No abstract

Genes can be regulated by the interaction of proteins with specific sequences in DNA (1). Proteins called repressors specifically turn off transcription, and positive regulatory proteins enhance specific transcription. In this article we describe the complete sequence of two control regions in the DNA of a bacteriophage. We show how interaction of these sequences with a regulatory protein mediates intricate patterns of gene regulation. In particular, we show that one of these sequences is arranged so that a single protein can function both as a positive and a negative regulator. Moreover, we argue that this same control region may contain information important for posttranscriptional control.

We have cloned and characterized a fragment of the Escherichia coli chromosome spanning ginA, the structural gene for glutamine synthetase [L-glutamate:ammonia ligase (ADPforming), EC 6.3.1.2]. The fragment also carries ginG, whose product is necessary for regulation of ginA expression, and a previously unidentified gene whose function we have not discovered. Transcription of ginA and the newly identified gene occurs divergently from a region between the two genes. Transcription ofginA proceeds toward ginG, which is transcribed in the same direction. A region of DNA between ginA and ginG contains genetic information whose loss may result in the inability to reduce expression of ginA and other operons in response to ammonia (the GInC phenotype).Glutamine synthetase [L-glutamate:ammonia ligase (ADPforming), EC 6.3.1.2], the product of the glnA gene, plays a central role in the assimilation of ammonia in enteric bacteria. Cellular glutamine synthetase activity is regulated in response to the quality and abundance of the nitrogen source in the growth medium both at the level ofenzymatic activity (by covalent modification-adenylylation) and at the biosynthetic level (for review, see ref. 1). Moreover, at least some of the mechanisms that regulate the biosynthesis of glutamine synthetase also regulate the expression of other operons and genes whose products are responsible for the utilization of nitrogenous compounds. One such operon, hut, produces the conveniently assayed enzyme histidase (2). We have cloned the glnA region of the Escherichia coli chromosome. This region also includes the ginG gene, whose product is involved in the regulation of glutamine synthetase biosynthesis (3, 4), and a third gene, whose role in nitrogen assimilation is unclear. We have determined the direction oftranscription ofeach ofthese genes, and we have identified their products in minicells. MATERIALS AND METHODS

An abbreviated procedure has been developed for the purification of restriction endonucleases. This procedure uses chromatography on phosphocellulose and hydroxylapatite and results in enzymes of sufficient purity to permit their use in the sequencing, molecular cloning, and physical mapping of DNA.

XKH54h8o; in some cases, both se~tions were performed simultaneously. The immunity of plasmid containing strains was further tested by streaking single colonies across a streak of phage on an agar plate. Colonies of transformants which constitiltively synthesized f3-galactosidase appeared blue on agar plates containing a chromogenic, noninducing substrate, 5-chloro-4-bromo-3-indolyl-,B-D-galactoside (40 Ag/ml). Assays of X repressor (12), fl-galactosidase (f3-D-galactoside galactohydrolase, EC 3.2.1.23) (13), and the isolation of operator containing DNA fragments on nitrocellulose filters (14) have been described previously. Purified X repressor was a gift of R. Sauer and purified lac repressor was a gift of A. Maxam. Experiments were carried out in a P1 (EK1) facility. RESULTS Preliminary considerationsWe wished to proceed in two steps; first, to clone cI flanked by as little extraneous phage DNA as possible on a plasmid and then to insert a DNA fragment bearing the lac promoter near the beginning of cI. Two problems arose. No single restriction endonuclease cleaves X DNA just outside the ends of cI without also cleaving within it, and no restriction endonuclease site suitable for inserting the lac promoter exists in X DNA near the beginning of cI. We adopted a strategy based on the following considerations: gene cI can be neatly isolated on two DNA fragments, one of which bears two HindIII ends and the other of which bears one HindIII end and one HaeIII end (see Fig. 1). Proper joining of these fragments reconstitutes cI on a larger fragment bearing one HindIII end and one HaeIII end. HindIII ends are staggered and readily anneal to each other, whereas HaeIII ends are flush. Staggered ends which anneal to each other can be joined by T4 polynucleotide ligase (15), and flush ends can be joined to other flush ends by that enzyme (9). To clone our cI fragment, we sought a plasmid which could be opened so as to produce one HindIII end and one flush end. The plasmids we used, pCR11 and pMB9, each have a single HindIII site and a single EcoRI site. Although EcoRI produces staggered ends, we anticipated that these ends could be converted to flush ends by treatment with DNA polymerase I and the four deoxyribonucleotide triphosphates, because the recessed 3' end can be extended by copying the protruding 5' end of the complementary strand (a process we hereafter refer to as filling-in). Precise joining of a filled-in EcoRI end to a HaeIII end should produce a molecule recognized by EcoRI at the junction (see Fig. 2). This regenerated EcoRI site near the beginning of cI could then be used as a site to insert the lac promoter.