A 7-Deazaadenosylaziridine Cofactor for Sequence-Specific Labeling of DNA by the DNA Cytosine-C5 Methyltransferase M.HhaI

DNA methyltransferases (MTases) catalyze the transfer of the activated methyl group of the cofactor S-adenosyl-l-methionine (AdoMet or SAM) to the exocyclic amino groups of adenine or cytosine or the C5 ring atom of cytosine within specific DNA sequences. The DNA adenine-N6 MTase from Thermus aquaticus (M.TaqI) is also capable of coupling synthetic N-adenosylaziridine cofactor analogues to its target adenine within the double-stranded 5′-TCGA-3′ sequence. This M.TaqI-mediated coupling reaction was exploited to sequence-specifically deliver fluorophores and biotin to DNA using N-adenosylaziridine derivatives carrying reporter groups at the 8-position of the adenine ring. However, these 8-modified aziridine cofactors were poor substrates for the DNA cytosine-C5 MTase from Haemophilus haemolyticus (M.HhaI). Based on the crystal structure of M.HhaI in complex with a duplex oligodeoxynucleotide and the cofactor product, we synthesized a stable 7-deazaadenosylaziridine derivative with a biotin group attached to the 7-position via a flexible linker. This 7-modified aziridine cofactor can be efficiently used by M.HhaI for the direct, quantitative and sequence-specific delivery of biotin to the second cytosine within 5′-GCGC-3′ sequences in short duplex oligodeoxynucleotides and plasmid DNA. In addition, we demonstrate that biotinylation by M.HhaI depends on the methylation status of the target cytosine and, thus, could provide a method for cytosine-C5 DNA methylation detection in mammalian DNA.


Introduction
Deoxyribonucleic acid (DNA) carries the genetic information of all living organisms, and the information is stored in the sequence of the four nucleobases adenine, cytosine, guanine and thymine. In most organisms, the information content of DNA is increased by enzymatic methylation of the exocyclic amino groups of adenine or cytosine or the C5-ring atom of cytosine within short DNA sequences (epigenetic information). All three types of DNA methylation are found in bacteria and serve a broad range of biological processes, including protection against endogenous restriction endonucleases, direction of DNA mismatches repair after DNA replication and regulation of gene expression. In mammalian cells, DNA methylation is restricted to cytosine-C5 within 5 1 -CG-3 1 (CpG) sequences and involved in transcriptional regulation of gene expression and cell differentiation.  The prokaryotic DNA cyotsine-C5 MTase from Haemophilus haemolyticus (M.HhaI) is of special interest, because it methylates the second cytosine within the sequence 5′-GCGC-3′ (target cytosine in bold) which includes the 5′-CG-3′ recognition sequence of mammalian DNA MTases [14]. Thus, M.HhaI could serve as useful tool for studying mammalian DNA methylation by SMILing DNA. Unfortunately, the 8-modified N-adenosylaziridine derivatives 3 and 4 are poor substrates for M.HhaI. To overcome this limitation, we report here the design and synthesis of a 7-modified aziridine cofactor, which is suitable for quantitative labeling with M.HhaI in one step and for CpG-methylation detection.

Design and Synthesis of an N-Adenosylaziridine Cofactor for DNA Labeling with M.HhaI
Although M.HhaI and M.TaqI can utilize N-adenosylaziridine (2) equally well, M.HhaI in contrast to M.TaqI cannot efficiently couple the 8-modified N-adenosylaziridine derivatives 3 and 4 with DNA (vide infra). This result is in agreement with the crystal structure of M.HhaI in complex with a duplex ODN and S-adenosyl-L-homocysteine (AdoHcy), the demethylated cofactor formed after methyl group transfer from the natural cofactor AdoMet 1 [15]. AdoHcy is buried in the cofactor binding pocket of M.HhaI and the 8-position of the adenine points towards the enzyme (Figure 1). Thus, attaching groups to this position could lead to sterically unfavorable interactions and loss of binding affinity. In contrast, the 7-position of the adenine ring is pointing towards the solvent, and modifications at this position should not interfere with cofactor binding. However, alkylation of the N7 atom will formally place a positive charge onto the adenine ring, which will weaken the glycosidic bond and enhance depurination. This problem can be overcome by replacing nitrogen with carbon and synthesizing the stable 7-deazaadenosylaziridine derivative 5 with biotin attached via a flexible linker to the 7-position. The prokaryotic DNA cyotsine-C5 MTase from Haemophilus haemolyticus (M.HhaI) is of special interest, because it methylates the second cytosine within the sequence 5 1 -GCGC-3 1 (target cytosine in bold) which includes the 5 1 -CG-3 1 recognition sequence of mammalian DNA MTases [14]. Thus, M.HhaI could serve as useful tool for studying mammalian DNA methylation by SMILing DNA. Unfortunately, the 8-modified N-adenosylaziridine derivatives 3 and 4 are poor substrates for M.HhaI. To overcome this limitation, we report here the design and synthesis of a 7-modified aziridine cofactor, which is suitable for quantitative labeling with M.HhaI in one step and for CpG-methylation detection.

Design and Synthesis of an N-Adenosylaziridine Cofactor for DNA Labeling with M.HhaI
Although M.HhaI and M.TaqI can utilize N-adenosylaziridine (2) equally well, M.HhaI in contrast to M.TaqI cannot efficiently couple the 8-modified N-adenosylaziridine derivatives 3 and 4 with DNA (vide infra). This result is in agreement with the crystal structure of M.HhaI in complex with a duplex ODN and S-adenosyl-L-homocysteine (AdoHcy), the demethylated cofactor formed after methyl group transfer from the natural cofactor AdoMet 1 [15]. AdoHcy is buried in the cofactor binding pocket of M.HhaI and the 8-position of the adenine points towards the enzyme (Figure 1). Thus, attaching groups to this position could lead to sterically unfavorable interactions and loss of binding affinity. In contrast, the 7-position of the adenine ring is pointing towards the solvent, and modifications at this position should not interfere with cofactor binding. However, alkylation of the N7 atom will formally place a positive charge onto the adenine ring, which will weaken the glycosidic bond and enhance depurination. This problem can be overcome by replacing nitrogen with carbon and synthesizing the stable 7-deazaadenosylaziridine derivative 5 with biotin attached via a flexible linker to the 7-position.  The synthesis of the 7-modified aziridine cofactor 5 (Scheme 3) started by coupling 4-chloro-5-iodo-7H-pyrrolo[2,3-d]pyrimidine (6) with 5-O-(t-butyldimethylsilyl)-2,3-O-isopropylidene-α-Dribofuranosyl chloride (7). Initial attempts to deprotonate 6 with sodium hydride according to Ugarkar et al. [16] gave only small amounts of product 8. We, therefore, adapted the method by Rosemeyer and Seela [17], which employs potassium hydroxide in combination with the cryptand tris[2-(2methoxyethoxy)ethyl]amine (TDA-1) and obtained nucleoside 8 in an acceptable yield. After removal of the TBS protecting group with tetrabutylammonium fluoride, the chlorine atom in nucleoside 9 was substituted in saturated methanolic ammonia under elevated pressure to yield the 7-deazaadenosine derivative 10. Removal of the TBS protecting group at this stage of the synthesis was advantageous, because the TBS group was partially cleaved off by the ammonia treatment, leading to a product mixture. The side chain at the 7-position was introduced by palladium-catalyzed Sonogashira coupling with N-(2-propynyl)-2,2,2-trifluoroacetamide (11) in analogy to a procedure by Seela et al. [18], and   The synthesis of the 7-modified aziridine cofactor 5 (Scheme 3) started by coupling 4-chloro-5-iodo-7H-pyrrolo[2,3-d]pyrimidine (6) with 5-O-(t-butyldimethylsilyl)-2,3-O-isopropylidene-α-Dribofuranosyl chloride (7). Initial attempts to deprotonate 6 with sodium hydride according to Ugarkar et al. [16] gave only small amounts of product 8. We, therefore, adapted the method by Rosemeyer and Seela [17], which employs potassium hydroxide in combination with the cryptand tris[2-(2methoxyethoxy)ethyl]amine (TDA-1) and obtained nucleoside 8 in an acceptable yield. After removal of the TBS protecting group with tetrabutylammonium fluoride, the chlorine atom in nucleoside 9 was substituted in saturated methanolic ammonia under elevated pressure to yield the 7-deazaadenosine derivative 10. Removal of the TBS protecting group at this stage of the synthesis was advantageous, because the TBS group was partially cleaved off by the ammonia treatment, leading to a product mixture. The side chain at the 7-position was introduced by palladium-catalyzed Sonogashira coupling with N-(2-propynyl)-2,2,2-trifluoroacetamide (11) in analogy to a procedure by Seela et al. [18], and  (7). Initial attempts to deprotonate 6 with sodium hydride according to Ugarkar et al. [16] gave only small amounts of product 8. We, therefore, adapted the method by Rosemeyer and Seela [17], which employs potassium hydroxide in combination with the cryptand tris[2-(2-methoxyethoxy)ethyl]amine (TDA-1) and obtained nucleoside 8 in an acceptable yield. After removal of the TBS protecting group with tetrabutylammonium fluoride, the chlorine atom in nucleoside 9 was substituted in saturated methanolic ammonia under elevated pressure to yield the 7-deazaadenosine derivative 10. Removal of the TBS protecting group at this stage of the synthesis was advantageous, because the TBS group was partially cleaved off by the ammonia treatment, leading to a product mixture. The side chain at the 7-position was introduced by palladium-catalyzed Sonogashira coupling with N-(2-propynyl)-2,2,2-trifluoroacetamide (11) in analogy to a procedure by Seela et al. [18], and the alkyne linker of 12 was reduced to the more flexible alkyl linker by catalytic hydrogenation. The next steps followed our synthesis of the 8-modified aziridine cofactor 4 [7]. The 5 1 -hydroxy group was activated as a mesyl ester and the isopropylidene protecting group removed under acidic conditions. The mesylate was replaced with aziridine in a nucleophilic substitution reaction, the trifluoroacetyl protecting group removed under basic aqueous work up and the resulting primary amine coupled with N-hydroxysuccinimidyl biotin (NHS biotin) to yield the desired 7-deazaadenosylaziridine derivative 5 with biotin attached to the 7-position.

Labeling Short Duplex Oligodeoxynucleotides with M.HhaI and 5
To test whether the new 7-deazaadenosylaziridine derivative 5 functions as a cofactor for M.HhaI (Scheme 4), we initially used the short duplex ODN I¨II (13 base pairs). Duplex I¨II with the palindromic 5 1 -GCGC-3 1 recognition sequence was prepared by annealing Strand I with the complementary Strand II. The target cytosine in the lower Strand II was blocked by methylation, i.e., cytosine is replaced with C5-methylcytosine (C Me ), and only the target cytosine in the upper Strand I can be modified by M.HhaI. Formation of Duplex I¨II was analyzed by anion-exchange HPLC, and the duplex eluted as a single peak after 18.4 min.
Molecules 2015, 20, page-page the alkyne linker of 12 was reduced to the more flexible alkyl linker by catalytic hydrogenation. The next steps followed our synthesis of the 8-modified aziridine cofactor 4 [7]. The 5′-hydroxy group was activated as a mesyl ester and the isopropylidene protecting group removed under acidic conditions. The mesylate was replaced with aziridine in a nucleophilic substitution reaction, the trifluoroacetyl protecting group removed under basic aqueous work up and the resulting primary amine coupled with N-hydroxysuccinimidyl biotin (NHS biotin) to yield the desired 7-deazaadenosylaziridine derivative 5 with biotin attached to the 7-position.

Labeling Short Duplex Oligodeoxynucleotides with M.HhaI and 5
To test whether the new 7-deazaadenosylaziridine derivative 5 functions as a cofactor for M.HhaI (Scheme 4), we initially used the short duplex ODN I·II (13 base pairs). Duplex I·II with the palindromic 5′-GCGC-3′ recognition sequence was prepared by annealing Strand I with the complementary Strand II. The target cytosine in the lower Strand II was blocked by methylation, i.e., cytosine is replaced with C5-methylcytosine (C Me ), and only the target cytosine in the upper Strand I can be modified by M.HhaI. Formation of Duplex I·II was analyzed by anion-exchange HPLC, and the duplex eluted as a single peak after 18.4 min. Duplex I·II was incubated with M.HhaI and 7-deazaadenosylaziridine derivative 5, and the coupling reaction was followed by anion exchange HPLC ( Figure 2). Directly after mixing of all components, the formation of a new compound with a retention time of 9.8 min was observed ( Figure 2a, Trace 1). In absence of either M.HhaI or 5, no formation of any additional compound was detected, indicating that the formation requires both M.HhaI and 5. Based on the altered UV-absorption ratio (260 nm/280 nm), the new compound was assigned to be a protein-DNA complex consisting of the coupling product I 5 ·II and M.HhaI (E260/280 (I·II) = 1.88; E260/280 (M.HhaI-I 5 ·II) = 1.54). All duplex ODN I·II had reacted to the M.HhaI-I 5 ·II complex after incubation at 37 °C for 3 h (Figure 2a, Trace 2). The protein-DNA complex between M.HhaI and the coupling product I 5 ·II is very stable and does not dissociate during anion exchange chromatography. A similar behavior has been observed before for the coupling of 8-modified N-adenosylaziridine cofactor 3 with a short duplex ODN by the DNA MTase M.TaqI [4]. The sample was heated to 65 °C for 30 min to denature the protein-DNA complex, and the liberated product duplex I 5 ·II (E260/280 = 1.80) eluted slightly earlier (17.7 min) than the starting Duplex I¨II was incubated with M.HhaI and 7-deazaadenosylaziridine derivative 5, and the coupling reaction was followed by anion exchange HPLC ( Figure 2). Directly after mixing of all components, the formation of a new compound with a retention time of 9.8 min was observed ( Figure 2a, Trace 1). In absence of either M.HhaI or 5, no formation of any additional compound was detected, indicating that the formation requires both M.HhaI and 5. Based on the altered UV-absorption ratio (260 nm/280 nm), the new compound was assigned to be a protein-DNA complex consisting of the coupling product I 5¨I I and M.HhaI (E 260/280 (I¨II) = 1.88; E 260/280 (M.HhaI-I 5¨I I) = 1.54). All duplex ODN I¨II had reacted to the M.HhaI-I 5¨I I complex after incubation at 37˝C for 3 h (Figure 2a, Trace 2). The protein-DNA complex between M.HhaI and the coupling product I 5¨I I is very stable and does not dissociate during anion exchange chromatography. A similar behavior has been observed before for the coupling of 8-modified N-adenosylaziridine cofactor 3 with a short duplex ODN by the DNA MTase M.TaqI [4]. The sample was heated to 65˝C for 30 min to denature the protein-DNA complex, and the liberated product duplex I 5¨I I (E 260/280 = 1.80) eluted slightly earlier (17.7   The product duplex I 5 ·II was characterized by enzymatic fragmentation with a mixture of DNase I, phosphodiesterase from Crotalus adamanteus, phosphodiesterase from calf spleen and alkaline phosphatase. The fragmentation products were analyzed by reverse-phase HPLC (Figure 2b). Beside the deoxynucleosides 2'-deoxycytidine (dC), C5-methyl-2'-deoxycytidine (dC Me ), 2'-deoxyinosine (dI) (formed from 2'-deoxyadenosine by contaminating adenosine deaminase activity), 2'-deoxyguanosine (dG) and 2'-deoxythymidine (dT), a new compound eluting after 22.9 min was observed. This new compound was isolated and detected as a positively-charged ion at m/z 824.4 by electrospray ionization mass spectrometry (ESI-MS). The observed mass is in good agreement with the calculated mass (824.3) for 2′-deoxycytidine modified with 5 (dC 5 ) and a sodium ion ([dC 5 + Na] + ). Taken together, these results demonstrate that 7-deazaadenosylaziridine derivative 5 is a cofactor for M.HhaI and can be used to covalently attach biotin to short duplex ODN.

Biotin Labeling of pUC19 Plasmid DNA with M.HhaI and 5
Next, we verified that M.HhaI and aziridine cofactor 5 can be used to label longer DNA fragments. pUC19 plasmid DNA (2686 bp) containing 17 recognition sequences for M.HhaI was selected as the test substrate. The pUC19 plasmid DNA was linearized by treatment with the restriction endonuclease R.XmnI (LpUC19) to facilitate analyses by agarose gel electrophoresis and electron microscopy (vide infra). To follow the labeling reaction, we employed a modification-restriction assay, which utilized the property of the restriction endonuclease R.HaeII to be unable to cleave its 5′-RGCGCY-3′ recognition sequence (R = G or A, Y = C or T, overlapping M.HhaI recognition sequence in bold) when modified by M.HhaI [19]. LpUC19 contains three R.HaeII recognition sequences, and fragmentation by R.HaeII (Trace 4) analysis of streptavidin binding to I 5¨I I; (b) Enzymatic fragmentation of the product duplex I 5¨I I with DNase I, phosphodiesterase from Crotalus adamanteus, phosphodiesterase from calf spleen and alkaline phosphatase analyzed by reverse-phase HPLC. The modified nucleoside (dC 5 ) elutes after 2'-deoxycytidine (dC), C5-methyl-2'-deoxycytidine (dC Me ), 2'-deoxyinosine (dI), 2'-deoxyguanosine (dG) and 2'-deoxythymidine (dT) (2'-deoxyadenosinewas converted to dI by contaminating adenosine deaminase activity during the fragmentation reaction).
The product duplex I 5¨I I was characterized by enzymatic fragmentation with a mixture of DNase I, phosphodiesterase from Crotalus adamanteus, phosphodiesterase from calf spleen and alkaline phosphatase. The fragmentation products were analyzed by reverse-phase HPLC (Figure 2b). Beside the deoxynucleosides 2'-deoxycytidine (dC), C5-methyl-2'-deoxycytidine (dC Me ), 2'-deoxyinosine (dI) (formed from 2'-deoxyadenosine by contaminating adenosine deaminase activity), 2'-deoxyguanosine (dG) and 2'-deoxythymidine (dT), a new compound eluting after 22.9 min was observed. This new compound was isolated and detected as a positively-charged ion at m/z 824.4 by electrospray ionization mass spectrometry (ESI-MS). The observed mass is in good agreement with the calculated mass (824.3) for 2 1 -deoxycytidine modified with 5 (dC 5 ) and a sodium ion ([dC 5 + Na] + ). Taken together, these results demonstrate that 7-deazaadenosylaziridine derivative 5 is a cofactor for M.HhaI and can be used to covalently attach biotin to short duplex ODN.

Biotin
Labeling of pUC19 Plasmid DNA with M.HhaI and 5 Next, we verified that M.HhaI and aziridine cofactor 5 can be used to label longer DNA fragments. pUC19 plasmid DNA (2686 bp) containing 17 recognition sequences for M.HhaI was selected as the test substrate. The pUC19 plasmid DNA was linearized by treatment with the restriction endonuclease R.XmnI (LpUC19) to facilitate analyses by agarose gel electrophoresis and electron microscopy (vide infra). To follow the labeling reaction, we employed a modification-restriction assay, which utilized the property of the restriction endonuclease R.HaeII to be unable to cleave its 5 1 -RGCGCY-3 1 recognition sequence (R = G or A, Y = C or T, overlapping M.HhaI recognition sequence in bold) when modified by M.HhaI [19]. LpUC19 contains three R.HaeII recognition sequences, and fragmentation by R.HaeII results four DNA fragments. Upon treatment of LpUC19 with M.HhaI and AdoMet 1 or another cofactor, the R.HaeII recognition sequences become modified, and R.HaeII is no longer capable of cleaving the modified LpUC19 plasmid DNA (Figure 3a).
Directly after mixing M.HhaI and the 7-modified aziridine cofactor 5 with LpUC19 plasmid DNA, some target sequences of R.HaeII were already blocked by coupling of 5 with M.HhaI recognition sequences (Figure 3b, Lane 4). After 30 min of incubation at 37˝C, the LpUC19 plasmid DNA was fully protected against cleavage by R.HaeII, indicating complete DNA modification (Figure 3b, Lane 6). Full protection was also observed when proteinase K was added after the modification reaction, and the plasmid purified before incubation with R.HaeII. This again indicates that the DNA becomes covalently modified. In contrast, complete fragmentation was observed in control experiments in the absence of either M.HhaI or cofactor 5, which demonstrates that both DNA MTase and cofactor are required for DNA modification (Figure 3b, Lanes 2 and 3).  It should be noted that biotinylation of LpUC19 plasmid DNA with the 7-modified aziridine cofactor 5 was performed with a stoichiometric (two-fold) excess of M.HhaI over 5′-GCGC-3′ recognition sequences. This is because the coupling reaction of aziridine cofactor 7 with the target cytosine results in one covalently-linked product, which stays tightly bound to the enzyme (see Section 2.2) and prevents further turnover. With the natural cofactor AdoMet 1, a transfer reaction occurs, resulting in two products, the methylated DNA and the demethylated cofactor S-adenosyl-L-homocysteine, and dissociation of the enzyme from the product complex can readily occur to allow further turnover. It should be noted that biotinylation of LpUC19 plasmid DNA with the 7-modified aziridine cofactor 5 was performed with a stoichiometric (two-fold) excess of M.HhaI over 5 1 -GCGC-3 1 recognition sequences. This is because the coupling reaction of aziridine cofactor 7 with the target cytosine results in one covalently-linked product, which stays tightly bound to the enzyme (see Section 2.2) and prevents further turnover. With the natural cofactor AdoMet 1, a transfer reaction occurs, resulting in two products, the methylated DNA and the demethylated cofactor S-adenosyl-L-homocysteine, and dissociation of the enzyme from the product complex can readily occur to allow further turnover.

Localization of Streptavidin-Biotin Complexes on Plasmid DNA by Electron Microscopy
The sequence-specificity of the labeling reaction was directly verified by electron microscopy. LpUC19 plasmid DNA was biotinylated with M.HhaI and the 7-modified aziridine cofactor 5 as described above. The biotinylated plasmid DNA was purified and incubated with streptavidin. The excess of streptavidin was removed by gel filtration and the plasmid prepared for electron microscopy. Using this procedure, only a small amount of streptavidin-crosslinked plasmid molecules was found by electron microscopy. Streptavidin molecules bound to DNA were observed as bulges on LpUC19 plasmid DNA (Figure 4a). Analysis of 84 plasmid molecules with added streptavidin demonstrated a clear preference for the predicted positions of M.HhaI recognition sequences (Figure 4b).
Molecules 2015, 20, page-page described above. The biotinylated plasmid DNA was purified and incubated with streptavidin. The excess of streptavidin was removed by gel filtration and the plasmid prepared for electron microscopy. Using this procedure, only a small amount of streptavidin-crosslinked plasmid molecules was found by electron microscopy. Streptavidin molecules bound to DNA were observed as bulges on LpUC19 plasmid DNA (Figure 4a

CpG-Methylation Detection
DNA biotinylation with aziridine cofactor 5 and M.HhaI was used to specifically detect DNA methylation of LpUC19 plasmid DNA by electromobility shift assay (EMSA) (Figure 5a). Cytosine residues within the 5′-CG-3′ DNA sequences (CpG-motifs) were methylated using the DNA cyotsine-C5 MTase from Spiroplasma sp. strain MQ1 (M.SssI) and the natural cofactor AdoMet 1. Non-and CpG-methylated LpUC19 were treated with biotinylated aziridine cofactor 5 and M.HhaI, which can only alkylate the first cytosine residue within the 5′-GCGC-3′ DNA sequence if this residue is not blocked by methylation of the inner CpG-motif. The different plasmids were then treated either with the cognate restriction endonuclease R.HhaI, which is blocked by C5-alkylation of the first cytosine within the 5′-GCGC-3′ recognition sequence, or streptavidin, leading to fragmentation, no reaction or an electrophoretic mobility shift. Methylation-sensitive enzymatic biotinylation of LpUC19 was analyzed by agarose gel electrophoresis (Figure 5b).
Analysis by restriction endonuclease protection was performed by treating samples of unmodified (A), CpG-methylated (B), CpG-methylated and treated with 5 and M.Hhal (C) or biotin-labeled (D) LpUC19 with R.HhaI. Fragmentation by R.HhaI occurred only with unmodified LpUC19 (A), while CpG-methylated (B or C) or biotinylated (C) LpUC19 was protected against cleavage by R.HhaI

CpG-Methylation Detection
DNA biotinylation with aziridine cofactor 5 and M.HhaI was used to specifically detect DNA methylation of LpUC19 plasmid DNA by electromobility shift assay (EMSA) (Figure 5a). Cytosine residues within the 5 1 -CG-3 1 DNA sequences (CpG-motifs) were methylated using the DNA cyotsine-C5 MTase from Spiroplasma sp. strain MQ1 (M.SssI) and the natural cofactor AdoMet 1. Non-and CpG-methylated LpUC19 were treated with biotinylated aziridine cofactor 5 and M.HhaI, which can only alkylate the first cytosine residue within the 5 1 -GCGC-3 1 DNA sequence if this residue is not blocked by methylation of the inner CpG-motif. The different plasmids were then treated either with the cognate restriction endonuclease R.HhaI, which is blocked by C5-alkylation of the first cytosine within the 5 1 -GCGC-3 1 recognition sequence, or streptavidin, leading to fragmentation, no reaction or an electrophoretic mobility shift. Methylation-sensitive enzymatic biotinylation of LpUC19 was analyzed by agarose gel electrophoresis (Figure 5b).
Analysis by restriction endonuclease protection was performed by treating samples of unmodified (A), CpG-methylated (B), CpG-methylated and treated with 5 and M.Hhal (C) or biotin-labeled (D) LpUC19 with R.HhaI. Fragmentation by R.HhaI occurred only with unmodified LpUC19 (A), while CpG-methylated (B or C) or biotinylated (C) LpUC19 was protected against cleavage by R.HhaI (Figure 5b, Lanes 2). These results confirm complete modifications of LpUC19.
For the functional analysis of biotinylation samples of unmodified (A), CpG-methylated (B), CpG-methylated and treated with 5 and M.Hhal (C) or biotin-labeled (D) LpUC19 were incubated with streptavidin. In the presence of streptavidin (Figure 5b, Lanes 3), a reduced electrophoretic mobility caused by the binding of streptavidin was only observed with biotinylated LpUC19 (D), whereas unmodified LpUC19 (A) and CpG-methylated LpUC19 (B or C) did not change their electrophoretic mobility. This result clearly confirms that the labeling reaction with aziridine cofactor 5 and M.HhaI is sequence specific and blocked by CpG methylation. Thus, this system can be used to distinguish between non-methylated and CpG-methylated DNA sequences.
Molecules 2015, 20, page-page electrophoretic mobility. This result clearly confirms that the labeling reaction with aziridine cofactor 5 and M.HhaI is sequence specific and blocked by CpG methylation. Thus, this system can be used to distinguish between non-methylated and CpG-methylated DNA sequences.

Experimental Section
Nomenclature: In this section, we have applied the systematic nomenclature of the International Union of Pure and Applied Chemistry (IUPAC) for naming synthesized compounds and assigning NMR data. This is different from the other sections, where the common purine nomenclature is used, e.g., the 7-position in the purine nomenclature translates into the 5-position in the IUPAC nomenclature. Material

Experimental Section
Nomenclature: In this section, we have applied the systematic nomenclature of the International Union of Pure and Applied Chemistry (IUPAC) for naming synthesized compounds and assigning NMR data. This is different from the other sections, where the common purine nomenclature is used, e.g., the 7-position in the purine nomenclature translates into the 5-position in the IUPAC nomenclature.
General Procedures: All air-or water-sensitive chemical reactions were carried out in dried glassware under an argon atmosphere. Silica gel 60 F 254 glass plates (Merck, Darmstadt, Germany) were used for TLC. Flash chromatography was carried out using Merck silica gel 60 (40-63 µm). HPLC was performed using a Waters Breeze System equipped with a binary programmable pump system 1525, a dual wavelength absorbance detector 2487 and a Waters inline degasser. NMR spectra were recorded using a Mercury 300 (75 MHz for 13 C), Inova 400 (400 MHz, 100 MHz and 376 MHz for 1 H, 13 C and 19 F, respectively) and a Unity 500 (500 MHz for 1 H) (all Varian) in the NMR spectroscopy facility of the institute. CDCl 3 (δ H = 7.24 and δ C = 77.0) or [D 6 ]DMSO (δ H = 2.49 and δ C = 39.5) were used as solvents. Assignments of 13 C signals are based on 1 H-, 13 C-correlated 2D-NMR and on 13 C-DEPT spectra. Electrospray ionization mass spectra (ESI-MS) were obtained using a Finnigan LCQ DECA XP Plus in the mass spectrometry facility of the institute. Measurements were carried out in the positive ion mode. UV absorption measurements were performed in methanol or water using a Varian Cary 3E spectrometer.
The Netherlands). Contour lengths were measured with a LM4 (Brühl, Nürnberg, Germany) and the data analyzed with a computer program [27].

Conclusions
We designed and synthesized a new 7-deazaadenosylaziridine nucleoside 5 with biotin attached to the 7-position and demonstrate that 5 functions as a cofactor for the DNA MTase M.HhaI. The quantitative coupling reaction is sequence specific, as demonstrated by electron microscopy, and biotin is functionally attached to DNA, as evidenced by streptavidin binding to the modified DNA. Several other cofactor analogues have been prepared and used to modify DNA with M.HhaI. They include N-adenosylaziridine with azide at the 8-position [8], N-adenosyl mustards (which are thought to form reactive aziridinium intermediates) with an azide or terminal alkyne group attached via short linkers to the 8- [12] or N6-position [13], as well as AdoMet analogues with extended methyl group replacements for transfer of amino [28,29], terminal alkyne or azide groups [30,31]. These cofactor analogues have in common that unique functional groups, like primary amines, terminal alkynes or azides, are attached to the DNA and a second chemical step is required for DNA labeling. Although this offers flexibility with regard to the reporter groups, it has the disadvantage that the second step can be difficult to follow, especially when labeling long plasmid DNA. This can result in partially-labeled DNA. The 7-modified aziridine cofactor 5 already contains biotin as the reporter group, and DNA labeling is achieved in one step, which can be easily analyzed with suitable restriction endonucleases.
It is interesting to note that some of these cofactor analogues for two-step DNA labeling contain substituents at the 8-position. Apparently, the steric demand of these substituents, like azide, propargylamine or 4-azidobutylamine, is smaller than the N-(4-aminobutyl)biotinamide in the 8-modified N-adenosylaziridine 4, which is not accepted by M.HhaI (Figure 3b, Lanes 9-12).
The new 7-modified cofactor 5 should also be a substrate for other DNA MTases. In fact, parallel experiments with the adenine-specific DNA MTase M.TaqI show that the 7-modified cofactor 5 is coupled with DNA even more rapidly than the 8-modified cofactor 4. In addition, three-dimensional structures of further DNA MTases in complex with AdoMet 1 or adenosine derivatives indicate that they should be able to bind either the 7-or the 8-modified aziridine cofactors 5 and 4 for coupling with their DNA recognition sequences [5].
One interesting application of M.HhaI in combination with AdoMet analogues for DNA labeling could be DNA methylation detection in mammalian DNA. Using M.HhaI and the aziridine cofactor 5 non-methylated CpG sites within the 5 1 -GCGC-3 1 recognition sequence can be labeled with biotin for affinity enrichment and DNA sequencing, while CpG-methylated recognition sequences will be protected against biotinylation. Such an approach has been recently reported for the M.SssI DNA MTase and AdoMet analogues with extended methyl group replacements [31]. It is also feasible that M.HhaI will be capable of coupling other 7-modified aziridine cofactors carrying fluorophores with DNA and use them for CpG-methylation-dependent DNA labeling and detection by DNA mapping [29,32,33]. Thus, SMILing DNA could not only become an interesting tool for sequence-specific labeling of DNA, but also for the exploration of DNA methylation in mammalian cells.