional telomere-binding proteins. In this paper we describe our identification of two novel proteins, telomere-binding factors, TBFa and TBF^, and our studies on some of their properties. We discuss the possible function for one of these proteins, TBFa.
Results The binding of RAPl to polyfCj^sA) is not sufficient for the mitotic segregation of plasmids RAPl binds to multiple sites on a 300-bp poly(Ci_3A) yeast telomeric sequence (Longtine et al. 1989). These binding sites conform to the consensus binding sequence 5'-(A/G)(A/C)ACCCANNCA(T/C)(T/C)-3' determined previously for the RAPl protein at the regulatory regions of the silent mating-type loci, HMLa and HMRa, and other promoter sequences (Buchman et al. 1988a). Recent reports suggest that RAPl may be associated with the nuclear scaffold (Hofmann et al. 1989). This scaffold association may explain the stability of plasmids containing the HMRE RAPl-binding site, which contributes a centromere-like segregation function (Brand et al. 1987; Kimmerly and Rine 1987). A plausible unified function served by RAPl at the many seemingly unrelated locations is that it mediates scaffold associations at these sites. If telomeric sequences are also scaffold bound via the action of RAPl, they should exhibit the same centromere-like segregation behavior as HMRE. Thus, one would expect that poly(Ci_ 3A) would be antagonistic to the presence of "another" centromere, rendering CEN plasmids unstable, but would stabilize ARS plasmids without a centromere. We cloned a poly(Ci_3A) sequence of —300 bp con-
Telomere binding proteins taining multiple RAP 1-binding sites from the left telomere of chromosome III into two almost identical plasmids: YRP74, which contains ARSl URA3, and YCP86, which contains CEN3 ARSl URA3 (plasmids were gifts from K. Struhl). As a control, we also cloned the HMRE sequence into these same plasmids. The stabilities of the resulting plasmids were examined in transformed yeast strains. We found that poly(Ci_3A) does not possess centromere-like functions as observed in HMRE (Table 1, see YRP74 vs. pRHE3 and YCP86 vs. pCHE3). It neither stabihzes ARS plasmids (cf. YRP74 with pRYT3) nor destabilizes CEN plasmids (cf. YCP86 with pCYT3). These results suggest that either the binding of RAPl alone has no effect on plasmid segregation or that RAPl does not bind to telomeric sequences in vivo. These results caution us not to overinterpret the role of RAPl in telomere function based solely on its binding to poly(Ci_3A) sequences in vitro. RAPl binds to yeast but not to other telomeric sequences
heterologous
Previous experiments showed that telomeric sequences from Tetrahymena (Szostak and Blackburn 1982), Oxytricha (Pluta et al. 1984), and human (Cross et al. 1989) can serve as primers for the extension of telomeres in yeast. We argue that proteins involved in the extension mechanism, whether as part of a telomerase-like machinery or a recombination machinery, should be able to bind to heterologous as well as yeast telomeric sequences. To further assess the role of RAPl in telomere functon, we examined the binding specificity of RAPl and found that among all telomeric sequences tested, it binds only to the yeast sequence (Fig. 2a). This result suggested to us that there must be other yeast telomerebinding proteins with DNA-binding properties consistent with the in vivo data on telomere propagation in yeast. Carrying out the same experiment using crude extracts indicated that there are proteins that bind specifically to the human (lanes 2 - 4 ) and Tetrahymena (lanes 6-8) telomeric sequences in yeast (Fig. 2b). Furthermore, cross competition of these substrates (lane 9) suggested that the same protein or protein complex is re-
Table 1. sites
Mitotic
stability of plasmids
bearing
RAPl-binding Stability
Plasmid
Insert
CEN3
YRp74 YCp86 pRHE3 pCHE3 pRYT3 pCYT3
-
+ -
HMRE HMRE (Ci_3A)„ (Ci_3A)n
+
-t-
(%)" 16.7 91.6 85.7 19.6 19.3 71.6
± ± ± ± ± ±
2.3 1.6 1.8 0.7 1.1 2.6
'Percent of cells with plasmid after 10 generations of nonselective growth. Plasmids were transformed into the yeast strain TDl {MATa his4-32 trpl-289 ma3-52].
Lane 1 2 3 Protein O 3 30
4 5 6 7 8 9 1 0 1 1 '3 ?. 30 300 O 3 30 30Q
tic Probe poly(q.3A) (C3TA2}27
(C3TA2)27
(04^2)50
(C4A2)5o
Figure 2. [a] Binding of RAPl protein to different telomere sequences. The binding reaction includes 3 fmoles of specific telomeric DNA probes, 2 jig of sonicated E. coli DNA, and partially purified £. coii-expressed RAPl protein, [b] Telomerebinding activities in crude yeast extracts. The binding reaction includes 3 fmoles of telomeric DNA and 8 |jLg (lanes 2-4] or 16 |jLg (lanes 6-9) of yeast extracts. Competitor DNA used are 2 \xg of sonicated E. coli DNAs (lanes 3 and 7), 100 fmoles of (C3TA2)27 (lanes 4 and 9), or 100 fmoles of (C4A2)5o (lane 8). (v) Vector sequence; (c) protein-DNA complex; (p) probe.
sponsible for the observed DNA-binding activity with specificity for telomeres in general. Other telomere-binding
proteins
Using different telomeric sequences as binding substrates in an agarose gel mobility shift assay (Berman et al. 1987), we were able to identify three activities from fractionated yeast extracts that bind to one or more of the telomeric sequences tested (Fig. 3a). Figure 3b shows the protein profile and the DNA-binding activity profile of eluates from phosphocellulose chromatography. The first activity peak (Fig. 3a, lanes 4-11) binds to the yeast telomeric sequence [138 bp of poly(Ci_3A) and 27 bp of X sequence; Fig. 3a(D)], as well as a restriction fragment containing a RAPl-binding site within the UAS of the TEF2 gene [Fig. 3a(A)]. This activity peak has been confirmed as that of the RAPl protein by antisera to RAPl protein expressed in Escherichia coli (Z. Liu, unpubl.). The second activity peak, which we call TBFa (Fig. 3a, lanes 10-14) binds to telomeric sequences of yeast [poly(Ci_3A) • X; Fig. 3a (D)], Tetrahymena [poly(C4A2); Fig. 3a(C)], and human [poly(C3TA2); Fig. 3a(B)]. Like RAPl, the third activity peak, TBFp (lanes 17-20), binds only to the yeast telomeric sequence, but unlike RAPl, it does not hiad to TEF2. Immunoprecipitation of fraction II incubated with yeast and vertebrate telomeric sequences using RAPl antibody enriches for the poly(Ci_ 3A)-binding activity but not the poly(C3TA2)-binding activity (data not shown). Thus, all three proteins, RAPl, TBFa, and TBFp, bind to poly(Ci_3A) • X but can be dif-
GENES & DEVELOPMENT
51
Liu and Tye
A.
probe: TEF-2
U •••KW* —P
!!»•»'w
B
r 0.10
orobe. i C ^ T A j l j y
10
C. probe- (C4 Ag) 50
15
2C
25
30
Fractiori
35
40
45
Number
|tff«r#fiPi||g!!;«ifiif*'-»
Figure 3. Phosphocellulose column chromatography of yeast telomere-binding proteins. Dialyzed L i it< . •' ;''.i i i ; i!i» *> '-i.• .oi JLii& j s ijs ifraction II [25-50% (NH4)2S04 cut] was loaded on QL £ sS O a S o . a 100-ml PC colurrm and developed with a o m ^ °in e 2 + probe: poly(C^ ,jA).X 150-600 mM linear KCl gradient in buffer A. [a] 2 H- o CL a . 2 3 4 5 6 7 9 10 11 12 13 14 15 16 17 18 19 20 Telomere-binding activities in the column frac12 3 4 5 6 mmmtmUmmmtim tions were assayed as described in Fig. 2 using one -V •f - f - l S f - f - l I—V ic of the following substrates: TEF2 [A], (CaTA^lj/ {Bl (C4A2)5o (C), poly(Ci_3A) • X (D). (Lane 1] No 4a( « • « • • • • - ' -p protein added; (lane 2) onput material (fraction II); (lane 3) flowthrough from PC column; (lanes 4-20] even-numbered fractions from 8 to 40. A 10-|xl aliquot of the fractions was added to each binding reaction. The same lane in each panel corresponds to the same fraction assayed, (v) Vector sequence; (c) complex; (p) probe, [b] Elution profiles of the different telomere-binding activities in PC chromatography. All column fractions were assayed. RAPl was determined by binding activity to the TEF2 promoter. TBFa was determined by binding activity to (C3TA2)4, which coincides with that for (C3TA2)27 and (C4A2)5o in a. TBFp was determined as binding activity that binds only to poly(Ci_3A) • X. (D) The protein concentration in each fraction, (c) Inactivation of TBFa-binding activity. Agarose gel band-shift assay was carried out as in a using (C,TA,) 27 as substrate. (Lane 1) No protein; (lanes 2-6] TBFa (1 |xg fraction III). Inactivation by 0.1% SDS (lane 3); 65°C, 10 min (lane 4); 1 )jLg proteinase K, 26°C 10 min (lane 5); 1 \xg proteinase K -I- 1 mM PMSF (lane 6). p
*•• »y*
•
••: ;•
::•;. p^ SB »
':^
:• •
.-:•:••
j c
' c S
ferentiated from one another based on their DNAbinding specificities for other substrates, their antigenicity, and their distinct chromatographic behavior on phosphocellulose chromatography (Fig. 3b). We pooled the peak fractions of TBFa and show^ed that this binding activity to poly(C3TA2) is sensitive to SDS, heat, and proteinase K treatment (Fig. 3c, lanes 3 - 5 ) . The effect of proteinase K on the binding activity was specific, because the presence of a serine protease inhibitor blocked the inactivation (lane 6). These results suggest that TBFa is a protein or at least contains a protein component. TBFa is distinct from the RAPl
protein
Because of the abundance of RAPl, the TBFa activity peak (fraction III) from the phosphocellulose column is still rich in the RAPl protein. To purify TBFa from RAPl, vv^e carried out D N A column chromatography (Alberts and Flerrick 1971) by passing fraction III through a DNA cellulose column coupled v\rith plasmid DNA containing (C3TA2)27. Because RAPl does not bind specifically to poly(C3TA2), it w^as eluted at low^ salt, whereas TBFa w^as eluted at higher salt as a sequence52
GENES & DEVELOPMENT
CO ^ Q ^
05 ) ^ 9 ?0 21 22 23
Figure 4. DNA column chromatography of TBFa. Fraction III was loaded onto a plasmid DNA column containing (C3TA2)27After an extensive wash at 0.1 M KCI, the column was developed using two linear KCI gradients, 0.1-0.3 M and 0.3-0.6 M, with a three-column wash at 0.3 M KCI in between. The elution of RAP I and TBFa-binding activities was monitored by agarose gel band-shift assay with four different DNA probes (3 fmoles DNA per binding reaction), shown as four different panels. A 10-|JL1 aliquot of the fractions was added to each binding reaction. (Lane 1) No protein added; (lane 2) onput material (fraction III); (lanes 3-8] even-numbered fractions from the low salt gradient; (lanes 9 and 10) fractions from 0.3 M KCI step wash; (lanes 11-20) even-numbered fractions from the high salt gradient elution. (v) Vector DNA; (p) probe DNA; (c) proteinDNA complex.
205 Kd —
116 Kd — 97 Kd — 66 Kd —
45 Kd — 29 Kd —
FRACTION PC S 0
to proteinase K treatment, we wanted to identify the polypeptide that copurifies with the binding activity by examining column fractions using SDS-polyacrylamide gel electrophoresis. To remove DNA released from the DNA column as a result of nuclease contamination in fraction III, we pooled the activity peak (fraction IV) of TBFa resolved by D N A chromatography and once again passed it through a phosphocellulose column. The TBFa activity, assayed by gel shift (Fig. 5b, fractions 15-20), copurifies with a protein with an apparent molecular mass of 63 kD (Fig. 5a, fractions 15-20), which is considerably lower than that of RAPl (120 kD; Shore and Nasmyth 1987). This 63-kD protein also copurifies with the TBFa-binding activity on the DNA column (data not shown). There are other minor proteins that seem to copurify with this activity. At this time we are unable to discern whether TBFa contains multiple subunits or
(CaTAg)^.
11:-^
^^^.0f, ^ ^ mmmmm^mim -p
Figure 5. A 63-kD protein copurified with TBFa-binding activity. The DNA column fractions containing TBFa-binding activity were pooled (fraction IV) and concentrated on a 1.0-ml phosphocellulose column. TBFa was eluted with a 0.15-0.6 M linear KCI gradient, [a] Silver staining of the elution profile. Aliquots of 70-|xI of each fraction, from fraction 8 to 23, were precipitated with 15% TCA, resuspended in cracking buffer, and loaded on a 8% SDS-polyacrylamide gel. [Lane 1 (PC)] 6 |xg protein from fraction III (DNA column onput) was loaded. Arrowhead indicates position of 63-kD protein, {b) Agarose gel band-shift assay for the column fraction. A 5-^,1 aliquot of each fraction, from fractions 8 to 23, was added to the binding reaction containing 3 fmoles of (C3TA2)4 oligonucleotide, and proceeded as described in Materials and methods. (Lane 1) The binding reaction contained 1.2 |xg proteins from fraction III and 3 fmoles of (C3TA2)4 oligonucleotide. GENES & DEVELOPMENT
53
Liu and Tye
C-rich (Fig. 6a, left) and the G-rich (Fig. 6a, right) strands is shown. A region of —34 bp located at the junction of X and poly(Ci_3A) sequence is protected from DNase I digestion by TBFa. There are two copies of the sequence ACCCTA(A/C) located in this protected region at the junction (Fig. 6b). Although gel shift assay indicates that there is only a single retarded complex formed under the condition in which the footprinting was carried out, we believe that this region consists of two tandem footprints of TBFa, one (I) spanning the junction sequence (J) and another (II) lying within the X sequence, each pro-
a c-rich Strand
G-rich Strand
< I-QI-ECJP3£ Q i - a :
