(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Local mitochondrial physiology defined by mtDNA quality guides purifying selection [1]

['Felix Thoma', 'Faculty Of Biology', 'Ludwig-Maximilians-Universität München', 'Planegg-Martinsried', 'Graduate School Life Science Munich', 'Johannes Hagen', 'Romina Rathberger', 'Francesco Padovani', 'Institute Of Functional Epigenetics', 'Molecular Targets']

Date: 2026-01 

Our previous research had demonstrated that the first cell division of a zygote plays a major role in the segregation of mtDNA variants [ 10 ]. This led us to test if segregation might be faster when analysing populations derived from the first daughter cell of a heteroplasmic zygote, rather than populations derived from the zygote itself. To test this, we isolated the first daughter cells from 30 heteroplasmic zygotes, transferred them to a separate location on the agar plate, and examined mtDNA segregation after 20 hours ( S1E Fig ). This approach indeed revealed a more complete segregation of mtDNA variants compared to the analysis performed directly on the zygotes: the two clusters were more distinct, and the proportion of heteroplasmic cells fell from 3.32% to 1.8% ( S1B and S1D Fig ). Hence, we applied this procedure for all following mtDNA segregation experiments.

Atp6-NG may persist even after loss of the mtDNA ATP6-NG by which it is encoded, which could lead to an overestimation of the fraction of cells containing mtDNA ATP6-NG . To assess the extent of this effect, we quantified the decrease in Atp6-NG fluorescence per cell over time after inhibiting mitochondrial translation with chloramphenicol. From this analysis, we estimate an apparent fluorescence half-life of 2.7 h in cells, which likely reflects a combination of dilution by cell division and protein turnover. Comparison of Atp6-NG signal intensities after 2, 4, 6, and 8 hours of chloramphenicol treatment with untreated controls revealed a progressive decrease, with fluorescence clearly reduced and readily distinguishable from untreated cells by 4 hours ( S1C Fig ). These data indicate that only cells that have lost mtDNA ATP6-NG within approximately the last two hours and hence the previous cell cycle would still appear Atp6-NG–positive in our assays. Furthermore, as shown previously, populations derived from heteroplasmic founder cells segregate almost completely within 20 hours [ 12 ]. Hence, although a minor residual signal cannot be fully excluded, the stability of Atp6-NG is unlikely to have a substantial impact on our FAST quantifications.

In the next step, we tested whether flow cytometry enables rapid and quantitative analysis of mtDNA segregation from heteroplasmic founder cells. On glucose containing medium we crossed the mtDNA ATP6-NG (intron-containing) strain with a WT strain harboring intact intron-containing mtDNA (mtDNA ic ), which lacks a gene encoding a fluorescent protein, to generate heteroplasmic zygotes. Note that, depending on the experimental context, both intron-containing and intron-less mtDNA (mtDNA il ) variants are used throughout this study. After zygote formation on agar plates, we isolated 30 zygotes using a microdissection microscope and transferred them to a single separate area on the plate. The plates were then cultured for 20 hours to allow microcolony formation, after which the colony was collected and subjected to flow cytometry analysis ( Fig 1 D). Flow cytometry revealed clear mtDNA segregation, with 45.8% of cells displaying fluorescence and 50.9% showing no fluorescent signal ( Fig 1 E). Notably, the size distribution in Fig 1 E is narrower than in the liquid-grown controls ( Fig 1 A– 1 C), likely because the microcolonies came from cells grown on solid medium, while the controls were from log-phase liquid cultures, which show more size variation. While the segregation of cells into two clusters was evident, a subset of cells displayed intermediate fluorescence (3.32%, highlighted in Fig 1 E). Those intermediate cells were defined based on a narrow fluorescence intensity window ( 10% of the distance between peaks) centered around the local minimum between the two major peaks in the kernel density estimate (KDE) of the NG signal distribution ( S1D Fig ). We hypothesized that these might represent residual heteroplasmic cells. To explore transition from heteroplasmic to homoplasmic cells, we monitored the cell population over a time course at 8, 12, 20, and 24 hours after mating. Indeed, we observed a gradual shift toward two virtually completely separated discrete clusters: one fully fluorescent and one non-fluorescent ( Fig 1 F).

Flow-cytometric density plot of yeast cells lacking NG (A) , containing mtDNA ATP6-NG (B) , and a mixed population of both strains (C) with fluorescence intensity on the y-axis and forward scatter (FSC-A) on the x-axis. Note that non-fluorescent cells are not visible in (B) due to their low abundance. (D) Schematic of the FAST experimental workflow. (E) FAST density plot of mtDNA ATP6-NG mtDNA ic heteroplasmic microcolonies after 20 h: the area highlighted with a red dashed circle indicates heteroplasmic cells; percentages of fluorescent, heteroplasmic, and non-fluorescent populations are annotated next to each cluster. (F) mtDNA ATP6-NG cells were mated with mtDNA ic cells. 310, 150, 50, and 40 zygotes were isolated to discrete coordinates on an agar plate and microcolonies were analysed 8, 12, 20, and 24 h post-mating, respectively.

First, we asked whether flow cytometry allows reliable discrimination between non-fluorescent cells and cells harboring mtDNA ATP6-NG . To this end, we analysed three distinct populations: (i) ‘dark’ cells containing mtDNA not encoding fluorescent proteins, (ii) mtDNA ATP6-NG cells and (iii) a mixture of both ( Fig 1 A– 1 C). Flow cytometric analysis revealed that mtDNA ATP6-NG cells could be readily distinguished from dark cells. Although non-fluorescent and mtDNA ATP6-NG cells exhibited similar cell size based on forward scatter (FSC-A), only the latter showed markedly increased fluorescence intensity. Approximately 1.9% of cells from the mtDNA ATP6-NG strain lacked detectable fluorescence ( S1A Fig ). These likely correspond to petite cells that have partially or completely lost mtDNA including the Atp6-NG locus or elements required for mitochondrial translation. Notably, this proportion is lower than the 5% petite frequency typically observed in our background. A plausible explanation is that some petite cells retain the ATP6-NG reporter locus while harboring mutations or deletions elsewhere in the mtDNA that render them respiration-deficient. Such cells would remain fluorescent despite being petite, thereby reducing the apparent fraction of non-fluorescent cells. The difference between mtDNA ATP6-NG and dark cells was particularly evident in the mixed population, where the two cell types formed clearly separated clusters ( Fig 1 C).

In the following, we describe the development of a novel flow cytometry–based pipeline that enables quantitative analysis of mtDNA variant segregation in cell populations derived from single heteroplasmic zygotes carrying two distinct mitochondrial genomes. This approach builds on previous microscopy-based strategies that employed heteroplasmic yeast zygotes containing mtDNA encoding the fluorescent fusion protein Atp6-mNeonGreen (mtDNA ATP6-NG ) alongside a non-fluorescent mtDNA (“dark” mtDNA) [ 12 ].

Competitive advantage of intact over mutant mtDNA

We next sought to determine whether mtDNA harboring deletions of different genes encoding subunits of different respiratory chain complexes would exhibit altered segregation patterns. In our previous studies, we demonstrated a pronounced preference for WT mtDNA inheritance when mating WT strains with strains harboring a deletion of the COB gene in their mtDNA (mtDNAΔcob) using microscopic or genetic readouts [10,12]. Of note, the mtDNAΔcob strain was generated from intron-less mtDNA. In intron-containing mtDNA, a COB intron encodes a maturase required for COX1 mRNA splicing. Deleting COB in that context would therefore also block COX1 expression [20]. To control for effects of the intron-less nature of mtDNAΔcob, we included mating experiments as controls in our FAST analysis in which strains containing mtDNAATP6-NG were crossed with strains containing either intact mtDNAic or mtDNAil [21](Fig 2A and S1B Fig). Note that all FAST analyses were performed on cells mated and grown on glucose containing medium. We confirmed the purifying selection against cob mtDNA : mating WT mtDNAATP6-NG with mtDNAΔcob resulted in a significant skew towards cells containing intact mtDNA, with approximately 82.1% WT mtDNAATP6-NG-containing cells and only around 17.9% dark cells (Fig 2B). Because FAST relies on competition between intact and mutant mtDNA contributed by the two parental cells, it was important to test whether mtDNA is stably maintained in the mtDNAΔcob parent. If mtDNAΔcob cells were to become or at high frequencies, mating them with mtDNAATP6-NG cells would alter the competitive dynamics and could bias the FAST outcome. To rule out this possibility, we screened for cells by DAPI staining, which did not reveal any cells completely lacking mtDNA (S2H Fig). To test for partial mtDNA loss ( ), we plated single mtDNAΔcob cells and crossed the resulting colonies to tester strains carrying defined mitochondrial deletions. All 50 mtDNAΔcob colonies fully complemented the tester strains, indicating retention of the corresponding mtDNA regions (S2G Fig). Thus, we found no evidence for cells or for a substantial fraction of cells in the mtDNAΔcob strain.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Competitive advantage of WT over mutant mtDNA. FAST results for matings of mtDNAATP6-NG with (A) mtDNAil or (B) mtDNAΔcob. 30 first daughter cells from formed zygotes were isolated by micromanipulation, cultured for 20 h, and analysed by flow cytometry. (C) Schematic of the post-growth mixing assay workflow: 30 cells of mtDNAATP6-NG or non-fluorescent strains were microdissected to a defined, separate position, cultured for 20 h, then pooled for flow-cytometric analysis. (D) Bar graph depicting the fraction of non-fluorescent cells for each FAST mtDNAATP6-NG mating combination in WT and atp20 strains. Individual replicates are indicated by dots. (E) Results of the post-growth mixing assay showing the proportion of non-fluorescent cells. (F) Comparison of FAST and simulation results. In the simulation, mtDNA fate was modeled over time without assuming any selective intracellular advantage for either mtDNA variant. (G) Simulation outcomes for mtDNAΔcob, mtDNAΔcox2, and mtDNAΔatp6 matings across a range of initial mtDNA ratios (0.1–1), each simulated 10 times. The experimentally determined mtDNA ratio for each strain was simulated separately (100 iterations as for (F)) and is highlighted in blue. Statistical significance in panels (D) and (E) was determined by unpaired Student’s t-test on the means of replicates. Statistical significance in panel (F) was determined by Monte Carlo comparison of each FAST replicate to the simulation distribution, with replicate one-sided p-values combined using Fisher’s method to test for a significant reduction of cells containing ‘dark’ mtDNA in simulations compared to the FAST experiments. (*P < 0.05; **P < 0.005; ***P < 0.001). https://doi.org/10.1371/journal.pgen.1011836.g002

In contrast to the mtDNAATP6-NG mtDNAΔcob cross, mating strains containing mtDNAATP6-NG with WT strains harboring intact mtDNAic or mtDNAil resulted in a more balanced segregation (Fig 2A–2B and S1B Fig). Notably, we observed a slight competitive advantage of mtDNAic over mtDNAATP6-NG, resulting in a higher proportion of cells retaining the ‘dark’ mtDNAic. This suggests a modest selection against mtDNAATP6-NG within the arising populations and that Atp6-NG might be functionally compromised, despite the absence of detectable growth defects in strains harboring mtDNAATP6-NG, which was assessed in previous studies [10].

Next, we asked whether the observed selective disadvantage is specific to the COB gene or whether similar selection patterns apply to mutations in other mtDNA-encoded genes. To address this, we extended our analysis to mtDNA deletion strains lacking either COX2 (mtDNAΔcox2), which encodes a core subunit of complex IV, or ATP6, a core component of the ATP synthase complex (mtDNAΔatp6) [22–24]. Like the mtDNAΔcob strain, mtDNAΔcox2 and mtDNAΔatp6 strains were unable to grow on non-fermentable carbon sources (S1F Fig). Moreover, the mtDNAΔcox2 and mtDNAΔatp6 strains did not exhibit pronounced loss of mtDNA (S1G and S1H Fig). We then performed matings between strains containing intact mtDNAATP6-NG and strains harboring either mtDNAΔcox2 or mtDNAΔatp6. Both mtDNAΔcox2 and mtDNAΔatp6 contain introns, and thus results from the FAST analyses involving these strains must be compared to those from a mtDNAATP6-NG mtDNAic cross. In both cases, the amount of non-fluorescent cells was reduced (27.6% and 30.8%, respectively), pointing to a preferential maintenance of intact mtDNA in the resulting colonies (Fig 2D).

We validated these findings by repeating the analysis using fluorescence microscopy instead of flow cytometry. For this analysis, individual colonies derived from the first daughter cells of heteroplasmic zygotes were imaged separately to assess cell-to-cell variability. This approach highlighted a stochastic element in the segregation process, with some colonies from heteroplasmic zygotes composed almost entirely of non-fluorescent (dark) cells and others containing predominantly fluorescent cells (S2A and S2B Fig). Nonetheless, when results were averaged across multiple colonies, they consistently converged on a reproducible mean, which was comparable to our findings obtained through FAST.

To interpret the outcomes of FAST and to infer modes of selection for or against mtDNA variants, the following two parameters must be taken into account. First, the two parental cells may contribute unequal amounts of each mtDNA variant to the heteroplasmic zygote and bias the predominance of a certain mtDNA variant in the arising population. Second, strains carrying mutant mtDNA variants may exhibit slower growth rates, even on fermentable medium, compared to their corresponding WT strains, potentially contributing to a lower percentage of cells containing mutant mtDNA in populations resulting from heteroplasmic founder zygotes.

First, we tested for putative growth defects associated with the presence of mutant mtDNA variants. Assessment of doubling times for strains containing either intact (mtDNAil or mtDNAic) or mutant (mtDNAΔcob, mtDNAΔcox2, or mtDNAΔatp6) mtDNA in liquid fermentable medium did not reveal significant growth differences (S2C and S2D Fig). However, the FAST analysis is performed on solid medium, where spatially heterogeneous growth conditions within colonies may impose additional metabolic challenges [25–27]. To better reflect these conditions, we sought to assess the growth behavior of the different yeast strains under similar settings, and therefore repeated the FAST analysis without the mating step. Specifically, thirty haploid budding cells were isolated and deposited at a defined position on the agar plate containing glucose as a carbon source, giving rise to a microcolony derived from these founder cells. This procedure was performed for all strains harboring the various mtDNA variants. After 20 hours of growth, the resulting colonies were harvested and pooled in combinations mimicking those used in the mating-based FAST experiments. The pooled populations were then immediately analysed by flow cytometry. We refer to this assay as ‘post-growth mixing assay’ (Fig 2C).

For mixtures of WT mtDNAATP6-NG and either mtDNAΔcob or mtDNAΔcox2 cells, approximately 50% of the population exhibited fluorescence, consistent with similar growth rates between strains (Fig 2E). Thus, the observed reduction of non-fluorescent mutant cells in the FAST experiments involving mtDNAΔcob or mtDNAΔcox2 is unlikely to be caused by growth differences. In contrast, when we performed the post-growth mixing assay with mtDNAATP6-NG and mtDNAΔatp6 cells, a lower proportion of mtDNAΔatp6 cells was observed, indicating a growth disadvantage for this strain under plate-based conditions. This growth defect likely contributes to a decreased percentage of mtDNAΔatp6 cells in FAST analysis of populations derived from heteroplasmic zygotes.

In our post-growth mixing assay, cells were homoplasmic for either mtDNAATP6-NG or mtDNAΔatp6, which is an extreme scenario that would arise if complete segregation of mtDNA would occur during the first cell division. In contrast, mtDNA segregation starts from heteroplasmy and proceeds gradually [12]. Hence, the growth disadvantage due to mtDNAΔatp6 is likely minor at first and increases as cells approach homoplasmy.

The second factor potentially biasing the segregation experiment is that unequal mtDNA copy numbers in the two parental cells lead to unequal variant ratios in the resulting heteroplasmic zygote. We first quantified mtDNA copy number in strains carrying exclusively intact or mutant mtDNA by quantitative PCR. As a reference, we used the mtDNAATP6-NG strain that was used in all matings. Unexpectedly, strains harboring mtDNAil displayed elevated mtDNA copy numbers compared to the mtDNAATP6-NG strain (S2E Fig). Strains containing mtDNAΔcob also showed increased mtDNA copy number relative to the mtDNAATP6-NG strain, although slightly lower than in mtDNAil. mtDNAic strains carried mtDNA amounts comparable to mtDNAATP6-NG, whereas mtDNAΔcox2 and mtDNAΔatp6 strains showed a modest reduction.

To better resolve how different mtDNA copy numbers and growth rates influence FAST outcomes, we used a previously established mtDNA-segregation simulation framework [12]. This model simulates mtDNA segregation at the population level and incorporates major features of mitochondrial dynamics, including fusion, fission, and mtDNA partitioning during cell division (S2G Fig). We extended the model by adding a growth-correction factor for heteroplasmic cells containing intact and mutant mtDNA, assuming a linear relationship between the intracellular percentage of each mtDNA variant and its contribution to cellular growth rate.

To parameterize the simulation, we first determined the doubling time of mtDNAATP6-NG cells on plates containing glucose (1.48 h) by seeding 10, 20, or 30 cells and quantifying cell numbers after 20 hours via flow cytometry (S2F Fig). Combined with the fluorescence ratios obtained in the post-growth mixing assay, these measurements allowed us to calculate the relative growth rate of cells containing mutant mtDNA compared with mtDNAATP6-NG cells, which served as the growth-correction factors in the model. We also incorporated the measured starting copy-number differences between mtDNA variants. Thus, starting cells were initialized with mtDNA amounts proportional to those found in the corresponding haploid parental strains. For example, in a cross between mtDNAATP6-NG and mtDNAΔcox2 parental cells, the starting cell contained a ratio of mtDNAATP6-NG:mtDNAΔcox2 of 1:0.71.

Using these parameters, the model predicted the proportion of mtDNAATP6-NG in the population after 20 h of growth, mirroring the conditions of the FAST assay. Importantly, the simulations do not assume any intracellular disadvantage of mutant mtDNA; they incorporate only growth-rate differences and the measured initial mtDNA abundances. Given the lack of a growth defect or reduced mtDNA copy number in the case of mtDNAΔcob, the simulations predicted equal numbers of mtDNAATP6-NG and mtDNAΔcob cells for the mtDNAATP6-NG mtDNAΔcob cross, which was substantially different from the results of the respective FAST analysis, where mtDNAΔcob cells amounted to less than 20% (Fig 2F). Simulations for the mtDNAATP6-NG mtDNAΔcox2 cross, predicted reduced amounts (39.5%) of mtDNAΔcox2 cells after 20 h of growth. However, this amount was still larger compared to the FAST analysis (27.7%). Thus, growth differences and unequal mtDNA contributions alone do not appear to be responsible for the experimentally observed depletion of mtDNAΔcob or mtDNAΔcox2 cells in the respective populations. To further test this, we systematically varied the starting ratios of mtDNAATP6-NG to mutant mtDNA to identify conditions under which the simulations would reproduce the FAST results. The ratios required to match the experimental data were considerably lower—approximately 1:0.25 for mtDNAΔcob and 1:0.4 for mtDNAΔcox2—than the empirically measured starting amounts (Fig 2G).

In contrast to crosses involving mtDNAΔcob or mtDNAΔcox2, for the mtDNAATP6-NG mtDNAΔatp6 cross, the simulated proportion of dark cells was only slightly higher (33.3%) than experimentally observed by FAST (30.8%), indicating that growth defects associated with mtDNAΔatp6 cells in combination with reduced amounts of mtDNAΔatp6 can account to a large extent for the reduction of mtDNAΔatp6 cells in populations arising from cells heteroplasmic for mtDNAATP6-NG and mtDNAΔatp6.

In summary, our analyses indicate that segregation outcomes in populations derived from heteroplasmic zygotes are in part shaped by the combined effects of growth differences linked to mutant mtDNA and parental mtDNA copy-number disparities. These factors, which are explicitly incorporated into our simulations, are sufficient to explain the behaviour of mtDNAΔatp6, whose depletion in FAST assays can be accounted for by its growth disadvantage together with its reduced mtDNA abundance in the parental haploid cell. In contrast, the experimentally observed loss of mtDNAΔcob—and, to a lesser extent, mtDNAΔcox2—is stronger than predicted by simulations that include only growth and mtDNA copy number effects. This discrepancy indicates the presence of an additional intracellular mode of selection against mutant mtDNA.

As shown previously, mitochondrial ultrastructure plays an important role in maintaining purifying selection [10]. It has been proposed that cristae-mediated compartmentalization of mitochondria allows each mitochondrial genome to supply proteins specifically to the respiratory chain complexes within its immediate vicinity, thereby enabling intracellular processes to selectively maintain intact genomes based on physiological differences in the mitochondrial network [10,11].

In our previous work, we demonstrated that deletion of ATP20, a gene required for proper cristae formation, compromises this spatial organization and impairs purifying selection against mtDNA lacking the COB gene [10]. To confirm these findings, to test our system, and to examine whether cristae formation is also critical for selection against mtDNAΔcox2 and mtDNAΔatp6, we performed FAST with strains lacking ATP20. Indeed, purifying selection was strongly reduced in atp20 cells and segregation shifted towards similar percentages of mtDNAATP6-NG cells and cells containing mtDNAΔcob, mtDNAΔcox2 and mtDNAΔatp6, implying that normal cristae formation via assembly of the dimeric ATP synthase [28] is critical for counterselection against mutant mtDNA (Fig 2D). The loss of counterselection against mtDNAΔatp6 in the absence of atp20 was unexpected, as our analyses indicate that the disappearance of mtDNAΔatp6 mtDNA in the FAST assays predominantly depends on growth defects and altered mtDNA copy number and should therefore be unaffected by the loss of ATP20, which has been implicated in intracellular selection against mutant mtDNA. We therefore examined the growth of mtDNAΔatp6 strains in the atp20 background and found that the growth defect of mtDNAΔatp6 cells was fully rescued, yielding growth indistinguishable from that of mtDNAATP6-NG cells (S2H Fig). Moreover, mtDNA copy numbers in atp20 mtDNAΔatp6 cells were not reduced compared to the atp20 mtDNAATP6-NG strain, but rather increased. These findings explain the absence of mtDNAΔatp6 depletion in the FAST assay, although the molecular basis for the suppression of growth defects and increase of mtDNA copy number by absence of Atp20 remain unclear.

Importantly, mtDNA copy numbers and growth rates were unchanged in atp20 strains containing mtDNAic or mtDNAΔcox2 compared with the corresponding WT strains containing mtDNAic or mtDNAΔcox2 (S2I Fig). This indicates that the counterselection against mtDNAΔcox2 observed in FAST assays in a WT background is independent of mtDNA copy number differences and requires mechanisms that are absent in atp20 cells. Unexpectedly, mtDNA copy numbers increased substantially in atp20 strains carrying either mtDNAil or mtDNAΔcob relative to WT strains harboring the same mtDNA variants, a change that is likely responsible for the elevated proportion of dark cells lacking mtDNAATP6-NG observed in the FAST analysis of mtDNAATP6-NG mtDNAil and mtDNAATP6-NG mtDNAΔcob crosses in a atp20 mutant background. The basis for this increase in intron-less mtDNA copy number upon loss of Atp20 remains unclear. Crucially, however, the relative ratio between mtDNAil and mtDNAΔcob is only mildly altered in a atp20 background (1:1) compared with WT (1:0.8). Although the overall increase in mtDNA abundance may influence segregation dynamics to some extent, we consider it unlikely to fully explain the complete loss of counterselection against mtDNAΔcob.

[END]
---
[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1011836

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/