Trans-activation of HIV-1 transcription by the viral regulatory protein Tat has been proposed recently to overcome a block to RNA polymerase II elongation in vivo imposed by 5'-untranslated leader sequences. Interestingly, however, only full-length transcripts, rather than prematurely terminated HIV RNAs, are synthesized in most cell-free transcription extracts. Here, we describe an in vitro system in which induction of a highly efficient RNA polymerase II termination or cotranscriptional RNA processing event creates short HIV RNAs with 3' ends that map to a region immediately downstream of the HIV-1 or HIV-2 promoters. Termination in vitro is sequence dependent, generating short HIV-1 RNAs of 58-61 nucleotides that resemble in vivo transcripts observed in the absence of Tat, and a distinct, longer transcript of approximately 125-130 nucleotides from the HIV-2 promoter. Deletion of promoter-proximal HIV-2 downstream sequences results in the loss of a discrete RNA but also fails to restore wild-type transcription, indicating that termination actually is specified at the promoter and occurs at a site positioned by one or more elements located immediately upstream of the 3' end of the short RNAs. Experiments with recombinant HIV-2 promoters and nucleoside analogs indicate that this event involves a concerted interaction between the promoter and orientation-dependent leader sequences and that RNA secondary structure formation may also be required. These data provide direct evidence for abbreviated HIV transcripts and an in vitro approach to understanding the roles of cellular and viral regulatory proteins that mediate this process at the HIV promoters.

The activity of the mouse mammary tumor virus promoter was assessed in various sequence contexts with a transient transfection assay in which promoter activity was determined by way of expression of a linked gene encoding chloramphenicol acetyltransferase, as well as by direct analysis of RNA transcripts. The results indicate that the proviral long terminal repeat contains a negative transcriptional control element in addition to the glucocorticoid-responsive transcriptional enhancer that has been described previously. The negative element is able to function in both orientations and, at least to some extent, at multiple positions with respect to the regulated transcription unit. The effects on gene expression cannot be explained by alterations in transfection efficiency. The element has been localized to a 91 base pair fragment located immediately 5' of binding sites for the glucocorticoid receptor protein that have been defined in vitro. The role of the negative element may be to repress the inherent activity of the proviral promoter in the absence of glucocorticoids, resulting in an increased ratio of gene expression in the presence and absence of hormone.

Transcription from the promoter of mouse mammary tumor virus is subject to induction by several classes of steroid hormones as well as to repression by a negative regulatory element present in the long terminal repeats of proviral DNA. In order to characterize the functional elements of the promoter that in some way must respond to these regulatory signals, a number of promoter mutations were constructed, including a set of linker-scanning mutations across the entire promoter region. Analysis of these mutated promoters with a transient-transfection assay defined at least three mutation-sensitive promoter elements that are required for both basal and hormone-induced transcription. One mutation-sensitive region contains a TATA element located at approximately position-30 with respect to the start of transcription. A second mutation-sensitive region contains two 10-base-pair direct repeats located between positions-60 and-38, within which are embedded three copies of octamer-related sequences; complete disruption of this region of the promoter leads to a more severe decrease in transcription than do any of the linker-scanning mutations, suggesting that the repeated sequences may be at least partially functionally redundant. Gel electrophoresis mobility shift assays were used to demonstrate specffic binding of a nuclear protein to this region of the promoter. A third mutationsensitive region contains a binding site for nuclear factor 1 (NF-1) located between positions-77 and-63. Sitedirected mutations in the NF-1-binding site which increase the apparent affinity of NF-1 for the promoter in vitro do not decrease the hormone dependence of transcription, suggesting that transcriptional activation mediated by steroid hormone-receptor complexes cannot be explained by facilitation or stabilization of the interaction of promoter sequences with NF-1 and consistent with the idea that binding of NF-1 is not rate determining in transcription from the mouse mammary tumor virus promoter. None of the promoter mutations functionally separates basal from glucocorticoid-induced transcription, suggesting that hormone induction does not make the promoter independent of any of the DNA-binding factors required for its basal activity.

Sets of genes under a common regulatory control in a given cell type are often differentially transcribed. The possibility that this differential transcription can be modulated by the number or strength of cis-acting regulatory sequences associated with a given gene was tested by using the glucocorticoid-responsive enhancer element associated with the mouse mammary tumor virus promoter. Results indicate that differential levels of hormone-inducible gene expression can be modulated in an additive way by the number of glucocorticoidresponsive enhancers associated with this promoter. Realization of these effects shows little preference for position of the additional elements with respect to the promoter. When sequences that bind the glucocorticoid receptor in vitro with somewhat lower affinity than the enhancer were tested, these additive effects were not detected. The results support the view that differential transcription of genes subject to a common regulatory control can be mediated, at least in part, by the number or strength of their associated cis-acting regulatory sequences.The control of gene expression can, in principle, occur at a number of levels, the most basic of which is transcription initiation. For eucaryotic genes transcribed by RNA polymerase II, the DNA sequences required in cis for this kind of control have been extensively studied in several systems (for a review, see reference 12). One class of these sequences, the enhancer elements (for a review, see reference 22), has been shown to play a role in the determination of cell specificity of viral gene transcription (5, 17) as well as in the tissue-specific expression of endogenous cellular genes (10,63). These elements appear to act as binding sites for trans-acting factors (35,47,51) that are required for the enhancer-directed stimulation of transcription from a linked promoter (57,65). An emerging view of enhancer action is that different enhancers may act independently of one another to confer different controls on a gene by two or more distinct biochemical pathways (66). An extension of this view is that differential regulation of transcription for a set of genes under a common control could be conferred by the strength or number of enhancer elements associated with each gene (13,33,52).The regulated transcription of mouse mammary tumor virus (MMTV) provides a relevant biological system in which to test the proposal that differential levels of transcription can be mediated by multiple functionally related enhancer elements of different strength or number. The glucocorticoid-regulated transcription of the proviral genes of MMTV (Fig. 1A), a retrovirus, is mediated by cis-acting sequences that have been termed the glucocorticoid response element (GRE). These sequences are specifically bound by a partially purified hormone-receptor complex in vitro (9, 16, 40-42, 48, 49) and serve to increase the rate of transcription initiation from the MMTV promoter (60) in a manner reminiscent of enhancer elements (6, 43). Functional GRE sequences have be...

The sequence organization of mouse mammary tumor virus DNA endogenous to the C57BL/6 inbred mouse strain was characterized by Southern blot analysis, utilizing probes specific for particular regions of the mouse mammary tumor virus provirus and by molecular cloning of endogenous mouse mammary tumor virus DNA. The genome of C57BL/6 mice contains three apparently intact, endogenous proviral units; two of these units comprise the Mtv-8 (unit II) and Mtv-9 (unit III) genetic loci that are also present in the DNA of BALB/c mice. The third unit is defined by EcoRI restriction fragments of 10.0 and 8.4 kilobases that contain the 5' and 3' portions of the provirus, respectively. This unit, termed unit XI and encoded by the genetic locus Mtv-17, has not been previously recognized in C57BL/6 DNA, but it can be clearly distinguished from the proviral units at Mtv-8 and Mtv-9 by Southern blot analysis under appropriate conditions. The proviral unit at Mtv-17 is not present in BALB/c DNA. DNAs comprising the entire Mtv-8 locus and the 3' portions of Mtv-9 and Mtv-17 were cloned. Analysis of the cloned DNA revealed no obvious deletions or rearrangements that would render proviral DNA defective; however, these endogenous genes are normally not transcriptionally active.