HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 275, Issue 51, 40273-40281, December 22, 2000

This Article Abstract Full Text (PDF) All Versions of this Article: 275/51/40273    most recent M007021200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Viswanath, R. L. Articles by MacDonald, R. J. Search for Related Content PubMed PubMed Citation Articles by Viswanath, R. L. Articles by MacDonald, R. J. Social Bookmarking                      What's this?

A Binary Mechanism for the Selective Action of a Pancreatic -Cell Transcriptional Silencer^*

Raghu L. Viswanath, Scott D. Rose^§, Galvin H. Swift, and Raymond J. MacDonald^¶

From the Department of Molecular Biology, the University of Texas Southwestern Medical Center, Dallas, Texas 75235-9148 and ^§ Integrated DNA Technologies, Coralville, Iowa 52241

Received for publication, August 3, 2000

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The pancreatic elastase I gene (ELA1) is selectively transcribed to high levels in pancreatic acinar cells. Pancreatic specificity is imparted by a 100-base pair enhancer that activates transcription in -cells of the islets of Langerhans as well as in acinar cells. Adjacent to the enhancer is a silencer that renders transcription specific to acinar cells by selectively suppressing the inherent -cell activity of the enhancer. We show that the selective repression of -cell transcription is due neither to a -cell specific activity of the silencer nor to selective interference with -cell-specific transcriptional activators acting on the enhancer. Rather, the silencer is effective in both pancreatic endocrine and acinar cell types against all low and moderate strength enhancers and promoters tested. The silencer appears to act in a binary manner by reducing the probability that a promoter will be active without affecting the rate of transcription from active promoters. We propose that the ELA1 silencer is a weak off switch capable of inactivating enhancer/promoter combinations whose strength is below a threshold level but ineffective against stronger enhancer/promoters. The apparent cell-specific effects on the ELA1 enhancer appear due to the ability of the silencer to inactivate the weak -cell activity of the enhancer but not the stronger acinar cell activity.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The complementary action of positive and negative transcriptional control is a common regulatory strategy for cell-specific genes (1). In many instances transcriptional enhancers direct expression to several cell types in an organ, generally due to the action of an organ-specific, but not cell-specific, transcriptional activator or combination of activators. Cell-specific expression is then resolved by the action of a repressor/silencer region that prevents transcription in the inappropriate cell type(s). The basis for cell type restriction by transcriptional silencers is not well understood, although recent evidence indicates repressive chromatin structure may be initiated by targeted recruitment of histone deacetylases or other chromatin-modifying enzymes (for reviews see Refs. 2 and 3). Switching between open (accessible) and closed (inaccessible) forms of chromatin (4, 5) is consistent with all-or-none mechanisms of transcriptional control by enhancers (6) and presumably silencers as well.

Selective transcription of pancreatic genes in exocrine versus endocrine cells requires both positive and negative forms of transcriptional control. For example, the restricted transcription of insulin in pancreatic -cells (7-9) and elastase in acinar cells (10) is mediated by the complementary action of enhancers and silencers. -Cells and acinar cells are constituents of the endocrine and exocrine compartments, respectively, of the compound pancreatic gland. Most of the pancreas is exocrine, organized into acinar structures that secrete a set of digestive hydrolases such as amylase, ribonuclease, and elastase into the intestine. Only approximately 2% of the pancreas is endocrine tissue, organized into islets of Langerhans comprising four cell types, each producing one principal polypeptide hormone. The majority of islet cells are -cells, which produce insulin. During embryonic pancreatic development mature acinar and islet cells arise from transitional epithelial cells that express both acinar- (e.g. amylase) and islet (e.g. insulin)-specific markers (11). Differentiation of these transitional cells leads to silencing of acinar genes in islet cells and vice versa. The elastase I gene (ELA1) is one of the complement of pancreatic genes encoding digestive enzymes that are expressed selectively in the acinar cells. In this report we describe a transcriptional silencer that prevents expression of rat ELA1 in islet cells while permitting expression in acinar cells.

Activation and repression of ELA1 gene transcription is mediated through 5'-proximal gene flanking sequences (Fig. 1). A 100-bp,¹ three element transcriptional enhancer (195 to 96 relative to the transcriptional start) contains regulatory information that directs pancreatic expression to both acinar and -cells of transgenic mice (10, 12-14). The endogenous ELA1 gene as well as transgenes containing as little as 501 bp of 5'-flanking sequences are not expressed in pancreatic islets, however (15). The appropriate acinar-specific transcription is imposed by a negative regulatory region located immediately upstream of the enhancer (between 501 and 202) (10).

The acinar and -cell specific activities of the ELA1 enhancer have been assigned to two separate transcriptional elements (Fig. 1), based on their activity in transgenic animals and transfected pancreatic cell lines (10, 14, 16). The 25-bp A element is the sole positively acting transcriptional element for acinar transcription. The latent -cell specific activity of the three-element enhancer is due to a 12-bp transcriptional element, the B element. The B element is also active in acinar cells, but in these cells it has a different function, which is to augment the activity of the acinar-specific A element (13). A third element (C) is inactive on its own, but in combination with either the A or the B element augments their activity nearly 10-fold without contributing to or affecting cell specificity (13). In -cells the activity of the enhancer is mediated by the combination of the B and C elements, whereas in acinar cells all three elements are active. The silencer region (501 to 202) suppresses the BC element activity in -cells of mice without detectably altering the activity of the three-element (ABC) enhancer in acinar cells (10).

We set out to determine the regulatory strategy that the silencer uses to suppress selectively the action of the ELA1 enhancer in pancreatic -cells. For example, the silencer might selectively antagonize the action of the enhancer B element, which is responsible for the -cell activity. Alternatively, the silencer may be active in -cells but not in acinar cells; for example, the factors that mediate its activity may be present selectively in -cells. We report that the action of the silencer is specific neither for the B element nor to -cells. Instead we present evidence that the apparent selective action of the enhancer is based on its ability to suppress effectively the relatively weak two-element (B + C) enhancer active in -cells but not the complete enhancer with the three elements (A + B + C) working in concert in acinar cells. We show that the silencer acts in a binary manner to block effectively transcription from a promoter with a weak enhancer.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Construction of the Silencer Test Genes-- Fusion genes containing fragments of transcriptional control regions and the human growth hormone (hGH) reporter gene in pUC119 (construct pUC119.hGH) were constructed by standard recombinant DNA techniques (17), polymerase chain reaction (18), and site-specific mutagenesis (19). Details of the constructions are available upon request. The construction of the fusion genes, 5B.EIp.hGH, 92/+8EIp.hGH, 205EI.hGH, 500EI.hGH, and 4.5EI.hGH (all in pUC119), has been described (10, 12, 20). The limits of each gene fragment are indicated in the figures by the numbers indicating the nucleotide distances from the start site of transcription of the gene from which each fragment was derived.

The reporter plasmids to map the functional domain of the repressor region by deletion mapping were constructed by introducing EcoRV sites through site-specific mutagenesis (18) at 436, 368, and 319. After restriction endonuclease digestion the desired fragments were isolated, and HindIII and SalI linkers were added to the distal and proximal ends, respectively, and inserted into the HindIII/SalI sites immediately upstream of the B element multimer of the 5B.EIp.hGH reporter plasmid.

Cell Transfections and Reporter Gene Assays-- Ten µg of the hGH fusion gene constructs in 120 µg of total DNA were transfected into 1 × 10⁷ RIN1046-38 cells by electroporation (21) at 280 V and 950-microfarad capacitance in a Bio-Rad Gene Pulsar. The cells were plated on 10-cm dishes immediately following the electroporation, washed twice with phosphate-buffered saline 20 h after transfection, and cultured in DMEM containing 10% fetal calf serum. 0.5 × 10⁶ 266-6 cells and Rat2 fibroblasts were transfected with 10 µg of test plasmid by the calcium phosphate procedure (22), washed twice in 1× phosphate-buffered saline containing 3 mM EGTA 20 h following the transfection, and cultured in DMEM. Expression of the transfected fusion genes was monitored at 48 h following the replacement of media with fresh DMEM by measuring hGH accumulation in the medium (23) using a radioimmunoassay (Nichols Institute, San Juan, Capistrano, CA). All constructs were transfected and assayed in duplicate and represent the average of at least three independent transfections unless stated otherwise in the figure legends. hGH expression levels were corrected for the efficiency of transfection by measuring the activity of a cotransfected Rous sarcoma virus (RSV) chloramphenicol acetyltransferase or cytomegalovirus (CMV) lacZ fusion gene constructs and quantification of enzyme activity.

For distinguishing the binary and graded mechanisms of the silencer, 5 × 10⁶ cells were electroporated as described above, resuspended in 4 ml of medium, and divided into four 35-mm culture plate wells. The cells were cultured as above. 48 h after changing the medium (approximately 72 h after electroporation), the medium from each well was collected for the hGH assay, and the cells were either fixed with 4% paraformaldehyde in phosphate-buffered saline or (in one of the plates) lysed to assay -galactosidase activity from the pCMV-lacZ internal reporter plasmid (CLONTECH). Immunohistochemical analysis was performed under standard conditions on the cells fixed in the culture dishes. The anti-hGH primary antibody (Dako) was diluted 1:500. The biotinylated secondary antibody, streptavidin-linked peroxidase, and AEC chromogen were provided by a Zymed Laboratories Inc. Histostain Plus Kit, used according the manufacturer's instructions. The use of subsaturating amounts of DNA helped ensure the presence of single active templates in hGH-expressing cells, as required for the test (see "Results"). The number of hGH-staining cells increased in direct proportion to DNA dose, whereas the ratio of hGH produced in the medium per hGH-staining cell remained constant up to at least 10 µg of plasmid DNA (data not shown); therefore, 2 and 6 µg of plasmid DNA were used in the test of the binary action of the silencer.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Initially we tested two likely mechanisms for the -cell-specific suppression of the ELA1 enhancer by the silencer. For mechanism 1, the silencer might selectively antagonize the action of the pancreatic transcription factor PDX1, which mediates the action of the B element in islet -cells (16). For mechanism 2, the silencer may be inactive in acinar cells; for example, the factors that bind to and mediate the activity of the silencer may be absent from acinar cells.

To test these mechanisms, it was first necessary to verify that the silencer, active in -cells of transgenic mice, was also active in transfected pancreatic -cell lines. RIN1046-38 (hereafter RIN38) is a well differentiated, insulin-synthesizing -cell line (24). At low passages (as in these experiments) RIN38 cells retain many differentiated characteristics of pancreatic -cells, including moderate levels of insulin mRNA and glucose-regulated secretion of insulin (25). The -cell activity of the B element can be mimicked in transgenic animals and cultured cells by linking a pentamer of the B element to the minimal ELA1 promoter spanning nucleotides 92 to +8 (Fig. 1 and Ref. 18). In animals the minimal ELA1 promoter is inactive, and the addition of the B pentamer activates transcription selectively in -cells (10). In transfected RIN38 cells the low activity of the minimal promoter is increased 20-fold by the B element pentamer (Fig. 2, constructs A and B). The B pentamer is also active in a second independently derived and well differentiated -cell line, TC3 (Ref. 16 and data not shown), but is inactive in non--cell lines, such as pancreatic acinar effectively, NIH3T3, and Rat2 fibroblast lines (16).

View larger version (6K): [in this window] [in a new window]   Fig. 1.   Positive and negative transcriptional regulatory elements of the gene-proximal 5' region of ELA1. The principal transcriptional control elements of the ELA1 gene are confined to the 500-bp region immediately upstream of the transcriptional start site (see text). The reporter gene construct containing five tandem repeats of the B-element linked to the 92 region of the ELA1 promoter is shown.

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Repression of B element activity in transfected RIN1046-38 -cells. The activity of each test gene and the effects of the silencer were measured by transient transfection in RIN38 cells as described under "Experimental Procedures." The gene fragments in each fusion construct are discussed in the text and are identified as follows: EIp, the ELA1 minimal promoter; B, the B element of the ELA1 enhancer; Sil, the ELA1 silencer region in the normal orientation (501/202); and liS, the silencer region in the reverse orientation (202/501); hGH, the 2.2-kb human growth hormone reporter gene. The nucleotide numberings are relative to the start site of transcription of the gene from which each fragment was derived. Expression was measured by assaying hGH production as described under "Experimental Procedures." The level of activity of each construct is expressed as a percent of the activity of 5B.EIp.hGH (construct B) after correction for relative transfection efficiencies by monitoring the activity of a cotransfected RSV.mCAT plasmid.

The 300-bp ELA1 silencer region (501 to 202) reduced the activity of the B pentamer in transfected RIN38 cells (Fig. 2C), consistent with its ability to suppress the action of the ELA1 enhancer in -cells of transgenic mice. The silencer decreased the activity of the B pentamer 12-15-fold, so that the residual activity was approximately that of the minimal promoter alone. The silencer also was effective in reverse orientation (Fig. 2D). However, the silencer region was not effective when moved to the other side of the hGH reporter gene; this position is 2.2 kb downstream and 2.8 kb upstream of the promoter in the circular plasmid (Fig. 2, E and F). Thus, the action of the silencer is independent of its orientation but affected by distance from the promoter.

The ELA1 Silencer Is Not Specific for the ELA1 Minimal Promoter or the B Element-- To test whether the specificity of the silencer may be due to selective interference with the islet-specific activity of the B element (mechanism 1), we measured the effects of the silencer on the activity of a variety of other promoters and enhancers, both cellular and viral, containing a wide variety of transcriptional elements. First, the activity of the silencer is not restricted to constructs containing the cognate ELA1 minimal promoter, because it also repressed the activity of the B pentamer coupled to the hsp70 minimal promoter (Fig. 3, compare constructs C and D). Moreover, the silencer inhibited the activity of the 49-bp H-2 K^b cellular enhancer driving either the hsp70 or the ELA1 minimal promoter (85 and 95% reduction, respectively; Fig. 3, compare E with F and H with I). The H-2 K^b enhancer has an exceedingly broad cell-type expression pattern (26) and therefore is likely driven by ubiquitous factors. The silencer region also reduced the activity of the HSV tk promoter by 95% (Fig. 3, J and K) and the SV40 enhancer/promoter by 80% (Fig. 3, L and M). The viral HSV tk promoter contains binding sites for common factors and is active in a wide variety of cell lines (27). The SV40 early enhancer/promoter is also active in a variety of cell types, although individual elements within the enhancer display a unique pattern of cell-specific activity (28, 29). These results demonstrated that the silencer region can interfere with the activity of a wide variety of heterologous promoters and enhancers. Consequently, its ability to repress selectively the ELA1 enhancer in transgenic mouse islets appears not due to an effect specific to the B element or to PDX1, which binds and mediates the activity of the B element (mechanism 1).

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Repression of heterologous promoters and enhancers in RIN38 -cells. The silencer is not specific to the action of the B element of the ELA1 enhancer. The gene fragments in each fusion construct are as follows: hsp, the minimal heat shock promoter of hsp70; H-2Kbe, the enhancer of the major histocompatibility complex class I H-2Kb gene; HSV tkp, the herpes simplex virus thymidine kinase promoter; and SV40p/e, the promoter and enhancer of early gene transcription of simian virus 40. Levels of activity are expressed as a percentage of the activity of 5B.EIp.hGH (construct A). All other designations are as described in the legend for Fig. 2.

The Silencer Is Also Active in Pancreatic Acinar Cell Lines-- To test whether the silencer is active in - but not acinar cells (mechanism 2), we took advantage of the observation that the heterologous enhancers and promoters it can suppress in the RIN38 -cell line (Fig. 3) are also active in a wide variety of other cell types. 266-6 cells are derived from a mouse pancreatic acinar adenocarcinoma induced by an SV40 T-antigen transgene (15); they retain properties of differentiated acinar cells, including the expression of cell-specific genes for the digestive enzymes elastase I, trypsinogen I, and amylase (30). Strikingly, the silencer region inhibited the activity of the cellular H-2 K^b enhancer driving either the hsp70 or the ELA1 minimal promoter in transiently transfected 266-6 acinar cells (Fig. 4, compare C with D and F with G). The silencer also suppressed the activity of the viral HSV tk promoter in this acinar cell line (Fig. 4, compare H and I). When the activity of the silencer with these identical enhancer and promoter constructs are compared for the acinar and -cell lines, the silencer was at least as effective in the acinar cells. These results suggest that the differential action of the silencer in islet versus acinar cells of mice is due neither to cell-specific activity (mechanism 2) nor to specificity toward the B element (mechanism 1).

View larger version (23K): [in this window] [in a new window]   Fig. 4.   Repression of heterologous promoters and enhancers in 266-6 acinar cells. These data show that the silencer is also active in an acinar cell line derived from transformed pancreatic acinar cells. EIe represents the ELA1 gene enhancer. The designations for all other fusion genes are the same as described in the legends for Figs. 1 and 2. Levels of activity are expressed as a percentage of the activity of the ELA1 enhancer/promoter construct (A).

Far Upstream Elements Relieve Repression in Acinar but Not Endocrine Cells-- Although the silencer had no apparent effect on the ELA1 gene enhancer in acinar cells of transgenic mice (10), its ability to repress heterologous promoters in the 266-6 acinar cell line (Fig. 4) led us to test whether it was effective against the ELA1 enhancer in these cultured cells. The silencer did indeed repress the activity of the ELA1 enhancer about 90% (Fig. 5, compare B and C). The residual activity in the presence of the silencer was still at least 10-fold greater than that of the ELA1 minimal promoter without the enhancer (Fig. 5; compare A and C). Therefore, the silencer appears unable to suppress the activity of the enhancer in the acinar tumor cells as effectively as it does the B element multimer in insulinoma cells. This significant residual acinar activity may account in part for the continued expression of the identical silencer construct (as C of Fig. 5) in the pancreatic acinar cells of transgenic mice (10). Due to the difficulty in comparing expression levels of different constructs in transgenic animals (e.g. Ref. 30), quantitative effects of the silencer in acinar cells in vivo may not be evident.

View larger version (24K): [in this window] [in a new window]   Fig. 5.   A distant upstream region relieves repression of the ELA1 enhancer in acinar cells but not in -cells. These transfection data show the activity of the silencer in ELA1 gene constructs in 266-6 acinar cells (top panel) and RIN38 -cells (bottom panel). The dotted lines in the figure represent a deletion of the silencer region from 501 to 202 and a fusion of the boundary at 502 to the boundary of either the ELA1 enhancer or the B element multimer. Levels of activity are expressed as a percent of the activity of the ELA1 enhancer/promoter (construct A) in 266-6 cells and as a percentage of the activity of 5B.EIp.hGH (construct G) in RIN38 cells. All the designations for the gene fragments are described in the legends for previous figures.

To determine whether additional regulatory elements contribute to the control of the endogenous ELA1 gene, we examined the regulatory properties of the region upstream of the silencer. When an additional 4 kb of ELA1 5'-flanking DNA was included with the silencer, enhancer and promoter, the repression by the silencer in 266-6 acinar cells was lost (Fig. 5, compare C and D). However, in the absence of the silencer (Fig. 5E) the upstream region did not increase the expression from the enhancer. Therefore, the upstream region may interfere directly with the silencer rather than indirectly by increasing the strength of the enhancer and thereby overcoming the influence of the silencer.

In contrast to the effects in transfected 266-6 acinar cells, the upstream region could not overcome the action of the silencer in RIN38 -cells (Fig. 5, compare H and I). Moreover, this upstream region suppressed the activity of the B element in RIN38 cells in the absence of the 501 to 202 silencer region (Fig. 5J), suggesting the presence of additional silencing activity in the upstream region that is not manifested in 266-6 acinar cells (Fig. 5D). In summary, the region 4 kb upstream of the silencer contains two classes of regulatory elements that help confer cell type-specific expression, additional negative control elements that suppress -cell expression and others that interfere with the action of the silencer in acinar cells. The overall effect of the upstream region is to reinforce expression in acinar cells and to suppress it in -cells.

The Silencer Is Ineffective with Strong Enhancer/Promoter Regions-- An apparent discrepancy remains for the selective action of the 300-bp silencer in animals compared with cultured cells. In vitro the silencer is neither specific for the ELA1 B element, because it acts on heterologous promoters and enhancers (Fig. 3), nor for -cells, because it also acts in acinar cells (Figs. 4 and 5). Although other regulatory sequences in the endogenous locus may complement the action of the silencer (Fig. 5), in transgenic animals the 300-bp silencer effectively silenced the ELA1 enhancer in islet cells while acinar activity was maintained (10). There are at least three possible explanations for this discrepancy. First, the RIN38 -cell and 266-6 acinar cell lines may not accurately reproduce the differential activity of the silencer in -cells and acinar cells in situ. Because islet and acinar cells have a common developmental origin, retro-differentiation of the acinar cell line may acquire some endocrine properties, including -cell proteins for the silencer. Second, the silencer may also decrease the activity of the ELA1 enhancer in the acinar cells of animals but not completely; because of the difficulty of accurately quantifying transgenic expression in mice, this effect may have been overlooked.

A third possibility is a previously unsuspected basis for the differential action of a silencer. In transfected RIN38 cells the silencer is able to repress the B element pentamer activity completely to the level of the ELA1 minimal promoter (Fig. 2), whereas in acinar cells it is unable to repress the ELA1 enhancer completely to the level of the same minimal promoter (Fig. 5). The post-repression activity of the 500EIhGH construct in acinar cells is at least 10-fold greater than that of the ELA1 minimal promoter (Fig. 5, construct C). Thus, the silencer is only partially effective on the three-element ELA1 enhancer but is much more effective on the synthetic B element multimer, which is a much less potent enhancer. This raises the possibility that the effectiveness of the silencer may depend on the strength of the enhancer with which it is paired. Thus, the 300-bp silencer may be able to suppress the action of a weak enhancer completely but only partially suppress a stronger enhancer.

To test whether the effectiveness of the silencer depends on the strength of the paired enhancer/promoter, we measured the ability of the silencer to repress the activity of the potent RSV and CMV promoters in both acinar and -cell lines. The silencer had no significant effect on the activity of these extremely strong promoters in either cell type (Fig. 6), consistent with the notion that the effectiveness of the silencer is dependent on the strength of the positively activating control sequences it antagonizes. The ability to affect weak but not strong promoters is also a property of a silencer from the chicken lysozyme gene (31). Because only the B and C elements of the ELA1 enhancer are active in -cells, whereas all three elements are active in acinar cells (13, 32), the enhancer may be more susceptible to the silencer in -cells.

View larger version (28K): [in this window] [in a new window]   Fig. 6.   Lack of repression of the potent RSV and CMV promoters in both acinar and islet cells. RSV and CMV LTR represent the long terminal repeats containing the promoter/enhancer region of the early genes of the Rous sarcoma virus and the cytomegalovirus, respectively. The activity of the various constructs are expressed relative to the activity of the ELA1 promoter (EIp.hGH), which was arbitrarily designated as 1.

Binary Action of the Silencer-- Many enhancers appear to affect transcription by increasing the probability of a promoter being active without increasing the rate of initiation at the promoter (6). This binary model for enhancer action proposes that an enhancer determines whether a promoter is on or off, not the rate of RNA polymerase initiation. This contrasts with the generally accepted notion of a graded mechanism in which the enhancer affects the rate of initiation of a constitutively active promoter in a progressive manner. Measuring the effect of an enhancer in individual cells rather than a population can distinguish these two mechanisms (33-35). If an enhancer acts in a strictly binary manner, addition of an enhancer to a functional, minimal promoter would increase the number of active templates but not the activity of already active ones. In contrast, an enhancer acting progressively would increase the rate of initiation of already active templates without increasing the number of active templates. If transfection conditions are controlled to foster single active templates in individual cells, then the number of expressing cells will be a measure of the number of active templates, and the level of expression in individual cells will be a measure of the rate of transcription from a single template.

For example, only about 0.1% RIN38 cells in a population transfected with an hGH reporter gene driven by the minimal ELA1 promoter expressed hGH (EI.hGH, Fig. 7B). Addition of the synthetic 5B enhancer expanded the number of expressing cells about 20-fold (5B.EIp.hGH, Fig. 7B) without changing the expression level per cell, as judged by the proportional increase in secreted hGH levels as well as the relative staining intensities of individual cells (Fig. 7A). The strength of the 5B enhancer may be compared with that of the RSV enhancer/promoter, which is active in ~10% of the cells under the same conditions.

View larger version (38K): [in this window] [in a new window]   Fig. 7.   The silencer decreases the number of cells with an active template but not the level of transcription from an active template. The transient transfection conditions to help ensure the presence of single active templates in individual transfected cells are described under "Experimental Procedures." A, immunohistochemical detection of hGH in individual RIN38 cells transfected with hGH reporter constructs driven with either the 92 to +8 minimal ELA1 promoter (EIp.hGH), the B-element pentamer attached to the ELA1 promoter (5B.EIp.hGH), or that configuration plus the silencer (Sil.5B.EIp.hGH). B, quantification of the effect of the silencer and the B-pentamer enhancer on the activation of the minimal ELA1 promoter. The results for electroporation of 2 and 6 µg (see under "Experimental Procedures") are shown.

We tested whether a silencer may function in a complementary manner, that is by decreasing the probability that a transfected template would be active rather than decreasing the level of transcription per template. Indeed, coupling the ELA1 silencer to the construct containing the B pentamer (Sil.5B.EIp.hGH) reduced the fraction of expressing cells 10-fold (Fig. 7B). The hGH staining intensity and the secreted hGH/stained cell were approximately equivalent for cells expressing the 5B construct with or without the silencer, indicating that, although there were many fewer cells expressing the construct with the silencer, individual expressing cells had levels of transcription equal to those expressing the construct without the silencer. These results suggest that the silencer acts as a binary switch that reduces the probability that the template will form a stably active transcription complex. The level of transcription (i.e. the rate of RNA polymerase loading) is a property of the 92 to +8 ELA1 promoter, which is common to the two constructs.

To substantiate the requirement that only one or a very few templates are active per expressing cell, we showed that colonies of transfected RIN38 cells predominantly had single hGH-stained cells (Fig. 7A). Because these cultures were immunostained 72 h after transfection, a number of cell divisions had taken place. Consequently, a single active cell in a clonal cluster indicates the presence of one or a few active templates, because many active templates should segregate upon cell division. Occasional colonies had two stained cells, generally juxtaposed, suggesting they were daughters of a recent cell division, and hGH staining in one was likely due to the persistence of hGH protein accumulated before division. The approximately equal staining intensity of each expressing cell is also consistent with the presence of a single active template; if multiple active templates were present in each cell, both the graded and binary models would predict that cells with the silencer construct would produce substantially less hGH than cells with the construct containing the enhancer.

Increasing the transfected amount of each construct 3-fold increased both the number of expressing cells and the total level of hGH production accordingly (Fig. 7B), for all the constructs, with and without the enhancer and silencer. These results affirm the presence of only one or a few active templates per cell and that the hGH-stained cells account for the vast majority of expressing cells. These results indicate that the silencer does not decrease the number of active templates per cell from many to a few, nor does it reduce the load of RNA polymerase on those templates that are active. If either of these occurred, the staining intensity of expressing cells would not be equivalent between constructs with and without the silencer, and the level of hGH production by the population would not be proportional to the number of expressing cells. Rather, the silencer would be expected to generate similar numbers of expressing cells, but with less intense staining in proportion to the lower hGH production. Thus, as for enhancers (6) and the synthetic B-multimer, above, that act in a binary manner to increase the number of active templates, the silencer acts similarly, but to decrease the number of active templates.

The Silencer Inhibits Both Promoters and Enhancers-- To determine whether the silencer antagonizes enhancer or promoter functions, we took advantage of its ability to repress the SV40 enhancer/promoter and its inability to act from a distance. The silencer reduced transcriptional activity 4-fold when placed adjacent to the SV40 enhancer/promoter (Fig. 8, compare B and C). However, it had no effect from a position at the 3' end of the hGH reporter gene 2.2 kb downstream of the promoter (Fig. 8D). The SV40 region can be divided into enhancer and promoter functions (36). The promoter domain alone, which contains the transcriptional start sites and reiterated Sp1-binding sites, conferred a low level of expression (Fig. 8, A and E). The enhancer domain, which is composed of two 72-bp repeats containing multiple transcription factor binding sites (37, 38), increases transcription 10-fold at its normal position adjacent to the promoter and 4-fold when placed 2.2 kb downstream at the 3' end of the hGH gene (Fig. 8, B and F). By using the construct with the SV40 enhancer distant from the promoter, it was possible to distinguish effects of the silencer (which is proximity-dependent) on the enhancer from effects on the promoter. When the silencer was adjacent to the promoter (so that its effect was limited to the promoter), it effectively suppressed expression (Fig. 8G). Similarly, when the silencer was adjacent to the distant enhancer, so that its effect was limited to the enhancer, it also suppressed expression. These results indicate that the silencer works in a manner that is independent of the functional differences between promoters and enhancers.

View larger version (23K): [in this window] [in a new window]   Fig. 8.   Action of the silencer on both promoters and enhancers in transfected RIN38 -cells. SV40p denotes the SV40 minimal promoter and SV40e the SV40 enhancer. In the upper panel the levels of activity are expressed as percent of SV40p/e.hGH activity (construct B) and in the lower as percent of the activity of SV40p.hGH.SV40e (construct F). All other designations are the same as described for previous figures.

The Functional Silencer Spans 81 bp-- To identify functional domains within the 300-bp silencer region, we tested a series of partial deletions for their ability to repress the activity of the B element pentamer in transfected RIN38 cells (Fig. 9). A single small domain of 81-bp near the proximal end of the 300-bp silencer was nearly as effective as the full-length silencer region. The remaining domains had little or no repressing activity.

View larger version (21K): [in this window] [in a new window]   Fig. 9.   Mapping the functional domains of the silencer. A schematic of the ELA1 silencer from 501/202 is shown; the EcoRV sites were introduced by site-specific mutagenesis. The position of a TG repeat is shown. For each deletion construct the open boxes represent the region tested by fusion directly to the B element pentamer. The nucleotide boundaries of the gene fragments are relative to the start site of the ELA1 gene. The shaded box represents the minimal functional silencer domain and the filled arrowheads the human growth hormone reporter gene. The level of activity of each construct is expressed as a percent of activity of the 5B.EIp.hGH construct after correction for relative transfection efficiencies monitored by the activity of cotransfected RSV.mCAT. The nucleotide sequence of the 81-bp silencer core is shown at the bottom. The positions of potential binding sites for GFI-1, SP1-like, and YY1 repressors are indicated together with the selective mutations (vertical arrows) used to inactivate each site.

The 81-bp minimal silencer domain contains potential binding sites for three known transcriptional repressors as follows: two candidate-binding sites for GFI-1, one for SP1-like repressors, and one for YY1 (Fig. 9). Mutation of the prospective GFI-1 motifs, singly or together, of the SP1-like or YY1 motifs, however, did not affect silencing activity (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The term silencer was originally derived to emphasize similarities between a certain class of negative regulatory elements and transcriptional enhancers (39, 40) and to distinguish this class from repression elements that interfere directly with the DNA binding of transcriptional activators or their transactivation function (41, 42). Aside from opposite effects on transcription, silencers and enhancers share many experimental properties; they are effective with noncognate promoters, in either orientation and from a distance.

The ELA1 negative regulatory region (502 to 202) has the properties of a transcriptional silencer. The ELA1 silencer was effective against all four promoters tested, both viral and cellular, and four cellular and viral enhancers. It was ineffective only against the potent CMV and RSV enhancers. It was equally effective in either orientation. Although ineffective at 2 kb, the ability of the 300-bp region to act (i) in either orientation when its 80-bp functional region is located at one end and (ii) with various promoters and enhancers whose functional elements were differently distributed indicates that the distance requirement is not stringent. Thus, the ELA1 silencer is remarkably nonspecific for promoters (and enhancers), orientation, and to some extent distance. In these regards, it shares properties with many other transcriptional silencers, i.e. those associated with the yeast mating type (39), lysozyme (31), glutathione transferase P (43), neural SCG10 (44, 45), and rat growth hormone (46) genes.

The function of the ELA1 silencer in situ is to suppress selectively activation of the acinar-specific ELA1 gene in pancreatic -cells. However, our analysis of the properties of the silencer in transfected cells indicates that its action is neither confined to -cells nor specific to the transcription factor (PDX1) that mediates the -cell specificity of the ELA1 enhancer. These results demonstrated that the silencer is active in acinar as well as -cells; therefore, its action is not restricted by -cell-specific repressor proteins. Furthermore, the silencer is active against promoters and enhancers with a wide variety of transcription factor binding sites; therefore, it is not selective against a particular cell-specific transcriptional activator. Instead, the apparent cell type specificity appears to be derived from the different strengths of the ELA1 enhancer in -cells versus acinar cells due to the action of all three enhancer elements in acinar cells but only two of the elements in -cells. This is consistent with a binary mechanism for silencing, in which the silencer is effective against enhancer/promoter combinations below a threshold level of strength and ineffective against stronger combinations.

Binary Versus Graded Silencer Mechanisms-- Transcriptional regulation is most commonly thought of as a progressive mechanism in which the efficiency of initiation by a promoter is controlled in a graded manner, for example by changing the rate of RNA polymerase loading at a promoter. However, a better understanding of the role of chromatin in transcriptional control suggests that enhancers and silencers may compete in a winner-take-all contest to activate and silence promoters, respectively (3, 6). Because the action of a silencer is generally measured as the averaged expression of a population of transfected cells, it has not been clear whether silencers control transcription in a graded manner (by decreasing the rate of initiation events on the promoter in many or all cells of the population) or in a binary manner (by completely silencing the promoter in a fraction of the cells).

We distinguished these two mechanisms using the single cell assay devised by Weintraub (34) and elaborated by Martin and colleagues (35, 47) to determine whether the ELA1 silencer decreases the level of promoter activity in each transfected cell (graded) or decreases the number of expressing cells (binary). Our results (Fig. 7) showed that the silencer decreased the number of hGH-stained cells and the total production of hGH congruently under conditions in which the majority of the expressing cells has a single active template. The level of expression per cell did not decrease; rather, expressing cells disappeared. Consequently, the silencer appears to act principally in a binary manner by preventing the activation of templates without appreciably affecting the rate of transcription on active templates.

The Basis of Silencer Action-- Two distinctive properties of the ELA1 silencer suggest that it acts by controlling chromatin structure. First, the ability of the silencer to disrupt the activity of either the enhancer or promoter of SV40 indicates that the silencer acts at a level common to both, such as by affecting chromatin structure. For instance, if enhancers and promoters open repressive chromatin and keep it open through chromatin remodeling complexes and histone acetylation (5, 48), then silencers may counteract this action through the opposing activity of other chromatin remodeling complexes and histone deacetylation (3). Second, the binary action of the silencer as a simple on-off switch is consistent with a mechanism that competes with an enhancer to sequester a gene in repressive chromatin (47).

The recruitment of histone deacetylase (HDAC) by the silencer is one possible mechanism involving chromatin. However, the HDAC inhibitors sodium butyrate and trichostatin did not inhibit the ELA1 silencer (data not shown). Not all repressors work through HDAC, however; in particular, repression of the SV40 enhancer by the retinoblastoma protein does not require HDAC activity (49). The inability of the HDAC inhibitors to block the action of the ELA1 silencer (which is also effective against the SV40 enhancer) indicates that the silencer acts independently of the known mammalian HDAC enzymes. Further investigation is necessary to understand the molecular mechanism of this silencer, including possible HDAC-independent mechanisms or the action of a trichostatin-resistant HDAC (such as the yeast Sir2 enzyme (50)).

Summary of the Positive and Negative Regulation of ELA1 Transcription-- The interactions among the ELA1 regulatory regions that lead to acinar-specific transcription are summarized in Fig. 10. All three elements of the enhancer activate the basal ELA1 promoter in acinar cells, whereas only the B and C elements activate the promoter in -cells (13). Transgenes driven by the complete enhancer in the absence of the silencer are expressed in virtually all acinar cells of mice but only in about 10% of the -cells (10). This result is consistent with the notion that the combined strength of the transcription factors interacting with the three enhancer elements in acinar cells is greater than those binding the two elements in -cells. The silencer has the potential to inhibit the action of both the enhancer and promoter in -cells as well as acinar cells (e.g. Figs. 2 and 4). A control region upstream of the silencer interferes with the action of the silencer in acinar cells but not -cells (Fig. 5). The lower effectiveness of the silencer on the three-element enhancer, coupled with the interference of silencer action by an upstream region in acinar but not -cells, accounts for the -cell-specific action of the silencer in the endogenous ELA1 gene.

View larger version (13K): [in this window] [in a new window]   Fig. 10.   Summary of the positive and negative regulatory control of ELA1 gene transcription. See the text for discussion.

This overview of ELA1 regulation emphasizes the competing binary actions of the enhancer and silencer. The three element enhancer activates transgenes in most, if not all, acinar cells of the pancreas (13, 20). For the endogenous ELA1 gene, an additional upstream control region helps ensure that the silencer does not inactivate either allele in acinar cells. Once active, the rate of transcription is set by other elements, e.g. the proximal promoter. Ensuring the activation of both alleles in all acinar cells maximizes production of this hydrolytic digestive enzyme, normally expressed in massive amounts by the pancreatic gland. In contrast, complete inactivation of both ELA1 alleles is necessary to prevent the deleterious expression of even small amounts of this enzyme in pancreatic endocrine cells. The ability of the silencer to shut off the ELA1 enhancer in -cells of transgenic mice is a clear demonstration of its effectiveness in situ (10). Thus, the effect of the competing binary actions of the silencer and weak enhancer in islet cells is to decrease the probability of gene activation to zero.

Through a regulatory strategy of competing on-off switches, the silencer has no effect on expression in acinar cells where the enhancer is dominant, but successfully suppresses expression in endocrine cells where the silencer is dominant. This would not be the effect in competing graded mechanisms, in which the silencer would be expected to reduce the rate of transcription in acinar cells and incompletely block transcription in endocrine cells. Thus, the collusion of opposing binary mechanisms might facilitate maximum activation in the appropriate cell type and complete suppression in the inappropriate one.

	ACKNOWLEDGEMENTS

We thank Drs. U. G. Sathyanarayana and R. Urrutia for help with the mutational tests of the candidate repressor binding proteins and Shirley Hall and the Macromolecular Sequencing Core for automated DNA sequencing.

	FOOTNOTES

^* This work was supported by United States Public Health Service Grant DK27430 and Grant DK55266 from the National Institutes of Health.The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Current address: Genomics Applications, Monsanto, Inc., 800 N. Lindberg Blvd., St. Louis, MO 63176.

^¶ To whom correspondence should be addressed: Dept. of Molecular Biology, Universtiy of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75390-9148. Tel.: 214-648-1923; Fax: 214-648-1915; E-mail: raymond.macdonald@UTSouthwestern.edu.

Published, JBC Papers in Press, September 19, 2000, DOI 10.1074/jbc.M007021200

	ABBREVIATIONS

The abbreviations used are: bp, base pair(s); hGH, human growth hormone; kb, kilobase pair(s); DMEM, Dulbecco's modified Eagle's medium; RSV, Rous sarcoma virus; CMV, Rous sarcoma virus; HSV tk, herpes simplex virus thymidine kinase; HDAC, histone deacetylase.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This Article Abstract Full Text (PDF) All Versions of this Article: 275/51/40273    most recent M007021200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal