The alternative reality of plant mitochondrial DNA: One ring does not rule them all

Abstract

Plant mitochondrial genomes are usually assembled and displayed as circular maps based on the widely-held view across the broad community of life scientists that circular genome-sized molecules are the primary form of plant mitochondrial DNA, despite the understanding by plant mitochondrial researchers that this is an inaccurate and outdated concept. Many plant mitochondrial genomes have one or more pairs of large repeats that can act as sites for inter- or intramolecular recombination, leading to multiple alternative arrangements (isoforms). Most mitochondrial genomes have been assembled using methods unable to capture the complete spectrum of isoforms within a species, leading to an incomplete inference of their structure and recombinational activity. To document and investigate underlying reasons for structural diversity in plant mitochondrial DNA, we used long-read (PacBio) and short-read (Illumina) sequencing data to assemble and compare mitochondrial genomes of domesticated (Lactuca sativa) and wild (L. saligna and L. serriola) lettuce species. We characterized a comprehensive, complex set of isoforms within each species and compared genome structures between species. Physical analysis of L. sativa mtDNA molecules by fluorescence microscopy revealed a variety of linear, branched, and circular structures. The mitochondrial genomes for L. sativa and L. serriola were identical in sequence and arrangement and differed substantially from L. saligna, indicating that the mitochondrial genome structure did not change during domestication. From the isoforms in our data, we infer that recombination occurs at repeats of all sizes at variable frequencies. The differences in genome structure between L. saligna and the two other Lactuca species can be largely explained by rare recombination events that rearranged the structure. Our data demonstrate that representations of plant mitochondrial genomes as simple, circular molecules are not accurate descriptions of their true nature and that in reality plant mitochondrial DNA is a complex, dynamic mixture of forms.

Fig 1. Primary structural units of the Lactuca mitochondrial genome and secondary building blocks.

A) Comparison of primary structural units between the L. sativa and L. saligna mitochondrial genomes. Polished contigs generated from CLC assembly form the primary structural units of the mitochondrial genome and are shown in pairs with L. sativa on top and L. saligna on the bottom. The stoichiometry (1x or 2x) and whether the units were common or unique between L. sativa and L. saligna are indicated. Different termini of a pair of contigs are labeled with an asterisk. The differences between termini can be up to several hundred nucleotides. A dark green bar indicates a large insertion from the chloroplast genome. The contig from L. saligna, UV, was a single contig that correspond to major parts of contigs U and W of L. sativa and was conceptually split into two segments (color coding) to simplify and clarify the visualization and interpretation of contig relationships. A segment of the S inverted repeat sequence is duplicated in both species on the termini of units W and V, and in unit N of L. sativa (yellow arrows). B-F) Secondary building blocks of mitochondrial genomes with a defined sequential order of primary units determined by reverse read mapping (see Materials and methods) for L. sativa (B-D) and L. saligna (E,F). Junctions within boxes were quantified using PacBio reads and the counts are displayed as numbers above the boxes.

Fig 2. Isoforms of Lactuca mitochondrial genomes.

Putative major isoforms of L. sativa (A) and L. saligna (B) were derived from analysis of secondary building blocks (Fig 1B–1F) and the primary structural unit stoichiometry (Fig 1A and S2 Fig). For L. sativa, the dashed green arrow in A shows rearrangement of unit U and the dashed yellow arrow shows rearrangement of unit P between the two major isoforms. For L. saligna, putative recombination events that result in a transition from one isoform to another are indicated by solid blue arrows. Isoforms are labeled by the configuration of the units M, S, and Z, with > and < symbols designating where recombination has taken place. Yellow triangles indicate repeat X-01. Displayed isoforms are simplified models that fit the sequential order of primary structural units and stoichiometry data and make no assumptions about the underlying form of mtDNA molecules.

Fig 3. Lactuca mitochondrial genome annotations.

The annotations for genes and other sequence features for L. sativa (A) and L. saligna (B) are displayed along the primary structural units (see Fig 1A) for each genome, which are indicated by thick gray arrows. Intronless genes are indicated by red arrows, exons of spliced genes are indicated by blue arrows, ribosomal RNA genes by white arrows, and plastid insertions are indicated by green arrows. Genes that span junctions between primary units are indicated by a jagged line at the division point. Thin gray arrows show alternative junctions between primary structural units that result in different models for the genes that are split over a junction.

Fig 4. Long distance analysis of mitochondrial genomes using Illumina mate pair and Hi-C libraries.

Plots of distances between read-pairs Illumina mate pair libraries with 2.5 and 10 kb insert sizes in 1 kb bins for L. sativa (A) and L. serriola (B) exhibit essentially identical long-distance sequence connections. C) Plot of Hi-C contact frequencies for L. sativa. Major isoform α was arbitrarily chosen as a reference for visualization of mate-pair distances and Hi-C contact frequency. The color gradient displayed below panels reflects the number of read-pairs (out of a total of 320,000) in each 2-dimensional-1-kb bin and applies to all panels.

Fig 5. Detection of minor isoforms using Illumina mate pair data.

Above, schematic illustrations of the transition between major isoform α and two minor isoforms through recombination at repeat X-01b (yellow triangles and blue arrows). Below, high-resolution-Illumina-mate-pair plot showing sequence connections (indicated by gray ovals) that are expected for the minor isoforms shown above. Mate-pair library sizes were 2.5 and 10 kb. The major mitochondrial isoform of L. sativa includes the following order of basic units: M-N-K-M-N-W-Z. There is a medium size repeat X-01b of length ~500 bp (yellow triangle) at the N and W termini that can cause two distinct recombination events as shown on the right side of the figure. Upon X-01b recombination, two new isoforms have three distinct junctions M-W (common for both), W-N-Z, and K-N-Z (specific for each isoform). A long distance plot of mate-pair libraries clearly demonstrated the existence of both isoforms and all three junctions. Ovals highlight the diagonals of long distance interactions that were evidence of the existence of minor isoforms for the L. sativa mitochondrial genome. Similar patterns were detected using L. serriola mate pair libraries (Fig 4B and S5C Fig). The color gradient displayed below panels reflects the total number of read-pairs (out of a total of 320,000) in each 2-dimensional-1-kb bin and applies to all panels.

Fig 6. Interconversion between L. sativa and L. saligna genomes.

Linear representation of one of the L. saligna major mitochondrial genome isoforms (GenBank accession MK759657) on the top. The middle structure is a putative minor isoform of L. saligna derived from recombination at repeat X-07 (black triangles), which leads to a similar arrangement of primary structural units as that of L. sativa major isoform α at bottom (GenBank accession MK642355). Collinear segments (longer than 500 bp) are indicated by gray shading.

Fig 7. Structural analysis of L. sativa mitochondrial DNA by in-gel fluorescence microscopy and pulsed-field gel electrophoresis.

(A-F) DNA obtained from mitochondria isolated from seven-day-old dark-grown seedlings (roots and shoots) was stained with either ethidium bromide or QuantiFluor dye. Images are representative of branched linear (A), circular (B), linear (C), degraded (D), comet (E), and branched circular (F) structures. The scale bar in panel F applies to panels A-F and is 10 μm, corresponding to the length of approximately 30 kb of DNA. Branched linear: interconnected linear forms with or without a densely staining central core. Circular: closed loop without any additional branches. Linear: linear fiber with no branches. Degraded: many small molecules of undetermined structure. Comet: bright core with short connected fibers and no other visible branch points. Branched Circular: closed loop structure with linear branches. G) Quantification of the primary structures observed among 98 total microscopic fields. H) Size distribution of the circular molecules based on length measurements of their circumferences. I) Pulsed-field gel electrophoresis. Each lane represents an independent gel run. Sizes (in kb) are determined from the migration of lambda concatemers run in each gel experiment. In some cases, the DNA migrated at a slight angle to the left.