In this thesis, I studied mtDNA replication in mung bean cotyledons to understand the association with the dynamics and changes of physical
structure/conformation during cotyledon development under specific experimental conditions. Combining flow cytometry, PFGE, EM, real-time PCR and biochemical analyses, the results support the following conclusions. (1) MtDNA replication in cotyledon proceeds by a multiple recombination-dependent initiation process generating an organizationally complex physical structure that mediates
recombination and replication. (2) The “rosette core” might contain highly condensed mtDNA with a replication origin and plays an initial and central role in mtDNA replication. (3) A freeze-thaw treatment on mtDNA opened up the unusually large and complex physical structure observed during the active phase of mtDNA replication to reveal a collection of conventional recombination markers, which further confirmed its functional role as an RDR intermediate. (4) Short repeats are actively involved in mtDNA recombination-dependent initiation and likely also in formation of
recombination-mediated structures. Our findings support the earlier interpretation that RDR is the principal mechanism for plant mtDNA replication (Bendich, 1996;
Oldenburg & Bendich, 1996; Backert et al., 1997a; Backert & Borner, 2000;
Oldenburg & Bendich, 2001) and provide certain new insights into plant mtDNA replication and its association with the changes of structure complexity during plant development.
Throughout mung bean cotyledon development, a large fraction of mtDNA migrated as a smear without bands into the PFGE gel with size equivalent to approximately 50 to 200 kb linear DNA markers; most of the remaining mtDNA remained in the well, with a minor fraction falling in between the fast-moving (fm)
and well-bound (wb) zones. The relative amount of fm and wb forms varied significantly during cotyledon development (Fig. 2). This result agrees with earlier findings indicating that plant mtDNA consists of a mixture of monomers and head-to-tail concatermers of circular and linear molecules together with highly complex branched structures (Bendich, 1993; Backert et al., 1996; Bendich, 1996;
Oldenburg & Bendich, 1996; Backert & Borner, 2000; Oldenburg & Bendich, 2001).
The mobility of the fm form of the 27 ºC-12 hr cotyledon mtDNA was significantly retarded and the degree of this retardation was augmented further in day-1 and -2 samples so that most of the fm constituents moved slower than did the 200-kb linear marker (Figs 2B-a,b, lanes 5 -7). A proteinase pre-treatment regressed this retardation in day-1 samples (Fig. 2B-a,b, lane 8). These results suggest that a large fraction of the 27 ºC-12 hr, day-1 and -2 cotyledon mtDNA contains a proteinase K-sensitive structure that migrated slower than the 50- to 200-kb linear DNA marker in PFGE (Fig. 2). Protein(s) associated with this fast-moving component are likely involved in the DNA replication process or in mitochondrial gene function.
To slow the development process, our germination condition included a
prolonged low temperature (4 ºC) imbibition at the onset to enhance the synchrony for capturing the detailed mtDNA replication with accompanying changes to its physical structure. At this time point (4 ºC-12 hr), EM data indicated that 93.5% of total mtDNA length was taken up by linear plus branching linear molecules, with 5.5% by simple rosette plus simple reticular rosette-like structures (Table 2). Also, many long linear molecules (150~210 kb; most > 50 kb) was found (Fig. 7D). Flow cytometry also revealed a very low amount of DAPI-stained DNA present in individual mitochondria at this time point (Fig. 1A-a). Throughout all cotyledon development stages, the relative amount of circular molecules was low. The highest value, 2.2%,
with size < 3 kb was found in the 27 ºC-12 hr sample, with mtDNA synthesis at the onset of its most active phase (Table 2 and Fig. 5).
At the early cotyledon developmental stage, 4 ºC-12 hr, P32-dCTP was
incorporated into the fm form, with very little incorporation into the wb form during a short 60-min pulse under our in organello experimental conditions. This finding is in contrast to that detected for the 27 ºC-12 hr sample, in which the wb fraction was clearly labeled. The two intervening samples, 4 ºC-12 hr + 27 ºC-2 hr and 4 ºC-12 hr + 27 ºC-6 hr, exhibited a gradual increase in wb labeling (Fig. 5a). In day-1 and -2 samples, considerable more P32-dCTP labeling was found in the newly synthesized fm than wb fraction (Fig. 5b). During early cotyledon development stages, the
newly-synthesized fm DNA predominated over the wb component, which constituted only a minor fraction. This situation lasted until the onset of the most active synthesis phase at 27 °C-12 hr, when the newly-synthesized wb fraction became sizable. Thus, formation of highly complex mtDNA (wb form) was closely related to high mtDNA synthesis activity. The quantity of mtDNA per mitochondrion was also coordinately increased and decreased with mtDNA synthesis activity (Figs. 1 and 5). This high fluctuation and exceedingly dynamic characteristics of the organelle DNA content and complexity was found previously (Backert & Borner, 2000; Bendich, 2004;
Oldenburg & Bendich, 2004; Bogenhagen, 2010).
As seen by EM, all mtDNA samples analyzed during mung bean cotyledon development invariably contained mixtures of different structures and sizes (linear, branching linear, rosette, reticular rosette and a few circular, of < 12 kb, see Table 2).
The summation of each samples' composition in terms of percentage of overall length of each structure type over the total sum of length was measured for all types
regardless of the size of each individual molecule (Table 2). Giant rosettes > 500 kb
were scarce in number: 8, 10 and 2 found in 27 ºC-12 hr, day-1 and day-2 cotyledon mtDNA, respectively, on EM. The data for giant rosettes were not included in Table 2.
The scarce amount of giant rosettes may not represent what exists in vivo, because the structure of such complexes and huge mtDNA may easily change conformation and integration during mitochondria isolation and mtDNA purification. An image from each time was chosen to emphasize the representativeness and uniqueness (Figs. 7 and 8).
Our findings øf a rosette-like structure during mtDNA replication agrees well with that shown previously in Chenopodium album (Backert & Borner, 2000). From the architecture and timing of the appearance of regular rosette structure, including giant rosettes, corresponds well with the multi-fibered, comet-like forms of ethidium- bromide-stained mtDNA seen in moving pictures during electrophoresis of several plants (Bendich, 1996). Our results are consistent with and support the earlier interpretations that those large complex structures with size often greater than the genome size (presented in Figs. 9 to 11) represent the RDR intermediates and their formation has been attributed to multiple 3' overhang invasion of homologous regions spanning the repeats to initiate mtDNA replication. After multiple rounds of RDR, the newly replicated DNA results in a very complex branched network (Backert & Borner, 2000; Oldenburg & Bendich, 2001; Bendich, 2010; Gerhold et al., 2010; Gualberto et al., 2014).
The steepest fluctuation in structure composition was observed with the
transition from dry seed to the 4 ºC-2 hr sample, showing a decrease in proportion of linear + branching linear molecules from 85.8% to 8.8% and a corresponding increase in proportion of simple rosette + simple reticular rosette molecules from 13.6% to 91%. This fluctuation pattern closely resembled that observed for the transition from
4 ºC-12 hr to 4 ºC-12hr + 27 ºC-2 hr, in which linear + branching linear molecules decreased in proportion from 93% to 12% and simple rosette + simple reticular rosette molecules increased from 5% to 87.7%. This parallel trend of changing mtDNA composition (Table 2 and Fig. 7) suggesting prolonged low temperature imbibition may have rendered the imbibing seeds to a state somewhat resembling that of dry seeds in terms of the physical structure. The active mtDNA synthesis with
regular/giant rosettes but not simple rosettes is shown in Figures 5 and 7. These results clearly show that the linear and rosette mtDNA structures can be converted in vivo during early cotyledon development, which is affected by the temperature and
duration applied to the seed incubation.
The highly complex giant rosette mtDNA molecule in Figure 10A contains some unique sub-units. Several mtDNA rosettes, presumably replication units that are surrounded by a large circular DNA, are connected to each other. Individual rosette structures with similar (Fig. 10B-a) or various (Fig. 10B-b) complexities in their structures are linked to each other (core to core) by DNA junctions. Figure 10B-a presents three rosette structures with more similar structural complexity, which
indicates their similarity in terms of the re-initiation time (see also box b of Fig. 10A).
The rosettes with different RDR re-initiation times may result in rosettes with various complexities (Fig. 10B-b). This result confirms that the connected rosette structures are part of one giant mitochondrial molecule (Fig. 10A). Putative replication forks are often found near the center of the rosette core and appear as components that extend out from the center core (Figs. 9 and 10). The stretched-out forks and layers of concentric, petal-like structures develop during mtDNA RDR. The satellite-rosette cores represent re-initiation sites of mtDNA replication that surround the parental replicating core and form satellite architecture.
In the freeze-thaw treatment study, the regular-sized rosette structure (~40-200 kb) lost its core composition and converted to linear DNA with a size that was much longer than genome sizes plus various sub-structure attachments (>2000 kb) (Fig. 8).
The core in the rosette-like structure contained condensed plastid DNA, which was clearly demonstrated by Oldenburg and Bendich (2004). These exceedingly long DNA molecules may be expanded from the relaxation of rosette core, which probably contain the whole or partial mitochondrial genome and replication origin. The center core (arrow in Figs. 8 and 9) might be the initial site for mtDNA replication, and the satellite-cores (arrowheads in Fig. 9) might be the subsequent re-initiation mtDNA replication sites on the same mtDNA molecule when the original mtDNA replication process is not yet complete. The continuous serial re-initiation of mtDNA replication on the same molecule may result in a complex branched structure and expand a single mtDNA into many genome-equivalent mtDNAs and form a giant complex mtDNA compartment (Figs. 9-11). These recombination replication intermediates that contain putative replication forks, D-loops and Holliday junctions all indicate that RDR occurs in mung bean cotyledon mitochondria. Because freeze-thaw treatment is effective in opening up the tightly-folded rosette structures, single-stranded DNA regions, with or without associated protein(s), may be involved in holding the tightly folded state of the rosette structures.
After freeze-thaw treatment, the rosette structure converted to a much larger molecule; also an abundance of small networks appeared on the same DNA molecule (~0.5-2 kb for each entity, long dotted arrow in Fig. 11A-b). Moreover, some large square networks (~10-20 kb per square entity, open arrowhead in Figs. 11A-b, 11B) were also found. The large networks arranged in rows showed Holliday junctions (short dotted arrows). Interestingly, similar small reticular networks around the core
structure were present in the very early mtDNA replication stages, especially in the 27 °C-6 hr, 27 °C-12 hr mitochondria and, to a lesser extent, day-1 mitochondria (Fig.
12). The small reticular networks might be produced by short repeat DNA sequences and may be associated with the initiation of RDR. According to real-time PCR analysis (Fig. 15), short repeats in cotyledon mtDNA showed recombination activity in vitro, which supports that short repeats may be involved in RDR initiation. The
proposed progressive development of small networks around the core is shown in Figure 12. Small networks gradually convert to loose petal-like structures and then form the rosette-like structure. The RDR forks are formed during this progressive process (Fig. 12F). The small networks, which appear around the core in large numbers, may be related to the short repeat recombination in mung bean mtDNA for initiation of RDR.
Recombination occurs in shorter repeats, and the production of rearranged mtDNA molecules was reported previously (Andre et al., 1992; Shedge et al., 2007;
Woloszynska & Trojanowski, 2009; Cappadocia et al., 2010; Marechal & Brisson, 2010; Woloszynska, 2010; Sloan, 2013). Some short repeats in various plant
mitochondrial genomes were suggested as hot spots for recombination (Shedge et al., 2007; Cappadocia et al., 2010). Although the mung bean mitochondrial genome lacks the large repeat, a PCR survey of 6 short (38 to 175 bp) repeats revealed evidence for recombination across all of them (Alverson et al., 2011b). In agreement with these data, my findings also indicate that mung bean short repeats are active in
recombination (see Fig. 15). Moreover, the recombination activity detected by quantitative PCR did not vary throughout cotyledon development. Whether recombination activity of short repeats in vivo could be accurately inferred from real-time PCR results alone is questionable; a recombinant clone involving a 175-bp
mung bean short repeat recovered in a high-depth mung bean mitochondrial genome sequencing project indicates that at least some, and perhaps many, of the mung bean short repeats actually do recombine in vivo. Indeed, a number of similarly short repeats have been found in the mitochondrial genome of Arabidopsis (Shedge et al., 2007; Arrieta-Montiel et al., 2009) and Phaseolus (Woloszynska & Trojanowski, 2009; Woloszynska et al., 2012). The conclusion derived from these PCR-based studies were further substantiated by direct sequencing data. Nevertheless, the involvement of short repeat recombination in RDR initiation requires further
investigation. Both intramolecular and intermolecular recombination may be involved in recombination events (Lonsdale, 1984; Oldenburg & Bendich, 1996; D'Aurelio et al., 2004; Marechal & Brisson, 2010; Gualberto et al., 2014).
Nucleoid proteins binding with mtDNA are dynamic, and the composition changes during development. The progressive alteration regulated mtDNA replication and mitochondria encoding gene expression. In addition, it may result in DNA
released from membrane and degraded by nucleases (Oldenburg & Bendich, 2015).
On LC-MS/MS of PFGE of DNA–protein complexes at different cotyledon stages (Table 3A), some mitochondria membrane proteins always found in different stages included ATPase-, ATPase-, ANT and porin. Therefore, plant mtDNA also attaches on mitochondria membrane and binds with transmembrane proteins as for mtDNA in mammalian cells (Bogenhagen et al., 2003; Bogenhagen et al., 2008).
Hsp70 and chaperonin CPN60, which belongs to Hsp60, appear in almost all stages of the cotyledon mt-nucleoid. Recent studies indicated that Hsp60 interacts with the yeast mtDNA Ori sequence and is involved in nucleoid division determination and mtDNA maintenance (Kaufman et al., 2003; Chen & Butow, 2005; Kucej et al., 2008). Hsp70 also can be found in mammal mt‐nucleoids. Hsp70 has an improtant
role in mitochondrial protein import and protein-quality control, but whether Hsp70 has specific functions in nucleoid organization is unclear (Kucej & Butow, 2007). The cytoskeletal protein actin appear after seed germination stages. Similar reports are from both plant and mammalian mt-nucleoids. These reports suggest that actin directly contacts nucleoids in mitochondrial matrix and and may be involved in mtDNA segregation and mitochondrial division (Lo et al., 2011; Reyes et al., 2011;
Bogenhagen, 2012). Some metabolic proteins associate with all stages of cotyledon mt-nucleoids. Recent studies suggest that more metabolic proteins associate with mt-nucleoids with mitochondria with higher physiology activity and may be involved in organization and protection of mtDNA against oxidative damage during respiratory growth (Kucej & Butow, 2007).
MtSSB (single-stranded DNA-binding protein) is the marker protein in
mt-nucleoids. In my study, this protein existed in cotyledon mitochondria (Table 3B-b) but was not detected by proteomics analysis after PFGE analysis. It could be detected in 3-day-old seedling mt-nucleoids purified by sucrose gradient purification followed by LC-MS/MS analysis in our lab. Mitochondrial transcription factor A is the other essential protein in mt-nucleoids that directly binds with mtDNA in mammalian cells.
This protein is best characterized and abundant among nucleoid proteins (Kang et al., 2007; Bogenhagen, 2012). However, mitochondria in Arabidopsis, the T7
bacteriophage RNA polymerase (RNAP), as a single-polypeptide enzyme, performs all steps of transcription, including promoter recognition, initiation, and elongation.
RNAPs that localize to mitochondria (RpoTm) appears to be the enzyme that transcribes most mitochondrial genes (He et al., 2007). In this study, the T7 bacteriophage RNA polymerase (DNA-directed RNA polymerase) was found in mt-nucleoids of day-1 and day-2 cotyledons by proteomic analysis. Another problem
with the proteomics data was that in organello mt-DNA synthesis was driven by γ-like DNA polymerase in mung bean (Fig. 5). The γ-like DNA polymerase was identified in Arabidopsis (Sakai et al., 2004; Gualberto & Kuhn, 2014). No DNA polymerase homologs were found after analysis, which may due to the low recovery from the PFGE gel or peptid sequence differences resulting in the data not matching up in the database.
From the results, I derived a possible model of mtDNA RDR during cotyledon development (Fig. 16). A short linear fragment mtDNA is stored in dormant seed mitochondria. The DNA fragments form head-to-tail concatemers before seed germination and then develop into simple rosette or linear-branch structures with a core structure. RDR may then occur because of double-stranded homologous recombination breakage or double-stranded or single-stranded break repair mechanisms (Bendich, 1996; Marechal & Brisson, 2010). The center core in the rosette, which may consist of a partial or complete mitochondrial genome, acts as an active replication origin with the necessary associated replication factors. During replication, the rosette structure is enlarged and increased in complexity by re-initiation of replication in the same molecule before the previous replication is completed. At the end of DNA replication, the complex mtDNA rosette is formed. At the late cotyledon development stage, the complex rosette structure starts to degrade and finally turn into linear fragments, and only a scarce amount of mtDNA is reserved in the organelles (Dai et al., 2005).