Site-specific methylated reporter constructs for functional analysis of DNA methylation

Abstract

Methods to experimentally alter and functionally evaluate cytosine methylation in a site-specific manner have proven elusive. We describe a site-specific DNA methylation method, using synthetically methylated primers and high fidelity PCR coupled with ligation of reporter constructs. We applied this method to introduce methylated cytosines into fragments of the respective DAPK and RASSF1A promoters that had been cloned into luciferase reporters. We found that methylation of 3–7 residue CpG clusters that were 5′ adjacent to the transcription start site (TSS) of the DAPK gene produced up to a 54% decrease in promoter activity (p < 0.01). Similarly, for RASSF1A promoter reporter constructs, the methylation of either of two clusters of four CpGs each, but not an intervening cluster, produced a 63% decrease in promoter activity (p < 0.01), suggesting that precise mCpG position is crucial, and factors other than simple proximity to the TSS are at play. Chromatin immunoprecipitation analysis of these reporter constructs demonstrated that transcription factor Oct-1 and Sp1 preferentially bound the unmethylated vs. methylated DAPK or RASSF1A promoter reporter constructs at the functional CpG sites. Histone H1, hnRNP1, and MeCP2 showed preferential binding to methylated sequence at functional sites in these reporter constructs, as well as highly preferential (>8–80-fold) binding to native methylated vs. unmethylated chromatin. These results suggest that: (1) site-specific, precision DNA methylation of a reporter construct can be used for functional analysis of commonly observed gene promoter methylation patterns; (2) the reporter system contains key elements of the endogenous chromatin machinery.

Keywords: :

• DNA methylation
• site-specific DNA methylation
• functional analysis
• epigenetics
• transcription factor

Introduction

DNA methylation is an important process during normal development and homeostasis, and frequently becomes deregulated during disease pathogenesis, including carcinogenesis.^1,2 DNA methylation profiles at the level of individual CpGs exhibit substantial variation across molecules, cells, tissues, and phenotypes within the normal human population.^3-9 So far, little is known about the functional implications of the precise spatial distribution patterns of DNA methylation among tissues/individuals.

Functional analysis of individual CpG dinucleotides is desirable for better understanding of normal cellular physiology, inter-individual variation of DNA methylation, disease pathogenesis and for the optimization of cancer biomarker searches for robustness and precision, where it is critical to select CpG dinucleotides that exhibit a central role in gene silencing.

The high-resolution analysis of CpG methylation, including evaluating its functional implications, is an obligatory step in rational approaches to both epigenetic biomarker development, and targeted epigenetic interventions. Development of a site-specific DNA methylation technique is critical for functional analysis of individual CpG dinucleotides. In 2002, Curradi et al.¹⁰ reported a patch methylation technique, but the method has not worked in our lab, and there are no further publications about this technique. Other techniques for experimental manipulation of DNA methylation have included siRNA directed DNA methylation, found in plant and yeast; there are a few papers supporting siRNA directed DNA methylation in human cells.^11-15 Zinc finger-DNA methyltransferase has also been used for site-directed DNA methylation in vitro at low precision.^16-19 Teng et al. reported that transfection of methylated DNA complementary to target DNA could induce the DNA methylation of the target region also at low precision.^20-22 Recently, a CpG-free luciferase reporter vector was reported for functional analysis of DNA methylation using methyltransferase SssI.²³ Among the limitations of these techniques is that the methylation cannot be precisely targeted to individual single CpGs as a site-specific intervention. Recently, Arakawa et al. developed PCR base method for in vitro site-specific DNA methylation, but this method is only used for linear vectors, and is quite cumbersome.²⁴

Here we report a new protocol which is based on high fidelity PCR coupled with ligation using synthetically methylated primers for site specific methylation of reporter constructs; subsequent testing of the impact of specific methylation clusters was performed. We also explored the DNA-binding proteins associated with these methylation-specific effects. This PCR-ligation reaction technique could be used to make single or multiple CpG methylation changes in a reporter construct of interest. Therefore, it becomes possible to couple high-resolution observations of CpG methylation patterns^25-27 to experimental recapitulation of the precise methylation sequence, and proceed to functional analysis. This technique could be useful in functional epigenetic studies, epigenetic biomarker development, and targeting for precise epigenetic interventions.

Results

Principle of site-specific DNA methylation

The principle of the method is shown in Figure 1 in four steps. Step 1, PCR-ligation coupled reaction: A reporter construct is used for site specific methylation modification and two primers with desired CpG methylation and 5′-end phosphorylated are synthesized for the amplification of the entire constructs. The primers are extended during temperature cycling by high fidelity DNA polymerase without strand-displacement activity, and generate new complementary strand DNA containing a nick, which can then be ligated by Taq DNA ligase. After 25 to 30 cycles, ~2 μg DNA product could be produced. Step 2, Dpn1 and T7 exonuclease digestion: Following temperature cycling, the product is purified and digested with Dpn1 and T7 exonuclease, in which the parental DNA template is digested by Dpn1 that recognizes 5′-G^m6ATC-3′ sequences in the parental plasmid derived from E. Coli; the linear DNA is then digested by T7 exonuclease. Step 3, Dnmt1 modification: The DNA is purified and treated with human DNA Methyltransferase (Dnmt1, New England Biolabs Inc.). Dnmt1 is specific for hemi-methylated DNA and is used to methylate the DNA strand complementary to the methylated primer. Step 4, Bisulfite sequencing: The methylated CpG site is verified by bisulfite genomic DNA sequencing.

Figure 1. Strategy for site-specific DNA methylation. Step1, Conventional reporter constructs are adapted as template. A pair of primers (arrows) containing synthetically methylated CpGs (stars) are phosphorylated at 5′ end. Both primers are extended by high-fidelity DNA polymerase to generate new complementary strand DNA containing a nick. The new complementary strand DNA is ligated to form circular DNA by Taq DNA ligase. Step 2, The parental DNA template is digested by Dpn1 which recognizes 5′-G^m6ATC-3′, which are initially derived from E. coli., and the resulting linear DNA is digested by T7 exonuclease. Step 3, Dnmt1 is used to methylate the remaining DNA strand complementary to the methylated primer. Step 4, The integrity of the intended methylated CpG sites is confirmed by bisulfite sequencing before transfection.

PCR-ligase coupled amplification of DAPK reporter constructs

A 268 bp DAPK promoter fragment inserted into Luciferase reporter constructs (5011 bp) was effective as template. Two primers designed for effecting the site-specific DNA methylation by the PCR-ligation reaction included the synthetically methylated forward primer: 5′-GCCGCCTGCC GGCCGCCCTC CG-3′, in which the four bold C’s are methylated cytosine. Two high fidelity DNA polymerases (pfuUltra II Fusion HS DNA polymerase and Herculase II Fusion DNA polymerase were tested for PCR-ligase coupled amplification with Taq DNA ligase. PCR-ligation products were digested with Dpn1 and T7 exonuclease. Results showed that Herculase II Fusion DNA polymerase coupled with Taq DNA ligase produced high yield of circled DNA products (Fig. 2).

Figure 2. PCR-ligase coupled amplification of DAPK reporter constructs. Two primers containing four methylated CpGs were used for PCR-ligase coupled amplification of DAPK reporter constructs (5011bp). (A) pfuUltra II Fusion HS DNA polymerase (lane 1) and Herculase II Fusion DNA polymerase (lane 2) could amplify the constructs along with Taq DNA ligase. (B) PCR-ligation products were digested with Dpn1 and T7 exonuclease, lane 1: pfuUltra II Fusion HS DNA polymerase product; lane 2: Herculase II Fusion DNA polymerase product; lane 3: 1kb plus DNA ladder.

Verification of site-specific methylation in DAPK promoter reporter constructs

The methylation status of the newly-synthesized construct was confirmed with tag modified bisulfite sequencing (tBGS)²⁵ and then the constructs were transfected into A549 cells. Thirty-six hours after transfection, total DNA was extracted for methylation status reassessment of the reporter construct. After bisulfite treatment, the DAPK promoter was amplified with vector-specific primers and the methylation status was again determined by tag-modified bisulfite sequencing transfection. The results showed that only the intended four target CpG sites were specifically methylated in the constructs prior to transfection, and they remained methylated after transfection for at least 36 h (Fig. 3; Fig. S1).

Figure 3. Bisulfite sequencing of the DAPK promoter in site-specific methylated reporter constructs. Site-specific methylated constructs (MP2) were used for bisulfite treatment. The promoter region was amplified with vector specific primers flanking the promoter. PCR product was subject to direct Sanger sequencing.²⁵ (A) The sequencing result before transfection into A549 cells. (B) The sequencing result after transfection into A549 cells for 36 h. (C) The anti-sense strand sequencing result after transfection into A549 cells for 36 h. The upper sequence in each pair is genome sequence in which the underlined target sequence containing four methylated cytosines are in red characters. In the bisulfite sequencing, all unmethylated cytosines were converted to thymine. Only the four intended target cytosines were methylated in the bisulfite sequencing result. Corresponding chromatograms are depicted in the supplement (Fig. S1).

Functional analysis of DAPK promoter methylation

We created six site-specific methylated DAPK promoter reporter constructs (MP1-MP6) using six methylated PCR primers (Table S1) containing the same methylated CpGs clusters. As controls, unmethylated PCR primers were used for creating unmethylated DAPK promoter reporter constructs by the same protocol. Results showed that the methylation of the three CpGs (MP1) (out of 31 CpGs in the promoter) adjacent to the transcription start site (TSS) produced a 54% decrease in promoter activity (p = 0.002) as compared with the unmethylated DAPK promoter (Fig. 4). The methylation of four CpGs (MP2) at -44~-61 and seven CpGs at -130~-156 (MP4) decreased the promoter activity by 31% (p = 0.01) and 32% (p = 0.02). The methylation of the 12 CpG sites most upstream from the TSS had no significant effect on promoter activity. We then methylated the 12 CpGs combining the CpGs from each of three methylated oligos (MP1, MP2, MP3) together, to determine if there are any synergetic effects on the promoter activity to larger regions of methylation. Results showed that there were no synergetic effects of these combined CpGs on DAPK promoter activity repression (Fig. S2).

Figure 4. Functional analysis of CpG methylation impact on DAPK promoter activity. Regions ~60 bp in length (30 bp forward and 30 bp reverse primers fit end-to-end) were tested. The numbers of methylated CpG sites were determined by the endogenous sequence in that 60 bp region (e.g., four CpGs at -10, -23, and -34 for MP1, three CpGs for MP2, five CpGs for MP3, seven CpGs for MP4, nine CpGs for MP5 and three CpGs for MP6). Unmethylated DAPK reporter constructs (unM) and six site-specific methylated DAPK reporter constructs (MP1 to MP6) were transfected into A549 cells, PRL-TK which expresses Renilla luciferase was co-transfected as an internal control. The empty vector (Basic) was also transfected as negative control. All constructs were transfected in triplicates. After 36 h, luciferase activities were determined by dual luciferase assay. Firefly luciferase activity was normalized to Renilla luciferase activity (internal control). Here, the most 3′ cluster (the three CpGs methylated together by the MP1 patch, which covers the region -34 to -10, reference to TSS) appears to halve the promoter activity. P values represent t test comparisons with the unmethylated DAPK promoter insert (unM) control. KEY: Basic: the reporter construct without promoter; unM: The reporter construct with unmethylated DAPK promoter; MP1 to MP6: reporter constructs with DAPK promoter methylated at different CpG sites; mCpG: Methylated CpG.

Functional analysis of DNA methylation in RASSF1A promoter reporter constructs

To further demonstrate our site specific methylation technique is reproducible and applicable to other genes, we tested it on RASSF1A promoter reporter constructs with two different fragment sizes of the same promoter sequence (~300 bp and 1 kb). We created four site-specific methylated RASSF1A promoter reporter constructs using four pairs of methylated PCR primers (Table 1) containing 4 to 5 methylated CpGs each. Results showed that the methylation of two clusters of four CpGs each (out of 18 analyzed CpGs in the promoter) adjacent to the TSS produce a 63% decrease in promoter activity (p < 0.01), as compared with the unmethylated RASSF1A promoter (Fig. 5). The methylation of the other 10 CpG sites had no significant effect on promoter activity. The results were robust and identical, whether the smaller or larger encompassing region was cloned into the reporter vector.

Table 1. Proteins differentially binding to methylated or unmethylated DNA oligos repeating DAPK and RASSF1A proximal promoters by LC-MS/MS

Protein	Methylated DNA oligo	Function
Histone H1	DAPK-MP1 RASSF1A-MP1	Involved in the formation of higher order chromatin structures, modulate the accessibility of regulatory proteins, chromatin-remodeling factors, and histone modification enzymes to their target sites^27^-^29
Heterogeneous nuclear ribonucleoprotein H1/A1 (hnRNPH1/A1)	DAPK-MP1 RASSF1A-MP1	RNA binding proteins, involved in pre-mRNA processing and mRNA metabolism and transport.^30^,^31
Ribosomal protein L6 (RPL6)	RASSF1A-MP1	A component of the Ribosome 60S subunit. It may participate in tax-mediated transactivation of transcription^32
ATP-dependent RNA helicase A (DHX9)	RASSF1A-MP1	As a transcriptional regulator, which catalyzes the ATP-dependent unwinding of double-stranded RNA and DNA-RNA complexes^33^,^34
MYB binding protein (P160) 1a (MYBBP1A)	RASSF1A-MP1 RASSF1A-P3	A transcription factor, interacting with P53 and P300 to regulate gene expression^35^,^36
X-ray repair cross-complementing protein 6 (XRCC6)	RASSF1A-MP1	Function as a single-stranded DNA-dependent ATP-dependent helicase and involved in the repair of non homologous DNA ends such as that required for double-strand break repair^37^,^38
SRp55–3	RASSF1A-MP3	Involved in pre-mRNA procession^39
Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)	RASSF1A-MP3	Catalyze an important energy-yielding step in carbohydrate metabolism and the reversible oxidative phosphorylation of glyceraldehyde-3-phosphate in the presence of inorganic phosphate and nicotinamide adenine dinucleotide^40
Activating signal co-integrator 1 (ASCC1)	RASSF1A-P1 RASSF1A-P3	A transcriptional co-activator that plays an important role in gene transactivation by multiple transcription factors including activating protein 1 (AP-1), nuclear factor kappa-B (NF-kB) and serum response factor (SRF)^41^,^42
DEK oncogene (DNA binding) (DEK)	RASSF1A-P1	This protein binds to cruciform and superhelical DNA and induces gene expression.^43^,^44
DNA topoisomerase 1 (TOP1)	RASSF1A-P1 RASSF1A-P3	An enzyme that controls and alters the topologic states of DNA during transcription.^45^,^46
Ribosomal protein L19 (RPL19)	RASSF1A-P1 RASSF1A-P3	Ribosomal protein that is a component of the 60S subunit^47
Ribosomal protein L13 (RPL13)	RASSF1A-P3	Ribosomal protein that is a component of the 60S subunit^48
Elongation factor Tu GTP binding domain containing 2 (EFTUD2)	RASSF1A-P3	A GTPase which is a component of the spliceosome complex which processes precursor mRNAs to produce mature mRNAs.^49
Heterogeneous nuclear ribonucleoproteins C1/C2 (hnRNPC1/c2)	RASSF1A-P3	RNA-binding protein that is associated with pre-mRNAs in the nucleus and appear to influence pre-mRNA processing and other aspects of mRNA metabolism and transport.^50
Nucleolar transcription factor 1 (UBTF)	RASSF1A-P3	A member of the HMG-box DNA-binding protein family, which plays a critical role in rRNA transcription.^51
PRP8 pre-mRNA processing factor 8 (PRPF8)	RASSF1A-P3	A component of both U2- and U12-dependent spliceosomes, which is essential for the catalytic step II in pre-mRNA splicing process.^78

Figure 5. Functional analysis of DNA methylation impact on RASSF1A promoter reporter constructs. Unmethylated RASSF1A reporter constructs (unM) and four site-specific methylated reporter constructs (MP1 to MP4) were transfected into A549 cells. PRL-TK, which expresses Renilla luciferase was co-transfected as an internal control. The empty vector (Basic) was also transfected as negative control. All constructs were transfected in triplicates. After 36 h, luciferase activities were determined by dual luciferase assay. Firefly luciferase activity was normalized to Renilla luciferase activity (internal control). Here, each of two clusters, each containing four methylated CpGs at +21,+15,+2,-6 and -138, -129,-125,-102, reference to TSS, appear to halve promoter activity. P values represent t test comparisons with the unmethylated RASSF1A promoter insert (unM) control. Basic, the reporter construct without promoter; unM, the reporter construct with unmethylated RASSF1A promoter; MP1 to MP4, reporter constructs with RASSF1A promoter methylated at different CpG sites; mCpG, methylated CpG.

Protein binding analysis on methylated and unmethylated promoter

To begin to understand how the critical CpG clusters in DAPK and RASSF1A promoter impact on gene expression, we compared nuclear proteins binding to the methylated vs. unmethylated oligos, by performing gel-shift assays. The results (Fig. 6) showed that the methylated DAPK probe (DAPK-MP1) did bind nuclear extract proteins, which could then be competitively inhibited by unlabeled methylated oligo DNA (Fig. 6A, lanes 1, 2). The unmethylated probe (DAPK-P1) displayed some non-specific binding (Fig. 6A, lanes 3, 4). Similarly, methylated RASSF1A probes (RASSF1A-MP1 and RASSF1A-MP3) also bind nuclear extract proteins, which could be partially inhibited by unlabeled methylated oligo DNA (Fig. 6B and C, lanes 1, 2). The unmethylated RASSF1A probes (RASSF1A-MP1 and RASSF1A-MP3) also displayed some non-specific binding (Fig. 6B and C, lanes 3, 4).

Figure 6. LightShift EMSA for biotin labeled methylated and unmethylated DAPK-MP1, RASSF1A –MP1, and MP3 oligos from the DAPK and RASSF1A promoter. (A): Lane 1: Nuclear extract proteins specifically bind to the methylated DAPK-MP1 probe. Lane 2: The binding of nuclear extract proteins to the methylated DAPK-MP1 was competed with unlabeled methylated DAPK-MP1. Lane 3: Nuclear extract proteins nonspecifically bind to unmethylated DAPK-P1 probe. Lane 4: The nonspecific binding of nuclear extract proteins to the unmethylated DAPK-P1 was not competed with unlabeled unmethylated DAPK-P1. Lane 5: Methylated DAPK-MP1 only as negative control. (B): Lane 1: Nuclear extract proteins specifically bind to the methylated RASSF1A-MP1. Lane 2: The binding of nuclear extract proteins to the methylated RASSF1A-MP1 was partially inhibited by unlabeled methylated RASSF1A-MP1. Lane 3: Nuclear extract proteins bind to unmethylated RASSF1A-P1. Lane 4: The binding of nuclear extract proteins to the unmethylated RASSF1A-P1 was competed with unlabeled unmethylated RASSF1A-P1. Lane 5: RASSF1A-MP1 only as negative control. (C): Lane 1: Nuclear extract proteins specifically bind to the methylated RASSF1A-MP3. Lane 2: The binding of nuclear extract proteins to the methylated RASSF1A-MP3 was partially inhibited by unlabeled methylated RASSF1A-MP3. Lane 3: Nuclear extract proteins bind to unmethylated RASSF1A-P3. Lane 4: The binding of nuclear extract proteins to the unmethylated RASSF1A-P3 was competed with unlabeled unmethylated RASSF1A-P3. Lane 5: RASSF1A-MP3 only as negative control. NE: nuclear extracts; Mb: methylated DNA labeled with biotin; M: methylated DNA unlabeled; uMb: unmethylated DNA labeled with biotin; uM: unmethylated DNA unlabeled.

Identification of proteins binding to the methylated oligo

In order to identify the responsible proteins binding to the methylated oligos (DAPK -MP1, RASSF1A-MP1, and RASSF1A-MP3), we immobilized the methylated and unmethylated DNA oligo to the Dynabeads (Invitrogen), and pulled down the binding proteins by affinity purification. The results of LC-MS/MS analysis (Fig. S3) showed that: for DAPK promoter there are two proteins (histone H1 and hnRNPH1) specifically binding to the methylated DAPK-MP1 (Table 1), and five proteins non-differentially binding to both methylated DAPK-MP1 and unmethylated DAPK-P1 (Table S2). For the RASSF1A promoter, there are five proteins (including histone H1 and hnRNPH1) differentially binding to the methylated RASSF1A-MP1, and two proteins specifically binding to the methylated RASSF1A-MP3 (Table 1). There are additional four proteins specifically binding to the unmethylated RASSF1A-P1 and nine proteins specifically binding to the unmethylated RASSF1A-P3 (Table 1). There are 10 proteins non-differentially binding to both methylated RASSF1A-MP1 and unmethylated RASSF1A -P1, 14 proteins non-differentially binding to both methylated RASSF1A-MP3 and unmethylated RASSF1A –P3 (Table S2).

Chromatin immunoprecipitation analysis of reporter constructs containing functional CpG sites

In silico sequence analyses revealed that the three CpG sites methylated in the DAPK-MP1 encompass a putative binding site of transcription factor Oct-1, and that there are 16 putative Sp1 binding sites in the DAPK promoter and 17 putative Sp1 binding sites in the RASSF1A promoter. We did not find any transcription factors (Oct-1 or Sp-1) nor known methylated CpG binding proteins in the LC-MS/MS results; only histone H1 and hnRNP uniquely bound the methylated DNA oligos (DAPK-MP1 and RASSF1A-MP1) containing the functional CpG sites. We considered the limitation of detectability of LC-MS/MS for lower abundant proteins, which is about 200 pmol.⁵² We therefore performed ChIP assay in A549 cells transfected with DAPK-MP1, RASSF1A-MP1, and RASSF1A-MP3 reporter constructs using anti-Histone H1 and anti-hnRNP, anti-Oct1, anti-Sp1, and anti-MeCP2 antibodies (Table 2; Fig. S4). The results of real time PCR using reporter construct specific primers indicated that Oct-1 was uniquely bound to unmethylated DAPK construct and not the methylated DAPK-MP1. Sp1 was also preferentially bound to promoters in unmethylated reporter constructs of both DAPK and RASSF1A; the binding ability to methylated DAPK-MP1, RASSF1A-MP1, and RASSF1A-MP3 reporter constructs was reduced to 82%, 10%, and 16% of their unmethylated counterparts (Table 2; Fig. S4). MeCP2 was bound to each of methylated promoter constructs (Fig. S4). The histone H1 and hnRNPH1 were highly preferentially bound to methylated constructs (Table 2; Fig. S4), which is consistent with the results of ChIP assay from the unmethylated promoter in native chromatin from NHBE cells and methylated promoter in native chromatin from MCF7 cells (Table S3; Fig. S5). The ChIP results therefore indicate that the reporter system contains many elements of the endogenous chromatin machinery.

Table 2. Real time PCR results for ChIP assay

Antibody	DAPK -MP1 * (n = 3)	DAPK -UM * (n = 3)	DAPK -MP1/-UM **	RASSF1A-MP1 * (n = 3)	RASSF1A-MP3 * (n = 3)	RASSF1A -UM * (n = 3)	RASSF1A -MP1/-UM **	RASSF1A -MP3/ -UM **
Histone H1	0.016 ± 0.0012	5.9 × 10^−5 ± 9.4 × ± 10^−6	271	8.3 × 10^−3 ± 7.8 × 10^−4	2.3 × 10^−3 ± 6.6 × 10^−4	1.9 × 10^−6 ± 4.5 × 10^−7	4368	1210
hnRNPH1	0.055 ± 0.02	2.3 × 10^−4 ± 7.0 × 10^−5	239	8.2 x10^−3 ± 7.4x10^−4	6.1 × 10^−3 ± 9.1 × 10^−4	0	∞	∞
Oct-1	0	0.025 ± 0.007	0	0	0	0	0	0
Sp1	1.8 × 10^−3 ± 4.4 × 10^−4	2.2 × 10^−3 ± 3.0 × 10^−4	0.82	1.9 × 10^−2 ± 3.7 × 10^−3	0.029 ± 0.0044	0.18 ± 0.036	0.10	0.16
MeCP2	0.21 ± 0.042	0	∞	4.2 × 10^−2 ± 3.1 × 10^−3	0.047 ± 0.01	0	∞	∞

* The precipitated DNA was normalized to input DNA by 1/2^∆CT, ∆CT equals CT value of the immuno-precipitated DNA minus CT value of input DNA. ± SD, n = 3. **The ratio of the immuno-precipitated DNA from methylated constructs vs. that from the unmethylated constructs was calculated by 1/2^∆∆CT, ∆∆CT equals ∆CT of the immuno-precipitated DNA from methylated constructs minus ∆CT of the immuno-precipitated DNA from unmethylated constructs. The final real-time PCR product is shown in Figure S4.

Discussion

DNA methylation sequencing in a variety of systems and contexts reveals complex patterns of promoter methylation. We have here shown that methylation of certain specific CpG sites drive the repression of a reporter gene more than other CpGs and confirmed some of the proteins mediating that effect. The methylation reporter method used is novel in that synthetically methylated oligos are precisely incorporated into the reporter constructs by a PCR + ligation reaction to create precisely methylated clusters, mimicking same of the detail of in vivo chromatin patterns we have previously observed.^25,53

The site-specific DNA methylation technique reported here is based on PCR coupled with DNA ligation, and is suitable for a circular vector. In our method, because the PCR is coupled to a Taq DNA ligase reaction, the reaction conditions of the DNA polymerase must be compatible with Taq DNA ligase. Therefore, we chose a compatible DNA polymerase for the DNA ligase combination base on the yield of the circular PCR product. Our method is quite different from a recently reported method for linear reporter vector creation,²⁴ in which PCR and ligation are in different reactions, the PCR products are prepared in single-strand form before ligation, and that strategy, as a consequence, is somewhat more complex to execute.²⁴

For our primer design, the target methylation CpGs are integrated into primers, and the 5′ end must be phosphorylated for DNA ligation. The primer size could be up to 60 bp, as tested in our lab to date. The forward and reverse primers do not have to be 5′ end adjacent since the DNA extension is circular, but the extension time should be adjusted to the reporter construct size. In PCR-ligation DNA products, some products may have one strand ligated and circular, the other strand is still nicked because of ligation inefficiency. Addition of a ligation step using ligase specific buffer may increase circular DNA up to 20% (not shown).

We made the arbitrary choice to test promoter sequences from two tumor suppressor genes. DAPK (Death-associated protein kinase) is a serine/threonine, microfilament-bound kinase involved in interferon-, TNF- or FAS-induced apoptosis.^54,55 Loss of function of this kinase could significantly affect cellular apoptosis. Methylation of the DAPK promoter is seen in both lung adenocarcinoma and squamous-cell carcinoma at prevalences ranging from 30–48%,^56-58 and appears to occur early in carcinogenesis.^59,60

When we inserted a 300bp DNA fragment from the DAPK promoter directly upstream of the transcription start site into the luciferase reporter construct and progressively methylated the 31 CpGs in individual clusters of convenience, encompassing every CpG site in the fragment, we found that three CpGs (out of 31 CpGs in the promoter fragment) adjacent to the transcription start site (TSS) produce the most significant decrease in promoter activity as compared with the unmethylated DAPK promoter. Other clusters of four methylated CpGs at one site (-44~-61) and five methylated CpGs at another (-54~-112) also decreased the promoter activity. When methylating all 11 proximate CpGs from DAPK-MP1, DAPK-MP2, and DAPK-MP3 together in one reporter, there were no apparent synergetic effects on the promoter activity. We speculate that this lack of synergy may be due to the fact that these CpGs are very close together (within 100 bp), and it is plausible that one large complex of DNA methylation proteins could potentially block the entire 100 bp region from transcription factors access. This is a hypothesis requiring further testing.

We also verified the methylation reporter using part of the RASSF1A promoter. This member of the RAS superfamily of GTP-binding proteins has an important role in signal-transduction pathways that controls cell proliferation, differentiation, and death,^61,62 and it has been reported that RASSF1A could be inactivated in 40–60% and 100% of non-small-cell and small-cell lung cancers, respectively.^63,64

Two promoter DNA fragments (~1000 bp, and a ~300 bp subfragment of the first) from the same RASSF1A gene promoter sequence were each inserted into identical reporter constructs, and the same cluster of methylation was created in each by site-specific methylation. The results indicate that the methylation effects on the promoter activities were reproducible, and not affected by the size of inserted promoter, suggesting some robustness to broader sequence context. There were two specific regions (MP1 and MP3) that showed reduction of reporter expression, in both the larger and smaller RASSF1A promoter fragment constructs, spanning another (MP2) where no such effect was apparent. This non-response to the MP2 cluster methylation suggests sequence and/or position specificity of the methylation effect on reporter expression, and that the effect is not attributable to mere proximity to the TSS alone. Recent reports^65,66 showed that RASSF1A methylation could be a candidate biomarker for lung cancer. We compared the methylated CpG sites, Deng et al.⁶⁵ examined and some were included in our MP1 and MP2 region, Fraipont et al.⁶⁶ checked the CpGs downstream of the TIS.

As to local mechanism of methylation impact on reporter expression, the proteins associating with the methylation effects in the DAPK and RASSF1A reporters tested were initially explored by EMSA for the relevant methylated/unmethylated oligo pair. The subsequent LC-MS/MS analysis detected differential binding of histone H1 and hnRNPH1, but did not find methylated CpG binding proteins (MeCP) or any transcription factors, we presume due to sensitivity limits of LC-MS/MS. We therefore proceeded to test expected transcription factors (Oct-1 for DAPK, Sp1 for both) and MeCP2 with ChIP of the reporter constructs themself. Sp1 is a sequence-specific transcription factor that recognizes GC boxes.⁶⁷ In silico sequence analyses revealed that the TATAless, GC-rich promoter of the DAPK and RASSF1A contained many Sp1 binding sites, but only one Oct-1 binding site in the DAPK-MP1 region of interest. Oct1 is widely expressed in adult tissues, and is implicated in crucial and versatile biological events, such as embryogenesis, neurogenesis, immunity, and body glucose and amino acid metabolism.⁶⁸ ChIP analysis demonstrated that MeCP2 uniquely bound to the methylated reporter constructs and transcription factor Oct-1 and Sp1 could bind to unmethylated DNA but the binding was inhibited by DNA methylation at these binding sites. Histone H1 is an abundant and essential component of chromatin and has strong competition with MeCP for nucleosome binding sites.^69,70 _.There are papers in the literature suggesting that certain histone H1 subclasses bind preferentially to GC-rich DNA^71-75 and other H1 subclasses bind primarily to AT rich sequences.^76,77 hnRNPH1 has not previously been associated with DNA methylation regulation. One can speculate this RNA binding protein is a logical regulatory point for pre-mRNA processing, mRNA metabolism, and transport, but this clearly requires further dissection. Overall, the results indicate that the reporter system contains many elements of the endogenous chromatin regulatory machinery.

In conclusion, we have developed a simple PCR-ligation based method for site-specific methylation in reporter constructs. This method has three advantages: complete precision in site specificity of methylation, simplicity, and high methylation efficiency. The limitation is that the methylation alterations are limited to the two PCR primer regions, and therefore strategies to circumvent this size constraint need to be developed. This method can be used to aid the validation of specific CpG methylation sites as physiologically relevant, patho-physiologically relevant, and/or as potential disease biomarkers. This technical development could also help guide development of sequence-specific alteration of DNA methylation for preventive and therapeutic purposes.

Materials and Methods

Construction of promoter-luciferase constructs

The respective 5′ flanking regions of the DAPK and RASSF1A genes were amplified from genomic DNA of normal human bronchial epithelial cells (NHBE; Lonza, Inc.) by high fidelity PCR. PCR reactions were prepared in 50-μl volumes containing 1× Pfx Amplification buffer (invitrogen) 0.3 μM of each primer, 0.3 mM of each of the four deoxynucleoside triphosphates (dNTPs), 200 ng of genomic DNA. One unit of AccuPrime™ Pfx DNA polymerase (Invitrogen). Thermocycling was conducted using either the GeneAmp® PCR System 9600 (Applied Biosystems) for 95 °C 5 min, 35 cycles of 95 °C 15 s, 60 °C 30 s and 68 °C 1 min. Three pairs of primers were used (the underlined sequences are Sac1 and Nco1 restriction sites for cloning): DAPK-F: CATTATAGAG CTCGCGTGGG TGTGGGGCGA GTG and DAPK-R: TAATTCATCC ATGGCGTTGG CTCGGTCCGG CTGT for amplification of 268 bp (-243~+25) DAPK promoter. In order to test sequence context, two differently sized fragments (288 bp and 1023 bp) of the proximal RASSF1A promoter were obtained using two pairs of primers: RASSF1A-F1: CATTATAGAG CTCAAGTGTG TTGCTTCAGC AAACC and RASSF1A-R1: TAATTCACCA TGGCGGCGGG AAGGAGCTGA GG for amplification of 288 bp (-266~+22) RASSF1A promoter. RASSF1A-F2: CATTATAGAG CTCCCTAGAT CCCAGAAATC TGGGAG, and RASSF1A-R1 for amplification of 1023 bp (-1000~+22) RASSF1A promoter. PCR products were purified from an agarose gel and digested with Sac1 and Nco1. The three digested fragments were cloned into the Sac1/Nco1 site of the promoterless pGL3-basic firefly luciferase reporter vector (Promega) to produce three constructs. All constructs were sequenced prior to use in order to ensure a correct insertion sequence and direction.

Overall strategy of site-specific DNA methylation

To methylate the specific CpG site in the promoter of reporter constructs, we used a site-specific DNA methylation technique, where a pair of primers containing synthetically methylated CpGs is employed to amplify the recombinant reporter construct with high-fidelity PCR (Fig. 1).

The two primers are 5′-end adjacent for the amplification of the entire constructs, and 5′-end phosphorylated for ligation. The primers are extended during temperature cycling by high fidelity DNA polymerase without strand-displacement activity, and generate new complementary strand DNA containing a nick, which can then be ligated by Taq DNA ligase. Following temperature cycling, the parental DNA template is digested by Dpn1 which recognizes methylated 5′-G^m6ATC-3′ sequences in the parental constructs, which are initially derived from E. coli. The linear DNA is digested by T7 exonuclease. After purification, the product is treated with human DNA Methyltransferase (Dnmt1, New England Biolabs Inc.). Dnmt1 is specific for hemi-methylated DNA) and is used to methylate the DNA strand complementary to the methylated primer. The methylated CpG site is verified by bisulfite genomic DNA sequencing.

Site-specific methylation of promoter reporter construct

We created six site-specific methylated DAPK promoter reporter constructs and eight site-specific methylated RASSF1A promoter reporter constructs using PCR primers containing 3 to 9 methylated CpGs each (Table 1). As controls, unmethylated PCR primers were used for creating unmethylated promoter reporter constructs by the same protocol. The PCR-ligation reaction was performed in 50 µl 1× Herculase II reaction buffer (Stratagene), 1mM NAD⁺ (NEB), 0.5 µl Herculase II fusion DNA polymerase, 160 U Taq DNA ligase and 200 µM each dNTP, using 10 ng reporter constructs as template, 250nM each primer. Reaction condition: denature at 98 °C for 2 min, then 98 °C × 20 s, 60 °C × 30 s, 72 °C × 3 min, and 65 °C ×5 min for 26 cycles. The PCR-ligation reaction is the critical step in this site-specific DNA methylation protocol. The denature temperature 98 °C is necessary for supercoiled plasmid templates. The DNA polymerase should have no strand displacement activity. The reaction condition should be compatible with tag DNA ligase (e.g., pH value, salt concentration). The PCR-ligation product was purified with QIAquick PCR purification kit (Qiagen) and subject to digestion with Dpn1 and T7 exonuclease in NEB buffer 2 overnight for removing the parental DNA and non-circular DNA. The circular DNA constructs containing the methylated primer were purified with QIAquick PCR purification kit (Qiagen) and treated with Dnmt1 to methylate the complementary strand of the methylated primer. Dnmt1 treatment was performed in 50 μl reaction containing 50 mM Tris-HCl, 1 mM Dithiothreitol, 1 mM EDTA, 5% Glycerol, 100 μg/ml Bovine Serum Albumin and 160 μM S-adenosylmethionine, 2 μg DNA, 4 units of Dnmt1, pH 7.8 and incubated at 37 °C for 4 h. The removing of linear DNA and parent DNA by Dpn1 and T7 exonuclease before Dnmt1 treatment may increase the efficiency of Dnmt1 due to the removing of unnecessary substrate. The site-specific methylated constructs were subject to bisulfite treatment and the promoter region was amplified for DNA sequencing. Site-specific methylated and unmethylated promoter reporter constructs were transfected into A549 cells. After 36 h incubation, promoter activity was determined by dual luciferase assay.

Cell culture and transfection

A549 cells were maintained in F-12K nutrient mixture medium (Invitrogen) supplemented with 10% (v/v) FBS 10 µg/ml gentamicin at 37 °C in a humidified 5% CO₂ atmosphere. Transfections were performed in triplicate in 24-well plates. Each well was plated with 2 × 10⁵ cells 24h before transfection. Transfection was performed using 0.5 μg of reporter gene construct, together with 20ng of reference PRL-TK vector (Promega) expressing Renilla luciferase constitutively, which serves as an internal control of transfection efficiency. DNA was combined with 3 μl of Lipofectamine LTX and 1 μl of plus reagent in Opti-MEM reduced serum medium (Invitrogen). After a 36 h incubation, the cells were lysed by Passive Lysis Buffer (Promega). The activities of firefly and Renilla luciferases were determined by Dual-Luciferase reporter assay system (Promega). Firefly luciferase activity was then numerically normalized to Renilla luciferase activity.

GC tag-modified bisulfite genomic DNA sequencing (tBGS) ²⁵

Bisulfite treatment was performed with DNA methylation kit (Zymo Research), with the reaction condition optimized to 50 °C for 20 h. DNA was eluted in 10 µl of elution buffer. Bisulfite-treated genomic DNA was amplified by two step PCR as described previously.²⁵ Step 1: PCR amplification with the vector specific primers. Step 2: Nest PCR using tag-modified primers: an 18–45 base tag containing at least 10 cytosines is added to the 5′end of each of the forward primer, similarly, an 18–45 base tag containing at least 10 guanines is added to the 5′end of each of the reverse primer. The product of the first step PCR is used as template. The nested PCR product is purified by DNA gel extraction kit (Qiagen), and subject to directly sequencing, using the tag as sequencing primer. The GC tag-modified primer was used because for bisulfite treated DNA, due to the bisulfite conversion of unmethylated C ≥ U in the template, there is low GC content in the PCR product. This is especially apparent when the target sequence is unmethylated.¹⁹ This low GC content PCR product often produces C-noisy signal background in sense strand sequencing result, or G noisy signal background in antisense strand sequencing result, making the result impossible to interpret. Therefore, we used GC enriched tag-modified PCR primers, which yields a 5′ and 3′ GC-tagged PCR product that sufficiently enhances the GC content to allow successful direct PCR product sequencing.

Electrophoretic mobility shift assay (EMSA)

Synthesized methylated and unmethylated double strand oligo DNA containing critical functional CpG sites of DAPK and RASSF1A promoter were end labeled with biotin (Integrated DNA Technologies). Nuclear extracts were prepared from A549 cells, using NE-PER nuclear extraction kit (Pierce Biotechnology). EMSA was performed with LightShift Chemiluminescent EMSA Kit (Pierce Biotechnology).

LC-MS/MS analysis

Approximately 9.0 mg of magnetic Dynabeads (Invitrogen) was incubated with 5.0 μmol biotin labeled DNA for 10 min at room temperature in the binding and washing buffer containing 5mM Tris-HCl (pH7.5), 0.5 mM EDTA and 1 M NaCl. The DNA-coupled magnetic Dynabeads (Invitrogen) were washed with 500 µl of binding and washing buffer twice and incubated with 4 μg of poly(deoxyinosinic-deoxycytidylic; Pierce Biotechnology) and 4 mg of nuclear extracts from A549 cells for 30 min at 37 °C in Binding buffer (Pierce Biotechnology). The Dynabead-DNA–protein complex was precipitated, and washed with 500 µl of Binding buffer (Pierce Biotechnology) three times and then suspended in 50 µl of 25 mM buffer ABC (ammonium bicarbonate, pH 8.0). The bound proteins were released by incubating at 95 °C for 5min. and reduced using TCEP (tris[2-carboxyethyl]phosphine, final concentration 10 mM) for 30 min at room temperature, then alkylated using iodoacetamide (final concentration of 50 mM) for 30 min at room temperature in the dark. The final volume was adjusted to 100 μl using 25 mM buffer ABC for all approaches, then 1 μl of 1% ProteaseMax^TM trypsin enhancer and 200 ng of trypsin (Promega) were added and the digestion was performed at 37 °C overnight with gentle shaking. The reaction was stopped by adding 1 μl of 50% trichloroacetic acid (TCA) and centrifuged at 20 000 × g for 30 min. The supernatant was transferred to a new tube and directly analyzed by LC-MS/MS.

LC-MS/MS results were created from the raw mass spectra using Proteome Discoverer 1.2, merged and searched against the mammalian NCBI database (May 27, 2011) using the in-house Mascot Protein Search engine. The following parameters for searches were used: trypsin 2 missed cleavages; fixed modification of carbamidomethylation (Cys); variable modifications of deamidation (Asn and Gln) and oxidation (Met); monoisotopic masses; peptide mass tolerance of 3.0 Da; product ion mass tolerance of 0.6 Da. Proteins are considered identified having at least two bold red significant (p < 0.05) peptides.

Chromatin Immunoprecipitation (ChIP) analysis

Approximately 10⁶ cells were transfected with promoter reporter constructs using Lipofectamine as described above. After 36 h, cells were treated with 1% formaldehyde (final concentration, v/v) for 10 min at room temperature to cross-link proteins to DNA. Chromatin immunoprecipitation was performed using Pierce Agarose ChIP kit (Pierce Biotechnology). Briefly, the cross-linked DNA was fragmented with Micrococcal nuclease and precipitated with antibodies: anti-Histone H1 (Thermo Scientific), anti-hnRNP (Thermo Scientific), Anti-Oct1 (Santa Cruz), anti-Sp1 (Santa Cruz), and anti-MeCP2 (LifeSpan Biosciences). The target DNA was detected by real time PCR, using reporter construct specific primers: Forward: 5′-TATCGATAGG TACCGAGCTC-3′; Reverse: 5′-GTTTTTGGCG TCTTCCATGG-3′; After 45 cycles of real time PCR, the precipitated DNA was normalized to input DNA by 1/2^∆CT, (∆CT equals CT of the immunoprecipitated DNA minus CT of input DNA), The ratio of the immunoprecipitated DNA from methylated constructs vs. that from the unmethylated constructs was calculated by 1/2^∆∆CT (∆∆CT equals ∆CT of the immunoprecipitated DNA from methylated constructs minus ∆CT of the immunoprecipitated DNA from unmethylated constructs). The PCR product size for DAPK promoter reporter construct was 308 bp. The PCR product size for DAPK reporter construct was 308 bp and for RASSF1A construct was 328 bp. Final PCR product was analyzed on 1.5% agarose gel.

Data analysis

The promoter activities of the various methylated reporter constructs were determined by Dual-Luciferase reporter assay system (Promega). Firefly luciferase activity was then normalized to Renilla luciferase activity (internal transfection efficiency control). T test was used to compare triplicate luciferase levels of the variously methylated promoters with that of unmethylated promoters, using SigmaSTAT software.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by NIH grants: RC1 CA145422; R21 CA 121068; R03 CA132145; R01 CA106186; 1 R21 CA104812.

Acknowledgments

David Reynolds and the Genomics Core at Albert Einstein College of Medicine for help with DNA sequencing; Edward Nieves, Myrasol Callaway, and Berta Burd Proteomics Core of Albert Einstein College of Medicine for the LC-MS/MS analysis. Also, we offer gratitude to Drs Matthew Scharff, Eric Bouhassira, Vern Schramm, Quan Du, Joseph Locker, and Arthur Skoltchi for helping to clarify previous drafts of the manuscript.

Patent

Patent application for this method was submitted in 2011.

Supplemental Materials

Supplemental materials may be found here: www.landesbioscience.com/journals/epigenetics/article/26195

10.4161/epi.26195