Site-Specific 5-Formyl Cytosine Mediated DNA-Histone Cross-Links: Synthesis and Polymerase Bypass by Human DNA Polymerase η

Abstract

DNA-Protein Cross-links (DPCs) between DNA epigenetic mark 5-formylC and lysine residues of histone proteins spontaneously form in human cells. Such conjugates are likely to influence chromatin structure and mediate DNA replication, transcription, and repair, but are challenging to study due to their reversible nature. Here we report the construction of site specific, hydrolytically stable DPCs between 5fdC in DNA and K4 of histone H3 and an investigation of their effects on DNA replication. Our approach employs oxime ligation, allowing for site-specific conjugation of histones to DNA under physiological conditions. Primer extension experiments revealed that histone H3-DNA crosslinks blocked DNA synthesis by hPol η polymerase, but were bypassed following proteolytic processing.

Graphical Abstract

DNA-Histone cross-links (DPCs) have been synthesized efficiently via oxime ligation. These DPC lesions block DNA replication but the replication block can be removed via proteolytic processing.

Irreversible entrapment of cellular proteins on genomic DNA following exposure to bis-electrophiles, transition metals, and free radical species gives rise to DNA-protein cross-links (DPCs).^[1] DPCs are unusually bulky DNA lesions that have the ability to block DNA transactions including transcription, replication, and repair, potentially leading to genomic instability and cell death.^[2] DPCs accumulate in the heart and brain tissues with age and are hypothesized to play an important role in aging, cancer and neurodegenerative diseases.^[3] Individuals deficient in SPRTN, a critical gene required for DPC repair, develop the Ruijs–Aalfs syndrome, a human autosomal recessive disorder characterized by accelerated aging, genomic instability, and early-onset of hepatocellular carcinoma.^[4]

In addition to their role in disease, DPCs may also be involved in epigenetic regulation. We and others have shown that 5-formylcytosine (5fC), an endogenous oxidation product of the major DNA epigenetic mark 5-methylcytosine,^[5] forms reversible Schiff’s base conjugates with lysine and arginine residues of histone proteins in cells (Scheme 1).^[6] Histones H2A, H2B and H4 are reported to participate in formation of DPCs with 5fC in nucleosomal core particles.^[6a] Such conjugates influence chromatin structure and stability and are thought to mediate gene expression levels in cells.^[6b] In addition to their effects on gene expression,^[6b] 5-formylC-histone crosslinks are likely to interfere with DNA replication and repair.^[6]

Scheme 1.

Reversible histone-DNA cross-linking and their stabilization via reduction of the resulting Schiff bases.

Specialized translesion synthesis (TLS) DNA polymerases such as hPol η and κ allow for DNA replication to continue in the presence of DNA damage. Unlike replicative DNA polymerases, TLS polymerases have open and flexible active sites which can accommodate many bulky DNA lesions.^[7] TLS polymerases are recruited to stalled replication forks and can traverse the damaged sites, preventing cell death.^[7] Specifically, we found that hPol η bypasses DNA-peptide cross-links in an error-free or error-prone manner depending on lesion structure and the local sequence context.^[8]

In the present work, we aimed to elucidate the effects of site specific histone-5fC conjugates on DNA replication in the presence of human DNA polymerase η. In our previous studies, DNA-histone cross-links were created via reductive amination reactions of fC-containing DNA with histones H4 and H2A, generating a complex mixture of products as a result of cross-linking at multiple Lys and Arg residues (Scheme 1).^{[6a, 9]} A major limitation of this approach is poorly defined conjugate structures due to crosslinking at multiple sites and potential protein denaturation upon treatment with NaCNBH₄.^[10]

We recently developed a novel method for creating site specific, hydrolytically stable 5fC-protein crosslinks via oxime ligation reactions between 5fC in DNA and oxy-Lys residues within proteins (Scheme 2).^[11] Oxy-Lys is a structural analogue of lysine, where the Lys ε-CH₂ group is replaced with an oxygen atom (Scheme 2). This new approach bypasses the use of harsh reducing agents and produces structurally defined conjugates via a bioorthogonal reaction between 5fC in DNA and site-specific oxy-Lys in proteins and peptides.^[11]

Scheme 2.

Oxime ligation to generate histone H3-DNA cross-links.

Herein, we employed the oxime ligation approach^[11] to prepare hydrolytically stable, site-specific DPCs between 5fC in DNA and oxy-Lys placed on K4 of human histone H3. Further, the ability of human polymerase η to bypass structurally defined histone H3-DNA cross-links in vitro was investigated.

Initial conjugation reactions were conducted using a short oxy-Lys containing peptide derived from the N-terminus of histone H3. Synthetic 8-mer peptide containing oxy-Lys at K4 [Pep 1: NH₂-ARTKQTAR (K = Oxy-Lys), Table 1] was prepared by solid phase synthesis (for details, see Supporting Information). The protected oxy-Lys amino acid was synthesized in our laboratory starting with glutamic acid.^[11] Oxy-Lys containing peptide was purified by HPLC and characterized by electrospray ionization mass spectrometry (Supporting Information, Figure S1). DNA strands containing 5fC [DNA-1, 5’-d(GCC AGT GCC AAG CTT GCA TGC CT)-3’ and DNA-2, 5’-d(ATG GCG TGC TAT, where C= 5fdC] were prepared by solid phase synthesis on a DNA synthesizer using 2’-deoxynucleoside phosphoramidites and characterised by ESI⁺ MS (Table 1, Supporting Information, Figures S2 and S3).

Table 1.

Mass spectrometry characterization of synthetic oligonucleotides and polypeptides employed in this work.^[a]

	Sequences	M Calc.	M Obs.
Pep 1	NH2-ARTKQTAR-COOH	933.1	933.05
Pep 2	NH2-ARTKQTARKSTGGKAPRK QLATKAARKSA PATG-COOH	3454.6	3453.8
DNA-1	5’-d(GCC AGT GCC AAG CTT GCA TGC CT)-3’	7030.1	7030.5
DNA-2	5’-d(ATG GCG TGC TAT)-3’	3704.5	3704.7

^[a]Calculated mass and experimentally observed deconvoluted mass. Peptides were analyzed in the ESI⁺ mode and DNA in the ESI⁻ mode. K = oxy-Lys, T = Threonine glycolic acid, C = 5fC.

Bioconjugation reactions between 5fC containing 23-mer DNA (DNA-1, 5’-d(GCC AGT GCC AAG CTT GCA TGC CT)-3’) and oxy-Lys containing peptide Pep 1 (DNA-1:Pep 1; 1:100) were conducted in 100 mM ammonium acetate buffer, pH 4.5 at 37 °C for 16 h, with 100 mM aniline present as a catalyst (for details see Supporting Information). HPLC purification (Supporting Information, Figure S4) and HPLC-ESI-MS analyses confirmed the identity of the conjugation product (M = 7942.3 Da, corresponding to the sum of DNA-1 (M = 7030.53 Da) and Pep 1 (M = 933.05 Da) (Supporting Information, Figure S5).

To generate human histone H3 containing site specific oxy-Lys modification at K4, sortase mediated ligation^[12] was employed (Scheme 3). First, the 33-mer peptide corresponding to the N-terminal domain of histone H3 [Pep 2: NH₂-(ARTKQTARK STGGKAPRKQ LATKAARKSA PATG, K=oxy-Lys, T=threonine ester)] was prepared by solid phase synthesis. Threonine glycolic acid and oxy-Lys were synthesized in our laboratory according to the literature (Supporting Information, Scheme S1, S2).^{[11, 13]} As described elsewhere,^[13] the use of threonine glycolic acid instead of threonine improves sortase ligation yields by making the ligation reaction irreversible. The 33-mer peptide Pep 2 was prepared via solid phase synthesis, purified by RP-HPLC, and characterised by mass spectrometry (Supporting Information, Figure S6). Sortase F40^[12] -catalyzed reaction was used to ligate Pep 2 with truncated histone H3 missing 32 N-terminal amino acids (gH3) (for details, see Supporting Information). The truncated histone H3 (102 aa in length) was expressed in E. coli as described previously.^[14] Synthetic peptide Pep 2 was ligated with truncated histone H3 in the presence of F40 sortase in 20 mM PIPES, 1 mM CaCl₂ at pH 7.0, and 37 °C overnight, followed by dialysis to obtain the full length histone H3 containing site-specific oxy-Lys at position K4 [oxy-Lys(K4)-H3, (Scheme 3, Supporting Information, Figure S7].^[14] The integrity of the site-specifically modified histone H3 was confirmed by SDS-PAGE (Supporting Information, Figure S7) and mass spectrometry (Supporting Information, Figure S8).

Scheme 3.

Construction of K4-Oxy-Lys-H3 via sortase mediated ligation.

To create site specific histone H3-DNA cross-links, purified K4-Oxy-Lys-H3 was incubated with 5fdC containing DNA [DNA-1, 5’-d(GCC AGT GCC AAG CTT GCA TGC CT)-3’ at 37 °C for 18 h in 100 mM ammonium acetate buffer (pH 4.5) in the presence of 100 mM aniline as a catalyst (Scheme 4, Supporting Information). Reaction progress was monitored by denaturing PAGE, and the desired product was observed as a reduced mobility band on the gel (Figure 2).

Scheme 4.

Oxime ligation reaction between oxy-Lys(K4)-H3 protein and 5fdC containing DNA. Conditions:100 mM NH₄OAc, 100 mM aniline, pH 4.5. 37 °C, 16 h.

Figure 2.

SDS-PAGE analysis of reductive animation (Lane 4) and oxime ligation reactions (Lane 5). DNA 23-mer (5’-d(GCC AGT GCC AAG CTT GCA TGC CT)-3’, C = 5-formylC( DNA-1)) was reacted with histone H3 or oxy-Lys-histone H3. A. Molecular weights of the reactants and the product. B. Representative SDS-PAGE gel image. Lane 1. Protein ladder. Lane 2. Histone H3. Lane 3. oxy-Lys(K4)-H3. Lane 4. Native histone H3 conjugation to DNA-1 (23-mer) via reductive amination. (a) DNA-1 (500 pmol) was incubated with 2-fold molar excess of native histone H3 (1 nmol) in 4.5 mM sodium phosphate buffer, pH 7.2, 37° C for 3 h; (b) The resulting Schiff base was reduced with 25 mM NaCNBH₃, 37° C for overnight. Lane 5. Engineered oxy-Lys(K4)-H3 conjugation to DNA-1 via oxime ligation. DNA-1 (500 pmol) was incubated with 2-fold molar excess of oxy-Lys(K4)-H3 (1 nmol) in 20 μL of ammonium acetate buffer (100 mM, pH 4.5) containing 100 mM aniline at 37 °C overnight. The reaction mixture was then heated to 90 °C for 10 min prior to loading on denaturing 4–12% bis-tris PAGE. The gel was stained with Simply Blue.

As a positive control, native histone H3 was also conjugated to DNA-1 via reductive amination to generate a DPC (Scheme 1).^[9b] We found that histone-DNA conjugates obtained via oxime ligation (Lane 5) and reductive amination (Lane 4) moved identically on denaturing PAGE, further confirming the identity of the conjugate. However, oxime ligation was more efficient as compared to reductive amination (90% vs 30% yield, Figure 2).

In order to optimize oxime ligation yields, reaction conditions such as pH, temperature, and DNA:protein ratios were systematically varied (Figure 3, Supporting Information). Radiolabeled DNA 12-mer (DNA-2) was incubated with increasing amount of protein in ammonium acetate buffer and at different pH. As shown in Figure 3, up to 95% reaction conjugation yields were obtained when using 100 mM ammonium acetate buffer, 100 mM aniline, 37 °C, pH 4.5, for 18 h, and protein:DNA ratio of 2:1.

Figure 3.

Optimization of oxime ligation reactions. ³²P-endlabeled, 5fdC containing DNA 12-mers (DNA-2, 250 pmol) were incubated with increasing amounts of oxy-Lys(K4)-H3 in 20 μL of ammonium acetate buffer (100 mM, pH 4.5) containing 100 mM aniline at 37 °C overnight. The reaction mixture was then heated to 90 °C for 10 min prior to loading on denaturing 20% (w/v) PAGE containing 7 M urea. DNA and DPCs were visualized on the gel via autoradiography. A. Reaction yields at different protein:DNA ratios; B. Reaction yields at different pH.

The identity of the covalent DNA–histone H3 crosslinks was confirmed by mass spectrometry. The conjugates were subjected to in gel digestion with nucleases and alkaline phosphatase to cleave the DNA to single nucleosides and trypsin to cleave the protein to peptides (for details, see Supporting Information). The resulting tryptic peptides were analyzed by nanoLC-ESI⁺-MS/MS on an Orbitrap Elite mass spectrometer. Only oxy-Lys residue at position 4 (K4) contained a cross-link to 5-formyl-dC (Figure 4), while the remaining Lys and Arg residues were intact (Supporting Information, Table S1). These results confirm that oxime ligation was site-specific for oxy-Lys residue within histone H3 (K4). Other Lys residues of histone H3 may form transient Schiff base conjugates with 5fC in DNA, which are not expected to survive gel purification.

Figure 4.

NanoLC- NSI⁺-MS/MS spectrum of oxy-Lys-histone H3 tryptic peptide containing 5fc cross-link to oxy-Lys at K4 (K).

To determine to what extent the presence of 5fC-mediated DNA–histone cross-links influences the efficiency and the fidelity of DNA replication, primer-template complexes for in vitro DNA replication DPC containing DNA strand (5’-GCCAGTGXCAAGCTTGC ATGCCT-3’ where X = 5fC crosslink to histone H3, DNA-1 in Table 1) were prepared by oxime ligation and by reductive amination. These DPCs were separated from unreacted protein by SDS-PAGE (Figure 2) and extracted from the gel using ammonium bicarbonate buffer, pH 8.0, containing 0.1% SDS (Supporting Information).^[15] Desalting of the DPC was performed with Amicon 3K filters.^[9a] DNA primer 5’-AGGCATGCAAGTTCG-3’ was radiolabeled by incubating with T4 PNK and ³²P-ATP at 37 °C for 1 h, followed by the removal of excess ³²P-ATP by microspin G-25 columns (for details see Supporting Information) and annealed to DPC containing templates.

In vitro DNA replication experiments were conducted with templates containing unmodified dC, 5fC, and 5fC-histone H3 DPCs generated by reductive amination or oxime ligation (for details, see Supporting Information). We selected hPol η (6:1 DNA:enzyme ratio) based on our previous studies demonstrating the ability of this enzyme to catalyze polymerase bypass across other DNA-polypeptide cross-links.^{[2a, 7a, 16]} The polymerase was expressed in E. coli and purified as described previously.^[17] Primer extension reactions were conducted under physiological conditions for 0.5, 15, 30 or 60 min.

Primer extension products were separated by denaturing PAGE (Figure 5). For templates containing unmodified dC or 5fC at position X, full extension of the primer (23-mer product) was observed within the first 5 min of the polymerizarion reaction, confirming the catalytic activity of the recombinant polymerase η (Figures 5A,B and Supporting Information). In contrast, primer extension was completely blocked in the presence of histone H3 DPCs generated by reductive amination (Figure 5C) or site specific DPCs created via oxime ligation (Figure 5D). Even after 1 h incubation, no primer extension products past DPC were observed (Figure 5). However, polymerase activity was restored when the protein was cleaved with proteinase K prior to polymerase reaction (Figure 5E, F). Our previous studies revealed that such treatment generates DNA-Lys crosslinks; such lesions are much smaller in size and can be accommodated in the polymerase active site.^{[9b, 16]}

Figure 5.

Primer extension across DPCs generated by reductive amination and oxime ligation. Radiolabeled 15-mer primer was annealed to 23-mer DNA template containing site specific DPCs at position X. In vitro replication reactions were conducted in the presence of recombinant hPol η and all four dNTPs. Primer extension products were resolved by denaturing PAGE and visualized by autoradiography. A. Replication of DNA templates containing standard dC; B. Replication of DNA templates containing 5fdC at X; C. Replication of DNA templates containing Histone H3 DPC prepared by reductive amination at X; D. Replication of DNA templates containing DPCs prepared by oxime ligation at X; E. Polymerase bypass of reductive amination created DPCs to histone H3 after treatment with proteinase K; F. Polymerase bypass of oxime ligation created DPCs to K4-oxy-Lys H3 after treatment with proteinase K.

It has been proposed that in living cells, proteins covalently linked to DNA are proteolytically degraded to smaller polypeptide conjugates.^[18] This proteolytic processing can be catalyzed by specialized metalloproteases such as Spartan and/or via ubiquitin/proteasomal degradation pathways.^{[8c, 19]} The resulting DNA-peptide cross-links can be removed via nucleotide excision repair (NER) mechanism^[20] or bypassed by TLS polymerases.^{[2a, 16, 21]}

Our ongoing studies will determine whether 5fC mediated cross-links formed at other positions within the H3 protein (e.g. K14, K27, K36, Supporting Information, Table S2) and at other histones exert a similar effect on DNA polymerases. Furthermore, the role of reversible 5fC-histone cross-links in controlling gene expression and chromatin structure requires further investigation. As the N-terminal lysines of histone proteins are subject to posttranslational modifications such as acetylation and methylation,^[22] reversible cross-linking of histones to DNA may block with these reactions, potentially eliciting additional biological effects. Our methodology to prepare site specific, hydrolytically stable histone-DNA conjugates under mild conditions opens the door to future studies of the role of 5fC-histone conjugates in epigenetic regulation.

In conclusion, we have prepared model hydrolytically stable, site-specific DNA-histone H3 crosslinks via oxime ligation and reductive amination and investigated their effects on DNA replication in the presence of human DNA polymerase η. The advantage of employing oxime ligation over reductive amination methodology reported previously^[11] is the ability to prepare site-specific conjugates via a biorthogonal reaction between 5fC in DNA and oxy-Lys in histone proteins. This approach allows for efficient, site specific bioconjugation of protein to DNA without the need for harsh chemicals such as NaCNBH₃. Molecular dynamics simulations reveal that the presence of an oxygen in place of CH₂ group at the ɛ position of lysine has minimal effect on protein and DPC structure,^[8b] allowing for future use of these model lesions in biological experiments in vivo and in vitro.

Primer extension results reveal that histone K4-H3 DPC lesions completely block DNA replication, but that the replication block can be removed via proteolytic processing (Figure 5). Literature suggests that proteolytic degradation of DPCs is important for their repair and replication bypass in vivo.^{[4, 18a, 19c, 23]} It should be noted that Schiff base DPCs generated in cells are reversible and can dissociate spontaneously or in the presence of accessory proteins: their dynamics in chromatin requires further investigation.

Figure 1.

Sortase mediated ligation to construct site-specifically modified histone H3 ARTK*QTARKSTGGKAPRKQLATKAARKSAPATGGVKKPHR YRPGTVALREIRRYQKSTELLIRKLPFQRLVREIAQDFKTDLRFQSSAVMALQEACEAYLVGLFEDTNLCAIHAKRVTIM PKDIQLARR IRGERA (K* = Oxy-Lys,). gH3 represents globular histone H3 lacking the N-terminal tail (32 residues) and K4-Oxy-Lys-H3 is the full length histone containing oxy-Lys residue at K4.