Promoter and Riboswitch Control of the Mg²⁺ Transporter MgtA from Salmonella enterica

Abstract

The MgtA protein from Salmonella enterica serovar Typhimurium mediates Mg²⁺ uptake from the periplasm into the cytoplasm. Here we report that the PhoP/PhoQ two-component regulatory system, which responds to periplasmic Mg²⁺, governs mgtA transcription initiation at all investigated Mg²⁺ concentrations and that the Mg²⁺-sensing 5′ leader region of the mgtA gene controls transcription elongation into the mgtA coding region when Salmonella is grown in media with <50 μM Mg²⁺. Overexpression of the Mg²⁺ transporter CorA, which is believed to increase cytoplasmic Mg²⁺ levels, decreased mgtA transcription in a manner dependent on a functional mgtA 5′ leader.

Expression of the Mg²⁺ transporter gene mgtA from Salmonella enterica serovar Typhimurium is regulated at both the transcription initiation and elongation steps. Transcription initiation is dependent on the two-component regulatory system PhoP/PhoQ (9), which is activated in response to low Mg²⁺ (8), acid pH (16), and antimicrobial peptides (1) sensed in the periplasm by the PhoQ protein. Transcription elongation into the mgtA coding region is controlled by the 5′ leader region of the mgtA transcript, which functions as a Mg²⁺-sensing device or riboswitch (5) and renders the transcript susceptible to degradation by RNase E (20). The mgtA leader region can adopt alternative stem-loop structures that favor or hinder transcription elongation into the mgtA coding region at low and high Mg²⁺, respectively (5). In addition, overexpression of the regulatory gene rob promotes mgtA transcription from a site located 44 nucleotides downstream of the PhoP-dependent transcription start site (2), thereby generating an mgtA transcript with a shorter leader region that could lack some of the Mg²⁺-sensing elements.

To examine the contributions that the PhoP-dependent mgtA promoter and the mgtA riboswitch make to the Mg²⁺-regulated expression of the mgtA gene, we constructed a set of four isogenic strains with alterations in the promoter and/or riboswitch regions of the chromosomal copy of the mgtA gene, as well as a lac transcriptional fusion at position 977 in the mgtA open reading frame (position 1 corresponds to the PhoP-dependent transcription start site in the wild-type strain) (Fig. 1A). All four strains are derived from wild-type strain 14028s and have a cat cassette (conferring resistance to chloramphenicol) upstream of the promoter, which does not alter mgtA expression and was used as a selectable marker when moving mutations into different genetic backgrounds. One strain—YS773—retains the wild-type PhoP-dependent promoter and mgtA 5′ leader region. The second strain—YS783—contains nucleotide substitutions in the DNA sequence corresponding to positions 151 to 160 of the mgtA 5′ leader region, where the sequence AGAUGUUUC replaced the original GUAAGACAGU, which was anticipated to interfere with the formation of stem-loop B, a structure normally formed in cells experiencing high Mg²⁺ (5). The third strain—YS802—lacks the PhoP-dependent wild-type promoter and harbors a derivative of the lac promoter—designated p_lac1-6—(12) that responds neither to PhoP nor to Mg²⁺ (5). (This strain retains the normal Mg²⁺ response of the mgtA riboswitch despite lacking the first 31 nucleotides of the mgtA leader region [5].) The fourth strain—YS812—combines the PhoP-independent p_lac1-6 promoter present in strain YS802 with the mutant mgtA 5′ leader of strain YS783. The construction of strains YS773 and YS802 has been reported elsewhere (5). Strains YS783 and YS812 were made by the one-step gene disruption method (6), using chromosomal DNA from strains YS773 and YS802, respectively, as templates to create PCR-generated DNA fragments for substitution using primers 4416 (5′-TGATTTCCCTACGCCGCTCAGGCGGGCGATGTCTTTGATAG TGTAGGCTGGAGCTGCTTC) and 4479 (5′-CCTTGCCCGATGAGCAATGTTTAAATAAAAACAGGGACGTTA TTGTGTCGAAAACATCTACACCGGTAAGACAGCAGAGG). The resulting DNA was integrated into the chromosome of strain EG9521 (8), harboring a lac transcriptional fusion in the mgtA coding region.

FIG. 1.

Distinct roles for the mgtA promoter and mgtA riboswitch in mgtA expression. (A) Schematic representation of four isogenic strains with wild-type (P_mgtA) or variant (P_lac1-6) promoters and wild-type (black line) or mutant (black line interrupted by white box) mgtA leader regions. (B) β-Galactosidase activity produced by the four strains depicted in panel A grown for 4 h in N-minimal medium with the indicated Mg²⁺ concentrations. Data correspond to the average of two experiments conducted in duplicate.

We determined the β-galactosidase activity produced by the four isogenic strains following 4 h of growth in N-minimal medium, pH 7.4 (19), supplemented with 0.1% Casamino Acids, 38 mM glycerol, and different Mg²⁺ concentrations spanning a 1,000-fold range as described previously (13). There was no β-galactosidase activity in strain YS773 following growth at 5 or 10 mM Mg²⁺, presumably because the PhoP protein is not activated under such high Mg²⁺ concentrations (8). However, the β-galactosidase levels increased as the Mg²⁺ concentration decreased with a dramatic jump between 50 and 10 μM Mg²⁺ (Fig. 1B; notice log scale of x axis), as previously reported (5, 21). In contrast, there was little mgtA transcription when strain YS802 was grown in media with >50 μM Mg²⁺, some transcription in 50 μM Mg²⁺, and maximum levels in 10 μM Mg²⁺ (Fig. 1B). The similar β-galactosidase activity produced by strain YS802 when grown in 10 mM and 100 μM Mg²⁺ reflects the Mg²⁺-insensitive nature of the p_lac1-6 promoter (5). For both YS773 and YS802, the largest difference in mgtA expression was observed following growth in 50 versus 10 μM Mg²⁺ (Fig. 1B). Because these strains differ in the promoter transcribing the mgtA gene (Fig. 1A), we hypothesized that the shared sequences in the 5′ leader region were likely responsible for the significant mgtA derepression displayed in organisms grown in 10 μM Mg²⁺. Consistent with this notion, strain YS783, with the mutant mgtA 5′ leader but wild-type mgtA promoter (Fig. 1A), produced similar β-galactosidase activity when grown in media containing 50 and 10 μM Mg²⁺ (Fig. 1B). Finally, strain YS812 made β-galactosidase at the same levels at all seven Mg²⁺ concentrations (Fig. 1B), indicating that it lacks the sequence information to modulate mgtA expression in response to changes in the levels of Mg²⁺.

Strain YS802 synthesized less β-galactosidase than the double mutant YS812 when grown in the presence of 100 μM to 10 mM Mg²⁺ (Fig. 1B), which may reflect the dampening effect exerted by the riboswitch at high Mg²⁺. This effect is highlighted also by the consistently higher levels of mgtA transcription displayed by strain YS783 than strain YS773, both of which share the PhoP-activated mgtA promoter but differ in that the latter harbors a functional riboswitch but the former does not, when bacteria were grown in the presence of 50 μM to 1 mM Mg²⁺ (Fig. 1B). Yet, at 10 μM Mg²⁺, the strains with the wild-type mgtA leader region produced higher β-galactosidase activity than their respective isogenic strains with a mutant mgtA leader (Fig. 1B). This indicates that the mgtA leader region both promotes transcription of the mgtA coding region in low Mg²⁺ and decreases its transcription in high Mg²⁺. Cumulatively, our data demonstrated that the PhoP-dependent mgtA promoter controls mgtA expression over the whole range of Mg²⁺ concentrations tested (i.e., 10 μM to 10 mM Mg²⁺) whereas the mgtA riboswitch exerts its regulatory effect primarily when bacteria experience <50 μM Mg²⁺.

Because transcription elongation into the mgtA coding region responds to cytoplasmic Mg²⁺ via the mgtA riboswitch, we reasoned that synthesis of the MgtA protein might also take place under other conditions promoting a drop in the cytoplasmic Mg²⁺ concentration. In other words, the MgtA protein might be observed earlier if Salmonella is grown in media with a Mg²⁺ concentration of <10 μM but at later times if the Mg²⁺ concentration is >10 μM, as one would anticipate that Mg²⁺ would be exhausted at earlier and later times, respectively. To test this hypothesis (and because anti-MgtA antibodies were not available), we engineered a Salmonella strain expressing a C-terminal FLAG-tagged MgtA protein from its normal promoter and chromosomal location and harboring the wild-type 5′ leader region. The FLAG epitope was introduced as described previously (22) by amplifying the cat gene from plasmid pKD3 by the PCR with the following primers: MgtA-FLAG, 2048 (5′-GTTGGTGAAAGGGTTTTACAGCAGACGTTATGGCTGGCAGGACTACAAGGACGACGATGAC AAGTAACATATGAATATCCTCCTTAG-3′) and 2049 (5′-TCGGGGATTAAGCACGCTGGCGAATCCCCGACGAA AGTGTGTGTAGGCTGGAGCTGCTTC-3′). Addition of the FLAG tag does not appear to disrupt normal MgtA protein function because the corA mgtB mgtA⁺-FLAG strain MJC116 grew as well as the isogenic corA mgtB mgtA⁺ strain EG10983 in LB medium (data not shown), whereas the corA mgtB mgtA triple mutant did not grow in LB unless supplemented with high concentrations of Mg²⁺, as described previously (11).

The MgtA-FLAG-expressing strain EG13250 was grown in 10 mM Mg²⁺ and then harvested at different times after organisms were switched to media with different Mg²⁺ concentrations. Western blot analysis was carried out with crude cell extracts prepared from the different cell cultures and developed with anti-FLAG antibodies (Fig. 2). We determined that the lower the Mg²⁺ concentration, the earlier the MgtA-FLAG protein was produced. For instance, MgtA-FLAG was detected after 2.5 h in organisms switched to 2 μM Mg²⁺, after 4 h when Salmonella was grown in 20 μM Mg²⁺, and only after 5 h in organisms experiencing 40 μM Mg²⁺ (Fig. 2). As expected, the levels of the Mg²⁺ transporter CorA, which was used as control because CorA expression is PhoP/PhoQ independent and nonresponsive to changes in the concentration of Mg²⁺ in the growth media (3), were similar at the investigated times and Mg²⁺ concentrations (Fig. 2). Thus, even though the PhoP-activated mgtA and phoP promoters are bound by the PhoP protein at the same time and transcription of the mgtA leader sequence happens concurrently with that of the phoP gene (17), production of the MgtA protein takes place >4 h later than that of the PhoP protein (Fig. 2) (17).

FIG. 2.

Synthesis of the MgtA protein is determined by the length of time and the Mg²⁺ concentration in which Salmonella is grown. Western blot analysis of crude extracts prepared from strain EG13250 coding for an MgtA-FLAG protein following growth in N-minimal medium with the indicated Mg²⁺ concentrations and incubation times. Blots were probed with both anti-FLAG (top) and anti-CorA (bottom) antibodies.

If MgtA synthesis is triggered when the cytoplasmic Mg²⁺ drops below a certain threshold, as one would expect from MgtA being regulated by a Mg²⁺-responding riboswitch, artificially increasing cytoplasmic Mg²⁺ levels may hinder synthesis of the MgtA protein. To explore this possibility, we examined the chromosomally encoded MgtA-FLAG protein levels by Western blot analysis in bacteria carrying plasmid pUC-corA, which expresses the Mg²⁺ transporter gene corA from the vector plac promoter, or the plasmid vector pUC19 (23). MgtA-FLAG could be detected in the latter but not in the former strain (Fig. 3A). (As expected, there was no reactivity in extracts prepared from the wild-type strain lacking the FLAG tag [Fig. 3A].) Likewise, when streaked onto 1% agar plates containing N-minimal media, pH 7.4, supplemented with 0.1% Casamino Acids, 38 mM glycerol, 10 μM MgCl₂, and the chromogenic LacZ substrate X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside; 60 μg/ml; Gold Biotechnology, Inc.), strain EG9170, harboring a chromosomal mgtA-lac transcriptional fusion, formed blue colonies when containing pUC19 but white colonies when carrying pUC-corA (Fig. 3B).

FIG. 3.

Overexpression of the Mg²⁺ transporter gene corA turns off mgtA transcription in an mgtA riboswitch-dependent manner. (A) Western blot analysis of crude extracts prepared from strains 14028s (wild type) or EG13250 (mgtA⁺-FLAG) harboring plasmid pUC-corA or the plasmid vector pUC19. (B) Lac phenotype of isogenic strains with lac transcriptional fusions at positions 6 (EG9220) and 2190 (EG9170) with respect to the PhoP-dependent transcription start site for mgtA or at position 977 and with the 100-bp sequence corresponding to positions 148 to 247 in the wild-type mgtA 5′ leader replaced by the 84-bp “scar” sequence (6) (EG17425) and harboring plasmid pUC-corA or the plasmid vector pUC19.

The silencing effect of the corA-expressing plasmid is exerted on the mgtA riboswitch because the pUC-corA plasmid had no effect on the expression of strain EG9220 harboring a lac fusion at the sixth nucleotide of the mgtA 5′ leader region and thus lacking the mgtA riboswitch (Fig. 3B), but could silence mgtA expression in strain EG9170 with a lac fusion at position 2190. Moreover, mgtA expression in strain EG17425, with a defective mgtA riboswitch due to replacement of the 100-bp sequence corresponding to positions 148 to 247 in the wild-type mgtA 5′ leader (5) by the 84-bp “scar” sequence (6), was refractory to repression by the pUC-corA plasmid (Fig. 3B).

In sum, we have established that production of the Mg²⁺ transporter MgtA is ultimately governed by the Mg²⁺ levels in the cytoplasm, which are sensed by the mgtA riboswitch. Expression of the mgtA gene is also controlled at the transcription initiation step via the two-component system PhoP/PhoQ responding to periplasmic Mg²⁺. Because the PhoP/PhoQ system is a major regulator of virulence functions (9) and the mgtA gene is not required for pathogenicity, the mgtA riboswitch may enable Salmonella to produce the MgtA protein only when the cytoplasmic Mg²⁺ concentration falls below a certain level and thus differentially from other gene products belonging to the PhoP regulon. Indeed, there is no mgtA expression when Salmonella experiences acid pH (4) even though the PhoP/PhoQ system is activated by acid pH (16). The mgtA expression behavior is in contrast to that of the Fe²⁺ transporter gene feoB, which is turned on via the PhoP-activated RstA protein when Salmonella faces acid pH but not in response to low Mg²⁺ (4). As rstA transcription is also induced in low Mg²⁺ (14, 20, 24), these findings suggest that acid pH is necessary for activation of the RstA protein.

Finally, there is increasing evidence suggesting the intriguing possibility that signals other than Mg²⁺ may act on the mgtA leader region to promote transcription elongation into the mgtA coding region. First, mutations in the mgtA leader region resulting in heightened mgtA expression enhanced the thermotolerance of Salmonella experiencing high osmolarity (15). Second, rob overexpression confers resistance to cyclohexane in an mgtA-dependent manner (2). And third, inactivation of the RNA chaperone Hfq promoted mgtA expression (7, 18) while decreasing the mRNA levels of other PhoP-activated genes (18). (The mgtA gene appears to be regulated in a different manner in Escherichia coli because hfq inactivation led to increased mgtA transcription [10].) Moreover, the dual control of mgtA expression at the transcription initiation and elongation steps is unusual for genes regulated by riboswitches, as they are typically transcribed from constitutive promoters. Therefore, we hypothesize that metabolic signals, environmental or cellular cues, and/or regulatory trans-acting factors may act on the mgtA leader region so that mgtA transcription initiated at intermediate Mg²⁺ concentrations in a PhoP-dependent manner (5) (Fig. 1B) can continue into the mgtA coding region, resulting in the synthesis of the MgtA protein even if the Mg²⁺ concentration in the cytoplasm is relatively high.