INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The serovar Typhimurium of Salmonella enterica is a facultative intracellular parasite that can cause gastroenteritis and enteric fevers, depending on the particular host-parasite combination. One virulence determinant important for Salmonella's survival in (and exploitation of) the host intestine is the type III secretion system 1 (TTSS-1) (19, 31, 32, 34). Electron micrographs reveal that the type III secretion apparatus resembles the flagellar basal body (22). The appearance of the apparatus has also been likened to a syringe with a long, presumably hollow, needle passing from inside the basal structure and out into the supernatant. The effectors secreted by TTSS-1 contribute to a diverse array of in vitro phenotypes and are required for virulence in both the mouse model of typhoid fever and the bovine model of gastroenteritis (10, 42).

The more than 15 proteins that are thought to comprise the type III secretion apparatus of TTSS-1 are all encoded by genes of the Salmonella pathogenicity island 1 (SPI1) (Fig. 1A). The core of the complex, probably found embedded in the inner membrane and crossing the periplasmic space, is comprised of PrgH and PrgK, while the barrel of the syringe is predominantly PrgI, a protein that requires PrgJ for its stability (20). The genes encoding these proteins are in an operon (prgHIJK-orgABC) at one end of SPI1 (20-23) (Fig. 1A). All type III secretion systems include a secretin family outer membrane protein, possibly forming a channel through which the needle of the type III secretion apparatus passes. In the case of TTSS-1, the protein InvG probably serves this purpose (7, 8). InvG is also encoded by SPI1, but at the opposite end of the more than 40-kb pathogenicity island (Fig. 1A). InvG is encoded by an operon with 15 other genes: invFGEABC-spaMNOPQRS-sicA-sipBCDA-iacP-sicP-sptP (reviewed in reference 10) (C. A. Lee, unpublished data). With the exception of InvF and InvB, the other Inv and Spa proteins are probably directly involved in secretion (reviewed in reference 10).

View larger version (27K): [in this window] [in a new window]   FIG. 1.   HilA-dependent regulation of the TTSS-1. (A) A map of SPI1. Each gene is depicted, approximately to scale, as an arrow, with the direction of the arrow indicating the direction of transcription. The shading of each gene reflects the putative or demonstrated function(s) of its protein product. The functions of the proteins encoded by the unshaded genes are unknown. The part of the island containing sitABCD is not included in this depiction. The sit operon maps to the left of avrA. (B) HilA-dependent cascade of transcriptional activation. Under conditions that do not favor TTSS-1 gene expression, the hilA promoter is repressed by a putative regulatory protein (R). When environmental conditions become favorable for TTSS-1 gene expression, HilD derepresses the hilA promoter (14, 15, 37, 40, 41). HilA activates PinvF-1 (formerly PinvF) and PprgH and, thus, the expression of TTSS-1 structural genes (26 and this work). InvF, encoded by the first gene in the PinvF-1 transcript, interacts with SicA and activates the expression of effectors on and off SPI1 (9, 11, 12). Under inducing conditions, HilD also activates a second promoter far upstream of the invF translation start site, termed PinvF-2 (dotted arrow). PinvF-2 makes a minor contribution to invF production under inducing conditions in vitro (S. Akbar, unpublished data).

The production of TTSS-1 proteins is not constitutive in vitro and is probably regulated in vivo as well. In vitro inducing conditions that result in the optimal expression of TTSS-1 phenotypes include high osmolarity (10 g of NaCl/liter), low aeration, and slightly basic pH (2, 27). Transcriptional regulation of TTSS-1 genes is a primary mechanism for controlling the production of TTSS-1 factors in response to environmental and physiological cues (2, 10, 27, 28). The expression of SPI1 operons encoding components of the TTSS-1 apparatus and some of its effectors requires HilA, a transcription factor encoded by SPI1 (Fig. 1B) (1, 26).

HilA is an OmpR/ToxR family transcriptional regulator by virtue of the DNA binding and activation domain in its N terminus (1, 26). Consensus binding sites for OmpR/ToxR proteins have been difficult to agree upon, but the emerging trend is that they usually contain degenerate direct repeats (reviewed in reference 29). The OmpR DNA binding domain probably interacts with DNA as a dimer such that each monomer faces the same direction (18). Members of this family, however, can have binding sites that are more difficult to recognize as direct repeats, and they often make large footprints on DNA. Furthermore, in some cases they seem to require sites at different promoters that are completely unrelated in sequence (see, for example, references 25 and 35). Finally, many family members are posttranslationally modified or membrane associated, complicating biochemical analyses considerably (29).

HilA activates transcription at P_invF-1 directly through binding to an element termed the HilA box (Fig. 1B) (26). The HilA box is a 17-nucleotide sequence comprised of two tandemly arranged hexamers (a direct repeat) separated by 5 nucleotides. P_invF-1 (57 to +10), which contains only 3 nucleotides upstream of the HilA box, is sufficient for HilA-dependent activation at P_invF-1. Deletion of sequences upstream of 40 results in a complete loss of activation by HilA. Furthermore, two artificial promoters containing P_invF-1 (57 to 30) sequences fused in frame to unrelated (29 to +10) sequences are activated by HilA. Finally, double-stranded DNAs containing the HilA box bind to membrane-associated HilA in vitro. Thus, the HilA box in P_invF-1 is both necessary and sufficient for activation by HilA. Two additional hexamers similar in sequence to a half-site of the HilA box are found at 75 (TTTAAT) and 6 (TTACAT) in P_invF-1 (Fig. 2A). P_invF-1 (78 to +10) is twofold more activated than a minimal P_invF-1 encoding only the HilA box through the +10, suggesting that the half-site-like hexamer at 75 may contribute to HilA-dependent transcription even though it is not required. In contrast, the half-site-like hexamer at 6 appears to be fortuitous and is not involved in HilA-dependent activation (26 and unpublished data).

View larger version (24K): [in this window] [in a new window]   FIG. 2.   The cis elements in PprgH required for activation by HilA. (A) PinvF-1 and PprgH each contain a HilA box at the same position relative to the +1 of transcription. The +1 accompanied by a bent arrow above the line indicates the transcription start site of each promoter. The arrows inside the rectangle between 53 and 37 show the HilA box, composed of a hexameric direct repeat (consensus TTTCAT) separated by 5 nucleotides. Two copies of a half-site-like hexamer are found in each promoter outside the HilA box, as indicated by the additional arrows below each line. (B) The HilA box is necessary, but not sufficient, for HilA-dependent activation of PprgH. PCR products or oligonucleotides carrying WT and mutant PprgH sequences were cloned into the pAH125oriT reporter system and then placed in the E. coli chromosome in single copy (see Materials and Methods). Strains were transformed with either pACYC177 or pVV214 (HilA+) and then grown to early log phase under oxygenating conditions on a culture wheel at 37°C. Values represent the average of at least two assays performed on three independent transformants of each reporter strain and were not necessarily performed on the same day. Fold activation was calculated as the ratio of units expressed in a pVV214-containing strain to that expressed in a pACYC177-containing strain and were calculated not from the average units reported in the Miller units columns but from raw data. All standard deviations are equal to or less than 15% of the reported values.

A second HilA-activated promoter, P_prgH, also has a HilA box at the same position relative to the start (+1) of transcription. We have hypothesized that HilA activates P_prgH through a direct interaction with its HilA box. Like P_invF-1, P_prgH has two half-site-like hexamers outside the HilA box, but in this case they are located at 106 (TTTAAT) and 67 (TTTTAT) (Fig. 2A). In this paper, we report our investigations of the HilA box and the upstream half-site-like hexamers in P_prgH. We have found that the HilA box is required for activation of P_prgH. By using an in vivo binding assay, we found that the HilA box of P_prgH is also likely to be the site of a direct interaction between HilA and P_prgH DNA. Nevertheless, the HilA box is not sufficient for activation of P_prgH. Curiously, our results indicate that sequences between the HilA box and the +1 of transcription affect the ability of the HilA box from P_prgH to serve as an upstream activating site in the absence of distal P_prgH sequences. In contrast, the HilA box from P_invF-1 is a strong upstream activating site in every promoter we have tested, irrespective of the sequences between the HilA box and +1. Although we cannot detect differential regulation of P_invF-1 and P_prgH in vitro, we speculate that the two promoters could be regulated independently in vivo.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and growth of bacteria. S. enterica serovar Typhimurium and Escherichia coli strains used in this study are listed in Table 1 (4, 5). Bacterial cultures were grown at 37°C in Luria-Bertani (LB) medium comprised of 0.5% Bacto-yeast extract, 1% Bacto-tryptone, and 1% NaCl. When appropriate, the medium was supplemented with antibiotics as follows: 100 to 200 µg of ampicillin/ml and 10 to 25 µg of kanamycin/ml. hilA-mychis expression was induced from pCH112 by adding 20% arabinose to a final concentration of 0.2% to early log cultures. -Galactosidase assays were performed on Salmonella cultures grown under activating (limited oxygen, high salt) conditions as previously described (24) and on mid-log E. coli cultures grown with high aeration on a culture wheel; the -galactosidase activities were then quantified by the Miller method (30).

Crossover PCR. Crossover PCR was performed using a modified protocol from the thesis of Dereth Phillips from George Church's lab (D. Phillips and G. Church, personal communication; http://arep.med.harvard.edu/labgc/pko3_pcr.html). Two separate 50-µl reaction mixtures were used to amplify overlapping fragments of P_prgH while at the same time introducing point mutations into the amplicon. The reaction mixtures contained 20 pmol of each primer (e.g., ebe89 and ebe91 or ebe90 and ebe96), 10 µl of 10× Extaq buffer, 4 µl of a deoxynucleoside triphosphate stock containing each nucleoside triphosphate at 2 mM (TaKaRa), 1 µl of genomic template DNA, exTaq polymerase (TaKaRa), and water. After these reactions, the flanks were crossed over by taking 1 µl from each of these PCRs and using them together as template in a second round of PCR, this time using ebe46 and ebe47 to add EcoRI and PstI sites, respectively, to clone the resulting product into pAH125oriT, as described previously (26).

Construction of strains. Plasmid DNA was electroporated into either S. enterica serovar Typhimurium or E. coli by standard methods. Plasmids transferred from E. coli into S. enterica serovar Typhimurium were first passaged through the restriction-negative modification-positive strain listed in Table 1 (4).

pAH125oriT reporter construction. To construct pAH125oriT reporter strains, oligonucleotides or PCR products (Table 2) were digested with EcoRI and PstI and ligated into pAH125oriT vector that had also been digested with EcoRI and PstI. Ligation mixes were transformed into E. coli BW25142, a strain with a pir-116 mutation that allows the plasmid to be maintained at high copy, and were plated on kanamycin (25 µg/ml) (17). Transformants containing proper inserts were identified by colony PCR using primers ebe20 and ebe22 (24), which flank the multiple cloning site in pAH125oriT. DNA was isolated from candidates using Qiagen miniprep columns or a Qiagen Biorobot, and then that DNA was sequenced by the Microbiology Core Facility to confirm the presence of correct inserts. Correct candidates were then electroporated into E. coli BW21355 (lac mutant) that already contained the plasmid pINT in order to construct chromosomal integrants and were plated on kanamycin (10 µg/ml) (17). Colonies were patched to identify clones that were ampicillin sensitive and so had lost pINT, and the presence of reporters in single copy was confirmed using primers P1, P2, P3, and P4 (17). Reporter strains with a single insertion were then transformed with either pACYC177, pVV214, pBAD vector, or pCH112 as indicated in Results.

Western blotting. Polyclonal antiserum to HilA was raised by injecting rabbits with His-HilA purified from E. coli in the presence of sodium dodecyl sulfate (L. M. Schechter and C. A. Lee, unpublished data). The serum was diluted 1:500 and used as primary antibody to bind to HilA that had been transferred to polyvinylidene difluoride membranes following sodium dodecyl sulfate-polyacrylamide gel electrophoresis through a 10% gel, as described previously (26). Horseradish peroxidase-conjugated anti-rabbit antibody (Sigma) was used as the secondary antibody, and a chemiluminescent detection system was used to detect antibody-protein complexes using X-ray film in a standard protocol that was described elsewhere (38).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

The HilA box is not sufficient for HilA-dependent activation of P_prgH. Based on the results with P_invF-1, we hypothesized that the HilA box at P_prgH would be necessary and sufficient for activation by HilA. We first cloned three 5' truncations of P_prgH into a lacZ reporter system in E. coli (PH17, PH27, and PH15) (Fig. 2B). The shortest is in PH15 (P_prgH, 57 to +8), which contains 3 bp upstream of the HilA box through 8 bp downstream of the start of transcription. The reporters were integrated in single copy in the E. coli chromosome. We then tested whether the reporters are responsive to HilA by transforming the reporter strains with pVV214 (hilA⁺) or pACYC177 (vector). We hypothesized that all three promoters would be activated by HilA, as is the case of an analogous series of P_invF-1 reporters (26). As predicted, full-length P_prgH in PH17 is strongly activated by HilA (by about 11-fold) (Fig. 2B). The removal of sequences upstream of the hexamer at 106 has little effect on the fold activation by HilA (compare PH17 to PH27). In contrast, PH15 is barely activated by HilA (1.5-fold compared to 9- to 11-fold for the longer constructs). The simplest interpretation of these results is that the HilA box of P_prgH is not sufficient to activate P_prgH. This finding is dramatically different from our analysis of P_invF-1.

The HilA box is required for HilA-dependent activation of P_prgH. Because the HilA box was, unexpectedly, not sufficient for activation of P_prgH, we tested whether the HilA box at P_prgH is required for HilA-dependent activation. To do so we took advantage of knowledge gained from our previous work with P_invF-1. In full-length P_invF-1, mutations T45C, T42C, and T37C in the HilA box result in promoters that are severely reduced in HilA-dependent activation (26). Thus, we constructed three full-length P_prgH reporters with one of these three point mutations in the HilA box (Fig. 2B). At P_prgH, these point mutations dramatically reduce activation by HilA, down from 11-fold to 2.0-, 1.7-, and 1.2-fold, respectively (see Fig. 2B, PH30, PH31, and PH32). From these results we conclude that the HilA box is required for activation of P_prgH by HilA.

Mutations in the half-site-like hexamers upstream of the HilA box reduce HilA-dependent activation of P_prgH. To test whether the half-site-like hexamers at 106 and 67 are required for HilA-dependent activation, we tested three full-length P_prgH promoters with mutations in one or both of these sequences (Fig. 2B). PH28, which has a TT-to-GG mutation in the first nucleotides of the hexamer at 106, loses half of its ability to respond to HilA, while PH29, which has a TT-to-GG mutation in the first nucleotides of the hexamer at 67, is even more severely compromised (Fig. 2B). PH39 has all of these mutations and is no longer activated by HilA. In fact, its phenotype is very similar to that of PH15 (P_prgH, 57 to +8) This result strongly suggests that the phenotype of PH15 is owing to the absence of upstream sequences that include the initial T's in each of the half-site-like hexamers.

HilA can be converted to a repressor by interrupting the ⁷⁰ recognition hexamers with a HilA box. The HilA box is necessary, but not sufficient, for activation of P_prgH. This led us to hypothesize that HilA may bind less well to the HilA box in P_prgH than it does to the HilA box in P_invF-1. We developed an in vivo binding assay to test this hypothesis. Genomic studies of E. coli K-12 have demonstrated that activator binding sites usually do not occur downstream of 30 (16). Moreover, forcing an activator to bind downstream of 30 can repress that promoter's activity (6, 13, 36, 43). In most cases, it appears that repression results from inhibiting the ability of RNA polymerase (RNAP) to bind the promoter by simple steric hindrance. To study HilA binding in vivo, we cloned the HilA box from P_invF-1 between consensus 35 and 10 hexamers that direct ⁷⁰-dependent transcription in E. coli. We cloned this construct into the same lacZ reporter system we used throughout this work, resulting in the construct designated -35INVWT-10 (Fig. 3A).

View larger version (22K): [in this window] [in a new window]   FIG. 3.   HilA behaves as a repressor when the HilA box from invF is cloned between the 35 and 10 hexamers of a 70-dependent promoter (-35INVWT-10). (A) Diagram of -35INVWT-10. Using long oligonucleotides, the HilA box from PinvF-1 was artificially placed between consensus 35 and 10 hexamers that direct 70-dependent transcription in E. coli, forming the promoter -35INVWT-10. This promoter was cloned into the pAH125oriT lacZ reporter system and then integrated in single copy in the E. coli chromosome. (B) HilA represses -35INVWT-10. -35INVWT-10 reporter strains containing either pBAD vector or pCH112 (HilA-MycHis+) were diluted 1:200 at time zero into LB containing 0.2% arabinose (see Materials and Methods). The Miller units produced by two independent transformants of each strain were determined for each time point, and a single representative experiment is shown. Strains grown in the absence of arabinose had approximately the same activity as the pBAD plus arabinose result shown (black boxes).

View larger version (32K): [in this window] [in a new window]   FIG. 4.   The HilA box in PprgH is a functional HilA binding site. Strains were constructed and grown as described in the legend to Fig. 3. The nucleotides separating the 35 and 10 hexamers in the reporter encode a WT HilA box from PinvF-1, one of three point mutations in the HilA box from PinvF-1, or a WT copy of the HilA box from PprgH, as indicated in the legend. The mutations are numbered as they are when the HilA box is at its native position; e.g., T37C is a mutation from T to C at the final T in the HilA box of PinvF-1 (see Fig. 5B for the sequence). Fold repression was calculated by growing the reporter strains, containing either pBAD vector or pCH112 (HilA-MycHis+) in the presence of arabinose, as for Fig. 3, and then dividing the units produced by a pBAD strain by those produced by a pCH112 strain. Values come from a single experiment on three independent colonies of each type. The standard deviations are indicated as error bars.

The prgH HilA box binds to HilA. To test if HilA binds to the HilA box sequence in P_prgH directly, we cloned the prgH HilA box into a repressor-reporter, -35PRGWT-10. As shown in Fig. 4, this reporter is strongly repressed by HilA. Using these results, we infer that the HilA box of P_prgH is a functional HilA binding site. It is even possible that the HilA box at P_prgH is a better HilA binding site than the HilA box at P_invF-1.

The HilA box from P_prgH is a functional cis determinant for HilA-dependent activation. We tested whether the HilA box from P_prgH can serve as an upstream activating site by creating heterologous hybrid promoters and testing their response to HilA (Fig. 5A). PH24 has the HilA box and the 35 region from P_invF-1 but has the 29 to +8 sequences from P_prgH. It is activated to the same extent as PH7, which contains P_invF-1 (57 to +10) (Fig. 3A, top line). This result indicates that the HilA box and the 35 region from P_invF-1 is sufficient to activate P_prgH basal (29 to +8) sequences (Fig. 5A). The reciprocal prg-inv hybrid in PH25 is also activated by HilA, demonstrating the HilA box and the 35 region from P_prgH can also function as a HilA-dependent activating region (Fig. 5A).

View larger version (25K): [in this window] [in a new window]   FIG. 5.   The HilA box from PprgH is a functional cis element. (A) PprgH (57 to 30) can confer HilA-dependent activation upon PinvF-1 (29 to +10). Double-stranded oligonucleotides carrying the hybrid promoters were cloned into the pAH125oriT reporter system and then placed in the E. coli chromosome in single copy (see Materials and Methods). Strains were transformed with either pACYC177 or pVV214 (HilA+) and then grown to early log phase under oxygenating conditions on a culture wheel at 37°C. Fold activation is calculated as the ratio of units expressed in a pVV214-containing strain to that expressed in a pACYC177-containing strain. The value for PH7 is as reported in reference 26. All standard deviations are equal to or less than 15% of the reported values. (B) PprgH and PinvF-1 sequences are most dissimilar in the region between 29 and +1. The invF and prgH promoter sequences were aligned at their respective +1 start sites for transcription. Identical nucleotides are connected by a vertical line, and the 10 hexamers, 35 hexamers, and HilA boxes are indicated. T's at positions 37, 42, and 45 are indicated with vertical arrows.

P_prgH and P_invF-1 respond equally to HilA in E. coli and S. enterica serovar Typhimurium under in vitro culture conditions. The cis requirements for HilA-dependent activation of P_prgH and P_invF-1 are somewhat different, and hence the two promoters may respond differently to various amounts of HilA. We investigated this possibility in two ways. First, we examined the activity of P_invF-1 (153 to +129) and P_prgH (254 to +8) in the presence of large or small amounts of HilA in E. coli. Small amounts of HilA expressed from uninduced pCH112 (HilA-MycHis⁺) activate full-length P_invF-1 2.9-fold and activate full-length P_prgH 3.5-fold (Table 3). To test the promoters' responses to somewhat higher amounts of HilA, we expressed HilA from pVV214. Estimating from a Western blot, pVV214 produces two to three times more HilA than uninduced pCH112 (data not shown), and HilA expressed from pVV214 activates P_invF-1 7.5-fold and P_prgH 11-fold (Table 3). So they respond about equally to both low and intermediate levels of HilA. To express very high levels of HilA, we induced pCH112 with 0.2% arabinose. After 2.5 h of induction, these populations of cells contain approximately 10 times more HilA than pVV214-expressing cells (data not shown), and P_invF-1 expression was activated 210-fold and P_prgH was activated 237-fold. Thus, it appears that, at least in E. coli, P_prgH and P_invF-1 have similar responses over a range of HilA levels.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We have observed that the HilA boxes in P_invF-1 and P_prgH are required for activation at each promoter and that HilA binds to each of them (Fig. 2 through 4 as well as reference 26). The HilA boxes of P_invF-1 and P_prgH are different from one another at only 5 of 17 positions and are in identical positions relative to the +1 of transcription at each promoter. The simplest model for activation by HilA is that the HilA box is a critical upstream activating sequence because it serves as the primary HilA binding site. Because positions 37, 42, and 45 are critical for binding and activation from both HilA boxes, we might assume that HilA makes the same base pair-specific contacts with the HilA box from P_prgH and P_invF-1. By extension, we would simplistically hypothesize that the same protein-protein interaction(s) between HilA bound to the HilA box and RNAP serve(s) to activate transcription at P_invF-1 and P_prgH.

But HilA-dependent activation may not be so straightforward. Our results demonstrate that the HilA box and 35 regions of P_invF-1 and P_prgH have different capacities to serve as upstream activating sites in the context of other sequences both upstream and downstream of the HilA box. Strikingly, sequences downstream of the HilA box in P_prgH appear to inhibit the prgH HilA box's activity in the absence of upstream sequences, including the half-site-like hexamers at 106 and 67 (Fig. 2). Changing the (29 to +8) sequences in P_prgH (57 to +8) to P_invF-1 (29 to +10) allows the prgH HilA box to serve as a HilA-dependent upstream activating site, even in the absence of the half-site-like hexamers (Fig. 5). In contrast, the P_invF-1 HilA box serves as a good upstream activating region when present in many (57 to +8 [or +10]) promoters with different (29 to +8 [or +10]) sequences (Fig. 5 and reference 26). In seeking to explain these results, we hypothesize that several molecular interactions influence the outcome of HilA bound to any given HilA box found between 53 and 37 at a promoter. We first discuss the role(s) the sequences upstream of the HilA box in P_prgH might play in order to relieve poisoning by the (29 to +8) sequences. Then we discuss how the sequences between 29 and +1 could affect transcriptional activation.

Sequences upstream of the HilA box in P_prgH. HilA activates P_prgH very efficiently in the presence of upstream sequences that include the half-site-like hexamers at 106 and 67. The initiation of transcription is a complex process, so these sequences could be important for any of several reasons. What complicates the analysis of P_prgH and P_invF-1 considerably is the fact that no matter what the molecular mechanism by which the half-site-like hexamers at 106 and 67 allow for activation of P_prgH, they are required only in the presence of particular 29 to +1 sequences, so that they are dispensable at prg-inv but not at P_prgH (57 to +8). In contrast, P_invF-1 (57 to +10), inv-phoE, inv-PRM, and inv-prg all have HilA box sequences from P_invF-1 and are activated strongly by HilA, regardless of the sequences between 29 and +1, and in the absence of any upstream half-site-like hexamers (Fig. 5 and reference 26).

The sequences downstream of the HilA box. We determined that the differences in the activities of P_invF-1 (57 to +10) and P_prgH (57 to +8) cannot be attributed to the half-site-like hexamer at 6 in P_invF-1. So some other sequences between (29 and +8 [or +10]) must account for their different activities. The 10 hexamers play a large role in determining the intrinsic properties of ⁷⁰-dependent promoters, and the 10 hexamers of P_invF-1 and P_prgH are different from one another at two of six positions (P_invF-1, TATTGT; P_prgH, TAATCT; nonidentical positions are underlined). So we hypothesize that the 10 hexamers determine whether P_prgH (57 to 30) is able to serve as an upstream activating site in the absence of distal half-site-like hexamers. This hypothesis could be tested directly by systematically mutating every position of the 10 hexamers in P_prgH and P_invF-1.