Identification of a cis-acting DNA antisilencer element which modulates vimentin gene expression

Influence of untranslated regions on retroviral mRNA transfer and expression

Background

Deliberate cellular reprogramming is becoming a realistic objective in the clinic. While the origin of the target cells is critical, delivery of bioactive molecules to trigger a shift in cell-fate remains the major hurdle. To date, several strategies based either on non-integrative vectors, protein transfer or mRNA delivery have been investigated. In a recent study, a unique modification in the retroviral genome was shown to enable RNA transfer and its expression.

Results

Here, we used the retroviral mRNA delivery approach to study the impact of modifying gene-flanking sequences on RNA transfer. We designed modified mRNAs for retroviral packaging and used the quantitative luciferase assay to compare mRNA expression following viral transduction of cells. Cloning the untranslated regions of the vimentin or non-muscular myosin heavy chain within transcripts improved expression and stability of the reporter gene while slightly modifying reporter-RNA retroviral delivery. We also observed that while the modified retroviral platform was the most effective for retroviral mRNA packaging, the highest expression in target cells was achieved by the addition of a non-viral UTR to mRNAs containing the packaging signal.

Conclusions

Through molecular engineering we have assayed a series of constructs to improve retroviral mRNA transfer. We showed that an authentic RNA retroviral genomic platform was most efficiently transferred but that adding UTR sequences from highly expressed genes could improve expression upon transfection while having only a slight effect on expression from transferred RNA. Together, these data should contribute to the optimisation of retroviral mRNA-delivery systems that test combinations of UTRs and packaging platforms.

A single heterochromatin boundary element imposes position-independent antisilencing activity in Saccharomyces cerevisiae minichromosomes

Chromatin boundary elements serve as cis-acting regulatory DNA signals required to protect genes from the effects of the neighboring heterochromatin. In the yeast genome, boundary elements act by establishing barriers for heterochromatin spreading and are sufficient to protect a reporter gene from transcriptional silencing when inserted between the silencer and the reporter gene. Here we dissected functional topography of silencers and boundary elements within circular minichromosomes in Saccharomyces cerevisiae. We found that both HML-E and HML-I silencers can efficiently repress the URA3 reporter on a multi-copy yeast minichromosome and we further showed that two distinct heterochromatin boundary elements STAR and TEF2-UASrpg are able to limit the heterochromatin spreading in circular minichromosomes. In surprising contrast to what had been observed in the yeast genome, we found that in minichromosomes the heterochromatin boundary elements inhibit silencing of the reporter gene even when just one boundary element is positioned at the distal end of the URA3 reporter or upstream of the silencer elements. Thus the STAR and TEF2-UASrpg boundary elements inhibit chromatin silencing through an antisilencing activity independently of their position or orientation in S. cerevisiae minichromosomes rather than by creating a position-specific barrier as seen in the genome. We propose that the circular DNA topology facilitates interactions between the boundary and silencing elements in the minichromosomes.

Role of vimentin in regulation of monocyte/macrophage differentiation

Maturation of blood cells depends on dramatic changes of expression profiles of specific genes. Although these changes have been extensively studied, their functional outcomes often remain unclear. In this study, we explored the identity and function of an unknown protein that was greatly overexpressed in v-myb-transformed BM2 monoblasts undergoing differentiation to macrophage-like cells. We identified this protein as vimentin, the intermediate filament protein. We show that an increased level of vimentin protein results from activation of the vimentin gene promoter occurring in monoblastic cells induced to differentiate by multiple agents. Furthermore, our studies reveal that the vimentin gene promoter is stimulated by Myb and Jun proteins, the key transcriptional regulators of myeloid maturation. Silencing of vimentin gene expression using siRNA markedly suppressed the ability of BM2 cells to form macrophage polykaryons active in phagocytosis and producing reactive oxygen species. Taken together, these findings document that up-regulation of vimentin gene expression is important for formation of fully active macrophage-like cells and macrophage polykaryons.

Stat3 enhances vimentin gene expression by binding to the antisilencer element and interacting with the repressor protein, ZBP-89

Vimentin exhibits a complex pattern of developmental- and tissue-specific expression and is aberrantly expressed in most metastatic tumors. The human vimentin promoter contains multiple DNA elements, some of which enhance gene expression and one that inhibits. A silencer element (at -319) binds the repressor ZBP-89. Further upstream (at -757) is an element, which acts positively in the presence of the silencer element and, thus, is referred to as an antisilencer (ASE). Previously, we showed that Stat1alpha binds to this element upon induction by IFN-gamma. However, substantial binding and reporter gene activity was still present in nontreated cells. Here, we have found that Stat3 binds to the ASE element in vitro. Transfection experiments in COS-1 cells with various vimentin promoter--reporter constructs show that gene activity is dependent upon the cotransfection and activation of Stat3. Moreover, activated Stat3 can overcome ZBP-89 repression. Coimmunoprecipitation studies demonstrate that Stat3 and ZBP-89 can interact and confocal microscopy detects these factors to be colocalized in the nucleus. Moreover, a correlation exists between the presence of activated Stat3 and vimentin expression in MDA-MB-231 cells, which is lacking in MCF7 cells where vimentin is not expressed. In the light of these results, we propose that the interaction of Stat3 and ZBP-89 may be crucial for overcoming the effects of the repressor ZBP-89, which suggests a novel mode for Stat3 gene activation.

Sp1 transactivation of the TCL1 oncogene

Cis-regions and trans-factors controlling TCL1 oncogene expression are not known. We identified the functional TCL1 promoter by mapping four transcriptional start sites 24-30 bp downstream of a TATA box. A 424-bp fragment upstream of the major start site showed robust promoter activity comparable with SV40 in both TCL1 expressing and non-expressing cell lines. Additional constructs spanning 10 kb upstream and 20 kb downstream of the start site showed only modest increases in reporter activity indicating that TCL1 expression is primarily controlled by the promoter. Ten putative Sp1-binding sites were identified within 300 bp of the start site, and three of these specifically bound Sp1. A dose-dependent transactivation of the TCL1 promoter with Sp1 addition in Sp1-negative Drosophila SL2 cells was observed, and mutation of the three identified Sp1-binding sites significantly repressed reporter gene expression in 293T cells, confirming a key role for Sp1 in activating the TCL1 promoter in vivo. In TCL1 silent cell lines, CpG DNA methylation was rarely observed at functional Sp1 sites, and methylation of a previously reported NotI restriction site was associated with dense CpG methylation rather than endogenous TCL1 gene silencing. Together, these results indicate that Sp1 mediates transactivation of the TCL1 core promoter and that TCL1 gene silencing is not dependent on mechanisms involving Sp1 and NotI site methylation.

A silencer inhibitor confers specific expression of intestinal trefoil factor in gobletlike cell lines

Intestinal trefoil factor (ITF) is selectively expressed in intestinal goblet cells. Previous studies identified cis-regulatory elements in the proximal promoter of ITF, but these were insufficient to recapitulate the exquisite tissue- and cell-specific expression of native ITF in vivo. Preliminary studies suggested that goblet cell-specific expression of murine ITF requires elements far upstream that include a silencer element that effectively prevents ITF expression in non-goblet cells. Transient transfection studies using native or mutant ITF 5'-flanking sequences identified a region that restores expression in goblet cells. This element, designated goblet cell silencer inhibitor (GCSI) element, enables human and murine goblet cell-like cell lines to override the silencing effect of more proximal elements. The GCSI has no intrinsic enhancer activity and regulates expression only when the silencer element is present. Ligation of GCSI and silencer elements to sucrase-isomaltase conferred goblet cell-specific expression. Goblet cells but not non-goblet cells possess a nuclear protein that binds to the GCSI regulatory element (GCSI binding protein; GCSI-BP). Both transient transfection and gel mobility shift assay studies localize the GCSI and GCSI-BP to -2216 to -2204. We conclude that goblet cell-specific transcription of ITF in vivo depends on a regulatory element designated GCSI.

Leukocyte-specific expression of the pp52 (LSP1) promoter is controlled by the cis -acting pp52 silencer and anti-silencer elements

pp52 (LSP1) is a leukocyte-specific phosphoprotein that binds the cytoskeleton and has been implicated in affecting cytoskeletal remodeling in a variety of leukocyte functions, including cell motility and chemotaxis. The expression of pp52 is restricted to leukocytes by a 549 bp tissue-specific promoter. Here, we show that promoter fragments smaller than the 549 bp pp52 promoter have activity in fibroblasts where pp52 is not normally expressed. Specifically, a truncated construct (+1 to -99) functioned as a basal promoter active in leukocytes and fibroblasts. We identified two upstream regions within the 549 bp pp52 promoter responsible for restricting pp52 promoter activity in fibroblasts. These two regions contained a silencer (pp52 NRE) and an anti-silencer (pp52 anti-NRE) with opposing activities controlling pp52 gene expression. The pp52 NRE was active in both leukocytes and fibroblasts while the pp52 anti-NRE was only active in leukocytes, thereby allowing pp52 gene transcription in leukocytes but not in fibroblasts. The pp52 NRE was localized to an 89 bp DNA segment between -324 and -235 in the 549 bp pp52 promoter and functioned as an active silencer element in a position and orientation independent manner. The pp52 anti-NRE was localized to a 33 bp segment between -383 and -350 of the 549 bp pp52 promoter and acted as an anti-silencer element against the pp52 NRE, but lacked any intrinsic enhancing activity on its own. These findings indicate that the tissue specificity of the pp52 promoter is determined by the pp52 anti-NRE anti-silencer which over-rides the general inhibitory activity of the pp52 NRE silencer.

The zinc finger repressor, ZBP-89, binds to the silencer element of the human vimentin gene and complexes with the transcriptional activator, Sp1

Vimentin is a component of the eukaryotic cytoskeleton belonging to the family of intermediate filament proteins. It exhibits a complex pattern of tissue- and development-specific expression. It is also a marker of the metastatic potential of many tumor cells. Previously, the human vimentin promoter was shown to contain several regions for the binding of positive and negative acting regulatory factors. Until now, the silencer element, which shuts down vimentin synthesis in selected tissues during development, was not precisely localized; nor was its binding protein known. In vivo DMS footprinting by ligation-mediated PCR delineated the position of guanine residues important to vimentin expression. Transient transfection assays in HeLa cells of various vimentin 5'-end promoter sequences and mutants thereof precisely defined two regulatory elements, a negative element and an adjoining positive acting element. Band shift assays, UV cross-linking, and Southwestern blot analysis confirm that the silencer element specifically binds a protein. Several lines of evidence show that ZBP-89, a zinc finger, Kruppel-like repressor protein is vimentin's silencer element binding factor. Co-immunoprecipitation and DNA affinity chromatography prove that Sp1 heterodimerizes with ZBP-89 when bound to the silencer element to yield a DNA-protein complex whose mobility is indistinguishable from that displayed by HeLa nuclear extract in band shift assays.

Mutually exclusive expression of beta(III)-tubulin and vimentin in adrenal cortex carcinoma SW13 cells

During embryogenesis, the maturation of neuroblasts into neurones is accompanied by the down-regulation of vimentin and by the expression of neuronal microtubular proteins. Here, we show that human adrenal cortex SW13 cells express beta(III)-tubulin, MAP2b and tau. Analysis of vimentin-positive and -negative subclones of SW13 cells revealed that, under defined cultured conditions, beta(III)-tubulin and MAP2b were present only in vimentin-deficient cells and that beta(III)-tubulin repression occurred at the transcriptional level in vimentin-positive cells. These results suggest that vimentin repression and beta(III)-tubulin expression are co-ordinated by an upstream mechanism relevant to the control of cytoskeletal protein expression during neuronal development.

Antisilencing: myelin proteolipid protein gene expression in oligodendrocytes is regulated via derepression

Antisilencer or antirepressor elements have been described, thus far, for only a few eukaryotic genes and were identified by their ability not to augment gene expression per se but to override repression mediated via negative transcription regulatory elements. Here we report the first case of antisilencing for a neural-specific gene, the myelin proteolipid protein (PLP) gene (Plp). PLP is the most abundant protein found in CNS myelin. The protein is synthesized in oligodendrocytes, and its expression is regulated developmentally. Previously we have shown that a PLP-lacZ transgene (which includes the entire sequence for Plp intron 1) is regulated in mice, in a manner consistent with the spatial and temporal expression of the endogenous Plp gene. In the present report, we demonstrate by transfection analyses, using various PLP-lacZ deletion constructs, that Plp intron 1 DNA contains multiple elements that collectively regulate Plp gene expression in oligodendrocytes. One of these regulatory elements functions as an antisilencer element, which acts to override repression mediated by at least two negative regulatory elements located elsewhere within Plp intron 1 DNA. The mechanism for antisilencing appears to be complex as the intragenic region that mediates this function binds multiple nuclear factors specifically.

An antisilencer element is involved in the transcriptional regulation of the human vimentin gene

Vimentin is an intermediate filament protein normally expressed in cells of mesenchymal origin. The promoter of the human vimentin gene was previously reported to contain two positive-acting regions, separated by a negative region (Rittling, S.R., Baserga, R., 1987. Functional analysis and growth factor regulation of the human vimentin promoter. Mol. Cell. Biol. 7, 3908-3915). Here, detailed studies reveal two additional regulatory elements, a new positive transcriptional element located between -717 and -757, and a new repressor element at -780 to -821. In transient transfections, the positive-acting element is able to completely override the effect of different silencer elements when fused to a heterologous promoter. However, this element does not enhance gene activity when the silencer element is absent and thus cannot be viewed as a true enhancer. Since it appears to overcome the effect of a silencer element, we refer to it as an antisilencer element. Gel mobility shift assays, UV-cross-linking experiments, and Southwestern blots reveal that a 105-kDa protein specifically binds to this region.

Modulation of TAT gene induction by glucocorticoids involves a neutralizing sequence

Recent studies have indicated that two elements in addition to the glucocorticoid response element (GRE) are involved in the induction of the endogenous TAT gene in FuS-5 rat hepatoma cells. The first is the 21 bp glucocorticoid modulatory element (GME) at -3648 bp, which causes reporter constructs to display both a left shift in the dose-response curve for glucocorticoids and increased percentages of agonist activity for antiglucocorticoids. The second is a negative element at -3340 to -3050 that blocks the action of the GME. This last observation raised the question of how GME activity can be expressed in Fu5-5 cells in the intact TAT gene that contains both the GME and the negative element. The present study identifies a third element, a "neutralizing" sequence, that restores the activity of the GME even when otherwise inactivated by the negative element. This neutralizing sequence was located within the region surrounding the GREs of the TAT gene but is separate from the GREs. The activity of the individual GME and negative elements was found to depend upon spacing. However, in combination with the natural GRE, the native TAT gene spacing of the GME and negative elements was able to reproduce the activity of the intact gene. Thus, a total of three additional elements (an activator, a negative element, and a neutralizer) appear to cooperate with the GREs in glucocorticoid induction of the TAT gene in Fu5-5 cells. While such a grouping of elements may be novel among steroid regulated genes, it is a not uncommon occurrence for the transcriptional control of other genes.

Transcriptional repression of type I IFN genes

Transcriptional regulation is a consequence of the combination of both activation and repression for establishing specific patterns of eukaryotic gene expression. The regulation of the expression of type I interferon (IFN-A and IFN-B) multigene family is controlled primarily at the transcriptional level and has been widely studied as a model for understanding the mechanisms of stable repression, transient virus induction and postinduction repression of the genes. The positive and negative regulatory elements required for this on/off switch have been defined within a complex 5' upstream region of their transcription start site. The differential expression pattern of type I IFN genes is thought to involve both substitutions in the virus responsive element (VRE) and presence or absence of negatively acting sequences surrounding the VRE. In this review we discuss several mechanisms of negative regulation due to the existence of common or specific elements in the IFN-B and IFN-A genes and we summarize recent studies on transcriptional repressors that bind to these promoters.

A regulatory element within a coding exon modulates keratin 18 gene expression in transgenic mice

Multiple tissue-specific, DNase-hypersensitive sites are correlated with known or potential regulatory regions of the human keratin 18 (K18) gene. One of these sites is found within exon 6, close to a potential AP-1 binding site. Footprint analysis confirmed that this site is capable of binding c-Jun and c-Fos in vitro. However, exon 6 can stimulate expression of a reporter gene driven by the K18 proximal promoter independent of AP-1 in F9 cells and additionally modulates AP-1 responsiveness when in combination with an intron enhancer. Analysis in transgenic mice and by transient transfections of mutant forms of the K18 gene showed that exon 6 contributes to the expression of the K18 gene. However, substitution of part of exon 6 with the corresponding part of the keratin 19 gene which lacks an AP-1 site decreased but did not destroy the regulatory activity of the exon. Furthermore, this mutation did not alter either the tissue specificity or the position-independent and copy number-dependent behavior of the K18 gene. In contrast, a frameshift mutation within exon 6 dramatically decreased the expression of the gene. K18 RNA expression from the frameshift mutation was less than 10% of the wild type K18 transgene. This decline in expression was the result of a combination of decreased stability of mutant K18 RNA and the creation of a negative regulatory element that can interact with the first intron regulatory elements and actively suppress K18 expression. These results demonstrate that a protein-coding portion of the K18 gene also has a regulatory function.

Silencer activity in the interferon-A gene promoters

Interferon-A (IFN-A) differential gene expression is modulated by a complex interplay between cis-acting DNA elements and the corresponding specific trans-regulating factors. Substitutions in the proximal virus-responsive element of the interferon-A (IFN-A) promoters contribute to their differential gene expression. The 5' distal silencing region in the weakly virus-inducible murine IFN-A11 gene has been previously delimited. DNase I footprinting experiments and transient gene expression assays demonstrate identical silencing activity in equivalent regions of the genes for IFN-A11 and IFN-A4 promoters. A minimal 20-mer distal negative regulatory element (DNRE) in both promoters is necessary and sufficient for the silencing and a region in the highly inducible IFN-A4 promoter located between the silencer and the virus-responsive element overrides the silencer activity. Mutations in the central region of the DNRE, causing derepression, also altered the formation of one of the two major DNA-protein complexes. One of these contains a protein related to or identical to the high mobility group I(Y) proteins, while the other complex contains a major protein present in uninduced and virus-induced cells with a molecular mass of 38 kDa, which may be related to the silencer activity. Similar DNREs are present in other virus-uninducible IFN-A promoters, and these data suggest that a common silencer may mediate the transcriptional repression in different genes of this family.

Identification and characterization of a human CD4 silencer

Using transgenic mice, we have identified a human CD4 silencer contained within a 484-bp fragment in the first intron of the CD4 gene. Further experiments have mapped a lineage-specific silencing activity to a region of 190 bp. This region contains two protein-binding sites detected by deoxyribonuclease I footprinting analyses. Tested in transient transfection assays, these two DNA elements showed significant silencing activity restricted to the CD8 phenotype. In CD4 cells, either no clear effect (FP I) or strong enhancing activity (FP II) was observed by transient transfection assays. Despite the lineage-specific activity of these two elements, electrophoretic mobility shift assays (EMSA) showed similar levels of protein binding to the silencer element FP I in CD4 and CD8 T cells. Base substitutions in the FP I fragment abolished the silencing activity in transfected CD8 cells as well as the protein binding in EMSA, suggesting an important role of this protein-DNA interaction in CD4 gene regulation.

Two homologous enhancer elements in the chicken vimentin gene may bind a nuclear factor in common with a nearby silencer element

Vimentin, a cytoskeletal protein belonging to the intermediate filament protein family, exhibits a complex pattern of expression. In the case of the chicken vimentin gene, several regulatory elements within the 5' region of the gene have been characterized, including an enhancer activity between -160 and -320, which may contribute to the down-regulation of vimentin expression during myogenesis. In this study, sequences within this region were examined via transient transfections of various deletion constructs, and two distinct enhancer elements were found, one on either side of a previously described silencer element. These two enhancer elements also enhanced transcription when fused separately to the basal promoter region of the chicken vimentin gene. Gel mobility shift assays, UV cross-linking experiments, and DNase I protection studies indicate that these two enhancer elements and the silencer element all contain a common binding site for the previously described 95-kDa silencer element binding protein, suggesting that this regulatory protein can act as both an activator and a repressor.

Structural and biological consequences of increased vimentin expression in simple epithelial cell types

Cytoskeletal intermediate filaments (IFs) constitute a diverse family of proteins whose members are expressed in tissue-specific patterns. Although vimentin IFs are normally restricted to mesenchyme, a variety of cell types express vimentin alone or together with cell-specific IFs during growth, differentiation, and neoplasia. In this study, we have investigated the influence of increased vimentin expression on the simple epithelial cell phenotype. An expression vector encoding a human vimentin cDNA was transfected into murine HR9 endoderm and F9 embryonal carcinoma cell lines, which serve as models for early extraembryonic epithelial differentiation. Stable clones that expressed varying levels of human vimentin were characterized by human vimentin were characterized by immunofluorescence and biochemical analysis. A relatively high level of vimentin expression in HR9 and differentiated F9 epithelial cells resulted in aberrant vimentin structures with co-collapse of keratin K8/K18 filaments and lowered amounts of keratin protein. In F9 epithelial cells, the desmosomal proteins DP I/II did not appear to localize to cell surface desmosome s but rather but rather co-aggregated with the perturbed IFs. Although overall cell morphology was not dramatically altered, individual nuclei were distorted by excess intracellular vimentin. Furthermore, cell proliferation as well as the cell spreading response time were slowed. Ther appears to be a threshold effect regarding overall vimentin levels as cells that expressed lower amounts of the human vimentin exhibited no obvious structural nor biological effects. Our results demonstrate that wild-type vimentin can act as a "mutant" protein when present at high intracellular levels, inducing a variety of phenotypic changes.

Characterization of protein interactions with positive and negative elements regulating the apoVLDLII gene

Synthesis of avian apo very-low-density lipoprotein (apoVLDL)II is estrogen dependent and liver specific. Competence to express the apoVLDLII gene is not acquired until days 7-9 of embryogenesis and thus lags 5-6 days behind appearance of the liver primordial bud. It is not known whether the delayed ability to activate the gene is attributable to hepatic estrogen receptor profiles, or a requirement for other transcription factors not expressed at earlier stages of embryogenesis. The latter possibility is supported by developmental alterations in nuclease hypersensitivity flanking the gene that occur independently of estrogen administration. We have examined the influence of these hypersensitive regions on expression from the apoVLDLII promoter and have characterized novel protein-DNA interactions at two of them. One is located in a copy of the CR1 family of middle repetitive elements approximately 3.0 kb upstream from the start of the gene. We demonstrate by DNase I footprinting that the site contains an element which matches a predicted consensus silencer sequence. The other site contains no previously identified binding motifs. It is located between nucleotides -228 and -245 and is adjacent to an imperfect estrogen response element (ERE) that we demonstrate acts additively with a canonical ERE 30 nucleotides downstream. We have identified ubiquitous and liver-specific factors that display overlapping DNA contacts with the site. Mutation of G residues contacted by these proteins decreases hormone-inducible expression from the promoter 5- to 8-fold. Hepatic levels of the liver-enriched factor interacting with this site increase abruptly between days 7 and 9 of embryogenesis, suggesting that it may be an important determinant of the ability to express the apoVLDLII and possibly other liver-specific genes.

Functional analysis of chicken vimentin distal promoter regions in cultured lens cells

Synthesis of the cytoskeletal intermediate filament protein vimentin (Vim) in the lens is unexpected due to the mesenchymal preference of Vim-encoding gene (Vim) expression and the epithelial origin of the lens. Previous studies indicated that chicken Vim gene expression in cultured lens cells is regulated by both positive- and negative-acting sequence elements within the first -767 nucleotides (nt) of its promoter. Here, we demonstrate the existence of additional upstream chicken Vim promoter elements which function in transfected lens cells. Sequences within the nt -1360/-1156 region repressed promoter activity in transfected lens cells to levels lower than that observed for the previously defined more proximal repressor elements. The -1612/-1360 region activated promoter activity to levels similar to those observed for the strongest previously defined proximal promoter. The nt sequence analysis of the upstream promoter region revealed the presence of multiple consensus repressor and activator transcription-factor-binding sites. Several of these sites have been implicated for lens expression of enzyme-crystallin-encoding genes (cry), suggesting that Vim expression may share features with the cry genes for recruitment and high-level expression in the lens.

A 300 bp 5′-upstream sequence of a differentiation-dependent rabbit K3 keratin gene can serve as a keratinocyte-specific promoter

Keratinocytes of the suprabasal compartment of many stratified epithelia synthesize as a major differentiation product a keratin pair, consisting of an acidic and a basic keratin, which accounts for 10–20% of the newly synthesized proteins. While genes of several differentiation-related keratins have been cloned and studied, relatively little is known about the molecular basis underlying their tissue-specific and differentiation-dependent expression. We have chosen to study, as a prototype of these genes, the gene of K3 keratin, which has the unique property of being expressed in the majority of corneal epithelial basal cells but suprabasally in peripheral cornea, the site of corneal epithelial stem cells. Using a monoclonal antibody, AE5, specific for K3 keratin, and a fragment of human K3 gene as probes, we have isolated several cDNA and genomic clones of rabbit K3 keratin. One genomic clone has been sequenced and characterized, and the identity of its coding sequence with that of cDNAs indicates that it corresponds to the single, functional rabbit K3 gene. Transfection assays showed that its 3.6 kb 5′-upstream sequence can drive a chloramphenicol acetyl transferase (CAT) reporter gene to express in cultured corneal and esophageal epithelial cells, but not in mesothelial and kidney epithelial cells or fibroblasts, all of rabbit origin. Serial deletion experiments narrowed this keratinocyte-specific promoter to within -300 bp upstream of the transcription initiation site. Its activity is not regulated by the coding or 3′-noncoding sequences that have been tested so far. This 300 bp 5′-upstream sequence of K3 keratin gene, which can function in vitro as a keratinocyte-specific promoter, contains two clusters of partially overlapping motifs, one with an NFkB consensus sequence and another with a GC box. The combinatorial effects of these multiple motifs and their cognate binding proteins may play an important role in regulating the expression of this tissue-restricted and differentiation-dependent keratin gene.