Proc. Nati. Acad. Sci. USA Vol. 89, pp. 10119-10123, November 1992 Biochemistry Methylation in the preinitiation domain suppresses gene transcription by an indirect mechanism (transient transfections/transcription factor TFHID and TFUA binding/in vitro trasription/transcription reinitiation) ALEX LEVINE*t, GiULIo L. CANTONIt, AND AHARON RAZIN* *Department of Cellular Biochemistry, The Hebrew University, Hadassah Medical School, Jerusalem, Israel 91010; and tLaboratory of General and Comparative Biochemistry, National Institute of Mental Health, Bethesda, MD 20892 Contributed by Giulio L. Cantoni, July 28, 1992 ABSTRACT Although the first observations of the inhib- The present study provides evidence that the inhibition of itory effect of methylation on gene activity were made almost promoter activity is correlated with the density ofmCpG sites a decade ago, the mechanism by which methyl groups affect at the preinitiation domain but not at sequences flanking this transcription is still obscure. Here we use engineered promot- region. The results also suggest that a methylcytosine- ers methylated in vitro in transient tranections to study the binding protein is involved in this inhibition, most probably mechanism by which methylation mediates promoter repres- by binding to the domain involved in formation of the sion. The results clearly show that the location of the methyl preinitiation complex. However, templates that escape inac- groups within the promoter region determines the extent of tivation and succeed in establishing a functional preinitiation promoter repression. The most effective suppression was ob- complex support transcription and repeated reinitiations served when methylation was in the preinitiation domain. The even in their methylated state. results also support a previous suggestion that a mediator protein is involved in the mechanism of promoter inhibition. The suppressor effect of methylation at sequences flanking the MATERIALS AND METHODS TATA box can be partially overcome in the presence of the Preparation ofConstructs, Ollgonucleotides, and PCR Prod- simian virus 40 enhancer. In addition, results obtained by ucts and in Vitro Methylation. pmMTIhGH used for the transient thymidine labeling ofLtk- cells that were transfected studies on the effect ofmethylation ofthe mouse metallothio- with a methylated thymidine kinase gene from herpes simplex nein I (mMTI) promoter (see map in Fig. 1) has been virus, at the level of approximately one template per cell, described previously (14). pSV2hGH (see map in Fig. 2) was further support the conclusion that methylation affects pri- constructed by excision of the simian virus 40 (SV40) early marily banscription preinitiation. promoter from pSV2CAT (HindIII/Pvu II fragment), filling in the ends, and ligation into the HincII site of the polylinker A correlation between gene-specific methylation patterns of the pOGH (Nichols Institute, San Juan Capistrano, CA). and expression is well established: many genes have been The orientation of the insert was determined by digestion shown to be inactive when methylated and expressed when with Bgi I. The SV40 enhancer was deleted by digesting the undermethylated. Several studies show that methylation at parental plasmid pSV2hGH with HindIII/Nsi I, which re- the promoter region is sufficient to repress gene activity (1). moves most of the 72-base-pair (bp) repeats, resulting in the Recent progress in the identification and characterization of loss of enhancer activity (15). Duplex oligonucleotides with transcription factors and in the clarification ofthe mechanism Nco I overhangs, synthesized by the Applied Biosystems of transcription initiation in eukaryotes has made it possible automatic oligonucleotide synthesizer, were annealed and to investigate the mechanism whereby methylation at the ligated into the Nco I site or end filled and inserted into Bgl promoter region results in repression of gene activity (1, 2). I, Xba I, or HindIII sites by blunt-end ligation. Methylation Three different mechanisms have been suggested: (i) Meth- with methylases M.Sss I, M.Hpa II, M.Hha I, and M.FnuDII ylation of specific DNA sites can directly inhibit the binding was performed as described previously (8). Completion ofthe of transcription factors, such as the major late transcription in vitro methylation reaction was verified by digestion of the factor (MLTF) (3) or the cAMP-responsive element binding methylated product with the corresponding restriction en- factor (CREB) (4). However, this mechanism does not apply zyme. to all transcription factors; the transcription factor Spl, for Transfections and Assay of Promoter Activity. Transient instance, binds and facilitates transcription even when Spl transfection experiments were performed as previously de- sites are fully methylated (5). (ii) The binding oftranscription scribed (8) with the following modifications: Ltk- cells, factors may be prevented indirectly by proteins capable of which are mouse L cells lacking a thymidine kinase (tk) gene, binding to methylated DNA. One such protein might be the were plated into six-well culture plates (3.5 x 105 cells per recently identified meCPl, which has been shown to bind plate). After 24 hr of incubation in Dulbecco's modified more efficiently to densely methylated DNA sequences in Eagle's medium (DMEM) + 10% fetal bovine serum the vitro (6) and has been postulated to do so in vivo (7, 8). The medium was replaced by a transfection mixture consisting of efficiency of meCP1 binding and the extent of promoter 300 ng of plasmid DNA in 250 ,ul of Opri-MEM (GIBCO) activity suppression is a function of the density of mCpG mixed with 12.5 Al of N-[1-(2,3-dioleoyloxy)propylJ-NN,N- (mC, methylcytosine) sites in the promoter region (6, 9). (iii) trimethylammonium methyl sulfate (DOTAP) transfection DNA methylation has also been correlated with the organi- reagent (Boehringer-Mannheim, Germany) in 250 A.l ofOPTi- zation of the chromatin structure (10-12), perhaps through MEM. After 6 hr with this mix the volume in the wells was the participation of mCpG-binding proteins in this process brought to 2.5 ml with DMEM + 10%6 fetal bovine serum. The (13). promoter activity was assayed after 48 hr by measuring the The publication costs of this article were defrayed in part by page charge Abbreviations: MLTF, major late transcription factor; mMTI, payment. This article must therefore be hereby marked "advertisement" mouse metallothionein I; SV40, simian virus 40; tk, thymidine in accordance with 18 U.S.C. §1734 solely to indicate this fact. kinase; hGH, human growth hormone. 10119 Downloaded by guest on September 24, 2021 10120 Biochemistry: Levine et al. Proc. Natl. Acad. Sci. USA 89 (1992) levels of human growth hormone (hGH) in the culture me- striking synergistic effect is consistent with the hypothesis dium as described previously (8). that the interaction between methylated sites may result in a In Situ Autoradiography. Ltk- cells were transfected with cooperative binding of suppressor protein molecules to the pHSVtk (plasmid containing the herpes simplex virus tk promoter region. However, other interpretations of the re- gene) that had been methylated in vitro by M.Hpa II, M.Hha sults are possible. It should be noted, for instance, that mMTI I, and M.FnuDII at a concentration of 500 or 100 ng per and other native promoters may contain more than one 60-mm plate. After 16 hr of transfection in Opri-MEM, the recognition site for various transcription factors (17). Meth- medium was replaced by 1 ml of DMEM + 10% fetal bovine ylation of these recognition sites could interfere directly with serum containing 10 ,Ci (370 kBq) of [3H]thymidine (48 the binding of the factors. For example, the relatively strong Ci/mmol) (Amersham) for another 24 hr. Cells were washed effect obtained by methylation of the FnuDII site might be and plated on microscope slides and incubated for 6 hr. As a explained by the location of these sites within the MLTF control of expression of single-copy tk gene we used CHO element. In addition, the results presented in Fig. 1 do not (Chinese hamster ovary) cells labeled with [3H]thymidine allow us to determine whether the extent of inhibition is under the same conditions. By lowering the DNA concen- dependent on the density ofthe methyl groups, their position, trations that were used in the transfection experiments we or a combination of the two. Therefore, this type of exper- have arrived at a DNA concentration that resulted in a iment cannot distinguish between the three mechanisms smaller number of positive cells with a relative homogenous outlined above-namely, a direct effect of methylation on the grain density comparable to that obtained with CHO cells. binding of transcription factors, an indirect effect mediated Cells were fixed with methanol/acetone (1:1, vol/vol) for 2 by a repressor which binds to the methylated sites, or an min at room temperature, washed with phosphate-buffered overall effect on chromatin structure. Previous reports in the saline, and dehydrated by incubating the slides in gradually literature with regard to these questions were also inconclu- increasing concentrations ofethanol (60% to 100%1o in steps of sive. A study with the late SV40 promoter revealed that a 10%). After drying, the slides were submerged in 1:1 dilution single methylated site is sufficient to inhibit promoter activity of NTB (Kodak) in water and exposed for 12 days. Devel- (18). On the other hand, when the y-globin promoter was oping was with D-19 developer (Kodak) for 2 min at 80C. studied it was suggested that a minimum stretch of unmeth- Slides were stained with Giemsa stain and visualized and ylated DNA in the promoter region is required for expression photographed in a Zeiss Axioplan microscope equipped with of the gene, suggesting a global effect on chromatin structure a 35-mm camera. (19). To distinguish between the various interpretations it would be advantageous to use promoters in which the position and RESULTS number of methyl groups along the promoter region can be We have previously demonstrated that in vitro methylation of controlled with precision.