Elongation factor P controls translation of the mgtA gene encoding a Mg2+ transporter during Salmonella infection

Abstract Ribosome often stalls on mRNA sequences harboring consecutive proline codons. Elongation factor P (EF‐P) is required for the stalled ribosome to continue translation and thus the absence of EF‐P affects translation of the associated open reading frame. Here we report that EF‐P controls translation of the mgtA gene encoding a Mg2+‐transporting ATPase from the intracellualr pathogen Salmonella enterica serovar Typhimurium. EF‐P's effect on mgtA translation is dependent on the 550th and 551st proline codons in the coding region and thus substitution of those proline codons eliminates EF‐P‐mediated control of MgtA protein without affecting the Mg2+‐transporting activity of the mgtA gene. The Pro550 and Pro551‐substituted mgtA gene promotes Salmonella's intramacrophage survival and mouse virulence, suggesting that EF‐P‐mediated translational control of the mgtA gene is required for Salmonella pathogenesis.


| INTRODUC TI ON
If mRNA sequences harbor two or more consecutive proline codons, a ribosome often stalls at those consecutive proline codons (Doerfel et al., 2013;Peil et al., 2013;Tanner, Cariello, Woolstenhulme, Broadbent, & Buskirk, 2009;Woolstenhulme et al., 2013) due to a steric constraint imposed upon the peptide bond formation between preexisting peptidyl-prolyl tRNA and incoming prolyl-tRNA (Blaha, Stanley, & Steitz, 2009;Pavlov et al., 2009;Wohlgemuth, Brenner, Beringer, & Rodnina, 2008). EF-P is a protein factor that is specifically required for the stalled ribosome to continue translation (Doerfel et al., 2013;Lassak, Wilson, & Jung, 2016;Ude et al., 2013). EF-P promotes the peptidyl transferase activity of the ribosome by binding to E-site within the ribosome and stabilizing the CCA end of the P-site tRNA (Blaha et al., 2009;Huter et al., 2017). As expected from EF-P's role in the elongation step of translation, a strain lacking EF-P induces ribosome stalling on polyproline mRNA sequences and therefore it has an impact on expression of many genes harboring the polyproline motifs in either the coding region or the short open reading frame (ORF) within the leader RNA. If polyproline motifs are located in the coding region of a target gene, protein levels of the gene are expected to be lower in the strain lacking EF-P than in the strain harboring EF-P. A different scenario is also possible in bacteria if polyproline codons are located at a short ORF within the leader RNA. In this case, lack of EF-P induces ribosome stalling at the short ORF and uncouples between translation of the short ORF and transcription of the leader RNA, which affects the formation of an attenuator stem-loop structure. Depending on whether ribosome stalling promotes or inhibits the formation of the attenuator stem-loop, it controls transcription elongation into the downstream genes and affects mRNA levels of those target genes. In the mgtCBR virulence operon from the intracellular pathogen Salmonella enterica serovar Typhimurium, EF-P controls both transcription of the mgtCBR operon (Nam, Choi, Shin, & Lee, 2016) and translation of the mgtB gene encoding the MgtB Mg 2+ transporter located within the mgtCBR operon . At the level of transcription, EF-P controls transcription elongation into the mgtCBR operon via three consecutive proline codons at mgtP ORF in the leader RNA.
Removal of EF-P affects ribosome stalling at mgtP and thus increases mRNA levels of the mgtC and mgtB genes Nam et al., 2016). And substitution of the consecutive proline codons at mgtP eliminates the EF-P-mediated increase in mRNA levels of the mgtC and mgtB genes Nam et al., 2016). In addition to EF-P's role in transcriptional control, EF-P also controls translation of the mgtB gene via two consecutive proline codons located in the mgtB coding region .
In this case, removal of EF-P results in a decrease in protein levels of the MgtB Mg 2+ transporter and thus substitution of those two proline codons prevents the decrease in MgtB protein levels even in the absence of EF-P . Interestingly, given that mRNA levels of EF-P decrease during Salmonella infection (Nam et al., 2016), Salmonella is likely to produce an altered ratio of the MgtC virulence protein to the MgtB Mg 2+ transporter , which is required for Salmonella pathogenesis .
In addition to the mgtB gene in the mgtCBR operon, Mg 2+ transport in S. enterica is mediated by two additional loci, the mgtA and corA genes . Among them, the mgtA gene encoding the MgtA Mg 2+ transporter is similar to the mgtB gene in several aspects. First, both the mgtA and mgtB genes encode the P-type Mg 2+ transporting ATPases (51.3% identity) that involve the influx of Mg 2+ ions through phosphorylation of the key aspartic acid residues in the transporter proteins during each transport cycle Smith & Maguire, 1998). Second, transcription initiation of the mgtA and mgtB genes is controlled by the PhoP/ PhoQ two-component system (Soncini, Garcia Vescovi, Solomon, & Groisman, 1996), which is activated in response to low Mg 2+ (Garcia , as well as two other signals, mildly acidic pH and antimicrobial peptides (Bader et al., 2005;Prost et al., 2007). Third, the mgtA gene and mgtCBR operon are preceded by long leader RNAs, which harbor proline codon-rich ORFs, mgtL and mgtP, respectively (Lee & Groisman, 2012;Park, Cromie, Lee, & Groisman, 2010). The presence of proline codons in mgtL and mgtP allows EF-P to control transcription of the associated mgtA gene and mgtCBR operon (Gall et al., 2016;Nam et al., 2016). These similarities between the mgtA and mgtB genes lead us to examine whether EF-P controls translation of the mgtA Mg 2+ transporter gene, in addition to the previously identified EF-P's role in transcription elongation of the mgtA gene (Gall et al., 2016;Nam et al., 2016). In this paper, we determined that EF-P controls translation of the mgtA gene via two consecutive proline codons located in the coding region. Substitution of the proline codons prevents EF-P-mediated translational control in MgtA protein levels without affecting the Mg 2+ -importing activity of the MgtA protein. Moreover, removal of EF-P's control in the mgtA gene promotes Salmonella virulence, implying that EF-P compromises the Mg 2+ transporting activity of the MgtA protein during Salmonella infection.

| RNA extraction and quantitative real-time polymerase chain reaction (RT-PCR)
Total RNA was isolated using RNeasy Kit (Qiagen) according to the manufacturer's instructions. A quantity 500 ng to 1 μg of the purified RNA was synthesized to cDNA, using PrimeScript TM RT reagent kit (TakaRa). The mRNA levels of the mgtA and mgtB gene were measured by SYBR Green PCR Master Mix (TOYOBO) and appropriate primers (mgtA: 4308/4309, mgtB: 7763/7764, and rrsH: 6970/6971) and monitored, using a 7500 Fast Real-Time PCR system (Applied Biosystems, Foster City).

| Western blot analysis
Cells were grown for 5 hr or 6 hr in 35 ml of N-minimal mediumcontaining 0.01 mM or 10 mM Mg 2+ . Cells were normalized by measuring optical density at 600 nm (OD 600 ) and pelleted by centrifugation. Crude extracts were prepared in TBS (Tris-buffered saline) buffer by sonication and analyzed as described . The intensity of bands in the blots was quantified, using Image J software. The data are representative of two independent experiments, which gave similar results.

| Measuring growth of strains lacking Mg 2+ transporters
To address whether the mgtA proline substitution mutant carries the Mg 2+ -transporting activity in a strain lacking two other Mg 2+ transporters, MgtB, and CorA, cells were grown in N-minimal medium containing 0.01 mM or 100 mM Mg 2+ . Growth was determined at 37°C for 7 hr in a 96-well plate with orbital shaking and absorbance was measured at OD 600 every 2.5 min, using Synergy H1 (BioTek).

| Mouse virulence assays
Six-to eight-week-old female C3H/HeN mice were inoculated intraperitoneally with ~10 3 colony-forming units (CFU) of Salmonella strains. Mouse survival was followed for 21 days. Virulence assays were conducted twice with similar outcomes, and data correspond to groups of five mice. All procedures were performed according to approved protocols by the Institutional Animal Care and Use Committee from Kangwon National University.

| Construction of a strain with the chromosomal mgtA deletion
The one-step gene inactivation method (Datsenko & Wanner, 2000) was used for the chromosomal mgtA gene deletion. For construction of the mgtA deletion strain, the Km R cassette from plasmid pKD4 (Datsenko & Wanner, 2000) was amplified using primers DE-mgtA-F and DE-mgtA-R and integrated into 14028s chromosome to create EN394 (mgtA::Km R ). The Km R cassette was removed using plasmid pCP20 (Datsenko & Wanner, 2000) to generate EN396 (mgtA).

| Construction of chromosomal mutant strains with the mgtA proline codons substituted by alanine codons
We substituted consecutive proline codons in the mgtA coding region to alanine codons using the fusaric acid-based counterselection method . First, we introduced Tet R cassettes in two different regions of the mgtA gene as follows: We generated PCR products harboring tetRA genes using primers KHU546/ KHU547 (for 39th and 40th proline codons), KHU548/KHU549 (for 550th and 551st proline codons substitution in the mgtA-HA background) and KHU602/KHU603 (for 550th and 551st proline codons substitution in the wild-type background) and MS7953s genomic DNA as a template. The PCR products were purified using a QIAquick PCR purification kit (QIAGEN) and used to electroporate EN336 (mgtA-HA) or 14028s strain containing plasmid pKD46 (Datsenko & Wanner, 2000). The resulting mgtA(up)-HA::tetRA (EN921), mgtA(down)-HA::tetRA (EN922) and mgtA(down)::tetRA (EN981) strains containing plasmid pKD46 were kept at 30°C for the following step. Then, we replaced the tetRA cassettes by DNA fragments carrying proline to alanine substitutions in mgtA at positions 39 and 40 or 550 and 551. The DNA fragments were prepared by a two-step PCR process. For the first PCR, we used two sets of primer pairs KHU550/KHU553 and KHU552/KHU551 (for 39th and 40th proline codon), KHU554/KHU557 and KHU556/KHU555 (for 550th and 551st proline codon substitution in the mgtA-HA background) and KHU606/KHU557 and KHU556/KHU607 (for 550th and 551st proline codon substitution in the wild-type background) and 14028s genomic DNA as template. For the second PCR, we mixed the two PCR products from the first PCR as templates and amplified DNA fragments using primers KHU550/KHU551 (for 39th and 40th proline codon substitution) and KHU554/KHU555 (for 550th and 551st proline codon substitution in the mgtA-HA background) and KHU606/KHU607 (for 550th and 551st proline codon substitution in the wild-type background). The resulting PCR products were purified and integrated into the EN921, EN922 and EN981 chromosomes and selected against tetracycline resistance with media containing fusaric acid  to generate EN932 (mgtA Pro 39,40 Ala -HA), EN933 (mgtA Pro 550,551 Ala -HA) and EN982 (mgtA Pro 550,551 Ala ), tetracyclinesensitive, ampicillin-sensitive chromosomal mutants, respectively. The presence of the expected nucleotide substitutions was verified by DNA sequencing. A P22 phage lysate grown in strain DN337 (efp::Cm R ) was used to transduce strains EN932, and EN933 Salmonella selecting for chloramphenicol resistance to generate EN940 (mgtA Pro 39,40 Ala -HA, efp::Cm R ), and EN941 (mgtA Pro 550,551 Ala -HA, efp::Cm R ), respectively.

| efp deletion increases mgtA mRNA levels in a manner dependent on the interspersed proline codons at mgtL located in the leader region
The 264-nt long leader region controls transcription elongation of the mgtA gene by a transcription attenuation mechanism, whose critical step is the formation of the attenuator stem-loop (stem-loop B, Figure 1a). The formation of stem-loop B is determined by a degree of the coupling/uncoupling between transcription of the leader region and translation of mgtL located within the leader region of the mgtA gene (Park et al., 2010). Therefore, any conditions that slow down translation of mgtL ORF could uncouple transcription of the leader region and translation of mgtL and inhibit the formation of stem-loop B, resulting in an increase in transcription of the mgtA coding region (Figure 1a). The presence of alternating proline codons at 3rd, 5th, 7th, and 9th positions in mgtL raised the possibility that efp deletion affects ribosome stalling at mgtL and thus increases mgtA transcription. To test this possibility, we measured the effect of efp deletion on mRNA levels of the mgtA gene in the wild-type or a derivative Salmonella where we substituted the mgtL proline codons by other codons (Leu, Thr, His, Leu) (Please note that the mgtL proline substitution eliminates proline codons but retains a base-pairing required for the Mg 2+ -sensing riboswitch (Park et al., 2010;Cromie et al., 2006)). As expected, the efp deletion mutant increased mRNA levels of the mgtA gene by ~12 fold (Gall et al., 2016;Nam et al., 2016) (Supporting Information Figure S1A). The increase in mgtA mRNA levels is dependent on interspersed proline codons at mgtL because the elevation of mgtA mRNA levels was not detected in the mgtL derivative with the Pro substitution (Supporting Information Figure S1A). This indicates that efp deletion affects ribosome stalling at those proline codons in mgtL even though they are not consecutive and allows the leader to increase mgtA mRNA levels.
Control experiments proved that the Pro substitution at mgtL did not affect the expression behaviors of mgtB and efp genes (Supporting Information Figure S1B and C).

| Salmonella lacking EF-P decreases MgtA protein levels compared to Salmonella producing EF-P in low Mg 2+
In addition to the previous finding that EF-P controls transcription of the mgtA gene by the leader RNA (Gall et al., 2016;Nam et al., 2016) (Supporting Information Figure S1), we suspected that EF-P might control translation of mgtA gene based on following: First, MgtA protein harbors two polyproline motifs at positions 39 and 40, and 550 and 551 (Supporting Information Figure S2A and B). The location of the consecutive proline codons and the sequences neighboring the proline codons are well conserved with those in the MgtB protein, which is another Mg 2+ -importing transporter in Salmonella F I G U R E 1 Regulation of the mgtA gene encoding the MgtA Mg 2+ transporter by EF-P. (a) The mgtA gene has the 264 nt-long leader region and harbors a short ORF (mgtL: encoded by nucleotide 71-124) within the leader region that controls transcription elongation of the associated coding region in response to proline-tRNA pro levels. The mgtL have conserved four proline codons with an alternating arrangement. Also, the mgtA gene has consecutive proline codons at positions 550 and 551 in the coding region. When EF-P is present, ribosomes translate mgtL and allow the leader to form stem-loop B, decreasing mgtA mRNA levels. However, EF-P maintains MgtA protein levels by helping ribosomes to translate the mgtA gene. By contrast, when EF-P is absent, ribosomes stall at proline codons in mgtL and enable the formation of stem-loop C, resulting in an increase of mgtA transcripts. Interestingly, even though mgtA mRNA levels are elevated, MgtA protein levels decrease because ribosomes stall proline codons in the mgtA coding region due to lack of EF-P. Amino acid substitutions used in this study are indicated in blue. (b-e) Salmonella lacking EF-P decreases MgtA protein levels. Western blot analysis of crude extracts prepared from strains with the C-terminally HA-tagged mgtA gene in either the wild-type (EN336) or efp deletion mutant Salmonella (EN897). Blots were probed with anti-HA (b), anti-MgtB (c), anti-EF-P (d), or anti-Fur (e) antibodies to detect MgtA-HA, MgtB, EF-P, and Fur proteins, respectively. Bacteria were grown for 5 hr in N-minimal media containing 10 mM or 0.01 mM Mg 2+ as described in Materials and Methods. Numbers below the blots correspond to the relative amount of proteins when the amount present in the wild-type strain at low Mg 2+ is set to 100. RI represents relative intensity ( Figure 2a and Supporting Information Figure S2A and B) (Snavely et al., 1991). Second, our previous study identified that EF-P controls translation of the mgtB gene via one of those consecutive proline codons . To explore this, we created chromosomal mutant strains with the C-terminally HA-tagged mgtA gene in the wild-type or efp mutant Salmonella. Because transcription of the mgtA gene is controlled by the PhoP/PhoQ two-component regulatory system , we measured MgtA protein levels in a PhoP-inducing (0.01 mM Mg 2+ ) or PhoP-repressing (10 mM Mg 2+ ) condition. Wild-type Salmonella increased MgtA protein production in low Mg 2+ media, whereas Salmonella lacking EF-P produced lower levels of the MgtA protein compared to those of the wild-type in the same media (Figure 1b). And, this is similar to what we observed in MgtB protein levels ( Figure 1c)  , indicating that EF-P controls translation of both the mgtA and mgtB genes encoding the MgtA and MgtB Mg 2+ transporters respectively. Control experiments proved as follows: Both MgtA and MgtB proteins were not detected in high Mg 2+ media (Figure 1b,c). And efp deletion abolished EF-P protein production (Figure 1d) but had no effect on Fur protein levels (Figure 1e).

| The mgtA with the Pro550 and Pro551 substitution has no defect of the Mg 2+ -transporting activity in the strain lacking other Mg 2+ transporters
Next, we asked whether the Pro550 and Pro551 substitution has an impact on the Mg 2+ -importing activity of the MgtA protein because the Pro550 and Pro551 residues and the phosphorylated Asp377 Blots were probed with anti-HA (b), anti-MgtB (c), anti-EF-P (d), or anti-Fur (e) antibodies to detect MgtA-HA, MgtB, EF-P, and Fur proteins, respectively. Bacteria were grown for 5 hr in N-minimal media containing 0.01 mM Mg 2+ as described in Materials and Methods. Numbers below the blots correspond to the relative amount of proteins when the amount present in the wild-type strain at low Mg 2+ is set to 100. RI represents relative intensity It is interesting to note that the proline substitution mutant grew faster than Salmonella carrying the wild-type mgtA gene because MgtA protein levels in the proline substitution mutant are higher than those in the wild-type (Figure 3c). By contrast, an introduction of the mgtA deletion completely abolished growth in low Mg 2+ media ( Figure 3a), further supporting that the ability to grow in low Mg 2+ media is mediated by the presence of the functional mgtA gene in the strains lacking other Mg 2+ transporters. As a control, supplementation of 100 mM Mg 2+ supports the growth of all strains tested ( Figure 3b) (Hmiel, Snavely, Florer, Maguire, & Miller, 1989).

| The mgtA with the Pro550 and Pro551 substitution increases Salmonella's survival inside macrophages and virulence in mice
Because EF-P mRNA levels decrease during Salmonella infection (Nam et al., 2016) and also because the Pro550 and Pro551 sub-

| D ISCUSS I ON
Here, we showed a new example of EF-P's role in translation of the mgtA Mg 2+ transporter gene. We determined that lack of EF-P decreases MgtA protein levels and the decrease is mediated by the 550th and 551st proline codons in the mgtA coding region (Figures 1   and 2). Substitution of the Pro550 and Pro551 codons prevents the EF-P-mediated decrease of MgtA protein levels, thereby producing significant levels of the MgtA protein even in Salmonella lacking EF-P ( Figure 2). This property of the Pro550 and Pro551-substituted mgtA gene also influences Salmonella's pathogenicity (Figure 4) because Salmonella decreases mRNA levels of EF-P inside macrophages (Nam et al., 2016). Accordingly, Salmonella with the Pro550 and Pro551-substituted mgtA enhances survival inside macrophages and virulence in mice by increasing the amounts of the MgtA Mg 2+ F I G U R E 3 MgtA with the Pro 550, 551Ala substitution in the mgtA gene supports growth in low Mg 2+ in a strain lacking other Mg 2+ transporters. (a-b) Growth curves of a strain with the wild-type mgtA gene (EN977), the mgtA derivative with the Pro 550, 551 codons substituted by Ala codons (EN998), or the mgtA deletion (EL498) in a genetic background where both corA and mgtB genes are deleted. Bacteria were grown in N-minimal medium containing 0.01 mM (a) or 100 mM Mg 2+ (b) at 37°C for 7 hr in a 96-well plate with orbital shaking and measured absorbance at OD 600 every 2.5 min. (c-d) Western blot Analysis of crude extracts prepared from Salmonella strains listed above. Blots were probed with anti-HA (c) and anti-Fur (d) antibodies to detect MgtA and Fur proteins, respectively. Bacteria were grown for 6 hr in N-minimal medium containing 0.01 mM Mg 2+ transporter ( Figure 4). Given that the mgtA Pro550 and Pro551 substitution renders Salmonella hypervirulent, it reflects that wild-type Salmonella needs to compromise the Mg 2+ -transporting activity of the MgtA protein during infection. And, this is further supported by the previous finding that EF-P mediates the decrease in the protein levels of another MgtB Mg 2+ transporter during infection .
It is interesting to note that EF-P controls translation of both the mgtA and mgtB genes encoding the Mg 2+ -transporting ATPases via two consecutive proline codons in the respective coding regions. Moreover, elimination of EF-P's control from either the mgtA or mgtB genes by substituting the proline codons renders Salmonella hypervirulent, reinforcing the notion that EF-P limits Mg 2+ transport during infection by a similar regulatory mechanism imposed upon translation of the mgtA and mgtB Mg 2+ transporter genes. By decreasing Mg 2+ uptake during infection, Salmonella seems to coordinate intracellular Mg 2+ levels with intracellular ATP levels, which are also limited by the MgtC virulence protein within a macrophage phagosome Pontes, Lee, Choi, & Groisman, 2015).
One might wonder why Salmonella increases mgtA mRNA levels via the proline codons of mgtL in the leader RNA despite decreasing MgtA protein levels via the proline codons in the mgtA coding region when Salmonella is inside macrophages and thus decreases efp mRNA levels ( Figure 1). Does it have a potential role for the mgtA transcripts independent of producing the MgtA proteins? In addition to a previous example (Loh et al., 2009), a recent study may provide a clue for a potential role for the mgtA transcripts because one of the small leader transcripts of the mgtCBR operon is highly expressed inside macrophages and functions as a trans-acting riboregulator for downregulating synthesis of the Salmonella flagellin protein during infection (Choi, Han, Cho, Nam, & Lee, 2017). Therefore, a future exploration seeking a potential role(s) of the mgtA transcripts independent of its protein functions is expected.
A previous systemic approach found that lack of EF-P induces ribosome pausing the consecutive motif (VPPS) located near the Nterminus of the MgtA Mg 2+ transporter from Escherichia coli (Elgamal et al., 2014). Based on 90% identity between the mgtA genes from E. coli and S. enterica, the identified pause sequence corresponds to the sequence including Pro39 and Pro40 (IPPS) ( Figure S2). Even though they are highly similar to each other, the Pro39 and Pro40 substitution has no impact on MgtA protein levels upon efp deletion whereas the Pro550 and Pro551 substitution has. Therefore, it needs further attention to understand additional information to explain the discrepancy between the EF-P-mediated ribosomal pause site detected on ribosomal profiling data and the substitution effect on production of the MgtA proteins detected the strain lacking EF-P.

ACK N OWLED G EM ENTS
This work was supported, in part, by the Basic Science Research