The plant materials used in this study ( Table 1 ) were selected to explore primer amplification across across six angiosperm orders and to investigate whether intron amplification products could, at least, distinguish congener species of agricultural importance within these orders. These included two commercial Raphanus sativus mitotypes confirmed by PCR markers as described by Kim et al. (2007), representatives of the two major Phaseolus vulgaris gene pools (Bhakta et al., 2017), congener species representatives of Cenchrus, Citrus, Cynodon, Solanum and Vaccinium, along with Poncirus trifoliata, which is cross-compatible with Citrus species (Moreira et al., 2002) and considered by some to fall within the genus Citrus (Ollitrault et al., 2020). Citrus materials were from the University of Florida Citrus Research and Education Center, Lake Alfred, Florida and Harris Citrus Nursery, Lithia, FL. The Cynodon entries were from the USDA National Plant Germplasm System. The Phaseolus, Cenchrus, Solanum and Vaccinium entries were obtained from the University of Florida research programs of Dr. C.E. Vallejos, Dr. L. Sollenberger, Dr. C.E. Vallejos, and Dr. J. Olmstead, respectively. Total cellular DNA was extracted from leaf samples by a modification of the cetyl trimethylammonium bromide (CTAB) method in which 50 mg of tissue was combined with 750 μl of CTAB buffer (Murray et al., 1980) and 10 μg of DNase-free RNase A in a FastPrep™ Lysing Matrix A tube, disrupted for 40 s in a FastPrep^®-24 Instrument (MP Biomedicals LLC, Santa Ana, CA) and incubated at 65°C for 5 min. Cellular and lysing matrix debris was removed by centrifugation at 13,000 xg for 10 min at room temperature. Supernatants were extracted with an equal volume of chloroform-isoamyl alcohol mixed in a ratio of 24:1. DNA was precipitated from the aqueous phase by the addition of a 2/3 volume of isopropyl alcohol and recovered by centrifugation at 13,000 xg for 10 min at room temperature. The pellets were washed in 750 μl of 70% ethanol, air dried and rehydrated in 80 ul of 1 mM Trizma base, 0.1 mM di-sodium ethylene diamine tetra acetic acid (Na₂EDTA), 1 mM NaCl, pH 8. The concentration of DNA samples was determined from the absorbance at 260 nm.

The PCR primers used in this work ( Table 2 ; Grosser, 2011) were designed against introns of the mitochondrial nad1, nad2, nad4, nad5, nad7 and cyctochrome c maturation Fc (ccmFc) genes because these introns were common to seven plant species’ mitochondrial genomes: A. thaliana (Unseld et al., 1997), B. napus (Handa, 2003), B. vulgaris (Kubo et al., 2000), N. tabacum (Sugiyama et al., 2005), O. sativa (Notsu et al., 2002), T. aestivum (Ogihara et al., 2005), and Z. mays (Allen et al., 2007). The National Center for Biotechnology Information (NCBI) accession numbers for these genomes are NC_001284, NC_002511, NC_008285, NC_006581, NC_007886, NC_007579, and NC_007982, respectively (https://www.ncbi.nlm.nih.gov/genome/organelle/, accessed 1/20/2023). Primer pairs were designed manually based upon the highly conserved coding regions flanking intron sequences or, in some cases, from conserved sequences within introns.

PCR amplification reactions were performed on replicate DNA preparations made from different plants of each entry, with the exception of the two Cynodon entries. For these only a single pot culture was available, so replicate DNA extractions were prepared from the single culture of each. PCR reactions of 50 μl contained 25-100 ng of DNA, 0.2 µM of each primer, 0.125 mM dNTPs, 1.25 units of high fidelity, TAKARA EXTAQ Hot Start DNA polymerase (Clontech, Mountain View, CA) in 1X TAKARA Hot Start reaction buffer. This high-fidelity polymerase was selected due to the length of the amplified introns and the intent to sequence PCR products. Amplification was for 30 cycles of 1 min at 94°C, 2 min at 55°C, and 3 min at 72°C. Electrophoresis through 1% agarose gels was performed to survey PCR reactions for successful amplification. The DNA Hyperladder II (Bioline Inc., Cambridge, MA) was used as a size marker. Electrophoresis was at 100V for 100 min in Tris-Borate-EDTA (TBE) buffer (10 mM Trizma base, 10 mM boric acid, 2.5 mM Na₂EDTA, pH 8.2). Gels were stained in 0.5 μg/ml ethidium bromide for 20 min and viewed over a UV transilluminator in a Molecular Imager^® Gel Doc™ XR System (Bio Rad Laboratories, Inc. Hercules, CA). Gel images were captured with the Quantity One^® 1-D Analysis Software (Bio Rad Laboratories, Inc.) and exported as.tif files. The AdvanCE™ FS96 capillary electrophoresis system (Advanced Analytical Technologies Inc., Ames, IA) was used to estimate the length of PCR amplification products in DNA base pairs (bp). Amplification products were diluted 1:15 in TE buffer (10 mM Trizma Base 1 mM Na₂EDTA, pH 8) and fractionated by use of the DNF-915 dsDNA 915 Reagent Kit (Advanced Analytical Technologies Inc.) according to the supplier’s instructions. Indel polymorphisms were confirmed by electrophoresis of DNA amplification products, individually and mixed, through Criterion™ precast 5% polyacrylamide gels (Bio-Rad Laboratories Inc., Hercules, CA) run in TBE buffer for 740 Volt-h and imaged as described above.

The amplification products of ccmFci1, nad5i4, nad7i1were purified for DNA sequencing through use of the QIAquick PCR Purification Kit (Qiagen Inc., Valencia, CA) according to supplier’s instructions. Purified amplification products were fully sequenced in both directions by the University of Florida Interdisciplinary Center for Biotechnology Research (ICBR) Sanger Sequencing Core Laboratory in Gainesville, FL or by Eurofins USA. Intron sequences and their corresponding GenBank Accession numbers are listed in Table 1 . The sequences were aligned on the MultAlin web server (Corpet, 1988) (http://multalin.toulouse.inra.fr/multalin/, accessed 9/2/2022). Nucleotide substitutions per site (K₀) were calculated by the formula of Kimura (1980) based upon pairwise alignments of sequences with all indels removed. Indels per site (I) were calculated as the number of indels in a pairwise alignment divided by the number of nucleotides in the alignment with indels removed (Laroche et al., 1997). Intron sequences found to differ between congener species were also analyzed for potential restriction fragment polymorphisms with the NEB cutter V 2.0 tool (Vincze et al., 2003) (http://nc2.neb.com/NEBcutter2/index.php, accessed 1/26/2023).

Prediction of intron amplification products across a wider range of plant taxa was performed through application of the Primer-BLAST tool (Ye et al., 2012) (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) to selected plant mitochondrial genomes in the NCBI organelle genome database (https://www.ncbi.nlm.nih.gov/genome/organelle/) (both accessed 1/20/2023). Genomes queried included early andiosperms Magnolia biondii (NC_049134.1) (Dong et al., 2020) and Magnolia officinalis (NC_064401) (unpublished), which could potentially differ in sequence from later diverged andiosperms. Additional orders of monocots were selected to complement the single order (Poales) investigated experimentally. These included Allium cepa male-sterilizing (KU318712.1) (Kim et al., 2016) and normal (AP018390.1) (Tsujimura and Terachi, 2018) cytoplasms representing monocot order Asparagales; Cocos nucifera (KX028885.1) (Aljohi et al., 2016) representing monocot order Arecacales; and Zostera japonica (NC_068803.1) (Chen et al., 2022) and Zostera marina (KX808392.1)(Petersen et al., 2017) representing monocot order Alismatales. Also included were dicots Silene conica (JF40490.1-JF50629.1), Silene noctiflora (KP053825.1-KP053880.1), Silene latifolia (HM562727.1) and Silene vulgaris (JF750427.1-JF750430.1). Silene is an important model genus that includes species exhibiting unusual patterns of mitochondrial genome expansion and nucleotide substitution, which potentially affect primer performance and utility. Silene conica and Silene noctiflora provide tests of primers on expanded mitochondrial genomes that exhibit accelerated nucleotide substitution rates in comparison to Silene latifolia and Silene vulgaris (Sloan et al., 2012).

The intron primer sets ( Table 2 ) successfully amplified the target intron in each of the 19 entries investigated ( Figure 1 , Table 3 ). PCR reactions generally produced a single major product, although additional products of low abundance were detected for some nad2i1, nad5i4 and nad7i3 amplifications ( Figure 1 ).

The increasing number of complete plant mitochondrial genome sequences enabled investigation of the potential for these primer sets to amplify target introns in additional taxa. Primer-BLAST analysis of selected fully sequenced mitochondrial genomes predicted successful application of the introns in early angiosperms represented by Magnolia biondii and Magnolia officionalis; additional orders of monocots represented by Allium cepa, Cocos nucifera, Zoster japonica and Zoster marina; and an additional order of dicots represented by Silene conica, Silene latifolia, Silene noctifolora and Silene vulgaris ( Table 4 ). Of the 121 primer-accession combinations tested, 79 predicted a single amplification product produced by perfectly matched primers. An additional 18 combinations predicted a single amplification product produced by primers with only one or two mis-matched nucleotides between the target genome and primer set. The 11 primer sets are therefore predicted to be useful for the amplification of mitochondrial introns across the angiosperms. Primers were predicted to be less effective for plant mitochondrial genomes that exhibit exceptionally high rates of genome expansion and nucleotide substitution. Silene conica and Silene noctiflora represent expanded mitochondrial genomes with accelerated nucleotide substitution rates in comparison to Silene latifolia and Silene vulgaris (Sloan et al., 2012). While all primer sets were predicted to amplify single products in Silene latifolia and Silene vulgaris, most primer sets predicted multiple, weak matches to Silene conica and Silene noctiflora. Nevertheless, 3-4 primer sets were still predicted to work well for these two templates ( Table 4 ).

The fractionation of experimentally produced intron amplification products by gel electrophoresis ( Figure 1 ) and AdvanCE™ FS96 capillary electrophoresis ( Table 3 ) demonstrated significant intron length variation among diverse angiosperm genera, in agreement with primer-BLAST observations ( Table 4 ). Intron lengths, as estimated by the AdvanCE™ capillary technique, varied across genera by as few as 40 nucleotides in the case of nad5i1 to as many as 539 nucleotides in the case of nad7i2. This was excluding the extreme size (4284 nucleotides) of Phaseolus ccmFci1, which likely reflects a split intron. Length polymorphisms between congener species were, however, few in number and challenging to detect by electrophoresis. The well-to-well variation of the AdvanCE™ FS96 precluded use of length values to detect small indel polymorphisms in relatively large DNA amplification products. Length polymorphisms were identified by fractionation of amplification products on polyacrylamide gels ( Figure 1 ) and confirmed by acrylamide gel electrophoresis of PCR product mixtures ( Figure 2 ) for congeners of Cenchrus (ccmFci1 and nad2i4), Cynodon (nad7i2) and Citrus (nad7i1 and nad7i2) species. Intron length polymorphisms are summarized in Table 5 . The three Citrus maternal lineages and C. japonica were individually distinguished by the combination of nad7i1 and nad7i2 polymorphisms. C. paradisi (grapefruit) and C. sinensis (orange) were not distinguished from their respective C. maxima and C. reticulata maternal lineages. Citrus species were distinguished from P. trifoliata by length polymorphisms in ccmFci1, nad2i1, nad7i1, and nad7i2. Electrophoresis did not, however, distinguish the introns of the two Vaccinium or Solanum species, Phaseolus gene pools, or Raphanus sativus mitotypes.

Primer-BLAST demonstrated that short length polymorphisms often distinguish congener species’ mitochondrial introns ( Table 4 ). In Allium, introns differing by 8, 10 and 28 nucleotides distinguished the male sterilizing cytoplasm, derived by interspecific introgression (Manjunathagowda et al., 2021), from the normal cytoplasm. Magnolia biondii and Magnolia officinalis varied in seven introns with length differences ranging from 6 to 46 nucleotides. Silene vulgaris differed from Silene latifolia in nine introns having length variations ranging from 4-24 nucleotides. Zoster japonica and Zoster marina were polymorphic with respect to length in eight introns. While five of these differences ranged from 3-61 nucleotides, length polymorphisms of 316, 320 and 276 nucleotides were predicted for ccmFci1, nad2i4 and nad7i3, respectively. DNA sequence information clearly allows detection of mitochondrial intron length polymorphisms that distinguish related plant species.

CcmFci1, nad5i4 and nad7i1 introns amplified from 16 entries were sequenced to further characterize indels detected by electrophoresis and to search for additional indels, along with SNPs ( Figures S1 , S2 , and S3 , respectively). Citrus sinensis (sweet orange with the C. reticulata maternal lineage) was not included, and heteroplasmy or seed mixtures in the two commercial Raphanus sativus accessions precluded obtaining quality sequences for comparison of these two mitotypes within this species. With respect to intra-species variation, nad5i4 and nad7i1 sequences did not distinguish the two gene pools of Phaseolus vulgaris. The Phaseolus ccmFci1 shared 632 5’ nucleotides and 133 3’ nucleotides with other species separated by a 3353 nucleotide insertion ( Figure S1B ). The two Phaseolus accessions were polymorphic for one SNP and a 4 base indel within the 3353 nucleotide insertion, but were not polymorphic with respect to the intron regions. Moreover, the three C. paradisi introns were not polymorphic with respect to those of their C. maxima maternal ancestor. Sequencing further characterized indels detected by gel electrophoresis and revealed additional length polymorphisms ( Table 5 ). The ccmFci1 length polymorphism differentiating P. trifoliata from Citrus entries was due to separate deletions of 8, 9, and 17 nucleotides in P. trifoliata compared to Citrus ( Figure S1A ). Similarly, the polymorphism in nad7i1 was caused by separate deletions of 9, 8, and 9 nucleotides in P. trifoliata relative to C. maxima, C. medica and C. japonica. C. reticulata shared the 8 nucleotide deletion with P. trifoliata, while C. medica carried a unique 8 nucleotide deletion ( Figure S3 ). The ccmFci1 sequence distinguishing Cenchrus congeners was a 4 nucleotide indel ( Figure S1A ). Additional length polymorphisms identified by sequencing included a 4 nucleotide nad5i4 indel that distinguished Cynodon congeners ( Figure S2 ) and a 4 nucleotide nad7i1 that distinguished Cenchrus congeners ( Figure S3 ). Sequencing did not reveal indel polymorphisms between Vaccinium or Solanum congeners.

Sequences of three introns identified only nine SNPs that distinguished congener species ( Table 6 ). The nad5i4 sequence alignment ( Figure S2 ) revealed a SNP that distinguished V. corymbosum from V. virgatum. This was the only Vaccinium polymorphism identified in this study. Three nad5i4 SNPs distinguished C. dactylon from C. transvaalensis ( Figure S2 ). In addition to the nad7i1 indels, a nad7i1 SNP was found to distinguish C. reticulata from other Citrus species ( Figure S3 ). Of the nine SNPs, only one was a C/T difference that could possibly be erased at the RNA level by plant mitochondrial C-to-T RNA editing. In these comparisons, the frequency of SNPs per site (K₀) within genera was low - zero in the case of ccmFci1. The average K₀ for nad5i4 and nad7i1 in congeneric species comparisons was 0.03 and 0.01, respectively, that of comparisons among dicot genera ( Table 7 ). The frequency of indels per site (I) within genera was also low, 0.02-0.10 of I for comparisons among dicot genera ( Table 7 ).

While sequence analysis is the most direct means of identifying length and SNP variation in amplified introns, these polymorphisms also create restriction pattern differences. Analysis of sequenced introns with NEB Cutter ( Table S1 ) associated unique restriction patterns with the variant alleles reported in Tables 5 and 6 . The only exception was the SNP that distinguished Vaccinium corymbosum and Vaccinium virgatum nad5i4 created no RFLPs across the 112 enzymes predicted by the NEB Cutter tool to cut these templates.

The 11 PCR primer sets used in this work demonstrated robust amplification of the target mitochondrial introns across 16 species representing eight plant genera and seven plant orders. Primer-BLAST analysis with these same primer sets predicted successful amplification of mitochondrial introns from early angiosperms and additional orders of monocots and dicots. This expands and improves the available universal primers for plant mitochondrial introns (Demesure et al., 1995; Dumolin-Lapegue et al., 1997; Duminil et al., 2002). Aleksić (2016) found limited applicability of previously developed universal mitochondrial primers to legume (Fabaceae) species and suggested family-specific primers as a more practical approach. The primer sets employed here successfully amplified the mitochondrial introns of P. vulgaris as a representative legume. Previous universal primer design strategies (Duminil et al., 2002; Froelicher et al., 2011) utilized mitochondrial sequences conserved between A. thaliana and B. vulgaris only. Primer design based on conserved introns and flanking sequences from seven plant species (Grosser, 2011) likely contributed to the extended applicability of the current primer sets. Although most plant species’ mitochondrial genomes evolve slowly with respect to coding sequences (Wolfe et al., 1987; Palmer and Herbon, 1988), plant genera containing taxa with widely varying rates of mitochondrial nucleotide substitution have been identified (Cho et al., 2004; Parkinson et al., 2005; Mower et al., 2007; Sloan et al., 2009). Primer-BLAST analysis did predict that some, but not all, of the 11 primer sets would work reliably on the Silene species having rapidly evolving mitochondrial coding sequences. A further complication with Silene conica and Silene noctiflora is that their highly expanded genomes apparently contain multiple, degenerate targets for the intron flanking primers ( Table 4 ).

Mitochondrial intron length polymorphisms detectable by electrophoretic techniques were frequently observed between genera, whereas comparisons within genera revealed primarily short intron length variations. Large indels are therefore tolerated within introns, but rates of such variation are low within genera. These contrasting observations likely reflect the evolutionary processes that shaped modern plant organellar group II introns from their self-splicing, progenitor introns. On the one hand, altered intron sequences combined with novel nuclear and organelle-encoded splicing factors to maintain competence for splicing while shifting away from the group II ribozymic, self-splicing structures (Bonen, 2008; Brown et al., 2014). At the same time, the requirement for splicing factors to evolve in concert with the intron structure likely constrained variants that can be successfully spliced (de Longevialle et al., 2010; Zimmerly and Semper, 2015). Plant organelle introns retain significant common structural features (Bonen, 2008). Moreover, they reside within genes essential to photosynthesis or respiration, creating selective pressure for the maintenance of efficient splicing (Brown et al., 2014; Zimmerly and Semper, 2015; Best et al., 2020). Arrays of protein factors are required for the splicing of plastid and mitochondrial introns. These include members of the maturase family, descended from the maturases encoded in ribozymic, self-splicing group II introns (Schmitz-Linneweber et al., 2015), along with APO, CRM, PORR, PPR and TERF families of RNA binding proteins. With the exception of one plastid and one mitochondria-encoded maturase, these proteins are encoded by the nuclear genome and imported into the organelles where they act in combinatorial fashion for the splicing of particular introns or groups of introns (de Longevialle et al., 2010; Brown et al., 2014; Zimmerly and Semper, 2015; Wang et al., 2022). The complexity and specificity of this process may explain the lack of large intron indels found within genera.

The plant mitochondrial intron length differences between congener species as predicted by by Primer-BLAST averaged 17 nucleotides, excepting the three Zoster introns with larger differences. The experimentally characterized indels that distinguished congeners or cross-compatible species averaged less than 10 nucleotides in length, necessitating high-resolution acrylamide gels or DNA sequencing for discernment. Gel-resolved intron length polymorphisms differentiated Citrus, Cenchrus, and Cynodon congeners, but intron sequencing provided a more accurate picture of indel polymorphisms. The nad7i1 amplicons of C. medica and C. reticulata, for example, carried different indels of the same length. Even when larger intron size differences were apparent within genera, sequencing revealed them to result from multiple short indels. This is consistent with prior reports that short indels (1-10 bp) comprise greater than 50% of indels in plant mitochondrial introns and probably originate from slipped strand mispairing events during replication (Laroche et al., 1997).

For each of the three introns sequenced in the present study, the average frequency of nucleotide substitutions per site (K₀) was also low within genera, but SNPs that distinguished congeners of Citrus, Cynodon, Cenchrus, and Vaccinium were identified. These can serve as useful markers through workflows such as cleaved amplified polymorphic sequence (CAPS), PCR combined with sequencing, or amplification refractory mutation analysis strategies (Lo, 1998; Ciarmiello et al., 2013). With one exception, SNPs and short indels that distinguished congeners’ introns also created CAPS markers ( Table S1 ). More broadly, K₀ values for comparisons between Citrus, Vaccinium and Solanum as representative dicot genera ( Table 7 ) were similar to those reported by Laroche et al. (1997) for comparisons of six mitochondrial introns between two to three dicot genera. In the present study, K₀ values for intron sequence comparisons between congeneric species were 0.01-0.03 times those for comparisons between dicot genera. The low nucleotide substitution rates likely result from the low frequency of nucleotide substitutions characteristic of most plant mitochondrial genomes, typically three to ten times lower than nuclear nucleotide substitution rates (Wolfe et al., 1987; Palmer and Herbon, 1988; Drouin et al., 2008).

The limited variation of mitochondrial introns within plant genera contrasts with the extensive diversity of nuclear introns, which show a high frequency of length and substitution polymorphisms within species of Oryza (Wang et al., 2005), Solanum (Wang et al., 2010), Allium (Jayaswal et al., 2019) and Medicago (Shilpa and Lohithaswa, 2021) among others. While no mitochondrial intron polymorphisms distinguished Solanum lycopersicum from S. pennellii, two studies document extensive nuclear intron polymorphisms within Solanum lycopersicum (Van Deynze et al., 2007; Wang et al., 2010). Organellar group II introns are considered the ancestors of nuclear introns, which lack folding constraints because they share the use of spliceosomal RNAs that have taken on the functions of the group II intron domains (Sharp, 1991). The spliceosome is a complex that is highly malleable in order to accommodate diverse exon ends and alternative splicing, perhaps permitting more varied intron sequences (Chen and Moore, 2014).

When present, organelle intron polymorphisms have valuable applications for determining inheritance in sexual crosses or somatic hybridizations. The markers investigated in this study have proved useful for determining organelle inheritance in Citrus cybrids. Cybrids are produced spontaneously as a by-product of protoplast fusion and are characterized by the diploid nuclear genome of the mesophyll fusion partner, the mitochondrial genome of the embryogenic callus partner, and random inheritance of chloroplast DNA (Grosser et al., 1996; Cabasson et al., 2001; Guo et al., 2004; Guo et al., 2013). This contrasts with typical protoplast fusion products, which possess tetraploid nuclei inherited from both parents. Cybrids provide a means to quickly create novel combinations of nuclear and organellar genotypes and to evaluate their phenotypic consequences. Specific organelle genotypes are associated with beneficial traits in cybrids. For example, grapefruit cybrids with mandarin mitochondrial DNA exhibit an extended season of high-quality fruit (Satpute et al., 2015), whereas grapefruit cybrids with kumquat plastid DNA exhibit increased resistance to citrus canker regardless of mitochondrial origin (Omar et al., 2017). The nad7i1 and nad7i2 primer sets ( Table 2 ) were utilized, respectively, for verification and characterization of mitochondrial DNA inheritance in these two sets of cybrids, illustrating application for mitochondrial intron markers.

The currently reported nad7i1 marker overlaps with and confirms one of the three markers that Froelicher et al. (2011) demonstrated to be polymorphic in Citrus. Because our primer set amplified the entire intron, a new SNP was added to the previously published indels. Moreover, the list of intron markers polymorphic for Citrus species was expanded to include nad7i2, ccmFc, and nad2i1. The additional polymorphic markers did not, however, further distinguish differences within the seven citrus mitotypes identified by Froelicher et al. (2011). For example, C. maxima and its maternal derivative C. paradisi remained indistinguishable for all introns sequenced in this study.

Due to the lack of conserved gene order among plant mitochondrial genomes, often even between closely related taxa, assembling plant mitochondrial genome sequences presents special challenges and can preclude the universal application of intergenic sequences for distinguishing between closely related groups (Duminil and Besnard, 2021). Mitochondrial intron markers have demonstrated applicability in studies of population genetics, genotype characterization, detection of past hybridizations, and biogeographic studies of gene pool distributions (Ciarmiello et al., 2013; Aizawa et al., 2014; Xiang et al., 2014; Kersten et al., 2015). The current study documents a widely applicable set of primers for the mitochondrial marker toolbox and provides insights into the conservation and variation of plant mitochondrial introns.