Abstract
Vitamin C / L-ascorbic acid is a critical co-factor for a wide range of enzymes including the TET (Ten-Eleven Translocase) enzymes. The TET enzymes are involved in the oxidation of 5-methylcytosine in DNA (5mC) to 5-hydroxymethylcytosine (5hmC) and subsequently to 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC). These reactions are steps in the pathway of active DNA demethylation. Accordingly, these enzymes are important regulators of the epigenome. In my thesis, I have explored how vitamin C availability modulates the TET-enzyme catalysed demethylation of 5-methylcytosine and how this may translate into altered gene expression in different contexts. This mechanism was investigated in two different experimental models. I investigated how states of vitamin C deficiency altered TET activity in cultured leukaemia cells and also whether vitamin C deficiency during pregnancy disrupted epigenetic programming during guinea pig development.
I observed that the addition of either ascorbate or redox-stable phosphoascorbate to cell culture media altered the phenotype of the SKM-1 acute monoblastic leukaemia cell line. At the transcriptional level, ascorbate supplemented SKM-1 cells exhibited an altered transcriptome, with over 1000 differentially expressed genes relative to untreated cells. These differentially expressed genes were highly enriched for monocyte specific genes, suggesting a change in cell function mirroring differentiation. At the protein level, I also observed changes in monocyte specific cell surface markers, which can also be interpreted as representing a more differentiated phenotype.
TET2 has a critical role in haematopoiesis, regulating the differentiation and self-renewal properties of haematopoietic stem cells. Accordingly, I sought to investigate whether increased TET activity, by providing these cells with vitamin C, might be a driver of this altered phenotype. SKM-1 cells have extremely low levels of global 5hmC relative to human tissues, a phenomenon that is apparent in other cell lines. Providing SKM-1 cells with ascorbate or phosphoascorbate resulted in a time and concentration dependent increase in 5hmC. Global 5hmC increased from around 0.002% - 0.003% of total cytosine species to around 0.013% – 0.024% with ascorbate or phosphoascorbate supplementation. While baseline 5hmC increased, ascorbate supplementation did not induce large changes in global or CpG island methylation. Furthermore, no correlation was apparent between purported methylation changes at transcription start site CpG islands and the expression of differentially expressed genes. This analysis suggested that the transcriptional response to ascorbate in cell culture was directed by mechanisms that did not involve a change in methylation.
To further understand the effects of TET activity manipulation in leukaemia cells, I treated SKM-1 and THP-1 leukaemia cells with the novel TET inhibitor, TETi76 diethyl ester. I confirmed that TETi76 was preferentially cytotoxic towards TET2 heterozygous mutant SKM-1 cells compared with THP-1 TET2 wild type cells. However, TETi76 did not decrease global 5hmC in SKM-1 cells, raising doubts as to whether the molecule was acting as a TET inhibitor. Further, I discovered that sublethal doses of TETi76 appeared to induce SKM-1 differentiation along the monocytic trajectory with transcriptional similarities to the effects of ascorbate. This was an unexpected finding, and the mechanism underlying this observation is unknown.
I also studied how vitamin C deficiency during pregnancy altered the transcriptomes and methylomes of offspring in a guinea pig model. In this model of maternal vitamin C deficiency, guinea pig offspring from low vitamin C pregnancies had significantly lower tissue and plasma vitamin C concentrations at birth. An analysis of global levels of cytosine species in these animals highlighted an organ specific 5hmC distribution, with brain tissues having significantly higher levels than other tissues. However, maternal vitamin C deficiency resulted in no appreciable change in global methylation or hydroxymethylation as measured with mass spectrometry.
I then applied higher resolution bisulphite sequencing methods to identify a methylation signature in animals from low vitamin C pregnancies. I initially confirmed that the TET-oxidation product 5hmC was highly enriched at CpG islands with intermediate methylation density in the guinea pig genome – demarcating these regions as sites of TET-mediated demethylation. In newborn guinea pig offspring from low vitamin C pregnancies, increased methylation was apparent across CpG islands with intermediate methylation density. This effect was also apparent in DNA extracted from the liver, adrenal glands or hippocampus and suggested an impairment of TET-mediated demethylation during early development. The strongest evidence in support of the hypermethylation phenotype came from an analysis of 10 DNA samples from the newborn male hippocampus with whole genome bisulphite sequencing. This comparison indicated that animals from low vitamin C pregnancies had a ~2.5% increase in CpG cytosine modification across intermediate methylation density CpG islands compared with animals from optimal vitamin C pregnancies. I did not detect large transcriptional changes in offspring from vitamin C deficient pregnancies, despite analysing 183 guinea pig transcriptomes. Small transcriptional changes due to vitamin C deficiency were however detected in the newborn liver, with 47 differentially expressed genes including a 2-fold upregulation of superoxide dismutase 2.
In performing bioinformatics analysis of guinea pig sequencing data, I overcame the limitations of the poorly annotated guinea pig genome by calculating and validating the positions of CpG islands. This allowed me to perform differential methylation analysis to detect tissue and sex-specific differences in CpG island methylation in the guinea pig genome, including clear evidence of X-inactivation via CpG island methylation. I also observed a profound transcriptional difference (2595 differentially expressed genes) between the livers of juvenile and newborn guinea pigs, with an upregulation of several CYP450 enzymes – possibly reflecting a metabolic response to a post-weaning change in diet.
The work in my thesis builds on the concept that TET activity is highly sensitive to vitamin C availability. The properties of leukaemia cells are altered by the addition of non-cytotoxic quantities of ascorbate to the cell culture media, with evidence of increased TET activity, transcriptional changes and altered growth rates. This raises the important consideration of whether vitamin C should be added to cell culture media for more physiologically relevant studies. Further, non-scorbutic, vitamin C deficiency during guinea pig pregnancy resulted in a clearly detectable legacy in the form of offspring DNA hypermethylation. If this effect translates to humans, this may have clinical implications.