The Nuclear DNA Base 5-Hydroxymethylcytosine Is Present in Purkinje Neurons and the Brain

See allHide authors and affiliations

Science  15 May 2009:
Vol. 324, Issue 5929, pp. 929-930
DOI: 10.1126/science.1169786


Despite the importance of epigenetic regulation in neurological disorders, little is known about neuronal chromatin. Cerebellar Purkinje neurons have large and euchromatic nuclei, whereas granule cell nuclei are small and have a more typical heterochromatin distribution. While comparing the abundance of 5-methylcytosine in Purkinje and granule cell nuclei, we detected the presence of an unusual DNA nucleotide. Using thin-layer chromatography, high-pressure liquid chromatography, and mass spectrometry, we identified the nucleotide as 5-hydroxymethyl-2′-deoxycytidine (hmdC). hmdC constitutes 0.6% of total nucleotides in Purkinje cells, 0.2% in granule cells, and is not present in cancer cell lines. hmdC is a constituent of nuclear DNA that is highly abundant in the brain, suggesting a role in epigenetic control of neuronal function.

To investigate Purkinje and granule cell nuclei, we took advantage of the fact that ribosomal proteins are assembled in the nucleolus of all cells. Consequently, transgenic mice containing ribosomal protein L10a fused to enhanced green fluorescent protein (EGFP)—for example, the Pcp2 bacTRAP and NeuroD1 bacTRAP lines that express in cerebellar Purkinje and granule neurons, respectively—contain fluorescent nucleoli (1, 2) (figs. S1, D and E, and S2, C and D). This allowed us to obtain ~95% pure preparations of Purkinje or granule cell nuclei by fluorescence-activated cell sorting. While determining the total amount of cytosine methylation in CpG content in Purkinje and granule cell genomic DNA, we consistently observed a significantly smaller amount of 5-methylcytosine (mC) in Purkinje DNA (Fig. 1C) and the presence of an unidentified spot (“x”) on thin-layer chromatography (TLC) plates (Fig. 1A). Treatment of the DNA samples overnight with a mixture of ribonucleases (RNases) A and T1, followed by a 1-hour incubation with RNase H and V1 (hydrolyzing single-stranded RNA, RNA-DNA hybrids, and double-stranded RNA, respectively), did not remove spot “x” (fig. S3A). When the sample was treated with RNase-free deoxyribonuclease, spot “x” disappeared, together with most of the signal from other spots (fig. S3A). Using the same labeling procedure on total cerebellar RNA, we did not observe spot “x.” Spot “x” was not 2'-deoxyuridine monophosphate, which could appear after cytosine deamination, because a hydrolysate of uracil containing DNA generated a spot that migrated substantially below spot “x” (fig. S3B). Fragmenting DNA with Fok I restriction endonuclease and incorporating [α-32P]dATP (2′-deoxyadenosine 5′-triphosphate) into the resulting 3′ end did not yield spot “x” on the resulting TLC plates, demonstrating that “x” is preferentially found in the context of xpG dinucleotides (fig. S3C). From the total number of nucleosides neighboring G, we estimated that “x” constitutes 0.59 ± 0.05% in Purkinje cell DNA and 0.23 ± 0.01% in granule cell DNA. We noticed that the actual increase in the amount of xpG is proportional to the decrease in mCpG (Fig. 1C). Considering the quantitative “x” relation with mC and that the abundance of the other nucleosides was not different between these cell types (Fig. 1C and Fig. 1 legend), we tested the possibility that “x” is 5-hydroxymethyl-2′-deoxycytidine, which is found in T-even bacteriophage DNA (3). The results showed that hmdC monophosphate generated a spot that comigrated with the “x” spot on a TLC plate (Fig. 2A). When a synthetic DNA hydrolysate was separated with reverse-phase high-pressure liquid chromatography (HPLC), hmdC eluted just after the cytosine peak, consistent with the published observations (4) (Fig. 2B). HPLC analysis of cerebellar genomic DNA resulted in a small, but reproducible, peak on the HPLC chromatogram in the same position (Fig. 2B). To provide definitive proof that “x” is hmC, we analyzed the corresponding fraction with high-precision mass spectrometry (MS). Mass spectra identified the presence of two ions, with a mass/charge ratio (m/z) of 142.06 ± 0.01 and 280.11 ± 0.02, which matched the theoretical isotopic molecular weights of ions derived from hmdC—m/z 142.06 and 280.09, respectively (Fig. 2C). MS collision–induced fragmentation of the hmdC corresponding fraction from synthetic DNA produced the same ions (fig. S4). Together, these data demonstrate the presence of hmC in mouse cerebellar DNA. We were unable to detect hmC in four different cell lines of mouse and human origin (fig. S5A). The distribution of hmC in mouse tissues displays the enrichment exclusively in the brain, with higher abundance in the cortex and brainstem (fig. S5B).

Fig. 1

Quantification of mC and “x” abundance in Purkinje and granule neurons. (A) Two-dimensional (2D) TLC separation of nucleoside monophosphates from genomic DNA in Purkinje and granule cells. (B) Reference map of the TLC spots (A, dAMP; C, dCMP; G, dGMP; T, dTMP; mC, 5-methylcytosine) (14), with the added “x” position. (C) Percentage shows the abundance of a nucleotide neighboring G. Error bars represent the SEM (n = 11); P values were derived from Matt-Whitney statistics. Abundance of dTMP or dAMP did not differ between the samples (P = 0.743 and P = 0.793, respectively).

Fig. 2

Two-dimensional TLC, HPLC, and MS identification of hmC. (A) Two-dimensional TLC analysis of synthetic DNA templates indicates that hmC comigrates with the “x” spot (Fig. 1). (B) HPLC chromatograms (A, 254 nm) of the nucleosides derived from synthetic and cerebellum DNA. The peaks were identified by MS. The arrow points to the peak, which elutes at the same time as hmdC. (C) MS of the fraction corresponding to the HPLC peak indicated above. Closed arrows indicate the masses of 5-hydroxymethylcytosine and 5-hydroxymethyl-2′-deoxycytidine sodium ions (structures are shown in the insets). Open arrows indicate the ions generated by 2′-deoxycytidine, which elutes in a large nearby peak and spills over into the analyzed fraction.

It is unlikely that the hmC that we observed in vivo is a product of DNA damage (5, 6). We did not observe any other DNA damage products such as 8-oxoguanine, a preferential target for oxidants (7), or thymidine glycol, which is produced in vitro by the oxidation of mC (6). In addition, hmC is more abundant in brain, but not in other metabolically active nonproliferating tissues (fig. S5B). Finally, contrary to what one would expect for oxidative DNA damage, we found no correlation between the age of adult mice and the amount of hmC in Purkinje and granule cells (fig. S6).

An early publication suggesting the presence of hmC in mammalian genomes (8) has not been reproduced by others (9, 10). It has been suggested that hmC, if treated with bisulfite, will produce cytosine 5-methylsulfonate, which would be deaminated even at a slower rate than mC (11), leading to the interpretation of hmC as mC after bisulfite sequencing. Although active DNA demethylation is considered to occur, no enzyme has been found that can remove the methyl group from 5-methylcytosine (12). The presence of a hydroxylated methyl group could indicate either an intermediate for oxidative demethylation or a stable end-product, which eliminates the need for removal of the methyl group, by modulating the affinity of proteins that bind to the mC signal in nondividing neuronal cell types. The finding that the methyl-CpG binding domain of MeCP2 protein has a lower affinity toward sequences containing hmC supports this notion (13). It is notable that hmC is nearly 40% as abundant as mC in Purkinje cell DNA. Given the critical role of mC in epigenetic regulation of the genome, we believe that hmC has an important biological role in vivo.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

  1. We thank B. Gauthier for technical assistance; S. Mazel, C. Bare, and X. Fan for flow cytometry advice and nuclei sorts; and H. Deng and J. Fernandez for acquirement of MS data and help with HPLC. We are grateful to members of the Heintz laboratory for discussions and support. This work was supported by the Howard Hughes Medical Institute and the Simons Foundation Autism Research Initiative.
View Abstract

Navigate This Article