A novel enhancer regulates MGMT expression and promotes temozolomide resistance in glioblastoma

Abstract

Temozolomide (TMZ) was used for the treatment of glioblastoma (GBM) for over a decade, but its treatment benefits are limited by acquired resistance, a process that remains incompletely understood. Here we report that an enhancer, located between the promoters of marker of proliferation Ki67 (MKI67) and O6-methylguanine-DNA-methyltransferase (MGMT) genes, is activated in TMZ-resistant patient-derived xenograft (PDX) lines and recurrent tumor samples. Activation of the enhancer correlates with increased MGMT expression, a major known mechanism for TMZ resistance. We show that forced activation of the enhancer in cell lines with low MGMT expression results in elevated MGMT expression. Deletion of this enhancer in cell lines with high MGMT expression leads to a dramatic reduction of MGMT and a lesser extent of Ki67 expression, increased TMZ sensitivity, and impaired proliferation. Together, these studies uncover a mechanism that regulates MGMT expression, confers TMZ resistance, and potentially regulates tumor proliferation.

Fig. 1

Enhancer marks are altered in a TMZ-resistant PDX line. a Schematic diagram for the development of 5199 and 3080 xenograft lines from parental GBM12 cells. Mice with GBM12 flank tumors were treated with placebo or three cycles of TMZ (50 mg/kg/d for 5 days every 28 days). Tumors were dissected after reaching >1500 mm³. The xenograft subline established from placebo-treated tumor was named 5199 and that from TMZ-treated tumor was named 3080. b Analysis of MGMT, H3K27ac and H3K4me1 levels in protein lysates of 5199 and 3080 xenograft tissues by western blotting. α-tubulin and histone H3 was used as a loading control for cytosolic proteins and nuclear proteins, respectively. c Analysis of MGMT promoter methylation in 5199 vs. 3080 xenograft lines by MS-PCR. Universal methylated DNA was used as positive control and normal human brain DNA was used as negative control. d–g Aggregate plots showing the average ChIP-seq reads distribution for the various histone marks: H3K4me1 (d) and H3K27ac (e) ChIP-seq reads surrounding the published enhancer regions; H3K4me3 (f) ChIP-seq reads surrounding transcription start site (TSS); H3K36me3 ChIP-seq reads across gene bodies (g). TTS: transcription termination site. h Heat maps showing cluster analysis, based on H3K27ac and H3K4me1 alterations, and expression change for genes closest to each genomic locus. The ratio of ChIP-seq reads density (3080 reads/5199 reads) for H3K4me1 and H3K27ac was calculated and subsequently analyzed by k-means cluster analysis. Changes in gene expression were calculated as RNA-seq reads ratio (3080 reads/5199 reads). Color scale represents decreased (green) and increased (magenta) signal intensity (intensity = log₂$\frac{3080\mathrm{reads}}{5199\mathrm{reads}}$₎. i Bar graphs showing ingenuity pathway analysis for Group-1 genes. Pathways with a p value <0.05 are presented

Fig. 2

Delineation of a putative enhancer-associated with MGMT gene in 3080 subline. a IGV snapshots showing H3K4me1, H3K27ac, H3K4me3, and H3K36me3 ChIP-seq reads density at the putative enhancer region, MGMT promoter, and gene body in 5199 (TMZ-sensitive) and 3080 (TMZ-resistant) xenograft lines. b H3K4me1 (bottom left) and H3K27ac (bottom right) occupancy at the putative K-M enhancer in 5199 (parental) and 3080 (TMZ-resistant) xenograft tumors were analyzed by ChIP-qPCR. Three different sets of primers (PE1-3) were designed to analyze against putative enhancer region. A set of primers amplifying 5 kb away from the putative enhancer (PE + 5 kb) was used as negative control. Top: magnified view of H3K27ac ChIP-read peaks and the location of each primer set. c Effect of each fragment on transcription of luciferase reporter gene in 5199 and 3080 cells. Fragments R1–R10 covering putative enhancer region (top) were cloned upstream of luciferase promoter (middle). The ChIP-seq peak covered by R7 region is labeled in blue. Each construct was transfected into 5199 or 3080 cells along with a plasmid expressing Renilla luciferase. Firefly and Renilla luciferase activities were measured 36 h after transfection. Luciferase activity was normalized to the Renilla luciferase activity (internal control) first and subsequently normalized to an empty vector control. d Enhancer–promoter interaction analyzed by chromatin conformation capture (3C) assay. Relative interaction frequency of each restriction fragment (F1–F6) was calculated as described in the experimental procedures and was plotted on genomic location of the 3′ cutting site of each fragment (X axis). Values reported were derived from three biological repeats (*p < 0.05, **p < 0.01, Student’s t test). The genomic region covered by H3K27ac peak is labeled in orange, location of R7 fragment was colored in blue to highlight the overlap with the F4 fragment

Fig. 3

Enhancer activity and MGMT expression in three pairs of GBM patient samples. a, b Relative occupancy of H3K4me1 (a) and H3K27ac (b) in the K-M enhancer region in primary and matched recurrent tumors from three GBM patients was analyzed by ChIP-qPCR using primers described in Fig. 2. c MGMT expression level in each tumor was analyzed by immunofluorescence. The MGMT signal (area × average intensity) from 100 individual nuclei was measured and one-way ANOVA was used for the calculation of significance (****p < 0.0001, ns not significant, n = 100) d Analysis of MGMT expression in proliferating tumor cells. Multicolor immunofluorescence was performed using antibodies against CD45, Ki67, and MGMT. Proliferating tumor cells (Ki67 staining without CD45 staining) were identified and the percentage of MGMT-positive proliferating cells was calculated. Two-tail paired Student's t test was used for calculation of significance (*p < 0.05). e, f Representative images showing co-localization of MGMT and Ki67 in tissue sections from primary tumor (e) and recurrent tumor (f) obtained from patient #1

Fig. 4

Targeting p300 to the K-M enhancer increases MGMT gene transcription. a Schematic diagram showing procedure of targeting p300 catalytic domain (p300 core) by CRISPR/dCas9 system. R7 region identified by previous experiment was labeled as blue to indicate the targeting loci. Experimental design involved co-transfection of dCas9^{p300 Core} (or control dCas9) vectors along with five-guide RNAs (or a control empty vector) followed by brief selection of transfected cells under 5 µg/ml puromycin and subsequently analyzed by ChIP-qPCR. b, c Analysis of the occupancy of Flag-tagged dCas9/dCas9^{p300 Core} protein (b) or histone H3K27ac (c) at the K-M enhancer region by ChIP-qPCR in HEK293T cells 72 h after transfection. An empty vector (lentiGuide puro vector) or a mixture of guide RNA constructs were transfected along with a dCas9 or dCas9^{p300 core} expression vector. d MGMT expression in HEK293T cells was analyzed 72 h after co-transfection with either an empty vector or guide RNA construct along with a dCas9 or dCas9^{p300 core} expression vector. MGMT transcript level was first normalized to actin and subsequently calculated as fold change relative to double-negative (Vector dCas9) control (*p < 0.05, n = 3 independent experiments, Student’s t test). e, f Analysis of the occupancy of Flag-tagged dCas9/dCas9^{p300 core} protein (e) or histone H3K27ac (f) at the K-M enhancer region by ChIP-qPCR in 5199 cells 72 h after infection. 5199 cells were infected with a mixture of virus containing an empty vector (lentiGuide puro vector) or five-guide RNA constructs and transfected with a dCas9 or dCas9^{p300 Core} construct. g The MGMT transcript level in 5199 cells after targeting p300 core to the enhancer by CRISPR/dCas9. The experiments were performed as described from b-d (*p < 0.05, n = 3 independent experiments, Student’s t test)

Fig. 5

Enhancer deletion reduces MGMT expression and increases sensitivity to TMZ. a An outline of deletion strategy using CRISPR/Cas9 system. Cells were infected with a mixture of two virus containing guide RNAs surrounding K-M enhancer locus. After puromycin selection, single clones were isolated and tested by PCR. b MGMT expression in SKMG3 parental cell, SKMG3 wild-type clone, and three K-M enhancer-deleted clones were analyzed by quantitative RT-PCR. MGMT transcript level was first normalized to actin and subsequently calculated as fold change relative to SKMG3 parental line. c MGMT protein levels in SKMG3 parental cell, SKMG3 wild-type clone, and three K-M enhancer-deleted clones were tested by western blotting. α-Tubulin was used as loading control. d SKMG3 parental line, wild-type clone, and K-M enhancer-deleted clones were treated with indicated concentrations of temozolomide (0–1000 µM/L final concentration). Cell viability was determined using clonogenic assay. e SKMG3 parental line, wild-type clone, and K-M enhancer-deleted clones were treated with 10 µM O₆BG 1 h prior to temozolomide (0–1000 µM/L final concentration). Cell viability was determined using clonogenic assay. f Bar graph showing log₂(TMZ IC₅₀) in SKMG3 clones with or without O₆BG pre-treatment. g MGMT expression in SKMG3 parental cells and three Del1 deleted clones were analyzed by quantitative RT-PCR. MGMT transcript level was first normalized to actin and subsequently calculated as fold change relative to SKMG3 parental line. h SKMG3 parental cells and three Del1 deleted clones were treated with indicated concentrations of temozolomide (0–1000 µM/L final concentration). Cell viability was analyzed using clonogenic assay. i MGMT transcription levels in 3080 parental cells and 3080 Del1 deleted cells were analyzed by quantitative RT-PCR. MGMT transcript level was first normalized to actin and subsequently calculated as fold change relative to 3080 parental line j 3080 parental cells and 3080 Del1 deleted cells were treated with different concentration of temozolomide (0–1000 µM/L final concentration). Cell viability was determined using neurosphere formation assay (***p < 0.001, ****p < 0.0001, n = 3 independent experiments, Student’s t test)

Fig. 6

Deletion of the K-M enhancer affects cell proliferation and Ki67 expression. a Schematic diagram showing the relative genomic location of MKI67 gene, the K-M enhancer, and MGMT gene. b, c The Hi-C analysis from SK-N-MC cells (b) and human fetal brain cortical plate (c) are taken from the 3D genome browser site. The boundaries of the topologically associated domains (TADs) and the location of MGMT, MKI67, and the K-M enhancer is indicated below. d Cell proliferation rate was analyzed by Incucyte. The cell confluency read out for each cell was normalized by the cell confluency acquired at the first time point. e MKI67 transcription levels in SKMG3 wild-type clone and two enhancer-deleted clones were analyzed by quantitative RT-PCR. Ki67 expression level was first normalized against actin and subsequently calculated as fold change relative to wild-type clone. f Growth rate of 3080 parental and 3080 Del1 cells injected flank tumors. Equal numbers of 3080 parental cells and 3080 Del1 cells were injected into the flank of mice (n = 5/group). Tumor volume was used to represent growth rate. g MKI67 transcription levels in 3080 parental cells and 3080 Del1 deleted clone cells were analyzed by quantitative RT-PCR. Ki67 expression level was first normalized against actin and subsequently calculated as fold change relative to the parental line