Folate metabolite profiling of different cell types and embryos suggests variation in folate one-carbon metabolism, including developmental changes in human embryonic brain
- First Online:
- Cite this article as:
- Leung, KY., De Castro, S.C.P., Cabreiro, F. et al. Mol Cell Biochem (2013) 378: 229. doi:10.1007/s11010-013-1613-y
- 1.3k Downloads
Folates act as co-factors for transfer of one-carbon units for nucleotide production, methylation and other biosynthetic reactions. Comprehensive profiling of multiple folates can be achieved using liquid chromatography tandem mass spectrometry, enabling determination of their relative abundance that may provide an indication of metabolic differences between cell types. For example, cell lines exposed to methotrexate showed a dose-dependent elevation of dihydrofolate, consistent with inhibition of dihydrofolate reductase. We analysed the folate profile of E. coli sub-types as well as cell lines and embryonic tissue from both human and mouse. The folate profile of bacteria differed markedly from those of all the mammalian samples, most notably in the greater abundance of formyl tetrahydrofolate. The overall profiles of mouse and human fibroblasts and mid-gestation mouse embryos were broadly similar, with specific differences. The major folate species in these cell types was 5-methyl tetrahydrofolate, in contrast to lymphoblastoid cell lines in which the predominant form was tetrahydrofolate. Analysis of embryonic human brain revealed a shift in folate profile with increasing developmental stage, with a decline in relative abundance of dihydrofolate and increase in 5-methyl tetrahydrofolate. These cell type-specific and developmental changes in folate profile may indicate differential requirements for the various outputs of folate metabolism.
KeywordsFolate Embryo Methotrexate Bacteria Liquid chromatography tandem mass spectrometry
Abnormalities in FOCM have been implicated in a variety of pathological conditions including cancer, neural tube defects (NTDs), cardiovascular disease, anaemia and neurological conditions [3, 8, 9, 10, 11]. Notably FOCM is implicated in diseases that may occur throughout life, including birth defects that arise during early development. For example, maternal supplementation with folic acid reduces the risk of NTDs whilst sub-optimal folate status and elevated homocysteine are associated with increased predisposition to an affected pregnancy (reviewed in ). The observation of defective thymidylate biosynthesis in some human NTD cases and mouse models [12, 13, 14] supports the hypothesis that altered FOCM may contribute to development of NTDs. Folate status has also been investigated in relation to risk of several different cancers, colorectal cancer being probably the most intensively studied. Epidemiological studies have shown an association of low folate intake with risk of colorectal adenomas and cancer, and use of folic acid supplements has been reported to reduce risk and mortality. However, folic acid supplements may also promote progression of pre-existing preneoplastic lesions, thereby increasing cancer risk [10, 15]. A similar non-linear relationship, in which both low and high folate intake are associated with risk, has also been reported for postmenopausal breast cancer .
The multiple outputs of FOCM imply several different biochemical mechanisms by which impaired function may influence embryonic development and post-natal pathologies. Nucleotide biosynthesis is essential to support DNA replication and cell proliferation, which may be of particular relevance to embryonic development, whilst impaired thymidylate biosynthesis causes an increase in erroneous incorporation of uracil into nuclear DNA . The supply of methyl groups is also required for cellular methylation reactions and may impact DNA methylation, possibly leading to altered transcriptional regulation. In this context, there is increasing evidence of a potential link between FOCM and the foetal epigenome, although whether altered methylation contributes to birth defects has yet to be established .
Alterations in FOCM can be investigated by analysis of downstream biomarkers such as homocysteine concentration, DNA methylation and uracil incorporation . Mathematical modelling also allows predictions of the effects of folate status or enzyme inhibition on methylation and nucleotide biosynthesis [18, 19]. It will also be informative to directly quantify FOCM intermediates to evaluate possible alteration in the ratio of abundance of individual metabolites that may reflect disturbance of a particular step(s). For example, 10-formyl-THF provides one-carbon units for purine biosynthesis and a reduced abundance of this metabolite, relative to its precursor THF, could lead to diminished purine synthesis. Similarly, alteration in the relative abundance of 5-methyl THF may be informative about the contribution of methyl groups to the methylation cycle.
Folates are transported into cells in the monoglutamated form and multiple glutamate residues are then added by the action of folylpolyglutamate synthetase. Polyglutamation is required for cellular retention of folates and an optimal analytical approach should therefore distinguish these forms. Previously, we used liquid chromatography tandem mass spectrometry (LC–MS/MS) for quantification of s-adenosyl methionine (SAM) and s-adenosylhomocysteine (SAH) in order to identify perturbations in the methylation cycle [20, 21]. In the current study, we applied LC–MS/MS to analyse FOCM intermediates in bacteria and for comparison of human and mouse cell lines and embryonic tissue, which have not previously been subject to comprehensive analysis.
Materials and methods
Folate standards including dihydrofolate (DHF), tetrahydrofolate (THF), 5,10-methylene THF, 5-formyl-THF, 5,10-methenyl THF, 5-methyl THF, folic acid (PteGlu), PteGlu3, PteGlu4, PteGlu5, PteGlu7 and methotrexate were purchased from Schircks Laboratories (Switzerland). Remaining reagents were methanol, acetonitrile (Fisher Scientific, UK), ammonium acetate, ascorbic acid, DTT, citric acid (Sigma-Aldrich) and N,N-dimethylhexylamine (Fluka).
Samples for analysis
Bacteria OP50 and HT115 Escherichia coli strains [22, 23, 24] were grown overnight in LB from a single colony at 37 °C. NGM plates  were seeded with 150 μL bacterial suspension and incubated for 96 h at 20 °C. Bacterial lawns were washed from the plates using M9, collected by centrifugation at 4 °C, 4,000 rpm for 20 min and the bacterial pellet stored at −80 °C prior to analysis. Cell lines EBV-transformed human lymphocytes were collected with ethical permission from normal Swedish individuals (Karolinska Institutet). Cells were cultured in RPMI 1640 media with 10 % FCS. For LC–MS/MS analysis, 2 × 107 cells were harvested, washed in PBS and cell pellets stored at −80 °C prior to sample preparation. Human tissue was obtained as frozen samples (at −80 °C) from the human developmental biology resource (www.hdbr.org). Mice wild-type (CBA/Ca and C57BL/6) strain mice were mated and mouse embryos were collected at embryonic day (E) 12.5 . Embryos were immediately frozen on dry ice and stored at −80 °C.
Buffer containing 20 mM ammonia acetate, 0.1 % ascorbic acid, 0.1 % citric acid and 100 mM DTT at pH 7 was added to cells, tissues or embryos. Where quantitation was performed, 3 μl of 10 μM methotrexate was added as internal standard. Sample suspensions were sonicated for 10 s using a hand-held sonicator at 40 % amplitude. A 10 μl aliquot of each of the homogenised samples was removed for DNA quantification using NanoDrop (thermo scientific). Protein was removed by precipitation by addition of 2 sample volumes of acetonitrile, mixing for 2 min and centrifugation for 15 min at 12,000×g and 4 °C. Supernatants were transferred to fresh tubes, lyophilised and stored at −80 °C prior to analysis.
Lyophilised samples were resuspended in 50 μl water (milli-Q) and centrifuged for 5 min at 12,000×g at 4 °C. Supernatants were transferred to glass sample vials for LC–MS/MS analysis. Metabolites were resolved by reversed-phase chromatography (Luna C18 column; 150 × 2.0 mm2; 5 μm bead size; Phenomenex, UK) using a 2795XE high performance liquid chromatography unit with solvent divert valve (Waters Corporation, UK). Solvents for HPLC were: Buffer A, 5 % methanol, 95 % Milli-Q water and 5 mM dimethylhexylamine at pH 8.0; Buffer B, 100 % methanol, 5 mM dimethylhexylamine. The column was equilibrated with 95 % Buffer A: 5 % Buffer B. The sample injection volume was 40 μl. The HPLC protocol consisted of 95 % Buffer A: 5 % Buffer B for 1 min, followed by a gradient of 5–60 % Buffer B over 9 min and then 100 % Buffer B for 6 min before re-equilibration for 4 min. The metabolites were eluted at a flow rate of 200 nl/min. The HPLC was coupled to a triple quadrupole tandem mass spectrometer (MicroMass Quattro, Waters Corporation, UK) operating in negative-ion mode using the following settings: capillary 3.54 kV, source temperature 150 °C, desolvation temperature 350 °C, cone gas flow rate 25 l/h and desolvation gas flow rate 950 l/h. Folates were measured by multiple reaction monitoring (MRM) with optimised cone voltage and collision energy for precursor and product ions (based on ).
Data analysis and statistics
The peak areas of individual folates were extracted using MassLynx software (Waters). The total peak area of each folate (sum of mono- and polyglutamated forms) was expressed as a percentage of the total folate in each sample. Data were analysed using SigmaStat (Systat Software, Version 3.5). Pairwise comparisons were made by t test, and multiple comparisons by one way ANOVA with Holm-Sidak post hoc test.
MRM values for folate standards
MRM transition (m/z)
472.17 > 314.95
458.19 > 328.89
456.28 > 326.98
454.00 > 281.00
444.20 > 314.85
441.97 > 175.99
Folic acid: mono and polyglutamates
PteGlu (folic acid)
440.05 > 310.86
698.37 > 421.99
827.52 > 422.02
956.67 > 422.03
606.84 > 422.02
In contrast to comparisons between fibroblasts and mouse embryos, the folate profiles of these cell and tissue types differed markedly from those observed in lymphoblastoid cell lines. The principal difference was the predominance of 5-methyl THF as the major folate species in fibroblasts and embryos, compared with THF in lymphoblastoid cells (Fig. 2b, c).
In addition to comparison of the total abundance of each major folate, we also analysed glutamation patterns. In bacterial samples, multiple glutamated forms of each folate including all types from Glu2 to Glu7 were present in varying proportions (Fig. 3a shows 5-methylTHF in OP50 strain as a representative example). The most abundant polyglutamated form in bacteria was somewhat variable between different folates, but in general corresponded to Glu3, Glu4, Glu6 and Glu7. The Glu2–Glu7 forms of 5-methyl THF were also detected in mouse embryos, the major forms being Glu5 and Glu6 (Fig. 3b). However, in contrast to bacteria, the monoglutamate also made up a sizeable proportion (~20 %) of 5-methyl THF in embryos. Analysis of the glutamation state of several folates suggested that abundance of monoglutamate is principally a feature of 5-methyl THF, as the predominant forms of the other folates were the 5–7 glutamated forms (Fig. 3c).
Folate one-carbon metabolism plays a fundamental role in key cellular functions including nucleotide biosynthesis and methylation and is implicated in a number of diseases. Comprehensive profiling of folates within a cell or tissue type will provide an opportunity to investigate disturbance of metabolism and potentially to focus attention on particular step(s) that may be abnormal in a disease state or following treatment with pharmacological agents. Moreover, folate profiling will provide opportunities to investigate the metabolic effects of targeted mutation of FOCM enzymes in model systems. This may include loss-of-function alleles, a number of which have already been generated in mice, as well as specific alleles designed to mimic putative disease-causing mutations in humans.
The sensitivity of LC–MS/MS profiling to detect gross disruption of FOCM was tested by exposure of lymphoblastoid cells to methotrexate, a DHFR inhibitor. As predicted, the level of DHF was markedly increased in treated cells, a finding which parallels the elevation in DHF levels observed in lymphoblasts derived from patients with inherited DHFR deficiency .
LC–MS/MS methodology has previously been used for simultaneous profiling of folate metabolites in samples from different plant species . In the current study, we used an LC–MS/MS approach to perform comprehensive folate profiling in bacterial sub-types as well as mouse and human cell lines and embryonic tissue. Each of the analysed folates was detectable in each of the different types of samples but we observed notable variation in the relative abundance of the different folate metabolites. Perhaps unsurprisingly, the overall folate profile substantially differed between bacterial and mammalian samples, in particular with a greater proportion of formyl-THF in bacteria. These differences may reflect, in part, the variation in FOCM enzymes in bacteria and mammalian cells. In particular, in eukaryotes, the trifunctional enzyme MTHFD1 (methylene tetrahydrofolate dehydrogenase/methylene tetrahydrofolate cyclohydrolase/formyltetrahydrofolate synthetase) catalyses the interconversion of 5,10-methylene THF, 5,10-methenyl THF and 10-formyl-THF [1, 2]. However, in bacteria, these reactions are catalysed by two or three separate enzymes with conversion of THF to 10-formyl-THF by a monofunctional synthetase enzyme. In addition to possible variation in regulation of FOCM, differences in folate profile likely reflect the fact that, unlike mammalian cells, bacteria can synthesise folates de novo.
The overall folate profile patterns were similar in human and mouse fibroblast cell lines and mouse embryos, despite differences in the relative abundance of specific folate metabolites. In contrast, lymphoblastoid cell lines exhibited a rather different profile, with a significantly higher THF and lower 5-methyl THF than in fibroblasts. The metabolic basis of this variation in profiles has not been defined but computational modelling indicates that reduced activity of MTHFR or elevated activity of methionine synthase in lymphoblasts compared with fibroblasts could result in the observed difference in profile. The functional significance of differing profiles could be hypothesised to reflect the metabolic requirements of different cell types. For example, diminished MTHFR activity could indicate channelling of one-carbons towards nucleotide biosynthesis at the expense of methylation in lymphoblasts.
We also observed developmental shifts in folate profile in human embryonic brain, with an overall decline in DHF and THF and a corresponding increase in 5-methyl THF during the period from around 6 to 9 weeks of gestation. This change correlates with the onset of neuronal migration in the primate embryonic brain  and could potentially be effected through increased activity of MTHFR (or reduced activity of methionine synthase) with development. Increased availability of 5-methyl THF could potentially support methylation reactions at later stages as the overall behaviour of the cellular population shifts from proliferation towards differentiation.
In recent years, it has become increasingly clear that FOCM is compartmentalised at the sub-cellular level and that this is likely to be functionally important in regulation of flux through different pathways [2, 31]. For example, enzymes required for de novo thymidylate biosynthesis (SHMT1, TYMS and DHFR) localise to the nucleus during S and G2/M phase of the cell cycle . Folates must, therefore, be present in the nucleus to supply one-carbon units for dTMP synthesis. Mitochondrial folate metabolism plays a crucial role in provision of one-carbon units as formate, to the cytoplasm (and nucleus). Thus, the majority of one-carbons used for nucleotide biosynthesis and methylation are derived from serine, glycine, dimethylglycine or sarcosine in mitochondria . Mitochondrial-specific FOCM-related enzymes include MTHFD1L and MTHFD2L and the glycine cleavage system (GCS). Mutations in GLDC or AMT, encoding components of the GCS, cause non-ketotic hyperglycinemia and predispose to NTDs [32, 33]. The identification of diseases associated with FOCM in specific sub-cellular compartments suggests that fractionation prior to LC–MS/MS analysis will be important in evaluating the metabolic consequences of putative causative mutations. Comprehensive metabolite profiling, in combination with predictive modelling, may then provide a useful tool to investigate downstream functional effects on FOCM and indicate possible routes to therapeutic intervention.
The authors are grateful to Kevin Mills and members of ICH Biological Mass Spectrometry Facility for helpful discussion and advice. This study was supported by Sparks [08ICH03], the MRC [J003794] and the Wellcome Trust [Grant No. 087525). The human embryonic and foetal material was provided by the Joint MRC (Grant No. G0700089)/Wellcome Trust (Grant No. 082557) Human Developmental Biology Resource (http://hdbr.org).
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.