Molecular mechanism for sphingosine-induced Pseudomonas ceramidase expression through the transcriptional regulator SphR

Pseudomonas aeruginosa, an opportunistic, but serious multidrug-resistant pathogen, secretes a ceramidase capable of cleaving the N-acyl linkage of ceramide to generate fatty acids and sphingosine. We previously reported that the secretion of P. aeruginosa ceramidase was induced by host-derived sphingolipids, through which phospholipase C-induced hemolysis was significantly enhanced. We herein investigated the gene(s) regulating sphingolipid-induced ceramidase expression and identified SphR, which encodes a putative AraC family transcriptional regulator. Disruption of the sphR gene in P. aeruginosa markedly decreased the sphingomyelin-induced secretion of ceramidase, reduced hemolytic activity, and resulted in the loss of sphingomyelin-induced ceramidase expression. A microarray analysis confirmed that sphingomyelin significantly induced ceramidase expression in P. aeruginosa. Furthermore, an electrophoretic mobility shift assay revealed that SphR specifically bound free sphingoid bases such as sphingosine, dihydrosphingosine, and phytosphingosine, but not sphingomyelin or ceramide. A β-galactosidase-assisted promoter assay showed that sphingosine activated ceramidase expression through SphR at a concentration of 100 nM. Collectively, these results demonstrated that sphingosine induces the secretion of ceramidase by promoting the mRNA expression of ceramidase through SphR, thereby enhancing hemolytic phospholipase C-induced cytotoxicity. These results facilitate understanding of the physiological role of bacterial ceramidase in host cells.


Identification of gene(s) involved in sphingolipid-mediated CerN induction in P. aeruginosa.
We previously reported that the release of CerN into a culture medium of P. aeruginosa was strongly enhanced by the addition of sphingolipids such as SM, Cer, and Sph 21 . In order to elucidate the molecular mechanism of underlying sphingolipid-mediated CerN induction, we adopted a transposome-based gene disruption strategy in the present study. We initially generated a P. aeruginosa mutant (AN17-BGT), in which the cerN gene was replaced with the E. coli β-galactosidase gene (Fig. 1). Using this mutant, the induction of CerN with sphingolipids was easily monitored by β -galactosidase activity, which may be assessed using chromogenic substrates such as pNP-β -galactose. We then introduced EZ-Tn5 < KAN-2> Tnp Transposome into AN17-BGT and selected mutants lacking β -galactosidase activity in the presence of SM. After the screening of 1,700 mutants, we identified PA5324 as a candidate gene involved in sphingolipid-mediated CerN induction. In the genome database of P. aeruginosa PAO1 26,27 , PA5324 was assigned as the AraC-type transcriptional regulator, SphR 28 . Functions of SphR in hemolysis caused by P. aeruginosa. We generated SphR-deficient mutants of P. aeruginosa using the tetracycline resistance (Tc r ) cassette as described in the "Methods" section. The CerN activities of wild-type P. aeruginosa (WT) and SphR-deficient mutants were measured in the presence or absence of SM. As shown in Fig. 2a, CerN activity increased in WT when SM was added to the culture, while its activity was retained at the basal level in SphR-deficient mutants (KO) even in the presence of SM. Similarly, the deletion of SphR in P. aeruginosa resulted in hemolytic activity being significantly lower than that in WT, and this was restored in part by the complementation of SphR to the SphR-deficient mutant, indicating that hemolytic activity is also regulated by SphR (Fig. 2b). Since CerN enhances the hemolytic activity of PlcH 21 , these results strongly suggest that SphR plays an important role in hemolysis caused by P. aeruginosa through the regulation of CerN.

Roles of SphR in the transcriptional regulation of CerN.
We then examined the roles of SphR in the transcription of the cerN gene in P. aeruginosa using real-time PCR. The addition of SM resulted in cerN transcription levels being more than 80-fold higher in WT than in the absence of SM; however, this sphingolipid-mediated increase in cerN transcription was not observed in SphR-deficient mutants (Fig. 2c). The transcription of PA3001 (housekeeping gene, glyceraldehyde-3-phosphate dehydrogenase) did not increase in WT with the addition of SM. These results indicate that the sphingolipid-mediated induction of CerN occurs at the transcriptional level and is regulated by SphR.
Effects of the deletion of SphR on P. aeruginosa growth. We compared the cell growth of WT and SphR-deficient mutants in nutrition-rich medium (PY medium) and synthetic medium containing SM (SM synthetic medium) as the sole source of carbon. No significant differences were observed in cell growth between WT and SphR-deficient mutants in PY medium in the presence or absence of SM (Fig. 2d). On the other hand, the growth of SphR-deficient mutants was markedly slower than that of WT in SM synthetic medium (Fig. 2e), suggesting that SM is not utilized by SphR-deficient mutants as a source of carbon because CerN was not expressed in P. aeruginosa without SphR (Fig. 2a,c).

DNA microarray analysis. SM is an abundant lipid in the plasma membranes of host cells and may influ-
ence the expression of genes involved in SM metabolism in P. aeruginosa during the infectious process and colonization. Therefore, we performed a DNA microarray analysis in order to identify the genes induced in P. aeruginosa when SM was added to the culture. The mRNA quantity of P. aeruginosa was measured using cells cultured in the presence or absence of SM (control). The expression of 94 genes was significantly stronger by more than 1.5-fold (P < 0.05), in the presence of SM than in the control (Table 1). Of those, the expression of 27 genes increased by more than 2-fold, and that of 5 genes by more than 6-fold. SM added to the culture of P. aeruginosa may be sequentially converted to phosphorylcholine and Cer by PlcH (SMase), choline and phosphate by phosphatase, and fatty acids and Sph by CerN. Thus, the results of the DNA microarray may reflect gene expression induced not only by SM, but also by SM metabolites such as Cer, Sph, fatty acids, phosphorylcholine, and choline. As expected, the transcript of the cerN gene (PA0845) showed the greatest increase in P. aeruginosa in the presence of SM than in the control (~19-fold, P = 4.91E-6). This result is consistent with those shown in Fig. 2c. The expression of plcH (PA0844) and its accessory protein gene (plcR, PA0843) also increased in the presence of SM, however, the degree of this increase was lower than cerN (2-fold, P < 0.001). The transcriptional level of SphR did not change, whereas those of PA5325 -PA5328, which are adjacent to sphR (PA5324) in the P. aeruginosa genome, increased by 6~44-fold in the presence of SM; however, the biological relevance of these four genes remains to be clarified. Although transcripts of genes related to sulfur metabolism (PA0280-PA0283, PA1837-PA1838, and PA4442-PA4443) moderately increased (2~3-fold) in the presence of SM, the mechanism underlying the regulation of these genes by SM is unknown. The expression of PA5373 (betB) and PA5374 (betI), which form an operon for choline catabolism with PA5372 (betA), increased 2-fold in the presence of SM. Furthermore, PA5375 (betT1), a putative choline transporter, exhibited a significant increase (1.2-fold, P = 0.0138). The increase observed in the transcription of choline catabolism-related genes (bet genes) appeared to be mediated by choline, but not SM 25 . Similarly, the transcription of Fad (fatty acid degradation) genes (PA3014/FaoA/FadB5, PA3013/FaoB/FadA5, PA4994/probable acyl CoA dehydrogenase, and PA4995/probable acyl CoA dehydrogenase) was found to moderately increase in the presence of SM, suggesting that the expression of Fad genes in P. aeruginosa is regulated by fatty acids released from SM through the corporative reactions of PlcH (SMase) and CerN. Lipids inducing cerN expression. As previously reported, not only SM, but also its metabolites such as Cer and Sph induce CerN activity in P. aeruginosa 21 . However, we found that β -galactosidase activity was only strongly induced when P. aeruginosa AN17-BGT was cultured in the presence of Sph, but not SM, Cer, or palmitic acid (Pal) (Fig. 3a). In AN17-BGT, SM may be hydrolyzed to Cer by PlcH, whereas Cer is not converted to Sph because the cerN gene was replaced with the β-galactosidase gene in this mutant strain. This result suggests that the transcription of the cerN gene is induced by Sph, but not SM in P. aeruginosa through SphR. In order to elucidate the relationship between the Sph concentrations and extent of the transcription of the cerN gene, AN17-BGT was cultured with Sph at various concentrations and β -galactosidase activity was measured. As shown in Fig. 3b, β -galactosidase activity was detected in the presence of no less than 100 nM Sph and activity   Analysis of the cerN promoter. By using 5′ -RACE, we showed that the cerN transcription start point was located 80 nucleotides upstream of ATG (Fig. 4a). In order to identify the minimal promoter region required for cerN transcription, we generated plasmid constructs containing different sized promoter regions driven by LacZ. P. aeruginosa PAO1 was transformed with the constructs (deletion mutants) and the promoter activities of the mutants were evaluated by measuring β -galactosidase activity in the presence of SM. As shown in Fig. 4b and c, β -galactosidase activity was markedly weaker in the mutant possessing 100-bp, but not 200-bp nucleotides from the transcription start point than in the control possessing the putative full promoter sequence, indicating that an integral region for Sph-inducible transcription is present between − 200 and − 100 bp upstream of the cerN gene. Further investigations using several deletion mutants possessing 135-to 200-bp nucleotides identified the critical region in the cerN promoter, which was present between − 155 and − 135 bp upstream of the cerN gene ( Fig. 4b and d).
In order to elucidate the nucleotide sequence required for binding to Sph in the cerN promoter, the sequence of PA0845 (cerN) was aligned with that of the PA5325 and a PA5328-PA5326 operon, all of which may be regulated by SphR (Table 1). The conserved nucleotides among these regions were asterisked (Fig. 5a)  consensus sequences were underlined (Fig. 5b). The estimated consensus region appeared to consist of two repeating 15-bp nucleotides separated by a 7-bp spacer when SphR formed a dimer similar to other bacterial transcriptional regulators such as AraC. We generated several mutants (M1-M4), in which a few nucleotides were replaced with A in the conserved sequence and evaluated the promoter activities of these 4 mutants in the presence of SM. As a result, we found that the mutants M1, M2, and M4 completely lost their promoter activities, while the activity of the mutant M3 decreased to one fourth that of WT (Fig. 5c). This result indicates that the conserved sequence is critical for SM-inducible promoter activity.

Specificity of SphR toward various lipids.
The transcription of the cerN gene was induced in P. aeruginosa by Sph, but not SM, Cer, or fatty acids (Fig. 3a). We examined whether the Sph-inducible transcriptional regulation of CerN depends on SphR by evaluating the specificity of SphR toward various lipids. Recombinant MBP-fused SphR (MBP-SphR, 84 k) was used in this experiment because free SphR was not soluble under the conditions used. The binding of MBP-SphR to the cerN promoter was examined using an electrophoretic mobility shift assay (EMSA), in which the biotin-labeled cerN promoter (− 200~− 81) was used as a probe. As shown in Fig. 6a, MBP-SphR formed a complex with the probe in order to shift up on the gel in the presence of Sph; however, this mobility shift of the probe was not observed in the absence of Sph. The binding of MBP-SphR to the cerN promoter attenuated competitively with the addition of the unlabeled probe. No complex was formed with MBP per se and the cerN promoter. These results indicate that SphR binds to the cerN promoter in the presence of Sph. We then evaluated the specificity of SphR to various lipids by EMSA and found that the specificity of SphR was very strict, i.e., SphR bound to Sph, but not SM, Cer, Pal, DG, or PC under the conditions used (Fig. 6b). Among the long-chain bases tested, SphR bound to Sph, dihydroSph, and phytoSph, but not to S1P or 1-deoxy-Sph. Furthermore, SphR recognized the stereoisomers of Sph, i.e., it bound to D-erythro-Sph more strongly than L-threo-Sph (Fig 6c). These results clearly indicate that SphR recognizes the OH group at C1 and stereochemical configuration at C3 of Sph; however, the double bond between C4 and C5, or hydroxy group at C4 of Sph is not critical for binding to SphR. An acyl chain length of more than 14 in Sph is sufficient for binding to SphR.  In order to examine whether Sph directly binds to purified MBP-SphR, we performed a pull-down assay, in which a complex of MBP-SphR with biotin-labeled Sph was pulled down by avidin-labeled magnetic beads and detected on SDS-PAGE after staining with Coomassie Brilliant Blue. As shown in Fig. 6d, MBP-SphR was only detected in the presence of biotin-labeled Sph and MBP per se did not bind to biotin-Sph, indicating that SphR directly binds to Sph. In conclusion, we demonstrated that Sph specifically binds to SphR, through which the transcription of the cerN gene is initiated in P. aeruginosa, i.e., the Sph-inducible transcription of CerN entirely depends on SphR as a transcriptional regulator.

Discussion
In an attempt to elucidate the mechanism underlying the sphingolipid-mediated induction of CerN in P. aeruginosa, we examined gene(s) that regulate CerN transcription in response to sphingolipids. As a result, we identified the Sph-specific AraC-type transcriptional regulator, SphR, which is involved in Sph-mediated cerN gene expression. AraC-type transcriptional regulators are known to control a number of cellular processes such as carbon metabolism, stress responses, and virulence in Gram-positive and -negative bacteria 29,30 . Sixty-one putative AraC-type transcriptional regulators were identified in the P. aeruginosa genome when analyzed using a Pfam database 31 . Some AraC-type transcriptional regulators such as VqsM is related to quorum sensing 32 , PchR is involved in pyochelin (siderophore) biosynthesis 33 and GbdR is involved in phosphorylcholine catabolism 25,34 . These transcriptional regulators were previously reported to be important for P. aeruginosa infections. AraC-type transcriptional regulators are generally composed of an N-terminal ligand binding domain that is responsible for dimerization and a C-terminal helix-turn-helix (HTH) DNA-binding domain. The binding of a ligand to the N-terminal domain of the transcriptional regulator induces a conformational change, by which it binds to the promoter region of the target gene via a C-terminal domain. We speculated that SphR may be converted to an active form when Sph binds to the N-terminal region, and this active form may then bind to the promoter region of the cerN gene via the C-terminal HTH motif, thereby initiating cerN mRNA transcription.
LaBauve and Wargo also recently identified SphR in P. aeruginosa using a different approach. They performed a DNA microarray in the presence or absence of pulmonary surfactants and found that the expression of several genes including PA5325 and PA5328-PA5326 (these three genes are regulated as an operon) increased under the conditions employed. They ascertained that these genes were transcriptionally regulated by Sph in pulmonary surfactants through SphR as a transcriptional regulator 28 . Since Sph is toxic for bacteria and the deletion of SphR was found to decrease the tolerance of P. aeruginosa toward Sph, they speculated that PA5325 (sphA) and PA5328-PA5326 (sphB, C, D) are genes related to the import and metabolism of Sph in P. aeruginosa, respectively.
P. aeruginosa synthesizes a small amount of PC (~4% of all phospholipids) 35,36 and appears to possess all the enzymes required to metabolize this phospholipid. Thus, host-derived PC may be utilized by this bacterium as a nutrient for its growth and survival. Host PC is degraded by phospholipase C and DG lipases to phosphocholine, glycerol, and fatty acids in P. aeruginosa 23 . Fatty acids, phosphocholine, and glycerol are further metabolized by the β -oxidation pathway, Bet pathway, and Glp pathway in the bacterium, respectively. Son et al. found that these genes, which are involved in PC metabolism were up-regulated in P. aeruginosa by the addition of PC to the culture 23 . Sun et al. also reported that P. aeruginosa mutants lacking double or triple genes involved in PC metabolism exhibited a severe growth defect on synthetic medium containing PC as the sole carbon source as well as reduced survival rates in the lungs of mice 37 . These findings indicate that host PC is an integral nutrient for P. aeruginosa to infect and survive in the lungs of mice.
SM also appears to be a nutrient for P. aeruginosa. SM is a phosphocholine-possessing phospholipid, similar to PC; however, the lipid portion of SM is Cer instead of DG in PC. Host SM has been suggested to be hydrolyzed by two SM hydrolases (PlcB and PlcH) of P. aeruginosa in order to generate Cer and phosphocholine 38 . Cer is further decomposed by CerN into Sph and fatty acids. In the present study, we found that not only PlcH and its accessory protein (PlcR), but also the Fad (β -oxidation) and Bet (choline metabolism) genes were up-regulated in P. aeruginosa either in the presence of PC or SM. On the other hand, the DG-lipase and Glp (glycerol metabolism) genes were induced by PC, but not SM, while SphA and SphBCD were induced by SM, but not PC in P. aeruginosa. These results clearly indicate that SM is metabolized in part by the same pathway as PC; however, the metabolic pathways of Cer (Sph moiety) and DG (glycerol moiety) are completely different from each other in P. aeruginosa (see Supplementary Fig. S1).
Several genes were up-regulated in P. aeruginosa in the presence of SM, but not PC. Of these, PA0845 is the gene encoding neutral CDase (CerN) and PA5325 (sphA) is the gene encoding a homologue of porin, which makes a channel in the plasma membrane as a transporter. PA5328-PA5326 (sphBCD), three genes that form an operon, showed homology to cytochrome C oxidase (cbb3-type, subunit III), D-arabinono-1,4-lactone oxidase, and alanine racemase (N-terminal domain), respectively. These proteins may be involved in Sph transport and metabolism in P. aeruginosa; however, the physiological functions of these proteins have not yet been elucidated.
CerN may be induced in P. aeruginosa in the presence of not only Sph, but also SM or Cer 21 . However, we found that the promoter of cerN was activated by Sph, but not by SM or Cer when the strain ANT17-BGT was used instead of P. aeruginosa WT (Fig. 3a). This contradiction may be consistently explained, i.e., SM and Cer are converted to Sph by PlcH (SMase) and CerN in P. aeruginosa WT, but are not converted to Sph in the strain AN17-BGT because the CerN gene was replaced with the E. coli β-galactosidase gene and, thus, SM/Cer was not converted to Sph. On the other hand, the promoter of cerN in the strain AN17-BGT was activated by SM when we screened β -galactosidase-deficient strains (Fig. 1). We found that the egg yolk SM used in this experiment was contaminated with Sph (~0.5% of SM), and, thus, this experiment may be regarded as being performed in the presence of Sph. Thereafter, we used Sph-free SM in the experiments described in Figs 3a and 6b. Figure 7 shows a working model for the functions of SphR in P. aeruginosa. The presence of a positive feedback loop has been suggested between SphR and CerN, i.e., 1) Host Sph is incorporated into P. aeruginosa and induces the expression of CerN through the activation of SphR, 2) CerN is secreted into the host environment, in which host-cell Cer is hydrolyzed to generate Sph, 3) the Sph generated is then incorporated into P. aeruginosa and increases CerN expression through the activation of SphR. This positive feedback loop amplifies the generation of Sph in host environments and may exacerbate damage to host cells. In this model, PlcH also plays an important role in the production of Cer from host-derived SM. PlcH is induced by the PhoBR regulon under phosphate starvation conditions and by the GbdR regulon with the catabolism of choline from phospholipids such as PC and SM. On the other hand, SphR regulates intracellular Sph levels in P. aeruginosa through the expression of sphABCD, which may be involved in the transport and metabolism of Sph 28 . Since Sph exhibited the antibacterial activity toward many bacteria including P. aeruginosa 39,40 , the regulation of intracellular Sph levels is integral Scientific RepoRts | 6:38797 | DOI: 10.1038/srep38797 for bacteria exposed to Sph. According to this point of view, the study by Pewzner-Jung et al. is of interest. They reported that tracheal and bronchial epithelial Sph levels play important roles in the prevention of P. aeruginosa in healthy individuals and Sph levels are reduced in patients with cystic fibrosis due to reductions in acid CDase activity 41 . Collectively, SphR exerts two different physiological functions in P. aeruginosa, i.e., attacking host cells through the generation of Sph by the expression of CerN and defending against Sph through a decrease in intracellular Sph levels by the expression of SphABCD.
We found that Sph at a concentration of no less than 100 nM induced the transcription of cerN in P. aeruginosa (Fig. 3b). This concentration of Sph was markedly lower than its physiological concentration in human serum (~250 nM) 42 , and, thus, we speculate that the cerN gene is expressed in P. aeruginosa in some host environments. We isolated a CDase-producing bacterium, P. aeruginosa AN17, from the skin of patients with atopic dermatitis (AD) 10 and we demonstrated that the skin of these patients is significantly colonized by CDase-secreting bacteria, including P. aeruginosa, compared to that of healthy individuals 43 . The human epidermis contains a certain amount of Sph and dihydroSph 44 , which may induce the expression of cerN in P. aeruginosa-infected skin through SphR. We found that CerN hydrolyzes Cer in the human skin efficiently in the presence of anionic phospholipids derived from Staphylococcus aureus, which is found in atopic skin frequently, in place of detergents 45 . Recently, Oizumi et al. demonstrated that Sph produced by the action of CerN is converted to S1P and induces production of inflammatory mediators in human keratinocytes 46 . In addition, exogenous Sph, dihydroSph, and phytoSph are found to be modulators of differentiation and lipid metabolism in keratinocytes 47,48 . These observations indicate that CerN disturbs the skin barrier function by decreasing the levels of skin Cer and by increasing the long-chain bases. This study also suggests the important role of CerN as an exacerbating factor in P. aeruginosa-infected skin in conditions such as AD and related dermatitis. From another point of view, P. aeruginosa is an important opportunistic human pathogen, but is also found in diverse environmental conditions, such as soil and water. Because all eukaryotes and some bacteria possess sphingolipids as constituents of biological membranes, the bacterium may produce CerN through SphR, for decomposing SM or Cer from sphingolipid-producing organisms to obtain fatty acids as a nutrient source. In order to obtain a better understanding of the relationship between SphR and CerN in bacteria, we performed a blast search using the amino acid sequences of SphR and CerN. Among 56 bacteria with SphR homologues (E-value < e-30), we found that 12 bacteria (including P. aeruginosa) had highly homologous sequences of CerN (see Supplemental Fig. S2). Furthermore, their distribution was found in Gram-negative and Gram-positive bacteria, and frequently in the genus Mycobacterium. Mycobacterium are well-known pathogens towards humans and animals; however, other bacteria, except for Collimonas pratensis and C. fungivorans, which have both SphR and CerN, have also been identified as pathogenic bacteria towards humans and animals. Although the functions of SphR homologues currently remain unknown, the distribution of both proteins enables us to verify the importance of SphR in CerN production in pathogenic bacteria.
In the present study, we clarified the mechanism underlying the transcriptional regulation of cerN in P. aeruginosa, in which SphR functions as an Sph-inducible transcriptional regulator. Furthermore, we present detailed biochemical results on the interaction between Sph and SphR. The results described here may contribute to developing SphR inhibitors that have the potential to become anti-microbial drugs for SphR-possessing pathogens including P. aeruginosa. The AN17-BGT strain developed in this study is a useful tool for the high throughput (HST) screening of inhibitors of the interaction between SphR and the cerN promoter or SphR and Sph, which is easily performed by detecting β -galactosidase activity using artificial substrates. Bacterial strains and media. A type strain of P. aeruginosa PAO1 (JCM14847) was provided by the Japan Collection of Microorganisms, RIKEN BRC, which is participating in the National BioResource Project of MEXT, Japan. P. aeruginosa strains were cultivated in PY medium (0.5% polypeptone, 0.1% yeast extract, and 0.5% NaCl, pH 7.2) at 30 °C. E. coli strains were grown in Luria-Bertani (LB) medium at 37 °C. Media were supplemented with antibiotics such as ampicillin (100 μ g/ml for E. coli), carbenicillin (100 μ g/ml for E. coli), kanamycin (50 μ g/ml for E. coli and 1 mg/ml for P. aeruginosa), gentamycin (20 μ g/ml for E. coli and 30 μ g/ml for P. aeruginosa), and tetracycline (15 μ g/ml for E. coli and 200 μ g/ml for P. aeruginosa) where necessary.

Methods
Construction of AN17-BGT. All plasmids were constructed using PCR with PrimeSTAR MAX DNA polymerase (Takara Bio, Kusatsu, Japan) and the In-Fusion PCR cloning kit (Takara Bio). A 5-kbp fragment, which contained the 5′ -flanking region, open reading frame, and 3′ -flanking region of the cerN gene, was amplified by PCR using P. aeruginosa AN17 genome DNA as a template and the primers PaCD-5k-Infu1 and PaCD-5k-Infu2, and was then cloned into the BamHI site of pGEM-3Zf(+ ). The resulting plasmid was named pGEM-PaCD5k. In order to replace the cerN gene with the E. coli β-galactosidase gene, the cerN gene was removed by PCR using pGEM-PaCD5k as a template and the primers PaCD-5′ UTR and PaCD-3′ UTR. The fragment obtained was ligated with the β-galactosidase gene amplified from pGEM-LacZ using the primers PaCD-5′ UTR-LacZ-5′ and PaCD-3′ UTR-LacZ-3′ . The resulting plasmid was named pGEM-PaCD5k-LacZ. pGEM-PaCD5k-LacZ was linearized by PCR using the primers PaCD3′ UTR-3′ U and PaCD3′ UTR-5′ L to introduce the Tc r cassette into the 3′UTR region of the cerN gene in pGEM-PaCD5k-LacZ. The Tc r cassette was amplified by PCR using pBR322 as a template and the primers pBR322-U1-PaCD3′ and pBR322-L1-PaCD3′ , and was then ligated to linearized pGEM-PaCD5k-LacZ. The resulting plasmid was named pGEM-PaCD5k-LacZ-Tet. The cassette in pGEM-PaCD5k-LacZ-Tet was amplified by PCR using pGEM-PaCD5k-LacZ-Tet as a template and the primers PaCD Pro-5k-Infu1 and PaCD Pro-5k-Infu2, and was then ligated to the BamHI site of pK19mobsacB 50 . The sequences of the primers used in this study are summarized in Supplementary Table S1. The resulting plasmid was named pK19-PaCD5k-LacZ-Tet. The plasmid, which contains a mutant cerN disrupted with the β-galactosidase and Tc r genes, was transferred from the broad host range-mobilizing strain E. coli S17-1 51 to P. aeruginosa AN17 by biparental filter matings. Tc r plasmid integrants were selected on NAC medium containing Tc. Tc r colonies were then plated on LB agar containing 5% sucrose to identify strains that had lost the vector-associated sacB gene (resistant to sucrose). Gene replacement was ascertained by PCR of the cerN gene. The resulting strain was named AN17-BGT.
Construction of the AN17-BGT mutant library and screening of β-galactosidase activity-deficient cells. An EZ-Tn5 < R6Kγ ori/KAN-2> Tnp Transposome kit (Epicentre, Madison, WI, USA) was used to make the AN17-BGT mutant library according to the manufacturer's instructions. In order to obtain mutants, a 100-μ l aliquot of electrocompetent cells was mixed with 1 μ l of the EZ-Tn5 TM < R6Kγ ori/KAN-2> Tnp Transposome TM (Epicentre), and electroporation was then performed with a Gene Pulser II electroporator (Bio-Rad Laboratories, Hercules, CA, USA) in 0.2-cm cuvettes at 2.5 kV, 25 μ F, and 200 Ω. Cells were transferred to a 10-ml tube containing 2 ml of PY medium and incubated at 37 °C for 2 h with shaking. Transformants were selected on NAC agar containing 1 mg/ml kanamycin. The colonies obtained were added to the wells of a 96-well microplate containing 100 μ l of PY medium. On the next day, a 10-μ l aliquot of the culture was transferred to a new 96-well microplate containing 100 μ l of PY medium with 200 μ M SM from chicken egg yolk and incubated at 25 °C for 1 day. The β -galactosidase activity of each mutant was measured using CPRG. In order to permeabilize cells, 10 μ l of 0.1% SDS and 1 drop of CHCl 3 were added to each well and mixed well. The β -galactosidase assay was started by adding 50 μ l of 2x Z-buffer (120 mM Na 2 HPO 4 ·7H 2 O, 80 mM NaH 2 PO 4 ·H 2 O, 20 mM KCl, 2 mM MgSO 4 ·7H 2 O, and 100 mM β -mercaptoethanol) and 10 μ l of 1 mM CPRG. β -galactosidase activity was measured using a spectrophotometer at 570 nm.

Construction of the SphR mutant and complementation strains. The upstream and downstream
sequence fragments (1,000 bp) flanking the sphR gene were amplified by PCR using AN17 genome DNA as a template and the primers SphR-1000-Infu1 and SphR-1000-Infu2. The amplified product was cloned into the BamHI site of pK19mobsacB and the resulting plasmid was named pK19-SphR. In order to replace the sphR gene with the Tc r cassette, the sphR gene was removed by PCR using pK19-SphR as a template and the primers SphR-KO-InfuU1 and SphR-KO-InfuL1. The fragment obtained was ligated with the Tc r cassette amplified from pBR322 by using the primers pBR322U1new and pBR322L1331new. The primers used are summarized in Supplementary Table S1. The resulting plasmid was named pK19-SphR-Tet. The plasmid containing a mutant SphR disrupted with the Tc r gene was transferred from the broad host range-mobilizing strain E. coli S17-1 to P. aeruginosa PAO1 by biparental filter matings. Gene replacement was ascertained by PCR of the sphR gene. The resulting strain was named Δ SphR. In order to construct the SphR complementation vector, we first introduce the gentamicin resistance (Gm r ) cassette into the EcoRI site of pMMB66HE 52 . pMMB66HE was linearized by PCR using the primers pMMB66HE-U1 and pMMB66HE-L1. The Gm r cassette was amplified by PCR using pHP45Ωaac as a template and the primers aacC4-1-Infu1 and aacC4-1-Infu2, and was then ligated to linearized pMMB66HE. The resulting plasmid was named pMMB66HE-Gen. The sphR gene was amplified by PCR using pK19-SphR as a template and the primers SphR-U1-Infu1 and SphR-L1-Infu2, and was then ligated to the BamHI site of pMMB66HE-Gen, which was linearized by PCR using pMMB66HE-Gen as a template and the primers pMMB66HE-aacC4-U1 and pMMB66HE-aacC4-L1. The resulting plasmid was named pMMB66HE-Gen-SphR. pMMB66HE-Gen-SphR was transferred from the broad host range-mobilizing strain E. coli S17-1 to the strain Δ SphR by biparental filter matings. The primers used are listed in Supplementary Table S1.
Assay of hemolysis. Fifty microliters of the P. aeruginosa cell suspension (A600 = 0.4) in saline was mixed with 50 μ l of sheep erythrocytes in saline and incubated at 30 °C for an appropriate time. After being incubated, the reaction mixtures were centrifuged at 1,000 × g for 3 min, the supernatant was diluted with distilled water (3 volumes), and the release of hemoglobin was then measured using the Multiskan FC Microplate Photometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA) at 492 nm.
Assay of cell viability. The cell viabilities of the wild-type and SphR-deficient mutant were evaluated by Microbial Viability Assay Kit-WST (DOJINDO LABORATORIES, Kumamoto, Japan). This colorimetric microbial viability assay is based on the reduction of WST-8 reagent by the reducing ability of viable cells 53,54 . Strains were grown at 30 °C overnight on NAC agar. Bacteria were then scraped from the agar plate and suspended in saline to an A550 of 0.125. A 10-μ l aliquot of the cell suspension (A550 = 0.125) was inoculated into 180 μ l of PY medium containing 0.05% sodium taurodeoxycholate in the presence and absence of 200 μ M of SM and then incubated at 30 °C for an appropriate time. After the incubation, cell viability was analyzed by measuring the metabolic activity of cells using Microbial Viability Assay Kit-WST according to the manufacturer's instructions.
Real-time PCR. The WT and SphR-null mutant of P. aeruginosa PAO1 were cultured in PY medium with and without 200 μ M SM at 30 °C. After 8 h, cells were stabilized by RNAlater (Thermo Fisher Scientific) and harvested. Total cellular RNA was isolated from the cells using TRI reagent (Molecular Research Center, Inc., Cincinnati, OH, USA), and the SV Total RNA Isolation System (Promega Corporation, Madison, WI, USA). Reverse transcription was performed using the Prime Script RT reagent kit (Takara Bio) in accordance with the manufacturer's protocol. Real-time PCR was performed using the SYBR Premix ExTaq kit (Takara Bio) in an Mx3000P QPCR System (Agilent Technologies, Santa Clara, CA, USA). The data obtained were normalized against the expression levels of PA3001, which were not affected by the SM treatment, as shown in the microarray analysis. Relative changes in gene expression were analyzed using the ∆ ∆ CT method 55 . The sequences of the primer pairs are described in Supplementary Table S1.

Microarray analysis.
In microarray experiments, P. aeruginosa strain PAO1 was cultured in 2 ml of sterilized PY medium, and incubated at 30 °C for 1 day with vigorous shaking. An aliquot of the cell suspension (OD600 = 0.01) was inoculated into 2 ml of PY medium containing 0.05% sodium taurodeoxycholate and 200 μ M of SM, and then shaken at 30 °C for 8 h. Cells were stabilized by RNAlater and harvested. Total cellular RNA was isolated from cells using TRI reagent and the SV Total RNA Isolation System. The integrity of total RNA samples was assessed by the Experion Automated Electrophoresis System (Bio-Rad). Labeled cDNA probes were prepared according to the protocol supplied by the manufacturer (Affymetrix, Santa Clara, CA, USA) and hybridized to a GeneChip P. aeruginosa Genome Array (Affymetrix) according to the manufacturer's instructions. Data analysis was initially performed with Affymetrix Expression Console ™ Software. Microarray experiments and data analyses were supported by Cell Innovator Inc. (Fukuoka, Japan). Gene expression changes with significance were identified by Linear Models for Microarray Analysis (limma) package 56 of Bioconductor software. Fold changes were calculated as the average ratios between the signal in the absence (control) and presence of SM. Genes induced by 1.5-fold or more and a P value of < 0.05 were considered to be significant. Our results have been uploaded to the Gene Expression Omnibus database (accession number GSE83078).