Functional Diversification of Replication Protein A Paralogs and Telomere Length Maintenance in Arabidopsis

Abstract

Replication protein A (RPA) is essential for many facets of DNA metabolism. The RPA gene family expanded in Arabidopsis thaliana with five phylogenetically distinct RPA1 subunits (RPA1A-E), two RPA2 (RPA2A and B), and two RPA3 (RPA3A and B). RPA1 paralogs exhibit partial redundancy and functional specialization in DNA replication (RPA1B and RPA1D), repair (RPA1C and RPA1E), and meiotic recombination (RPA1A and RPA1C). Here, we show that RPA subunits also differentially impact telomere length set point. Loss of RPA1 resets bulk telomeres at a shorter length, with a functional hierarchy for replication group over repair and meiosis group RPA1 subunits. Plants lacking RPA2A, but not RPA2B, harbor short telomeres similar to the replication group. Telomere shortening does not correlate with decreased telomerase activity or deprotection of chromosome ends in rpa mutants. However, in vitro assays show that RPA^1B2A3B unfolds telomeric G-quadruplexes known to inhibit replications fork progression. We also found that ATR deficiency can partially rescue short telomeres in rpa2a mutants, although plants exhibit defects in growth and development. Unexpectedly, the telomere shortening phenotype of rpa2a mutants is completely abolished in plants lacking the RTEL1 helicase. RTEL1 has been implicated in a variety of nucleic acid transactions, including suppression of homologous recombination. Thus, the lack of telomere shortening in rpa2a mutants upon RTEL1 deletion suggests that telomere replication defects incurred by loss of RPA may be bypassed by homologous recombination. Taken together, these findings provide new insight into how RPA cooperates with replication and recombination machinery to sustain telomeric DNA.

TELOMERES serve two vital functions: they shield the ends of eukaryotic chromosomes from eliciting a DNA damage response that would lead to end-to-end fusion, and they solve the end replication problem, averting progressive loss of terminal DNA sequences during each cell division cycle (Zakian 2012). Telomerase is required to maintain the extreme ends of chromosomes, but the bulk of telomere tracts, which range from a few hundred base pairs in yeast to tens of thousands of bases in some plants, are replicated by the conventional DNA replication machinery (Fajkus et al. 1995; Liti et al. 2009; Greider 2016).

Telomeres reach a length homeostasis through the interplay of forces that allow elongation (e.g., telomerase action and recombination), and those that cause shortening (deletional recombination and telomere trimming) (Li and Lustig 1996; Marcand et al. 1997; Henson et al. 2002; Greider 2016; Li et al. 2017). Telomere attrition can also occur during DNA replication due to the repetitive and heterochromatic nature of telomere sequences (Martínez and Blasco 2015). G-rich telomere repeats are prone to forming a variety of intramolecular structures (Paeschke et al. 2010; Yang et al. 2017). One of the best studied is the G-quadruplex (G4)—four-stranded DNA bearing runs of guanine nucleotides paired via Hoogsteen hydrogen bonding (León-Ortiz et al. 2014). If not resolved, G4 DNA can hinder replication fork (RF) progression, causing fork stalling and collapse, and eventually loss of telomeric DNA via cleavage of the stalled fork at the G4 structure (León-Ortiz et al. 2014; Martínez and Blasco 2015; Simon et al. 2016). G4 structures have also been proposed to inhibit telomerase-mediated extension (Zahler et al. 1991), although this has been disputed (Moye et al. 2015).

Telomere maintenance requires augmentation of the core DNA replication machinery, including specialized DNA helicases, nucleases, checkpoint proteins, and homologous recombination (HR) proteins (Badie et al. 2010; León-Ortiz et al. 2014; Martínez and Blasco 2015). Telomere proteins within the shelterin complex assist in the recruitment of helicases such as BLM, WRN, and RTEL1, and nucleases such as FEN1 to unwind and cleave G4 DNA, respectively (Martínez and Blasco 2015). In addition, RF stalling triggers the recruitment of ATR, master regulator of the S phase checkpoint, to stabilize and restart RFs that stall naturally in telomeric DNA, and are exacerbated at chromosome ends devoid of shelterin components (Verdun and Karlseder 2006; Sfeir et al. 2009).

HR-related proteins contribute to telomere replication in both telomerase-positive and -negative cells (Badie et al. 2010). In cells lacking telomerase, HR can be employed to effectively maintain telomeres through a process termed alternative lengthening of telomeres (ALT) (Henson et al. 2002). HR is also proposed to restart stalled forks caused by G4-induced DNA damage (van Kregten and Tijsterman 2014). Recent studies indicate that HR may promote bypass of G4 structures at transcribed genes via template switching (van Wietmarschen et al. 2018). The action of HR on stalled forks is tightly regulated and antagonized by a variety of helicases including BLM and RTEL1 in human cells (Barber et al. 2008; van Wietmarschen et al. 2018), Srs2 (functional analog of RTEL1) in yeast (Marini and Krejci 2010), and RTEL1 in Arabidopsis (Recker et al. 2014).

Several single-stranded DNA (ssDNA) binding proteins bind and unfold telomeric G4 DNA in vitro. These include replication protein A (RPA) (Salas et al. 2006); the telomere-associated CST (CTC1;Cdc13/STN1/TEN1) complex, a heterotrimer with structural similarity to RPA (Gao et al. 2007; Zhang et al. 2019); and the shelterin component protection of telomeres 1 (POT1) (Zaug et al. 2005). In addition, Audry et al. (2015) reported that, in fission yeast, RPA prevents the formation of telomeric G4 structures in vivo, arguing that RPA may help to facilitate telomere maintenance by promoting RF progression through telomere tracts.

Several studies in single-celled organisms have implicated RPA in telomere length regulation (Smith et al. 2000; Ono et al. 2003; Schramke et al. 2004; Kibe et al. 2007; Luciano et al. 2012; Audry et al. 2015). In Saccharomyces cerevisiae, RPA is enriched at telomeric ssDNA during late S phase, where it is proposed to facilitate both lagging and leading strand replication. In both budding and fission yeast, mutation of RPA subunits leads to telomere shortening (Ono et al. 2003; Schramke et al. 2004). In the Schizosaccharomyces pombe RPA1 mutant rad11-D223Y, telomere shortening is proposed to reflect impaired lagging-strand replication as a result of G4 formation, since overexpression of the G4 resolving helicase Pfh1 rescues the short telomere phenotype (Audry et al. 2015).In addition to a role for RPA in semiconservative replication of telomeric DNA, RPA, and RPA-related proteins also engage telomerase and stimulate its activity. For example, RPA interacts with Ku, Cdc13, and Est2 to promote telomerase activity at chromosome ends in budding yeast (Luciano et al. 2012). Similarly, in S. pombe, RPA associates with the telomerase RNA subunit, TLC1, and stimulates telomerase activity (Luciano et al. 2012). More recently, Tetrahymena thermophila proteins paralogous to the general RPA subunits, Teb1 (RPA1), Teb2 (RPA2), and Teb3 (RPA3) were defined as components of the telomerase holoenzyme complex (Upton et al. 2017).

How RPA impacts telomere maintenance and stability in multicellular eukaryotes is less clear. Human RPA transiently associates with telomeres during S phase, but is actively excluded from telomeres by the combined action of POT1, hnRNPA1, and the long noncoding telomeric RNA TERRA in late S phase (Flynn et al. 2012). Consequently, it was suggested that RPA is required only for semiconservative telomere replication. Contrary to this model, telomerase-positive human cell lines bearing a mutant RPA1 allele undergo progressive telomere shortening (Kobayashi et al. 2010), implying a defect in telomere capping, telomerase activity, or C-strand fill-in. However, in another study, RPA depletion had no effect on telomere maintenance in telomerase-positive human cells (Grudic et al. 2007).

Unlike yeast and most mammals, which harbor a single RPA heterotrimeric complex, the flowering plant Arabidopsis thaliana employs a multi-RPA protein complex system to perform various types of DNA metabolism (Aklilu et al. 2014; Eschbach and Kobbe 2014; Aklilu and Culligan 2016; Liu et al. 2017). The RPA family in Arabidopsis includes five RPA1 (RPA1A-RPA1E), two RPA2 (RPA2A; RPA2B), and two RPA3 (RPA3A; RPA3B) paralogs (Aklilu et al. 2014). At least 12 distinct RPA heterotrimeric complexes are thought to form (Elmayan et al. 2005; Aklilu et al. 2014; Eschbach and Kobbe 2014; Liu et al. 2017). Because of partial functional redundancy, null mutations in Arabidopsis RPA genes are not lethal, making it feasible to define unique contributions of RPA subunits in a wide range of chromosome biology (Elmayan et al. 2005; Aklilu et al. 2014). Indeed, genetic analysis has revealed that each of the five RPA1 subunits has evolved a specialized function in DNA replication (RPA1B and RPA1D), repair (RPA1C and RPA1E) or meiotic recombination (RPA1A and RPA1C) (Aklilu et al. 2014).

In this study we exploit the genetic plasticity of the Arabidopsis RPA complex to explore the role of RPA in telomere maintenance. We report that RPA subunits play a critical role in establishing telomere length set point, and the extent of this contribution varies among different RPA paralogs. We provide evidence that RPA helps to sustain telomere length, and the particular subunits of RPA most important for telomere length regulation are sufficient to unfold G4 DNA in vitro. Our data further indicate that loss of ATR partially rescues the telomere shortening phenotype in RPA mutants, while loss of the RTEL1 helicase fully rescues this phenotype. Together, these findings expand our understanding of how RPA works with replication and recombination machinery to facilitate telomere maintenance in multicellular eukaryotes.

Materials and Methods

Plant materials

rpa1 (rpa1a, rpa1b, rpa1c, rpa1d, rpa1e, rpa1b rpa1d, rpa1c rpa1e) and atr mutants are Salk T-DNA insertion lines and were previously described (Aklilu et al. 2014). rpa2a (SALK_129173), rtel1-1 (SALK_113285) were obtained from the Arabidopsis Biological Resource Center (ABRC, Ohio State University) and have been characterized (Elmayan et al. 2005; Recker et al. 2014). rpa2b-1 (SALK_067322) was obtained from ABRC. The stn1-1 (SALK_023504) and ctc1-3 (SALK_083165) mutant alleles were previously described (Song et al. 2008; Surovtseva et al. 2009). Primers used for PCR genotyping to isolate homozygous lines are shown in Supplemental Material, Table S1.

Wild-type (Col-0) and mutant plants were germinated and grown under long-day conditions (16 hr light/8 hr dark) at 22°, ∼50% relative humidity, and a light intensity of 100–150 μmol/m²/sec. For experiments using seedlings, seeds were sterilized using 10% bleach for 5 min followed by three rinses with sterile double distilled water. Seeds were then plated on half strength MS medium (Murashige and Skoog 1962) (CAS #: M10500-50, RPI Research Products International) with 1% agar (CAS#: 9002-18-0, Caisson Labs), and stratified at 4° for 2 days in the dark before being placed in a growth chamber. For soil-grown plants, 7-day-old plate-grown seedlings were transferred to soil growing medium (SUNGRO Horticulture, Seba Beach, Canada) in pots and grown under conditions described above.

Telomere length analysis, telomerase activity assays, and telomere fusion PCR

Telomere length was determined by terminal restriction fragment (TRF) analysis as described (Nigmatullina et al. 2016). Briefly, ∼200 plate-grown seedlings or 10 20-day-old soil-grown plants were pooled and used for genomic DNA (gDNA) extraction with 2× CTAB (2% hexadecyltrimethylammonium bromide, 100 mM Tris–HCl, 1.4 M NaCl, and 20 mM EDTA) buffer. gDNA (50 µg) was digested with Tru1l enzyme, and digestion products were resolved on a 0.9% agarose gel, transferred to a Hybond-N+ membrane (GE Healthcare), and hybridized with [³²P] 5′ end-labeled telomere probe (TTTAGGG)₆. TeloTool software was used for quantification of mean telomere length (MTL) (Göhring et al. 2014). The software calculates a standardized MTL by assessing the entire range of signal intensity along the smear profile. A Student’s t-test was used to compare mean MTL between wild type and mutant or different mutant lines. A P-value ≤0.05 was considered significant.

Telomerase activity was measured using the telomere repeat amplification protocol (TRAP) (Fitzgerald et al. 1996) using protein extracts prepared from 6-day-old seedlings of wild type or rpa mutants. Quantitative TRAP (qTRAP) was performed as described (Kannan et al. 2008).

End-to-end chromosome fusions were assessed using telomere fusion PCR (Heacock et al. 2004). DNA (125 ng) was mixed with 1× ExTaq buffer (TaKaRa), 200 μM dNTPs, 0.4 μM of each chromosome-specific subtelomeric primer, and 1 μl ExTaq enzyme in a 20 μl total volume. PCR mixes were incubated at 96° for 3 min, followed by 40 cycles at 94° for 30 sec, 55° for 45 sec and 72° for 2.5 min, with a final incubation at 72° for 9 min. PCR reaction products were separated by 1% agarose gel, transferred to a nylon membrane, and probed with a [³²P] 5′end-labeled oligonucleotide probe (TTTAGGG)₆. Telomere signals were detected using a typhoon FLA 9500 phosphorImager (GE healthcare), and data were analyzed using Quantity1 software (Bio-Rad).

RNA isolation, cDNA preparation, RT-PCR

Total RNA was isolated from 7-day-old rpa2b-1 mutant seedlings using a Direct-zol RNA kit (Zymo Research). cDNA was synthesized from 1 µg of total RNA by using SuperScript III Reverse Transcriptase (Invitrogen) according to the supplier’s instructions. RT-PCR was performed with 125 ng of cDNA, 0.4 nmol of each forward (primer) and reverse (primer) primers, and 1× EmeraldAmp MAX PCR Master Mix (TaKaRa) in a total volume of 25 µl. PCR mixes were incubated at 96° for 2 min, followed by 40 cycles at 96° for 10 sec, 60° for 30 sec and 72° for 1.5 min, with a final incubation at 72° for 5 min. PCR products were separated by 1% agarose gel.

Purification of RPA^1B2A3B protein complex

Construction of pET-Duet(RPA1B)_amp and pET-Duet(RPA3B + RPA2A)_kana plasmids was described in (Eschbach and Kobbe 2014). For protein expression, the plasmids were cotransformed into BL21(DE3) cells. RPA^1B2A3B complex was purified on a ssDNA cellulose column that was pre-equilibrated in lysis buffer (25 mM Tris pH 7.5, 500 mM NaCl, 1 mM EDTA, 1 mM DTT, 10% glycerol), and eluted with elution buffer (25 mM Tris pH 7.5, 1.5 M NaCl, 50% ethylene glycol, 1 mM EDTA, 1 mM DTT, 10% glycerol). Fractions containing AtRPA^1B2A3B complex were pooled and dialyzed overnight against 2 l of dialysis buffer (25 mM Tris pH 7.5, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 10% glycerol). Coomassie staining indicated the protein was 80–90% pure. Formation of RPA heterotrimer was confirmed by size-exclusion chromatography followed by SDS gel. RPA^1B2A3B or human RPA1 (hRPA1) were injected onto Superdex 200 10/300 25 ml gel filtration column (GE Healthcare) equilibrated with GF buffer (25 mM Tris pH 7.5, 150 mM NaCl, 10% glycerol), and eluted with 30 ml of GF buffer with 0.2 ml/min flow and 0.5 ml fractions.

Electrophoretic mobility shift assay

A. thaliana telomeric DNA oligonucleotide (At24) (5′-GGGTTTAGGGTTTAGGGTTTAGGG-3′) was labeled with γ[³²P]ATP using T4 polynucleotide kinase. RPA^1B2A3B was diluted and preincubated (20 min at 4°) in buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM KCl or NaCl or LiCl, 1 mM DTT, 10% glycerol, 0.2 mg/ml of BSA, and 0.1 mM EDTA. Radiolabeled At24 (2 nM) was incubated with various amounts of protein in 10 µl of reaction buffer [50 mM HEPES (pH 7.9), 0.1 mg/ml of BSA, 100 mM KCl or NaCl or LiCl, and 2% glycerol] for 20 min at 20°. Samples were loaded on a native gel 1% agarose in 0.5× TBE buffer. After electrophoresis, the gel was dried and exposed on a phosphorimager screen, scanned with a Typhoon instrument (Molecular Dynamics). Quantification was conducted with ImageQuant version 5.1 as previously described (Lancrey et al. 2018).

Circular dichroism, fluorescence resonance energy transfer melting curves, and fluorescence titration

Circular dichroism (CD) spectra were recorded on a Jasco J-810 spectropolarimeter at 15° with 3 µM FAM-At24-TAMRA oligonucleotide (doubly labeled with a FAM and a TAMRA) in 10 mM cacodylic acid, 100 mM NaCl or KCl or LiCl, pH 7.2 as described (Tran et al. 2011). The melting of FAM-At24-TAMRA oligonucleotide was followed by fluorescence resonance energy transfer (FRET) on a FluoroMAX-3 spectrofluorimeter in 100 mM NaCl or 100 mM KCl as described in (Tran et al. 2011) except that the temperature was raised from 20 to 95°. Temperature of half dissociation (T_1/2) corresponds to an emission value of 0.5 of FAM emission normalized between 0 and 1. Fluorescence titrations were carried out on a SPEX Fluorolog spectrofluorimeter (Jobin Yvon Inc.) equipped with a circulating water bath to regulate the temperature of the cell holder. Fluorescence spectra of 100 nM FAM-At24-TAMRA were recorded at 20° in a buffer containing 100 mM KCl or NaCl, 2 mM MgCl₂, and 5 mM lithium cacodylate (pH 7.2). The protein (from 50 to 1000 nM) was directly added to the solution containing the oligonucleotides. The spectra were collected after 2 min of incubation while exciting at 470 nm. The percentage of opened labeled G4 units at 20° was calculated as previously described (Lancrey et al. 2018).

Data availability

The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article. Mutant lines and plasmids are available upon request. Supplementary Material R1 contains Table S1 and S2 and Figures S1–S8 including detailed descriptions of all supplemental files. Supplemental material available at figshare: https://doi.org/10.25387/g3.12045375. Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: RPA1A (AT2G06510), RPA1B (AT5G08020), RPA1C (AT5G45400), RPA1D (AT5G61000), RPA1E (AT4G19130), RPA2A (AT2G24490), RPA2B (AT3G02920), ATR (AT5G40820), CTC1 (AT4G09680), STN1 (AT1G07130), and RTEL1 (AT1G79950).

Results

RPA1 subunits required for DNA replication make a significant contribution to telomere length maintenance

Figure 1A shows a graphic depiction of the RPA protein family in A. thaliana. We began our analysis by testing if individual RPA1 proteins contribute to telomere maintenance. Bulk telomere length was analyzed in homozygous T-DNA insertion knockout mutants by the terminal restriction fragment (TRF) method. For each mutant line we grew and analyzed three successive generations propagated by self-pollination [generation 1 (G1) to generation 3 (G3)]. Telomeres in wild type A. thaliana range from 2 kb to 5 kb in the Col-0 accession (Shakirov and Shippen 2004). Because telomere length fluctuates within this range, we pooled and analyzed DNA from ∼10 individual plants to obtain a more comprehensive assessment of telomere length in a population of plants with a given genotype. We monitored MTL using TeloTool (Göhring et al. 2014). MTL for wild type telomeres in our study ranged from 2.9 kb to 3.2 kb. Analysis of the single rpa1 mutant lines revealed no statistically significant change in MTL compared to wild type through the three generations of the experiment (Figure 1B).

Figure 1.

RPA1 double mutants and the RPA2A single mutant reset telomere length equilibrium at a shorter set point. (A) Cartoon showing multiple paralogs of RPA subunits, defined functions of RPA1 subunits, and suggested heterotrimeric complex members. Solid arrows denote interactions supported by biochemical and molecular evidence, and dashed arrow inference from genetic data. (B–F) Results of TRF analysis for rpa (1a-1e) [B], rpa1b rpa1d [C], rpa1c rpa1e [D], rpa1a rpa1c [E], and rpa2a and rpa2b [F] mutants. G = Generation at homozygosity for the defect. For the TRF data shown for rpa1a rpa1c mutants (E), each lane contains DNA extracted from a different, single plant (= biological replicate). For other panels, each lane contains DNA extracted from a different pool (= biological replicate) of 10 20-day-old, soil grown plants (heterozygote and G1), or 200 plate-grown, 7-day-old, seedlings for G2–G7. Numbers below TRF images denote mean telomere lengths (MTL), over the generations shown, in kb (TRF length data from successive generations were combined for MTL analyses as they are stable across generations). TRF for each generation was performed independently with three biological replicates (Table S2). Only one biological replicate is shown in the figures. Asterisk denote statistically significant difference at P ≤ 0.05.

To test for functional redundancy among the RPA1 protein paralogs, we assessed telomere length through a multigenerational analysis (seven generations) for RPA1 paralogs with overlapping function in DNA replication (RPA1B and RPA1D), DNA repair (RPA1C and RPA1E), and meiosis (RPA1A and RPA1C) (Aklilu et al. 2014; Aklilu and Culligan 2016). The rpa1b rpa1d double mutant line was previously reported to have defects in DNA replication, growth, and development (Aklilu et al. 2014). TRF analysis revealed a dramatic drop in telomere length by an average of 1.4 kb in the first generation (Figure 1C). Notably, telomere length did not continue to decline over the seven subsequent generations analyzed. Instead, telomeres stabilized at an MTL of 1.5 kb. We conclude a new telomere length equilibrium or set point was established in plants lacking RPA1B and RPA1D that was about half the length of wild type.

RPA1C and RPA1E are mainly specialized for DNA damage repair. Since numerous DNA repair-related proteins are implicated in telomere maintenance (Vespa et al. 2005; Badie et al. 2010), we asked if RPA1C and RPA1E are also involved in this process. TRF analysis showed rpa1c rpa1e double mutants had short, stable telomeres over seven generations (Figure 1D). However, compared to the rpa1b rpa1d mutation, the reduction in MTL of 0.7 kb in rpa1c rpa1e mutants was more modest.

The RPA1A and RPA1C paralogs are partially redundant in meiotic-induced DNA double-strand break repair and recombination (Aklilu et al. 2014; Aklilu and Culligan 2016). As expected, (Aklilu et al. 2014), rpa1a rpa1c double mutants were completely sterile. Nevertheless, we were able to analyze telomeres of G1 homozygous rpa1a rpa1c plants segregated from double heterozygous parental lines. Similar to the DNA replication and DNA repair group mutants (Figure 1, C and D), rpa1a rpa1c had a short telomere length phenotype (2.5 kb) (Figure 1E). The length phenotype was even milder than in rpa1c rpa1e mutants with a loss of 0.5 kb MTL compared to wild type. As shown in Figure 1E, the TRF profile in rpa1c rpa1e and rpa1a rpa1c is more heterogenous overall, and the range of telomere lengths greater than in rpa1b rpa1d (Figure S2A). These findings suggest that individual RPA1 subunits have different impacts on telomere maintenance.

To test for other functional overlap among the larger RPA subunits, we assessed telomere length in all possible double mutant combinations of rpa1 (a–e). MTL in all of the double mutants was in the wild-type range (Figure S1B), indicating that there is no additional telomere-related functional overlap among the remaining pairs of RPA1 paralogs. We conclude among the five subunits of Arabidopsis RPA1, the DNA replication group, RPA1B and RPA1D, plays the leading role in maintenance of telomeres in Arabidopsis, followed by the DNA repair group, RPA1C and RPA1E, and lastly the meiotic group, RPA1A and RPA1C. Analysis of heterozygous RPA1 double and RPA2A single mutant plants (described below) suggest that the genes are not haploinsufficient for telomere maintenance since their TRF profile is similar to that of wild type.

RPA2A, but not RPA2B, contributes to telomere length maintenance:

Arabidopsis encodes two RPA2 subunits, RPA2A and RPA2B, which exhibit 40% and 60% amino acid identity and similarity, respectively (Figure S2A). Protein interaction studies and phylogenetic analysis suggest that the proteins may have separate functions (Eschbach and Kobbe 2014; Liu et al. 2017). The phenotype of rpa2a mutants (defects in cell division, plant growth, and development) is similar to that of rpa1b rpa1d double mutants (Elmayan et al. 2005; Aklilu et al. 2014), suggesting that the primary function of RPA2A resides in DNA replication. However, since rpa2a mutants are hypersensitive to DNA damage (Elmayan et al. 2005), it is also possible that RPA2A functions in DNA repair.

TRF analysis of rpa2a mutants revealed a telomere profile essentially the same as in rpa1b rpa1d mutants; MTL declined by ∼1.5 kb and this length was maintained for at least three generations (Figure 1F). To test the role of RPA2B, we characterized a new RPA2B allele, dubbed rpa2b-1, which harbors a T-DNA insertion in the eighth exon. Mutants homozygous for the insertion fail to produce mRNA indicating the rpa2b-1 allele is null (Figure S2, B and E). In contrast to rpa2a, TRF analysis of the rpa2b mutant revealed no change in telomere length (Figure 1F). This finding was somewhat surprising given that RPA2B preferentially assembles with RPA1A, RPA1C, and RPA1E (DNA repair and meiotic groups) over RPA1B or RPA1D (replication) in vitro (Eschbach and Kobbe 2014; Liu et al. 2017). Because both the DNA repair and meiosis group RPA1 mutants exhibit some degree of telomere shortening (Figure 1, D and E), we expected RPA2B would contribute to telomere length maintenance. Instead, our findings suggest that RPA2A can substitute for RPA2B, assembling with RPA1A, RPA1C, or RPA1E to help fulfill the RPA2B telomere-related functions. These genetic data, combined with previous biochemical analysis (Figure 1, A and F; Kobayashi et al. 2010; Flynn et al. 2012; Upton et al. 2017), raise the possibility that RPA2A can functionally associate with all five of the RPA1 subunits. Since the telomere maintenance defect associated with rpa2a deletion is essentially identical to that of rpa1b rpa1d double mutants, for some downstream experiments we simplified our genetic analysis by using rpa2a mutants.

Telomerase activity and chromosome end protection are intact in rpa mutants:

We considered several possible mechanisms for telomere shortening in RPA mutants. One is that telomerase activity or its recruitment to telomeres is decreased, causing a reduction in telomere repeat addition. To test this, we first used the TRAP to monitor telomerase activity. Conventional TRAP revealed no obvious difference in the level of telomerase enzyme activity in rpa1b rpa1d or rpa1c rpa1e relative to wild type (Figure 2A). We also employed a quantitative telomerase activity assay, qTRAP, to measure telomerase activity in different mutant backgrounds, including rpa2a single mutants and rpa1b rpa1d, rpa1c rpa1e, and rpa1a rpa1c double mutants. Strikingly, telomerase activity was demonstrably higher in all of the RPA mutants tested. We suspect that this result reflects greater access of telomerase to the ssDNA primer during the reaction with extracts lacking RPA subunits (Figure 2B). Further evidence that telomerase was fully active and can be successfully recruited to chromosome ends was the heterogenous “smeary” TRF profile in all of the RPA mutants (Figure 1, B–F). In marked contrast, plants with a telomerase deficiency exhibit a telomere profile of sharply defined bands, corresponding to individual chromosome ends that are no longer subjected to the stochastic action of telomerase (Riha et al. 2001). These data argue that telomerase enzyme activity and its ability to engage chromosome ends are not substantially perturbed in plants lacking RPA.

Figure 2.

Telomerase activity is not reduced in RPA mutants. (A) Results from the standard telomere repeat amplification protocol (TRAP) assay. Total protein was extracted from 30 7-day-old seedlings and ∼0.5 μg protein was used for the primer extension assay. Products were resolved by 8% PAGE. Pairs of lanes show biological replicates for each line. (B) Results for quantitative TRAP (qTRAP) assay. Total protein was extracted from 30 7-day-old seedlings or 20 flower buds from 10 plants (for rpa1a rpa1c) and ∼0.05 μg protein was used for the assay. Bar graphs show mean telomerase activity and standard error values from three biological replicates. Unpaired t-test was used for pairwise comparison only between a wild type and a specific mutant line. Asterisk below paired bars denote statistically significant difference at P ≤ 0.05. We suspect the increase in telomerase activity in qTRAP of RPA mutants reflect decreased competition of telomerase with ssDNA binding proteins for the ssDNA oligonucleotide primers used in the telomerase reaction.

Another possible explanation for shorter telomeres in RPA mutants is that chromosome ends are partially deprotected, and accessible to nuclease attack. Our data are not consistent with this notion for two reasons. (1) Telomeres in RPA mutants reach a length equilibrium that is maintained over many generations. (2) Telomere ends are protected from end-joining reactions in RPA mutants. Loss of chromosome end protection is associated with end-to-end chromosome fusion. We tested for this using telomere fusion PCR, a method that employs primers directed at unique subtelomeric sequences on different chromosome arms to amplify across the junction of covalently fused telomeres (Heacock et al. 2004). While abundant telomere fusions were detected in plants lacking core components of CST (CTC1 or STN1) (Song et al. 2008; Surovtseva et al. 2009), we found no telomere fusions in RPA mutants (Figure S3), indicating chromosome termini are functionally intact.

The Arabidopsis RPA^1B2A3B complex unfolds telomeric G4 structures in vitro:

Another way that RPA can promote telomere length maintenance is by preventing the formation of G4 structures, which are inhibitory to RF progression (Wold 1997; Audry et al. 2015). Unfortunately, repeated attempts to directly observe the accumulation of G4 DNA in vivo in rpa mutants using a G4 specific antibody were unsuccessful, likely due to technical issues. Therefore, we employed a biochemical approach to ask if the Arabidopsis RPA subunits implicated in establishing telomeric DNA length set point are capable of binding and unfolding G4 telomeric DNA in vitro. The replication group RPA1 subunit, RPA1B, was previously shown to preferentially form a complex with RPA2A and either of the smaller subunits, RPA3A or RPA3B (Eschbach and Kobbe 2014). We co-expressed RPA1B and RPA2A with RPA3A or RPA3B in Escherichia coli and purified RPA heterotrimeric complexes (Figure S4A). Proteins were estimated to be between 80 and 90% pure, and heterotrimer formation was confirmed by size-exclusion chromatography followed by SDS-PAGE (Figure S4, B and C).

We obtained a sufficient amount of pure RPA^1B2A3B heterotrimer for biochemical assays. G4 structures were generated from a 24mer Arabidopsis telomeric oligonucleotide (GGGTTTA)₃GGG (AtTelo). AtTelo G4 formed in both Na⁺ and K⁺ solutions, but showed higher stability in K⁺ (Tran et al. 2011). AtTelo G4 folding was prevented with Li⁺ cation (You et al. 2017). We next performed electrophoretic mobility shift assays (EMSA) to check for RPA^1B2A3B binding to AtTelo G4 DNA in Na⁺, K⁺, or Li⁺ buffers. As illustrated in Figure 3, a shift in the migration of AtTelo G4 was observed when incubated with a gradient concentration of RPA^1B2A3B, demonstrating that the RPA^1B2A3B complex efficiently bound G4 DNA. Quantification of the EMSA indicated that RPA^1B2A3B had the lowest affinity for the more stable telomeric G4 assembled in the presence of K⁺ and the highest affinity for the unstructured telomeric sequence, in the presence of Li⁺ (Figure 3E). Hence, like human RPA, the affinity of RPA^1B2A3B for telomeric G4 is inversely correlated with the stability of the G4 structure formed by telomeric sequence (Safa et al. 2014).

Figure 3.

Recombinant RPA^1B2A3B protein binds to Arabidopsis telomeric DNA (At24). (A) Cartoon showing the specific RPA complex and subunits used in this particular assay. (B and C) [³²P]-labeled At24 (2 nM) was incubated with increasing amounts of RPA^1B2A3B in the presence of 100 mM NaCl [Na⁺] (B), KCl [K⁺] (C) or LiCl [Li⁺] (D). (E) Percentages of At24 bound to RPA^1B2A3B in NaCl (red), KCl (blue) and LiCl (green) as a function of RPA^1B2A3B concentration.

To investigate if RPA^1B2A3B is able to unfold AtTelo G4, FRET experiments were conducted (Salas et al. 2006) with 24mer AtTelo (At24) labeled with the same two fluorophores, FAM-At24-TAMRA. CD spectra in Na⁺ and K⁺ solutions (Figure S5A) showed that FAM-At24-TAMRA folded into G4 structures in Na⁺ and K⁺ conditions, but with different conformations, while thermal melting showed a higher stability in K⁺ (Figure S5, B and C). As the ratio r= [RPA^1B2A3B]/[FAM-At24-TAMRA] increased, FAM emission was stimulated, while TAMRA fluorescence decreased (Figure 4, B and C), indicating that FRET was suppressed and that RPA^1B2A3B unfolded G4 in the presence of Na⁺ and K⁺. Quantification of G4 unfolding (Figure 4D) showed that, in our conditions, RPA^1B2A3B unfolded FAM-At24-TAMRA up to 90% in Na⁺ against 30% in K⁺. The very low opening of G4 observed in K⁺ is due to decreased RPA^1B2A3B binding (Figure S6). This may result from steric hindrance of fluorophores in different G4 conformations present in K⁺, and/or different G4 conformations and stability in K⁺ (Figure S5A). Any one of these factors is expected to decrease RPA^1B2A3B binding. We conclude that RPA^1B2A3B binds and unfolds Arabidopsis telomeric G4 in vitro. Our previous in vitro kinetics work with human RPA (Salas et al. 2006) showed that hRPA unfolds G4 in a time that is compatible with cellular processes. Considering G4 unfolding activity is conserved across eukaryotic RPAs (our current and previous work), we anticipate that AtRPA would behave in a similar way to hRPA.

Figure 4.

Arabidopsis RPA^1B2A3B unfolds Arabidopsis telomeric G4 units. (A) Cartoon showing the specific RPA complex and subunits used in this particular assay. (B and C) Fluorescence emission spectra (excitation wavelength 470nm) of FAM-At24-TAMRA oligonucleotides with increasing amounts of RPA^1B2A3B in Na⁺ (A) or K+ (B) solution. AtTelo (At24) oligonucleotide form an intramolecular (monomolecular) G4 structure as shown by Tran et al. (2011). Arrows indicate the evolution of emission spectra with increasing RPA^1B2A3B concentrations. r represents [RPA^1B2A3B]/[FAM-At24-TAMRA] ratio. (D) Percentages of unfolded labeled G4 (FAM-At24-TAMRA) units as a function of [RPA^1B2A3B]/[FAM-At24-TAMRA] ratio in Na⁺ (red) and K⁺ (blue) solution based on the data presented in (B and C). Data were derived from two experimental replicates.

ATR mutation partially rescues short telomeres in rpa mutants:

To further explore the possibility that telomere shortening in rpa mutants is related to RF stalling, we investigated how the simultaneous loss of RPA and ATR impacts telomere length maintenance. Previously it was shown that structural maintenance of chromosomes (SMC5/6 complex) not only stabilizes and resolves stalled RF, but also is required for loading RPA onto stalled RF to stabilize ssDNA for downstream HR-dependent replication restart (Irmisch et al. 2009). In Arabidopsis, mutation of SNI1, a subunit of the SMC5/6 complex, results in DNA fragmentation and growth defects. Interestingly, ATR inactivation reverses both phenotypes in the sni1 mutant (Yan et al. 2013).

We reasoned that, if stalled RF lead to telomere truncation in RPA mutants, the short telomere phenotype might be rescued by deleting ATR. Although plants lacking ATR are viable (Krysan et al. 1999), we were unable to isolate a homozygous rpa2a atr double mutant line (we genotyped 128 plants segregated from F2 RPA2A^−/− ATR^+/−), suggesting rpa2a atr is lethal. As a alternate approach, we obtained RPA2A^−/− ATR^+/− mutants and analyzed their telomere lengths. As shown in Figure 5A and Figure S7, MTL rose from 1.6 kb in the rpa2a mutant to 1.9 kb in RPA2A^−/− ATR^+/−. To test whether the impact of ATR depletion was specific to rpa2a, we generated rpa1c rpa1e atr triple mutants and compared their telomere lengths to rpa1c rpa1e double mutants. Strikingly, even though telomeres were longer in rpa1c rpa1e mutants compared to rpa2a, the same net increase in MTL was observed when ATR was deleted in this genetic background (increase from 2.3 kb to 2.6 kb) (Figure 5B). Thus, ATR mutation can partially rescue the short telomere phenotype in both rpa2a and rpa1c rpa1e mutants.

Figure 5.

ATR mutation partially rescues the short telomere phenotype in rpa mutants, but not the growth defects. (A) TRF analysis for rpa2a ATR ^+/−. Image is from one experiment. Figure S7 shows quantitative analysis from three biological replicates. (B) TRF results for rpa1c rpa1e triple mutant. DNA was pooled from ten soil grown plants for G1 and 200 plate grown seedlings for rpa1c rpa1e (G2-G3), rpa1c rpa1e atr (G2-G3). (C) Morphological phenotype of 43-day-old WT, atr, rpa2a, and rpa2a ATR ^+/− plants. (D) Siliques from 50-day-old wild type (WT), atr, rpa2a, rpa2a ATR ^+/− plants. (E) Siliques from 40-day-old WT and rpa1c rpa1e plants. (F) 30-day-old rpa1c rpa1e triple mutant plants showing a shoot phenotype. Compared to WT, the triple mutant has curlier, smaller rosette leaves.

We noted that ATR^+/− mutation worsened the growth and development phenotypes associated with the absence of RPA subunits. In the case of RPA2A^−/− ATR^+/−, inflorescences were shorter, rosette leaves were smaller and the siliques were shorter with fewer seeds (Figure 5, C and D). Likewise, rpa1c rpa1e atr mutants had shorter roots, smaller and curly leaves, and exhibited earlier flowering with smaller siliques bearing fewer seeds (Figure 5, E and F and Figure S8). These data imply that despite the partial rescue of short telomeres in plants deficient in RPA and ATR, genome integrity is compromised.

RTEL1 mutation completely rescues short telomeres in rpa2a mutants:

The multifunctional RTEL1 helicase protein has been reported to dismantle T-loops, suppress HR, and counteract telomeric G4 DNA to ensure RF progression and stability of telomeres (Vannier et al. 2012). To investigate whether any of these phenomena might contribute to the establishment of a shorter telomere length set point in rpa mutants, we created rpa2a rtel1 double mutants by crossing heterozygous plants for each gene. If accumulation of G4 DNA is responsible for RF stalling in the telomere tract in rpa mutants, we would expect further telomere shortening in the double mutant as RTEL1 would be unable to mitigate the impact of increased G4 DNA. Alternatively, if RTEL1 acts to suppress HR (Barber et al. 2008; Marini and Krejci 2010; Recker et al. 2014), which is normally activated to bypass stalled RFs (van Wietmarschen et al. 2018), then we might observe telomere lengthening (or rescue of short telomeres) when RTEL1 is mutated in a rpa2a background. Consistent with previous studies of Arabidopsis rtel1-1 mutants (Recker et al. 2014; Hu et al. 2015), plants lacking RTEL1 had a slightly longer MTL (by approximately 100 bp) compared to wild type (Figure 6A). Strikingly, MTL in rpa2a rtel1 double mutants was even higher than in rpa2a single mutants, and was similar to the MTL in rtel1-1 single mutants (Figure 6A). This finding suggests that RTEL1 is required to enforce telomere shortening in rpa2a mutants, and is consistent with the hypothesis that RTEL1 functions to inhibit HR-mediated alternative telomere lengthening (ALT) pathways. In contrast to rpa2a mutants deficient in ATR, the morphological phenotype of rpa2a rtel1 double mutants was essentially indistinguishable from rpa2a plants (Figure 6B), suggesting that genome integrity is largely retained in plants lacking both RPA2A and RTEL1.

Figure 6.

RTEL1 mutation rescues telomere shortening in rpa2a mutants. (A) Results of TRF analysis for rpa2a rtel1. Each lane includes DNA extracted from 10 plants for analysis. (B) RTEL1 mutation in the rpa2a mutant background does not lead to further growth and developmental defects. rtel1 mutants display minor growth defects (smaller leaves) (Hu et al. 2015). rpa2a rtel1 double mutants show a phenotype similar to the rpa2a single mutant.

Discussion

Proper replication of DNA is necessary to ensure stability and faithful transmission of genomic information from parent to daughter cells. The natural propensity of G-rich telomeric sequence to form secondary structures including G4 DNA impedes RF progression and can cause telomere attrition. ssDNA binding proteins and a range of DNA repair factors help to circumvent RF stalling at telomeres (Jain and Copper 2010; Martínez and Blasco 2015). While studies in yeast established a role for RPA in telomere maintenance (Smith et al. 2000; Ono et al. 2003; Audry et al. 2015), the contribution of this complex at telomeres in multicellular eukaryotes has been unclear. In this work, we exploit the multi RPA system in Arabidopsis to investigate how RPA impacts the maintenance of telomere length and integrity.

We found no evidence that the five single RPA1 mutants (rpa1a-rpa1e) alter MTL. However, analysis of telomere length in double RPA1 mutants revealed that all of the RPA1 subunits contribute to some extent to the establishment of telomere length equilibrium. We uncovered a functional hierarchy wherein RPA1 paralogs from the replication group (RPA1B and RPA1D) play the major role in determining telomere length set point, followed by paralogs from the repair (RPA1C and RPA1E) and then meiosis (RPA1A and RPA1C) groups. We also found separation-of-function between the two mid-size RPA subunits, with RPA2A, but not RPA2B, impacting telomere length equilibrium. Notably, RPA2A mutation resulted in comparable telomere shortening to replication group RPA1 mutants, supporting previous findings that RPA2A has the capacity to form a complex with RPA1B (Eschbach and Kobbe 2014; Liu et al. 2017), and likely other RPA1 subunits (Kobayashi et al. 2010; Flynn et al. 2012; Upton et al. 2017; this study).

Telomere length in rpa1 double and rpa2a single mutants was reduced in the first generation after RPA depletion and this shorter length was faithfully maintained through subsequent generations, indicating a new length equilibrium has been established. We suspect that telomeres do not continue to shorten in rpa1 double mutants because of partial functional redundancy among the five RPA1 subunits. This functional overlap would enable telomere length homeostasis to be established in rpa1 double mutants, albeit at a shorter length set point. Similarly, we contend that establishment of a new shorter telomere length in rpa2a mutants reflects functional redundancy of RPA2A and RPA2B in telomere replication.

In budding yeast, telomere shortening in cells carrying an RPA mutation (rfa10D228Y) has been linked to defects in chromosome end integrity (Smith et al. 2000). However, our data indicate that chromosome ends are fully protected in plants bearing rpa mutations. Since RPA is also implicated in telomerase regulation in yeast (Ono et al. 2003; Schramke et al. 2004; Luciano et al. 2012) and RPA paralogs assemble into the Tetrahymena telomerase holoenzyme (Upton et al. 2017), we also considered the possibility that short telomeres in rpa mutants were caused by decreased telomerase activity. We found telomerase enzyme activity was not diminished in the mutants. Further, the telomere tracts in rpa mutants displayed a highly heterogeneous profile, in marked contrast to the sharp banding pattern displayed by telomeres in telomerase null mutants (Riha et al. 2001). While we cannot formally exclude the possibility that G4 DNA at chromosome ends partially impedes telomerase access (Zahler et al. 1991), the loss of 1.5 kb of telomeric DNA in first generation rpa1b rpa1d double mutants and rpa2a single mutants is significantly greater than telomeric DNA attrition in a telomerase null mutant. Plants deficient in the telomerase catalytic subunit TERT lose only ∼500 bp per generation (Riha et al. 2001). Thus, our data do not support the conclusion that telomerase activity and/or recruitment to chromosome ends is defective in rpa mutants.

In contrast, several lines of evidence support the notion that shortened telomeres in rpa mutants reflect defective DNA replication. First, telomeres are shortest in mutants from the DNA replication group of RPA proteins, consistent with global defects in DNA replication. Interestingly, telomere length is more heterogenous and the range of telomere lengths broader for mutants from the repair and meiosis groups compared to the replication group. It is possible that RF stalling is not as severe for rpa mutations that primarily impact repair and meiotic recombination functions. In this scenario, intact RPA1B and RPA1D subunits would partially compensate for the contributions of repair and meiosis group subunits in telomeric DNA replication, but less so in the other direction.

Second, our biochemical data demonstrate that the A. thaliana RPA subunits implicated in establishing telomere length equilibrium can efficiently unwind G4 DNA in vitro. Although our analysis focused on the activity of RPA^1B2A3B, 1 of the 12 possible RPA complexes that can form in Arabidopsis (Figure 1A), an untested but intriguing possibility is that the varying contributions of large subunit RPAs in telomere length set point reflect differences in the G4 binding and unfolding activity. Audry et al. (2015) reported that fission yeast RPA (Rad11) prevents in vivo G4 formation on lagging strand telomeres and, similar to Arabidopsis RPA, mutation in Rad11 causes telomere shortening (Ono et al. 2003). RPA could potentially protect against telomere fragility in Arabidopsis by unfolding G4 DNA within the G-rich lagging strand to facilitate telomere replication. Alternatively, RPA might help to prevent the formation of G4 structures ahead of the RF when telomeres are transcribed to generate TERRA (Zhang et al. 2019).

Third, our genetic analysis of rpa atr double mutants suggests that these plants may have the capacity to circumvent stalled RF by enabling a replication bypass pathway such as HR to promote telomere length maintenance. We found that the shorter telomeres of rpa mutants are partially rescued in both RPA2A^−/− ATR^+/− double mutants and in rpa1c rpa1e atr triple mutants. The observation that ATR depletion can enable telomeres in rpa mutants to be extended is interesting, given the role of ATR in facilitating telomerase recruitment and chromosome end protection in other organisms. ATR mutation in S. pombe results in shorter telomeres due to impaired telomere protection and telomerase recruitment (Moser et al. 2009). Moreover, in human cells RF stalling promotes ATR-mediated recruitment of telomerase (Tong et al. 2015). ATR also promotes telomere maintenance in Arabidopsis. Telomeres in plants doubly deficient in ATR and TERT (Vespa et al. 2005) shorten at an accelerated pace, implying that ATR modulates a nontelomerase based mechanism for telomere maintenance. While we cannot formally rule out the possibility that rescue of telomere length in rpa atr mutants is due to unrestricted access of telomerase to broken DNA replication forks, an alternative explanation is that ATR negatively regulates HR, which would otherwise overcome RF stalling in telomere tracts devoid of RPA. Consistent with this notion, ATR mutation results in longer telomeres in telomerase-deficient mouse embryonic fibroblasts by relieving ATR-dependent suppression of HR at telomeres (McNees et al. 2010). The growth and developmental defects of plants lacking RPA and ATR can be explained by checkpoint bypass, but why should Arabidopsis suffer increased telomere attrition in atr tert mutants instead of telomere elongation? Vespa et al. (2005) proposed that ATR might play a role in loading end protection proteins onto telomeres. In this model, combined absence of ATR and TERT would preclude telomere extension by telomerase, and leave chromosome ends vulnerable to nucleolytic attack (Vespa et al. 2005). Perhaps HR activation is insufficient to compensate for telomere erosion under these circumstances.

An additional piece of evidence arguing that shortened telomeres in rpa mutants derive from defects in DNA replication comes from our analysis of the RTEL1 helicase. We show that rtel1 mutation fully rescues short telomeres in rpa2a mutants. This finding initially appears counterintuitive since RTEL1 has the capacity to resolve G4 DNA structures (Vannier et al. 2012) predicted to form in RPA deficient cells. Accordingly, loss of RTEL1 in an RPA deficient background is expected to impede RF progression through telomeres, leading to even shorter telomeres. Instead, the fully restored telomeres in rpa2a rtel1 double mutants are consistent with activation of an HR-mediated ALT mechanism for telomere replication. RTEL1 is known to suppress HR in Arabidopsis, humans, yeast, and worms (Barber et al. 2008; van Kregten and Tijsterman 2014; van Wietmarschen et al. 2018). Moreover, HR has been shown to bypass G4 induced stalled RF, without loss of DNA, via polymerase template switching, and this activity of HR is suppressed by BLM helicase (van Wietmarschen et al. 2018). Thus, while our data do not directly address the involvement of HR in the rescue of telomere length in rpa2a rtel1 mutants, we speculate that, in the absence of RTEL1, the G4 structures that accumulate in rpa2a-deficient plants can be efficiently bypassed by HR followed by fork restart. This pathway would enable efficient replication of telomeric DNA and restoration of wild-type telomere length homeostasis.

A final proposition stemming from our study is that HR contributes to telomere maintenance in parallel with telomerase in plants lacking RPA. We and others find that Arabidopsis telomeres are slightly elongated when RTEL1 is mutated (Recker et al. 2014; this study). Strikingly, however, in rtel1 tert double mutants, telomere shortening is exacerbated (Recker et al. 2014; Olivier et al. 2018). Arabidopsis RTEL1 has been postulated to suppress HR and dismantle T-loops to facilitate telomerase activity, similar to the role of RTEL1 in mice (Vannier et al. 2012). Olivier et al. (2018) recently suggested that RTEL1 could promote HR in Arabidopsis telomerase mutants by dismantling D-loops during HR dependent restart of telomeric stalled RFs. However, if RTEL1 promotes HR, we would expect similar-defects in telomere maintenance in rtel1 rpa2a double mutants as in rpa2a single, and not full rescue of the telomere length defect. Instead, our data are consistent with the hypothesis that RTEL1 deletion facilitates telomere replication in rpa2a mutants by enabling HR-dependent bypass of telomeric G4 structures.

In conclusion, our results provide new insight into the role of RPA in telomere replication, and the back-up pathways that ensure genome integrity when RPA is compromised. More broadly, our findings raise intriguing questions concerning the interplay of paralogous RPA complexes and closely related RPA-like proteins such as CST. The unprecedented ability in Arabidopsis to genetically dissect individual contributions of RPA and RPA-like complexes (Song et al. 2008; Surovtseva et al. 2009; Leehy et al. 2013; Aklilu et al. 2014) provides a unique opportunity to explore fundamental aspects of telomere biology and other facets of DNA metabolism.