An enhancer/locus control region is not sufficient to open chromatin

Gammaretroviral vector integration occurs overwhelmingly within and near DNase hypersensitive sites

Concerns surrounding the oncogenic potential of recombinant gammaretroviral vectors has spurred a great deal of interest in vector integration site (VIS) preferences. Although gammaretroviral vectors exhibit a modest preference for integration near transcription start sites (TSS) of active genes, such associations only account for about a third of all VIS. Previous studies suggested a correlation between gammaretroviral VIS and DNase hypersensitive sites (DHS), which mark chromatin regions associated with cis-regulatory elements. In order to study this issue directly, we assessed the correlation between 167 validated gammaretroviral VIS and a deep genome-wide map of DHS, both determined in the same cell line (the human fibrosarcoma HT1080). The DHS map was developed by sequencing individual DNase I cleavage sites using massively parallel sequencing technologies. These studies revealed an overwhelming preference for integrations associated with DHS, with a median distance of only 238 bp between individual VIS and the nearest DHS for the experimental dataset, compared to 3 kb for a random dataset and 577 to 1457 bp for two unrelated cell lines (p<0.001). Indeed, nearly 84% of all VIS were found to be located within 1 kb of a DHS (p=10(-43)). Further, this correlation was statistically independent from the association with TSS. The preference for DHS far exceeds that seen for other hallmarks of gammaretroviral VIS, including TSS, and may help explain several aspects of gammaretroviral vector biology, including the mechanism of VIS selection, as well as the relative frequency and underlying biology of gammaretroviral vector-mediated genotoxicity.

BEST1 expression in the retinal pigment epithelium is modulated by OTX family members

A number of genes preferentially expressed in the retinal pigment epithelium (RPE) are associated with retinal degenerative disease. One of these, BEST1, encodes bestrophin-1, a protein that when mutated causes Best macular dystrophy. As a model for RPE gene regulation, we have been studying the mechanisms that control BEST1 expression, and recently demonstrated that members of the MITF-TFE family modulate BEST1 transcription. The human BEST1 upstream region from -154 to +38 bp is sufficient to direct expression in the RPE, and positive-regulatory elements exist between -154 and -104 bp. Here, we show that the -154 to -104 bp region is necessary for RPE expression in transgenic mice and contains a predicted OTX-binding site (Site 1). Since another non-canonical OTX site (Site 2) is located nearby, we tested the function of these sites using BEST1 promoter/luciferase constructs by in vivo electroporation and found that mutation of both sites reduces promoter activity. Three OTX family proteins - OTX1, OTX2 and CRX - bound to both Sites 1 and 2 in vitro, and all of them increased BEST1 promoter activity. Surprisingly, we found that human and bovine RPE expressed not only OTX2 but also CRX, the CRX genomic region in bovine RPE was hypersensitive to DNase I, consistent with active transcription, and that both OTX2 and CRX bound to the BEST1 proximal promoter in vivo. These results demonstrate for the first time CRX expression in the RPE, and suggest that OTX2 and CRX may act as positive modulators of the BEST1 promoter in the RPE.

Deletion of the core region of 5' HS2 of the mouse beta-globin locus control region reveals a distinct effect in comparison with human beta-globin transgenes

The beta-globin locus control region (LCR) is a large DNA element that is required for high-level expression of beta-like globin genes from the endogenous mouse locus or in transgenic mice carrying the human beta-globin locus. The LCR encompasses 6 DNaseI hypersensitive sites (HSs) that bind transcription factors. These HSs each contain a core of a few hundred base pairs (bp) that has most of the functional activity and exhibits high interspecies sequence homology. Adjoining the cores are 500- to 1000-bp "flanks" with weaker functional activity and lower interspecies homology. Studies of human beta-globin transgenes and of the endogenous murine locus show that deletion of an entire HS (core plus flanks) moderately suppresses expression. However, human transgenes in which only individual HS core regions were deleted showed drastic loss of expression accompanied by changes in chromatin structure. To address these disparate results, we have deleted the core region of 5'HS2 from the endogenous murine beta-LCR. The phenotype was similar to that of the larger 5'HS2 deletion, with no apparent disruption of chromatin structure. These results demonstrate that the greater severity of HS core deletions in comparison to full HS deletions is not a general property of the beta-LCR.

Discovery of functional noncoding elements by digital analysis of chromatin structure

We developed a quantitative methodology, digital analysis of chromatin structure (DACS), for high-throughput, automated mapping of DNase I-hypersensitive sites and associated cis-regulatory sequences in the human and other complex genomes. We used 19/20-bp genomic DNA tags to localize individual DNase I cutting events in nuclear chromatin and produced approximately 257,000 tags from erythroid cells. Tags were mapped to the human genome, and a quantitative algorithm was applied to discriminate statistically significant clusters of independent DNase I cutting events. We show that such clusters identify both known regulatory sequences and previously unrecognized functional elements across the genome. We used in silico simulation to demonstrate that DACS is capable of efficient and accurate localization of the majority of DNase I-hypersensitive sites in the human genome without requiring an independent validation step. A unique feature of DACS is that it permits unbiased evaluation of the chromatin state of regulatory sequences from widely separated genomic loci. We found surprisingly large differences in the accessibility of distant regulatory sequences, suggesting the existence of a hierarchy of nuclear organization that escapes detection by conventional chromatin assays.

Spatial organization of gene expression: the active chromatin hub

Developmental and tissue-specific expression of higher eukaryotic genes involves activation of transcription at the appropriate time and place and keeping it silent otherwise. Unlike housekeeping genes, tissue-specific genes generally do not cluster on the chromosomes. They can be found in gene-dense regions of chromosomes as well as in regions of repressive chromatin. Depending on the location, shielding against positive or negative regulatory effects from neighboring chromatin may be required and hence insulator and boundary models were proposed. They postulate that chromosomes are partitioned into physically distinct expression domains, each containing a gene or gene cluster with its cis-regulatory elements. Specialized elements at the borders of such domains are proposed to prevent cross-talk between domains, and thus to be crucial in establishing independent expression domains. However, genes and associated cis-acting sequences often do not occupy physically distinct domains on the chromosomes. Rather, genes can overlap and cis-acting sequences can be found tens or hundreds of kilobases away from the target gene, sometimes with unrelated genes in between. Therefore the ability of a gene to communicate with positive cis-regulatory elements rather than the presence of specialized boundary elements appears to be key to establishing an independent expression profile. Our recent finding that active beta-globin genes physically interact in the nuclear space with multiple cis-regulatory elements, with inactive genes looping out, has provided a potential mechanistic framework for this model. We refer to such a spatial unit of regulatory DNA elements as an active chromatin hub (ACH). We propose that productive ACH formation underlies correct gene expression, requiring the presence of protein factors with the appropriate affinities for each other bound to their cognate DNA sequences. Proximity and specificity determines which cis-acting sequences and promoter(s) form an ACH, and thus which gene will be expressed. Other regulatory sequences can interfere with transcription by blocking the appropriate physical interaction between an enhancer and promoter in the ACH. Possible mechanisms by which distal DNA elements encounter each other in the 3D nuclear space will be discussed.

Genome-wide identification of DNaseI hypersensitive sites using active chromatin sequence libraries

Comprehensive identification of sequences that regulate transcription is one of the major goals of genome biology. Focal alteration in chromatin structure in vivo, detectable through hypersensitivity to DNaseI and other nucleases, is the sine qua non of a diverse cast of transcriptional regulatory elements including enhancers, promoters, insulators, and locus control regions. We developed an approach for genome-scale identification of DNaseI hypersensitive sites (HSs) via isolation and cloning of in vivo DNaseI cleavage sites to create libraries of active chromatin sequences (ACSs). Here, we describe analysis of >61,000 ACSs derived from erythroid cells. We observed peaks in the density of ACSs at the transcriptional start sites of known genes at non-gene-associated CpG islands, and, to a lesser degree, at evolutionarily conserved noncoding sequences. Peaks in ACS density paralleled the distribution of DNaseI HSs. ACSs and DNaseI HSs were distributed between both expressed and nonexpressed genes, suggesting that a large proportion of genes reside within open chromatin domains. The results permit a quantitative approximation of the distribution of HSs and classical cis-regulatory sequences in the human genome.

Beta-globin locus control region HS2 and HS3 interact structurally and functionally

The overall structure of the DNase I hypersensitive sites (HSs) that comprise the beta-globin locus control region (LCR) is highly conserved among mammals, implying that the HSs have conserved functions. However, it is not well understood how the LCR HSs, either individually or collectively, activate transcription. We analyzed the interactions of HS2, HS3 and HS4 with the human epsilon- and beta-globin genes in chromatinized episomes in fetal/embryonic K562 cells. Only HS2 activates transcription of the epsilon-globin gene, while all three HSs activate the beta-globin gene. HS3 stimulates the beta-globin gene constitutively, but HS2 and HS4 transactivation requires expression of the transcription factor EKLF, which is not present in K562 cells but is required for beta-globin expression in vivo. To begin addressing how the individual HSs may interact with one another in a complex, we linked the beta-globin gene to both the HS2 and HS3. HS2 and HS3 together resulted in synergistic stimulation of beta-globin transcription. Unexpectedly, mutated, inactive forms of HS2 impeded the activation of the beta-globin gene by HS3. Thus, there appear to be distinct interactions among the HSs and between the HSs and the globin genes. These preferential, non-exclusive interactions may underlie an important structural and functional cooperativity among the regulatory sequences of the beta-globin locus in vivo.

Locus control regions

Locus control regions (LCRs) are operationally defined by their ability to enhance the expression of linked genes to physiological levels in a tissue-specific and copy number-dependent manner at ectopic chromatin sites. Although their composition and locations relative to their cognate genes are different, LCRs have been described in a broad spectrum of mammalian gene systems, suggesting that they play an important role in the control of eukaryotic gene expression. The discovery of the LCR in the beta-globin locus and the characterization of LCRs in other loci reinforces the concept that developmental and cell lineage-specific regulation of gene expression relies not on gene-proximal elements such as promoters, enhancers, and silencers exclusively, but also on long-range interactions of various cis regulatory elements and dynamic chromatin alterations.

The murine beta-globin locus control region regulates the rate of transcription but not the hyperacetylation of histones at the active genes

Locus control regions (LCRs) are defined by their ability to confer high-level tissue-specific expression to linked genes in transgenic assays. Previously, we reported that, at its native site, the murine beta-globin LCR is required for high-level beta-globin gene expression, but is not required to initiate an open chromatin conformation of the locus. To further investigate the mechanism of LCR-mediated transcriptional enhancement, we have analyzed allele-specific beta-globin expression and the pattern of histone acetylation in the presence and absence of the LCR. In single cells from mice heterozygous for a deletion of the LCR, beta-globin expression from the LCR-deleted allele is consistently low ( approximately 1-4% of wild type). Thus, the endogenous LCR enhances globin gene expression by increasing the rate of transcription from each linked allele rather than by increasing the probability of establishing transcription per se. Furthermore, in erythroid cells from mice homozygous for the highly expressing wild-type beta-globin locus, hyperacetylation of histones H3 and H4 is localized to the LCR and active genes. In mice homozygous for the LCR deletion reduced histone hyperacetylation is observed in LCR proximal sequences; however, deletion of the LCR has no effect on the localized hyperacetylation of the genes. Together, our results suggest that, in its native genomic context, the LCR follows the rate model of enhancer function, and that the developmentally specific hyperacetylation of the globin genes is independent of both the rate of transcription and the presence of the LCR.

Role of NF-Y in in vivo regulation of the gamma-globin gene

The duplicated CCAAT box is required for gamma gene expression. We report here that the transcriptional factor NF-Y is recruited to the duplicated CCAAT box in vivo. A mutation of the duplicated CCAAT box that severely disrupts the NF-Y binding also reduces the accessibility level of the gamma gene promoter, affects the assembly of basal transcriptional machinery, and increases the recruitment of GATA-1 to the locus control region (LCR) and the proximal promoter and the recruitment of transcription cofactor CBP/p300 to the LCR. These findings suggest that recruitment of NF-Y to the duplicated CCAAT box plays a role in the chromatin opening of the gamma gene promoter as well as in the communication between the gamma gene promoter and the LCR.

Matrix attachment region-dependent function of the immunoglobulin mu enhancer involves histone acetylation at a distance without changes in enhancer occupancy

Nuclear matrix attachment regions (MARs), which flank the immunoglobulin mu heavy-chain enhancer on either side, are required for the activation of the distal variable-region (V(H)) promoter in transgenic mice. Previously, we have shown that the MARs extend a local domain of chromatin accessibility at the mu enhancer to more distal sites. In this report, we examine the influence of MARs on the formation of a nucleoprotein complex at the enhancer and on the acetylation of histones, which have both been implicated in contributing to chromatin accessibility. By in vivo footprint analysis of transgenic mu gene constructs, we show that the occupancy of factor-binding sites at the mu enhancer is similar in transcriptionally active wild-type and transcriptionally inactive DeltaMAR genes. Chromatin immunoprecipitation experiments indicate, however, that the acetylation of histones at enhancer-distal nucleosomes is enhanced 10-fold in the presence of MARs, whereas the levels of histone acetylation at enhancer-proximal nucleosomes are similar for wild-type and DeltaMAR genes. Taken together, these data indicate that the function of MARs in mediating long-range chromatin accessibility and transcriptional activation of the V(H) promoter involves the generation of an extended domain of histone acetylation, independent of changes in the occupancy of the mu enhancer.

A large upstream region is not necessary for gene expression or hypersensitive site formation at the mouse beta -globin locus

Developmental expression at the beta-globin locus is regulated in part by the locus control region, a region upstream of the genes containing at least five major DNase I hypersensitive sites (HSs) in mammalian erythrocytes. Sequences farther 5' of these HSs are conserved in mouse and human, and both loci are embedded within a cluster of functional odorant receptor genes. In humans, distant upstream sequences have been implicated in regulation of the beta-globin genes. In this study, the role of the 5'-most HSs and their adjacent sequence was investigated by deletion of an 11-kb region from the mouse locus, including 5'HS 4.2, 5'HS 5, 5'HS 6, and the 5'beta1 odorant receptor gene. Mice that were homozygous for this deletion were fully viable, and no significant effect on adult beta-globin gene expression was seen. 5'HSs 1-4, which are located downstream of the deletion, were still present in the mutant mice. In addition, two new upstream HSs, HS -60.7 and HS -62.5, were found in erythroid tissue of both wild-type and mutant mice. Therefore, although the possibility of a minor role still exists, neither the HSs nor the other regions deleted in this study are essential for beta-globin gene expression, and it is unlikely that chromatin structure is affected either upstream or downstream of the deletion. This is the largest deletion at the mouse locus control region to show no apparent phenotype, and focuses attention on the possible contribution of sequences even farther upstream.

The H19 endodermal enhancer is required for Igf2 activation and tumor formation in experimental liver carcinogenesis

The expression of the linked but reciprocally imprinted Igf2 and H19 genes is activated in adult liver in the course of tumor development. By in situ hybridization analysis we have shown that both the Igf2 and H19 RNAs are expressed in the majority of the neoplastic nodules, and that hepatocellular carcinomas are developed in an experimental model of liver carcinogenesis. H19 is also highly activated in smaller and less distinct hyperplastic regions. The few neoplastic areas showing Igf2 but no H19 RNA display loss of the maternally inherited allele at the Igf2/H19 locus. These data are compatible with the existence of a common activation mechanism of these two genes during liver carcinogenesis and with a stronger H19 induction in the pre-neoplastic lesions. By using mice carrying a deletion of the H19 endodermal enhancer, we show that this regulatory element is necessary for the activation of the Igf2 and H19 genes upon induction of liver carcinogenesis. Furthermore, multiple sites of the H19 endodermal enhancer region become hypersensitive to DNase I when the carcinogenesis process is induced. Lastly, liver tumors developed in mice paternally inheriting the H19 enhancer deletion are found to have marked growth delays, increased frequency of apoptotic nuclei, and lack of Igf2 mRNA expression, thus indicating that this regulatory element plays a major role in the progression of liver carcinogenesis, since it is required for the activation of the anti-apoptotic Igf2 gene.

Structural and functional cross-talk between a distant enhancer and the epsilon-globin gene promoter shows interdependence of the two elements in chromatin

We investigated the requirements for enhancer-promoter communication by using the human beta-globin locus control region (LCR) DNase I-hypersensitive site 2 (HS2) enhancer and the epsilon-globin gene in chromatinized minichromosomes in erythroid cells. Activation of globin genes during development is accompanied by localized alterations of chromatin structure, and CACCC binding factors and GATA-1, which interact with both globin promoters and the LCR, are believed to be critical for globin gene transcription activation. We found that an HS2 element mutated in its GATA motif failed to remodel the epsilon-globin promoter or activate transcription yet HS2 nuclease accessibility did not change. Accessibility and transcription were reduced at promoters with mutated GATA-1 or CACCC sites. Strikingly, these mutations also resulted in reduced accessibility at HS2. In the absence of a globin gene, HS2 is similarly resistant to nuclease digestion. In contrast to observations in Saccharomyces cerevisiae, HS2-dependent promoter remodeling was diminished when we mutated the TATA box, crippling transcription. This mutation also reduced HS2 accessibility. The results indicate that the epsilon-globin promoter and HS2 interact both structurally and functionally and that both upstream activators and the basal transcription apparatus contribute to the interaction. Further, at least in this instance, transcription activation and promoter remodeling by a distant enhancer are not separable.

A novel element upstream of the Vgamma2 gene in the murine T cell receptor gamma locus cooperates with the 3' enhancer to act as a locus control region

Transgenic expression constructs were employed to identify a cis-acting transcription element in the T cell receptor (TCR)-gamma locus, called HsA, between the Vgamma5 and Vgamma2 genes. In constructs lacking the previously defined enhancer (3'E(Cgamma1)), HsA supports transcription in mature but not immature T cells in a largely position-independent fashion. 3'E(Cgamma1), without HsA, supports transcription in immature and mature T cells but is subject to severe position effects. Together, the two elements support expression in immature and mature T cells in a copy number-dependent, position-independent fashion. Furthermore, HsA was necessary for consistent rearrangement of transgenic recombination substrates. These data suggest that HsA provides chromatin-opening activity and, together with 3'E(Cgamma1), constitutes a T cell-specific locus control region for the TCR-gamma locus.

The beta-globin promoter is important for recruitment of erythroid Krüppel-like factor to the locus control region in erythroid cells

Erythroid Krüppel-like factor (EKLF), which binds to the CACCC box in the beta-globin promoter, is required for the expression of the beta-globin gene in adult erythroid cells. It was recently demonstrated that EKLF is also required for the activity of the beta-globin locus control region (LCR) 5'HS3. Some evidence suggests that the LCR and the beta-globin promoter interact in adult erythroid cells, and the network of protein-protein interactions that exists between these two elements may regulate how EKLF is recruited to the LCR. In this report, we use the PIN*POINT assay to study the role of the promoter on the recruitment of EKLF to 5'HS2 and 5'HS3 of the LCR. We find that recruitment of EKLF to 5'HS2 requires the TATA box, but recruitment to 5'HS3 depends on the CACCC and TATA boxes of the beta-globin promoter. Furthermore, recruitment of EKLF to 5'HS3 only occurred in beta-globin-expressing murine erythroid leukemia cells, whereas recruitment of EKLF to 5'HS2 occurred in both gamma-globin-expressing K562 cells and murine erythroid leukemia cells. Unlike EKLF, Sp1, which also binds to CACCC boxes, is not recruited to 5'HS3. We have also examined how one 5'HS affects the recruitment of EKLF to another 5'HS. We have found that the recruitment of EKLF to 5'HS3 depends on the presence of 5'HS2 in cis, but the recruitment to 5'HS2 does not depend on 5'HS3. Based on these results, we present a model that illustrates how EKLF may be recruited to the beta-globin locus.

Life without white fat: a transgenic mouse

We have generated a transgenic mouse with no white fat tissue throughout life. These mice express a dominant-negative protein, termed A-ZIP/F, under the control of the adipose-specific aP2 enhancer/promoter. This protein prevents the DNA binding of B-ZIP transcription factors of both the C/EBP and Jun families. The transgenic mice (named A-ZIP/F-1) have no white adipose tissue and dramatically reduced amounts of brown adipose tissue, which is inactive. They are initially growth delayed, but by week 12, surpass their littermates in weight. The mice eat, drink, and urinate copiously, have decreased fecundity, premature death, and frequently die after anesthesia. The physiological consequences of having no white fat tissue are profound. The liver is engorged with lipid, and the internal organs are enlarged. The mice are diabetic, with reduced leptin (20-fold) and elevated serum glucose (3-fold), insulin (50- to 400-fold), free fatty acids (2-fold), and triglycerides (3- to 5-fold). The A-ZIP/F-1 phenotype suggests a mouse model for the human disease lipoatrophic diabetes (Seip-Berardinelli syndrome), indicating that the lack of fat can cause diabetes. The myriad of consequences of having no fat throughout development can be addressed with this model.

The locus control region is necessary for gene expression in the human beta-globin locus but not the maintenance of an open chromatin structure in erythroid cells

Studies in many systems have led to the model that the human beta-globin locus control region (LCR) regulates the transcription, chromatin structure, and replication properties of the beta-globin locus. However the precise mechanisms of this regulation are unknown. We have developed strategies to use homologous recombination in a tissue culture system to examine how the LCR regulates the locus in its natural chromosomal environment. Our results show that when the functional components of the LCR, as defined by transfection and transgenic studies, are deleted from the endogenous beta-globin locus in an erythroid background, transcription of all beta-globin genes is abolished in every cell. However, formation of the remaining hypersensitive site(s) of the LCR and the presence of a DNase I-sensitive structure of the beta-globin locus are not affected by the deletion. In contrast, deletion of 5'HS5 of the LCR, which has been suggested to serve as an insulator, has only a minor effect on beta-globin transcription and does not influence the chromatin structure of the locus. These results show that the LCR as currently defined is not necessary to keep the locus in an "open" conformation in erythroid cells and that even in an erythroid environment an open locus is not sufficient to permit transcription of the beta-like globin genes.

Variable and tissue-specific hormone resistance in heterotrimeric Gs protein alpha-subunit (Gsalpha) knockout mice is due to tissue-specific imprinting of the gsalpha gene

Albright hereditary osteodystrophy (AHO), an autosomal dominant disorder characterized by short stature, obesity, and skeletal defects, is associated with heterozygous inactivating mutations of GNAS1, the gene encoding the heterotrimeric G protein alpha-subunit (Gsalpha) that couples multiple receptors to the stimulation of adenylyl cyclase. It has remained unclear why only some AHO patients present with multihormone resistance and why AHO patients demonstrate resistance to some hormones [e.g., parathyroid hormone (PTH)] but not to others (e.g., vasopressin), even though all activate adenylyl cyclase. We generated mice with a null allele of the mouse homolog Gnas. Homozygous Gs deficiency is embryonically lethal. Heterozygotes with maternal (m-/+) and paternal (+/p-) inheritance of the Gnas null allele have distinct phenotypes, suggesting that Gnas is an imprinted gene. PTH resistance is present in m-/+, but not +/p-, mice. Gsalpha expression in the renal cortex (the site of PTH action) is markedly reduced in m-/+ but not in +/p- mice, demonstrating that the Gnas paternal allele is imprinted in this tissue. Gnas is also imprinted in brown and white adipose tissue. The maximal physiological response to vasopressin (urinary concentrating ability) is normal in both m-/+ and +/p- mice and Gnas is not imprinted in the renal inner medulla (the site of vasopressin action). Tissue-specific imprinting of Gnas is likely the mechanism for variable and tissue-specific hormone resistance in these mice and a similar mechanism might explain the variable phenotype in AHO.

The developmental activation of the chicken lysozyme locus in transgenic mice requires the interaction of a subset of enhancer elements with the promoter

The complete chicken lysozyme locus is expressed in a position independent fashion in macrophages of transgenic mice and forms the identical chromatin structure as observed with the endogenous gene in chicken cells. Individual lysozyme cis -regulatory elements reorganize their chromatin structure at different developmental stages. Accordingly, their activities are developmentally regulated, indicating a differential role of these elements in locus activation. We have shown previously that a subset of enhancer elements and the promoter are sufficient to activate transcription of the chicken lysozyme gene at the correct developmental stage. Here, we analyzed to which grade the developmentally controlled chromatin reorganizing capacity of cis -regulatory elements in the 5'-region of the chicken lysozyme locus is dependent on promoter elements, and we examined whether the lysozyme locus carries a dominant chromatin reorganizing element. To this end we generated transgenic mouse lines carrying constructs with a deletion of the lysozyme promoter. Expression of the transgene in macrophages is abolished, however, the chromatin reorganizing ability of the cis -regulatory elements is differentially impaired. Some cis -elements require the interaction with the promoter to stabilize transcription factor complexes detectable as DNase I hypersensitive sites in chromatin, whereas other elements reorganize their chromatin structure autonomously.

Distal enhancer regulation by promoter derepression in topologically constrained DNA in vitro

Long-range promoter-enhancer interactions are a crucial regulatory feature of many eukaryotic genes yet little is known about the mechanisms involved. Using cloned chicken betaA-globin genes, either individually or within the natural chromosomal locus, enhancer-dependent transcription is achieved in vitro at a distance of 2 kb with developmentally staged erythroid extracts. This occurs by promoter derepression and is critically dependent upon DNA topology. In the presence of the enhancer, genes must exist in a supercoiled conformation to be actively transcribed, whereas relaxed or linear templates are inactive. Distal protein-protein interactions in vitro may be favored on supercoiled DNA because of topological constraints. In this system, enhancers act primarily to increase the probability of rapid and efficient transcription complex formation and initiation. Repressor and activator proteins binding within the promoter, including erythroid-specific GATA-1, mediate this process.

Regulation of the myeloid-cell-expressed human gp91-phox gene as studied by transfer of yeast artificial chromosome clones into embryonic stem cells: suppression of a variegated cellular pattern of expression requires a full complement of distant cis elements

Identifying the full repertoire of cis elements required for gene expression in mammalian cells (or animals) is challenging, given the moderate sizes of many loci. To study how the human gp91-phox gene is expressed specifically in myeloid hematopoietic cells, we introduced yeast artificial chromosome (YAC) clones and derivatives generated in yeast into mouse embryonic stem cells competent to differentiate to myeloid cells in vitro or into mouse chimeras. Fully appropriate regulation was recapitulated with a 130-kb YAC containing 60 and 30 kb of 5' and 3' flanking sequences, respectively. Immunodetection of human gp91-phox protein revealed uniform expression in individual myeloid cells. The removal of upstream sequences led to decreased overall expression which reflected largely a variegated pattern of expression, such that cells were either "on" or "off," rather than pancellular loss of expression. The proportion of clones displaying marked variegation increased with progressive deletion. DNase I mapping of chromatin identified two hypersensitive clusters, consistent with the presence of multiple regulatory elements. Our findings point to cooperative interactions of complex regulatory elements and suggest that the presence of an incomplete set of elements reduces the probability that an open chromatin domain (or active transcriptional complex) may form or be maintained in the face of repressive influences of neighboring chromatin.

Sheltering of gamma-globin expression from position effects requires both an upstream locus control region and a regulatory element 3' to the A gamma-globin gene

Integration position-independent expression of human globin transgenes in transgenic mice requires the presence of regulatory elements from the beta-globin locus control region (LCR) in the transgene construct. However, several recent studies have suggested that, while clearly necessary, such elements are not by themselves sufficient to realize this effect. In the case of the human fetal gamma-globin genes, previous results have indicated that additional regulatory information required for sheltering of gamma-globin transgene expression from position effects may reside downstream from the A gamma gene. To investigate this possibility, we established 17 lines of transgenic mice carrying constructs comprising a micro-LCR (microLCR) element, an A gamma-globin gene fragment, and a variable length of 3' sequence information beyond the A gamma 3' HindIII site. gamma-Globin expression during development was studied in 170 individual F2 progeny from these lines. We find that gamma-globin expression becomes sheltered from position effects when the normally position-sensitive microLCR-A gamma construct is extended by 600 bp beyond the 3' HindIII site to include a previously identified regulatory sequence (the A gamma-globin enhancer), the functional significance of which in vivo had heretofore been unclear. The results suggest that the mechanism whereby an upstream LCR achieves sheltering of globin gene expression from position effects involves cooperation with a gene-proximal regulatory element distinct from the promoter region.

Essential role of NF-E2 in remodeling of chromatin structure and transcriptional activation of the epsilon-globin gene in vivo by 5' hypersensitive site 2 of the beta-globin locus control region

Much of our understanding of the process by which enhancers activate transcription has been gained from transient-transfection studies in which the DNA is not assembled with histones and other chromatin proteins as it is in the cell nucleus. To study the activation of a mammalian gene in a natural chromatin context in vivo, we constructed a minichromosome containing the human epsilon-globin gene and portions of the beta-globin locus control region (LCR). The minichromosomes replicate and are maintained at stable copy number in human erythroid cells. Expression of the minichromosomal epsilon-globin gene requires the presence of beta-globin LCR elements in cis, as is the case for the chromosomal gene. We determined the chromatin structure of the epsilon-globin gene in both the active and inactive states. The transcriptionally inactive locus is covered by an array of positioned nucleosomes extending over 1,400 bp. In minichromosomes with a (mu)LCR or DNase I-hypersensitive site 2 (HS2) which actively transcribe the epsilon-globin gene, the nucleosome at the promoter is altered or disrupted while positioning of nucleosomes in the rest of the locus is retained. All or virtually all minichromosomes are simultaneously hypersensitive to DNase I both at the promoter and at HS2. Transcriptional activation and promoter remodeling, as well as formation of the HS2 structure itself, depended on the presence of the NF-E2 binding motif in HS2. The nucleosome at the promoter which is altered upon activation is positioned over the transcriptional elements of the epsilon-globin gene, i.e., the TATA, CCAAT, and CACCC elements, and the GATA-1 site at -165. The simple availability of erythroid transcription factors that recognize these motifs is insufficient to allow expression. As in the chromosomal globin locus, regulation also occurs at the level of chromatin structure. These observations are consistent with the idea that one role of the beta-globin LCR is to maintain promoters free of nucleosomes. The restricted structural change observed upon transcriptional activation may indicate that the LCR need only make a specific contact with the proximal gene promoter to activate transcription.

RIGS (repeat-induced gene silencing) in Arabidopsis is transcriptional and alters chromatin configuration

We have previously reported repeat-induced gene silencing (RIGS) in Arabidopsis, in which transgene expression may be silenced epigenetically when repeated sequences are present. Among an allelic series of lines comprising a primary transformant and various recombinant progeny carrying different numbers of drug resistance gene copies at the same locus, silencing was found to depend strictly on repeated sequences and to correlate with an absence of steady-state mRNA. We now report characterization, in nuclei isolated from the same transgenic lines, of gene expression by nuclear run-on assay and of chromatin structure by nuclease protection assay. We find that silencing is correlated with absence of run-on transcripts, indicating that expression is silenced at the level of transcription. We find further that silencing is also correlated with increased resistance to both DNase I and micrococcal nuclease, indicating that the silenced state reflects a change in chromatin configuration. We propose that silencing results when a locally paired region of homologous repeated nucleotide sequences is flanked by unpaired heterologous DNA, which leads chromatin to adopt a local configuration that is difficult to transcribe, and possibly akin to heterochromatin.

Chromatin structure and gene expression

It is now well understood that chromatin structure is perturbed in the neighborhood of expressed genes. This is most obvious in the neighborhood of promoters and enhancers, where hypersensitivity to nucleases marks sites that no longer carry canonical nucleosomes, and to which transcription factors bind. To study the relationship between transcription factor binding and the generation of these hypersensitive regions, we mutated individual cis-acting regulatory elements within the enhancer that lies between the chicken beta- and epsilon-globin genes. Constructions carrying the mutant enhancer were introduced by stable transformation into an avian erythroid cell line. We observed that weakening the enhancer resulted in creation of two classes of site: those still completely accessible to nuclease attack and those that were completely blocked. This all-or-none behavior suggests a mechanism by which chromatin structure can act to sharpen the response of developmental systems to changing concentrations of regulatory factors. Another problem raised by chromatin structure concerns the establishment of boundaries between active and inactive chromatin domains. We have identified a DNA element at the 5' end of the chicken beta-globin locus, near such a boundary, that has the properties of an insulator; in test constructions, it blocks the action of an enhancer on a promoter when it is placed between them. We describe the properties and partial dissection of this sequence. A third problem is posed by the continued presence of nucleosomes on transcribed genes, which might prevent the passage of RNA polymerase. We show, however, that a prokaryotic polymerase can transcribe through a histone octamer on a simple chromatin template. The analysis of this process reveals that an octamer is capable of transferring from a position in front of the polymerase to one behind, without ever losing its attachment to the DNA.

A dynamic assembly of diverse transcription factors integrates activation and cell-type information for interleukin 2 gene regulation

The interleukin 2 (IL-2) gene is subject to two types of regulation: its expression is T-lymphocyte-specific and it is acutely dependent on specific activation signals. The IL-2 transcriptional apparatus integrates multiple types of biochemical information in determining whether or not the gene will be expressed, using multiple diverse transcription factors that are each optimally activated or inhibited by different signaling pathways. When activation of one or two of these factors is blocked IL-2 expression is completely inhibited. The inability of the other, unaffected factors to work is explained by the striking finding that none of the factors interacts stably with its target site in the IL-2 enhancer unless all the factors are present. Coordinate occupancy of all the sites in the minimal enhancer is apparently maintained by continuous assembly and disassembly cycles that respond to the instantaneous levels of each factor in the nuclear compartment. In addition, the minimal enhancer undergoes specific increases in DNase I accessibility, consistent with dramatic changes in chromatin structure upon activation. Still to be resolved is what interaction(s) conveys T-lineage specificity. In the absence of activating signals, the minimal IL-2 enhancer region in mature T cells is apparently unoccupied, exactly as in non-T lineage cells. However, in a conserved but poorly studied upstream region, we have now mapped several novel sites of DNase I hypersensitivity in vivo that constitutively distinguish IL-2 producer type T cells from cell types that cannot express IL-2. Thus a distinct domain of the IL-2 regulatory sequence may contain sites for competence- or lineage-marking protein contacts.

The effect of matrix attached regions (MAR) and specialized chromatin structure (SCS) on the expression of gene constructs in cultured cells and in transgenic mice

The flanking sequences of several genes have been shown to direct a position independent expression of transgenes. Attempts to completely identify the insulating sequences have failed so far. Some of these sequences contain a matrix attached region (MAR) located in the flanking part of the genes. This article will show that the MARs in cultured cells located in the 3' OH region of the human apolipoprotein B100 (Apo B100) and within the SV40 genome were unable to stimulate and insultate transgene expression directed by the promoters from a rabbit whey acidic protein (WAP) gene or from human cytomegalovirus (hCMV) early genes. In transgenic mice, the MAR from the Apo B100 and SV40 genes did not enhance the expression of a transgene containing the rabbit whey acid protein (WAP) promotor, the late gene SV40 intron (VP1 intron), the bovine growth hormone (bGH) cDNA and the SV40 late gene terminator. This construct was even toxic for embryos. Similarly, the specialized chromatin structure (SCS) from the Drosophila 87A7 HSP70 gene reduced chloramphenicol acetyl transferase (CAT) activity when added between a cytomegalovirus (CMV) enhancer and a Herpes simplex thymidine kinase (TK) gene promoter. This inhibitory action was almost complete when a second SCS sequence was added before the CMV enhancer. Sequences from the firefly luciferase and from the human gene cathepsin D cDNA used as control unexpectedly showed a similar inhibitory effect when added to the CMVTKCAT construct instead of SCS. When added before the CMV enhancer and after the transcription terminator in the CMVTKCAT construct, the SCS sequence was unable to insulate the integrated gene as seen by the fact that the level of CAT in cell extracts were by no means correlated with the number of copies in individual clones. From these data, it is concluded that i) a MAR containing the canonical AT rich sequences does not amplify the expression of all gene constructs ii) At rich MAR sequences do not have per se an insulating effect iii) Drosophila SCS from the 87A7 HSP70 gene has no insulating effect in all gene constructs (at least in mammalian cells) iv) and the addition of a DNA fragment between an enhancer and a promoter in a gene construct cannot be used as a reliable test to evaluate its insulating property.

Function of the upstream hypersensitive sites of the chicken beta-globin gene cluster in mice

We have shown previously that the chicken beta A-globin gene, with its 3' enhancer, is expressed in a copy number-dependent manner in transgenic mice. The expression level was low but increased approximately 6-fold upon inclusion of 11 kb of upstream DNA containing four DNase I hypersensitive sites. To study the effect of the individual upstream hypersensitive sites on transgene expression, we produced lines of mice in which the individual upstream sites were linked to the beta A gene and enhancer. RNA levels were measured in blood from adult animals. With each of these four constructs, the level of transgene RNA per DNA copy varied over a > 20-fold range. These data suggest that addition of a hypersensitive site to the beta A-globin/enhancer region abrogates its position independent expression. The average beta A-globin expression per copy in the lines carrying an upstream site was comparable with that in lines without an upstream site. Thus, no single upstream hypersensitive site accounts for the higher level of beta A-globin expression seen in mice containing the complete upstream region. We had shown previously that control of the chicken beta-globin cluster is distributed between at least two regions, the beta A/epsilon enhancer and the upstream region. Our current results suggest that the control mediated by the upstream DNA is itself distributed and is not due to a single hypersensitive site.

A novel cis-acting centromeric DNA element affects S. pombe centromeric chromatin structure at a distance

The chromatin structure of the central core region of Schizosaccharomyces pombe centromeric DNA is unusual. This distinctive chromatin structure is associated only with central core sequences in a functional context and is modulated by a novel cis-acting DNA element (centromere enhancer) within the functionally critical K centromeric repeat, which is found in multiple copies in all three S. pombe centromeres. The centromere enhancer alters central core chromatin structure from a distance and in an orientation-independent manner without altering the nucleosomal packaging of sequences between the enhancer and the central core. These findings suggest a functionally relevant structural interaction between the enhancer and the centromeric central core brought about by DNA looping.

Expression of the chicken beta-globin gene cluster in mice: correct developmental expression and distributed control

To investigate the regulation of gene clusters, we introduced the entire chicken beta-globin cluster into mice. This 35-kb region includes the four globin genes (rho-beta H-beta A-epsilon), the four upstream hypersensitive sites, and the intergenic beta A/epsilon enhancer. The chicken globins are not arranged in order of developmental expression, which is unlike the case for the human beta-globin cluster, in which gene order plays a role in the regulation of globin expression. Mice carrying the chicken cluster expressed the transgenes with the same developmental patterns as seen in the chicken. Therefore, stage-specific erythroid transcriptional milieus existed before the divergence of birds and mammals and have been conserved since then. Mice bearing the complete cluster except for a deletion removing the beta A/epsilon enhancer displayed markedly reduced expression of the beta H, beta A, and epsilon genes with efficient (but variable) rho expression. Mice carrying the four genes and beta A/epsilon enhancer but without the upstream hypersensitive sites showed reduced expression of rho, beta H, and beta A, with variable expression of epsilon. We conclude that (i) all of the genes (except possibly rho) are under the control of both the upstream hypersensitive sites and the enhancer, (ii) the influence of the control elements can extend beyond the nearest active gene, (iii) a single element (the enhancer) can influence more than one gene in a single developmental stage, (iv) the enhancer can work bidirectionally, and (v) neither the upstream sites (as a group) nor the enhancer showed developmental stage specificity. Thus, the regulation of this cluster is achieved by interaction of two distinct control regions with each of the globin genes.

The bulk chromatin structure of a murine transgene does not vary with its transcriptional or DNA methylation status

The DNA methylation status of HRD, a murine transgene, can be controlled by the genetic background upon which it is carried. We found the transgene to be transcribed in competent tissues only when undermethylated. Chromatin structure over the transgene was assayed by nuclear accessibility with DNase I, MspI, and PstI. While the transgene was up to fivefold more resistant to MspI when methylated than when not methylated, we observed no such difference with DNase I or PstI. We suggest that methyl-CpG-binding proteins are responsible for the difference observed with MspI, but that the chromatin structures are otherwise similarly compacted. Methylation could, therefore, play a regulatory role in gene expression beyond that which can be accomplished by bulk chromatin structure alone.

Dissection of the locus control function located on the chicken lysozyme gene domain in transgenic mice

The entire chicken lysozyme gene locus including all known cis-regulatory sequences and the 5' and 3' matrix attachment sites defining the borders of the DNase I sensitive chromatin domain, is expressed at a high level and independent of its chromosomal position in macrophages of transgenic mice. It was concluded that the lysozyme gene locus carries a locus control function. We analysed several cis-regulatory deletion mutants to investigate their influence on tissue specificity and level of expression. Position independent expression of the gene is lost whenever one of the upstream tissue specific enhancer regions is deleted, although tissue specific expression is usually retained. Deletion of the domain border fragments has no influence on copy number dependency of expression. However, without these regions an increased incidence of ectopic expression is observed. This suggests that the domain border fragments may help to suppress transgene expression in inappropriate tissues. We conclude, that position independent expression of the lysozyme gene is not controlled by a single specific region of the locus but is the result of the concerted action of several tissue specific upstream regulatory DNA elements with the promoter.

Chromosomal position effects in chicken lysozyme gene transgenic mice are correlated with suppression of DNase I hypersensitive site formation

The complete chicken lysozyme gene locus is expressed copy number dependently and at a high level in macrophages of transgenic mice. Gene expression independent of genomic position can only be achieved by the concerted action of all cis regulatory elements located on the lysozyme gene domain. Position independency of expression is lost if one essential cis regulatory region is deleted. Here we compared the DNase I hypersensitive site (DHS) pattern formed on the chromatin of position independently and position dependently expressed transgenes in order to assess the influence of deletions within the gene domain on active chromatin formation. We demonstrate, that in position independently expressed transgene all DHSs are formed with the authentic relative frequency on all genes. This is not the case for position dependently expressed transgenes. Our results show that the formation of a DHS during cellular differentiation does not occur autonomously. In case essential regulatory elements of the chicken lysozyme gene domain are lacking, the efficiency of DHS formation on remaining cis regulatory elements during myeloid differentiation is reduced and influenced by the chromosomal position. Hence, no individual regulatory element on the lysozyme domain is capable of organizing the chromatin structure of the whole locus in a dominant fashion.

Position independence and proper developmental control of gamma-globin gene expression require both a 5' locus control region and a downstream sequence element

We have analyzed the expression of human gamma-globin genes during development in F2 progeny of transgenic mice carrying two types of constructs. In the first type, gamma-globin genes were linked individually to large (approximately 4-kb) sequence fragments spanning locus control region (LCR) hypersensitive site 2 (HS2) or HS3. These LCR fragments contained not only the core HS elements but also extensive evolutionarily conserved flanking sequences. The second type of construct contained tandem gamma- and beta-globin genes linked to identical HS2 or HS3 fragments. We show that gamma-globin expression in transgenic mice carrying HS2 gamma or HS3 gamma constructs is highly sensitive to position effects and that such effects override the cis regulatory elements present in these constructs to produce markedly different developmental patterns of gamma-globin expression in lines carrying the same transgene. In contrast, gamma-globin expression in both HS2 gamma beta and HS3 gamma beta mice is sheltered from position effects and the developmental patterns of gamma-globin expression in lines carrying the same transgene are identical and display stage-specific regulation. The results suggest that cis regulatory sequences required for proper developmental control of fetal globin expression in the presence of an LCR element reside downstream from the gamma genes.

Position independence and proper developmental control of gamma-globin gene expression require both a 5' locus control region and a downstream sequence element

We have analyzed the expression of human gamma-globin genes during development in F2 progeny of transgenic mice carrying two types of constructs. In the first type, gamma-globin genes were linked individually to large (approximately 4-kb) sequence fragments spanning locus control region (LCR) hypersensitive site 2 (HS2) or HS3. These LCR fragments contained not only the core HS elements but also extensive evolutionarily conserved flanking sequences. The second type of construct contained tandem gamma- and beta-globin genes linked to identical HS2 or HS3 fragments. We show that gamma-globin expression in transgenic mice carrying HS2 gamma or HS3 gamma constructs is highly sensitive to position effects and that such effects override the cis regulatory elements present in these constructs to produce markedly different developmental patterns of gamma-globin expression in lines carrying the same transgene. In contrast, gamma-globin expression in both HS2 gamma beta and HS3 gamma beta mice is sheltered from position effects and the developmental patterns of gamma-globin expression in lines carrying the same transgene are identical and display stage-specific regulation. The results suggest that cis regulatory sequences required for proper developmental control of fetal globin expression in the presence of an LCR element reside downstream from the gamma genes.

The 5' flanking region of the rat LAP (C/EBP beta) gene can direct high-level, position-independent, copy number-dependent expression in multiple tissues in transgenic mice

The efficiency and tissue-specificity of transgene expression in animals is usually subject to the position of integration into the host chromatin. We have discovered that a 2.8kbp fragment flanking the rat gene encoding the transcription factor LAP (C/EBP beta) directs position-independent, copy number-dependent expression in transgenic-mouse livers. Concomitantly, the DNAse I hypersensitivity pattern normally observed in the liver is established in the integrated transgene construct demonstrating that this region is capable of creating chromatin structures equivalent to the endogenous situation. These observations are reminiscent of the locus control regions (LCR) described for several genes. Additionally, this LAP element functions with both intron-less and intron-containing genes. The tissue specificity of this element, however, is not restricted to liver. The 2.8kbp region is capable of allowing position-independent, copy number-dependent expression in brain, kidney, heart, spleen, and lung, but in a construct-dependent manner. This is, to our knowledge, the first transcription factor gene with which a cis-linked LCR-like element has been associated.

Sp1 is essential for both enhancer-mediated and basal activation of the TATA-less human adenosine deaminase promoter

Tissue-specific expression of the human adenosine deaminase (ADA) gene is mediated by transcriptional activation over a thousand-fold range. Cis-regulatory regions responsible for high and basal levels of activation include an enhancer and the proximal promoter region. While analyses of the T-cell specific enhancer have been carried out, detailed studies of the the promoter region or promoter-enhancer interactions have not. Examination of the promoter region by homology searches revealed six putative Sp1 binding sites. DNase I footprinting showed that Sp1 is able to bind these sites. Deletion analysis indicated that the proximal Sp1 site is required for activation of a reporter gene to detectable levels and that the more distal Sp1 sites further activate the level of expression. Inclusion of an ADA enhancer-containing fragment in these deletion constructions demonstrated that Sp1 sites are also essential for enhancer function. Apparently Sp1 controls not only low level expression but is also an integral part of the mechanism by which the enhancer achieves high level ADA expression. Mutagenesis of a potential TBP binding site at base pairs -21 to -26 decreased activity only two-fold indicating that it is not essential for transcriptional activation or enhancement.

Germ-line transmission and developmental regulation of a 150-kb yeast artificial chromosome containing the human beta-globin locus in transgenic mice

Sequential expression of the genes of the human beta-globin locus requires the formation of an erythroid-specific chromatin domain spanning > 200 kb. Regulation of this gene family involves both local interactions with proximal cis-acting sequences and long-range interactions with control elements upstream of the locus. To make it possible to analyze the interactions of cis-acting sequences of the human beta-globin locus in their normal spatial and sequence context, we characterized two yeast artificial chromosomes (YACs) 150 and 230 kb in size, containing the entire beta-globin locus. We have now successfully integrated the 150-kb YAC into the germ line of transgenic mice as a single unrearranged fragment that includes the locus control region, structural genes, and 30 kb of 3' flanking sequences present in the native locus. Expression of the transgenic human beta-globin locus is tissue- and developmental stage-specific and closely follows the pattern of expression of the endogenous mouse beta-globin locus. By using homology-directed recombination in yeast and methods for the purification and transfer of YACs into transgenic mice, it will now be feasible to study the physiological role of cis-acting sequences in specifying an erythroid-specific chromatin domain and directing expression of beta-globin genes during ontogeny.