AmpI Functions as an Iron Exporter To Alleviate β-Lactam-Mediated Reactive Oxygen Species Stress in Stenotrophomonas maltophilia

Stenotrophomonas maltophilia is an organism with a remarkable capacity for drug resistance with several antibiotic resistance determinants in its genome. S. maltophilia genome codes for L1 and L2, responsible for intrinsic β-lactam resistance.

ABSTRACT

Stenotrophomonas maltophilia is an organism with a remarkable capacity for drug resistance with several antibiotic resistance determinants in its genome. S. maltophilia genome codes for L1 and L2, responsible for intrinsic β-lactam resistance. The Smlt3721 gene (denoted ampI), located downstream of the L2 gene, encodes an inner membrane protein. The existence of an L2 gene-ampI operon was verified by reverse transcription-PCR (RT-PCR). For aerobically grown S. maltophilia KJ, inactivation of ampI downregulated siderophore synthesis and iron acquisition systems and upregulated the iron storage system, as demonstrated by a transcriptome assay, suggesting that AmpI is involved in iron homeostasis. Compared with the wild-type KJ, an ampI mutant had an elevated intracellular iron level, as revealed by inductively coupled plasma mass spectrometry (ICP-MS) analysis, and increased sensitivity to H₂O₂, verifying the role of AmpI as an iron exporter. The β-lactam stress increased the intracellular reactive oxygen species (ROS) level and induced the expression of the L1 gene and L2 gene-ampI operon. Compared to its own parental strain, the ampI mutant had reduced growth in β-lactam-containing medium, and the ampI mutant viability was improved after complementation with plasmid pAmpI in either a β-lactamase-positive or β-lactamase-negative genetic background. Collectively, upon challenge with β-lactam, the inducibly expressed L1 and L2 β-lactamases contribute to β-lactam resistance by hydrolyzing β-lactam. AmpI functions as an iron exporter participating in rapidly weakening β-lactam-mediated ROS toxicity. The L1 gene and L2 gene-ampI operon enable S. maltophilia to effectively cope with β-lactam-induced stress.

INTRODUCTION

Iron is an essential metal for bacterial growth since it is a critical cofactor for protein functioning in ATP generation, DNA replication, and the electron transfer chain (1). In addition, iron is a key determinant of bacterial infection potential (2). However, excess ferrous iron causes iron toxicity and leads to bacterial death. Therefore, it stands to reason that tight regulation between iron uptake systems and iron export systems is critical for bacterial viability.

Excess ferrous iron reacts with hydrogen peroxide to generate the highly toxic hydroxyl radicals via the Fenton reaction. Reactive oxygen species (ROS) are inevitable by-products of the normal metabolism in aerobic bacteria (3). Therefore, the surplus free iron in the cytosol is a crucial factor in potentiating ROS toxicity. Moreover, excess ROS can target iron-sulfur clusters of iron-containing enzymes and lead to the release of iron from these enzymes (4). In this regard, the toxicity of ROS can also be exacerbated by conditions that elevate the intracellular iron concentration. Accordingly, the toxicities of iron and ROS are mutually affected, with each potentially increasing the toxicity of the other.

Antibiotics are mainly classified by their action targets in bacterial cells, such as that of β-lactam toward penicillin-binding protein. It is well recognized that the bactericidal effect of antibiotics can be attributed to the interaction between antibiotics and their action targets, which in turn affects the target functions and causes bacterial death. Recently, the interplay among antibiotic exposure, ROS generation, and bactericidal effect has attracted much attention. Kohanski et al. have proposed that the antibiotic-target interactions induce bacterial changes in metabolism and stimulate intracellular accumulation of superoxide and hydrogen peroxide, which is linked to the bactericidal activity of antibiotics (5). This phenomenon is demonstrated in β-lactam-, aminoglycoside-, and quinolone-treated bacteria (6–7). However, somewhat contradictory opinions regarding antibiotic-induced ROS increases and the role of ROS in antibiotic killing are also proposed (8–9), as reviewed in detail elsewhere (10). It remains accepted so far that antibiotic-induced ROS production is not universal and varies because of the genetic background of the strains and the type of antibiotics investigated.

Stenotrophomonas maltophilia occurs naturally in water, soil, and plants and is recognized as an important opportunistic and nosocomial pathogen, causing pneumonia and bacteremia (11). The importance of iron in S. maltophilia biofilm formation, oxidative stress response, exopolysaccharide (EPS) production, outer membrane protein (OMP) regulation, quorum sensing, and virulence has been reported previously (12). Therefore, iron homeostasis is critical for S. maltophilia to adapt to stresses. It has been reported that both siderophore synthesis and iron uptake systems were upregulated when the clinical isolates of S. maltophilia were grown under iron-depleted conditions (13). To date, however, no proteins mediating iron export have been described in S. maltophilia.

The genome of S. maltophilia K279a reveals this organism to have a remarkable capacity for drug and heavy metal resistance (14). Consistent with this notion, clinical isolates of S. maltophilia are intrinsically resistant to β-lactam, aminoglycosides, and macrolides. S. maltophilia harbors two inducible β-lactamases, L1 and L2, whose expression contributes to β-lactam resistance (15). The inducible expression of L1 and L2 is regulated by ampR, which is divergently transcribed from the L2 gene (16). The ampR-L2 gene module and its underlying regulatory mechanism have been reported in our previous studies (15–20). Smlt3721 (denominated AmpI here), downstream of the L2 gene, attracted our attention due to the possibility of an L2 gene-Smlt3721 operon. In this study, we demonstrated that the L2 gene and ampI form an operon and that AmpI functions as an iron exporter. The significance of the L2 gene-ampI operon for S. maltophilia adaption to β-lactam stress was also addressed.

RESULTS

L2 and ampI genes form an operon and the ampI gene has its own promoter (P_ampI).

Analysis of the genomic sequence of S. maltophilia K279a revealed that the L2 gene (Smlt3722) and two genes downstream (Smlt3721 and Smlt3720) organize themselves into a probable operon. Smlt3721 and Smlt3720 were annotated as an Na⁺/H⁺ antiporter (denoted the ampI gene, where “I” indicates iron) and a hypothetical protein (HP), respectively (Fig. 1A). To elucidate whether these three genes are cotranscribed by the L2 gene promoter, reverse transcription-PCR (RT-PCR) was performed. Primers AmpI-C and HP-C (Fig. 1A and Table S1 in the supplemental material), which target the internal sequences of ampI and the HP gene, were used to produce cDNAs from total RNA prepared from S. maltophilia KJΔDI. KJΔDI is an in-frame ampD_I deletion mutant (where ampD_I is one of two ampD homologs) in which the L2 gene is highly expressed in the absence of β-lactam (17). The cDNAs were PCR amplified using primer sets HpQ-F/HpQ-R, AmpIQ-F/R, L2I-F/R, and L2Q-F/R (Table S1 and Fig. 1A) and should have amplicon lengths of 162, 205, 248, and 221 bp, respectively (Fig. 1A). The 205-, 248-, and 221-bp amplicons were observed in the AmpI-C-derived cDNA group but not in the HP-C derived group (Fig. 1B), supporting the presence of an L2 gene-ampI operon and ruling out the possibility of an L2 gene-ampI-HP gene operon.

FIG 1.

The L2 and ampI genes form an operon, and ampI has its own promoter (P_ampI). (A) Genomic organization surrounding the L2 gene-ampI operon of S. maltophilia. (B) Agarose gel electrophoresis of the products of reverse transcription-PCR. The cDNAs of S. maltophilia KJ were obtained by RT-PCR using the primers AmpI-C and HP-C, and then PCR was performed using different primer sets: lanes 1, primers L2Q-F and L2Q-R; lanes 2, primers L2I-F and L2I-R; lanes 3, primers AmpIQ-F and AmpIQ-R; lanes 4, primers HpQ-F and HpQ-R; lanes 5, primers SmeXQ-F and SmeXQ-R. The SmeX gene, which is intrinsically unexpressed in strain KJ, was used as a control for a check of DNA contamination during cDNA preparation. The S. maltophilia KJ chromosome DNA was used as a control for primer reliability. (C) The L2 gene promoter and P_ampI responded to β-lactam challenge. Overnight-cultured KJ(pL2_xylE) and KJ(pAmpI_xylE) cells were subcultured into LB medium with an initial OD₄₅₀ of 0.15. After a 5-h incubation, the bacterial culture was divided into two parts: one was treated with 30 μg/ml cefuroxime and the other was left untreated. After a further 1-h incubation, C23O activities were determined. Bars represent the average values from three independent experiments. Error bars represent the standard errors of means. *, P < 0.001, by Student's t test.

Next, we considered whether ampI possesses its own promoter (P_ampI), in addition to being driven by the L2 gene promoter. The plasmid-borne promoter transcriptional fusion constructs pL2_xylE and pAmpI_xylE (Fig. S1) were used to check this probability. KJ(pAmpI_xylE) displayed a higher basal level of catechol 2,3-dioxygenase (C23O) activity than KJ(pL2_xylE), and the expressed C23O activity was constant regardless of the presence of β-lactam (Fig. 1C). These results supported the idea that the ampI gene has its own promoter (P_ampI) and that the expression of P_ampI is independent of β-lactam.

AmpI protein is less involved in sodium homeostasis.

The ampI gene was predicted to encode a 404-amino-acid (aa) transmembrane (TM) protein of 12 predicted TM segments (TMHMM server, version 2.0 [http://www.cbs.dtu.dk/services/TMHMM-2.0/]). Since the ampI gene is annotated as an nha gene (Na⁺/H⁺ antiporter) in the S. maltophilia K279a genome, whether AmpI protein confers sodium tolerance was considered. Strains KJ, KJΔAmpI, and KJΔDI are, respectively, the representatives of normal-, non-, and high-expression models of AmpI protein, and their growth characteristics in the culture containing 0, 0.2, and 0.5 M NaCl were assessed. Upon shifting the bacterial culture to the NaCl-containing medium, the growth of KJ, KJΔAmpI_, and KJΔDI was retarded, especially in medium with 0.5 M NaCl. After this retardation period, the three tested strains exhibited similar growth patterns despite the concentrations tested (Fig. 2A), revealing that the major function of AmpI protein may not relate to sodium homeostasis.

FIG 2.

Role of AmpI in sodium tolerance. Bars represent the average values from three independent experiments. Error bars represent the standard errors of means. (A) The growth curves of strains KJ, KJΔAmpI, and KJΔDI in XOLNG medium and with different concentrations of NaCl (0 M, 0.2 M, and 0.5 M). Cell growth was measured by reading the OD₄₅₀ at 1-h intervals. (B) The L2 gene promoter and P_ampI responded to different NaCl concentrations. Overnight-cultured KJ(pL2_xylE) and KJ(pAmpI_xylE) cells were subcultured into XOLNG medium containing 0, 0.2, and 0.5 M NaCl. After a 5-h incubation, the C23O activities from each culture were determined.

The expression of the Na⁺/H⁺ antiporter gene is generally influenced by the NaCl concentration, and this notion was assessed. The C23O activities of KJ(pL2_xylE) and KJ(pAmpI_xylE) were determined in XOLNG medium, XOLN medium (33) with 10% glucose, containing 0, 0.2, and 0.5 M NaCl. Figure 2B shows that the C23O activities of both strains were relatively unchanged in response to various NaCl concentrations. These results further supported that AmpI is not involved in sodium homeostasis.

Inactivation of AmpI downregulates siderophore synthesis and iron acquisition systems, upregulates iron storage system, and increases the intracellular iron concentration and H₂O₂ sensitivity.

To elucidate the functions of AmpI, gene expression levels in logarithmic-phase KJ and KJΔAmpI cells were compared by transcriptomic analysis. Transcriptome sequencing (RNA-seq) data representing the alignment of sequences (short reads) to coding sequences (CDS) were quantified as the number of reads per kilobase of CDS length per million reads analyzed (RPKM). We defined statistical significance as the absolute change in transcript level (RPKM) that is equal to or greater than 3-fold. Based on this criterion, expression levels of 121 (2.6%) and 102 (2.2%) genes out of a total of 4,522 genes were increased and decreased, respectively (Table S2). Remarkably, the genes involved in siderophore production (ent cluster) and iron uptake systems (such as RhuR-, FiuA-, FepA-, FpvA, and FecA-like outer membrane proteins and FeoAB) were highly downregulated; in contrast, the iron storage-associated genes (hms cluster and bacterioferritin) and the putative membrane-associated transporter genes (Smlt2191 and Smlt3771) were upregulated in the KJΔAmpI strain relative to levels in KJ (Table 1). This observation led us to speculate on the involvement of AmpI in iron homeostasis. It seemed that an iron-overload stress occurred in KJΔAmpI cells and that this stress led to the downregulation of iron acquisition systems and upregulation of iron storage systems.

TABLE 1.

Transcriptomic analysis of selected genes differentially expressed in the ampI mutant, KJΔAmpI, compared to the level of the wild-type S. maltophilia KJ

Gene group and locus	RPKMa		Fold changeb	Encoded protein
	KJ	KJΔAmpI
Downregulated genes
Smlt0794	656.03	14.83	–44.23	Heme uptake protein
Smlt0795	392.60	32.74	–11.99	PhuA-like TonB-dependent OMP
Smlt0796	159.91	26.51	–6.03	Periplasmic protein
Smlt0797	510.23	57.71	–8.84	Periplasmic protein
Smlt1144c	15.99	2.39	–6.69	TonB-dependent OMP
Smlt1145	71.18	7.19	–9.89	Transmembrane protein
Smlt1146	72.00	34.12	–2.11	Cytoplasmic protein
Smlt1147	86.50	36.15	–2.39	Periplasmic protein
Smlt1148	941.94	263.25	–3.58	FiuA-like TonB-dependent OMP
Smlt1149	141.31	24.99	–5.65	Periplasmic protein
Smlt1233	212.69	16.37	–12.99	TonB-dependent OMP
Smlt1426	1083.25	70.05	–15.46	FepA, TonB-dependent OMP
Smlt1762	58.02	8.53	–6.81	FpvA-like TonB-dependent OMP
Smlt2210	184.82	28.35	–6.52	Ferrous iron transport protein A
Smlt2211	130.80	64.32	–2.03	Ferrous iron transport protein B
Smlt2353	116.14	22.48	–5.17	Periplasmic hydrolase
Smlt2354	916.79	43.05	–21.29	Periplasmic ATP-binding protein
Smlt2355	630.01	22.03	–28.59	Periplasmic protein
Smlt2356	169.29	12.47	–13.58	FepD/FepG, inner membrane protein
Smlt2357	106.03	5.69	–18.62	FepC, cytoplasmic protein
Smlt2358	139.15	71.34	–1.95	Periplasmic protein
Smlt2664	57.71	19.28	–2.99	Sigma factor
Smlt2665	57.29	10.18	–5.63	FecR-like TonB-dependent receptor
Smlt2666	22.86	5.95	–3.84	FecA-like receptor
Smlt2712	48.59	3.98	–12.22	Outer membrane protein
Smlt2713	776.70	41.58	–18.68	Extracellular protein
Smlt2714	106.72	10.19	–10.48	FecA-like TonB-dependent receptor
Smlt2715	15.83	10.58	–1.50	FecR-like protein
Smlt2716	40.08	8.61	–4.66	Sigma factor
Smlt2817	48.79	5.72	–8.54	EntA enterobactin synthesis
Smlt2818	119.40	8.55	–13.97	EntF
Smlt2819	156.58	11.21	–13.97	EntD
Smlt2820	223.37	15.99	–13.97	EntB
Smlt2821	139.51	14.00	–9.97	EntE
Smlt2822	84.58	9.69	–8.73	EntC
Smlt2823	61.87	6.04	–10.24	MFS transporter
Smlt2858	119.65	24.31	–4.92	Iron transporter
Smlt2935	29.70	2.84	–10.48	Sigma factor
Smlt2936	5.45	0.1	–54.5	FecR-like protein
Smlt2937	9.53	1.17	–8.15	HasR-like FecA-like TonB-dependent
Smlt2938	8.41	1.81	–4.66	Iron regulated lipoprotein
Smlt2939	21.22	4.05	–5.24	Energy transducer TonB
Smlt3898	12.69	3.47	–3.66	FecA-like TonB-dependent receptor
Smlt3899	28.77	9.27	–3.10	FecR-like protein
Smlt3900	47.04	0.5	–94.08	RpoE-like protein
Smlt3999	159.75	11.38	–14.04	TonB-dependent OMP
Smlt4135	92.14	13.43	–6.86	TonB-dependent OMP
Upregulated genes
Smlt0767	16.26	61.12	+3.76	5,10-Methylenetetrahydrofolate reductase
Smlt1524	327.01	851.63	+2.6	Bacterioferritin
Smlt1631	26.58	81.18	+3.05	Protein disulfide reductase
Smlt2191	36.96	112.56	+3.05	Putative transmembrane protein
Smlt2597	2.81	9.64	+3.44	NAD(P)H-dependent FMN reductase
Smlt3289	2.49	13.55	+5.44	Hemin storage system protein, HmsH
Smlt3290	1.78	6.13	+3.44	Hemin storage system protein, HmsF
Smlt3291	1.34	13.80	+10.31	Hemin storage system protein, HmsR
Smlt3292	0.1	18.42	+1842	Hemin storage system protein, HmsS
Smlt3601	181.18	384.09	+2.12	Bacterioferritin
Smlt3647	36.67	137.04	+3.74	Peroxidase bpoA
Smlt3771	93.12	253.50	+3.05	Putative transmembrane protein
Smlt4145	57.66	209.77	+3.64	Beta-barrel domain-containing OMP
Smlt4297	253.73	567.96	+2.24	Bacterioferritin

^aRPKM, reads per kilobase of transcript per million mapped reads.

^bFold change in expression compared to that in the wild-type KJ strain was determined by RNA-seq.

Given the transcriptome results, we speculated that AmpI is an iron exporter candidate and that inactivation of ampI leads to accumulation of iron intracellularly, leading to iron-overload stress. To test this assumption, the intracellular iron concentrations of logarithmic-phase KJ and KJΔAmpI cells were quantified by inductively coupled plasma mass spectrometry (ICP-MS). The specimen collection conditions of KJ and KJΔAmpI cells for the ICP-MS assay were the same as those for the transcriptome assay. The iron level in KJΔAmpI cells was approximately 3.4-fold higher than that in the wild-type KJ (Fig. 3A). Given the Fenton reaction, it can be reasoned that the intracellular iron concentration should be related to the bacterial H₂O₂ susceptibility, and the H₂O₂ susceptibility test supported this notion (Fig. 3B).

FIG 3.

Roles of AmpI in iron export and H₂O₂ susceptibility. Bars represent the average values from three independent experiments. Error bars represent the standard errors of the means. *, P < 0.001, by a Student's t test. (A) The intracellular iron levels of KJ and KJΔAmpI. Overnight-cultured bacterial cells were subcultured into fresh LB medium and incubated for 5 h. The amount of iron in the wild-type KJ and ampI mutant, KJΔAmpI, was determined by inductively coupled plasma mass spectrometry (ICP-MS). (B) Hydrogen peroxide sensitivities of KJ and KJΔAmpI. Bacterial suspension was uniformly spread onto LB agar, and filter paper with 15 μl of 10% hydrogen peroxide was placed onto the agar. The growth inhibition zone was recorded after a 24-h incubation at 37°C.

The expression of ampI is highly enhanced by β-lactam.

Considering the function of a gene in dealing with stresses, the gene expression response is a helpful clue. The expression of ampI in response to different stress signals was assessed. Given that ampI expression is regulated by both promoters (the L2 gene promoter and P_ampI) (Fig. 1), we used chromosomally in situ transcriptional fusion constructs, KJL2-23 and KJAmpI23 (Fig. S1), to explore individual expression levels of the L2 gene and ampI. After treatment of cells with FeSO₄ or menadione (MD), the C23O activities of KJAmpI23 had a mild but reproducible increase (approximately 1.5-fold) regardless of the concentrations tested; nevertheless, no change was observed in the C23O activity of KJL2-23 (Table 2). In contrast, the C23O activities of KJL2-23 and KJAmpI23 were significantly increased by β-lactam treatment in a concentration-dependent manner (Table 2).

TABLE 2.

The C23O activities expressed by KJL2-23 and KJAmpI23 under the different stresses

Treatment and concn	C23O activity (Uc/OD450)a
	KJL2-23	KJAmpI23
None	6 ± 1	23 ± 4
FeSO4 (μM)
35	3 ± 2	34 ± 5
175	5 ± 1	31 ± 6
350	6 ± 2	39 ± 8
Menadione (μg/ml)
20	4 ± 2	32 ± 4
30	6 ± 2	37 ± 6
Cefuroxime (μg/ml)
30	80 ± 10	103 ± 15
50	107 ± 12	143 ± 19
Cefoxitin (μg/ml)
30	150 ± 15	168 ± 16
50	196 ± 21	202 ± 22

^aOne unit of enzyme activity (Uc) was defined as the amount of enzyme that converts 1 nmol of substrate per minute.

AmpI is involved in alleviating β-lactam-mediated ROS toxicity.

The facts (i) that ampI and the L2 genes form an operon (Fig. 1A), (ii) that inactivation of ampI increases the intracellular iron concentration (Fig. 2A), and (iii) that ampI is highly expressed in response to challenge by β-lactams (Table 2) made us speculate whether AmpI contributes to lessening β-lactam-mediated stress by exporting iron. Therefore, we hypothesized that β-lactam may induce an increase in reactive oxygen species (ROS) in S. maltophilia and that iron exportation by AmpI may limit the damage caused by the Fenton reaction. To test this, we first investigated the impact of β-lactam treatment on ROS generation in KJ and KJ2 cells by a dihydrorhodamine 123 (DHR123) assay. KJ2 is a double-deletion mutant of L1 and L2 that was constructed in our previous study (21). When KJ2 cells were grown in the presence of piperacillin or cefoxitin, fluorescence induction occurred (Fig. 4A), indicating that β-lactams increase ROS generation in S. maltophilia. However, this phenomenon was less obvious in KJ cells (Fig. 4A), likely owing to the potent hydrolyzing abilities of L1 and L2 β-lactamases.

FIG 4.

Impact of β-lactam stress, ROS toxicity, and AmpI. (A) β-Lactam treatment induces ROS generation in S. maltophilia. The bacterial cells tested were cultured in LB medium containing dihydrorhodamine 123 and a β-lactam, as indicated, for 5 h, and the fluorescence at 550 nm was determined. The fluorescence induction is shown as the ratio of fluorescence in the presence of β-lactam to that in the absence of β-lactam. Bars represent the average values from three independent experiments. Error bars represent the standard errors of means. *, P < 0.001, by Student's t test. (B) Impact of AmpI on cell variability under β-lactam and/or iron depletion stress. The logarithmic-phase bacterial cells at 5 × 10⁵ CFU/μl were 10-fold serially diluted. Five microliters of bacterial suspension was spotted onto the LB agars containing β-lactam, 2,2′-dipyridyl (DIP) (30 μg/ml), and/or FeCl₃ (35 μM), as indicated. After 24 h of incubation at 37°C, the growth of bacterial cells was observed. KJ and its derived mutants were inoculated on an LB plate containing 16 μg/ml piperacillin (PIP), and KJ2 and its derived mutants were inoculated on an LB plate containing 2 μg/ml PIP. (C) The impact of AmpI on intracellular ROS levels. The bacterial cells tested were cultured in LB medium containing dihydrorhodamine 123 and a β-lactam as indicated for 5 h, and the fluorescence at 550 nm was determined. The relative fluorescence intensity was calculated using the fluorescence of KJ cells grown in LB without β-lactam as 1. Bars represent the average values from three independent experiments. Error bars represent the standard errors of means.

Next, we sought to determine whether β-lactam-induced ROS reinforces iron toxicity in the case of ampI inactivation. Inactivation of ampI compromised bacterial growth in a β-lactam-containing medium in either a β-lactamase-positive (KJ) or β-lactamase-negative (KJ2) genetic background. Complementation of the ampI mutant with an intact ampI gene reverted growth (Fig. 4B). To further survey the contribution of iron on the increased β-lactam susceptibility of the ampI mutant, the susceptibility of an ampI mutant to β-lactam was measured in iron-depleted medium and in medium complemented with iron. To achieve the condition of iron depletion, 2,2′-dipyridyl (DIP) at 30 μg/ml was added into the assay medium (22). In β-lactam- and DIP-containing medium, the viability of the ampI mutant was comparable to that of the wild type in either the KJ or KJ2 genetic background (Fig. 4B), supporting the observation that iron contributes to the β-lactam susceptibility of the ampI mutant. When iron was added to the β-lactam- and DIP-containing medium, the β-lactam susceptibility of the ampI mutant was not reverted, as expected (Fig. 4B). This observation strongly suggests that DIP may have some impact on the interplay of ROS and iron homeostasis.

Since β-lactam-challenged S. maltophilia has an elevated intracellular ROS level (Fig. 4A), a question may arise as to whether AmpI can export ROS in addition to iron. To answer this question, the β-lactam-induced ROS levels of KJ2 and KJΔ2AmpI were determined by the DHR123 assay, and the two strains demonstrated comparable ROS levels (Fig. 4C).

DISCUSSION

The bactericidal potency of β-lactam mainly depends on its ability to block peptidoglycan synthesis and can be reinforced by β-lactam-mediated ROS stresses (5, 23) although there are of contradictory viewpoints on this issue (8–9). If the β-lactam-mediated ROS stress indeed exists, is there any defensive mechanism for bacteria to deal with the problem? No convincing answers have been proposed thus far. In this study, we characterized a novel inner membrane protein, AmpI (Smlt3721), which was originally annotated as an Na⁺/H⁺ antiporter gene in the genome sequencing project (14). We validated that AmpI, as an iron exporter, contributes to β-lactam resistance by limiting β-lactam-induced ROS toxicity and thus provided support for the linkage between β-lactam-mediated killing and ROS stress.

Based on protein analysis, AmpI is predicted to contain a domain of a sodium/hydrogen exchanger superfamily and to function as a monovalent cation-proton antiporter (https://blast.ncbi.nlm.nih.gov/Blast.cgi). When AmpI protein was compared with other proteins in the GenBank, it was found to have 94% to 99% identity with the annotated cation-proton antiporters of Pseudomonas aeruginosa, Xanthomonadaceae, and Lysobacter enzymogenes. This may be the reason why the gene is annotated as an Na⁺/H⁺ antiporter in the S. maltophilia K279a genome (14). However, our results support the idea that the function of AmpI is less involved in sodium homeostasis than in iron homeostasis. It cannot be totally denied that AmpI may have the capacity to export sodium under specific conditions but not under the conditions we tested.

Our window into the role of AmpI in iron export first came from transcriptome analysis (Table 1) and was further verified by an intracellular iron level assay by ICP-MS (Fig. 3A). Loss of function of AmpI results in elevated intracellular iron concentrations, which provides a reasonable explanation for the altered expression patterns of iron homeostasis-associated genes revealed in the transcriptome assay. As revealed in a variety of bacteria (24–30), there are five different groups of iron exporters documented: ABC-type transporters, P_1B-type ATPases, cation diffusion facilitator (CDF) family proteins, major facilitator superfamily (MFS) proteins, and membrane-bound ferritin-like proteins. Interestingly, the sequence similarity between these iron exporters and AmpI is unexpectedly low (12% to 19% identity). Phylogenetic linkage between AmpI and these known iron exporters was analyzed. Figure 5 demonstrates that AmpI displays the closest phylogenetic relationship with members of the CDF family, such as PieF of Escherichia coli and FeoE of Shewanella oneidensis MR-1, among the microorganisms considered. The CDF family is a group of inner membrane proteins that can export divalent metal cations using the proton motive force as an energy source (31, 32). A typical bacterial CDF contains an N-terminal domain (NTD), six transmembrane (TM) helices, and a C-terminal cytoplasmic domain (CTD). However, bioinformatics analysis of AmpI appears not to support AmpI as a member of the CDF family according to the known criteria since the TMHMM server (version 2.0) predicts that AmpI has 12 TM domains. Therefore, AmpI of S. maltophilia may represent a novel group of iron exporters, in addition to the five known groups. However, the possibility cannot be immediately ruled out that AmpI may represent a novel subfamily of the CDF family.

FIG 5.

The phylogenetic relationship of AmpI of S. maltophilia and previously characterized iron exporters documented in other bacteria. The dendrogram was constructed by the neighbor-joining method using the amino acid sequences of the proteins. The bootstrap numbers at the branch points refer to 1,000 replications. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Poisson correction method, and units are of the number of amino acid substitutions per site.

Based on the findings in this study, we propose a model to illustrate the significance of the L2 gene-ampI operon in S. maltophilia. Under normal growth conditions, the L2 gene is feebly expressed, and ampI has a moderate level of expression driven by its own promoter (P_ampI). AmpI acts as an iron exporter and may have an accessory role in mitigating the aerobic metabolism-generating ROS stress (Fig. 6A). In this instance, inactivation of AmpI results in an elevated intracellular iron level, which leads to global regulation by downregulating siderophore synthesis and iron acquisition systems and upregulating the iron storage system, as evidenced in the transcriptome assay (Table 1). Upon challenge with β-lactam, the L1 gene and L2 gene-ampI operon are upregulated, and the intracellular ROS level is also elevated. L1 and L2 β-lactamases contribute much to β-lactam resistance by hydrolyzing β-lactam. In this instance, AmpI functions to export cytosol iron, which can protect bacteria from ROS toxicity by limiting the Fenton reaction (Fig. 6B). Due to the potent L1 and L2 β-lactamases induced by β-lactam, the β-lactam-mediated ROS is slightly increased in wild-type KJ but significantly increased in a β-lactamase-null mutant KJ2 (Fig. 4A). Therefore, the contribution of AmpI to the alleviation of β-lactam-mediated ROS stress is more obvious in KJ2 background (Fig. 4B).

FIG 6.

Proposed model for the role of AmpI in the alleviation of β-lactam-mediated ROS stress. (A) In physiologically grown S. maltophilia, ampI is moderately expressed driven by its own promoter. In this instance, AmpI may have an accessory role in alleviating aerobic metabolism-mediated ROS stress. (B) During the β-lactam challenge, the L1 gene and the L2 gene-ampI operon are upregulated in an AmpR-dependent manner. Meanwhile, the intracellular ROS level is also elevated. The induced L1 and L2 β-lactamase activities contribute to β-lactam resistance by hydrolyzing β-lactam. The increased AmpI functions to export cytoplasmic iron and limits the occurrence of the Fenton reaction.

MATERIALS AND METHODS

Bacterial strains, media, plasmids, and primers.

The bacterial strains, plasmids, and primers used in this study are listed in Table S1 in the supplemental material. The XOL minimal medium (33) contained the following basal salts (per liter): K₂HPO₄, 0.7 g; KH₂PO₄, 0.2 g; (NH₄)2SO₄, 1.0 g; MgCl2·6H2O, 0.1 g; FeSO4·7H2O, 0.01 g; MnCl₂, 0.001 g, pH 7.15. XOLN medium was XOL supplemented with 0.0625% yeast extract and 0.0625% tryptone (33). XOLNG medium consisted of XOLN minimal medium with 10% glucose as a carbon source (Fig. 2). NaCl, menadione (MD), and FeSO₄ were added at the indicated concentrations when necessary.

Construction of an in-frame ampI deletion mutant, KJΔAmpI, and complementation plasmid, pAmpI.

The ampI gene of S. maltophilia KJ was obtained by PCR with primers AmpI-F and AmpI-R (Fig. S1 and Table S1) and cloned into pEX18Tc, yielding pEXAmpI. Plasmid pΔAmpI was obtained by deleting an internal 606-bp SphI-SphI fragment from pEXAmpI. A deletion mutant, KJΔAmpI (Fig. S1), was constructed by double-crossover homologous recombination between the KJ chromosome and the recombinant plasmid pΔAmpI as described previously (17).

Given that the expression of ampI was driven by two promoters, the L2 gene promoter and P_ampI, the complementation plasmid pAmpI was constructed by placing the ampI gene under the control of both promoters. A 2,170-bp DNA segment containing the intact ampI gene and its own promoter (P_ampI) was excised from pEXAmpI and cloned into pL2_xylE by replacing the xylE gene with the ampI gene, yielding pAmpI (Fig. S1).

Construction of promoter-xylE transcriptional fusion constructs pL2_xylE and pAmpI_xylE.

A 248-bp DNA segment containing the L2 gene promoter region was obtained by PCR using the primer sets L2P-F and L2P-R (Table S1) and cloned into pRKxylE, yielding pL2_xylE. The P_ampI-containing BglII-SphI DNA fragment excised from pAmpI was cloned into pRKxylE, generating pAmpI_xylE (Fig. S1). The orientation of the xylE gene was opposite that of P_lacZ of the pRK415 vector.

Transcriptome analysis.

Overnight cultures of bacterial cells were subcultured into fresh LB broth with at an initial optical density at 450 nm (OD₄₅₀) of 0.15 and incubated for 5 h to the mid-exponential growth phase (OD₄₅₀ of 3.0). Total RNA isolation, rRNA depletion, adapter-ligated cDNA library construction and enrichment, and cDNA sequencing were performed as described previously (34). For RNA-seq analysis, the total number of reads per gene between the replicates was normalized by RPKM (reads per kilobase million) values.

RT-PCR.

Total RNA extraction and cDNA preparation were carried out as previously described (17). The primers used for cDNA preparation and reverse transcription-PCR (RT-PCR) are listed in Table S1. The RT-PCR products were separated on 2% agarose gels and stained with ethidium bromide.

Determination of C23O activity.

The xylE gene product is a catechol 2,3-dioxygenase (C23O), which converts the colorless substrate catechol to an intensely yellow hydroxymuconic semialdehyde, which can be spectrophotometrically quantified. The C23O activity in intact cells was measured as described previously (15). One unit of enzyme activity (Uc) was defined as the amount of enzyme that converts 1 nmol of substrate per minute. Specific activity (Uc/OD₄₅₀) of the enzyme was defined in terms of units per OD₄₅₀ unit of cells.

ICP-MS.

The bacterial culture preparation for the inductively coupled plasma mass spectrometry (ICP-MS) experiment was the same as that for the transcriptome assay. The cell pellets were washed twice with Milli-Q water and resuspended in 2 ml of Milli-Q water. Cell numbers in the cell suspension were quantified by CFU counts. For the ICP-MS assay, the cell suspension was sonicated, centrifuged, and then filtered through a 0.45-μm-pore-size Millipore membrane. ICP-MS was performed using an Agilent 7700e instrument (Agilent Technologies, USA).

Determination of intracellular ROS.

Overnight cultures were subcultured to fresh LB medium containing 0.9 μg/ml dihydrohodamine 123 (DHR123) and β-lactam at different concentrations (0, 3, 10, 20, or 30 μg/ml) with an initial OD₄₅₀ of 0.15. DHR123 can be oxidized by intracellular reactive oxygen species (ROS) and further form the fluorescent rhodamine 123. After a 5-h incubation, rhodamine 123 was detected using 500 nm as the excitation wavelength and 550 nm as the emission wavelength.