P16-specific DNA methylation by engineered zinc finger methyltransferase inactivates gene transcription and promotes cancer metastasis

P16 DNA methylation directly inactivates gene transcription

In order to determine whether P16 DNA methylation directly inactivates gene transcription, a P16 promoter-specific DNA methyltransferase (P16-dnmt) was initially constructed using the pcDNA3.1_myc/His vector as described in the
methods section (Fig. 1a). Western blot analysis confirmed that endogenous P16 was greatly reduced in HEK293T
cells 48 h following transient transfection with the P16-dnmt vector (Fig. 1b). A dual-luciferase reporter assay further illustrated that P16 promoter activity was significantly inhibited in the P16-dnmt-transfected cells (Fig. 1c). Notably, methylation of CpG islands within both the P16 promoter and exon-1 regions was detected using denatured high performance liquid
chromatography (DHPLC) and bisulfite-sequencing (Fig. 1d and e). An additional control lacking approximately 80 % of the DNA methyltransferase activity
(R882H mutant) was constructed to evaluate the impact of steric hindrance from P16-Dnmt
DNA binding on gene transcription. As expected, chromatin-immunoprecipitation (ChIP)-PCR
analysis showed that the mutant still bound with the P16 promoter DNA fragment (Fig. 2a), but did not induce P16 DNA methylation (Fig. 2b). Furthermore, its capacity to repress P16 expression was sharply decreased in both HEK293T and BGC823 cells (Fig. 2c and d). These data suggest that P16 DNA methylation is directly responsible for P16 repression as opposed to steric hindrance. Taken together, these results indicate
that P16-dnmt encodes an active methyltransferase for P16 CpG islands, and P16 DNA methylation is sufficient to inactivate endogenous P16 expression.

Fig. 1. P16-specific methyltransferase (P16-dnmt) induces methylation of P16 CpG islands and represses gene transcription in HEK293T cells. a Construction of P16-dnmt using P16 promoter–specific seven-zinc finger protein (7ZFP) and the catalytic domain of mouse DNMT3a; b Western blot analysis for P16-Dnmt and endogenous P16; c Reporter assay results following P16-dnmt transfection; d DHPLC methylation analysis of the P16 promoter in HEK293T cells; the 392-bp methylated (M) and unmethylated (U) P16 fragments were detected at the partial denaturing temperature of 57.0 °C; e bisulfite clone-sequencing results from HEK293T cells transiently transfected with
P16-dnmt

Fig. 2. Comparison of DNA binding capacity, DNA methylation induction, and P16 expression levels between P16-dnmt and the R882H mutant in HEK293T and BGC823 cells 72 h following transient transfection.
a Chromatin-immunoprecipitation (ChIP)-PCR results comparing P16 promoter DNA binding for P16-Dnmt and R882H protein; b Methylation-specific PCR (MSP) detecting methylated and unmethylated P16 alleles; c Quantitative RT-PCR indicating P16 mRNA levels; d Western blot of P16 and Myc/P16-Dnmt protein levels

In order to specifically methylate P16 CpG islands, the P16-dnmt coding-sequence was then integrated into the pTRIPZ lentivirus vector carrying a
‘Tet-on’ switch to allow the gene expression to be controlled. Expression of P16-Dnmt
protein was induced in GES-1 cells stably transfected with the P16-dnmt pTRIPZ vector after treatment with 0.25 ?g/mL doxycycline for 3 days (61KD; Fig. 3a). Significant inhibition of endogenous P16 expression was observed in Western blot and quantitative RT-PCR analysis when compared
to GES-1 cells transfected with the dnmt3a and 7ZFP control vectors (Fig. 3a and b). Confocal microscopy revealed that the average density of nucleic P16 gradually
decreased in the P16-dnmt expressing cells (Fig. 3c). In fact, after treatment with doxycycline for 3 and 7 days, P16 expression levels were decreased by 21.4 % and 53.3 %, respectively (P 0.001). Most importantly, intensive methylation of P16 CpG islands was induced in the GES-1 cells stably transfected with P16-dnmt and treated with doxycycline, but not in cells transfected with the control vectors,
nor in cells that did not receive doxycycline treatment (Fig. 3d). Similarly, P16 DNA methylation and subsequent repression of P16 expression was also induced by P16-Dnmt in the BGC823 cell line (Additional file
1: Figure S1).

Fig. 3. P16 DNA methylation analysis of GES-1 cells stably transfected with the P16-dnmt pTRIPZ ‘Tet-on’ vector. a Western blot analysis of P16 and P16-Dnmt (61KD, arrow) and control vectors following
treatment with 0.25 ?g/mL doxycycline for 3 days. b Quantitative RT-PCR results for the P16-dnmt stably transfected, doxycycline-treated cells. c Confocal analysis with P16 labeling following 3 and 7 days of doxycycline treatment.
d Confirmatory DHPLC methylation analysis of P16-specific methyltransferase expressing cells (arrow) and controls following 7 days
of doxycycline treatment. The 567-bp methylated (M) and unmethylated (U) P16 promoter fragments were analyzed at the partial denaturing temperature of 54.0 °C.
DNA samples from HCT116 cells containing both the methylated and unmethylated P16 alleles were used as controls (M.U.Ctrl)

ChIP-PCR analysis also showed that P16-Dnmt specifically bound the P16 promoter, but not the P14 promoter (Additional file 1: Figure S2). Similarly, ChIP-sequencing confirmed that the P16-Dnmt binding fragment
was only detected in the promoter of P16-Dnmt/Myc antibody immunoprecipitated DNA
from the P16-dnmt-expressing BGC823 cells, but not in the IgG control, nor the cells transfected with
the control vector (Fig. 4a, red fragment; Additional file 2: File S1, Additional file 3: File S2, and Additional file 4: File S3). Although most of the P16-Dnmt binding fragments were found in intergenic
and intron sequences (Additional file 1: Figure S3A), the main P16-Dnmt binding motif was found to closely match the antisense
strand of the targeted fragment in the P16 promoter with a similarity of 21/23 (91.3 %) base pairs (Additional file 1: Figure S3C, red-framed motif). Genome-wide methylation analysis of P16-Dnmt expressing
BGC823 cells was performed using an Infinium Methylation 450 K array. The results
illustrated that 647 of 481,615 informative CpG sites (0.13 %) were significantly
hypermethylated (?? 0.50). Interestingly, 229 of these 647 CpG sites were located
in intragenic CpG islands and shores corresponding to 203 genes (Additional file 5: File S4). The targeted P16 CpG island was included in the list of differentially hypermethylated sites (Fig. 4a, blue arrow). Furthermore, DNA methylation was not induced in the CpG islands of
two control genes, P14 (located within the same CDKN2A locus as P16) and ZNF382 (located on a different chromosome) (Fig. 4b). These results suggest that doxycycline-induced P16-dnmt expression could specifically methylate P16 CpG islands.

Fig. 4. Characterization of the P16-Dnmt binding fragments and methylation status of the promoter
CpG islands in P16 and control genes in P16-dnmt stably transfected cells. a Chromatin-immunoprecipitation-sequencing (ChIP-Seq) and Illumina methylation 450
K array results in the ink4a locus in BGC823 cells; b DHPLC methylation analysis of CpG islands in P14 and ZNF382 promoters in GES-1 cells expressing P16-Dnmt. DNA samples with and without M.sssI methylation were used as methylated and unmethylated controls

P16-specific DNA methylation promotes migration and invasion of cancer cells

Various assays were then conducted to further characterize the biological behaviors
of cancer cells following P16-specific inactivation by DNA methylation. Transwell assays revealed that the migration
ability of GES-1 and BGC823 cells was significantly increased following P16-specific DNA methylation (Fig. 5a and b). Similarly, Matrigel assays showed that the invasion capacity of these cell lines
was also significantly enhanced by P16-specific DNA methylation (Fig. 5c and d).

Fig. 5. Migration and invasion assays carried out with cell lines stably transfected with
the P16-dnmt pTRIPZ vector in vitro. a, b Results of Transwell migration assays for GES-1 and BGC823 cell lines following 48-h
and 36-h incubation, respectively; c, d Results of Matrigel invasion assays for GES-1 and BGC823 cell lines following 108-h
and 96-h incubation, respectively; the average cell number and s.d. are displayed
(Right). Dox (+), with 0.25 ?g/mL doxycycline treatment; Dox (?), without doxycycline
treatment. These experiments were independently repeated in triplicate

Four weeks after BGC823 cells stably transfected with P16-dnmt were injected into the tail vein of the NOD SCID mice, metastatic nodules were observed
in the lung (Fig. 6a). The average lung weight, which correlates with the number of metastatic cells,
in the P16-dnmt group was 152.5 % that of the empty vector control group (Mann–Whitney test, P 0.001; Fig. 6b). The average proportion of metastatic nodule area to total lung area in the P16-dnmt group was also significantly higher than the control group (P 0.004, Fig. 6c).

Fig. 6. P16-specific methylation promotes experimental pneumonic metastasis of BGC823 cells.
a Images of representative metastatic nodules in the lung of SCID mice (HE staining).
b The lung weights of mice in the P16-dnmt pTRIPZ and control groups at day 19. c The ratio of metastatic nodule area to lung area of mice in the P16-dnmt pTRIPZ and control groups

In addition, P16-specific DNA methylation was found to slightly, but significantly, inhibit proliferation
of GES-1 cells, while the proliferation of BGC823 cells was not affected (Additional
file 1: Figure S4). However, growth inhibition of the P16-dnmt transfected GES-1 cells was not observed in the SCID mice despite detection of methylated-P16 alleles in the xenografts (Additional file 1: Figure S5).

In order to confirm whether the enhanced migration of cancer cells is P16 DNA methylation-specific, a rescue assay was carried out in the P16-dnmt expressing BGC823 cells through transient transfection of a P16 expression vector. Results of the Transwell assay demonstrated that enforced P16 overexpression significantly reversed the enhanced migration phenotype of these cells
(Fig. 7a). Similar results were also observed in HONE-1 cells (Additional file 1: Figure S6). In contrast, downregulation of endogenous P16 expression through transient siRNA transfection significantly enhanced the migration
of BGC823 and GES-1 cells (Fig. 7b). Furthermore, the migration capacity of A549 cells, which lack P16 alleles, was not changed following stable transfection of P16-dnmt and 7 days of doxycycline treatment (Fig. 7c). Taken together, these results imply that the enhanced migration and invasion phenotypes
of cancer cells are P16-specific.

Fig. 7. Effect of varying P16 expression on migration. a Rescue assay demonstrating the effect of enforced P16 overexpression on migration of the P16-dnmt stably transfected BGC823 cells treated with doxycycline for 2 weeks. b Migration assay results after 24-h and 48-h transient siRNA transfection in GES-1
and BGC823 cell lines. P16 qRT-PCR results are also presented. c Migration capacity of A549 cells (lacking P16 alleles) following stable transfection of P16-dnmt. After treatment with 0.25 ?g/mL doxycycline for 1 week, these cells (4.0?×?10
⁴
) were seeded into each well and incubated for 28 h. Western blot analysis of P16-Dnmt
and P16 expression is also presented. Migration assays were independently repeated
in triplicate

P16-specific DNA methylation promotes RB phosphorylation and upregulates NF?B subunit
P65 expression

To confirm that P16 DNA methylation affects its downstream signal pathway, P16-CDK4/6-RB, the phosphorylation
level of RB protein was analyzed using Western blot analysis. As expected, increased
levels of phosphorylated RB were detected in the P16-dnmt transfected BGC823 and GES-1 cells treated with doxycycline when compared to those
without doxycycline induction and those transfected with the control vector. Total
RB protein levels were not changed (Fig. 8a and b). Furthermore, the expression level of nuclear factor NF?B subunit P65 was also increased
in the P16-dnmt transfected cells.

Fig. 8. Western blot analysis of P16, RB, phosphorylated RB, and NF?B subunit P65 expression
in the P16-dnmt stably transfected cells. BGC823 and GES-1 cell lines were tested 14 and 8 days following
doxycycline treatment, respectively. a WB image; b Relative density of tested protein bands normalized against GAPDH