The MRT-1 nuclease is required for DNA crosslink repair and telomerase activity in vivo in Caenorhabditis elegans

Abstract

The telomerase reverse transcriptase adds de novo DNA repeats to chromosome termini. Here we define Caenorhabditis elegans MRT-1 as a novel factor required for telomerase-mediated telomere replication and the DNA-damage response. MRT-1 is composed of an N-terminal domain homologous to the second OB-fold of POT1 telomere-binding proteins and a C-terminal SNM1 family nuclease domain, which confer single-strand DNA-binding and processive 3′-to-5′ exonuclease activity, respectively. Furthermore, telomerase activity in vivo depends on a functional MRT-1 OB-fold. We show that MRT-1 acts in the same telomere replication pathway as telomerase and the 9-1-1 DNA-damage response complex. MRT-1 is dispensable for DNA double-strand break repair, but functions with the 9-1-1 complex to promote DNA interstrand cross-link (ICL) repair. Our data reveal MRT-1 as a dual-domain protein required for telomerase function and ICL repair, which raises the possibility that telomeres and ICL lesions may share a common feature that plays a critical role in de novo telomere repeat addition.

Introduction

The ends of linear chromosomes, telomeres, pose two major challenges to the maintenance of chromosome integrity and overall genome stability: telomeres need to be adequately replicated, to compensate for the inability of canonical DNA polymerases to replicate the chromosome terminus, and they need to be protected from being mistakenly sensed and repaired as DNA double-strand breaks (DSBs).

Telomeric DNA is composed of simple, repetitive sequences. The 5′-to-3′ telomeric DNA strand is G-rich and terminates as a 3′ single-stranded overhang. Telomeric repeats are replenished by telomerase, a ribonucleoprotein composed of the telomerase reverse transcriptase (TERT) and an RNA component (TR) containing the telomere-repeat template (Greider and Blackburn, 1989; Collins, 2006). The catalytic subunit of telomerase, TERT, and its RNA component are sufficient to confer telomere-repeat addition in vitro. Additional proteins, many of which are essential, facilitate processing of TR and telomerase holoenzyme function (Collins, 2006; Fu and Collins, 2007; Venteicher et al, 2008). Mutations in the essential proteins dyskerin, NOP10, NHP2, as well as in TR and TERT, confer shortened telomeres and reduced in vitro telomerase activity in patients with heritable forms of Dyskeratosis Congenita and Pulmonary Fibrosis (reviewed by Vulliamy and Dokal, 2008). Thus, telomere maintenance defects can limit proliferation of cells in lymphatic or pulmonary systems in vivo, consistent with evidence that telomerase can limit the proliferative lifespan of human primary cells in vitro (Garcia et al, 2007).

Double mutants deficient for fission or budding yeast DNA-damage sensor proteins ATM and ATR display progressive telomere erosion, suggesting that DNA-damage signalling may be required for telomerase-mediated telomere maintenance in vivo (Naito et al, 1998; Ritchie et al, 1999; Nakamura et al, 2002). Conceptually related results were reported for Caenorhabditis elegans DNA-damage response mutants, where telomerase-mediated telomere replication was abolished in vivo by single mutations in subunits of the Rad9-Rad1-Hus1 (9-1-1) PCNA-like sliding clamp, mrt-2 (the worm rad1) or hus-1, or its large RFC clamp loader subunit, hpr-17 (Ahmed and Hodgkin, 2000; Hofmann et al, 2002; Boerckel et al, 2007). Further, subunits of the homologous mammalian 9-1-1 complex, as well as its RFC clamp loader, RAD17, were shown to physically interact with the telomerase holoenzyme, to bind to telomeric DNA in vivo, and to facilitate telomerase activity in vitro (Francia et al, 2006). However, knockdown of these mammalian DNA-damage response proteins is cell-lethal and results in dramatic, rapid effects on telomere length, precluding analysis of effects on telomerase-mediated telomere length maintenance in vivo (Francia et al, 2006). The former results suggest that DNA-damage response proteins may function at chromosome ends as a prerequisite for telomere repeat addition by telomerase. These proteins can respond to DSBs (d'Adda di Fagagna et al, 2004), suggesting that telomeres may be sensed as ‘aberrant' DSBs when they are replicated during S-phase and that the 9-1-1 complex in conjunction with its clamp loader may facilitate recruitment of telomerase to chromosome termini.

Here we identify C. elegans MRT-1 as a novel factor required for in vivo telomerase activity. MRT-1 encodes a dual-domain protein with an N-terminus homologous to the second OB DNA-binding fold found in POT1 (Protection Of Telomeres 1) proteins and a C-terminus bearing homology to the SNM1 nuclease family. SNM1 proteins function in DNA repair and checkpoint responses to interstrand cross-links (ICLs), stalled replication forks and DSBs, as well as in telomere protection (Henriques and Moustacchi, 1980; Dronkert et al, 2000; Demuth et al, 2004; Jeggo and Lobrich, 2005; Freibaum and Counter, 2006; Lenain et al, 2006; van Overbeek and de Lange, 2006; Bae et al, 2008; Hazrati et al, 2008; Hemphill et al, 2008). Although previously described C. elegans DNA-damage response mutants that are deficient for telomerase activity in vivo are hypersensitive to DSBs and ICLs, mrt-1 mutants are only deficient for ICL repair. Thus, MRT-1 defines a dual-domain ICL DNA-damage response protein that may process and interact with chromosome termini prior to telomerase-mediated telomere repeat addition.

Results

MRT-1 is required for telomerase activity in vivo

To identify non-essential mutations that compromise telomerase activity in vivo in C. elegans, genetic screens for ethylmethanesulphonate (EMS)-induced mortal germline (mrt) mutations that resulted in progressive telomere shortening and progressive sterility, accompanied by telomere–telomere fusions, were performed. Two alleles of mrt-1, e2661 and yp2, were identified in such screens (Y Liu and S Ahmed, unpublished data) (Ahmed and Hodgkin, 2000). mrt-1 mutants showed progressive reduction in progeny and eventual sterility comparable to mrt-2 mutants and mutants defective for the C. elegans catalytic subunit of telomerase, trt-1, (Table I) accompanied by progressive telomere shortening over successive generations (Figure 1A and Figure 5). Although telomere length does fluctuate as N2 wild-type strains are propagated for multiple generations (Ahmed et al, 2001), wild-type telomeres appeared as diffuse bands on Southern blots, whereas telomeres of all mrt-1 alleles appeared as discrete bands that shorten progressively (Figure 1A, right panel), as previously observed for C. elegans mutants that are deficient for telomerase mediated-telomere replication such as trt-1, mrt-2, hus-1 and hpr-17 (Ahmed and Hodgkin, 2000; Hofmann et al, 2002; Meier et al, 2006; Boerckel et al, 2007). In crosses between mrt-1(e2661) and wild-type, mrt-1^−/− F2 siblings displayed telomere erosion accompanied by progressive sterility, whereas wild-type F2 siblings displayed neither phenotype, but did possess discrete telomeric restriction fragments inherited from the mrt-1 background (Figure 1A, left panel, and data not shown). In addition, mrt-1 strains displayed late-onset chromosome fusions, as indicated by reduced numbers of metaphase-arrested meiotic chromosomes in late-generation mrt-1 mutants (Figure 1B and Supplementary Figure 1). The presence of end-to-end chromosome fusions was verified by isolation of X-autosome fusions from three independent mrt-1 strains (Figure 1C). Genetic mapping of the dominant-chromosome-loss phenotype of these independent fusions, which occurs when an X-autosome chromosome fusion is in trans to unfused chromosomes during meiosis, revealed tight genetic linkage of one end of the X chromosome with an end of an autosome in each case, confirming the formation of covalent end-to-end chromosome fusions (Figure 1C; see Supplementary data) (Ahmed and Hodgkin, 2000; Meier et al, 2006; Boerckel et al, 2007; Lowden et al, 2008). The former mrt-1 phenotypes resemble C. elegans strains that are deficient in telomerase activity in vivo (Ahmed and Hodgkin, 2000; Hofmann et al, 2002; Meier et al, 2006; Boerckel et al, 2007).

Table 1.

Progressive brood size reduction and loss of viability in late generation mrt-1, mrt-2 and trt-1 mutants

	Generation
	F2	F4	F6	F8	F10	F12	F14	F16	F18	F20	F22	F24	F26
Wild-type
1	W	W	W	W	W	W	W	W	W	W	W	W	W
2	W	W	W	W	W	W	W	W	W	W	W	W	W
3	W	W	W	W	W	W	W	W	W	W	W	W	W
4	W	W	W	W	W	W	W	W	W	W	W	W	W
mrt-1(e2661)
1	W	W	W	M	M	F	VF	S
2	W	W	W	W	W	M	M	M	F	S
3	W	W	W	W	W	M	W	M	F	VF	S
4	W	W	W	M	W	M	M	M	F	VF	S
mrt-1(yp2)
1	W	W	W	W	W	M	W	M	F	F	F	VF	S
2	W	W	W	W	W	W	W	W	M	M	M	F	VF
3	W	W	W	W	W	W	M	M	M	M	M	M	F
mrt-1(tm1354)
1	W	W	W	W	W	M	W	M	F	S
2	W	W	W	W	W	M	M	F	F	F	VF	S
mrt-2(e2663)
1	W	W	W	W	M	M	F	M	F	VF	VF	S
2	W	W	M	M	M	M	M	M	F	VF	S
3	W	W	M	W	M	M	F	F	VF	S
4	W	W	M	F	S
trt-1(ok410)
1	W	W	W	W	M	M	M	M	F	F	VF	S
2	W	W	W	W	W	W	M	F	VF	S
3	W	W	W	W	W	W	W	M	M	F	F	VF

W, wild-type, ∼250 progeny per animal; M, medium, ∼80 progeny per animal; F, few, ∼20 progeny per animal; VF, very few, ∼3--5 progeny per animal; S, sterile.

Mutants were backcrossed twice against wild-type to restore telomere length and two to four homozygous lines of the indicated genotype were followed by picking six L1s each line every two generations, as described previously (Ahmed and Hodgkin, 2000), for 26 generations. Brood size and sterility are indicated.

Figure 1.

Characterization and genetic mapping of mrt-1. (A) mrt-1 mutants display progressive telomere shortening. Southern blotting of genomic DNA with a telomere repeat-specific probe was performed as described previously (Ahmed and Hodgkin, 2000; Meier et al, 2006). F4 and F12 generations of three homozygous mrt-1(e2661) mutant and three homozygous wild-type siblings from a single outcross are shown in the left panel and three progressive generations of wild-type, mrt-1(yp2), two lines of mrt-1(e2661) and mrt-1(tm1354) each are shown in the right panel. Internal-repeat signals (Wicky et al, 1996) and telomere signals are indicated. (B) DAPI staining of late-generation wild-type or mrt-1 worms. Representative oocyte nuclei, indicated by dashed circles, are shown. (C) X-autosome fusions, eT3, eT7 and 9u, isolated from independent mrt-1(e2661) strains. Visible markers used for mapping are indicated. (D) Map position of mrt-1 as determined by three-factor crosses. The number of recombination events scored between mrt-1 and unc-29 is indicated in brackets. Cosmids covering the approximate genetic position of mrt-1 are shown. (E) mrt-1 gene structure and mutations. Point mutations are indicated in bold italics. Solid black boxes depict exons that are not translated due to the e2661 premature stop codon or are missing as a consequence of altered mrt-1 splicing in tm1354. In tm1354, the deletion within introns 3 and 5 leads to two alternatively spliced mRNAs indicated as (a) and (b) (see Supplementary Figure 2B), resulting in downstream exons to be out of frame (indicated in grey). (F) Western blot of protein extracts from wild-type and mrt-1 mutant strains. The arrow indicates MRT-1 protein, asterisks indicate nonspecific bands.

Two- and three-factor crosses were used to map mrt-1 to approximately +2.92 on Chromosome I (Figure 1D). Although trt-1 is located nearby at +3.08, mrt-1 mutations complemented trt-1(ok410) for progressive sterility when propagated as trans-heterozygotes, whereas failure to complement was observed between the mrt-1 mutations e2661 and yp2, indicating that these mutations correspond to a single gene that is distinct from trt-1 (Table II; and data not shown). Failure to complement trt-1(ok410) was previously reported for three independent alleles of trt-1 (e2727, yp1 and tm899), thereby clearly defining the C. elegans telomerase reverse transcriptase (Meier et al, 2006). BLAST searches of predicted proteins to the left of trt-1 revealed an open reading frame, F39H2.5, encoding a protein with an N-terminal domain homologous to the second OB-fold of POT1 telomere-binding proteins and a C-terminal domain containing the metallo-β-lactamase and β-CASP motifs characteristic of the SNM1 family of nucleic acid processing factors (Figures 1E and 2, see below). Isolation and sequencing of the mrt-1 cDNA confirmed the predicted 608-amino-acid POT1 OB-fold/SNM1 dual-domain protein (Figure 1E and Supplementary Figure 2). Sequencing of F39H2.5 from wild-type, mrt-1(e2661) and mrt-1(yp2) revealed independent C-to-T transition mutations in e2661 and yp2, predicted to create stop codon and missense mutations, respectively (Figure 1E). The mrt-1(yp2) missense mutation results in an H127Y amino-acid change, thus altering an amino acid whose charge is conserved in the OB2-fold of most POT1 proteins (Figure 2A).

Table 2.

mrt-1 complementation

Strain	Lines	Sterility
trt-1(ok410) unc-29/++	3	No*
trt-1(ok410) unc-29/trt-1(ok410) unc-29	6	Yes
mrt-1(e2661) dpy-5/++	4	No*
mrt-1(yp2) dpy-5/++	3	No*
++trt-1(ok410) unc-29/dpy-5 mrt-1(e2661)++	4	No*
++trt-1(ok410) unc-29/dpy-5 mrt-1(yp2)++	4	No*
+trt-1(ok410) unc-29/mrt-1(tm1354)++	3	No*
dpy-5 mrt-1(e2661)/+mrt-1(tm1354)	4	Yes
dpy-5 mrt-1(yp2)/+mrt-1(tm1354)	4	Yes

Trans-heterozygous analysis of mrt-1 alleles with trt-1. mrt-1 alleles were placed in trans to trt-1 or to a different mrt-1 allele (as shown) and progeny of several independent F1s were propagated as trans-heterozygotes (see Supplementary data) until sterility, while lines marked with an asterisk did not show any visible reduction in viability when propagated up to 15–20 generations.

Figure 2.

MRT-1 shares sequence homology with POT1 and SNM1 proteins. (A) Protein domain structure of hPOT1, MRT-1 and the three additional C. elegans OB-fold domain proteins with homology to POT1. The conserved histidine H127Y (asterisk) mutated in the MRT-1 protein of mrt-1(yp2) is indicated. The region of homology around H127Y of the three C. elegans (C.e.) OB2-fold proteins aligned with the respective POT1 domains of Homo sapiens (H.s.), Xenopus laevis (X.l) and Saccharomyces pombe (S.p.) is shown. No clear alignment could be obtained for Arabidopsis thaliana POT1 within this region. Sequence alignments were generated using Pole BioInformatique Lyonnais ClustalW (http://pbil.ibcp.fr/htm/index.php). Red shading reflects sequence identity, green strong and blue weak similarity. (B) Protein domain structure of C.e. MRT-1, S. cerevisiae (S.c.) Pso2p and human SNM1A, SNM1B/Apollo and SNM1C/Artemis. A multiple sequence alignment of the HxHxDH and β-CASP motif-4 nuclease domains is shown below. Amino acids depicted on top of the alignment indicate amino-acid changes introduced into MRT-1 to generate MRT-1(4mut) (see Figure 3 and Supplementary data).

Upon identification of the mrt-1 gene based on our forward genetic experiments, a deletion of this locus, tm1354, was kindly generated by Shohei Mitani. The tm1354 deletion eliminates several exons of the C-terminal SNM1 nuclease domain, including conserved amino-acid motifs that are relevant for ICL repair in yeast Pso2p (Niegemann and Brendel, 1994; Li and Moses, 2003 and Figure 1E). RT–PCR of mrt-1 cDNA from tm1354 animals revealed two mRNAs predicted to result in truncated, out-of-frame proteins (Figure 1E and Supplementary Figure 2B). The tm1354 deletion was isolated under conditions that generate many additional lesions in a strain's genome (Gengyo-Ando and Mitani, 2000). Therefore, two- and three-factor crosses were performed to show that a locus tightly linked to the tm1354 deletion conferred progressive telomere erosion phenotypes characteristic of C. elegans telomerase mutants (Figure 1A and B and Table I, and data not shown). Further, the tm1354 deletion failed to complement the mrt-1 alleles e2661 and yp2 for progressive sterility and late-onset end-to-end chromosome fusion phenotypes, but complemented trt-1(ok410) (Table II). Although a strain containing the tm1354 deletion was previously mentioned to display progressive telomere shortening (Raices et al, 2008), the identification of independent alleles of this locus, as well as the genetic mapping and complementation tests described here, indicate that the telomere shortening observed in the tm1354 strain is caused by a defect in the mrt-1/F39H2.5 gene.

A polyclonal antibody raised against full-length MRT-1 detected equivalent levels of full-length MRT-1 and MRT-1(H127Y) in wild-type and mrt-1(yp2) worm extracts, respectively (Figure 1F). Thus, the POT1-related OB2 domain of MRT-1 is required for telomerase activity in vivo. In contrast, no MRT-1 protein was detected in the e2661 nonsense mutation and tm1354 deletion extracts (Figure 1F and data not shown), indicating that these mutations are likely to be null alleles of mrt-1/F39H2.5, a non-essential gene required for de novo telomere repeat addition in C. elegans.

mrt-1 encodes a dual-domain protein

The N-terminal domain of C. elegans MRT-1 shares sequence homology with the second OB-fold of POT1 proteins (Figure 2A and Supplementary Figure 3). Single-stranded telomeric DNA-binding proteins commonly contain two adjacent N-terminal OB-folds, OB1 and OB2 (Horvath et al, 1998; Lei et al, 2003, 2004; Theobald and Wuttke, 2004). In addition to MRT-1, the C. elegans genome encodes two short proteins with homology to the second OB2-fold of POT1, F57C2.3 (CeOB1) and 3R5.1 (Figure 2A and Callebaut et al, 2002; Raices et al, 2008). The OB2-folds of MRT-1, F57C2.3 (CeOB1) and 3R5.1 are closely related and likely evolved from a single ancestral OB2-fold gene. A fourth C. elegans gene, B0280.10 (CeOB2), is homologous to the first OB-fold of POT1, OB1 (Figure 2A and Raices et al, 2008). The tandem OB-fold structure typical of POT1 proteins has been subjected to fission and duplication in C. elegans. Thus, we originally identified mrt-1/F39H2.5 based on the genetic map position of the mrt-1(e2661) telomerase-deficient mutant, and three additional C. elegans genes were identified based on their homology to POT1: B0280.10, F57C2.3 and 3R5.1. While this study was in progress, these genes were independently identified as POT1 homologues (Raices et al, 2008). We designate the C. elegans gene name for these genes as ‘pot', for ‘homologous to Protection of Telomeres 1(Pot1)', where pot-1 is B0280.10 (CeOB2) pot-2 is F57C2.3 (CeOB1) and pot-3 is 3R5.1 (Figure 2A) (Raices et al, 2008; Lowden et al, 2008). These genes display sequence similarity to, and evolved from, POT1, and their functions may reflect (1) one or more functions of ancestral POT1, including but not limited to ‘protection of telomeres', or (2) derived functions that may be unrelated to the ancestral protein. Since F39H2.5/mrt-1 contains homology to two conserved proteins, POT1 and SNM1, the gene name mrt-1 is used, based on the Mortal Germline phenotype of mrt-1 mutants (Ahmed and Hodgkin, 2000). MRT-1 is the only C. elegans POT1-like OB-fold protein that is required for telomerase activity in vivo (Figure 1), whereas the pot-1(CeOB2) and pot-2(CeOB1) genes may repress telomerase activity or recombination at telomeres (Raices et al, 2008; M Lowden and S Ahmed, unpublished data).

The C-terminus of MRT-1 corresponds to the sole C. elegans homologue of SNM1 proteins (Figure 1E and Figure 2B), which are members of the nucleolytic DNA- and RNA-processing β-CASP (metallo-β-lactamase-associated CPSF-Artemis-SNM1/PSO2) protein family (Aravind, 1999; Callebaut et al, 2002; Dominski, 2007). Saccharomyces cerevisiae Pso2p, and mammalian SNM1A and SNM1B/Apollo promote ICL repair (Henriques and Moustacchi, 1980; Demuth et al, 2004; Bae et al, 2008; Hazrati et al, 2008; Hemphill et al, 2008). In addition, SNM1B/Apollo and SNM1C/Artemis contribute to telomere end protection (Rooney et al, 2003; Freibaum and Counter, 2006; Lenain et al, 2006; van Overbeek and de Lange, 2006).

The OB-fold/SNM1 dual-domain structure of MRT-1 is observed for the closely related Caenorhabditis species remanei and briggsae, but was not predicted from genome sequences of the distantly related parasitic nematodes Brugia malayi and Trichinella spiralis (data not shown). Thus, fusion of POT1 OB2 and SNM1 domains to create the mrt-1 gene may have occurred within the Nematode phylum.

MRT-1 acts as a nuclease in vitro

To determine whether MRT-1 harbours nucleolytic activity as implied by its sequence homology to SNM1 proteins, we purified wild-type, the MRT-1(H127Y) OB-fold mutant, and two putative nuclease-dead mutant versions of MRT-1, MRT-1(D245A) and MRT-1(4mut), from Escherichia coli (Supplementary Figure 4A). The MRT-1(D245A) protein contains an aspartate to alanine mutation in the conserved HxHxDH signature motif, which comprises residues predicted to participate in zinc coordination (histidines) and hydrolysis (aspartate) at the active site (Figure 2B and Carfi et al, 1995). In budding yeast, this mutation diminishes the in vitro 5′-to-3′ nuclease activity of Pso2p and leads to an ICL-repair defect in vivo (Li et al, 2005). The corresponding mutation abolishes the in vitro 5′ exonuclease activity of mammalian SNM1A (Hejna et al, 2007) and reduces the endonucleolytic activity of SNM1C/Artemis that is observed in the presence of DNA-PK (Pannicke et al, 2004). However, as this single amino-acid change does not abolish SNM1 nuclease activity in all cases (Pannicke et al, 2004), we also disrupted the putative catalytic core of MRT-1 with four mutations (4mut): a HxHxDH-to-AxAxAH triple mutation and a D335A substitution in motif-4 (Figure 2B and Poinsignon et al, 2004). Wild-type and mutant versions of recombinant MBP-6 × His–TEV–MRT-1 were purified over a TALON column resulting in a ∼120-kDa MBP-6 × His–TEV–MRT-1 band (Supplementary Figure 4A). Cleavage with TEV protease followed by a second purification step yielded untagged full-length MRT-1 (Supplementary Figure 4A and Supplementary data). Cleaved and purified MRT-1 protein exhibited 3′ nuclease activity as revealed by complete removal of one or more 3′ nucleotides from a 5′-end-labelled substrate (Figure 3A, arrow, and Figure 3B, arrow), accompanied by a smear of additional degradation products, which includes minor stalling points and release of the terminal 5′ nucleotide. This activity was not observed for TEV protease alone (Figure 3A, lane 11). Importantly, no nuclease activity was observed with MRT-1(D245A) or MRT-1(4mut) mutants, indicating that this activity requires conserved residues in the MRT-1 nuclease domain (Figure 3A, lanes 5–10; Figure 3B). A time-course experiment using MRT-1 and MRT-1 (H127Y), which contains the POT1 OB-fold substitution that abolishes telomerase activity in vivo, indicated that the nuclease activity of MRT-1, apparent by the complete removal of one or more 3′ nucleotides (arrow), as well as a smear of degradation products, is not affected by the OB-fold mutation (Figure 3B). Removal of the N-terminal epitope tags from MRT-1 promoted degradation of 5′-end-labelled substrates, accompanied by release of a small amount of 5′ mononucleotide, which might correspond to either a weak 5′ nuclease activity or a processive 3′ nuclease activity that completely degrades the oligonucleotide substrate (Figure 3A and B). To distinguish between these possibilities, MRT-1 was incubated with a 3′-end-labelled substrate. A single 3′ mononucleotide was released from 3′-end-labelled substrate in Figure 3C, in a nuclease domain-dependent manner. Thus, MRT-1 functions as a 3′–5′ nuclease, similar to the 3′–5′ exonuclease ExoI control (Figure 3C). If MRT-1 were a processive 5′-to-3′ nuclease, then a ladder of products would have been observed for the 3′-end-labelled substrate in Figure 3C. Consistently, incubation of various dilutions of MRT-1 protein with 3′-end-labelled G-strand substrate failed to reveal any cleavage intermediates expected for 5′-to-3′ exonuclease activity (Supplementary Figure 4E). We considered the possibility that the 5′ phosphate of 5′-end-labelled substrates might elicit 5′-to-3′ nuclease activity by MRT-1, as has been observed for SNM1 (Hejna et al, 2007), but addition of a cold 5′ phosphate to 3′-end-labelled ssDNA substrate did not promote the formation of an n-1 product by MRT-1, and resulted exclusively in release of the terminal 3′ nucleotide (data not shown). While MRT-1 and MRT-1(H127Y) showed comparable efficiency in cleaving the terminal 3′-labelled substrate, MRT-1(D245A) and MRT-1(4mut) exhibited strongly reduced kinetics and absence of activity, respectively (Figure 3C). Thus, our results indicate that MRT-1 functions as a 3′-to-5′ nuclease in vitro.

Figure 3.

MRT-1 acts as a 3′ nuclease in vitro. (A) 70 nM of MRT-1, MRT-1(D245A) and MRT-1(4mut) from different purification steps were incubated with 5 nM of a C. elegans telomeric G-strand oligonucleotide labelled at the 5′ end (asterisk). The line before the oligonucleotide sequence shown depicts an invariant linker sequence (see Supplementary data). 6 × His–TEV1 extracts were tested for contaminating nuclease activity (lane 11). The 5′-to-3′ nuclease RecJf and the 3′-to-5′ nuclease ExoI were used as controls (lanes 12 and 13, respectively). 1-nt products generated by RecJf and 2- to 4-nt products generated by ExoI are indicated. The arrow indicates the reduced size oligonucleotide band due to nuclease activity. For panel A, lane 2, it is difficult to assess the degree of nucleolytic activity for uncleaved MRT-1 due to a gel-running artefact. (B) Time-course experiment of MRT-1, MRT-1(H127Y), MRT-1(D245A) and MRT-1(4mut) nuclease activity on a 5′ labelled (asterisk) C. elegans telomeric G-strand oligonucleotide. 70 nM of protein were incubated with 5 nM of 5′ labelled oligonucleotide and aliquots were taken at the time points indicated. 2- to 4-nt products generated by ExoI and 1-nt products are indicated. The line drawn before the oligonucleotide sequence depicts an invariant linker sequence (see Supplementary data). The arrow indicates the reduced size band due to nuclease activity. (C) Time-course of MRT-1, MRT-1(H127Y), MRT-1(D245A) and MRT-1(4mut) nuclease activity on a 3′ labelled (asterisk) C. elegans telomeric G-strand oligonucleotide performed as described in panel B. (D) 70 nM MRT-1 and MRT-1(4mut) were incubated with 5 nM of various 5′ labelled single-stranded oligonucleotides for 1 h. (E) 70 nM MRT-1 and MRT-1(4mut) proteins were incubated with 1 nM of single- or double-stranded 5′ labelled DNA substrates as indicated (see Supplementary data). 1-nt products as generated by RecJf are indicated. The arrow indicates the reduced size band due to nuclease activity. (F) MRT-1 and MRT-1(4mut) proteins were incubated with single- or double-stranded 3′ labelled (asterisk) DNA substrates as described in panel E. 1-nt products as generated by ExoI are indicated.

Use of n-1, n-2 and n-3 C. elegans G-strand oligonucleotide size markers revealed that an n-3 molecule is the major product of the MRT-1 nuclease for the 5′-end-labelled C. elegans G-strand substrate (Supplementary Figure 4C). Dilutions of MRT-1 protein revealed a smear of n-1, n-2 and n-3 degradation products, which occur during the rapid generation of the n-3 product (Supplementary Figure 4D). Thus, MRT-1 does not act as an endonuclease that chops off three 3′ nucleotides at the 3′ end of the C. elegans G-strand substrate, but rather as a processive 3′-to-5′ exonuclease that stalls at the n-3 position. Various 5′-end-labelled substrates, including telomeric G-strand and C-strand oligonucleotides and (dT)₃₅, were efficiently processed by MRT-1, although different patterns of nucleolytic activity were observed (Figure 3D). The oligo(dT) substrate was more severely degraded, suggesting that the activity of MRT-1 may depend on the sequence or structure of its substrates. In an attempt to address the various nucleolytic patterns of MRT-1 on different oligonucleotides (Figure 3D), we examined the effect of MRT-1 on n-1, n-2 and n-3 C. elegans G-strand substrate molecules, where the major product of the full-length G-strand substrate is an n-3 molecule. The n-1 or n-2 substrates were rapidly cleaved to yield an n-3 product, whereas the n-3 substrate was resistant to the strong 3′-to-5′ exonuclease activity of MRT-1, which rapidly removes 3′ nucleotides from full-length, n-1 and n-2 G-strand substrate molecules (Supplementary Figure 4F). Note that MRT-1 created a weak smear of degradation products for all single-stranded G-strand oligonucleotides, including the n-3 substrate. A major n-3-mer product was also generated when MRT-1 was incubated with double-stranded C. elegans G-strand substrate, where the G-strand was 5′-end-labelled (Figure 3E and data not shown). Thus, when three nucleotides are removed from the C. elegans G-strand substrate terminating in GGCTTA to yield an n-3 molecule terminating in TTAGGC, the processive 3′-to-5′ nuclease activity of MRT-1 is inhibited. Raising the incubation temperature of the MRT-1 cleavage reaction from 20 to 37°C resulted in uniform laddering of a C. elegans telomeric G-strand substrate (Supplementary Figure 4B), analogous to the oligo(dT) substrate at 20°C. This temperature-sensitive effect may reflect altered structure of the 3′ end of the n-3-mer, or perhaps increased processivity of MRT-1 protein, at 37°C. Formal proof that MRT-1 harbours a 3′-to-5′ exonuclease activity that may possess a degree of structure or sequence specificity awaits detailed characterization of various substrate molecules and how they are processed by MRT-1.

Analysis of the activity of MRT-1 on 5′-end-labelled, blunt, double-stranded DNA substrates or on a substrate with a 3′ telomeric overhang resulted in release of both 3′ and 5′ nucleotides (Figure 3E). To test whether release of the 5′ nucleotide from either substrate was due to 5′–3′ nuclease activity, analogous 3′-end-labelled substrates were examined, but only the 3′ nucleotide was released, and no n-1 band or additional banding pattern was observed (Figure 3F). n-1, n-2 and n-3 size markers indicated that an n-1 product could have been detected for 3′-end-labelled substrates if it occurred as a consequence of a non-processive 5′ nuclease activity (Supplementary Figure 4C, D and F). Thus, the relatively strong release of the 5′ terminal nucleotide from 5′-end-labelled dsDNA substrates occurs as a consequence of the 3′-to-5′ polarity of the MRT-1 nuclease.

We conclude that MRT-1 is a processive 3′-to-5′ exonuclease, which can degrade single-stranded substrates, which can be repressed by specific substrate configurations, and which can act to release the terminal 5′ nucleotides of substrates that are degraded with 3′-to-5′ polarity. The nuclease activity of MRT-1 is dependent on its SNM1 nuclease domain, but this activity is not affected by the H127Y POT1 OB-fold mutation. Note that the nuclease activities displayed by MRT-1 in vitro may be affected by protein modification or protein–protein interactions, and could be substrate dependent. The strong activity of MRT-1 on non-telomeric substrates is significant and not unexpected, as MRT-1 is not telomere-specific but also functions a general ICL-repair protein (see Figure 6, below).

MRT-1 binds single-stranded DNA in vitro

Although two adjacent N-terminal OB-folds in human POT1 are required for POT1 to bind single-stranded telomeric DNA in vitro (Lei et al, 2004), the sequence homology of MRT-1 is restricted to the second OB-fold of POT1 (Figure 2A). Electrophoretic mobility-shift assays revealed that 6 × His–MRT-1 bound to a radiolabelled single-stranded (GGCTTA)₄ C. elegans telomeric G-strand oligonucleotide (Figure 4A, lane 9). Furthermore, MRT-1 bound to various single-stranded oligonucleotides, including a C. elegans C-strand, an inverse C-strand and a human G-strand, with comparable affinity (Figure 4A, lanes 3, 6 and 12). DNA competition experiments revealed that an unlabelled C. elegans telomeric G-strand oligonucleotide was consistently able to compete with an MRT-1 complex bound to a radiolabelled G-strand oligonucleotide with at least 10-fold greater affinity than an unlabelled C. elegans telomeric C-strand oligonucleotide (Figure 4B and Supplementary Figure 5A). For reasons that are unclear, epitope-tagged MRT-1 band shifts were resolved as either single or double bands in different experiments, even when MRT-1 protein from the same protein purification was used (Figure 4B and Supplementary Figure 5A).

Figure 4.

MRT-1 binds single-stranded DNA in vitro. (A) Gel-shift analysis of 200 nM purified 6 × His–MRT-1 or 6 × His–MRT-1(H127Y) with 5 nM of the indicated oligonucleotide substrates (top; also listed in Supplementary data) in the presence of poly(dI–dC). The position of radioactive labels is indicated by asterisks and the invariant linker sequence of each oligonucleotide is indicated as described for Figure 3. DNA-binding activities of a bacterial in vitro translation extract were used as a DNA-binding control (see lane 2, ‘positive control'). DNA–protein complexes and free oligonucleotides are indicated. Percentages shown on the bottom of the gel indicate relative intensities of the shifted DNA–protein complexes (the shift of lane 6 being set to 100%). (B) Competition experiments of 6 × His–MRT-1 bound to a 5′ labelled (asterisk) single-stranded G-strand oligonucleotide followed by addition of increasing amounts of the indicated unlabelled oligonucleotides. Note that an inverse C-strand oligonucleotide is used for competition with the labelled G-strand oligonucleotide to avoid formation of double-stranded DNA. In this experiment, MRT-1 protein shifts as two different complexes with the radiolabelled oligonucleotide, which are equally competed by addition of unlabelled competitor oligonucleotides. Percentages shown on the bottom indicate relative amounts of free radiolabelled oligonucleotide with the unbound oligonucleotide (lane 1) set to 100%. (C) Gel-shift analysis of TEV-cleaved and dialysed MRT-1 wild-type and mutant proteins. Wild-type and mutant proteins were incubated with 5 nM 5′ labelled single-stranded G-strand oligonucleotide using increasing protein concentrations of 17 nM, 50 nM, 150 nM, 450 nM in the presence of poly(dI–dC) as described in panel A. (D) Gel-shift analysis of TEV-cleaved and dialysed MRT-1 wild-type with single-stranded and double-stranded substrates. TEV-cleaved and dialysed wild-type MRT-1 protein (50, 150 and 450 nM) was incubated with 5 nM 5′ labelled single-stranded or double-stranded DNA substrates (see Supplementary data) in the presence of poly(dI–dC). Single-stranded and double-stranded DNA and protein–DNA complexes are indicated on the right. Note that single-stranded and double-stranded DNA run differently in native gels. Relative intensities of DNA–protein complexes were measured between lanes for single-stranded and double-stranded DNA containing identical protein concentrations after subtraction of the background signal. The intensities of the DNA–protein complexes formed with double-stranded DNA were 9 and 7% of that formed with single-stranded DNA for 150 and 450 nM MRT-1 (lane 7 compared to lane 3 and lane 8 compared to 4), respectively.

The mrt-1(yp2) allele deficient for telomerase activity in vivo encodes an H127Y protein mutation that affects a conserved charged residue in its OB-fold (Figures 1E and 2A) and thus may affect DNA binding. In comparison to 6 × His–MRT-1, an equivalent concentration of 6 × His–MRT-1(H127Y) had strongly diminished affinity for all single-stranded oligonucleotides tested, including C. elegans G-strand telomere repeats (Figure 4A, lanes 4, 7, 10 and 13). To assess the DNA-binding activity of untagged MRT-1, MBP-6 × His–TEV–MRT-1 fusion proteins were purified and cleaved with TEV protease. Similar to 6 × His-tagged MRT-1, binding of TEV-cleaved MRT-1(H127Y) to a G-strand oligonucleotide was reduced five- to eight-fold in comparison with wild-type MRT-1 (Figure 4C). Untagged MRT-1 protein does not show the crisp band shifts observed for epitope-tagged MRT-1 (Figure 4 and Supplementary Figure 5). We hypothesized that the smeary band shifts might occur as a consequence of the nuclease activity of MRT-1, but they remained smeary for nuclease-dead versions of MRT-1, which bound C. elegans G-strand substrate with an affinity similar as MRT-1 (Figure 4C). Smeary band shifts were consistently observed using several different protein purification protocols to purify MRT-1, and for many electrophoresis conditions, suggesting that non-static DNA–protein interactions may be an inherent biophysical property of MRT-1. The positively charged 6 × His-tag may non-specifically stabilize the interaction of MRT-1 with DNA (Figure 4A and B). Using concentrated untagged MRT-1 we found that 1.8 μM MRT-1 was required to fully shift the oligonucleotide substrate, whereas 600 nM MRT-1 shifted approximately 50% of the substrate (Supplementary Figure 5B). MRT-1 binding to double-stranded DNA substrates generated by annealing telomeric G-strand and C-strand oligonucleotides was reduced at least 10-fold in comparison with binding to single-stranded DNA (Figure 4D, and data not shown). Thus, MRT-1 can interact with various DNA substrates in vitro, and this biochemical property may be biologically meaningful as it is disrupted by an OB-fold mutation that abolishes telomerase activity in vivo.

Efforts to determine MRT-1 localization in vivo by indirect immunofluorescence using an anti-MRT-1 antibody (Figure 1F) were unsuccessful (B Meier, T Lee, S Ahmed, A Gartner, unpublished data), consistent with other reports that have failed to detect endogenous mammalian SNM1 proteins by immunofluorescence (Dronkert et al, 2000; Lenain et al, 2006; van Overbeek and de Lange, 2006). Further, the MRT-1 antibody was not effective for chromatin immunoprecipitation of telomeric DNA (J Hall and S Ahmed, unpublished data). Introduction of the wild-type mrt-1(+) genomic DNA locus or MRT-1::GFP constructs in the mrt-1(tm1354) mutant backgrounds using either complex extrachromosomal arrays or microparticle bombardment resulted in undetectable or low expression of MRT-1 protein as assessed by western blotting and failed to rescue the mrt-1 mutant ICL hypersensitivity phenotype (B Meier, L Barber, S Boulton, S Ahmed, A Gartner, unpublished data).

MRT-1 acts in the same pathway as MRT-2 and TRT-1 for telomere replication

To determine whether mrt-1, telomerase and mrt-2 act in the same genetic pathway of telomere elongation, mrt-1 trt-1 and mrt-1; mrt-2 double mutants were generated for comparison with the respective single mutants. Mutants were grown to sterility, which occurred at approximately the same generation (data not shown), and the rate of telomere shortening was analysed by Southern blotting (Figure 5A and B). The rates of telomere shortening in mrt-1(yp2) trt-1 or mrt-1(e2661); mrt-2 double mutants, 106±42 and 128±30 bp per generation, respectively, were not significantly different from those of the single mutants, mrt-1(yp2) (106±35 bp); mrt-1(e2661) (114±26 bp); trt-1 (125±22 bp) and mrt-2 (129±30 bp) (Figure 5C). Therefore, these three genes appear to act in a single pathway to facilitate telomere repeat addition by telomerase in vivo.

Figure 5.

mrt-1, mrt-2 and trt-1 function in the same genetic pathway. (A, B) Southern blots of various mrt-1 single and double mutant combinations performed as described for Figure 1A. Internal-repeat signals (Wicky et al, 1996) and telomere signals are indicated on the right. (C) Statistical analysis of the rate of telomere shortening in single and double mutants. The rate of telomere shortening indicates the mean of individual telomere shortening rate measurements over at least three analysed generations (see Supplementary data) and error bars show standard deviations.

MRT-1 defines an ICL-damage response pathway

Subunits of the 9-1-1 DNA-damage response complex and its RFC clamp loader hpr-17 orchestrate DNA-damage signalling and repair of ionizing radiation (IR)-induced DSBs. Additionally, they are required for telomerase-mediated telomere length maintenance (Ahmed and Hodgkin, 2000; Hofmann et al, 2002; Boerckel et al, 2007). This dual function, coupled with observations that many additional proteins that interact with DSBs also function at normal telomeres (d'Adda di Fagagna et al, 2004), suggested that the DSB-repair function of the 9-1-1 complex may be relevant to telomerase-dependent telomere replication. Given that mrt-1 and mrt-2 act in a single telomere replication pathway (Figure 5A and C), we asked whether mrt-1 shares the DSB-repair functions of the 9-1-1 complex by testing their sensitivity to ionizing radiation. DNA DSBs in germ cells are predominately repaired by homologous recombination in a manner that is dependent on the 9-1-1 DNA-damage response complex (Clejan et al, 2006). Quantification of the survival rates of progeny derived from γ-irradiated L4 larvae revealed that mrt-2 was hypersensitive to IR-induced DSBs (Ahmed and Hodgkin, 2000), whereas mrt-1 displayed a dose–response survival curve that was not significantly different from that of the wild-type (Figure 6A).

Figure 6.

mrt-1 mutants are hypersensitive to UV/TMP but not to ionizing radiation. (A) Sensitivity of wild-type, mrt-1(e2661) and mrt-2(e2663) strains to increasing doses of ionizing radiation was assayed as described (Ahmed and Hodgkin, 2000). (B) Sensitivity of wild-type and mrt-1(yp2), mrt-1(e2661), mrt-2(e2663) and hus-1(op244) mutants to UV/TMP with 10 μg/ml TMP and the indicated doses of UVA light (bottom) in Joules/cm² was assayed as described (Collis et al, 2006). (C) Sensitivity of wild-type, mrt-1(yp2) and mrt-2(e2663) in comparison with mrt-1(yp2); mrt-2(e2663) double mutants to UV/TMP as described in panel B. (D) Sensitivity of wild-type, mrt-1(e2661) and mrt-2(e2663) in comparison with mrt-1(e2661); mrt-2(e2663) double mutants to UV/TMP as described in panel B. (A–D) Results show averages from three independent experiments and error bars represent standard error of the mean for all graphs depicted.

The mammalian SNM1C/Artemis protein is known to facilitate telomere capping and plays a role in non-homologous end-joining (NHEJ)-mediated DSB repair (Richie et al, 2002; Rooney et al, 2003). Since MRT-1 is the only C. elegans protein bearing sequence homology to the SNM1 family, we examined whether it too plays a role in NHEJ. In contrast to strains deficient for any of the core NHEJ subunits lig-4, cku-70 or cku-80 (Clejan et al, 2006), mrt-1 mutant strains did not display defects in an assay for NHEJ-mediated DSB repair (Supplementary Figure 6). Thus, mrt-1 is not required for NHEJ- or homologous recombination-mediated DSB repair.

As we did not observe a role for MRT-1 in either DSB repair or NHEJ, we considered whether it may instead be required for ICL repair, analogous to yeast Pso2p and human SNM1A (Hazrati et al, 2008). We tested mrt-1(yp2) and mrt-1(e2661) for their sensitivity to DNA ICLs and found that both mutant alleles displayed hypersensitivity to trimethylpsoralen photoactivated by UVA radiation (UV-TMP) compared with wild-type (Figure 6B). Given the specific defect of MRT-1 in DNA cross-link repair, we also tested the previously identified telomerase-deficient 9-1-1 DNA-damage response complex mutants mrt-2 and hus-1, both of which were even more sensitive to UV-TMP than mrt-1 (Figure 6B). Interestingly, mrt-1(e2661); mrt-2(e2663) and mrt-1(yp2); mrt-2(e2663) double mutants were not more sensitive than mrt-2 single mutants (Figure 6C and D), indicating that mrt-1 and the 9-1-1 complex may function in a common pathway to promote ICL repair.

Discussion

We have identified MRT-1 as a new factor required for telomerase-mediated telomere repeat addition in vivo using unbiased genetic screening and positional cloning. MRT-1 is a dual-domain protein comprised of an N-terminal region with a POT1-like OB2-fold domain and a C-terminus homologous to the Pso2/SNM1 family of nucleases.

Four POT1-related OB-fold proteins exist in C. elegans, yet MRT-1 is the only POT-1 OB-fold protein that is essential for telomere-repeat addition by telomerase. POT1 duplications have occurred several times during evolution and, once duplicated, POT1-like proteins tend to evolve rapidly and adopt distinct telomere-related functions. Mutation of one of three Arabidopsis POT1 genes abrogates telomerase activity in vivo and in vitro (Shakirov et al, 2005; Surovtseva et al, 2007). The former results mirror evidence from organisms that contain a single essential POT1-related gene (Baumann and Cech, 2001; Veldman et al, 2004; Hockemeyer et al, 2005). For example, human POT1 can facilitate telomerase activity in vitro when in a complex with its interacting protein TPP1 (Lei et al, 2005; Wang et al, 2007; Xin et al, 2007). Thus, POT1 OB-fold proteins that interact with single-stranded telomeric DNA may perform a conserved function that enables telomerase to act at chromosome termini in vivo. This function is difficult to address in vivo in organisms where deficiency for POT1 OB-fold proteins results in acute telomere uncapping and/or end-to-end chromosome fusions (Baumann and Cech, 2001; Veldman et al, 2004; Hockemeyer et al, 2005). In contrast, viable null alleles of mrt-1 display telomere-erosion phenotypes that are indistinguishable from those of telomerase reverse transcriptase mutants (Figures 1 and 5 and Supplementary Figure 1).

Insight into the telomerase-promoting function of POT1-like proteins is provided by the H127Y mutation in the MRT-1 OB2-fold that abolishes telomerase-mediated telomere replication and affects a conserved histidine (Figure 2A). This amino acid corresponds to His245 of human POT1, which forms hydrogen bonds with the second and fourth nucleotides from the 3′OH terminus of a telomeric oligonucleotide in a crystal structure, suggesting that it may mediate interaction with the chromosome terminus in vivo (Lei et al, 2004). Our results demonstrate that this amino acid is required for interaction of MRT-1 with single-stranded DNA in vitro and that this residue is essential for in vivo telomerase activity. Thus, the DNA-binding activity of POT1-related OB-fold proteins is likely to be critical for telomerase activity in vivo.

Our genetic analysis of mrt-1; mrt-2 double mutants indicates that MRT-1 and the 9-1-1 DNA-damage response complex act together in a single pathway to facilitate telomerase-dependent telomere replication. While C. elegans 9-1-1 complex mutants display defects in responding to DSB and ICL lesions, the mrt-1 alleles (including two probable null alleles) are only deficient for ICL repair. It is intriguing that MRT-1 and MRT-2 function in both telomerase-dependent telomere replication and ICL repair. Although this may be coincidental, we favour the possibility that telomerase-dependent telomere replication and some ICL lesions may share common features. Human SNM1B/Apollo interacts with telomere-binding proteins and is required for telomere function in vivo (Freibaum and Counter, 2006, 2008; Lenain et al, 2006; van Overbeek and de Lange, 2006; Demuth et al, 2008), whereas human SNM1C/Artemis facilitates telomere capping (Rooney et al, 2003). This telomere association may be specific to metazoans, as independent large-scale genetic screens have failed to reveal a role for S. cerevisiae PSO2/SNM1 in telomere homeostasis (Askree et al, 2004; Downey et al, 2006). Our results indicate that MRT-1 possesses dual, separable biochemical activities: binding to single-stranded DNA (OB-fold) and nucleolytic processing (SNM1 nuclease domain). These activities are likely to act sequentially in vivo and may facilitate both ICL repair and telomere-repeat addition by telomerase, although the role of the nuclease domain of MRT-1 in telomere replication is presently unclear. Given that the mammalian 9-1-1 complex physically interacts with telomerase (Francia et al, 2006), and that subunits of this complex, as well as MRT-1, are required for de novo telomere repeat addition at C. elegans chromosome ends, we favour the possibility that an ancient metazoan DNA-damage response pathway may be triggered at chromosome termini prior to telomere-repeat addition by telomerase.

Secondary DNA structures arising at telomeres, some of which could resemble intermediates formed at ICL lesions, might serve as substrates for MRT-1. At present we can only speculate on the structure of the in vivo substrates of MRT-1. Despite evidence that Pso2p/ SNM1 nuclease activity is needed for ICL repair and that Pso2p/ SNM1 acts downstream of the initial cleavage event at a cross-link lesion, the relevant in vivo substrates processed by Pso2p are still unknown (Magana-Schwencke et al, 1982; Li et al, 2005). Of the mammalian homologues, DNA substrates have only been defined for SNM1C/Artemis, which targets hairpin intermediates generated during V(D)J recombination (Rooney et al, 2003). It is of interest that the ciliate TEBP β-protein has been implicated in resolution of the telomeric DNA G-quadruplex secondary structure during DNA replication, perhaps coincident with telomerase activity (Paeschke et al, 2008). Additional unusual DNA structures at telomeres include (1) the T-loop, where the terminal 3′ overhang folds back to form a structure that resembles a recombination intermediate, (2) stalled replication forks that commonly occur at repetitive DNA sequences (Griffith et al, 1999; Fouche et al, 2006) and (3) simple 3′ overhangs that are ubiquitous telomeric structures and would be excellent substrates for the 3′-to-5′ nuclease activity of MRT-1 (Figure 3 and Supplementary Figure 4). Finally, leading-strand DNA synthesis at chromosome termini has been suggested to yield the preferred substrate of telomerase (Chai et al, 2006), and blunt chromosome termini could be a substrate for the nuclease activity of MRT-1. In this case, 5′-to-3′ nuclease activity at blunt chromosome termini would be required to generate a 3′ overhang for telomerase, and MRT-1 did not possess such an activity under our experimental conditions (Figure 3E and F). However, studies of yeast and vertebrate SNM1 homologues of MRT-1 indicate that SNM1 nucleases often display 5′-to-3′ nuclease activity in vitro (Li et al, 2005; Hejna et al, 2007), which raises the possibility that MRT-1 may function analogously to create 3′ overhangs at blunt chromosome termini in vivo. Note that telomerase enzymes from various species are associated with a nuclease activity that can remove non-telomeric nucleotides from 3′ ends of primers in in vitro telomerase assays (Collins and Greider, 1993; Autexier and Greider, 1994; Cohn and Blackburn, 1995; Melek et al, 1996; Bhattacharyya and Blackburn, 1997; Prescott and Blackburn, 1997; Bednenko et al, 1997; Greene et al, 1998; Lue and Peng, 1998; Niu et al, 2000; Huard and Autexier, 2003, 2004; Oulton and Harrington, 2004), although studies with recombinant telomerase suggest that this nuclease activity may be a property of telomerase itself (Collins and Gandhi, 1998; Huard and Autexier, 2004; Oulton and Harrington, 2004). We conclude that the C. elegans ICL repair protein MRT-1 may interact with single-stranded telomeric DNA to promote telomere-repeat addition by telomerase and may function in a nucleolytic processing event at chromosome ends.

Materials and methods

Protein expression and purification

Proteins encoded by pAG472 (6 × His–MBP–TEV–MRT-1), pAG474 (6 × His–MBP–TEV–MRT-1(H127Y), pAG538 (6 × His–MBP–TEV–MRT-1(D245A) and pAG473 (6 × His–MBP–TEV–MRT-1(4mut)) (see Supplementary data) were expressed in E. coli BL21(DE3) pRil (Invitrogen). Cultures (500 ml) were induced at OD₆₀₀ ∼ 1.0 with 5 μM IPTG and grown overnight at 12°C. Bacteria were lysed in one-pellet volume of lysis buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl, 10 mM imidazole, 0.1% Tween, 10% glycerol, protease inhibitor cocktail complete (Roche) and 1 mg/ml lysozyme) by sonication, centrifuged and purified over a TALON metal affinity resin (Clontech) (see Supplementary data). MBP-6 × His–TEV–MRT-1 containing fractions eluted with 120 mM imidazole were cleaved with 6 × His–TEV1 protease at 4°C overnight. Cleavage efficiency was monitored by SDS–PAGE and Coomassie staining. Extracts were adjusted to a final concentration of 40 mM imidazole and incubated with NiNTA agarose (Qiagen) (see Supplementary data). The flow through was collected and assayed for MRT-1 by SDS–PAGE, Coomassie staining and western blotting, dialysed against 50 mM KCl, 20 mM potassium phosphate, pH 8.0, 20% glycerol, 0.2% Tween 20, 5 mM β-mercaptoethanol and concentrated up to 10-fold in an Amicon Ultra Centricon 30K MWCO (Millipore).

Proteins encoded by pAG368 (6 × His–MRT-1) and pAG369 (6 × His–MRT-1(H127Y)) were expressed as described above, but with IPTG induction for 2 h at 25°C. Bacteria were lysed by sonication in one-pellet volume of lysis buffer (20 mM sodium phosphate, pH 8.0, 500 mM NaCl, 20 mM imidazole, 0.2% Tween, 20% glycerol, 5 mM β-mercaptoethanol, protease inhibitor cocktail complete (Roche)) containing 1 mg/ml lysozyme, centrifuged and purified over a NiNTA agarose column (Qiagen) according to manufacturer's instructions. Protein fractions were pooled and then dialysed and concentrated as described above.

Nuclease assay

Radiolabelled oligonucleotide (5 nM) was incubated with 70 nM of cleaved and purified MRT-1 wild-type or mutant protein in a final volume of 10 μl reaction buffer (50 mM Tris–HCl, pH 8.0, 50 mM NaCl, 10 mM MgCl₂, 0.5 mM EDTA, 10% glycerol) unless indicated otherwise. Thirty units of RecJf (New England Biolabs) and 5–10 U ExoI (New England Biolabs) were used as controls. Reactions were incubated at RT for 1 h unless stated otherwise. An equal volume of formamide buffer (95% formamide, 18 mM EDTA, 0.025% SDS, 0.025% bromphenol blue, 0.025% xylene xyanol) was then added to each sample. Samples were heated at 80°C for 3–5 min and loaded onto a denaturing 12% polyacrylamide gel (40% acrylamide 29:1).

Electrophoretic mobility-shift assay

Radiolabelled oligonucleotide (5 nM) was incubated with 6 × His–MRT1 alleles or TEV-cleaved and dialysed MRT-1 protein versions at the indicated concentrations for 20 min at RT in 25 mM Hepes pH 7.5, 50 mM NaCl, 1 mM EDTA, 5% glycerol, 1 mM DTT in the presence of 100 ng of poly(dI–dC) as nonspecific competitor. For DNA competition experiments, proteins were incubated with a radiolabelled oligonucleotide in the absence of poly(dI–dC) for 20 min at RT. Unlabelled oligonucleotide was then added to the samples followed by incubation for 20 min. Samples were loaded onto a pre-run 6% polyacrylamide (29:1) gel and a marker dye was loaded in parallel into one well. The gel was then run for 3–5 h with 7 V/cm in 0.5 × TBE at 4°C, dried and exposed to film. DNA binding was quantified using Fujifilm FLA-5100 and LAS-4000 scanners and AIDA Image Analyzer v3.27 software (Raytest).

Supplementary Material

Supplementary Figure 1

Supplementary Figure 2

Supplementary Figure 3

Supplementary Figure 4

Supplementary Figure 5

Supplementary Figure 6

Supplementary Information