Methylation on CpG repeats modulates hydroxymethylcytosine induced duplex destabilization

Abstract

Thermodynamic stability of hydroxymethylcytosine (hmC) containing CpG repeats showed that though hmC destabilizes hypomethylated CpG, this effect can be reversed in a heavily methylated duplex. This dual-epigenetic modulation on duplex stability was also observed in cellular-mimic crowding conditions for both A-form DNA–RNA hybrids and B-form DNA–DNA duplexes.

---

Cytosine (C) in DNA strands can be methylated to form 5-methylcytosine (mC) by the DNA methyltransferase family (DNMT) in the presence of the methyl group donor, S-adenosyl-L-methionine (SAM).^1–4 mC has a profound impact on heritable gene silencing and plays an essential role in cell development in almost all of the mammalian species.^5–7 As the fifth base in the genome, mC has been considered a critical epigenetic marker, which is capable of modulating core biological process. Although the genomic methylation pattern is stably reserved over cell division, genome wide demethylation is observed in specific cell pluripotent states and carcinoma cells,^8–12 while the mechanism of active cytosine demethylation remains largely unknown.

5-Hydroxymethylcytosine (hmC) was not discovered in mammalian cell until 2009,¹³ though it was firstly identified in bacteriophage genome in 1953.¹⁴ hmC is viewed as an oxidative product of mC converted by ten eleven translocation proteins (TETs).^15,16 Together with TETs, hmC could contribute to genomic hypomethylation maintenance, either by blocking DNA methylation machinery or by initiating rapid oxidation-associated active demethylation of mC.¹⁷ hmC exists in most mammalian tissues as a stable base with abundance range from 0.03% to 0.69% of dG.^18,19 High resolution sequencing of hmC in a genomic DNA reveals a unique distribution of hmC over regulatory genomic segments, which suggests that hmC is deeply associated with transcriptional regulation.^20–24 Therefore, not only key functions in active demethylation process, hmC could also serve as an alternative epigenetic marker besides mC. However, how hmC regulates transcription still remains as an enigma.

Recent studies suggested that apart from epigenetic bases involved regulation of protein–DNA interaction, modulation of thermodynamic stability of duplex DNA may account for the epigenetic regulation.^25,26 High resolution melting (HRM) had been used to reveal that hmC modified DNA duplex exhibited declined thermostability comparing to methylated DNA.²⁷ This destabilization effect of hmC was also reported for CpG repeats.²⁵ However, there is so far no report about whether and how the effect of hmC on duplex thermostability may correlate to helical structures and the methylation level, which would unravel the regulatory functions of hmC in RNA transcription and intron–exon recognition during mRNA processing.^28,29 In current study, we considered the possibility that hmC in either hypo- or hypermethylated CpG repeats would alter the effects on duplex stability upon varying the helical structure and solvent environments. By introducing mC and hmC to CpG repeats in B-form DNA and A-form DNA–RNA hybrid duplexes, we report here that hmC effect on decreasing duplex thermostability is highly dependent on the methylation level of CpG regions. This phenomenon can be observed both in aqueous buffer and under cellular-mimic crowding conditions.

To investigate how hmC affects thermostability of A-form and B-form duplexes, two series of DNA sequences, CXC and MXM, containing three CpG or mCpG repeats flanked by the same scramble sequences, were used (Table 1). The central cytosine in (CpG)₃ were substituted by mC or hmC in CMC/MMM or CHC/MHM sequences, respectively. The number of mC in sequence CCC, CMC, MCM, MMM, increases from 0 to 3. The ratio of hmC to mC varies from 1 : 0 (CHC) to 1 : 2 (MHM). CHC and MHM strands were prepared as earlier report³⁰ and all of the oligonucleotides used here were synthesized by solid phase phosphoramidite chemistry and characterized by ESI-MS (Table S1†). DNA duplexes and DNA–RNA hybrids were formed by annealing complementary DNA (G) and RNA (g) strands to CXC and MXM series. Helical structures of DNA duplexes, (denoted as CXC-G and MXM-G series), and DNA–RNA hybrids, (denoted as CXC-g and MXM-g series) were confirmed by CD spectra (Fig. S1†). All of DNA duplexes and DNA–RNA hybrids showed canonical B-form and A-form helical structures, respectively, regardless of mC and/or hmC substitutions. Epigenetic modifications on C5 of cytosine showed no significant alteration on helical structures of CpG containing duplexes.

Table 1 DNA and RNA sequences used in this study

DNA	Sequence (5′-3′)
a X = C/M/H, M and H represent mC and hmC respectively.
CXC^a	TTC CAC GXG CGT TCC TGA CTG ACT C
MXM^a	TTC CAmC GXG mCGT TCC TGA CTG ACT C
G	GAG TCA GTC AGG AAC GCG CGT GGA A
A	GAG TCA GTC AGG AAC GCA CGT GGA A
C	GAG TCA GTC AGG AAC GCC CGT GGA A
T	GAG TCA GTC AGG AAC GCT CGT GGA A
g	gag tca gtc agg aac gcg cgt gga a

---

Melting temperatures (T_m) of DNA duplexes were measured to reflect their thermodynamic stabilities. The high resolution of thermal denaturation allowed T_m to be measured with high repeatability (ESI.4†). In aqueous buffer, we observed a small increase (0.35 °C) in melting temperature of duplex CMC-G, comparing with CCC-G duplex (Fig. 1A). Whereas, T_m of CHC-G is significantly lower than those of CMC-G (−1.4 °C) and CCC-G (−1.05 °C). Similar experiments were carried out for MXM-G duplexes. T_m of MXM-G series of DNA duplexes were higher than those of CXC series. However, T_m of MHM-G was no longer lower, but was comparable to that of MCM-G and MMM-G, though similar as CXC-G series, MMM-G showed highest melting temperature. Destabilization effect of hmC was almost offset by flanking methylation sites in CpG repeats.

Fig. 1 Thermostability of DNA–DNA and DNA–RNA hybrids in sodium phosphate buffer (20 mM pH = 7.0). (A) T_m of CXC and MXM with both G (to form B-form DNA duplex) and g (to form A-form DNA–RNA duplex); (B) ΔT_m = T_m(CXC) − T_m(CCC) or ΔT_m = T_m(MXM) − T_m(MCM); X = mC or hmC. All experiments are performed in triplicates.

Furthermore, the effect of mC and hmC on the thermostability of DNA–RNA hybrids was examined, since DNA–mRNA stability is pivotal to mRNA transcription and intron–exon recognition.²⁹ DNA–RNA hybrids were prepared by annealing well matched complementary RNA strand (g) to CXC and MXM series of DNA strands, and were applied to thermo-denaturation. As shown in Fig. 1A, T_m of all DNA–RNA duplexes were higher than those of DNA duplexes with the same epigenetic patterns. Upon methylation, CMC-g and MXM-g hybrids showed significantly higher T_m than CCC-g. Single, double and triple mC in CMC-g, MCM-g and MMM-g enhanced T_m to 0.77 °C, 1.02 °C and 1.97 °C, respectively. Stabilization from methylation was much better pronounced in DNA–RNA hybrids than in DNA duplexes. Whereas, hmC effects on hybrid duplexes showed the same trend as on DNA duplexes. hmC in hypomethylated hybrid, CHC-g, decreased T_m (−0.63 °C), while single hmC in MHM-g enhanced the thermostability of hybrid duplex (0.65 °C). Unlike mC, the destabilization effect of hmC is not always in accordance with the presence of hmC, but show strong dependence on the level of mC modification on flanking genomic regions.

In order to elucidate how cytosine modifications influence the duplex stability, we calculated hydrogen bonding interactions in Watson–Crick base pairs including C-G, mC-G and hmC-G. The initial structures of all base pairs extracted from crystal structure,³¹ were subjected to optimized at MP2/6-31G(d) level (Fig. S3†). The energy of base pairs and nucleobases were calculated by MP2/aug-cc-pvdz and M06-2X/6-31+G(d,p) methods.^32–35 The interaction energy (E_int), which indicates the H-bonding energy of base pair, was obtained as the difference between the minimized energy of the base pair (E_tot) and the energy sum of the two individual bases (E_G + E_C). The results from both calculation methods showed that mC-G has highest E_int value followed by C-G, while hmC-G has the smallest E_int value, which indicated that hmC-G base pair maintained the weakest H-bonding interaction. Therefore hmC-G is the least stable base pair among the three epigenetic and normal C-G base pairs (Fig. S4, Tables S5 and S6†). The results were consistent with UV-melting observations of CHC series of duplexes. Comparing to C-G base pair, mC stabilizes and hmC destabilizes C-G base pair.

However, in the mCpG repeats, the destabilization effect of hmC is mitigated. This could due to the fact that consecutive methylation may increase the rigidity of helical structures, because of the larger hydrophobic area around mC.^36,37 The small difference of E_int between C-G and hmC-G base pairs can be easily compensated in hypermethylated CpG repeats. When this rigidity is greater than the destabilization effect of hmC, the overall duplex stability would reflect as stability enhancement. Furthermore, the adjacent mC provides enhanced hydrophobicity against the polar environment and would limit the accessibility of hmC to the aqueous solvent. Destabilization effects of hmC on duplexes, due to H-bonding network between hmC and environment water, and/or between hmC and neighboring nucleobase/phosphate backbone via water molecules, would be significantly attenuated.²⁶ At last, methyl group on mC increases nucleobase polarizability, and would increase dispersion energy and stacking interaction among base pairs, which provides extra structural stability to hypermethylated CpG region.³⁵

Thermodynamic stability of nucleic acid in aqueous buffer is not sufficient to predict the situation in living cells. The cellular environments are crowded by the presence of a large number of biomacromolecules. The thermostability of DNA duplexes and DNA–RNA hybrids were studied in Ficoll solution to mimic the macromolecular crowding environments. In normal eukaryotic cells, 100–400 mg mL⁻¹ biomolecules exists in the nuclei, 100–200 mg mL⁻¹ was in nuclear organelle.³⁸ Therefore, 160 mg mL⁻¹ of Ficoll 400 (as final concentration) was applied to DNA–DNA and DNA–RNA duplexes. CD spectra of both CXC-G and CXC-g duplexes showed that no significant global structural changes were observed upon introducing mC or hmC into A-from or B-form duplexes in Ficoll solution (Fig. S2†). As shown in Fig. 2, T_m of all the A-form and B-form DNA duplexes were increased around 3 °C in the Ficoll solution, regardless of epigenetic modifications. Under crowding condition, the destabilization effect of hmC was also apparent in hypo-methylated CpG repeats, while similar negative ΔT_m values were observed for CXC-G and CXC-g in both aqueous buffer and Ficoll solution. For duplexes with high methylation status, both MHM-G and MHM-g showed higher T_m than those of MCM-G and MCM-g duplexes, respectively. Herein, the trends of epigenetic effects on the thermostability of double helixes in cellular crowding condition were consistent with that in aqueous solution. The dependence on methylation density of CpG repeats was maintained for the destabilizing effects from hmC.

Fig. 2 Thermostability of DNA–DNA and DNA–RNA hybrids in crowding condition (sodium phosphate buffer mixed with 160 mg mL⁻¹ Ficoll). (A) T_m of CXC and MXM with both G (to form B-form DNA duplex) and g (to form A-form DNA–RNA duplex) under macromolecular crowding condition; (B) ΔT_m = T_m(CXC) − T_m(CCC) or ΔT_m = T_m(MXM) − T_m(MCM); X = mC or hmC. T_m were measured in sodium phosphate (20 mM pH = 7.0) with 160 mg mL⁻¹ Ficoll. All experiments are performed in triplicates.

To investigate if epigenetic cytosines would compromise the specificity in base pairing to guanine, which is essential to the fidelity of genetic information transferring through central dogma, we studied the thermostability of CXC or MXM series of duplexes with mismatched X–Y pair (Y = A, T, C as shown in Table 1). ΔT_m(X,Y) as in eqn (1) was used to show the reduced stability of duplexes with mismatched mC or hmC.

ΔT_m(X,Y) = T_m(X,Y) − T_m(X,G) (1)

T_m(X,G) was melting temperature of well-matched duplex, CXC-G or MXM-G, and T_m(X,Y) was that of mismatched duplexes, CXC-Y or MXM-Y (Y = A, C, T). As shown in Fig. 3, all of the mismatched duplexes showed negative ΔT_m, larger than 5 °C, while ΔT_m can reach 8 °C for pyrimidine mismatches. Within CXC or MXM series, T_m(X,Y) showed no obvious difference among various epigenetic cytosines while it base paired with the same Y. Furthermore, no significant deviations in T_m(X,Y) were observed between two series of duplexes, CXC and MXM. Similar decrements in duplex thermostability were observed, regardless of either epigenetic cytosines or epigenetic levels, which indicated that base pairing specificity to G over A, C, and T is unlikely varied among C, mC and hmC. Therefore the presence of hmC or mC in template DNA remains synthetic fidelity during replication and transcription.

Fig. 3 ΔT_m between well matched CXC or MXM duplex and mismatched duplexes ΔT_m = T_m(CXC − Y) − T_m(CXC − G), or ΔT_m = T_m(MXM − Y) − T_m(MXM − G). X = C, H, or M; Y = A, C, or T.

The mechanism of how hmC regulates gene expression is still under debate, since recent studies have found hmC is enriched in both gene promoters with suppressed transcription and intragenic gene regions with high expression level.^{20–24,39,40} hmC displayed this unique dual regulation behaviour in embryonic stem cells, though it is inconsistent with previous literatures that hmC has shown only destabilization effects to duplex DNA.^25–27 By examining the thermostability of both hypo- and hypermethylated CpG repeats, a dependence on methylation pattern of flanking regions was observed for the effects from single hmC modification on CpG sequences, which can offer a rationale to address this dual regulation of hmC. The promoters of transcriptional inactive genes, which were often heavily methylated at cytosines, would innately have high duplex stability. Since MHM-G showed same or higher thermostability than MCM-G in our experiments, initiation of mC oxidation to form hmC would not significantly compromise the duplex stability and hence transcription inhibition would not be removed. Instead, hmC in the hypermethylated promoter marks the gene as a “poised gene” for transcription.^17,23 Later during cell development, the promoter of “poised gene” may be activated though hmC-mediated demethylation to enter hypomethylated stage. On the contrary, oxidation of mC to hmC in hypomethylated CpG repeats as in CHC-G would induce immediate mitigation of duplex stability. For gene promoter or gene body with low mC contents, hydroxymethylation on cytosine would lead to activation of gene expression.²⁵ Though hmC forms less stable base pair with G and may induce duplex destabilization, the degree of destabilization can be toned by existing epigenetic modifications and a more refined modulation on gene transcription via duplex stability can be achieved by varying hmC and mC patterns simultaneously.

It was reported that similar as mC, hmC is enriched in gene bodies especially in exons with high CpG content and methylation state.^23,39,41 It is observed here that thermodynamic stability of methylated DNA/RNA duplexes was manipulated upon the combination pattern of mC and hmC, which indicated that both epigenetic cytosines may work simultaneously and cooperatively in intron–exon recognition to regulate mRNA splicing, and consequently control gene expression.^28,29

Conclusions

We have shown that the destabilization effect of hmC on CpG repeats is highly dependent on the overall methylation level of duplex DNA or DNA–RNA hybrids. In hypomethylated (CpG)_n segment, hmC can significantly reduce duplex thermostability. However, in a hypermethylated CpG repeats, conversion of single mC to hmC would show nearly no attenuation on duplex melting temperature. The phenomena can be observed in either aqueous buffer or crowding condition when either A-form DNA–RNA hybrids or B-form DNA–DNA duplex were investigated. Our data suggested that though single hmC may induce apparent destabilization to both DNA duplex and DNA–RNA hybrids, the effect can be modulated by the epigenetic patterns in the flanking regions. The unique phenomena provides better understanding of epigenetic regulation via manipulating helical thermostability and may supply knowledge to alternative mechanisms of the interactions between genetic regions and binding proteins involved in essential biological processes, including central dogma and de/methylation of nucleic acids.

Notes and references

1. M. Okano, S. P. Xie and E. Li, Nat. Genet., 1998, 19, 219–220.
2. M. Okano, D. W. Bell, D. A. Haber and E. Li, Cell, 1999, 99, 247–257.
3. M. G. Goll and T. H. Bestor, Annu. Rev. Biochem., 2005, 74, 481–514.
4. H. Denis, M. N. Ndlovu and F. Fuks, EMBO Rep., 2011, 12, 647–656.
5. R. Jaenisch and A. Bird, Nat. Genet., 2003, 33, 245–254.
6. J. A. Law and S. E. Jacobsen, Nat. Rev. Genet., 2010, 11, 204–220.
7. Z. D. Smith and A. Meissner, Nat. Rev. Genet., 2013, 14, 204–220.
8. W. Mayer, A. Niveleau, J. Walter, R. Fundele and T. Haaf, Nature, 2000, 403, 501–502.
9. J. Oswald, S. Engemann, N. Lane, W. Mayer, A. Olek, R. Fundele, W. Dean, W. Reik and J. Walter, Curr. Biol., 2000, 10, 475–478.
10. P. Hajkova, S. Erhardt, N. Lane, T. Haaf, O. El-Maarri, W. Reik, J. Walter and M. A. Surani, Mech. Dev., 2002, 117, 15–23.
11. H. Sasaki and Y. Matsui, Nat. Rev. Genet., 2008, 9, 129–140.
12. S. Feng, S. E. Jacobsen and W. Reik, Science, 2010, 330, 622–627.
13. S. Kriaucionis and N. Heintz, Science, 2009, 324, 929–930.
14. G. R. Wyatt and S. S. Cohen, Biochem. J., 1953, 55, 774–782.
15. M. Tahiliani, K. P. Koh, Y. Shen, W. A. Pastor, H. Bandukwala, Y. Brudno, S. Agarwal, L. M. Iyer, D. R. Liu, L. Aravind and A. Rao, Science, 2009, 324, 930–935.
16. S. Ito, A. C. D'Alessio, O. V. Taranova, K. Hong, L. C. Sowers and Y. Zhang, Nature, 2010, 466, 1129–1133.
17. H. Wu and Y. Zhang, Cell, 2014, 156, 45–68.
18. D. Globisch, M. Münzel, M. Müller, S. Michalakis, M. Wagner, S. Koch, T. Brückl, M. Biel and T. Carell, PLoS One, 2010, 5, e15367.
19. M. Münzel, D. Globisch, T. Brückl, M. Wagner, V. Welzmiller, S. Michalakis, M. Müller, M. Biel and T. Carell, Angew. Chem., Int. Ed., 2010, 49, 5375–5377.
20. H. Wu, A. C. D'Alessio, S. Ito, Z. Wang, K. Cui, K. Zhao, Y. E. Sun and Y. Zhang, Genes Dev., 2011, 25, 679–684.
21. M. Yu, G. C. Hon, K. E. Szulwach, C.-X. Song, L. Zhang, A. Kim, X. Li, Q. Dai, Y. Shen, B. Park, J.-H. Min, P. Jin, B. Ren and C. He, Cell, 2012, 149, 1368–1380.
22. C.-X. Song, K. E. Szulwach, Y. Fu, Q. Dai, C. Yi, X. Li, Y. Li, C.-H. Chen, W. Zhang, X. Jian, J. Wang, L. Zhang, T. J. Looney, B. Zhang, L. A. Godley, L. M. Hicks, B. T. Lahn, P. Jin and C. He, Nat. Biotechnol., 2011, 29, 68–72.
23. W. A. Pastor, U. J. Pape, Y. Huang, H. R. Henderson, R. Lister, M. Ko, E. M. McLoughlin, Y. Brudno, S. Mahapatra, P. Kapranov, M. Tahiliani, G. Q. Daley, X. S. Liu, J. R. Ecker, P. M. Milos, S. Agarwal and A. Rao, Nature, 2011, 473, 394–397.
24. K. Williams, J. Christensen, M. T. Pedersen, J. V. Johansen, P. A. C. Cloos, J. Rappsilber and K. Helin, Nature, 2011, 473, 343–348.
25. A. Thalhammer, A. S. Hansen, A. H. El-Sagheer, T. Brown and C. J. Schofield, Chem. Commun., 2011, 47, 5325–5327.
26. L. Lercher, M. A. McDonough, A. H. El-Sagheer, A. Thalhammer, S. Kriaucionis, T. Brown and C. J. Schofield, Chem. Commun., 2014, 50, 1794–1796.
27. C. M. Rodríguez López, A. J. Lloyd, K. Leonard and M. J. Wilkinson, Anal. Chem., 2012, 84, 7336–7342.
28. R. I. Kraeva, D. B. Krastev, A. Roguev, A. Ivanova, M. N. Nedelcheva-Veleva and S. S. Stoynov, PLoS One, 2007, 2, e290.
29. M. N. Nedelcheva-Veleva, M. Sarov, I. Yanakiev, E. Mihailovska, M. P. Ivanov, G. C. Panova and S. S. Stoynov, Nat. Commun., 2013, 4, 2101.
30. S. Xuan, Q. Wu, L. Cui, D. Zhang and F. Shao, Bioorg. Med. Chem. Lett., 2015, 25, 1186–1191.
31. D. Renciuk, O. Blacque, M. Vorlickova and B. Spingler, Nucleic Acids Res., 2013, 41, 9891–9900.
32. P. Hobza and J. Šponer, J. Am. Chem. Soc., 2002, 124, 11802–11808.
33. Q. Song, Z. Qiu, H. Wang, Y. Xia, J. Shen and Y. Zhang, Struct. Chem., 2013, 24, 55–65.
34. Y. Zhao and D. G. Truhlar, Acc. Chem. Res., 2008, 41, 157–167.
35. C. Acosta-Silva, V. Branchadell, J. Bertran and A. Oliva, J. Phys. Chem. B, 2010, 114, 10217–10227.
36. M. Wanunu, D. Cohen-Karni, R. R. Johnson, L. Fields, J. Benner, N. Peterman, Y. Zheng, M. L. Klein and M. Drndic, J. Am. Chem. Soc., 2010, 133, 486–492.
37. B. Xu, G. Devi and F. Shao, Org. Biomol. Chem., 2015, 13, 5646–5651.
38. S.-i. Nakano, D. Miyoshi and N. Sugimoto, Chem. Rev., 2013, 114, 2733–2758.
39. Y. Xu, F. Wu, L. Tan, L. Kong, L. Xiong, J. Deng, A. J. Barbera, L. Zheng, H. Zhang, S. Huang, J. Min, T. Nicholson, T. Chen, G. Xu, Y. Shi, K. Zhang and Y. G. Shi, Mol. Cell, 2011, 42, 451–464.
40. L. Tan, L. Xiong, W. Xu, F. Wu, N. Huang, Y. Xu, L. Kong, L. Zheng, L. Schwartz, Y. Shi and Y. G. Shi, Nucleic Acids Res., 2013, 41, e84.
41. S. Feng, S. J. Cokus, X. Zhang, P.-Y. Chen, M. Bostick, M. G. Goll, J. Hetzel, J. Jain, S. H. Strauss, M. E. Halpern, C. Ukomadu, K. C. Sadler, S. Pradhan, M. Pellegrini and S. E. Jacobsen, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 8689–8694.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08647k

---