SMRT Sequencing of Paramecium Bursaria Chlorella Virus-1 Reveals Diverse Methylation Stability in Adenines Targeted by Restriction Modification Systems

Abstract

Chloroviruses (family Phycodnaviridae) infect eukaryotic, freshwater, unicellular green algae. A unique feature of these viruses is an abundance of DNA methyltransferases, with isolates dedicating up to 4.5% of their protein coding potential to these genes. This diversity highlights just one of the long-standing values of the chlorovirus model system; where group-wide epigenomic characterization might begin to elucidate the function(s) of DNA methylation in large dsDNA viruses. We characterized DNA modifications in the prototype chlorovirus, PBCV-1, using single-molecule real time (SMRT) sequencing (aka PacBio). Results were compared to total available sites predicted in silico based on DNA sequence alone. SMRT-software detected N6-methyl-adenine (m6A) at GATC and CATG recognition sites, motifs previously shown to be targeted by PBCV-1 DNA methyltransferases M.CviAI and M. CviAII, respectively. At the same time, PacBio analyses indicated that 10.9% of the PBCV-1 genome had large interpulse duration ratio (ipdRatio) values, the primary metric for DNA modification identification. These events represent 20.6x more sites than can be accounted for by all available adenines in GATC and CATG motifs, suggesting base or backbone modifications other than methylation might be present. To define methylation stability, we cross-compared methylation status of each GATC and CATG sequence in three biological replicates and found ∼81% of sites were stably methylated, while ∼2% consistently lack methylation. The remaining 17% of sites were stochastically methylated. When methylation status was analyzed for both strands of each target, we show that palindromes existed in completely non-methylated states, fully-methylated states, or hemi-methylated states, though GATC sites more often lack methylation than CATG sequences. Given that both sequences are targeted by not just methyltransferases, but by restriction endonucleases that are together encoded by PBCV-1 as virus-originating restriction modification (RM) systems, there is strong selective pressure to modify all target sites. The finding that most instances of non-methylation are associated with hemi-methylation is congruent with observations that hemi-methylated palindromes are resistant to cleavage by restriction endonucleases. However, sites where hemi-methylation is conserved might represent a unique regulatory function for PBCV-1. This study serves as a baseline for future investigation into the epigenomics of chloroviruses and their giant virus relatives.

Keywords: DNA methylation; NCLDV; PBCV-1; algal-virus; chloroviruses; hemimethylation; restriction modification.

FIGURE 1

PBCV-1 motif enrichment and depletion determined in the context of local codon flexibility at different genomic increments. The PBCV-1 genome was analyzed for motif frequency of CATG (A) and GATC (B) tetramers. Panels show increasing window sizes, ranging from 4 to 40 kb moving from the outer to inner ring. Each ring increases incrementally by 4 kb, but the step size is always 4 kb. The Z-score is represented with color; red indicates more enriched scores and blue indicates more depleted scores. Green represents the expected frequency with no depletion or enrichment. Since the DistAMo software was created for circular chromosomes, there is artificial data coded across the termini (at 12 o’clock) for the linear PBCV-1 genome. This region has been shaded over to discourage faulty interpretation. Genes with significant Z-scores (greater than the absolute value of 2) are annotated by color (blue is enriched and red is depleted) and are included in Supplementary Table S2. Finally, the polycistronic region containing the viral tRNAs are not annotated in the DistAMo graphs since they are not annotated as genes at NCBI, though they are known to lie at the center of the genome between annotated genes A326L and a331L.

FIGURE 2

DNA modification induced kinetic events encountered during SMRT sequencing of one PBCV1 replicate (PBCV1-1C, Supplementary Figure S1). Starting from the outermost ring and going inward: (1) ipdRatio for each nucleotide on the forward strand, (2) genomic positions for PBCV-1 (kbp), 3) PBCV-1 potential protein coding sequences and tRNA encoding sequences (Dunigan et al., 2012), (4) CATG sites, (5) GATC sites, and (6) ipdRatio for each nucleotide on the reverse strand. The color coding of the ipdRatio is artificial to denote peak values, with a red peak denoting a value > 2. Black peaks indicate nucleotide kinetics similar to a non-modified base (ipdRatio ∼1), and blue peaks indicate a value < 0.5. This plot was made in CIRCOS (Krzywinski et al., 2009) with the intent of displaying ipdRatio peak height occurrence and diversity. The sequenced replicates exhibited CIRCOS plots with nearly imperceptible differences, which is why only one representative is shown here.

FIGURE 3

ipdRatio score for nucleotides in one replicate of the PBCV-1 genome at 30-fold read recruitment coverage. Dot color denotes association with a motif detected by motifMaker.sh. There are nearly imperceptible differences between the three replicates, which is why only one is shown here. The x-axis indicates genome position of each motif sequence.

FIGURE 4

Methylation stability of GATC and CATG tetramers among PBCV-1 replicates, analyzed at 30-fold read recruitment coverage. Each dot represents the average number of reads that were methylated (x-axis) and the deviation between three biological replicates (y-axis). An average value approaching one indicates stable methylation, meaning the site was methylated in 100% of the 30 reads. Position along the y-axis indicates stability of the methylation status, as a value of zero indicates the methylation status is consistent between the three replicates. Coordinates were used to easily bin methylation status and frequency as stably non-methylated (Q1 – 73 events, 21.9% CATG, 57% GATC); complete methylation in only one of the three replicates (Q2 – 457 events, 23.1% CATG, 76.9% GATC), complete methylation in only two of the three replicates (Q3 – 143 events, 24.3% CATG, 75.7% GATC), and stable complete methylation in all replicates (Q4 – 2,825 events, 58.1% CATG, 41.9% GATC). Histogram plots provide an estimate of how many events occur with the given coordinates.

FIGURE 5

Methylation status of GATC and CATG palindromes across three sequenced replicates of PBCV-1. The status for the forward and reverse strand of a single palindrome are plotted for the separated motifs without variation information. Histogram plots provide an estimate of how many events occur with the given coordinates, with lines marking thresholds for defining complete methylation, hemimethylation, and stochastic methylation. A threshold of methylation in 75% of reads is typically used by PacBio for positive identification of methylation.

FIGURE 6

Methylation averages of reads aligning to the forward or reverse strand of palindromes falling within viral genes encoding capsid proteins. The major capsid protein is encoded by A430L.