tRNA ADENOSINE DEAMINASE 3 is required for telomere maintenance in Arabidopsis thaliana

Abstract

Telomere length maintenance is influenced by a complex web of chromatin and metabolism-related factors. We previously reported that a lncRNA termed AtTER2 regulates telomerase activity in Arabidopsis thaliana in response to DNA damage. AtTER2 was initially shown to partially overlap with the 5’ UTR of the tRNA ADENOSINE DEAMINASE 3 (TAD3) gene. However, updated genome annotation showed that AtTER2 was completely embedded in TAD3, raising the possibility that phenotypes ascribed to AtTER2 could be derived from TAD3. Here we show through strand-specific RNA-Seq, strand-specific qRT-PCR and bioinformatic analyses that AtTER2 does not encode a stable lncRNA. Further examination of the original tad3 (ter2–1/tad3–1) mutant revealed expression of an antisense transcript driven by a cryptic promoter in the T-DNA. Hence, a new hypomorphic allele of TAD3 (tad3–2) was examined. tad3–2 mutants showed hypersensitivity to DNA damage, but no deregulation of telomerase, suggesting that the telomerase phenotype of tad3–1 mutants reflects an off-target effect. Unexpectedly, however, tad3–2 plants displayed progressive loss of telomeric DNA over successive generations that was not accompanied by alteration of terminal architecture or end protection. The phenotype was exacerbated in plants lacking the telomerase processivity factor POT1a, indicating that TAD3 promotes telomere maintenance in a noncanonical, telomerase-independent pathway. The transcriptome of tad3–2 mutants revealed significant dysregulation of genes involved in auxin signaling and glucosinolate biosynthesis, pathways that intersect the stress response, cell cycle regulation and DNA metabolism. These findings indicate that the TAD3 locus indirectly contributes to telomere length homeostasis by altering the metabolic profile in Arabidopsis.

Introduction

Telomeres safeguard the genome by preventing chromosome ends from eliciting a DNA damage response and ensuring that terminal DNA sequences can be faithfully maintained (Shay and Wright 2019). Due to the nature of eukaryotic DNA replication, telomeres culminate in a single-stranded extension termed the G-overhang (Sandhu and Li 2017), which acts as a substrate for the addition of telomeric repeats by telomerase. Plant telomeres are unusual in that one-half of their chromosome ends terminate in a G-overhang, and the other half in a blunt end bound by the Ku complex (Kazda et al. 2012). Loss of Ku triggers extensive telomerase-dependent telomere elongation, presumably because blunt ends are converted to telomerase-accessible G-overhangs (Kazda et al. 2012; Valuchova et al. 2017). This unusual telomere architecture may further enhance genome stability, which seems advantageous given the sessile lifestyle of plants (Nelson and Shippen 2012).

Telomere length homeostasis is modulated by a host of factors. At the telomere, components of the shelterin complex, particularly the TTP1/POT1 heterodimer, enhance telomerase activity and processivity on human telomeric DNA (Wang et al. 2007). In Arabidopsis POT1a associates with the telomerase ribonucleoprotein complex (RNP) and stimulates its repeat addition processivity (Surovtseva et al. 2007; Renfrew et al. 2014; Arora et al. 2016). Plants deficient in POT1a undergo telomeric DNA attrition at a rate similar to the amorphic telomerase (AtTERT) mutant (Surovtseva et al. 2007). The progressive loss of telomeric DNA in telomerase mutants ultimately causes a critical length threshold to be breached, activating a DNA damage response that leads to telomere fusion and genome-wide instability. Arabidopsis telomeres normally span 2–5 kb in length; telomere tracts shorter than 1kb have an increased probability of being recruited into end-to-end chromosome fusions (Heacock et al. 2004). Thus, an optimal telomere length set point must be established to maintain genome integrity (Watson and Shippen 2007; Chiang et al. 2010).

In addition to canonical telomere-associated factors, genetic screens performed in Saccharomyces cerevisiae and Schizosaccharomyces pombe demonstrate that telomere length is also influenced by a wide variety of “non-telomeric” genes that function in various aspects of DNA metabolism, chromatin modification, vesicular trafficking, RNA metabolism, ribosome metabolism and translation (Askree et al. 2004; Ungar et al. 2009). Perturbation of cell cycle progression can also alter telomere length. For example, mutation of RAD1, a component of the intra-S DNA damage checkpoint, leads to telomere shortening in S. pombe (Nakamura et al. 2002). Similarly, Rad1 functions as a positive regulator of telomere length in mammals, working in concert with Hus1 and Rad9 in the 911 complex (Francia et al. 2007).

Telomere dysfunction induces programmed cell death (PCD) in plant meristems to eliminate genetically unstable cells (Boltz et al. 2012; Amiard et al. 2014). PCD activation is essential for cell differentiation and proper development and is also involved in pathogen and environmental stress responses (Locato and De Gara 2018). PCD activation involves various kinds of molecular signals including plant hormones, calcium and reactive oxygen species (ROS) (Huysmans et al. 2017). Different hormonal pathways are interconnected to fine-tune PCD via transcriptional regulation. The auxin hormone regulates plant growth, and under normal conditions concentrates at the quiescent center of the root stem cell niche. Under abiotic stresses, many of which induce the accumulation of reactive oxygen species (ROS), auxin levels decline causing PCD in root tissues (KrishnaMurthy and Rathinasabapathi 2013; Hong et al. 2017). Auxin signaling also controls cell cycle progression by mediating activation of CdC2. CdC2/Cdk2 kinase activity is necessary for expression of telomerase activity at early S phase (Tamura et al. 1999; Yang et al. 2002; Ren et al. 2004). Thus, telomerase is a downstream target of auxin signaling pathway.

Telomerase is comprised of two core components, the catalytic subunit TERT and a long non-coding RNA (lncRNA) TER/TR (Musgrove et al. 2018) that serves as a template for telomere repeat addition (Egan and Collins 2012). In A. thaliana, two lncRNAs were initially identified as telomerase subunits (Cifuentes-Rojas et al. 2011). AtTER1 was uncovered through partial purification of telomerase, and proposed to be the canonical telomerase RNA subunit. AtTER2, expressed from a locus partially overlapping the tRNA Adenosine Deaminase 3 (TAD3) gene, was uncovered by BLAST based on its high sequence similarity to AtTER1 (Cifuentes-Rojas et al. 2011, 2012). Subsequent studies indicated that AtTER2 was stabilized and functioned to down-regulate telomerase activity in response to DNA double-strand breaks (Cifuentes-Rojas et al. 2012; Xu et al. 2015). We recently employed an unbiased RIP-seq approach to identify lncRNAs associated with active telomerase under native conditions and failed to recover AtTER1 (Song et al. 2019). Instead, a single lncRNA, AtTR, was significantly enriched. Further analysis by our lab and others revealed that AtTR was the bona fide telomerase RNA subunit in A. thaliana (Fajkus et al. 2019; Song et al. 2019; Dew-Budd et al. 2020).

A new annotation of the A. thaliana genome, Araport11, extended the 5’ UTR of TAD3 to now fully embed AtTER2. This updated annotation prompted us to re-examine the TER2/TAD3 locus to assess whether the phenotypes originally ascribed to TER2 might instead result from mutation of TAD3. tRNA Adenosine Deaminase 3 (TAD3) catalyzes the deamination of adenosine at position 34 of the tRNA anticodon loop into Inosine to facilitate wobble base pairing (Torres et al. 2014b). Yeast and plant TAD3 amorphic mutants are inviable (Zhou et al. 2014; Gerber and Keller 2017). Similarly, loss of TAD3 in fission yeast compromises cell survival by affecting cell cycle progression (Tsutsumi et al. 2007). Decreased expression of human TAD3 impacts RNA editing for several tRNA species and is associated with intellectual disability (Torres et al. 2014a). Notably, TAD3 was uncovered in a genetic screen in S. cerevisiae as one of the essential genes that impacts telomere length maintenance (Ungar et al. 2009).

Here we show through strand-specific RNA-Seq, strand-specific qRT-PCR, and bioinformatic analyses that AtTER2 does not encode a stable lncRNA, and the telomere-related functions from this locus derive from the TAD3 gene. Through analysis of additional TAD3 mutant alleles, we report that hypomorphic tad3 mutants are hypersensitive to DNA damage, but TAD3 is not required to regulate telomerase activity in response to DNA damage. However, TAD3 is required for telomere length maintenance. This unanticipated function is independent of telomerase, and appears to reflect a broader role for TAD3 in modulating cellular metabolism.

Materials and methods

Plant materials, genotyping and genetic complementation

Seeds for tad3–1 (SAIL_556_A04), tad3–2 (SALK_121147) and WT Col-0 accessions along with T87 cell culture for the Col-0 accession were obtained from the ABRC stock center. Seeds were sterilized using 70% ethanol, 10% bleach and 0.1% Triton X-100 followed by vernalization for 2 days at 4°C. Seeds were plated on half Murashige and Skoog (RPI M10500) and 1% agar (Caisson A038) supplemented with 1% sucrose. Plants were grown in soil in controlled growth chambers maintained at 22°C under long day light conditions. Photographs to assess plant growth and development were captured using a digital camera.

Genotyping (primer sequences in S1 table) was performed with leaf DNA and emerald enzyme master mix (Clontech RR310A). pot1a tad3–2 double mutants and ku70 tad3–2 double mutants were generated by crossing plants heterozygous for pot1a and tad3–2 or ku70 and tad3–2 followed by segregating progeny for multiple generations. For genetic complementation, 3-week-old G2 tad3–2 plants were transformed with Agrobacterium (GV3101) cells harboring the plasmid pCBK05::NP^TAD3::TAD3 using the floral dip method (Zhang et al. 2006). Resistance to BASTA and Carbenicillin was used to select for true transformants in the next generation (G3).

RNA-Seq, transcriptome data visualization and analysis, and qRT-PCR

RNA extracted from 6-day-old seedlings was used to make RNA libraries in triplicate using the Illumina TruSeq® Stranded Total RNA Library Prep Plant (Catalog no. 2002061). After trimming the raw sequences using the Trimomatic program (Galaxy Europe), datasets were concatenated for each biological replicate and aligned to the A. thaliana reference genome sequence (TAIR10_v90) using RNA_STAR. The Bed file generated by RNA_STAR was visualized in SeqMonk to determine the density of raw reads aligning to various locations in the genome. To obtain the dataset for the Differentially Expressed Genes (DEG), the bed file was processed using the featureCounts program followed by the limma-voom software. For Gene Ontology analysis, the list was fed into G:Profiler (Reimand et al. 2007). For qRT-PCR, the Zymo Research kit (R2051) was used for RNA extraction. Strand-specific qRT-PCR was performed using cDNA synthesized from 1 μg total RNA using Super Scriptase IV (Thermo Fisher:18090050) and strand-specific primers (primer sequences in Supplemental Table I) followed by qPCR using PowerUp SyBr Green (Thermo Fisher: A25741). For non-stranded cDNA synthesis, a cDNA synthesis kit (Quanta:95047) was used with the same protocol for qPCR.

Zeocin treatment, PI staining and pollen viability assays

4- or 5-day-old seedlings grown on 0.5X MS media with 1% sucrose and 1% agar were transferred to six well plates containing MS media (Mock) or MS media plus 20 μM of Zeocin (Thermo Fisher Scientific - R25001). Plates were wrapped in aluminum foil and left on a shaker (100 RPM) for 2, 4 and 6 h. After treatment, seedlings were transferred to six well plates filled with PI stain solution (10 mg/ml; Sigma P4170) dissolved in H₂0. After 30 sec, seedlings were washed in ddH₂O, transferred to slides in a droplet of H₂0, sealed with a cover slip and imaged at 10X using a dsRED filter and brightfield of a Zeiss fluorescence microscope. Pollen viability was assessed as described (Li 2011). For accuracy and highest yield, the assay was performed with flowers collected between 6 AM and 8 AM. Slides containing pollen grains were imaged using a GFP filter (blue light, wavelength = 495 nm) on a Zeiss fluorescence microscope.

Comet assay

The comet assay was performed with protoplasts using a comet assay kit from Trevigen (4250–050-K) following the manufacturer’s directions with minor modifications. Protoplasts were extracted (He et al. 2007) from 6- or 7-day-old WT, atr (At5g40820) and tad3–2 seedlings. A concentration of 2 × 10⁵ cells/ml was used for the assay. Slides were run in an electrophoretic set up at 18 V for 10 mins in complete darkness. After drying the agarose, slides were stained with PI stain (100 μg/ml), sealed with a cover slip and imaged using a Zeiss fluorescence microscope at 5X magnification with a dsRED filter. The parameters (Percentage DNA in tail, Tail Length, Tail Moment) were calculated using Open Comet Software (Gyori et al. 2014).

Telomere and telomerase analysis

TRF assays were performed with 3- to 4-week-old plants as described (Kobayashi et al. 2019). To obtain high quality DNA, phenol chloroform extraction was performed twice while extracting the DNA from plant tissues. Telomere length was quantified using TeloTool (Göhring et al. 2014). Telomere fusion PCR was performed using 2 μg of DNA as described [10]. Fusions were monitored between the right arm of chromosome 1 (1R) and left arm of chromosome 2 (2L), and between 1R and the left arm of chromosome 3 (3L) using primer indicated in Supplemental Table I. G-overhangs were assessed using in-gel hybridization as described previously (Riha et al. 2000) with slight modifications. Plants no older than 3 weeks were used for the assay to obtain high-quality DNA. To assess blunt end telomeres, the hairpin-ligation assay and the UDG PENT assays were performed using 150 μg of high quality DNA quantified using a Qubit Analyzer as described (Kazda et al. 2012). Quantitative TRAP was conducted as described (Song et al. 2019) with two minor modifications. Buffer W+ (1M Tris-Acetate pH 7.5, 1M MgCl₂, 2M KGlu, 0.5M EGTA, 30% PVP, glycerol, 1μM DTT, 0.6 nM VRC, 1μM PMSF) was used to resuspend ground tissues (flowers or seedlings). The protein pellet was resuspended in buffer W+ supplemented with RNaseOUT (Thermo - 10777019). Debris were removed before measuring the protein concentration using Bradford reagent. Primer extension was performed with primer sequences in Supplemental Table I for 45 min at 25°C followed by qPCR using Dynamo SyBr mix (Thermo: F410L).

Anaphase bridges

Mitotic spreads from flower pistils were prepared and analyzed as described (Heslop-Harrison 1998; Surovtseva et al. 2009a). The spreads were stained with commercial DAPI solution (IHC-Tek 1W-1404), and imaged at 100X using a DAPI filter in Nikon Ti fluorescence microscope.

Cell culture synchronization and flow cytometry

T87 cell culture was maintained in NT-1 media on a rotary shaker (120 RPM) under continuous light for 24 h and every 7 days cells were subcultured into fresh NT-1 media (1:2 v/v). For cell synchronization, 5 mL of early stationary phase T87 cell suspension (7 days after previous subculture) was subcultured into 75 ml fresh NT-1 medium in a 250 mL Erlenmeyer flask. The flask was incubated at 24°C, 120 rpm under constant light for 7 days. 12 mL of the cell suspension was transferred into 60 ml fresh NT-1 medium to achieve a dilution of 1:5. 10 mL cell suspension was cleaned by filtration through sterilized miracloth. Excess liquid was removed with a paper towel, and an aliquot of unsynchronized cells was frozen in liquid nitrogen. To block cells in G1/early S-phase 173 μl aphidicolin stock solution of 5mg/ml (Sigma Aldrich, Catalog no. A0781) was added to 72 mL of diluted cell suspension to obtain a final concentration of 12 μg/mL. The culture was incubated at 24°C, 120 rpm under constant light for 23 h. To release the block, cells were filtered through miracloth, washed vigorously with 500 ml NT-1 medium and resuspended in 60 ml NT-1 medium. Aliquots were taken at various times for DNA content analysis. The first aliquot was labeled “T0”. The remaining cell culture continued to incubate at 24°C, 120 RPM under constant light and samples were collected each hour. For FACS analysis, frozen cells were transferred to a clean petri dish and 1 ml of cold homogenization buffer (25 mM PIPES (pH 7), 10 mM NaCl, 5 mM EDTA (pH 8), 250 mM Sucrose, 0.15 mM Spermine, 0.5 mM Spermidine, 20 mM -mercaptoethanol, 1% NP-40, 1 mM PMSF) was added. Cells were chopped with a razor blade to release nuclei followed by addition of 1 ml homogenization buffer. Cells were resuspended using a p1000 pipet and transferred into a new tube for 2 min. A 40 μm cell strainer (Merck or BD Falcon) was placed into a 50 ml falcon tube and the tube placed on ice. Resuspended cells were strained and collected into the cold falcon tube. Nuclei were collected by centrifugation at 7000 RPM for 20 min at 4°C then resuspended in homogenization buffer. Samples were treated with RNaseA at a final concentration of 15 μg/mL followed by incubation at RT for 10 min. Nuclei were stained with 60 μg/ml of propidium iodine (PI) and samples were run on a Becton-Dickinson FACSCalibur at 488 nm at the Flow Cytometry Core Facility, VMBS, Texas A&M University. DNA content was analyzed using CellQuest (Becton-Dickinson) and ModFit LT (Verity) programs.

Results

Reexamination of AtTER2 locus

The initial characterization of AtTER2 was based on annotation of the Arabidopsis genome published by The Arabidopsis Information Resource, TAIR10 (Release date, November 2010) (Berardini et al. 2015). AtTER2 is located in the Crick strand on Chromosome 5, partially overlapping the 5’ UTR of TAD3, encoded in the Watson strand (Fig 1A) (Cifuentes-Rojas et al. 2012). The non-overlapping region of AtTER2 was used to design AtTER2-specific primers to trace the molecule by RT-PCR. Given that the current genome annotation for A. thaliana, Araport11 (Release date, June 2016), extended the 5’ UTR of TAD3 to fully embed AtTER2 (Fig 1A), we designed a strand-specific RT-PCR approach to exclusively detect the AtTER2 transcript. We were unable to detect AtTER2 in flowers, leaves, and seedlings from wild type plants grown under normal conditions (Fig S1A). Cq values > 31 were obtained for AtTER2 amplification compared to Cq ≅ 19 for the internal control ACT2 (AT3G18780) (Fig S1A).

Fig. 1. Reannotation of the TER2 locus based on TAIR10_v90.

(a) Schematic representation of the TER2 and TAD3 loci in Arabidopsis thaliana based on the Araport11 version of genome annotation. The TAD3 gene (AT5G24670) is represented in blue and TER2 in red. The previous genome annotation (TAIR10 + Araport11 5’ Ext) placed TER2 within the 5’ UTR of TAD3. The putative promoter would span the TAD3 gene. The positions of the tad3–1, tad3–2 and tad3–3 T-DNA insertions are indicated by the black triangles. The short horizontal blue (TAD3) and red (TER2) lines below top panel denote stranded RNA-Seq reads from six-day-old wild type (WT) Col-0 seedlings. A cryptic antisense transcript emanating from the tad3–1 insertion is indicated by the dotted green line. (b) qRT-PCR data for TAD3 mRNA in flowers, seedlings, leaves and cell culture. The mean of two biological replicates are shown as fold change with respect to WT flowers. The WT plants are homozygous for the TAD3 allele. Error bars indicate standard deviation. (c) qRT-PCR data for TAD3 mRNA in flowers from WT, tad3–1 and tad3–2 plants. The mean of two biological replicates are shown as fold change with respect to WT samples. Error bars indicate standard deviation.

Previous studies indicated that AtTER2 was stabilized and accumulated in response to DNA damage (Cifuentes-Rojas et al. 2012). To thoroughly explore AtTER2 expression, we performed total RNA sequencing on tad3–2 mutants (see below) and wild type seedlings with and without Zeocin treatment. Stranded RNAseq libraries were prepared from total RNA after depletion of ribosomal RNAs. Sequencing of untreated tad3–2 and wild type seedlings produced a total of 51,194.244 (91.17%) and 59,996.775 (91.47%) reads, respectively, uniquely mapped to the reference genome. While sequencing of Zeocin treated seedlings produced a total of 46,127.084 (91.59%) and 53,093.124 (89.75%) uniquely mapped to the reference genome in the tad3–2 and wild type, respectively. TAD3 expression in tad3–2 mutants was ~33% of wild type (see below). However, we found no change in TAD3 expression in wild type plants upon Zeocin treatment. Moreover, no reads aligned to AtTER2 in either the tad3–2 or wild type datasets from mock (Fig. 1A) or Zeocin treated seedlings. Together, these data indicate that AtTER2 is not a stable lncRNA, and the previously detected PCR products likely reflect artifactual amplification of the Crick strand of the TAD3 5’ UTR. Therefore, any functions previously ascribed to this locus derive from TAD3.

Identification of TAD3 mutant alleles

TAD3 is widely expressed, with peaks during bolting, formation of mature flowers and silique development (Fig S1B). In silico metanalysis of publicly available transcriptomic data using Genevestigator (Hruz et al. 2008) indicated TAD3 is most highly expressed in leaves, flowers and root apical meristem (Fig. S1C). We verified this finding experimentally using qRT-PCR, and also found high TAD3 expression in cell culture (Fig 1B).

Our previous analyses of the TAD3 locus utilized the T-DNA insertion line SAIL_556_A04 (ter2–1) (Cifuentes-Rojas et al. 2012), now designated tad3–1, which resides in the 5’ UTR of TAD3 (Fig 1A). Although no transcript spanning the T-DNA insertion in tad3–1 could be detected (Fig 1A), qRT-PCR with strand-specific primers targeting a region 770 nt downstream of the T-DNA revealed the presence of an RNA transcript (Fig S2A and 2B), suggesting the activation of a cryptic promoter within the T-DNA (Mengiste and Paszkowski 1999). As this transcript could have indirect effects, we considered the tad3–1 allele suboptimal for further studies, and characterized two additional T-DNA lines. One allele termed tad3–3 carries a T-DNA in the intron between exons 8 and 9, but embryonic lethality was previously reported in homozygous mutants (Zhou et al. 2014). The third T-DNA line (SALK_121147) termed tad3–2 contains a T-DNA 902 nt downstream from the start of the TAD3 5’ UTR (Fig 1A). In contrast to tad3–1, tad3–2 does not produce an antisense transcript (Fig S2B). qRT-PCR analysis of floral RNA indicated that TAD3 mRNA is reduced by ~75% (p-value=0.06) in tad3–1 and by 83% (p-value=0.01) in tad3–2 mutants respectively (Fig. 1C). Because of the higher knockdown and the absence of a potentially confounding antisense transcript, downstream analyses were performed using the tad3–2 allele.

Plants deficient in TAD3 exhibit hypersensitivity to DNA damage and elevated programmed cell death.

It was previously reported that tad3–1 mutants exhibit an increased incidence of programmed cell death (PCD) in the Root Apical Meristem (RAM) after Zeocin treatment (Cifuentes-Rojas et al. 2012). We re-examined this response in tad3–2 mutants by imaging the RAM of seedlings stained with Propidium Iodide (PI) four- and 6 h post-treatment with 20 μM Zeocin. At 4 h, 70% of the tad3–2 seedlings displayed PCD, compared to 0% of wild type seedlings (Fig 2A and 2B). Thus, tad3–2 mutants are hypersensitive to DNA damage, consistent with the previous results obtained in tad3–1 mutants (Cifuentes-Rojas et al. 2012). The prior study indicated that tad3–1 mutants have an intrinsically elevated accumulation of DDR-related transcripts, including BRCA1, PARP1 and PARP2 (Cifuentes-Rojas et al. 2012). However, transcriptomic analysis of tad3–2 and wild type seedlings grown under normal conditions revealed only a slight increase in BRCA1 expression (1.8-fold (FDR<0.05)) and PARP2 expression (1.96-fold (FDR<0.05)) in tad3–2 mutants. Data from the RNA-Seq experiment was confirmed through RT-qPCR analysis. While BRCA1 showed a consistent increase in gene expression in tad3–2 mutants, expression levels of PARP1 and PARP2 remained unchanged upon loss of TAD3 (Fig S3A).

Fig. 2. The TAD3 locus does not modulate DNA damage related pathways.

(a) Schematic representation of an A. thaliana root tip with the Root Apical Meristem (RAM) highlighted in gray (left). On the right, images of roots from A. thaliana seedlings stained with Propidium Iodide solution (PI) following treatment with 20 μM Zeocin. Photos are shown of representative 4-days old WT and tad3–2 seedlings treated with Zeocin for 4 h and 6 h followed by PI staining. Yellow box highlights the RAM and Programmed Cell Death. (b) Percentage number of seedlings with PI staining in the RAM of WT and tad3–2 seedlings treated with Zeocin for 4 or 6 h Numerical values indicate total number of roots imaged for each condition. 0 out of the 36 WT seedlings showed RAM PCD at 4 h post-Zeocin treatment. (c) Representative images of data obtained from comet assays performed on protoplasts extracted from seedlings. The length and intensity of the comet tail indicates the level of DNA damage. (d) Values for percentage DNA in tail (%PDT) from the comet assay plotted using a box and whisker plot. Top and bottom edges of the box represent the first and the third quartiles, respectively. The length of the whisker spans the minimum to maximum values. The straight line inside the box represents the median and ‘X’ stands for sample mean. Normally distributed data have an overlapping mean and median. More than 1000 comets were scored for each genotype. **p-value <0.001 and NS = not significant based on students t-test. (e) Data obtained for quantitative Telomere Repeat Amplification Processivity (qTRAP) assays performed with flower bundles from WT and tad3–2 mutants. The mean of three biological replicates are shown as fold change with respect to WT samples. Error bars indicate standard deviation. NS = not significant based on u-test. (f) qTRAP results for 7-day old seedlings untreated (mock) or treated with 20 μM Zeocin for 6 h. The mean of three biological replicates is shown as fold change with respect to WT samples at 0 h mock treated. Error bars indicate standard deviation. ** = p-value <0.01 based on u-test.

To test if increased PCD in tad3–2 seedlings correlates with accumulation of endogenous DNA damage, we performed a modified version of the single cell comet assay using protoplasts extracted from 7 days old seedlings. We measured Percentage DNA in the comet Tail (PDT) and Tail Length (TL) to calculate Tail Moment (TM) (Olive and Banáth 2006). Statistical analysis of any of these three parameters gauges the level of DNA damage (Beedanagari et al. 2014) and can be confirmed by the other two parameters. For convenience, we represented DNA damage as a function of PDT (%PDT). As a positive control, assays were performed on cells from plants lacking ATR, a master regulator of the DNA damage response machinery (Wang et al. 2016). As expected, PDT was significantly higher in atr mutants compared to wild type (Fig 2C and 2D). However, the level of PDT observed in tad3–2 mutant was similar to wild type. We conclude that loss of TAD3 does not lead to accumulation of damaged DNA, and existing DNA damage is not an underlying cause of the PCD in tad3–2 mutants. To test whether DNA damage sensing and repair capabilities were affected in tad3–2 mutants, we performed gene ontology analysis on the Zeocin-induced differentially expressed genes (DEGs) in wild type and tad3–2 seedling. We did not find any find conspicuous difference in the GO term enrichment between the genotypes (Fig S3C). We further analyzed our RNA-Seq data to compare the transcriptional response to Zeocin at the gene level, 77 of the top 100 Zeocin-induced DEGs in wild type and tad3–2 are commonly deregulated; notably, these genes are deregulated in the same magnitude (Supplementary Table 2). This provides ample and detailed evidence of the unaltered transcriptional response to Zeocin-treated tad3–2 mutants, as compared to wild type.

Finally, since AtTER2 was reported to negatively regulate telomerase activity in response to DNA double-strand breaks (Xu et al. 2015), we re-assessed this conclusion using the tad3–2 allele. We found no difference in telomerase activity levels of tad3–2 flowers or seedlings relative to wild type (Fig 2E and 2F). Two hours of Zeocin treatment induced a robust DNA damage response as evidenced by ~100-fold increase in BRCA1 expression (Fig S3B). Telomerase activity was decreased by ~50% in both wild type and tad3–2 mutants, after 6 h of Zeocin treatment (Fig 2F), arguing that the telomerase response to Zeocin is not dependent on TAD3. Altogether, these findings indicate that tad3 mutants are hypersensitive to DNA damage, but TAD3 does not regulate the response to DNA damage.

A telomere maintenance defect in tad3–2 mutants is independent of telomerase

As part of our characterization of the TAD3 locus, we monitored bulk telomere length over three consecutive generations in tad3–2 mutants. Terminal Restriction Fragment (TRF) analyses revealed a subtle but progressive loss of high molecular weight telomere tracts in the tad3–2 mutants relative to wild type siblings (Fig 3A). Genetic complementation was used to test if the telomere maintenance defect was due to the loss of TAD3. tad3–2 mutants were transformed with a full-length TAD3 gene under the control of its native promoter (P_TAD3::TAD3) (Fig 3B). Within a single generation, three of the nine independent transformants showed complete recovery of telomere length to wild type, and in five others some telomere tracts were longer than in tad3–2 mutants (Fig 3B), supporting a role for TAD3 in telomere length maintenance.

Fig 3. TAD3 maintains telomeres via a telomerase-independent pathway.

(a) Results of Terminal Restriction Fragment (TRF) analysis to measure bulk telomere length in tad3–2 mutants from second (G2), third (G3) and fourth (G4) generations of homozygosity along with the segregating WT siblings followed in parallel through successive generations. Red dotted line indicates the maximum telomere length for the WT samples from this cross. (b) Results of TRF analysis performed for genetic complementation of tad3–2 mutants. 4-week-old tad3–2 mutants were transformed with pCBK05::NP^TAD3::TAD3. Results are shown for WT (lane 1), DNA from the same generation of untransformed tad3–2 mutants as the control (lane 2), and complementation lines (lanes 3–11). Red dashed line indicates maximum telomere lengths for WT and tad3–2 samples. Lanes 1 and 2 contains DNA derived from a pool of ~100 seedlings and lane 3–11 contains DNA from individual transformants. (c) Results from TRF analysis of WT, G2 tad3–2, G2 pot1a and G2 pot1a tad3–2 mutants. DNA samples are derived from individual plants of each genotype. Red line indicates the critical telomere length threshold of 1 Kb. (d) Quantification of the TRF gel from panel C determined by TeloTool (Göhring et al. 2014). Data are represented as box and whisker plot. Red dot within the box represents the mean value. (e) Results from qTRAP assays performed with flowers from WT, G2 tad3–2, G2 pot1a and G2 pot1a tad3–2 samples. The mean of three biological replicates are shown as fold change with respect to WT samples (f) Results of TRF analysis for WT, tad3–2, tad3–2 ku70 and ku70 mutants. DNA was analyzed from individual segregating siblings belonging to G1 of the tad3–2 X ku70 cross. The gel has been sliced to highlight the lane for ku70. A vertical line separates the ku70 lane from the rest of the gel.

When telomerase activity is limiting, shorter telomeres are preferentially elongated (Marcand et al. 1997; Armanios et al. 2005; Goldman et al. 2005; Harrington 2012). To investigate if depletion of long telomeres in tad3–2 mutants reflects a defect in telomerase, we generated double mutant plants. Our initial goal was to obtain plants lacking TAD3 and TERT. Both genes are situated on chromosome 5, approximately 2.9 Mb apart with TAD3 proximal to the centromere (Berardini et al. 2015). Linkage calculations indicated that Mendelian segregation of the two loci was possible, and predicted 6.25% of the offspring of TAD3–2^+/− TERT^+/− would be tad3–2^−/− tert^−/−. Nevertheless, we failed to recover any homozygous double mutants among ~200 offspring analyzed, suggesting that TERT and TAD3 may cooperate for some essential non-telomeric function. As an alternative strategy, we made crosses to generate plants doubly deficient in TAD3 and POT1a. First-generation (G1) pot1a tad3–2 plants were readily obtained and were self-pollinated to produce second-generation (G2) pot1a tad3–2 mutants. In parallel, we propagated wild type, pot1a and tad3–2 single mutants. Each line was grown for several consecutive generations (G2–G4).

We assessed how the combined loss of POT1a and TAD3 impacted telomere length using TRF (Fig 3C). As expected, telomeres in G2 pot1a mutants were shorter than wild type and displayed a discrete banding pattern indicative of a telomerase deficiency (Fig 3C). Strikingly, telomeres in G2 pot1a tad3–2 were even shorter than the pot1a single mutants (Fig 3C). A banding pattern was visible for longer telomeres, but telomere tracts shorter than 1kb were more heterogeneous (Fig 3C). Quantification of telomere length using TeloTool (Göhring et al. 2014) showed wild type spanned 2.0–5.0 kb with a mean telomere length (MTL) of 3kb (Fig 3D). tad3–2 telomeres were similar though slightly shorter (range=1.2–4.0 kb; MTL= 2.1 kb). In contrast, telomeres in G2 pot1a tad3–2 plants were significantly shorter (range=0.5–2.1 kb; MTL=1 kb) than telomeres in G2 pot1a mutants (range=0.8–2.8 kb; MTL=1.7 kb) (Fig 3D). We conclude that combined loss of TAD3 and POT1a accelerates telomere shortening relative to the loss of POT1a alone.

Progressive telomere shortening ultimately causes profound developmental defects as a consequence of genome instability (Riha et al. 2001). Consistent with the hypothesis that TAD3 acts synergistically with telomerase, there was accelerated shortening of telomeres in pot1a tad3–2 mutants (Fig 3C and S4A), which correlated with an early onset of stem cell-related defects (Fig 4A and S4B). tad3–2 mutants displayed no visible developmental defects for three generations (Fig 4A and S4B). Conversely, gross morphological abnormalities were evident in pot1a tad3–2 mutants beginning in G2 and worsened over the generations (Fig 4A and S4B). Importantly, pot1a single mutants were indistinguishable from wild type in G2 (Fig 4A and S4B). We categorized pot1a tad3–2 plants into three groups: class I mutants were similar to wild type; class II plants had stunted growth with leaf abnormalities, constricted rosettes, and occasional hook-shaped siliques; and class III mutants were more severely impacted than class II (Fig 4A). The number of class II and class III mutants increased with each generation (Fig 4B). Pollen viability of G2 pot1a tad3–2 was decreased relative to wild type or either single mutants (Fig 4C), and later generation pot1a tad3 plants were sterile, failing to produce any siliques (Fig 4A).

Fig 4. Exacerbated reproductive and developmental defects and genome instability in pot1a tad3–2 mutants.

(a) Photos of rosettes, individual leaves and siliques from three-week-old WT, G2 tad3–2, G2 pot1a and G2 pot1a tad3–2 plants. Siliques and leaves were collected from the same position for all samples. For G2 pot1a tad3–2 mutants representative images from three phenotypic classes (I, II, and III) are shown. (b) Pie chart illustrating the relative fraction of plants belonging to each phenotypic class of G2 pot1a tad3–2 double mutants. (c) Viability of pollen grains produced by WT, G2 tad3–2,G2 pot1a and G2 pot1a tad3–2 assessed with the FDA staining protocol (Li 2011) in combination with PI staining. Live pollen metabolizes the FDA into green colored fluorescein. PI stains dead pollen. (d) Mitotic spreads of anaphase were made from flower pistils of four-week-old WT, G2 tad3–2, G2 pot1a and G2 pot1a tad3–2 plants using previously published protocol (Surovtseva et al. 2009). Chromatin was stained with DAPI and observed with 100X magnification on a fluorescent microscope. (e) Quantification of anaphase bridges obtained from analyzing mitotic fields in pistils from genotypes as indicated.

The worsening of developmental phenotypes correlated with an increased incidence of telomere tracts below the critical 1kb length threshold (Heacock et al. 2004) (Fig S4A). Analysis of mitotically dividing cells revealed 12% of the anaphases in G2 pot1a tad3–2 harbored bridged chromosomes, consistent with telomere-to-telomere fusion, compared to 1.9% in pot1a and 0% in tad3–2 and wild type siblings (Fig 4D and E). The percentage of anaphase bridges increased to 21% in G3 pot1a tad3–2 (Fig 4E). Telomere fusion PCR experiments confirmed that the chromatin bridges reflected end-to-end chromosome joining through telomeres (Fig S4C and D).

The data presented thus far suggest that TAD3 and telomerase act in parallel pathways to maintain telomere length. However, an alternative possibility is that TAD3 acts in a pathway overlapping with telomerase. Although repeat addition processivity of telomerase is severely compromised in pot1a mutants, enzyme activity is not entirely abrogated (Surovtseva et al. 2007). Thus, with both TAD3 and POT1a simultaneously inactivated, telomerase activity could be entirely abolished. To test this, qTRAP was performed with pot1a tad3–2 mutants. There was no difference in telomerase activity in pot1a tad3–2 mutants compared to pot1a (Fig 3E), indicating that TAD3 is not required for maximal telomerase stimulation.

Finally, we asked if TAD3 was required for telomerase recruitment and enzymology at chromosome ends in vivo by assessing how the loss of TAD3 impacted telomere elongation in plants lacking Ku70. If telomere elongation in ku70 mutants requires TAD3, then plants doubly deficient in both Ku70 and TAD3 should not have ultra-long telomeres. To test this hypothesis, we crossed ku70 and tad3–2 single mutants and segregated double mutants from Ku70+/− TAD3–2+/− parents. TRF analysis performed with the G1 siblings, revealed no difference in telomere length in G1 ku70 tad3–2 plants compared to G1 ku70 (Fig 3F). Thus, TAD3 does not appear to play a critical role in promoting telomerase engagement and extension at chromosome ends. Altogether, our results support the conclusion that TAD3 acts independently of telomerase for telomere length maintenance.

Telomere terminal architecture is unperturbed in tad3–2 mutants

Another explanation for the telomere shortening phenotype is that telomere architecture is compromised in tad3–2 mutants, leaving chromosome ends vulnerable to inappropriate nucleolytic processing. Telomere integrity cannot be grossly altered since tad3–2 mutants do not suffer end-to-end fusions, but to test for subtle perturbation, we measured the status of the G-overhang using in-gel hybridization (Riha et al. 2000). The G-overhang signal was increased in ku70 mutants by 2.5-fold (Fig 5A), consistent with the conversion of blunt-end telomeres into G-overhangs (Kazda et al. 2012). In contrast, we found no difference in the G-overhang signal in G2 pot1a tad3–2, G2 tad3–2, and G2 pot1a mutants compared to wild type (Fig 5A). Next, we examined the integrity of blunt end telomeres using the dUTP-PENT assay (Kazda et al. 2012). As expected, approximately 55% of the signal was retained in wild type samples after UDG treatment, confirming half the telomeres are blunt ended, while in ku70 mutants, the signal was reduced by ~89%, consistent with conversion of most blunt ends into G overhangs (Fig 5B). tad3–2 mutants exhibited a wild type level signal (~50%) after UDG treatment. We verified blunt end telomeres in tad3–2 mutants using a hairpin ligation assay (Kazda et al. 2012; Valuchova et al. 2017). Blunt-ended telomeres migrate as a higher molecular weight smear and then are lost upon BamHI digestion. A high molecular weight smear sensitive to BamHI was observed in both wild type and tad3–2 samples, but not in ku70 (Fig 5C). We conclude that TAD3 does not play an essential role in maintaining the proper architecture of chromosome termini.

Fig 5. Loss of TAD3 does not affect the G-overhang or blunt-end architecture of telomeres.

(a) Quantification of the G-overhang assay (G-OH) performed using the in-gel hybridization technique with a radioactive probe complementary to the telomeric G-rich strand. The mean of two biological replicates are shown as fold change with respect to WT samples. Error bars indicate standard deviation. DNA from WT and ku70 mutants serve as the negative and positive controls, respectively. (b) Results of UDG-PENT assays performed to assess telomere end architecture. The % signal was calculated using QuantityOne Software. DNA from a ku70 mutant served as the positive control. (c) Results for a hairpin ligation assay to confirm the presence of blunt ended telomeres. For each genotype, the first lane (lanes 1, 4, 7) shows untreated telomeric DNA; the second lane (lanes 2, 5, 8) shows telomeric DNA ligated to a hairpin; the third lane (lanes 3, 6, 9) shows DNA cleaved with BamHI enzyme. The downward arrows highlight evidence of hairpin ligation. The higher molecular weight products in the ku70 samples are expected since telomeres are elongated in the absence of Ku (Riha et al. 2002). The smeared area above the main hybridization signal (right side bracket) was quantified using QuantityOne and results are represented using the bar graph on the right side of the gel. Data correspond to quantitation for the reactions in each lane as described.

Loss of TAD3 impacts many cellular pathways

Given the essential role of tRNA deaminases in translation (Torres et al. 2014b), TAD3 is expected to impinge on many cellular pathways. To gain insight into the global impact of TAD3 mutation, we further analyzed RNA-seq data from tad3–2 and wild type seedlings to identify differentially expressed genes (DEGs). We used Limma-Voom on the web-based program Galaxy (Afgan et al. 2018), with FDR<0.05. DEG with more than two-fold change in tad3–2 compared with wild type was fed into G: Profiler (Reimand et al. 2007) to determine the functional enrichment of gene ontology (GO) terms. A total of 980 RNAs were differentially accumulated in tad3–2 mutants; 598 were upregulated and 382 were downregulated. Notably, no telomere-related gene was identified as a DEG.

GO terms are categorized by Molecular Function (MF), Biological Pathway (BP) and Cellular Compartment (CC). We observed significant enrichment of GO terms in the BP category, with a large number of downregulated genes associated with auxin signaling, auxin transport, and cellular response to auxin. Other downregulated genes were associated with the cellular response to chemicals and growth, both of which are also related to auxin-related processes (Fig. 6A). In contrast, upregulated genes showed significant enrichment of GO terms related to secondary metabolic processes, secondary metabolite synthesis, and particularly with the glucosinolate biosynthetic pathway (Fig 6A).

Fig 6. Transcriptomic analysis reveals changes in auxin signaling, plant secondary metabolism and cell cycle-related genes due to loss of TAD3.

(a) Gene ontology analysis performed with differentially regulated genes in 6-day-old tad3–2 seedlings compared to 6-day-old WT seedlings. Table contains GO term source, term name and with the numerical p-value expressed as a function of intensity of the green color. p-values greater than 10^–16 are highly significant (Reimand et al. 2007). Red and the green arrows represent genes downregulated or upregulated in the tad3–2 mutants, respectively. (b) Expression data for some critical cell cycle and DNA replication related genes in tad3–2 mutants derived from transcriptome data.

Since the transcriptional responses to Zeocin in tad3–2 mutants are almost indistinguishable from those of wild type plants, we re-examined our RNA-seq dataset in an effort to find more direct targets of TAD3 by looking at DEGs between Zeocin-treated tad3–2 and wild type seedlings. This stringent analysis resulted in 166 differentially accumulated RNAs in tad3–2 mutants, of which 105 were upregulated and 61 were downregulated (Fig. S5D). GO analysis of the new gene pool was consistent with the previous analysis: the downregulation of auxin homeostasis and signal transduction pathway and upregulation of glucosinolate biosynthesis as the most affected processes in the hypomorphic tad3–2 mutant (Fig S5A).

Finally, given the importance of TAD3 in cell cycle progression in fission yeast (Tsutsumi et al. 2007), we specifically looked for changes in expression of 150 critical cell cycle regulators and DNA replication factors in tad3–2 mutants. We found that the MCM gene cluster (MCM2, MCM3, MCM4, MCM5, MCM7) exhibited a 1.5- to 1.8-fold increase in tad3–2, while CDC6 and CDC6B expression increased by almost 2.1-fold (Fig 6B). Both CDC6 and MCM gene clusters initiate S-phase by licensing origins for DNA replication (Borlado and Méndez 2008; Das et al. 2014). Loss of TAD3 also led to elevated expression for some cell cycle regulators including CDKB11 (1.6 fold), HAC1 (1.66), CDC45 (1.69) and CDT1 (1.62) (Fig 6B). Thus, TAD3 modulates expression of numerous cell cycle related genes.

We also investigated the expression pattern of TAD3 gene expression across the cell cycle in Arabidopsis using synchronized T87 A. thaliana cell culture (Menges and Murray 2002). Cells were treated with Aphidicolin to arrest them in G1/early S-phase. FACS analysis was done at various points after releasing the block to monitor cell cycle progression (Fig S5B) and transcript levels were measured using qRT-PCR. Although, TERT and TAD3 mRNA levels peaked during early S phase, based on statistical analysis, changes in expression levels for these genes across the cell cycle were not significant (Fig S5C). In contrast, POT1a mRNA level peaked during early S/G2 phase transition and statistical analysis revealed that POT1a expression changed significantly across the cell cycle (Fig S5D). Overall, based on the FACS analysis, TAD3 seems to be constitutively expressed across the cell cycle, possibly owing to its fundamental role in tRNA editing.

Discussion

Telomere length maintenance is essential for the stability of linear genomes. Over the past two decades, multiple genetic screens, interactome assays, and QTL mapping experiments illustrate the influence of “noncanonical” pathways in telomere length regulation. Remarkably, genome-wide studies in S. cerevisiae revealed that >5% of nonessential (Askree et al. 2004) and >11% of essential (Ungar et al. 2009) genes are necessary for telomere maintenance. Recently, translation-related factors have emerged as critical determinants of telomere length homeostasis (Lin and Zakian 1996; Heiss et al. 1998; Maas et al. 1999; Askree et al. 2004; Gatbonton et al. 2006; Fu and Collins 2007; Walne et al. 2007; Ungar et al. 2009; Gupta et al. 2013; Abdulkina et al. 2019). One of essential gene affecting telomere length in budding yeast is YLR317W, a transcript produced from the TAD3 locus (Ungar et al. 2009). Here we demonstrate the importance of TAD3 in telomere length maintenance in A. thaliana. We further show that this function is mediated by a noncanonical, telomerase-independent mechanism, highlighting the importance of cross-functional pathways in telomere biology.

Previously we described a telomerase regulatory function for the long non-coding RNA AtTER2 encoded on the opposite strand and partially overlapping with the 5’ UTR of TAD3 (Cifuentes-Rojas et al. 2012). In considering updated A. thaliana genome annotation (Araport 11) showing that AtTER2 is fully embedded into the 5’ UTR of TAD3 and the demonstration that TER1 was not the true telomerase RNA subunit (Fajkus et al. 2019; Dew-Budd et al. 2020) led us to revisit the TER2 locus using strand-specific qRT-PCR and transcriptomic analyses. We report that TAD3 does not give rise to a stable lncRNA, and hence telomere-related functions derive from the TAD3 gene itself.

Because a null mutation in TAD3 leads to embryonic lethality (Agorio et al. 2017), we obtained a new hypomorphic tad3 mutant (tad3–2) to further explore its function in telomere biology. We discovered that in tad3–2 mutants, the longest telomere tracts shortened progressively over successive generations, while shorter telomeres remained unchanged. A similar profile is observed in cells haploinsufficient for key telomerase components (Armanios et al. 2005; Goldman et al. 2005; Harrington 2012). However, ex vivo qTRAP assays indicated wild type levels of telomerase activity in tad3–2 mutants. In addition, analysis of ku70 tad3–2 mutants revealed that telomerase can fully access and extend telomeres in plants deficient in TAD3. Strikingly, defective telomere maintenance in tad3–2 mutants is strongly exacerbated in plants also lacking the telomerase processivity factor POT1a, with double mutants exhibiting an early onset of developmental defects and genome instability arising from telomere dysfunction. Thus, TAD3 facilitates telomere length homeostasis via a telomerase-independent pathway.

How could TAD3 promote telomere maintenance? TAD3 encodes a tRNA-editing deaminase that converts adenosine to inosine at the wobble 34 position of the tRNA anticodon loop (Torres et al. 2014b). This modification expands pairing to A, U, C at the 3rd position of a codon (Crick 1966; Grosjean et al. 2010). I34 is critical for reading and translating C-ended codons (Lim 1995) for Ala, Ser, Pro, and Thr (Rafels-Ybern et al. 2015, 2018). Consequently, compromising TAD3 is expected to impact many cellular pathways (Schimmel 2018). Analysis of human transcriptome and proteome data confirm the importance of adenosine deaminases (ADATs) in translating transcripts rich in these same four codons (Rafels-Ybern et al. 2015). Because such translation-related data are unavailable for A. thaliana, we performed a transcriptome analysis on tad3–2 mutants to examine how decreased expression of AtTAD3 impacts plant metabolism.

Over 6000 genes are differentially regulated upon loss of TAD3, but intriguingly none are associated with known telomere pathways. Instead the genes are concentrated in two major areas with significant downregulation of the auxin signal transduction pathways and significant upregulation of the glucosinolate biosynthetic pathway. Notably, both metabolic processes intersect stress response, cell cycle regulation and DNA metabolism. Reduced auxin signaling may account for the elevated PCD in the RAM of tad3–2 mutants in response to Zeocin. Our RNA-seq data and comet assays showed that tad3–2 mutants mount a normal DDR and do not accumulate more DNA damage than wild type under normal conditions. Auxin inhibits PCD during plant development and in response to stress (Awwad et al. 2019). Under normal conditions, auxin concentrations in root stem cell niche peak in the quiescent center and follow a local gradient at the root tip. However, in response to environmental stress, auxin levels decline, leading to PCD induction in roots (Hong et al. 2017). Thus, lower levels of auxin in tad3–2 mutants may sensitize plants to PCD in response to stress. Alternatively, down regulation of auxin signaling may render chromatin more vulnerable to Zeocin treatment. Auxin has recently been shown to increase chromatin compaction, and its inhibition results in increased DNA damage upon Zeocin treatment (Hasegawa et al. 2018).

Our transcriptomic data analyses also revealed that tad3–2 mutants significantly upregulate genes in the glucosinolate biosynthetic pathway. Glucosinolates are secondary metabolites in cruciferous plants that serve as antimicrobials and defend against herbivory. Interestingly, glucosinolate accumulation regulates cell cycle progression in Arabidopsis and reduces the rate of DNA replication in wild type plants, causing cells to accumulate in S phase (Åsberg et al. 2015; Chezem and Clay 2016). Despite the wide array of mutant phenotypes expected for TAD3 mutation, the predominant feature of tad3 mutation in fission yeast is a cell cycle defect (Tsutsumi et al. 2007). While the changes were not as dramatic as in other metabolic pathways, we observed a surge in expression of genes that regulate cell cycle and promote DNA replication. Telomere replication and processing require a dynamic switch from a protective state to an open conformation and back again (Gobbini et al. 2014), and thus cell cycle perturbation can alter telomere length and terminal architecture (Verdun et al. 2005; Vodenicharov and Wellinger 2007; Londoño-Vallejo and Wellinger 2012; Sarek et al. 2019). Although we saw no obvious change in the status of G-overhangs or blunt end telomeres in tad3 mutants, our experiments were performed on asynchronously growing seedlings. It is possible that a subtle shift in cell cycle progression in tad3 mutants decreases telomerase access to telomeres or increases access for nucleolytic processing enzymes, either of which would lead to telomere shortening.

We conclude that the TAD3 locus indirectly contributes to telomere length homeostasis in Arabidopsis by altering the metabolic profile. Understanding precisely how cross-functional pathways influence telomere biology may shed new light on how telomeres serve as both sentinels and elicitors of physiological stress.

Supplementary Material

299_2020_2594_MOESM1_ESM

Fig S1. TAD3 mRNA expression is regulated during plant development.

(a) Results from strand-specific qPCR. Cq values for TER2 and the ACT2 gene amplified using WT flowers, leaves and seedlings are shown. (b) Genevestigator-based analysis of TAD3 mRNA expression during different stages of plant growth and development. (c) Genevestigator analysis of organ-specific expression of TAD3 mRNA.

299_2020_2594_MOESM5_ESM

Fig S5. Gene ontology analysis of WT and tad3–2 transcriptomics data and expression of TAD3 and telomerase components across the cell cycle.

(a) Gene ontology analysis performed with differentially regulated genes in Zeocin-treated 6-day-old tad3–2 seedlings compared to Zeocin treated 6-day-old WT seedlings. (b) FACS data obtained from Aphidicolin-synchronized T87 cell culture. Graph shows a time course of the fraction of cells in each phase of the cell cycle post release from the drug. (c) RT-qPCR analysis of TAD3 and TERT performed on the RNA isolated from synchronized T87 cell culture at different timepoints. ΔCt values were normalized to the reference gene AT4G26410. Each data point represents the mean value ± SD (n=4 independent assays). p>0.05 (ANOVA). (d) RT-qPCR analysis of POT1a performed on RNA isolated from synchronized T87 cell culture at different timepoints. ΔCt values were normalized to the reference gene AT4G26410. Each data point represents the mean value ± SD (n=4 independent assays). *p<0.05 (ANOVA).

299_2020_2594_MOESM4_ESM

Fig S4. Combined loss of TAD3 and POT1a accelerates the onset of telomere dysfunction.

(a) Results of TRF analysis for consecutive generations of individual pot1a tad3–2 mutants from G2 (lane 1), G3 (lanes 2–4) and G4 (lanes 5–7). (b) Images of rosettes from three-week-old WT, G4 tad3–2, G4 pot1a and G4 pot1a tad3–2 plants. Examples of plants from the different classes of G4 pot1a tad3–2 mutants are shown. (c) and (d) Results from Telomere Fusion PCR assays with WT, tad3–2, pot1a and pot1a tad3–2 samples. DNA from a ctc1 null mutant (Surovtseva et al. 2009a) served as the positive control. The subtelomeric primers used for PCR amplification are indicated below each blot.

299_2020_2594_MOESM3_ESM

Fig S3. Results of zeocin treatment of WT and tad3–2 seedlings.

(a) Results for qRT-PCR experiments performed to detect BRCA1, PARP1 and PARP2 gene expression in tad3–2 and WT seedlings under normal conditions. The mean of three biological replicates is shown as fold change with respect to WT samples. Error bars indicate standard deviation. (b) Results for qRT-PCR experiments performed to detect BRCA1 gene expression samples treated with 20 μM zeocin for 2 h. The mean of three biological replicates is shown as fold change with respect to untreated WT samples. Error bars indicate standard deviation. (c) Gene ontology analysis performed using G profiler with the genes upregulated in 6-day-old WT and tad3–2 seedlings treated with 20 μM for 2 h.

299_2020_2594_MOESM2_ESM

Fig S2. Identification of a cryptic transcript produced from the TAD3 locus in tad3–1 mutants.

(a) Schematic representation of the TAD3 locus (see legend for Figure 1). Green arrows denote forward and reverse primers used to detect expression. (b) qRT-PCR results obtained with these primers with WT, tad3–1 and tad3–2 samples. PCR product size = 108 nts. The mean of two biological replicates are shown as fold change with respect to WT samples. Error bars indicate standard deviation.

Key message.

tRNA Adenosine Deaminase 3 helps to sustain telomere tracts in a telomerase-independent fashion, likely through regulating cellular metabolism.