Genome analysis uncovers telomere sequence variation between Mimulus species.

This project was originally motivated by an attempt to characterize telomere length variation across Mimulus species. In our previous study [50], we used the computational method k-Seek [51,52] to count k-mer repeats in raw un-mapped whole genome sequencing reads, and used the telomere repeat k-mer counts as a highly accurate method for approximating the amount of telomere repeats within an individual plant. We used k-Seek to analyze published whole-genome re-sequencing data of Mimulus species from three different sections (i.e., morphological groupings): M. aurantiacus from section Diplacus [53], M. guttatus from section Simiolus [54], and M. verbenaceus from section Erythranthe [55]. More recent phylogenetic analyses indicate polyphyly of the genus Mimulus [56] and taxonomic revision has split these focal species into separate genera, Diplacus and Erythranthe [45]. However, following other recent work in the Phrymaceae, we have elected to retain the use of the name Mimulus for its historic significance and recognizability (see [46]).

We analyzed Mimulus species with population genomic data to quantify the natural variation in k-mer counts (normalized by genome coverage) across multiple genotypes for each species (n = 43 for M. aurantiacus, n = 228 for M. guttatus, and n = 54 for M. verbenaceus). Initially, we did not expect there to be any sequence variation in the telomeres between Mimulus species, hence attempted to use the k-mer abundance as an approximation for quantifying telomere length variation. Note for M. aurantiacus, the population genomic data are from six closely related subspecies that we collectively refer to as M. aurantiacus only for the k-mer analysis for convenience, and importantly our k-mer results do not depend on the underlying subspecies classification. Results showed for all three species, most large sized k-mers (k > 3) had relatively low abundance (~100 copies per 1 × genome coverage), indicating the genome-wide tandem repeat variation was largely driven by small sized 1-, 2-, or 3-mers (S1 Fig). But in M. aurantiacus (Fig 2), the k-mer AAACCCT had substantial abundance for a large sized k-mer (~13,190 copies per 1 × genome coverage) and the sequence corresponded to the classical Arabidopsis-type telomere repeat sequence (note AAACCCT is an offset of the reverse complement of TTTAGGG [i.e., CCCTAAA] followed by tandem repetition).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. K-mer based genome sequencing analysis identifies changes in the candidate telomere sequence repeat for three Mimulus species.

For each species only the abundance of the top three k-mers that resemble a telomere sequence are shown on a log10 scale. Phylogenetic relationships between the three species are shown below with their divergence time [57]. Within a species the most abundant k-mer is hypothesized as the species telomere sequence, while the other two k-mers are likely to be low abundance interstitial repeats. Note the observed variability in k-mer repeats represents both natural variation and potential measurement or sequencing errors.

https://doi.org/10.1371/journal.pgen.1011738.g002

The k-Seek approach cannot discriminate the genomic locations the k-mer repeats originated from. Since there are interstitial telomeric sequences that are found outside of chromosome ends [58], we took an independent dataset to check whether the AAACCCT k-mer is terminally located and corresponded to the telomere repeat sequence. We analyzed the M. aurantiacus chromosome level reference genome assembly that was sequenced from the subspecies puniceus [53], and focused on the sequences at the ends of assembled chromosomes to identify the telomere repeat motif. The chromosome ends were abundant for the k-mer AAACCCT (S2 Fig) confirming that M. aurantiacus ssp. puniceus harbors the canonical Arabidopsis-type telomere sequence.

The genome sequencing k-mer analysis for M. guttatus and M. verbenaceus, on the other hand, did not detect AAACCCT as the most abundant 7-mer repeat (Fig 2). In fact AAACCCT was almost absent in M. guttatus while it had a relatively low abundance in M. verbenaceus. This suggested the telomere sequence motif had changed in M. guttatus and M. verbenaceus. To identify the alternative telomere motif we searched for k-mers that were similar to the Arabidopsis-type telomere sequence and had high abundance in the raw genome sequencing data. In M. guttatus the k-mer AAACCCG 7-mer had the highest abundance (~645 copies per 1 × genome coverage), and this k-mer was largely absent in M. aurantiacus and M. verbenaceus. Meanwhile in M. verbenaceus there was no highly abundant 7-mer sequence but the 6-mer AAACCG was particularly abundant (~6,070 copies per 1 × genome coverage). Furthermore this k-mer was largely absent in M. aurantiacus and M. guttatus.

To investigate if the two k-mers AAACCCG and AAACCG corresponded to the telomere repeat we analyzed the chromosome-level reference genome assemblies for M. guttatus and M. verbenaceus. We focused on the sequences at the chromosome ends and discovered M. guttatus assemblies were enriched for the AAACCCG repeat, while M. verbenaceus assemblies were enriched for the AAACCG repeat (S2 Fig). These data suggested the abundant k-mers that resemble telomere-like repeats from the M. guttatus and M. verbenaceus raw genome sequencing data (Fig 2) likely correspond to the true telomere sequences in these species.

Annotating the telomerase RNA gene in the Mimulus species with reference genomes

Our k-mer analysis suggested the telomere sequence had changed at least twice in the Mimulus genus (i.e., AAACCCT, AAACCCG, AAACCG). To verify this observation, we studied the evolution of the Mimulus TR gene. The template sequence within the TR determines the telomere DNA sequence (Fig 1), hence by investigating the evolution of the TR gene we aimed to gain a deeper understanding of the telomere sequence evolution in the Mimulus genus.

We first analyzed the Mimulus species with reference genomes (M. aurantiacus ssp. puniceus, M. cardinalis, M. guttatus, M. lewisii, M. parishii, and M. verbenaceus) to identify the TR gene. Mimulus cardinalis, M. lewisii, and M. parishii are species that were not analyzed in our k-mer analysis and belongs to the Erythranthe section, which includes M. verbenaceus that was included in our k-mer analysis (Fig 2). For each reference genome, we used a position weight matrix [59] and secondary structure-based algorithm [60] to search and annotate the TR gene locations (see S1 Table for genome coordinates). The size of the identified TR genes ranged between 278 bp to 317 bp, which are similar in size to other TRs found across the plant kingdom (ranging from 234–390 bp) [27,28]. During the TR annotation process we also obtained an intriguing result where the reference genomes from M. aurantiacus ssp. puniceus, M. cardinalis, and M. lewisii had evidence of two seperate TR gene paralogs. This intriguing discovery and the evolution of these TR duplicates will be further discussed in a later section.

To examine the potential functionality of the annotated TRs we extracted 100 bp upstream of the predicted start site of the TR transcript to search for the presence of conserved regulatory elements. Previous studies have found the upstream genetic regions of land plant TRs contained a highly conserved type III promoter motif called the Upstream Sequence Element (USE) and a TATA box motif [28,61]. All of our candidate TRs, including the duplicates, had the canonical USE motif TCCCACAT within the upstream region (S3 Fig). This sequence was identical to the USE motif found in all surveyed Lamiales species [28]. One exception was M. cardinalis where one of its TR duplicates had a single nucleotide substitution in the conserved USE motif (TCCCACGT; the nucleotide difference is bolded) and we verified this nucleotide change with Sanger sequencing. For all candidate TR sequences we also discovered the TATA box 28–30 bp downstream from the USE motif (S3 Fig). In sum, the presence of highly conserved transcription initiation sequences indicated our annotated TRs were likely to be expressed.

Identifying the telomerase RNA through total RNA transcriptomes of multiple Mimulus species

We aimed to obtain the TR sequence from multiple Mimulus species through de novo transcriptomics and assemble the TR transcript from the total RNA pool. We chose the three divergent Mimulus species (M. aurantiacus ssp. puniceus, M. guttatus, and M. verbenaceus) and synthesized cDNA from total RNA extracted from mature leaf, root, and floral meristem tissues and conducted a RT-PCR analysis. Prior studies have discovered high expression of the TR in plant tissues with actively dividing cells [27,61] and we also detected expression of the TR gene in the floral meristem for all three Mimulus species (Fig 3A). We observed no PCR bands when raw total RNA was used for the RT-PCR experiment (S4 Fig), indicating the RT-PCR positive bands did not originate from genomic DNA contamination. There was also some evidence of TR expression in other tissues that are largely composed of differentiated cells (root and mature leaf).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Molecular evolutionary analysis of the Mimulus Telomerase RNA (TR) gene.

(A) RT-PCR results amplifying TR transcripts on cDNA generated from RNA extractions of three tissues (root, mature leaf, and floral meristem). PCR on genomic DNA is shown as a positive control. (B) Phylogeny of the entire TR sequences assembled from 18 Mimulus species/subspecies. Sequences were obtained from total RNA transcriptome sequencing or by annotating the TR sequences from reference genomes (i.e., M. aurantiacus ssp. puniceus, M. cardinalis, M. parishii, and M. verbenaceus). Nodes with bootstrap support >95% are indicated with a red circle. Phylogenetic groups are colored according to the three Mimulus sections. Arrows point to the M. cardinalis and M. lewisii TR duplication. Blue arrows indicate TR1 and the red arrows indicate TR2. (C) Alignment of the TR templating region (see Fig 1 for the involvement of templating sequence in telomerase complex during telomere DNA synthesis). Sequences are highlighted according to the color scheme of (B). Based on the templating sequence, the Diplacus section will synthesize a TTTAGGG telomere repeat and the Simiolus section a TTTCGGG telomere repeat. In the Erythranthe section, all species will synthesize a TTTCGG telomere repeat. However, since in addition to TR1, M. cardinalis and M. lewisii have TR2 with a TTTCGGG templating sequence both species could have a potentially sequence heterogeneous telomere sequence (but see Fig 4 for results on the effects of M. cardinalis and M. lewisii TR duplication on telomere sequence).

https://doi.org/10.1371/journal.pgen.1011738.g003

Based on the RT-PCR results we targeted the floral meristem for total RNA transcriptome sequencing with a goal to assemble the TR sequence in multiple Mimulus species. We examined the floral meristem transcriptomes from 16 Mimulus species/subspecies and these include 8 (sub)species from the Diplacus section (M. aurantiacus ssp. aridus, M. aurantiacus ssp. aurantiacus, M. aurantiacus ssp. calycinus, M. clevelandii, M. aurantiacus ssp. grandiflorus, M. aurantiacus ssp. longiflorus, red flower ecotype M. aurantiacus ssp. puniceus, and yellow flower ecotype aurantiacus ssp. M. puniceus), 7 (sub)species from the Simiolus section (M. decorus, M. glaucescens, coastal perennial M. guttatus, inland perennial M. guttatus, inland annual M. guttatus, M. nasutus, and M. tilingii), and 1 species from the Erythranthe section (M. lewisii). We also took the TR sequences annotated from the reference genome (listed in S1 Table) for a combined total of 19 (sub)species (see S2 Table for list of analyzed species) TR sequences available for downstream analysis.

After quality control trimmed sequencing reads, we used the trinity pipeline [62] to conduct de novo transcriptome assembly. The contig N50 ranged from 617–2619 bp and the number of assembled transcripts ranged from 117,776–266,437 (S3 Table). BUSCO scores were calculated and ranged from 26.4–70.4%. These low BUSCO scores may be related to our total RNA based sequencing strategy. We depleted rRNA but omitted a polyadenylated transcript selection step to enrich noncoding RNA sequences such as the TR, but this does not control for the vastly over abundant organellar (chloroplast and mitochondria) transcripts that are not polyadenylated [63]. Consequently nuclear gene transcripts are likely to be under sampled, potentially resulting in the suboptimal assembly for coding sequences. Despite this limitation, we took the de novo transcriptome assembly and using the TR gene annotations from the reference genome assemblies, a BLAST-based nucleotide similarity analysis was able to identify the TR orthologs from 15 of the sequenced transcriptomes (no TR was successfully identified for the inland perennial M. guttatus). In the end, we determined the TR sequence from 15 Mimulus species from the transcriptome data and additional TR sequences from 3 Mimulus species annotated from the genome assemblies (M. cardinalis, M. parishii, and M. verbenaceus) for a total of 18 unique Mimulus species.

The telomerase RNA gene has duplicated in several Mimiulus species.

The Mimulus TR sequences were aligned to each other with MUSCLE [64] and the multi-sequence alignment was used to build a maximum-likelihood based phylogenetic tree to infer the evolutionary relationships (Fig 3B). The tree had high bootstrap support on internal nodes that grouped the species by Mimulus section (>99% after 1,000 bootstrap replicates) and the internal branch lengths between sections were deep. This phylogenetic evidence indicated there were substantial nucleotide differences between the TR sequences for species from different sections. This was also visually apparent in the multi-sequence alignment of the TR gene (S5 Fig), which displayed several insertions and deletions (indels) of variable sizes between Mimulus sections.

During the genome annotation and transcriptome-based analysis of the TR gene, we discovered the TR gene was duplicated in several Mimulus species (M. aurantiacus ssp. puniceus, M. cardinalis, and M. lewisii). We then focused our analysis on the Mimulus TR paralogs to understand the evolutionary origin of the duplication event.

In M. aurantiacus ssp. puniceus, the reference genome harbored two TR gene copies, but our de novo assembly from each Diplacus species showed evidence of only a single TR gene. We randomly designated the two TR copies from the M. aurantiacus ssp. puniceus reference genome as TR1 and TR2. Across the 244 aligned basepairs between M. aurantiacus ssp. puniceus TR1 and TR2 sequences, there were 11 nucleotide differences and a 50 bp insertion found only in TR2 (S5 Fig). We investigated the phylogenetic relationship between M. aurantiacus ssp. puniceus TR1 and TR2 from the reference genome, and the rest of the Diplacus section TR, but the bootstrap supports were too low to make any confident evolutionary inferences (Fig 3B). This suggested a recent evolutionary origin for the likely TR duplication in M. aurantiacus ssp. puniceus. In addition, our RT-PCR results (Fig 3A) detected the presence of only a single band, and given that the M. aurantiacus ssp. puniceus TR1 and TR2 genes from the reference genome have a 50 bp difference in size, if both genes were expressed we would expect to see two bands. However, the presence of a single TR band suggested that only one of the TR paralogs is likely being expressed.

M. cardinalis and M. lewisii were another pair of species where their reference genome assembly had two TR gene annotations. For M. lewisii we conducted total RNA transcriptomics analysis and we were able to de novo assemble both TR genes, indicating the two TR paralogs were expressed in the M. lewisii floral meristem. There was strong phylogenetic support (>95% after 1,000 bootstrap replicates) that grouped TR by paralog and not by species. We designated the TR paralog that grouped with M. parishii, and M. verbenaceus as TR1, while the duplicate only found in M. cardinalis and M. lewisii as TR2. Since TR1 and its orthologs were found in all four Erythranthe species, the TR1 paralog was hypothesized as the ancestral copy (see later analysis with phylogenetic outgroups that further indicates TR1 is the ancestral paralog) while TR2 was proposed to be the recently derived copy. The branch lengths that separated TR1 and TR2 were deep. Between M. lewisii TR1 and TR2 there were 241 basepairs aligned with 31 nucleotide differences, while between M. cardinalis TR1 and TR2 there were 251 basepairs aligned with 33 nucleotide differences. The divergence time between TR1 and TR2 was estimated to be ~ 9.3 million years ago, which is older than the species divergence time between M. cardinalis and M. lewisii (5.5 million years ago) [57].

The Mimulus family telomere sequence evolution

The functional domains within the TR gene (i.e., conserved regions) [27] were largely conserved across the Mimulus TR multisequence alignment (S5 Fig), and we then focused our analysis on the TR template domain as it determines the telomere sequence (Fig 1). Within the template domain, the annealing sequence was identical for all Mimulus species and it corresponded to the nucleotides AACC. Given that the annealing sequence is crucial for the proper template shift and double-stranded binding between the telomere DNA and TR RNA sequence, it was expected that the annealing sequence would be conserved. But the templating sequence harbored nucleotide differences, specifically between Mimulus species from different sections. We discovered all Diplacus section species had the templating sequence AACCCTA, all Simiolus section species had the templating sequence AACCCGA, and in the Erythranthe section the species with the TR1 orthologs had the templating sequence AACCGA (Fig 3C).

In summary, our results indicated the Mimulus genus had at least two mutational changes in the telomere sequence motif. Species of the same section had identical telomere sequence motifs, but between species of different sections there was variation in the telomere sequence motif suggesting the evolutionary changes occurred in the common ancestors of each section. Based on these evolutionary changes we propose a single-step mutational model of telomere sequence evolution to explain the genetic variation within the Mimulus telomere. The telomere sequence motif AAACCCT is the most commonly observed sequence in plants, hence it is likely to be the ancestral telomere sequence in Mimulus and is thus present in the Diplacus section species. A single nucleotide change then converted the ancestral sequence to AAACCCG in the common ancestor of the Simiolus and Erythranthe section. Subsequently in the Erythranthe section, a single nucleotide deletion resulted in the formation of AAACCG telomere sequence. A similar evolutionary model has also been proposed in fungi (i.e., termed as the step-by-step model), where progressive changes from an ancestral telomere sequence resulted in the diversity of yeast telomeric repeats [14]. Importantly, this scenario suggests a common evolutionary mechanism or constraint may underlie the telomere sequence variation across the telomeres of all eukaryotes. The model also predicts as the telomere sequence diverges, the telomere binding proteins would display complementary evolutionary changes, as was proposed earlier [65]. Further detailed analysis of this coevolutionary signatures in the telomere protein-DNA complexes would be an intriguing future study.

The M. lewisii TR duplication results in a sequence heterogenous telomere

We next examined the template domain in Mimulus species with TR duplications. If the templating sequence between the TR paralogs exhibits natural genetic variation, this could result in the species carrying a telomere with a mixed repeat motif (i.e., a heterogeneous sequence structure). For the TR sequences from the M. aurantiacus ssp. puniceus reference genome, both TR1 and TR2 paralogs had the identical templating sequence. But in M. cardinalis and M. lewisii the two TR paralogs differed in the templating sequence. The TR1 paralog had the templating sequence AACCGA, meanwhile the TR2 paralog had the templating sequence AACCCGA. This observation suggests that M. cardinalis and M. lewisii might synthesize a telomere with a heterogeneous sequence that comprised of both AAACCG and AAACCCG repeats.

We investigated the possible functionality of the M. cardinalis and M. lewisii TR duplicates by first conducting RT-PCR of the duplicate genes in three different tissues (root, mature leaf, and floral meristem). We first compared the expression profiles of the TR gene in M. cardinalis and M. lewisii, which have the TR duplications to M. parishii and M. verbenaceus, which also belong to the same Erythranthe section but without the TR duplication. The M. verbenaceus TR expression profile and telomere sequence motif analysis were shown in Figs 2, 3, and S2; and M. parishii had identical results to M. verbenaceus (Figs 4 bottom and S6). In summary, M. parishii and M. verbenaceus synthesize a single motif telomere sequence (AAACCG)_n from its single copy TR gene.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Erythranthe section Telomerase RNA (TR) gene duplication and its association with the telomere sequence.

Telomere sequence abundance at reference genome assembly chromosome ends (top) and RT-PCR results (bottom) from three tissues (root, mature leaf, and floral meristem) with genomic DNA is shown as a positive control for (A) M. lewisii, (B) M. cardinalis, and (C) M. parishii are shown. See S6 Fig for RT-PCR results from total RNA. Note M. verbenaceus is also a species from the Erythranthe section and its results have already been shown in Figs 3A and S2, and largely corroborated M. parishii results.

https://doi.org/10.1371/journal.pgen.1011738.g004

For M. cardinalis and M. lewisii, both TR1 and TR2 were expressed in multiple tissues (Fig 4 bottom), with a particularly relevant expression in the floral meristem, which harbors abundant actively dividing cells. The similar TR gene expression profiles between M. cardinalis and M. lewisii suggested both species may have a telomere with a heterogeneous sequence structure. We investigated this possibility by quantifying the telomere repeat motifs at chromosome ends of M. cardinalis and M. lewisii chromosome-level reference genome assemblies, and discovered the two species had contrasting telomere sequences. In M. cardinalis only the AAACCG sequence was found at chromosomal ends, meanwhile in M. lewisii both AAACCG and AAACCCG sequences were present at terminal chromosomal locations (Fig 4 top). This result for M.cardinalis would be consistent with the scenario where TR1 is the only RNA subunit utilized by telomerase, while TR2, though expressed, would not contribute to telomere repeat synthesis and, thus, remain functionally ineffective

M. lewisii has a sequence heterogenous telomere consisting of two telomere repeats.

Given that M. lewisii chromosome-level genome assembly suggested the presence of both AAACCG and AAACCCG repeats at chromosomal ends, we conducted a detailed examination of the M. lewisii telomere structure by using the Oxford nanopore sequencing platform. This allowed us to extract and analyze the long read sequences that span the entire telomere-subtelomere region. We nanopore sequenced M. lewisii and its two closely related species M. cardinalis and M. verbenaceus, and generated 1.11 Gbp, 0.74 Gbp, and 1.66 Gbp of sequencing data, which comprised of 15,990,860 reads, 10,699,772 reads, and 24,581,417 reads for M. cardinalis, M. lewisii, and M. verbenaceus respectively (see S4 Table for more sequencing information).

We matched the sequencing library to its species-specific reference genome, extracted reads that were longer than 10 kbp and aligned them to chromosome ends. For the three species, the total number of nanopore long reads that aligned to the telomere region were 82, 102, and 126 reads with a median read length of 24,995 bp, 22,733 bp, and 20,784 bp for M. cardinalis, M. lewisii, and M. verbenaceus respectively (Fig 5A). We then conducted a sliding window analysis by starting at the beginning of the telomere sequence (i.e., at the chromosome start or end depending on the arm of the chromosome) and examined the distribution and quantity of the 6-mer AAACCG or 7-mer AAACCCG repeats. Due to the anticipated sequencing errors associated with nanopore reads [66], counting perfectly matched 6-mer or 7-mer telomere repeat would underestimate the number of telomere repeats within the sequencing reads. Instead, a visual analysis of the raw telomere reads indicated that each species had a noticeable difference in the number of CC (and its complement GG) and CCC (and its complement GGG) repeats (S7 Fig). The former likely represented the 6-mer repeat while the latter represented the 7-mer repeats.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Nanopore long read sequencing of the telomeres in select Erythranthe section species.

(A) Density plot showing the lengths of sequencing reads that aligned to chromosome ends. Only reads that were longer than 10 kbp were analyzed. Vertical lines are showing the average read lengths which were 28,627 bp for M. cardinalis, 28,744 bp for M. lewisii, and 23,402 bp for M. verbenaceus. (B) 500 bp sliding window analysis of counting CC/GG and CCC/GGG repeats for M. cardinalis, M. lewisii, and M. verbenaceus. The red asterisk indicates the window where there is a significant reduction (Mann-Whitney U test p-value < 0.05) in CC/GG repeat counts. (C) Sum of the CCC/GGG repeat counts up to the 3,000 bp window. Significant differences (Mann-Whitney U test p-value < 0.05) are indicated with a black asterisk.

https://doi.org/10.1371/journal.pgen.1011738.g005

A 500 bp window sliding window analysis was conducted for each telomere read and within the window we counted the total number of CC/GG and CCC/GGG repeats (Fig 5B). Results showed for all three species the CC/GG repeat was the most abundant type (~58 counts per window) for all three species, indicating that the 6-mer AAACCG is the predominant telomere repeat. This finding is consistent with our reference genome based analysis results (Fig 4). In addition, the CCC/GGG repeat counts were significantly higher in M. lewisii compared to both M. cardinalis and M. verbenaceus (Fig 5B and 5C; Mann-Whitney U test p-value < 0.05). Noticeably, the CCC/GGG repeat counts were elevated across the windows and up to the candidate telomere-subtelomere boundary region (Fig 5B). This indicated the 6-mer and 7-mer repeats were interspersed throughout the M. lewisii telomere.

We then tested the window where there was a significant reduction in the CC/GG repeat count to detect the regions corresponding to the telomere-subtelomere boundary. In M. cardinalis and M. verbenaceus it was mapped to the 2,500 bp window, whereas in M. lewisii it was considerably larger and corresponded to the window of 3,000 bp (Fig 5B and significant reduction was determined after a Mann-Whitney U test p-value < 0.05). This indicated M. lewisii had a potentially 20% longer telomere compared to M. cardinalis and M. verbenaceus. We validated our telomere length estimates from the genome sequencing data by conducting the Terminal Restriction Fragment (TRF) analysis on M. cardinalis, M. verbenaceus, and M. lewisii DNA. Following restriction digestion of genomic DNA from all 3 species with Tru1I restrictase, we first hybridized the membrane with a 32P-labeled probe to target the AAACCG repeat (Fig 6A, left panel). The probe hybridized to digested DNA samples from all species, confirming our predictions that the AAACCG repeat is universally present on chromosomal ends of all tested Mimulus species. Furthermore, quantitative analysis of the TRF signals with the software WALTER [67] indicated that the mean telomere length in M. lewisii (3,511 bp) is up to 1,000 bp longer than in both M. cardinalis and M. verbenaceus (2,543 bp and 2,594 bp, respectively) (Fig 6B, two-tailed Student’s t-test p-value < 0.01). This finding further corroborates our telomere length estimates from the nanopore long read sequencing (Fig 5B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Terminal restriction fragment (TRF) analysis of M. cardinalis, M. verbenaceus, and M. lewisii DNA.

(A) TRF Southern blot of DNA samples from individual plants of M. cardinalis (n = 3), M. verbenaceus (n = 3) and M. lewisii (n = 4) species hybridized to the AAACCG-specific probe (left panel) or re-hybridized after stripping to AAACCCG-specific probe (right panel). Molecular weight DNA markers (in kb) are shown. (B) Telomere length (mean TRF) distributions in ≥3 biological replicates of each genotype from (A) are shown in boxplots. Data points represent mean TRF values from individual plants of corresponding species (biological replicates) analyzed with WALTER after hybridization to AAACCG probe. Whiskers indicate maximum to minimum values; boxes represent the lower and upper quartiles (25 and 75%); horizontal lines represent medians of the mean TRF values. ** indicates significant difference p-value < 0.01 after a two-tailed Student’s t-test.

https://doi.org/10.1371/journal.pgen.1011738.g006

To test for the presence of AAACCCG telomere repeats at the chromosomes of the three Mimulus species, we then stripped the AAACCG-specific probe (S8 Fig) and rehybridized the membrane with the AAACCCG-specific probe (Fig 6A, right panel). Remarkably, the AAACCCG-specific probe hybridized equally well to the M. lewisii DNA, indicating that chromosomes in this Mimulus species harbor a mixture of AAACCG and AAACCCG telomere repeats. Furthermore, the signal for AAACCCG-specific probe was nearly absent from the M. verbenaceus DNA, confirming our earlier predictions that telomeres in this species are made up uniquely of AAACCG repeats (Fig 2). Finally, we also detected much weaker AAACCCG-specific signals for the M. cardinalis DNA compared to the AAACCG probe in this species. This observation is consistent with the presence of very low amounts of AAACCCG telomere repeats at the M. cardinalis chromosomes.

Overall, our TRF analysis with AAACCG and AAACCCG specific probes confirmed that M. lewisii is unique among the tested Mimulus species in that it uses both types of telomeric templates interchangeably (from two different TR molecules), and that its telomeres are significantly longer than in the other two analyzed Mimulus species (two-tailed Student’s t-test p-value < 0.01) (Fig 6B). These experimental measurements are very similar to the genome sequencing based estimates we obtained earlier (~3,000 bp for M. lewisii versus ~2,500 bp for both M. cardinalis and M. verbenaceus).

The nanopore sequencing results and the TRF data were then additionally validated using fluorescence in situ hybridization (FISH) experiments to label the telomere repeat sequence of M. lewisii chromosomes and compared it to other Mimulus species. We labeled the telomeres of Mimulus species using a AAACCG probe or a AAACCCG probe and hybridized on metaphase chromosomes (Fig 7). Initially we hybridized the probes on chromosome spreads of M. verbenaceus and M. guttatus, since both species appears to have a single telomere repeat variant (Fig 2), allowing us to test the specificity of our probes. Results showed that in M. verbenaceus, the AAACCG probe hybridized to the chromosome ends but the AAACCCG probe did not show any signal (Fig 7 left panel). Conversely, for M. guttatus the AAACCCG probe hybridized to the chromosome ends but the AAACCG probe did not show any signal (Fig 7 middle panel). This indicated the AAACCG probe and AAACCCG probe were specific and hybridized to the corresponding telomeres that matched the probe sequence. In M. lewisii, however, both probes were detected at chromosome ends, further confirming that in this Mimulus species telomeres consists of a mixture of two different repeat sequences (Fig 7 right panel).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Fluorescence in situ hybridization (FISH) on metaphase chromosomes of M. verbenaceus (2n = 16), M. guttatus (2n = 28), and M. lewisii (2n = 16) using (A) AAACCG probe or (B) AAACCCG probe.

For each probe the top panel displays the telomere specific FISH signals and the bottom panel displays the DAPI (4’,6-diamidino-2-phenylindole) stained metaphase chromosomes with the telomere specific FISH signals. Bar represents 5 µm.

https://doi.org/10.1371/journal.pgen.1011738.g007

The telomerase synthesizes AAACCG and AAACCCG telomere repeats in M. lewisii but not in M. cardinalis.

The heterogeneity of M. lewisii telomeres with both AAACCG and AAACCCG telomere repeats suggest the telomerase is able to synthesize both repeat types in vivo (Figs 5B, 6 and 7). We conducted RT-qPCR of the TR1 and TR2 transcripts in leaf and meristem tissues and discovered the TR1 gene with the AAACCG template had a significantly higher expression (Mann-Whitney U test p-value < 0.05) compared to the TR2 gene with the AAACCCG template (S9 Fig). We hypothesize that the repeat composition of the M. lewisii telomeres is driven by the transcript abundance levels of the TR1 and TR2 genes.

We investigated the activity of the telomerase by conducting a Telomeric Repeat Amplification Protocol (TRAP) assay [68]. TRAP is an in vitro assay for examining the activity of the telomerase and involves a telomerase-mediated extension of a non-telomeric substrate. We conducted TRAP on M. lewisii, M. cardinalis, and M. verbenaceus protein extracts and sequenced the TRAP products to investigate the repeat sequence synthesized by the telomerase of each species. We used a previously published substrate primer (i.e., TS21) that worked in a variety of plant species and two previously published reverse primers (i.e., TelPr and HisPr_long) that amplified TRAP products with AAACGGG like repeats [49]. We also designed a reverse primer to amplify TRAP products with a AAACCG repeats (i.e., MvV1). All three reverse primers showed positive TRAP products and when the reactions were repeated following RNAse A treatments, the products were not observed indicating the TRAP assay was RNA dependent (left images on Fig 8). The TRAP products were then cloned for Sanger sequencing and only clones that had both primer sequences (i.e., TS21 substrate primer and the reverse primer) and a tandem repeat in between were analyzed. In M. verbenaceus we screened a total of 13 clones (TelPr = 3, MvV1 = 5, and HisPr_long = 5) and discovered 17 AAACCG tandem repeats and 1 AAACCCG repeats (see Fig 8 for examples from each reverse primer). This was expected since in M. verbenaceus the telomerase is predicted to synthesize the AAACCG repeat sequence. In M. lewisii we screened a total of 12 clones (TelPr = 4, MvV1 = 3, and HisPr_long = 5) and discovered 9 AAACCG tandem repeats and 26 AAACCCG repeats. This indicated the M. lewisii telomerase was likely using both TR1 and TR2 templates for synthesis and explained the sequence heterogenous M. lewisii telomere. In M. cardinalis we screened a total of 15 clones (TelPr = 5, MvV1 = 5, HisPr_long = 5) and discovered 28 AAACCG tandem repeats and 3 AAACCCG repeats. Despite both TR1 and TR2 being expressed in M. cardinalis, telomerase appears to preferentially synthesized the TTTGCC (i.e., the reverse complement of AAACGG) repeat. The M. cardinalis telomerase may have a preference for binding to TR1 or it may preferentially synthesizes the TTTGCC repeat regardless of the TR gene. This preference is also consistent with our TRF data and may also explain why the M. cardinalis telomeres largely consist of the AAACCG repeat despite both TR1 and TR2 being expressed in this species.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. TRAP assays of telomerase activity and sequencing results of cloned TRAP products.

Protein extracts from floral meristems of (A) M. verbenaceus, (B) M. lewisii, and (C) M. cardinalis were analyzed. (Left) TRAP products were separated on a polyacrylamide gel and stained for imaging. The (-) lane indicates TRAP products without the addition of RNAse A. The (+) lane indicates TRAP products from an assay including 2 ng of RNAse A and (++) lane indicates TRAP products from an assay including 5 ng of RNAse A. (Right) Representative images of Sanger sequencing of clones generated from positive TRAP assays. Telomere motifs are indicated with colored arrows where AAACCG (i.e., TTTCGG) motif is indicated with blue arrow, AAACCCG (i.e., TTTCGGG) motif is indicated with green arrow, and neither motif is indicated with grey arrow.

https://doi.org/10.1371/journal.pgen.1011738.g008

TR duplication event was ancient and both copies were retained in M. lewisii due to functionality.

It’s possible M. parishii and M. verbenaceus also had a TR2 gene but it was subsequently lost. This scenario would support the notion that the M. lewisii TR duplication was an evolutionarily old event. We conducted a comparative genomic analysis on the TR2 gene region in M. cardinalis and M. lewisii, and also examined the syntenic region in M. parishii and M. verbenaceus. Results showed clear synteny across the 4 species in the genomic region surrounding the TR2 gene, which is located on chromosome 7 for all 4 species (S10 Fig). We conducted nucleotide blast analysis using the TR2 sequence as the query and searched for presence of the TR2 in the syntenic regions in M. parishii and M. verbenaceus. In M. parishii we discovered a 36 bp hit (97% identity) and in M. verbenaceus a 80 bp hit (88% identity), but no match for the large majority of the TR2 gene was observed. These results indicated M. parishii and M. verbenaceus did not have a functional TR2 paralog and the sequences that do remain in the genome are potentially pseudogenized gene remnants.

To gain a deeper understanding of the evolutionary history underlying the M. lewisii TR duplication, we analyzed the draft genome assemblies [69] from two closely related outgroup species M. bicolor and M. primuloides. Evolutionarily, M. bicolor (section Monimanthe) is a phylogenetic sister to the Erythranthe section, while M. primuloides (section Monantha) is a third outgroup. The genomes for these two species were generated from short read (150 bp) Illumina sequencing and their assembly is fragmented, but through blast search we were able to discover the TR gene in both species and together with the Erythranthe section TR sequences we reconstructed a phylogenetic tree (S11 Fig). We also used all Mimulus TR sequence dataset from Fig 3B to build a phylogenetic tree and the M. bicolor and M. primuloides formed a monophyletic group with the Erythranthe section (S12 Fig). M. bicolor had three TR gene copies and one paralog grouped with the TR1 paralogs while the other two grouped with TR2 paralogs. In the outgroup M. primuloides there was only one TR gene and phylogenetically it clustered with the TR1 paralogs (but with low bootstraps). M. primuloides TR had the templating sequence AACCGA suggesting for all three Erythranthe, Monimanthe, Monantha section Mimulus species the (AAACCG)_n is the ancestral telomere and the TR1 paralog is the ancestral TR sequence. Further, our evidence suggests TR2 paralog arose from an evolutionarily old duplication event, occurring at the common ancestor of the Erythranthe and Monimanthe section (~8.2 million years ago) [57]. Most gene duplicates become nonfunctional [70], explaining the pseudogenization of the TR2 paralog in M. cardinalis, M. parishii, and M. verbenaceus. But in M. lewisii the TR2 paralog is clearly functional and appears to be directly involved in synthesizing the M. lewisii telomere. The initial TR duplication event was old and predates the speciation of M. lewisii, suggesting that both TR paralogs have been functional in M. lewisii since its evolutionary origin.

The TR duplication arises through transposition and the ancestral copy is eventually lost.

During the analysis of the M. lewisii TR gene duplication we noticed the TR paralogs were located at two different chromosomes. In M. lewisii, the TR1 gene was located at chromosome 6 while the TR2 gene was located at chromosome 7 (S1 Table). It was intriguing that the two duplicates were located at physically distant genomic positions and we investigated the evolutionary mechanism that resulted in this pattern using synteny analysis.

The genomic region surrounding the M. lewisii TR1 gene was highly syntenic with its sister species M. cardinalis, M. parishii, and M. verbenaceus (S10 Fig and S1 Table). We note TR1 in M. cardinalis is located at chromosome 7, which seemingly contradicts with the TR1 position of chromosome 6 in M. lewisii, but this is due to the TR1 gene located within a natural chromosomal translocation that occurred between chromosome 6 and 7 [71]. We then investigated the synteny of the M. lewisii TR1 gene in the evolutionarily divergent M. guttatus and M. aurantiacus ssp. puniceus reference genomes, and discovered those two species did not have a TR gene in the syntenic region (S1 Table). In fact for all three species (M. aurantiacus, M. lewisii, and M. guttatus), the TR gene was located at non-syntenic positions and on different chromosomes (S1 Table and Fig 9). The M. lewisii TR2 gene had the same results where syntenic region in the M. guttatus and M. aurantiacus ssp. puniceus reference genomes did not have a TR gene. In summary species in the same section shared TR synteny, but between species of different sections the TR was located at non-syntenic chromosomal positions (S1 Table). A possible scenario to explain these results is if the TR gene is able to duplicate and move to different chromosomes through a transposition-mediated mechanism of gene duplication [72]. We hypothesize this transposition event had occurred in the common ancestor of each Mimulus section but then the ancestral TR paralog was lost and the derived TR paralog survived, explaining why species from different Mimulus sections do not share TR gene synteny. This turnover of TR gene paralogs ultimately explains the evolutionary basis of telomere sequence variation in Mimulus. It is intriguing that each Mimulus section with the derived telomere sequence also has the TR gene on a different chromosome, indicating it was potentially the derived TR paralog that was consistently retained. Currently we do not have an evolutionary explanation for this observation, but the retention of the derived TR paralog suggests an unknown selection mechanism may be involved in the process. One hypothesis could be related to genetic conflicts [73,74], as both TR paralogs would interact with telomerase complex for proper function there, which may lead to competition for telomere binding proteins.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Synteny plot surrounding the Telomerase RNA (TR) gene.

Each panel is showing a focal species (A) M. aurantiacus ssp. puniceus, (B) M. guttatus, and (C) M. verbenaceus, and its TR chromosomal location (indicated with a red arrow) is plotted in the central position and highlighted with red letters. Orange arrows indicate genes and gray boxes indicate orthology between genes. For each focal species TR gene, we show the five genes upstream and downstream with an orthology. In each plot there is clear synteny between species surrounding a focal species TR gene, but in the non-focal species there is no TR gene and instead it is located at a different chromosome (compare the TR locations in panel A, B, and C).

https://doi.org/10.1371/journal.pgen.1011738.g009

The molecular mechanism that led to the transposition of the TR gene is not known, but we speculate transposable elements may have been involved. In six of the nine Mimulus species where we annotated the TR gene in the reference genome, there was a transposable element sequence within 2,000 bp of the TR gene annotation and all six of them were a DNA transposon (S5 Table). In plants, DNA transposons have been hypothesized as a major mediator of gene duplication through direct or indirect consequences from the transposition mechanism of the transposon [75]. Thus, we posit the transposition and duplication of the TR gene are potentially mediated through DNA transposons.

Analyzed reference genomes

All Mimulus reference genomes were downloaded from Mimubase (http://mimubase.org/). Specifically, the analyzed genome versions were M. guttatus v2.0 [77], M. aurantiacus ssp. puniceus v1.0 [53], M. cardinalis v2.0, M. lewisii v2.0, M. parishii v2.0, and M. verbenaceus v2.0. We note the genome sequences for M. cardinalis, M. lewisii, M. parishii, and M. verbenaceus are early access publicly available versions and a manuscript describing the biological properties of those genomes will be published elsewhere.

k-Seek based telomere sequence analysis.

We downloaded M. aurantiacus, M. guttatus, and M. verbenaceus whole genome resequencing data [53–55] from NCBI SRA with the identifiers PRJNA549183, PRJNA344904, and PRJNA813304. Raw reads were trimmed using BBTools (https://jgi.doe.gov/data-and-tools/bbtools/) bbduk.sh script v39.01 with parameters minlen = 25 qtrim = rl trimq = 10 ktrim = r k = 25 mink = 11 hdist = 1 tpe tbo. Trimmed reads were then analyzed with k-Seek [51,52] to quantify the total copy number of a tandem repeat sequence. K-Seek identifies tandem repeat motifs of up to 20 bp in the raw sequencing reads. It breaks down the sequencing read into smaller fragments (i.e., k-mers) and groups it into sequence identity groups. This allows identification of k-mers and its subsequent quantification by analyzing all sequencing reads.

k-Seek requires a k-mer repeat length to be at least 50 bp in order to avoid over counting small tandem repeats scattered across the genome. To account for the differences in genome coverage between samples we normalized each sample’s k-mer counts by the sample’s median genome coverage. Genome coverage was calculated by first aligning the whole genome resequencing reads to the reference genome of the respective species using bwa-mem v0.7.16a-r1181 [77] and then the per-sample average coverage was calculated using bedtools v2.25.0 [78].

Reference genome based telomere sequence analysis

All Mimulus species have a reference genome that is assembled into chromosomes (i.e., linkage groups), hence if the telomere sequence was assembled it would be at the ends of each chromosome assembly. From each species reference genome we extracted the starting and ending 500 bp of the assembled chromosome sequence using samtools v1.14 [79]. For each extracted sequence we then searched for the telomere repeat sequences AAACCCT, AAACCCG, or AAACCG using bash command grep with option -aob and counted the number of matches. The data was visualized using ggplot2 package [80] in R v4.3.1 (RStudio 2020).

Plant material

The Mimulus species that were used in this study with the growth conditions are mentioned in S2 Table. Total genomic DNA was isolated from the leaves of M. cardinalis, M. guttatus, M. aurantiacus ssp. puniceus, M.parishii, M. verbenaceus, and M. lewisii using the modified CTAB method [81].

RNA isolation and reverse transcription-polymerase chain reaction (RT-PCR).

The total RNA was extracted from 100 mg of floral meristem, root, and leaves using TRIzol (Invitrogen) according to the manufacturer’s protocol. The RNA was further treated with DNaseI (RNase-free, New England Biolabs) followed by column purification using Monarch Total RNA Miniprep Kit (New England Biolabs). The ProtoScript First Strand cDNA Synthesis Kit (New England Biolabs) was used for the synthesis of cDNA using the manufacturer’s protocol. One µg of RNA was reverse transcribed using random primers.

RT-PCR was performed on the root, leaves, and floral meristem of M. cardinalis, M. guttatus, M. aurantiacus ssp. puniceus, M.parishii, M. verbenaceus, and M. lewisii. Four µl of 10X diluted cDNA was used in a 20 µl reaction mix. Taq DNA Polymerase with Standard Taq Buffer (New England Biolabs) was used for the amplification of TR under conditions as follows: 10 min at 95°C; 30 sec at 95°C; 30 cycles of 30 sec at 95°C, 30 sec at 55°C, 15 sec at 68°C; and final extension at 68°C for 5 min. The primers used for the RT-PCR are listed in the S6 Table. The PCR products were separated using gel electrophoresis at 95 volts for 40 minutes on a 2% agarose gel stained with ethidium bromide.

RNA-seq library preparation

The total RNA extracted from the floral meristem of species listed in S2 Table were checked for quality using RNA ScreenTape (Agilent Technologies) and Qubit 2.0 fluorometer was used for determining the concentration. From the total RNA the ribosomal RNA (rRNA) were depleted using QIAseq FastSelect–rRNA Plant Kit. Using rRNA depleted RNA samples the RNA-seq libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit according to the manufacturer’s protocol. RNA-seq libraries were sequenced by the University of Kansas Genome Sequencing Core on a Illumina NextSeq 2000 platform using P3 Reagents and as 2 × 100 bp paired-end reads.

The total number of raw reads generated for each transcriptome ranged from 102–182 million reads and more than 98% of those reads were recovered after quality control trimming to remove adapters and low quality sequence regions (S3 Table). We then aligned the quality control trimmed reads back to a reference genome to check for the proportion of reads that originated from a genome. Using M. aurantiacus ssp. puniceus reference genome for Diplacus species transcriptomes, M. guttatus reference genome for Simiolus species transcriptomes, and M. lewisii reference genome for the M. lewisii transcriptome, we determined there were between 50.0–99.7% of the reads aligned back to a reference genome. The low alignment percentage for some species is likely due to the reference genome being too divergent from the species with RNA-seq reads to be mapped.

De novo transcriptome assembly

From the raw sequencing reads, the quality of reads were analyzed using fastqc v0.11.8 (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The reads were subjected to quality control trimming using the bbduk.sh script in the BBMap package v39.01 (https://sourceforge.net/projects/bbmap/). The processed data was used for de novo transcriptome assembly using the Trinity pipeline v2.8.5 [62] and following their assembly tutorial (https://github.com/trinityrnaseq/trinityrnaseq/wiki). The de novo assembly was runned as paired-end option with the command: Trinity --seqType fq --left “$FQ1” --right “$FQ2” --CPU 40 --max_memory 140G --output “$OUTDIR”. To align the reads back to the reference genome and check for read alignment statistics we used the program Bowtie2 v2.3.5.1 [82]. TrinityStats.pl script from the Trinity pipeline was used to compute transcript contig N50 values. To quantify the completeness of the transcriptome assemblies, the Benchmarking Universal Single-Copy Orthologs (BUSCO) v3.0.2 [83] was executed with the following command: run_BUSCO.py -c 20 -o OUTPUT_NAME --in SEQUENCE_FILE -l eudicotyledons_odb10 -m transcriptome.

Telomerase RNA gene bioinformatic analysis

Candidate TR gene annotation was conducted using a previously implemented bioinformatics methodology that was used for detecting the TR in various land plant species [27]. We used TR sequences from previously annotated plants to generate a position weight matrix (PWM) to search for the candidate TR sequences within the Mimulus reference genomes using fragrep2 [59], and the program Infernal [60] was used to search for candidate TRs using secondary structure and sequence similarity.

To search for the TR sequence within the de novo transcriptome assembly we used the TR gene that was annotated from the reference genome using fragrep2 and Infernal. Using the standalone blast+ v2.9.0 [84] the candidate TR gene in the transcriptome assemblies were identified using the following command: blastn -query “$QUERY_FASTA” -db “$REF_DB” -task blastn -outfmt 6. The closely related Mimulus species from the same section was used as the query sequence. The TR gene detected from each Mimulus species were aligned to each other using MUSCLE [64] with default parameters on MEGA v5.0 [85].

Annotating conserved and functional regions in the Mimulus TR

For species with the reference genome, the 100 bp upstream region of each species TR gene was retrieved for analyzing the conserved regulatory elements. The sequences were aligned to each other using MUSCLE [64] with default parameters on MEGA [85]. We specifically amplified the M. cardinalis TR2 upstream region and Sanger sequenced using primers mentioned in S6 Table. In addition, we annotated the functional domains with the Mimulus TR sequence using A. thaliana TR sequence. Since previous A. thaliana study had experimentally validated the secondary structure and functional domains of the TR [27], we aligned the A. thaliana TR sequence with the Mimulus TR multi-sequence alignment and annotated the Mimulus TR functional domains.

Phylogenetic analysis

The multi-species TR gene alignment was used for the phylogenetic analysis with RAxML v8.2.5 [86] to build a maximum likelihood-based tree. We used a general time-reversible DNA substitution model with gamma-distributed rate variation and bootstrap analysis was conducted with 1000 replicates.

Divergence time between the two TR duplicates was estimated using Nei’s genetic distance (Nei 1972) (D = 2μT where D is genetic divergence between the gene paralogs, μ is the mutation rate per generation, and T is number of generations). We used mutation rate estimate from A. thaliana (7E-9 base substitutions per site per generation) [87] and assumed 1 generations per year.

DNA isolation and library preparation for nanopore sequencing

For DNA extraction, leaves of M. lewisii, M. cardinalis, and M. verbenaceus were collected. High Molecular Weight DNA isolation was perfomed using the Wizard HMW DNA Extraction Kit (Promega). Library preparation was performed according to the Ligation Sequencing Kit V14 (SQK-LSK114) protocol from Oxford Nanopore Technologies (ONT). For each library, 1 μg of DNA was used as an input for DNA repair and end preparation for adapter attachment. DNA was treated with NEBNext FFPE DNA Repair Mix (NEB, M6630) and Ultra II End-prep Enzyme Mix (NEB, E7546) followed by purification using AMpure XP beads (AXP) and eluted in 61 μl of nuclease-free water. To attach sequencing adapters to DNA ends, 60 μl of eluted DNA from previous step was ligated with 5 μl of ligation adapter using NEBNext Quick T4 DNA Ligase (NEB, E6056) for 10 mins at room temperature. AXP beads were added and washed twice with Long Fragment Buffer (LFB). The final library was eluted in 15 μl of elution buffer and prior to sequencing DNA amount was quantified using Qubit. The library was kept on ice until ready to load on the MinION device.

MinION sequencing

The library was sequenced of a MinION Flow Cell (FLO-MIN114 R10.4.1 version) on a Mk1C (Oxford Nanopore Technologies) sequencer. Before sequence run start, a flow cell check was performed to determine the number of active pores available in flow cell. A total of 1000 μl of flow cell priming mix was loaded on priming port and a total of 75 μl of library mixed with Library Beads (LIB) was loaded on the SpotON sample port.

Data analysis

MinION base-calling was carried out by Guppy v3.6.1 + 249406c (https://nanoporetech.com/software/other/guppy). The reads were aligned back to reference genome using minimap2 v2.25 [88], using following command: “minimap2 -L -t “$CPU” -ax map-ont “$refgenome” “$fastq” > “$FILENAME”.sam. Nanopore reads aligning to reference genome telomere sequence region were extracted with samtools and the reads were analyzed with the grep command to count the telomere sequence repeat.

Terminal Restriction Fragments (TRF) analysis

Telomere length was measured by the Terminal Restriction Fragments (TRF) analysis [89] with minor modifications. Genomic DNA from individual plants was digested with Tru1l (ThermoFisher Scientific) restriction enzyme, and the digested DNA samples were separated by gel electrophoresis in 0.8% agarose gel at 55V for 18 h in 1X TAE buffer and transferred to a Hybond-N+ nylon membrane (GE Healthcare, Chicago, IL, USA). Digested DNA was hybridized with 32P-labeled CGGTTTCGGTTTCGGTTTCGGTTTCGGT probe for telomeric DNA sequence detection. CGGTTT-specific probe was subsequently stripped from the membrane (by incubating with 0.2N NaOH, 0.1% SDS at 42°C with shaking for 20 min twice), and reprobed with the 5’ end labeled CGGGTTTCGGGTTTCGGGTTTCGGGTTT probe. Radioactive signals were scanned with Amersham Typhoon IP PhosphorImager (Cytiva, Wilmington, DE, USA). Images were visualized with ImageQuantTL v10.2-499 software (Cytiva, Wilmington, DE, USA), and mean telomere length values (mean TRF) for each DNA sample were calculated using the WALTER program [67].

Real-time quantitative PCR (qPCR)

A quantitative real-time PCR was performed in M. lewisii leaves and floral meristem to check the level of expression of the TR paralogs. The qPCR was performed using PowerUp SYBR Green Master Mix (Applied Biosystems) on QuantStudio 3 Real-Time PCR System (Applied Biosystems). Each 10 μl reaction consisted of 5 μl of PowerUp SYBR Green Master Mix (2X), 1 μl of each primer, 2 μl of cDNA, and 1 μl of nuclease free water. The details of primer used in qPCR are mentioned in S6 Table. The reaction setup was as follows: initial activation at 50 °C for 2 min, 95 °C for 2 min followed by 40 cycles of denaturation at 95 °C for 15 s, and annealing/extension for 1 min at 60 °C in 96-well optical reaction plates. In the end dissociation step was performed in 3 steps, first at a ramp rate of 1.6°C/sec at 95 °C for 15 sec, second at 60 °C for 1 min and final at 0.15°C/second at 95 °C for 15 sec. The PCR efficiency was calculated using Data Analysis for Real-Time PCR (DART-PCR) [90].

Telomerase Repeat Amplification Protocol (TRAP) assay

The shoot apical meristem and floral buds of M. lewisii, M. verbenaceus, and M. cardinalis, were grounded in mortar and pestle using liquid nitrogen for protein isolation. Approximately, 100 mg of crushed tissue was added to 800 μl of Buffer W (50 mM Tris-acetate (pH 7.5), 5 mM MgCl₂, 100 mM potassium glutamate, 20 mM EGTA, 1.5% (wt/vol) polyvinylpyrrolidone, and 10% glycerol) containing 0.6 mM ribonucleoside vanadyl complex and 1.0 mM DTT. Extracts were incubated on a rotator for 10 minutes at 4°C, and centrifuge at 13,000 rpm for 15 min at 4°C. In the supernatant 350 µL of 50% PEG (6000 MW) was added and were incubated on a rotator for 30 min at 4°C. The products were then centrifuged at 8,000 rpm for 5 min at 4°C and the supernatant was discarded. A 100 μl of Buffer W was added and the pellet was gently resuspend and placed on a rotator for 30 min at 4°C [91]. The extracted protein was stored at -80°C until use. Protein concentration of extracted samples were determined using Bradford’s reagent.

Telomerase activity of extracted protein was tested using TRAP. Briefly, 500 ng of total protein was mixed with 1 µl of 10 µM TS21 (listed in S6 Table), 1X Taq PCR Master Mix (Cat. No. ID: 201445) and incubated at 37 °C for 45 mins. After extension, 1 µl of 10 µM reverse primer was added (listed in S6 Table), followed by PCR step of the TRAP assay (94°C for 15 min, 30 cycles of 94°C for 30 sec, 60°C for 30 sec, 72°C for 60 sec, with a final extension of 72°C for 1 min). For control experiment RNase was added in two different concentration, 40mg (+), and 80 mg (++). PCR products were separated on 10% native polyacrylamide gel in 0.5x TBE buffer and visualized after staining with SYBR Green I Nucleic Acid Gel Stain (Invitrogen, S7567) and visualized using LI-COR Odyssey M [92,93]. Reaction products from TRAP assay were cloned into a pCR2.1-TOPO plasmid using the TOPO TA Cloning Kit (Invitrogen, K450002). Positive clones were sent for Sanger sequencing.

Fluorescence in situ hybridization (FISH) analysis

Chromosome preparation and FISH procedure were performed following a previous method [94] with minor modifications. Mitotic chromosome spreads were prepared from the root tips of actively growing Mimulus plants, which were placed in a nitrous oxide gas chamber for 1.5 h. The root tips were then fixed overnight in a solution of ethanol and glacial acetic acid (3:1) and squashed with a drop of 45% acetic acid. All preparations were stored at -70°C until use.

After carefully removing the coverslips with a double-edged razor blade, the slides were pre-treated with 20 μg/mL pepsin in 10 mM HCl for 3 min at 37°C, followed by immersion in distilled water for 1 min to stop the pepsin reaction. The pre-treated slides were then washed three times with 2x SSC (300 mM Na-citrate, 30 mM NaCl, pH 7.0) for 5 min each. Subsequently, the slides were treated with 4% formaldehyde in 2x SSC for 5 min, dehydrated through a graded ethanol series (70%, 90%, and 100%, 2 min each), and air-dried.

The telomere specific DNA probes that targeted the AAACCCG and AAACCG repeat was developed by PCR in the absence of template DNA [95,96]. The PCR products were purified using QIAquick PCR purification kit (Qiagen) and labeled with digoxigenin-11-dUTP (Roche) using a nick translation reaction. Digoxigenin-labeled probes were detected with a rhodamine-conjugated anti-digoxigenin antibody (Roche). Chromosomes were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) in Vectashield antifade solution (Vector Laboratories). The images were captured using a Zeiss Axioplan 2 microscope equipped with a cooled CCD camera CoolSNAP HQ2 (Photometrics) and AxioVision 4.8 software. The final contrast of the images was adjusted using Adobe Photoshop 2024 software.

Synteny analysis

Synteny of the TR gene was determined through orthology of the coding sequence surrounding the TR. For each Mimulus species with a reference genome we took the chromosomal position of the TR and extracted the gene annotations that were up and downstream of the TR gene. For the focal species and its TR gene, orthology of the surrounding genes were determined through Orthofinder v2.5.5 [97,98] and using gene annotations of Mimulus species downloaded from Mimubase (http://mimubase.org/). In order to visualize the synteny, pyGenomeViz was used by following the tutorial (https://github.com/moshi4/pyGenomeViz).

Supporting information