SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

Mitochondrial DNA haplotypes are associated with various phenotypes, such as altered susceptibility to disease, environmental adaptations, and aging. Accumulating evidence suggests that mitochondrial DNA is essential for cell differentiation and the cell phenotype. However, the effects of different mitochondrial DNA haplotypes on differentiation and development remain to be determined. Using embryonic stem cell lines possessing the same Mus musculus chromosomes but harboring one of Mus musculus, Mus spretus, or Mus terricolor mitochondrial DNA haplotypes, we have determined the effects of different mitochondrial DNA haplotypes on chromosomal gene expression, differentiation, and mitochondrial metabolism. In undifferentiated and differentiating embryonic stem cells, we observed mitochondrial DNA haplotype-specific expression of genes involved in pluripotency, differentiation, mitochondrial energy metabolism, and DNA methylation. These mitochondrial DNA haplotypes also influenced the potential of embryonic stem cells to produce spontaneously beating cardiomyocytes. The differences in gene expression patterns and cardiomyocyte production were independent of ATP content, oxygen consumption, and respiratory capacity, which until now have been considered to be the primary roles of mitochondrial DNA. Differentiation of embryonic stem cells harboring the different mitochondrial DNA haplotypes in a 3D environment significantly increased chromosomal gene expression for all haplotypes during differentiation. However, haplotype-specific differences in gene expression patterns were maintained in this environment. Taken together, these results provide significant insight into the phenotypic consequences of mitochondrial DNA haplotypes and demonstrate their influence on differentiation and development. We propose that mitochondrial DNA haplotypes play a pivotal role in the process of differentiation and mediate the fate of the cell. STEM CELLS 2013;31:703–716


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

The mitochondrial genome (mtDNA) encodes 13 of the key subunits of the electron transfer chain (ETC), which generates ATP through oxidative phosphorylation (OXPHOS) [1]. OXPHOS is essential for the production of ATP, since it generates 32 molecules of ATP compared with only two produced by the other key energy-generating pathway, glycolysis [2]. The other 75 subunits of the ETC are encoded by the chromosomal genome and these proteins translocate to the mitochondrion where the nuclear- and mtDNA-encoded subunits are assembled into functional ETC complexes.

Over billions of years, mtDNA haplotypes have evolved [3]. These are a series of polymorphic variants within the mtDNA that are often associated with regional variation and distinguish breeds and strains of organisms. They can confer a positive or negative advantage to the individual [4]. For example, they influence milk quality in cows [5]; fertility in cows [6] and pigs [7]; growth and physical performance in mice [8]; sperm motility in humans [9]; adaptation to warm and cold climates [10]; organism longevity [11]; and an individual's predisposition to various age-associated disorders [4], such as cancer [12] and diabetes [13, 14].

Our understanding of how different mtDNA haplotypes influence the mature phenotype is very limited. During early mammalian development, there are significant reductions in mtDNA copy number as mtDNA passes from the fertilized oocyte into the inner cell mass and into its derivatives, embryonic stem cells (ESCs) [15, 16], which establishes the mtDNA set point. The mtDNA set point ensures that undifferentiated cells have a small population of mtDNA, which is then expanded in a cell-specific manner later during differentiation in order that specialized cells can meet their specific needs for OXPHOS-derived ATP [15, 17, 18]. Consequently, undifferentiated ESCs rely predominantly on glycolysis to produce ATP [19, 20], although not exclusively [20, 21]. As a result, they express high levels of key markers of glycolysis such as Gadph and Glut1 [22]. However, with cell differentiation, there is a cell-type-specific shift in energy metabolism [19, 21, 23, 24]. For example, during the differentiation of ESCs into cardiomyocytes, there is an increase in the number of mitochondria per cell and expression of mtDNA-specific replication factors [25] as cells become more OXPHOS dependent [24].

Although cell-specific function and gene expression profiles during differentiation are controlled by cytoplasmic, genetic, and epigenetic mechanisms [26–28], it is becoming increasingly evident that mtDNA is integral to cell differentiation and thus the cell phenotype [15, 24, 29–32]. This is highlighted by intraspecies and interspecies nuclear transfer embryos [33–36], where the chromosomal genome is not matched with a compatible mtDNA haplotype.

Central to understanding the role of mitochondria in development is the symbiotic relationship between the nuclear and the mtDNAs. Studies on the genetic and metabolic consequences of mitochondrial disease have used cytoplasmic hybrids (cybrids), where cells are depleted of their mtDNA complement and repopulated with exogenous mtDNA haplotypes, so that the nuclear genetic component remains constant and cellular changes are due to the transplanted mtDNA [37]. These fusions can be produced with different mtDNA haplotypes by using divergent strains or species of animal [37–40]. While high levels of mtDNA divergence between chromosomal and mtDNA result in OXPHOS defects [40, 41], low levels of divergence produce mild, often undetectable OXPHOS defects [38]. ESC lines have been generated that possess the same chromosomal genes (Mus musculus) but different mtDNA haplotypes, that is, Mus spretus or Mus terricolor (formally known as Mus dunni) [39]. In this instance, the degree of divergence between the Mus musculus and Mus spretus mitochondrial haplotypes is 2.0 million years or 38 amino acids [41], and that of Mus musculus and Mus terricolor haplotypes is 4.0 million years and 155 amino acids [41, 42]. These ESCs have been shown to give rise to chimeric mice and are germ line competent [39, 43], and thus provide excellent models to study the effects of mtDNA haplotypes on gene expression as pluripotent cells differentiate into mature cells, independent of any measureable OXPHOS defects.

Here, we show that ESCs harboring different mtDNA haplotypes display variable expression of genes associated with pluripotency, DNA methyltransferases, and energy metabolism. This is further amplified as differentiation progresses and is further influenced by their culture in 2D and 3D environments. To our knowledge, this is the first demonstration that mtDNA haplotypes can mediate cell fate.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

ESC Culture and Embryoid Body Formation

CC9.3.1 ESCs (Mus musculus; CC9mus), CC9.3.1 ESCs harboring divergent Mus spretus (CC9spretus), and Mus terricolor (Dunni; CC9dunni) mtDNA [39] were cultured on mitomycin C-treated mouse embryonic fibroblasts in ESC media, as previously described [15]. ESCs were differentiated by forming embryoid bodies (EBs), as previously described [16]. See supporting information data for a comprehensive protocol.

Electrospun 3D Scaffolds

Poly(ε-caprolactone) (PCL, Sigma, Mn = 80,000) was electrospun using a hhtp//:www.yflow.com. A 12% (w/v) solution in 3:1 chloroform:methanol (Merck Pty Ltd, Victoria, Australia, http://www.merckmillipore.com.au) was prepared at room temperature before being loaded into a 10-ml plastic syringe equipped with an 18 gauge needle for electrospinning. Electrospinning conditions were +20 kV and −5 kV with a 1.5 ml/hour flow rate and a working distance of 18 cm. Fibers were collected on an aluminum foil coated mandrel that had a diameter of 50 mm and which spun at 100 rpm. Temperature during spinning was 23°C and humidity was controlled to 15% ± 3% RH by means of nitrogen flow into an enclosed box. The needle from which the solution was ejected was translated back and forth (over a distance of 30 cm) parallel to the mandrel axis during spinning, to encourage more even coverage over the foil surface. The resulting mat was stored in a vacuum desiccator. The resulting fiber diameter was measured using ImageJ software to be 1.3 ± 0.8 μm. Samples were cut into 15 mm circles using a punch before being sterilized in 70% ethanol for 30 minutes and washed with sterilized phosphate buffered saline three times prior to cell culture.

Real Time Polymerase Chain Reaction (qPCR): Gene Expression and mtDNA Copy Number

Real time polymerase chain reaction (PCR) contained 2 μl of template (cDNA or DNA), 10 μl of 2× SensiMix (Bioline Pty Ltd, Sydney, Australia, https://www.bioline.com), 0.33 μM of forward and reverse primers (see supporting information Table S1), and ultrapure ddH2O (20 μl total volume). Reactions were performed in a Rotorgene-3000 real time PCR machine (Corbett Research, Cambridge, U.K., http://www.qiagen.com/corbett/) under primer-specific conditions. The number of mtDNA copies per cells and relative expression levels were quantified against known standards, using real time PCR, as previously described [15]. We assessed the reliability of β-Actin, Gapdh, and 18S as housekeeping genes. The expression of β-Actin and Gapdh varied between cell types, although 18S expression remained consistent across all samples analyzed. Primers are listed in supporting information Table S1.

ATP Quantification and Cellular Lactate Production

ATP content and lactate production were quantified using the ATP-LITE assay kit (Perkin Elmer, Melbourne, Australia, http://www.perkinelmer.com/au) and Lactate Assay Kit II (BioVision, San Francisco, USA, http://www.biovision.com), respectively, according to the manufacturer's conditions. See supporting information data.

Oxygen (O2) Consumption Rates

O2 consumption rates were determined by high-resolution respirometry (Oroboros Oxygraph-2K, Innsbruck, Austria, http://www.oroboros.at/), as previously described, with minor modifications [44]. See supporting information data.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 5.01 (GraphPad Software, Inc., San Diego, USA, http://www.graphpad.com). Data are expressed as mean ± SEM. The differences in gene expression were determined using one-way or two-way ANOVA in combination with Bonferroni post hoc tests. A Bartlett's test for equal variances was performed to determine the normal distribution of data. When one or all of the requirements for performing ANOVA were not met, a Kruskal-Wallis test was performed on Ranked values followed by Dunn's method.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

ESCs Harboring Divergent mtDNA Haplotypes Have Altered Pluripotent Gene Expression Profiles

To determine whether more closely related or more divergent mtDNA haplotypes can influence pluripotency, we analyzed the expression of several key transcriptional regulators of pluripotency [45] in the mouse ESC line, Mus musculus (CC9.3.1 cells; CC9mus). We then compared these outcomes with two other ESC lines that possessed the same chromosomal genome but with populations of mtDNA from either Mus spretus (CC9spretus) or Mus terricolor (CC9dunni). In order to ensure that we were only analyzing pure populations of undifferentiated cells, we isolated populations of undifferentiated ESCs by fluorescence-activated cell sorting following labeling with an antibody to SSEA1. We retrieved similar levels of enrichment for each of the ESC lines (supporting information Fig. S1A) and determined that each of the cell lines proliferated at the same rate (supporting information Fig. S1B).

In undifferentiated CC9mus and CC9spretus cells, we examined the levels of expression of the pluripotency-associated genes: Oct4 (Pou5f1), Sox2, Nanog, Klf4 [46], Rex1 [47], Dppa5 [48], Pramel 7 [49], and Ndp52L1 [50], which have been associated with mtDNA replication [15, 51]. For Oct4, Sox2, and Nanog, levels of expression were similar (Fig. 1A). However, there were increased levels of expression in the CC9spretus cells for Klf4 (p < .01) and reduced levels for Pramel 7 (p < .001) compared to CC9mus cells (Fig. 1A). While there were no significant changes in gene expression for Oct4 and Sox2 between CC9dunni and CC9mus ESCs, there were elevated levels of expression for Nanog (p < .05), Rex1 (p < .05), Klf4 (p < .01), Dppa5 (p < .01), and Ndp52L1 (p < .001) and reduced levels of expression for Pramel 7 (p < .001) in the CC9dunni cells. There were also significant differences between CC9spretus and CC9dunni cells for Nanog (p < .01), Rex1 (p < .001), Dppa5 (p < .01), Pramel7 (p < .001), and NDP52L1 (p < .001). We confirmed the nuclear localization of OCT4 and NANOG and the cell surface localization of SSEA1 by immunocytochemistry (supporting information Fig. S1C).

thumbnail image

Figure 1. Pluripotent characteristics of embryonic stem cells (ESCs) harboring divergent mitochondrial genome (mtDNA) haplotypes. (A): Real time polymerase chain reaction (PCR) analysis of Oct4 (Pou5f1), Sox2, Nanog, Rex1, Klf4, Dppa5, Pramel7, and Ndp52L1 gene expression in CC9mus, CC9spretus, and CC9dunni ESCs. (B): Relative enrichment for 5mC was analyzed by real time PCR within the promoter regions of Oct4, Nanog, Rex1, and Dppa5 in CC9mus, CC9spretus, and CC9dunni ESCs. (C): Relative enrichment for 5hmC was analyzed by real time PCR within the promoter regions of Oct4, Nanog, Rex1, and Dppa5 in CC9mus, CC9spretus, and CC9dunni ESCs. (D): Real time PCR analysis of OCT4 occupancy within the Nanog promoter region of CC9mus, CC9spretus, and CC9dunni ESCs. (E): Real time PCR analysis of Dnmt1, Dnmt3a, and Dnmt3b gene expression in CC9mus, CC9spretus, and CC9dunni ESCs. (F): MtDNA copy number analysis in CC9mus, CC9spretus, and CC9dunni ESCs. (G): ATP content in undifferentiated CC9mus, CC9spretus, and CC9dunni ESCs. (H): Oxygen consumption of undifferentiated CC9mus, CC9spretus, and CC9dunni ESCs. The data show basal oxygen consumption, nonphosphorylating respiration (5 mg/ml oligomycin), and uncoupled respiration (100 nM FCCP). (J): Lactate content of culture media after 24 hours of culture with CC9mus, CC9spretus, and CC9dunni ESCs. All values represent mean ± SEM. Real time PCR analysis of gene expression and mtDNA copy number was performed on SSEA1+ fluorescence-activated cell sorting sorted cells. Real time PCR gene expression levels are expressed relative to CC9mus cells. Significant differences between cell types are: *, p < .05; **, p < .01; ***, p < .001. Abbreviations: ETC, electron transfer chain; 5mC, 5-methylcytosine; 5hmC, 5-hydroxymethylcytosine.

Download figure to PowerPoint

As the transcriptional regulators of pluripotency are regulated by DNA methylation [28], we determined, using methylated DNA immunoprecipitation (MeDIP), whether there were similar levels of enrichment for 5mC, which is indicative of steady-state DNA methylation, within the promoter regions of Oct4, Nanog, Rex1, and Dppa5 for each of the mtDNA haplotypes (Fig. 1B). Enrichment within the promoter region of Nanog was significantly higher in CC9mus than CC9spretus (p < .001) and CC9dunni (p < .01) cells. We also observed significantly more enrichment in the promoter region of Dppa5 for CC9mus cells compared to the two more divergent haplotypes (p < 0.01). When MeDIP was performed for 5-hydroxymethylcytosine (5hmC), an indicator of transitory DNA methylation, enrichment in the promoter region of Oct4 was significantly lower in CC9mus (p < .001) and CC9dunni (p < .001) than in CC9spretus cells (Fig. 1C). We also observed significantly less enrichment in the promoter region of Rex1 for CC9mus cells compared to the two more divergent haplotypes (p < .01). Furthermore, the interaction of OCT4 with the promoter region of Nanog, which is essential for establishing the pluripotent network, was significantly less in the CC9dunni cells compared to CC9mus cells (p < .01) and CC9spretus cells (p < .05) (Fig. 1D). We then analyzed the levels of expression of the DNA methyltransferases (Dnmt) 1, 3a, and 3b (Fig. 1E). Overall, CC9spretus ESCs expressed significantly lower levels of Dnmt1 (p < .05), Dnmt3a (p < .001), and Dnmt3b (p < .001) when compared with CC9mus and CC9dunni ESCs. While the levels of expression were similar for CC9mus and CC9dunni cells for Dnmt1 and Dnmt3a, they were significantly different for Dnmt3b (p < .001; Fig. 1E). From these findings, it is evident that mtDNA haplotype influences the expression of pluripotent genes and has subtle effects on the DNA methylation of the promoters of these genes.

MtDNA Haplotypes Have Altered mtDNA Copy Number and Expression of OXPHOS Subunits

As the mtDNA set point establishes low levels of mtDNA copy number in undifferentiated cells, we determined whether the different mtDNA haplotypes affect mtDNA copy number and if this has subsequent effects on the generation of ATP through OXPHOS by the ETC. The more divergent CC9dunni ESCs contained fewer mtDNA copies than the CC9mus cell line (p < .05; Fig. 1F), although no significant differences in total ATP produced (Fig. 1G) or oxygen consumption (basal O2; Fig. 1H) were observed among the three ESC lines. Furthermore, the three ESC lines were actively respiring at their maximal capacity, as demonstrated by their uncoupled/basal respiration rates of approximately 1 (Fig. 1H). As the production of lactate is an indicator of anaerobic metabolism, we analyzed lactate production and observed similar levels for the three ESC lines (Fig. 1J). This is consistent with the low numbers of mtDNA copy and low levels of ATP generated by OXPHOS in undifferentiated ESCs [15, 16].

Despite these similarities in energy metabolism, divergent ESCs displayed altered levels of expression for the nuclear- and mtDNA-encoded subunits of the ETC. The levels of expression for the nuclear-encoded subunits, Ndufb5 (p < .05), Sdhb (p < .05), and Cox5b (p < .05) were reduced in CC9spretus cells (supporting information Fig. S1D), which were also reflected in the lower levels of mtDNA-encoded genes ND1 (p < .001), CytB (p < .001), COXI (p < .001), and ATPase6 (p < .001) (supporting information Fig. S1E). In marked contrast, there were elevated levels of expression for Uqcrc2 (p < .05; supporting information Fig. S1D) and COXI (p < .001; supporting information Fig. S1E) in CC9dunni ESCs, while CytB was reduced (p < .001; supporting information Fig. S1E). We also analyzed the expression of Glut1 and Gapdh (supporting information Fig. S1F), which are involved in glucose metabolism. While there were no differences in expression for Glut1, the more divergent CC9dunni ESCs expressed increased levels of the glycolytic enzyme, Gapdh (p < .05).

Energy Metabolism During Early Differentiation Is Similar Among mtDNA Haplotypes

To further test the influence of the different mtDNA haplotypes on the pluripotent potential of ESCs, we induced differentiation by generating EBs in hanging droplets in the absence of leukemia inhibitory factor. First, we analyzed the expression of OCT4, NANOG, and SOX2 over the first 7 days of differentiation by Western blot to determine that protein expression for the key regulators of pluripotency decreased during differentiation. In undifferentiated cells, protein expression was similar for SOX2 (supporting information Fig. S2A, S2B), OCT4 (supporting information Fig. S2A, S2C), and NANOG (supporting information Fig. S2A, S2D) among the three mtDNA haplotypes, except for NANOG in the CC9dunni cells. As expected, by day 7, protein levels for OCT4, SOX2, and NANOG had diminished. Surprisingly, we observed high levels of protein expression for each pluripotent factor on day 2 (supporting information Fig. S2A–S2D).

During the 7 days of differentiation, all three mtDNA divergent ESC lines produced comparable EBs of similar size (supporting information Fig. S2E), which proliferated at a similar rate (supporting information Fig. S2F) and contained comparable amounts of ATP (Fig. 2A). There were no significant differences in the resting O2 consumption (basal) rates, ETC coupling (basal/nonphosphorylating respiration) rates, or ETC capacity (uncoupled/basal respiration) among the three cell lines (Fig. 2B). In addition, on day 7 of differentiation, lactate production had reduced to 50% of the levels observed in undifferentiated ESCs. However, there was no significant difference among the cell lines (Fig. 2C). These outcomes demonstrate that the three mtDNA haplotypes are able to downregulate pluripotent gene expression during differentiation and maintain similar metabolic profiles.

thumbnail image

Figure 2. Embryonic stem cells (ESCs) harboring divergent mtDNA haplotypes demonstrate altered differentiation profiles. (A): ATP content in CC9mus, CC9spretus, and CC9dunni cells on day 7 of differentiation. (B): Oxygen consumption of differentiated (day 7) CC9mus, CC9spretus, and CC9dunni cells. The data show basal oxygen consumption, nonphosphorylating respiration (5 mg/ml oligomycin), and uncoupled respiration (100 nM FCCP). (C): Lactate content from culture media for CC9mus, CC9spretus, and CC9dunni cells on day 7 of differentiation. (D): The mean percentage of embryoid bodies (EBs) with beating foci with respect to the total number. CC9mus, CC9spretus, and CC9dunni ESCs were differentiated as EBs for 7 days when EBs were plated on gelatin. The numbers of beating EBs were counted every 24 hours until day 12 of differentiation (n = 4). (E): Real time polymerase chain reaction (PCR) analysis of Mesp1, Gata4, Tbx5, Nkx2.5, cardiac troponin T2 (Tnnt2), and myosin heavy chain 7 (Myh7) gene expression in CC9mus, CC9spretus, and CC9dunni cells on days 4, 7, and 12 of differentiation. (F): Real time PCR analysis of Sox2, Sox3, Musashi 1, Pax6, β-III-Tubulin (Tubb3), and Synaptophysin (Syn) gene expression in CC9mus, CC9spretus, and CC9dunni cells on days 5, 12, and 18 of neural-induced differentiation. Significant differences between cell types are: *, p < .05; **, p < .01; ***, p < .001. Abbreviation: ETC, electron transfer chain.

Download figure to PowerPoint

The Effects of mtDNA Haplotypes Are Amplified in Late-Stage Differentiation

As pluripotent stem cells have the potential to differentiate into multiple lineages, we first determined whether this was affected by the different mtDNA haplotypes. After 7 days of differentiation, each of the lines was analyzed using a cell lineage identification qPCR array to profile the expression of 84 genes of cellular differentiation. The CC9dunni cells overexpressed Slc32a1 and underexpressed Gfap, while the CC9spretus cells underexpressed Gfap, Map3k12, and Tyr compared to CC9mus cells (supporting information Table S2).

Since the differentiation of ESCs into cardiomyocytes is dependent on the transition from glycolysis to mitochondrial metabolism [24] and is indicative of mesodermal differentiation, we investigated whether the variation in mtDNA haplotypes influenced the potential of ESCs to develop beating foci, which characterizes cardiomyocyte potential. We transferred EBs from each cell line onto gelatin-coated culture dishes, on day 7 of differentiation, and monitored the number of EBs that produced rhythmically beating foci. Beating foci were evident after 24 hours (day 8) for all cell lines. However, the percentage of beating EBs for CC9spretus cells (3.57% ± 3.57%) was significantly lower (p < .05) compared to CC9mus (27.06% ± 2.82%) and CC9dunni (25.06% ± 5.14%; Fig. 2D) cells. The number of beating EBs increased up to day 12 of differentiation. At this time point, the number of beating EBs observed for CC9spretus cells (33.04% ± 1.72%) remained significantly lower compared to CC9mus (54.11% ± 2.82%; p < .01; Fig. 2D) and CC9dunni (68.9% ± 4.02%; p < .001; Fig. 2D) cells. Furthermore, CC9dunni cells (p < .05) produced significantly more beating EBs than CC9mus cells. We further demonstrated that beating EBs expressed the master regulators of cardiomyocyte differentiation, namely Mesp1, Gata4, Tbx5, and Nkx2.5, and the endpoint markers, Tnnt2 and Myh7 (Fig. 2E).

The above observations suggest that mtDNA haplotypes influence the pattern of cellular differentiation and alter the ability of the pluripotent cell to generate certain cell types. Therefore, differentiation was continued until day 18, when we analyzed the levels of expression of the markers of differentiation using the same cell lineage array (supporting information Table S3). When compared to CC9mus cells, CC9spretus cells differentially expressed 31 lineage-specific genes, 5 were overexpressed and 26 under-expressed (supporting information Table S3), while CC9dunni cells differentially expressed 18 genes, 9 overexpressed and 9 underexpressed (supporting information Table S3). These differences suggest that the in vitro differentiation of ESCs containing different mtDNA haplotypes influences the ability of these cells to produce terminally differentiated cell types. Indeed, the markers of mesodermal and endodermal terminal differentiation were preferentially underexpressed in CC9spretus cells (6/12 = mesoderm; 4/12 = endoderm; supporting information Table S3). Moreover, CC9dunni cells also underexpressed mesoderm and endoderm markers of terminal differentiation, as well as overexpressing immature progenitor and germ layer markers (supporting information Table S3).

To further highlight the impact of the different mtDNA haplotypes on differentiation, we induced each of the lines to undergo neural differentiation [52], which is indicative of ectodermal differentiation. We then assessed expression of the master regulators Sox2, Sox3, Musashi 1, and Pax6, and endpoint markers Tubb3 and Syp (Fig. 2F). For each of these genes, there were significant differences in the levels of expression among the mtDNA haplotypes. The most divergent line, CC9dunni, exhibited significantly less expression of each of the endpoint markers, Tubb3 and Syp. These results demonstrate lineage-specific biases among the mtDNA haplotypes.

Cells Harboring Divergent mtDNA Haplotypes Display Mild Respiratory Defects Following 18 Days of Differentiation

Since we detected no difference in energy metabolism on day 7 of differentiation but increased variation in chromosomal gene expression, we analyzed energy metabolism during the later stages of differentiation. On day 18 of differentiation, there were no significant differences in the ATP levels, although CC9dunni cells had less than the other two cell lines (Fig. 3A). There were also no significant differences for resting O2 consumption (basal) rates, ETC coupling rates (basal/nonphosphorylating respiration), and ETC capacity (uncoupled/basal respiration) among the three cell lines (Fig. 3B). However, lactate production was increased in CC9dunni cells compared to CC9mus and CC9spretus cells (p < .01; Fig. 3C), suggesting a greater utilization of glycolysis. These findings are consistent with increased lactate production, without measureable OXPHOS defects, as observed in CC9dunni fibroblast cybrids when compared to control cybrids [41].

thumbnail image

Figure 3. Cells harboring divergent mitochondrial genome haplotypes display mild respiratory defects. (A): ATP content in CC9mus, CC9spretus, and CC9dunni cells on day 18 of differentiation. (B): Oxygen consumption of differentiated (day 18) CC9mus, CC9spretus, and CC9dunni cells. The data show basal oxygen consumption, nonphosphorylating respiration (5 mg/ml oligomycin), and uncoupled respiration (100 nM FCCP). (C): Lactate content of culture media for CC9mus, CC9spretus, and CC9dunni cells on day 18 of differentiation. Significant differences between cell types are: **, p < .01. Abbreviation: ETC, electron transfer chain.

Download figure to PowerPoint

The 3D Environment Maintains the Influence of the mtDNA Haplotypes on Chromosomal Gene Expression During Early Differentiation

So far, we have shown that divergent mtDNA haplotypes alter the expression profiles of genes associated with pluripotency, differentiation and DNA methylation. In order to determine whether the culture microenvironment influenced the differentiation potential of the different mtDNA haplotypes, we examined all three cell lines in a 3D environment using nonwoven electrospun PCL scaffolds [53]. As pluripotent gene expression is altered during early differentiation, we first assessed the expression of the genes of pluripotency. We differentiated CC9mus, CC9spretus, and CC9dunni in hanging droplets for 4 days and then seeded whole EBs onto gelatin-coated 2D polystyrene culture dishes and gelatin-coated 3D electrospun scaffolds for a further 3 days (i.e., total of 7 days differentiation). In the 2D environment, Oct4 (p < .01), Sox2 (p < .05), and Nanog (p < .05) were elevated in CC9spretus cells when compared with the other cell types (Fig. 4A). In the 3D environment, the expression of Oct4 was increased in the CC9dunni cells compared to the CC9mus cells (Fig. 4A). There were elevated levels of expression for Oct4, Sox2, and Nanog for each of the lines in the 3D environment when compared to the 2D environment (Fig. 4A), although these levels were significantly reduced compared to undifferentiated ESCs (p < .001).

thumbnail image

Figure 4. Embryonic stem cells (ESCs) harboring divergent mitochondrial genome haplotypes demonstrate altered differentiation profiles on 3D scaffolds. Real time polymerase chain reaction (PCR) analysis of CC9mus, CC9spretus, and CC9dunni cells differentiated for 7 and 18 days on 2D polystyrene and 3D electrospun scaffolds. Day 7 samples were analyzed for the expression of Oct4, Sox2, and Nanog (A), and Dnmt1, Dnmt3a, and Dnmt3b (B). Samples taken on day 18 of differentiation were analyzed for the expression of Ibps, Myl3, and Myh7 (C); Krt14 and Gad1 (D); Scl2a2 and Sftpd (E). All values represent mean ± SEM. Real time PCR gene expression levels are expressed relative to CC9mus cells in the 2D culture environment, at that particular time point. Significant differences between cell types are: *, p < .05; **, p < .01; ***, p < .001.

Download figure to PowerPoint

We then determined the levels of expression of the DNA methyltransferases. In the 2D environment, Dnmt1 (p < .001) was higher for the CC9mus than the CC9spretus and CC9dunni haplotypes (p < .001), while Dnmt3a was only higher when compared to CC9spretus cells (Fig. 4B). For Dnmt3b, the CC9dunni haplotype was higher than the other haplotypes (p < .001). In the 3D environment, Dnmt1 (p < .001), Dnmt3a (p < .001), and Dnmt3b (p < .001) were elevated compared to the 2D environment for the CC9mus and CC9spretus haplotypes. However, for the CC9dunni haplotype, the levels of expression of Dnmt1 (p < .001) and Dnmt3a (p < .001) were elevated, but Dnmt3b (p < .001) expression was reduced (Fig. 4B). These outcomes indicate that pluripotent genes and DNA methyltransferases for the different mtDNA haplotypes are responsive to different environments during the early stages of differentiation.

The 3D Environment Alters the Differentiation Potential of the mtDNA Haplotypes During Late Stage Differentiation

In order to further understand how the mtDNA haplotypes influence gene expression in the 3D environment, we continued to differentiate the three lines in vitro. We selected seven genes (three mesoderm, two endoderm, and two ectoderm) and analyzed their expression in CC9mus, CC9spretus and CC9dunni cells on day 18 of differentiation. In the 2D environment, the levels of expression for the mesodermal genes, Ibps and Myh7, were elevated in CC9spretus cells compared to the CC9mus and CC9dunni cells, while the CC9dunni cells were lower than the CC9mus cells. In the 3D cultures, higher levels of expression were identified for CC9dunni cells, while the CC9spretus cells were lower than the CC9mus cells (Fig. 4C). Similar patterns of gene expression were also observed for the ectodermal genes, Krt14 and Gad1 (Fig. 4D), and the endodermal gene, Scl2a2 (Fig. 4E). Finally, the expression of Sftpd (endoderm) in the 2D environment was reduced for the CC9spretus and CC9dunni cells, while in the 3D environment there was an overall decrease in expression (Fig. 4E). These data demonstrate that variation in mtDNA haplotypes can result in variable patterns of gene expression and that these patterns are present in cells cultured in 2D and 3D environments. However, the 2D and 3D outcomes are dissimilar. For the DNA methyltransferases on day 18 of differentiation, we observed increases in expression for each gene in the 3D environment (Fig. 5A), with increased levels of expression for Dnmt1 (p < .001), Dnmt3a (p < .001), and Dnmt3b (p < .001) in the CC9dunni cells relative to CC9mus cells and CC9spretus; and for CC9spretus compared to CC9mus cells for Dnmt3b (Fig. 5A).

thumbnail image

Figure 5. Electrospun 3D scaffolds increase the expression of OXPHOS subunits. Real time polymerase chain reaction (PCR) analysis of CC9mus, CC9spretus, and CC9dunni cells differentiated for 18 days on 2D polystyrene and 3D electrospun scaffolds. Samples taken on day 18 of differentiation were analyzed for the expression of: (A) Dnmt1, Dnmt3a, and Dnmt3b; (B) Ndufb5, Sdhb, Uqcrc2, Cox5b, and Atp5o; (C) ND1, CytB, COXI, and ATPase6; (D) Glut1 and Gapdh. All values represent mean ± SEM. Real time PCR gene expression levels are expressed relative to CC9mus cells in the 2D culture environment, at that particular time point. Significant differences between cell types are: *, p < .05; **, p < .01; ***, p < .001.

Download figure to PowerPoint

MtDNA Haplotypes Alter the Expression of Nuclear- and mtDNA-Encoded Subunits of the ETC During Differentiation

Due to the differences between mtDNA haplotypes and their energy metabolism and expression of differentiation markers, we analyzed the levels of expression of the nuclear- and mtDNA-encoded subunits of the ETC on day 18 of differentiation of 2D and 3D cultures. There was a high degree of variation in the expression profiles for these genes between the mtDNA haplotypes. All genes were elevated in the 3D environment (Fig. 5B, 5C). Increasing mtDNA divergence altered the expression of the nuclear-encoded subunits, Uqcrc2 (p < .01) and Cox5b (p < .01) in the 3D environment (Fig. 5B). In the 2D environment, CC9spretus cells expressed lower levels of ND1 (p < .001), CytB (p < .001), and COXI (p < .001), whereas in the 3D environment, ND1 (p < .001) and ATPase6 (p < .001) levels of expression were elevated (Fig. 5C). Furthermore, the expression of ND1 (p < .001), CytB (p < .001), ATPase6 (p < .001), and COXI (p < .001) was significantly higher in the CC9dunni cells in the 3D environment (Fig. 5C). Given that increased genetic divergence between the nuclear- and mtDNA-encoded subunits of the ETC reduces the efficiency of ATP production [39, 41], we analyzed the expression of Glut1 and Gapdh. For Glut1, there was increased expression in the 3D environment for CC9dunni cells. Again, in the 3D environment, there was increased Gapdh expression for all three mtDNA haplotypes (Fig. 5D), with the CC9dunni cells producing a 20-fold increase. These findings show that there is a greater demand for upregulation of genes associated with OXPHOS and glycolysis in the 3D environment and that each of the haplotypes can respond accordingly.

MtDNA Haplotypes Influence Overall Gene Expression Profiles at Late Stage Differentiation

In order to determine the overall effects of the three mtDNA haplotypes on gene expression, we assessed the relative mean expression values for all the genes analyzed on days 7 and 18 of differentiation. On day 7 of differentiation, no differences were observed between CC9mus, CC9spretus, or CC9dunni cells in either the 2D or 3D environments (Fig. 6A, 6B). However, on day 18 of differentiation when cultured in 2D, the CC9spretus and CC9dunni cells displayed significantly reduced levels of gene expression (p < .001; Fig. 6C). However, in the 3D environment, the CC9spretus and CC9dunni cells had significantly altered gene expression (p < .01; Fig. 6D), with CC9dunni cells overall having higher levels of expression.

thumbnail image

Figure 6. Mitochondrial DNA haplotypes increase differential gene expression during differentiation. The mean relative real time polymerase chain reaction (PCR) expression values shown in Figures 4 and 5 were analyzed collectively. (A): Data points represent Log-mean values for gene expression on day 7 of differentiation in 2D and (B) 3D environments. (C): Data points represent Log-mean values for gene expression on day 18 of differentiation in 2D and (D) 3D environments. Significant differences between cell types are: *, p < .05; **, p < .01; ***, p < .001.

Download figure to PowerPoint

Coculture of ESCs with Differing mtDNA Haplotypes

To further determine how mtDNA haplotypes influence ESC differentiation, we cocultured CC9spretus and CC9dunni cells with CC9mus cells during differentiation at ratios of 80:20, 50:50, and 20:80 and cultured these combinations in a 3D environment. All combinations of CC9spretus-CC9mus cells expressed elevated levels of Ibps, Myh7, and Sftpd1 compared to CC9spretus cells during differentiation in 3D (Fig. 7). Furthermore, the 50:50 ratio of CC9spretus-CC9mus cells also increased the expression of Krt14 (p < .001) and Gad1 (p < .05), while the 20:80 ratio of CC9spretus-CC9mus cells expressed lower levels of Scl2a2 (p < .05; Fig. 7). It is evident for the CC9dunni-CC9mus combinations that there are increased levels of expression for Myl3 and Sftpd1, while the expression of Scl2a2 and Gad1 were reduced, when compared to CC9dunni differentiation alone (Fig. 7). The expression of Dnmt3a was also decreased for both CC9spretus and CC9dunni cells combined with CC9mus cells (supporting information Fig. S3), while increased CC9mus content (50:50 and 20:80) had the opposite effect for Dnmt3b (supporting information Fig. S3). Finally, analysis of Dnmt1 expression was only altered for the CC9dunni-CC9mus combinations (supporting information Fig. S3).

thumbnail image

Figure 7. Coculture of embryonic stem cells (ESCs) harboring divergent mtDNA haplotypes demonstrate altered differentiation profiles in a 3D environment. Differentiation was initiated using cocultures of CC9mus with CC9spretus or CC9dunni cells at 80/20, 50/50, and 20/80 ratios in embryoid bodies (EBs) for 4 days, at which point EBs were cultured on gelatin-coated 2D polystyrene or 3D electrospun scaffolds. On day 18 of differentiation, samples were analyzed for the expression of mesodermal (Ibps, Myl3, Myh7), ectodermal (Krt14, Gad1), and endodermal (Scl2a2, Sftpd) cell lineage specification markers. Real time PCR gene expression levels are expressed relative to their respective haplotypes (CC9spretus or CC9dunni) at 100% in the 3D culture environment. All values represent mean ± SEM. Significant differences between cell types are: *, p < .05; ** p < .01; *** p < .001.

Download figure to PowerPoint

Given that mtDNA haplotypes influence the expression of DNA methyltransferases and differentiation markers, we sought to determine if these coculture conditions would affect the expression of nuclear- and mtDNA-encoded genes associated with OXPHOS. The levels of expression of the nuclear-encoded Ndufb5, Cox5b, and Atp5o (supporting information Fig. S4A), as well as mtDNA-encoded ND1 and CytB (supporting information Fig. S4B) were significantly reduced for all combinations of CC9spretus and CC9dunni cells cocultured with CC9mus cells, when compared to their respective cohort cultured without CC9mus cells. The CC9spretus-CC9mus combination at 50:50 significantly increased the expression of COXI and ATPase6, although for the corresponding combination of CC9dunni-CC9mus cells, these genes were expressed at significantly reduced levels (supporting information Fig. S4B). Similarly, differences in Glut1 and Gapdh expression were observed between CC9spretus and CC9dunni cells cocultured with CC9mus (supporting information Fig. S4C). These outcomes demonstrate that there are distinct interactions between each of the mtDNA haplotypes, but their influence on gene expression profiles is dependent on their respective contributions.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

Over time, different mtDNA haplotypes have evolved that distinguish different breeds and strains of organisms. Several studies have suggested that mtDNA haplotypes confer positive and negative advantages to individuals, such as fertility [5, 6, 9] and disease [12–14]. For example, haplotype T is associated with decreased sperm motility in men suffering from infertility, while haplotype H is indicative of normal sperm motility [9]. However, the functional significance at a cellular level has so far been undetermined. To our knowledge, we demonstrate for the first time that different mtDNA haplotypes against the same nuclear background influence the expression of chromosomal genes involved in pluripotency, differentiation, mitochondrial energy metabolism, and the regulation of DNA methylation.

It is well-established that the strict control of the core-transcriptional regulators of pluripotency regulates lineage commitment at the onset of differentiation [54]. To this extent, as differentiation is initiated, pluripotent genes, such as Oct4, Sox2 and Nanog, are downregulated before the expression of stage-specific lineage genes takes place [55]. Furthermore, we have seen similar patterns for the pluripotency-associated genes Dppa5, Pramel7 and Ndp52L1, which were upregulated during days 1 and 2 of differentiation and downregulated during days 3 and 4 [15, 51]. Their levels of expression demonstrate a profile similar to days 2 and 3 of differentiation for the more divergent haplotypes. However, this appears to be compensated for by the upregulation of the transcription factor Klf4, known to regulate pluripotency [46], and Rex1, which loses expression abruptly as cells differentiate. Furthermore, the increased levels of expression of OCT4, SOX2, and NANOG during early differentiation in the CC9spretus and CC9dunni lines suggest that they may be slower to lose pluripotency. Added to this, we observed differences in levels of expression for the Dnmt genes for each of the haplotypes during pluripotency, the enrichment of 5mC in the promoter regions of Nanog and Dppa5, and of 5hmC in Oct4 and Rex1, along with the enrichment of OCT4 within the promoter region of Nanog, which indicate that differences in mtDNA haplotype subtly influence pluripotent gene regulation. However, it is very likely that the more divergent mtDNA haplotypes differentially induce other signaling pathways that impact on pluripotent gene regulation and thus account for the disparity between mtDNA methylation levels and pluripotent gene profiles.

The variability in pluripotent gene expression among the different mtDNA haplotypes resulted in subtle changes on day 7 of differentiation, which manifested into significant changes on day 18 of differentiation. We have shown this not only in a number of genes from all three lineages but also in the context of master regulators of gene differentiation that regulate the fate of cardiomyocyte and neural differentiation. Indeed, there are quite clearly different patterns of expression resulting from the different mtDNA haplotypes, which influence functional readouts such as the number of beating EBs generated. Consequently, it would appear that the slightest changes mediated by mtDNA haplotypes during pluripotency and the very earliest stages of differentiation cumulatively impact on downstream differentiation.

While these changes in the expression of pluripotency genes are significant, these ESCs are still amenable to the creation of viable xenomitochondrial mice [39]. Furthermore, both Mus spretus and Mus dunni mtDNA can be transmitted through the mouse germ line, with no overt effects on overall animal health [43]. Nevertheless, the expression of the immediate-early response genes Fos, Egr2, and Egr4 are downregulated in 3-week-old Mus dunni xenomouse brains, suggesting potential differences in early brain development [56]. In addition, the expression of the mtDNA-encoded gene CoxIII was upregulated in xenomouse brains, possibly as a compensatory mechanism to maintain mitochondrial respiratory activity at normal levels [56]. Although these xenomice show no signs of disease at a young age, it will be interesting to establish whether these early changes in gene expression contribute to age-related metabolic deficiencies as well as behavioral patterns in older Mus dunni xenomice.

The amplified variability in gene expression levels among the mtDNA haplotypes in later stages of differentiation can be explained in the context of Waddington's landscape model of differentiation. This model represents the differentiating cell as a marble stochastically traveling down a valley of branches and furrows, influencing the developmental pathways and restricting cellular plasticity [26]. Our data suggest that these changes in cell fate, which are essential to differentiation and development, are not only shaped by dynamic signaling pathways, transcription factors and epigenetic states but also are influenced by the mtDNA genotype of the cell. The impact of these changes is quite different to those arising from the introduction of different mtDNA haplotypes into depleted cancer cell lines, which can modulate transcription [57] and DNA methylation [58] in a mtDNA genotype-specific manner, as these cells have already committed to a specific fate.

Although a variety of in vitro differentiation protocols are performed on 2D polystyrene, we used 3D electrospun scaffolds in order to determine if the differences observed between haplotypes were intrinsic. For all mtDNA haplotypes, differentiation for 7 and 18 days in the 3D environment was associated with increased gene expression compared to the 2D environment. However, despite the increased gene expression observed in the 3D environment, mtDNA haplotypes continued to influence gene expression profiles during differentiation. Under these conditions, the expression patterns of differentiation markers, mtDNA- and nuclear-encoded OXPHOS subunits and regulators of DNA methylation were again mtDNA haplotype dependent. These observations further demonstrate that mtDNA plays an essential role during differentiation and development and is affected by its environment.

We further evaluated the influence of these haplotypes on chromosomal gene expression by combining different ratios of Mus spretus and Mus terricolor cybrids with Mus musculus ESCs prior to differentiation in the 3D environment. These ratios again altered the gene expression profiles during the later stages of differentiation and further highlight the effects of mtDNA haplotypes on chromosomal gene expression. This is of particular importance in human disease, as organs or tissues comprised of cells with different mtDNA haplotypes will have heterogeneous gene expression, potentially influencing tissue homeostasis and predisposing the tissue to a variety of complex disorders. This should also encourage us to reflect on whether allogeneic transplantation of stem cells harboring different mtDNA genotypes would always be beneficial to the individual or whether autologous transplantation is most suitable. Furthermore, it will help define the most appropriate sources of recipient oocytes for somatic cell-derived ESCs, as the recipient oocyte is often randomly chosen with little thought given to the implications of its mtDNA genotype [59].

The mechanisms behind the phenotypic consequences of mtDNA variation identified by others [60] and demonstrated here remain to be determined. In pluripotent ESCs harboring Mus terricolor mtDNA (CC9dunni), mtDNA copy content was lower than in Mus musculus ESCs. This difference may be attributed to genetic variance within the mitochondrial transcription factor A (TFAM) binding sites in the mtDNA [42], where discrete base changes can modify the expression of the mtDNA-encoded genes and mtDNA replication through altered binding efficiency of TFAM [61]. The regulation of mtDNA copy number during differentiation is essential; failure to restrict replication during pluripotency and early differentiation and the accumulation of the appropriate numbers of mtDNA copy during later stages of differentiation will reduce the ability of a cell to meet its specific demands for ATP, thus impacting on cell fate and function [15, 16]. However, in our experiments, the differences in mtDNA content seem an unlikely contributing factor to the variation in gene expression during differentiation, since the differences in gene expression between CC9spretus and CC9mus ESCs are not reflected by any differences in mtDNA content. Nevertheless, we identified altered expression of nuclear- and mtDNA-encoded OXPHOS subunits, which collectively could impact on respiratory capacity, thus modulating gene expression.

The differentiation of ESCs into cardiomyocyte progenitors is supported by anaerobic metabolism [62], with a shift to OXPHOS and decreased glycolysis during later stages of differentiation [24, 62]. Increased lactate production in differentiated cells with the Mus terricolor mtDNA haplotype may be attributed to subtle differences in respiratory output. However, we only observed changes in lactate production on day 18 of differentiation, downstream of cardiomyocyte specification. Furthermore, the Mus spretus haplotype produced fewer beating EBs compared to Mus musculus and the more divergent Mus terricolor mtDNA haplotypes, without any discernible difference in metabolic productivity. It is important to note that no differences in respiratory chain function in somatic cell cybrids constructed from Mus musculus and Mus spretus have been reported [38, 39].

The changes in the expression of the chromosomal- and mitochondrial-encoded genes induced by the mtDNA haplotypes likely alter the discrete changes in reactive oxygen species [63] and nitric oxide [64], which at low levels mediate intracellular signaling pathways. As a consequence, differentiating cells will commit to specific lineages accordingly, such as demonstrated for neuronal differentiation [65]. Additionally, both nitric oxide and reactive oxygen species have been implicated in the differentiation of ESCs into cardiomocytes [30, 66] and could account for the significant differences we observed for each mtDNA haplotype.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

To our knowledge, this is the first demonstration that mtDNA haplotypes influence chromosomal gene expression and determine the cell fate of a pluripotent ESC as it undergoes differentiation. These outcomes are significantly more pronounced in a 3D environment. Our outcomes indicate that the crosstalk between mtDNA and the nucleus is bidirectional and it is therefore not just a case of the chromosomal genome regulating the transcription, replication and translation of the mtDNA in order that cells generate sufficient energy to perform their specified functions. It is now evident that the mtDNA is integral to defining the phenotype of a cell from the earliest point of differentiation. These findings explain the associations of mtDNA variants with specific traits that confer selective advantages to the individual.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

This work was supported by the Victorian Government's Operational Infrastructure Support Program; Monash Institute of Medical Research start-up funds to J.C.St.J.; an NHMRC CDA Fellowship and the James and Vera Lawson Trust to M.McK.; a Monash Larkins Program and an NMHRC CDF to J.M.P.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICT OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
sc-12-0681_sm_SupplFigure1.tif2087KSupporting Figure S1
sc-12-0681_sm_SupplFigure2.tif789KSupporting Figure S2
sc-12-0681_sm_SupplFigure3.tif2642KSupporting Figure S3
sc-12-0681_sm_SupplFigure4.pdf1021KSupporting Figure S4
sc-12-0681_sm_SupplFigureLegends.pdf84KSupporting Figure Legends
sc-12-0681_sm_SupplInfo.pdf194KSupporting Information
sc-12-0681_sm_SupplTables.pdf31KSupporting Tables

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.