Kinetic studies of Escherichia coli AlkB using a new fluorescence-based assay for DNA demethylation

Strains and genomic DNA

HK82 [thr-1 araC14 leuB6(Am) DE(gpt-proA)62 lacY1 tsx-33 qsr′-0 glnV44 galK LAM- Rac-0 hisG4(Oc) rfbC1 mgl-51 rpoS396(Am) rpsL31(strR) kdgK51 xylA5 mtl-1 argE3(Oc) thi-1 nalA alkB22] was a kind gift from M. Sekiguchi. BH249 (=HK82λDE3) was constructed using the λDE3 Lysogenization Kit from Novagen (Madison, WI, USA). Escherichia coli strain BL21(DE3) was obtained from Novagen (Madison, WI, USA) Psuedomonas putida KT2440 genomic DNA was obtained from American Type Culture Collection (Manassas, VA, USA).

Chemicals and oligonucleotides

Methylated Substrates 5′-CGTCGXATTCTAGAGCCC C-3′, where X is 1-methyl-dA or 5′-CGTCGAATTXTA GAGCCCC-3′ where X is 3-methyl-dC, were synthesized by ChemGenes Corp (Wilmington, MA, USA). 3-Acetylpyridine adenine dinucleotide (APAD⁺), α-ketoglutaratic acid, l-ascorbic acid and HEPES were purchased from Sigma–Aldrich Co. (St Louis, MO, USA). Formaldehyde solution (16%) was obtained from Electron Microscopy Sciences (Hatfield, PA, USA).

Expression and purification of E. coli AlkB

The alkB⁺ gene was amplified from the E. coli B strain BL21 genomic DNA using PCR with the primers 5′-GGTGGTGAATTCATGTTGGATCTGTTTGCC-3′ and 5′-GGTGGTCTCGAGTTATTCTTTTTTACCTGC-3′. The amplification product was cloned into pGEX-4T-3 (Amersham) as an EcoRI–XhoI fragment to create the GST-AlkB fusion plasmid pGAlkB. The plasmids pGAlkB, and pET28a+ were digested with restriction enzymes BamHI and XhoI. The fragment containing the alkB⁺ gene was extracted and purified from agarose gel with the Agarose Gel-Out Kit (EURx Ltd., Gdańsk, Poland). The DNA fragment was then ligated into the digested pET28a+ vector and transformed into E. coli strain DH10B. Transformants were selected on LB plates with kanamycin. The resulting plasmid was designated pETAlkB and alkB gene was verified by DNA sequencing.

pETAlkB was introduced into E. coli strain BL21(DE3) and cells were grown in LB broth at 37°C to an OD₆₀₀ of 0.5. Fe(NH₄)₂(SO4)₂ was added to a concentration of 10 μM and the transcription of the fusion gene was induced by adding IPTG to the growth media to 1 mM. Cells were grown for another 3 h and harvested. The cell pellet was resuspended in 20 ml of 1× Ni/NTA Bind buffer [50 mM HEPES pH 8.0, 10 μM Fe(NH₄)₂(SO4)₂, 300 mM NaCl and 5 μM imidazole] and frozen at –70°C. The suspension was thawed on ice and sonicated for 10 pulses of 10 s duration separated by 1 min. The resulting lysate was centrifuged for 20 min at 12 000g in an Eppendorf 5810R centrifuge to pellet cell debris. The supernatant was then passed over a 500 μl bed volume of Ni/NTA His-Bind resin (Novagen, Madison, WI, USA). The bound protein was washed with three 2 ml volumes of wash buffer [50 mM HEPES pH 8.0, 10 μM Fe(NH₄)₂(SO4)₂, 300 mM NaCl and 20 μM imidazole] and eluted with four 500 μl volumes of elution buffer[50 mM HEPES pH 8.0, 10 μM Fe(NH₄)₂(SO4)₂ ,300 mM NaCl and 250 μM imidazole]. The fractions eluted were combined and dialyzed with a 30 kDa MWCO Amicon Ultra dialysis spin column at 4°C with SP–Sepharose running buffer [50 mM HEPES pH 7.3, 10 μM Fe(NH₄)₂(SO4)₂ and 2 mM DTT]. The concentrated protein solution was adjusted to a volume of 2 ml in SP–Sepharose running buffer and applied to an 8 ml SP–Sepharose cation exchange affinity column using an Amersham AKTA FPLC system. Proteins were eluted using a 0–1 M gradient of NaCl and collected in 1 ml fractions. The protein fractions were then analyzed by 15% SDS–PAGE eletrophoresis to identify the fractions containing AlkB. Fractions found to contain AlkB were concentrated and dialyzed with a 30 kDa MWCO Amicon Ultra dialysis spin column at 4°C against storage buffer [100 mM HEPES, pH 8, 20 μM Fe(NH₄)₂(SO4)₂, 4 mM EDTA and 6 mM DTT], glycerol was added to 50% v/v, aliquoted into 1 μl aliquots and stored at –70°C. Analysis of 5 μg of the final AlkB concentrate using 12% Tris–Tricine SDS–PAGE followed by staining with Coomassie blue showed to a single protein band (Supplementary Data, Figure S1).

Expression and purification of P. putida FDH

The P. putida FDH gene was PCR amplified from strain KT2440 genomic DNA (ATCC, Manassas, VA, USA) using the primers 5′-GGGGCATATGTCTGGCAATCG TGGAGTGG-3′ and 5′-GGGGCTCGAGTTACGCCG CACCCCACATTTTGTGC-3′. It was then cloned into pET28a+(Novagen) as an EcoRI–XhoI fragment to create the IPTG-inducible poly(His)–FDH fusion plasmid pETFDH.

The plasmid containing His-Tag FDH fusion, pETFDH was introduced into E. coli strain BH249. With the exception of cell culture incubation at 25°C, overproduction and purification over a Ni/NTA affinity column was performed in the same manner as described for AlkB. The eluent fractions from the Ni/NTA column were combined and dialyzed by a 30 kDa MWCO Amicon Ultra dialysis spin column at 4°C using 20 mM Tris buffer, pH 6.5. The concentrated protein solution was resuspended to a volume of 2 ml in 20 mM Tris pH 6.5 and applied to an 8 ml Q–Sepharose anion exchange affinity column on an Amersham AKTA FPLC system. Proteins were eluted using a 0–1 M gradient of NaCl and collected in 1 ml fractions. The protein fractions were then analyzed by 15% SDS–PAGE. The fractions found to contain FDH protein were concentrated and dialyzed with a 30 kDa MWCO Amicon Ultra dialysis spin column at 4°C against storage buffer (100 mM HEPES, pH 8, 4 mM EDTA and 6 mM DTT), glycerol was added to 50% v/v, aliquoted into 12 μl aliquots and stored at –70°C. Analysis of 5 μg of the final FDH concentrate using a 12% Tris–Tricine SDS–PAGE followed by staining with Coomassie blue gave rise to a single protein band (Supplementary Data, Figure S1).

Determination of FDH activity

Purified FDH (26 pmol) and 10 µmol formaldehyde were incubated in reaction buffer [20 mM HEPES, pH 8, 200 μM α-ketoglutarate, 2 mM l-ascorbic acid, 20 μM Fe(NH₄)₂(SO₄)₂, 1 mM NAD⁺ and 100 μg/ml BSA] at 22°C in a final volume of 100 μl. The production of NADH was monitored for 10 min by measuring absorbance at 340 nm using a Beckman DU 530 UV/Vis spectrophotometer. One unit of FDH activity is defined as 1.0 µmol formaldehyde oxidized per minute under the reaction conditions described above.

Formaldehyde standard curve

A standard curve of formaldehyde concentration versus intensity of APADH fluorescence was prepared by the following method. Various concentrations of formaldehyde (1–20 μM) were incubated with 1 U of FDH, APAD⁺ (1 mM) in AlkB reaction buffer [20 mM HEPES, pH 8, 200 μM α-ketoglutarate, 2 mM l-ascorbic acid, 20 μM Fe(NH₄)₂(SO₄)₂ and 100 μg/ml BSA] in a final volume of 120 μl. The incubations were monitored continuously for the production of APADH with a Varian Eclipse spectrofluorometer (excitation at 363 nm; emission at 482 nm) for 10 min at 25°C. Peak intensity values were plotted against formaldehyde concentration to create the standard plot (Supplementary Data, Figure S2).

Determination of AlkB steady-state kinetics

Double-stranded (DS) methylated substrates were generated by incubating the SS oligonucleotides containing the methyl lesions with 1.5-fold molar excess of the unmethylated complementary oligonucleotide at 37°C for 30 min in annealing buffer (50 mM HEPES, pH 8.0, 10 mM EDTA).

AlkB (final concentration 14 nM) was incubated with increasing concentration of substrate (2.1–42 μM) in the AlkB reaction buffer containing 0.3 U FDH and APAD⁺(final concentration 1 mM) in 120 μl. All the components of the reaction except substrate DNAs were mixed together and incubated at 25°C for 5 min. Reactions were initiated with the addition of DNA and were monitored continuously for the production of APADH with a Cary Varian Eclipse spectrofluorometer (excitation at 363 nm; emission at 482 nm). Typically, four reactions were monitored concurrently for 25 min, each set containing one blank (no DNA) and three cuvettes containing identical DNA concentration (i.e. triplicates). Initial reaction velocities for the rate of formaldehyde production were determined using the formaldehyde standard curve (Figure S2). The data was analyzed using the software Prism (GraphPad, San Diego, CA, USA) and K_m and k_cat values were determined by non-linear regression curve fitting of the data to the Michaelis–Menten equation.

Development of assay for demethylation

To detect formaldehyde released during the demethylation of DNA, we converted it to formate using FDH. During this conversion NAD⁺ is converted to NADH, and the latter molecule is usually detected based on its UV absorbance (17). In our experiments, we used APAD⁺ instead of NAD⁺ as the cofactor for FDH because of its higher redox potential leading to greater stability of the reduced form at various pHs and a higher extinction coefficient which contributes to greater sensitivity (18). We also chose to use fluorescence properties APADH to develop the demethylation assay because of its greater sensitivity over UV absorbance by a factor of ∼30 (Figure 2). Consequently, we were able to lower the amount of methylated DNA base used for the assay from several nanomoles—for a typical UV-absorbance assay—to subnanomole amounts. This allowed us to use synthetic oligonucleotides containing methylated bases in a defined sequence context as substrates for kinetic studies. Most previous studies have used randomly methylated poly-dA (or poly-dC) or DNAs containing both 1mA and 3mC for their assays (2,9,13,14,19,20).

During some preliminary experiments, we discovered that the commonly used buffering agent, Tris, was a substrate for FDH and led to conversion of APAD⁺ to APADH in the absence of AlkB or methylated DNA substrate in the reaction (Figure 3). This is probably because FDH also acts on some alcohols (21). Tris contains three primary alcohol groups that may be converted to acidic form by FDH. To avoid the contribution of this reaction to background fluorescence, HEPES was used as buffering agent in the reactions instead of Tris.

In coupled enzyme assays, there can be a lag period between the first and the second reaction. This is the time period before the coupling reaction converts the product of the first reaction (formaldehyde, in this case) to the final product (formate). This can affect the measurement of initial rates. However a ‘rule of thumb’ is that this lag period is negligible if a doubling of the coupling enzyme does not affect the rate of the overall reaction (22). This was found to be the case under the conditions used here (data not shown). In addition, if the maximal velocity of the coupling enzyme (FDH here) is much greater than the enzyme being studied, the lag time becomes inconsequential (23). In this case, the high specific activity of FDH (>250 µmol formaldehyde oxidized min⁻¹ mg⁻¹) as compared with the specific activity for AlkB (<0.14 µmol min⁻¹ mg⁻¹) satisfies this criterion.

However, we found that commercially available preparations of FDH were not homogeneous and contained material that contributed to background fluorescence even when DNA and AlkB were omitted from the reaction (data not shown). To avoid this problem, the FDH gene from P. putida was expressed in E. coli and purified to homogeneity (Supplementary Data, Figure S1). The conditions for the complete assay using purified FDH and AlkB enzymes are described in Materials and Methods section.

Steady-state kinetics for E. coli AlkB

Four substrates were used for studying AlkB kinetics. Two of them contained a 29-base DNA oligomer with a single 1mA and the other two substrates had the same base sequence but contained instead one 3mC. Unmethylated complementary DNAs were hybridized to the methylated DNAs to create the DS substrates and both SS and DS versions of the methylated DNAs were used for the demethylation assays. The initial velocities of APADH production from three or more independent reactions for five or six concentrations of substrates are shown in Figure 4 and Supplementary Figure S3.

The results were considered reproducible as indicated by the small standard deviations and could be fit to Michaelis–Menten equation or its alternate forms (Figures 4 and S3). It is seen that (Figures 4 and S3, insets) the [S]/V versus [S] plots are linear suggesting that the enzyme follows Michaelis–Menten kinetics under these conditions. The k_cat and K_m values for AlkB with the four substrates were calculated based on a fit of the V versus [S] plot to the Michaelis–Menten equation and these values are presented in Table 1. The reproducibility of the data collected using the new assay is evidenced by the small standard deviations seen in the Figures 4 and S3, and in Table 1.

Comparison of AlkB kinetic parameters

The k_cat and K_m values for the four substrates tested were fairly similar and the range of variation in catalytic specificity of the enzyme for the substrates was <2-fold (Table 1). While this shows that the enzyme does not have a strong preference for any one of the substrates, it did show a small preference for 1mA over 3mC and SS DNA over DS DNA. These results are qualitatively similar to some of the reported substrate preferences for AlkB, but they do show some important differences from the previous reports.

Falnes and colleagues (13) reported that AlkB prefers SS DNA over DS by a factor of 2 to 5. This conclusion was based on experiments where specific activity of AlkB was determined over a range of concentrations using DNA oligomers containing both A's and C's that were randomly labeled with ³H-methyl groups. In contrast, we find that the ratios of catalytic specificities (k_cat/K_m) of the enzyme for the SS DNA to DS DNA substrates is only 1.4 for the 1mA and 1.9 for 3mC (Table 1). It is possible that these differences in the conclusions from the two studies may reflect the differences in the base sequences of DNA used in the two studies. Additionally, we also note that the experiments reported by Falnes et al. (13) were only semi-quantitative and hence the numbers presented in Table 1 are likely to be more representative of substrate preferences of AlkB.

Koivisto and colleagues (14) studied AlkB using a randomly methylated poly-dA substrate and reported k_cat and K_m values respectively to be 11.7 ± 1.3 min⁻¹ and 1.4 ± 0.2 μM. This k_cat value is ∼3-fold higher, while the K_m value is ∼4-fold lower than the values reported for SS 1mA-containing DNA we studied (Table 1). Thus the catalytic specificity of AlkB for methylated poly-dA substrate appears to be ∼11-fold higher than the 1mA-containing substrate used here. These differences may reflect sequence differences between the two DNAs, but may also be due to secondary structure differences. While the substrate we used is likely to form secondary structure through intra- and inter-molecular hydrogen bonds, poly-dA cannot adopt such conformations. Thus the preference of AlkB for SS DNA may partly explain the differences between the two studies. Additionally, AlkB may act processively on substrates containing multiple methylated DNA bases. If so, the catalytic efficiency of the enzyme for demethylation of randomly labeled substrate would be higher than that containing a single methylated base such as the substrates used in our studies.

The results reported above add two additional useful insights into AlkB action. First, they show that both 1mA- and 3mC-containing DNA substrates are better substrates in SS form than DS form. Previous work on substrate specificity of AlkB have generally used DNAs containing more than one methylated base. Second, no catalytic constants have been reported for AlkB using a substrate containing 3mC as the only methylated base. The results reported here show that AlkB acts with roughly the same efficiency on the two methylated bases tested, regardless of whether they are in SS or DS form (Table 1).

Summary and future prospects

We have presented here a new assay for DNA demethylating enzymes and illustrated its use by determining AlkB kinetic parameters. The assay uses subnanomole amounts of DNA substrate per reaction, is continuous, rapid and non-toxic. We have done preliminary experiments with human orthologs of AlkB, ABH2 and ABH3 and these experiments show that these enzymes should also be amenable to study using this technique (data not shown). The assay may be adapted to stopped-flow studies and hence should be particularly useful to study the reaction mechanism of this class of enzymes. It may also be used to study mutant enzymes or to study inhibitors of this class of DNA repair enzymes. We have also adapted this assay to 96-well microtiter plate reader (data not shown) and in this format it may be used for high-throughput analysis of demethylating enzymes or their mutants.

We note that this assay is not specific for enzymes that demethylate DNA. Oxidative demethylation of histones methylated at lysines can also generate formaldehyde (24), which can be detected using FDH and APAD⁺ as described here. In fact Shi et al. (25) used FDH and the generation of NADH to monitor the production of formaldehyde by the histone demethylase LSD1. However, they used UV absorption at 340 nm to quantify NADH. Instead, using APAD⁺ and measuring fluorescence would improve the sensitivity and stability of this assay. Finally, many oxidative detoxification reactions in the liver also create formaldehyde and these enzymes could also be studied using the same assay. Thus the fluorescence-based assay for DNA demethylation described here may be adapted to the study of a wide variety of cellular enzymes.