High-Throughput Screening of Dipeptide Utilization Mediated by the ABC Transporter DppBCDF and Its Substrate-Binding Proteins DppA1-A5 in Pseudomonas aeruginosa

In this study, we show that the dppBCDF operon of Pseudomonas aeruginosa PA14 encodes an ABC transporter responsible for the utilization of di/tripeptides. The substrate specificity of ABC transporters is determined by its associated substrate-binding proteins (SBPs). Whereas in E. coli only one protein, DppA, determines the specificity of the transporter, five orthologous SBPs, DppA1–A5 are present in P. aeruginosa. Multiple SBPs might broaden the substrate specificity by increasing the transporter capacity. We utilized the Biolog phenotype MicroArray technology to investigate utilization of di/tripeptides in mutants lacking either the transport machinery or all of the five SBPs. This high-throughput method enabled us to screen hundreds of dipeptides with various side-chains, and subsequently, to determine the substrate profile of the dipeptide permease. The substrate spectrum of the SBPs was elucidated by complementation of a penta mutant, deficient of all five SBPs, with plasmids carrying individual SBPs. It became apparent that some dipeptides were utilized with different affinity for each SBP. We found that DppA2 shows the highest flexibility on substrate recognition and that DppA2 and DppA4 have a higher tendency to utilize tripeptides. DppA5 was not able to complement the penta mutant under our screening conditions. Phaseolotoxin, a toxic tripeptide inhibiting the enzyme ornithine carbamoyltransferase, is also transported into P. aeruginosa via the DppBCDF permease. The SBP DppA1, and with much greater extend DppA3, are responsible for delivering the toxin to the permease. Our results provide a first overview of the substrate pattern of the ABC dipeptide transport machinery in P. aeruginosa.


Introduction
Pseudomonas aeruginosa is a ubiquitous, opportunistic Gramnegative human pathogen that possesses high intrinsic drug resistance.Infections are difficult to eradicate since the bacterium rapidly acquires additional resistance mechanisms leading to multidrug or extensively drug resistance phenotypes.P. aeruginosa is the cause of hospital-acquired infections such as ventilator-associated pneumonia, bacteremia and urinary tract infections and may cause chronic infections in the lungs of cystic fibrosis patients [1][2][3][4][5].
Due to its metabolic versatility, P. aeruginosa is able to survive under harsh conditions in various environmental niches, where the uptake and metabolism of nutrients becomes crucial [1,6].
The capacity to import and utilize small peptides as carbon and nitrogen sources is a very common cellular function in eukaryotic and prokaryotic organisms [7][8][9].Furthermore, there are specific small peptides, which fulfill special biological functions as intracellular signals, antibiotics or toxins [6,10,11].In recent years, bacterial peptide uptake mechanisms regained interest in antibiotic development.The so-called ''Trojan-horse strategy'' has been employed to deliver peptide-coupled antibacterial substances into cells through the bacterial peptide transport machinery [12,13].Additionally, peptide transporters are pharmaceutically relevant due to their ability to take up peptidomimetics and other related therapeutic substances [14].
Studies about the physiology of peptide transport systems have been performed in the 1970s [15] and specific uptake of di-and oligopeptides has been investigated in Escherichia coli and Salmonella typhimurium decades ago [10,16].Homologous transporters from other bacteria such as Helicobacter pylori [17], Borrelia burgdorferi [18], Vibrio furnissi [19] and Lactococcus lactis [20] have only been characterized quite recently.
Most of these specific bacterial transport systems belong to the ABC transporter family.ABC-type transporter systems typically consist of two permease domains and various ATP-binding domains that are responsible for the energy supply [15].Translocation of molecules through the membrane is achieved by two hydrophobic transmembrane domains (TMD), usually driven by ATP hydrolysis.The direction of the translocation process can be deduced from the availability of so-called substrate-binding proteins (SBPs).While ABC transporters involved in the efflux of compounds do not possess SBPs, it is a necessary component of a functional ABC importer [15].SBPs are located in the periplasm and undergo a conformation change upon substrate binding.Due to this change, the bound substrate will be delivered to the integral membrane complex, released and further translocated into the cytoplasm [15,21,22].
It appears energetically more favorable to take up peptides instead of cleaving them in the extracellular space [23].Small peptides permeate the outer membrane of Gram-negative bacteria via non-specific porins [24].Once in the periplasmic space, they are further processed by proteolytic degradation such as the arginine-specific aminopeptidase [25] or they are bound by SBPs, which deliver them to their specific transporter [21].After translocation into the cytoplasm, membrane-associated peptidases hydrolyze the delivered peptides [26].
P. aeruginosa possesses more than 500 ORFs (about 10% of total ORFs) encoding proteins involved in the transport of nutrients and other small molecules [27].The dipeptide permease (Dpp) has been associated with the utilization of dipeptides as well as some tripeptides [9].The recognition of dipeptides has specific affinity to various L-amino acids depending on their side chains [16,21,23].Moreover, Dpp was also shown to mediate dipeptide chemotaxis [16,28] and haem metabolism [29,30] in other bacteria.During haem utilization, the SBP DppA was shown to bind haem and to deliver it to the ABC transporter [31].
Analysis of the sequenced genome of P. aeruginosa PA14 revealed two ABC-type transporter systems involved in the uptake of peptides: (i) an oligopeptide transporter, recently described to be involved in the uptake of uridyl peptide antibiotics [32], and (ii) a hitherto uncharacterized transporter system, which shows homology to the previously described dipeptide uptake machinery from other bacteria.In this study, we aimed to investigate and characterize the substrate profile of the putative dipeptide transporter system DppBCDF (PA14_58440-PA14_58490) from P. aeruginosa PA14.Investigation of peptide importers had been limited to the small number of available substrates.Our approach included the Biolog phenotype MicroArray (PM) technology to investigate diand tripeptides utilized as nitrogen source.For the first time, we show an extensive substrate profile comprising of almost 200 dipeptides and 14 tripeptides.Our analysis gives opportunity to assess the dipeptide transporter system of P. aeruginosa in much more detail as has been investigated before.A mutant of the dipeptide transport system DppBCDF showed reduced ability to utilize di/tripeptides as nitrogen source.Moreover, using this high-throughput approach, we unraveled the substrate profiles of five homologous SBPs, DppA1 (PA14_58350), DppA2 (PA14_58360), DppA3 (PA14_58390), DppA4 (PA14_58420), and DppA5 (PA14_70200), involved in peptide utilization via the DppBCDF transporter system.The emphasis of our analysis was on the properties of the amino acids side chains on peptide utilization to identify chemical groups which may be added to antibacterial compounds to promote their uptake by the dipeptide uptake machinery.

Computational analysis of the PA14 dipeptide transporter system
The dipeptide ABC transporter of PA14 is encoded in an operon-like structure containing four genes, two encoding the hydrophobic permeases PA14_58440 and PA14_58450 (dppC), and two encoding the hydrophilic nucleotide-and ATP-binding proteins PA14_58470 (dppD) and PA14_58490 (dppF) (Figure 1A).Using the protein sequence of PA14_58440, we identified orthologs of the dipeptide permease in Haemophilus influenzae (61% identity), E. coli (66% identity) and Burkholderia pseudomallei (74% identity), as depicted in the phylogenetic tree in Figure 2A (see also Table S3).Since the protein sequence of PA14_58440 shows a high degree of homology to other dipeptide ABC transporters, we suggest renaming this protein as DppB and refer to the operon as the dppBCDF operon.
Topology analysis of the two ABC-type permeases DppB and DppC using the TOPCONS prediction software [33] revealed the typical six a-helical transmembrane spanning domains (Figure S1).Further bioinformatical analysis indicated a promoter (s 32 ) in the dppB upstream region [34] and a single polycistronic dppBCDF transcript [35].
Analysis of the genome sequence upstream of the ABC transporter (Figure 1A) revealed the presence of a putative transcriptional regulator PsdR that appears to control the dppBCDF operon, and MdpA, a metallo-dipeptidase contributing to cytotoxicity in PAO1 [36].Additionally, several genes encoding periplasmic SBPs were identified: DppA1 (PA14_58350), DppA2 (PA14_58360), DppA3 (PA14_58390) and DppA4 (PA14_58420), whereas the later one appears to be co-transcribed in an operon with PA14_58410 encoding the porin OpdP (Figure 1A).Moreover, we identified a homologous SBP that is not located near the aforementioned SBPs, DppA5 (PA14_70200) (Figure 1B).Protein interaction modeling of the dipeptide transporter identified five SBPs interacting with the subunits of the transporter (Figure S2).
DppA1 and DppA3 are close paralogs (83% amino acid identity) and orthologues of DppA from E. coli, S. enterica and B. pseudomallei (50-60%) (Figure 2B).DppA2 and DppA4 are 59% identical to each other and show less than 60% amino acid sequence identity to the other dipeptide-binding proteins of PA14.DppA5 is not located near the dppBCDF operon.The DppA5 protein sequence shows about 50% identity to the other dipeptide-binding proteins of PA14 and appears to be an ortholog of BDL_1606 (bacterial extracellular solute-binding) and NikA (nickel-binding) from B. pseudomallei.An overview of the sequence identity data can be found in Table S4.Additionally, a graphical interpretation of the homology of the  2).Dipeptides containing positively charged amino acid residues were excluded from the analysis (see Materials and Methods).doi:10.1371/journal.pone.0111311.g003 Table 2. Total numbers of dipeptides utilized by PA14 (WT), the dppBCDF mutant (Dpp) and the SBP penta mutant (SBP) (see also Figure 3).Table 1.Effect of side chain properties of the terminal amino acid on utilization of dipeptides by PA14 (WT), the dppBCDF mutant (Dpp) and the SBP penta mutant (SBP).five SBPs is illustrated by the multiple sequence alignment in Figure S3.Investigation of the transcriptional profile of the dipeptidebinding proteins in PA14, using available RNA-seq data [35], revealed that the genes encoding DppA2 and DppA5 are expressed at a lower level than the genes encoding DppA1, DppA3 and DppA4 during growth at 28uC and 37uC, respectively.These data correlate with our observations from qRT-PCR analysis with PA14 grown at 37uC (data not shown).

Dipeptides used as nitrogen source by PA14
In order to investigate the substrate profile of dipeptides utilized by PA14, we took advantage of Biolog phenotype MicroArray plates that have numerous dipeptides implemented as nitrogen sources.This phenotype microarray approach allowed an extensive screening of the diversity of dipeptides utilized by PA14.Overall, 269 dipeptides (out of 400 possible combinations) are implemented in the Biolog plates PM6 to PM8, including 6 bpeptides containing an b-amino acid, which has the amino group bonded to the b-carbon, 2 c-peptides (both are Glu-Gly), and 12 dipeptides containing a D-form amino acid residue.Our analysis included a subset of 162 dipeptides and 13 tripeptides.Peptides that did not promote cell respiration above the threshold value of 2000 and/or showed a strong deviation from the other replicates were disqualified.
In total, PA14 was not able to use 36 dipeptides (18%) as nitrogen source.Ten of the excluded dipeptides contained a Dform amino acid residue.A full list of all excluded dipeptides is provided in the heat maps in Figure S4 and Figure S5.
The remaining set of dipeptides was classified based on the properties of the side-chain group of the single amino acid residues (acidic, basic, uncharged polar or nonpolar) and the ability of PA14 to utilize them as nitrogen source (Table 1, Figure 3).Since dipeptides contain two amino acids, we clustered each dipeptide into two groups.For example, if a dipeptide contains one nonpolar and one uncharged amino acid (N-or Cterminally attached), this dipeptide was clustered into the pool of 'nonpolar side-chains' as well as into the pool of 'uncharged sidechains'.This classification allowed us to identify whether specific side-chain groups are preferred by the dipeptide permease.Based on this grouping, we clustered 173 dipeptides as nonpolar, 43 as acidic, and 75 as uncharged.These clusters were then further divided based on the position of the amino acid residue in the  2).Basic amino acids were excluded from our analysis (see Materials and Methods).
We found that dipeptides containing uncharged residues, either at the N-or C-terminal end, were better substrates for PA14 (69 out of 75), than nonpolar (144 out of 173) and acidic ones (34 out of 43).Dipeptides with specific side-chains were used by PA14 with approximately the same efficiency irrespective of whether the specific side-chain was at the N-or C-terminal end of the dipeptide.

Substrate specificity of the ABC transporter DppBCDF
In order to investigate the substrate profile of the ABC transporter DppBCDF, a deletion mutant of the dppBCDF operon was constructed (Figure 1A).We used Biolog phenotype MicroArrays to elucidate the effect of the dppBCDF deletion on the ability of the mutant to utilize dipeptides.Dipeptides that did not support respiratory activity of the mutant indicated deficiency of the mutant to utilize the respective dipeptide as nitrogen source.Biolog respiration curve comparison between the wild type and the dppBCDF-deficient mutant for all PM plates are provided in Dataset S2.A direct comparison of utilized dipeptides in terms of the extent of respiratory activity is demonstrated in the heat map in Figure 4.
Numerous dipeptides could not be used as N-source by the mutant lacking the DppBCDF transporter system (as illustrated by the red colored squares in the heat map), identifying it as the main transporter system participating in the utilization of dipeptides.Overall, our data suggests that the transporter system is involved in the utilization of at least 66% of all dipeptides (107 out of 162) that are utilized by PA14 (Figure 3).The dppBCDF-deficient strain has a higher preference towards utilization of dipeptides containing acidic residues (29 out of 34), in contrast to dipeptides containing uncharged (40 out of 69) and nonpolar (95 out of 144) amino acids, respectively (Table 2).The utilization pattern of clustered dipeptides was not influenced by a N-or C-terminal localization of the amino acid residue.However, our data suggest that dipeptides coupled of uncharged-nonpolar (12 out of 22), nonpolar-uncharged (16 out of 29), and uncharged-uncharged (4 out of 9) residues are The number includes only dipeptides that were not used by the SBP penta mutant.doi:10.1371/journal.pone.0111311.t003 less efficiently taken up by the ABC transporter system (Table 1).
A closer look at single amino acid residues of the dipeptides revealed that only 1 out of 7 dipeptides that contained an Asn residue at the C-terminal end was utilized by the transporter (Figure 5).Moreover, only 2 out of 10 dipeptides containing Tyr at the N-terminal end could be used.Dipeptides containing Leu residues were poor substrates of the DppBCDF transporter.However, a localization of the Leu residue at the C-terminal end leads to a slightly increased utilization of these dipeptides by the ABC transporter.Moreover, less than 50% of dipeptides containing Ala or Gln at the N-terminal end were utilized by the transporter system (Figure 4, Figure 5).In contrast, almost all Thr-containing dipeptides were utilized regardless of their localization at the N-or C-terminal end.A similar observation was made with dipeptides containing Asp and Glu, which have negatively charged side chains.Several other dipeptides containing nonpolar amino acid residues such as Gly, Ile, Met, Pro, Thr, and Trp were good transporter substrates.While most of those residues were attached at the N-terminal end, Met and Pro showed higher preference of being utilized when residues were at the C-terminal end (Figure 5).

Role of substrate-binding proteins in dipeptides utilization
In order to investigate the role of the SBPs DppA1-A5 in dipeptide utilization, a penta deletion mutant, lacking all five SBPs, was constructed and investigated using Biolog phenotype MicroArrays.Biolog respiration curve comparison between the wild type and the penta mutant for all PM plates are provided in Dataset S2.A list of the excluded dipeptides is provided in the heat map in Figure S5.A comparison of the dipeptide utilization profiles of the wild type, the dppBCDF transporter mutant and the SBP penta mutant is shown as heat map in Figure 4.The utilization patterns of the transporter mutant and the SBP mutant are very similar.Both mutants are unable to utilize a wide range of dipeptides.
Our data suggest that about 52% of the dipeptides (84 out of 162) are recognized and delivered by the SBPs to the according permease (Table 2).The observed phenotypes indicate a higher preference towards dipeptides containing acidic residues (26 out of 34), in contrast to dipeptides containing uncharged (28 out of 69) and nonpolar (75 out of 144) amino acids (Table 2).
The substrate profiles of the SBP penta mutant and the dppBCDF transporter mutant overlap with only a few exceptions.The SBP penta mutant is able to utilize more substrates than the transporter mutant indicating that additional proteins might be involved in the delivery of substrates to the permease.Approximately 15% less dipeptides were recognized by the SBPs when compared to total substrate spectrum of the ABC transporter (Table 1).

Utilization of tripeptides as nitrogen source
The Biolog phenotype MicroArray plate PM8 contains 14 tripeptides as sole nitrogen sources.PA14 was able to use 13 tripeptides as nitrogen source (Figure 4).The tripeptide that PA14 could not use contains a D-form of alanine at the N-terminus (Figure S5).
Most of the tested tripeptides contain two glycine residues located either at the N-terminal or the C-terminal end.Tripeptides having two Gly residues at the N-terminal end are transported via DppBCDF and at least one is recognized by the SBPs.However, when the Gly-Gly is located at the C-terminal end, both the transporter mutant as well as SBP penta mutant showed still respiratory activity in the presence of the tripeptide, indicating that it is transported into the cells via a different transport mechanism or following proteolytic cleavage (Figure 4).

Substrate specificity of the dipeptide-binding proteins
In order to elucidate the substrate specificity of the five SBPs associated with dipeptide transporter, we complemented the SBP penta mutant with plasmids carrying individual SBPs.Comparisons of the Biolog respiration curves of the SBP penta mutant and of the penta mutant complemented with an individual SBP are provided in Dataset S2.A comparison of respiration promotion by the dipeptides is shown in the heat map in Figure 4. Excluded dipeptides are shown in Figure S5.We used qRT-PCR to determine the expression levels of the SBPs in the SBP penta mutant carrying complementation plasmids.As expected, expression of the SBPs was at higher levels in the complemented strains when compared to their native expression in the PA14 wild-type strain during early exponential phase: DppA1 (13-fold), DppA2 (11-fold), DppA3 (35-fold), DppA4 (24-fold), and DppA5 (18-fold).
The heat map of Figure 4 shows that dipeptides containing negatively charged (Asp or Glu) or nonpolar amino acids (Gly, Ile, Met, or Trp) did not support respiratory activity of the penta mutant complemented with the different SBPs or they enabled only a low level of respiration.
We observed that all SBPs are able to bind and deliver dipeptides with different side chains to the ABC transporter.DppA2 was able to recognize more dipeptides than any other SBP, about 15% of these dipeptides contain acidic residues (Figure 6).We observed the tendency that all SBPs, except DppA5, have a higher affinity for dipeptides containing acidic residues at the Cterminus than for dipeptides containing acidic residues at the Nterminus.The utilization of dipeptides was not strongly influenced by the type of amino acid residues located at the N-or C-terminal end.Although, it appears that DppA1, DppA3, and DppA4 prefer dipeptides with acidic residues at the C-terminal end (Table 3).Moreover, the SBP penta mutant was still able to utilize 118 clustered dipeptides as N source.In order to investigate whether the SBPs are able to increase the utilization of those peptides, we searched for dipeptides that increased the respiratory activity of the complemented strains at least 2-fold compared to the penta mutant.We found that DppA2 is able to increase the utilization of about 30% of all dipeptides, while the other SBPs were able to increase the utilization of less than 20% of dipeptides (Figure 7).
Furthermore, we determined the utilization efficiency of the different SBPs.Therefore, we compared the respiration of the penta mutant strains complemented with the different SBPs, considering differences more than 2-fold as significant.In total, we found almost 70 dipeptides showing a difference of at least 2-fold (Table 4).The strain expressing DppA2 was able to use 35 dipeptides more efficiently than the strains expressing other SBPs.
Moreover, almost all tripeptides were more efficiently used by strains expressing DppA2 and DppA4 than by strains expression DppA1, DppA3, and DppA5 (Figure 4).

Growth experiment with di/tripeptides as sole nitrogen source
Independent growth studies of batch cultures in minimal medium supplemented with di/tripeptides as sole nitrogen source were undertaken to confirm some of the respiration phenotypes of the strains from the Biolog phenotype MicroArray plates (Table 5).These growth assays confirmed most of the phenotypes of the PA14 strains, with some minor discrepancies with regard to the ability of specific substrates to act as sole nitrogen source.For example, the SBP penta mutant complemented with DppA3 showed a medium respiratory activity in wells of the Biolog plates containing the dipeptides Ala-Phe, Phe-Val, and Pro-Leu.However, the strain showed solid growth in minimal medium supplemented with these dipeptides.

Uptake of phaseolotoxin by the DppBCDF ABC transporter
The tripeptide phaseolotoxin [(N d-phosphosulfamyl)ornithylalanylhomoarginine] is produced by the plant pathogen P. syringae pv.phaseolicola, and shows phytotoxic and antibacterial activity.It has been demonstrated that oligopeptide transporters are responsible for the uptake of this toxin into bacterial cells [37].This prompted us to investigate whether P. aeruginosa PA14 is susceptible towards phaseolotoxin and whether the dipeptide transporter is involved in the uptake of phaseolotoxin.
Our results demonstrate that PA14 is sensitive to phaseolotoxin and that it is specifically taken up by the dipeptide ABC transporter DppBCDF.This finding is interesting since other bacteria use an oligopeptide transporter [37] and not a dipeptide transporter for the uptake of phaseolotoxin.
Furthermore, we elucidated which SBP is responsible for the delivery of phaseolotoxin to the ABC transporter.Our results demonstrate that DppA3 is involved in the transport of the toxin to the Dpp permease, as can be seen by the increased growth inhibition of the strain expression DppA3 caused by phaseolotoxin (Figure 8).In addition, DppA1 is also involved in the uptake of phaseolotoxin.

Discussion
Microorganisms need to take up nutrients from the environment as an inevitable consequence for survival and growth.Proteins, if available in adequate supply, are a major source of carbon and nitrogen molecules.However, due to the size limitation of molecules that can permeate the bacterial cell envelope, large proteins have to be degraded by extracellular proteases [38] before they can be utilized by the organism.
Once proteins are cleaved into small peptides and/or amino acids, they can be readily transported into the cell.In pseudomonads, uptake through the outer membrane is almost exclusively via specific porins [24].Once in the periplasm, specific transporter systems are required to translocate the molecules into the cytoplasm, where they are utilized as energy source [44].
The key objective of this study was to elucidate and characterize the specific uptake system of dipeptides in P. aeruginosa.We determined the substrate specificity of the dipeptide ABC transporter system DppBCDF and its associated substrate-binding proteins DppA1-A5.Therefore, we utilized the phenotypic MicroArray-based high-throughput screening method from Biolog that uses cellular respiration as a reporter.Using this approach, we were able to demonstrate that the DppBCDF transporter system is clearly responsible for the uptake and utilization of various di/tripeptides and we provide for the first time an extended overview of di/tripeptide substrates for this transport machinery.Additionally, we delineate a large pattern of various di/tripeptides that are recognized by the associated substrate-binding proteins.Such a high-throughput screening method to characterize the substrate specificity of the DppBCDF transporter system from P. aeruginosa has hitherto not been carried out.

The transport machinery
In order to study the substrate specificity of the dipeptide transporter system, we determined the dipeptide utilization in a mutant lacking the uptake machinery.However, our data revealed that several dipeptides could also be utilized by a dppBCDF-deficient mutant.When we examined these dipeptides, we found that specific aminopeptidases may contribute to peptide utilization by cleaving dipeptides into single amino acid residues.Hence, our screening approach had some restrictions.One example are Ala-containing dipeptides, where it has previously been reported that aminopeptidase N is involved in the degradation of Ala-Ala dipeptides [46].We found in our screening assay, that most Ala-containing dipeptides could be utilized by the dppBCDF-deficient strain; however, with lower efficiency compared to the wild type (Figure 4).This may correlate with the aforementioned involvement of peptidase N. Similar phenotypes were observed with dipeptides containing Leu, Tyr, Gln (Figure 4) as well as Arg, Lys, and His (Figure S5).
Nevertheless, the deletion of the dipeptide transporter abolished the ability of PA14 to use a broad range of dipeptides, which could otherwise serve as energy source (Figure 4).It has been reported that the Dpp transporter system is able to recognize and transport various dipeptides unrelated to their side chains [16].Our screening approach provides an overview of almost 200 dipeptides with different side chains that are translocated via the dipeptide permease.Our data suggest that the transporter system prefers acidic side chains over nonpolar or uncharged ones (Table 2).In general, it appears that there is no clear preference for side chains located at the N-or C-terminal end.Although some amino acid residues might be preferred at a specific position, such as Gly, Ile, Thr, and Trp at the N-terminal end and the hydrophobic Met and Pro at the C-terminal end (Figure 5).Our screening approach also showed that dipeptides composed of uncharged and nonpolar residues are poor substrates of the DppBCDF transporter (Table 1).
With the few dipeptides that contain a D-form amino acid and are present in the Biolog plates, it became apparent that the sterochemical preference of the dipeptide transporter is for L-form residues.Almost none of the D-form dipeptides could be utilized by PA14 (Figure S5), a result in accord with previous studies in E. coli [16,21,47].
Several studies have shown that dipeptide ABC transporters from other bacteria are able to transport also other compounds beside dipeptides.It has for example been demonstrated that a dipeptide transporter mutant of Helicobacter pylori was not able to utilize dipeptides as well as hexa-and nonapeptides [17].The heme precursor 5-aminolaevulinic acid is a specific substrate of the dipeptide transporter from E. coli and S. typhimurium [23,48].Moreover, ions such as nickel have been associated with dipeptide transporter systems [49].The eukaryotic di/tripeptide transporter Dal5p from Saccharomyces cerevisiae is able to transport the nitrogenous compound allantoate/ureidosuccinate [50].Interestingly, allantoin, the none-hydrolyzed form of allantoate, appears to be transported by the DppBCDF system in P. aeruginosa as observed in our screening (Figure S5).

The periplasmic delivery system
Substrate-binding proteins accomplish the recognition and binding of substrates in the periplasm and determine the substrate specificity of the associated ABC importer.The amount of SBPs varies among organisms.For example the dipeptide ABC transporter of E. coli possesses only one SBP, while Pseudomonas putida has two, and Pseudomonas fluorescence three SBPs [36].
The question arises why does the P. aeruginosa ABC dipeptide transporter possess even five SBPs?Do multiple SBPs have different binding specificity, and therefore broaden the specificity of the ABC transporter?If these SBPs occurred through gene duplication, they might have acquired a novel cellular function, thus providing additional capability to utilize nutrients.Since duplications will also share some common features, multiple SBPs would be able to complement each other in case of mutational inactivation or gene loss.Hence, we hypothesized that the five present SBPs may be biologically relevant in terms of the recognition of specific di/tripeptides, and therefore increase peptide transport efficiency and optimize nutrient utilization.
To study the impact of each SBP, we constructed a penta mutant of PA14 lacking all five SBPs, complemented the mutant with individual SBP and used these strains in our Biolog screening assay.We retrieved extensive information about the substrate specificity.However, our data suggests that several SBPs have no clear preference for dipeptides with a specific side chain.It became apparent that some dipeptides can be utilized with different affinity by various SBPs.A possible reason may be the fact that the binding proteins recognize rather the peptide backbone than the connected side chains [51].This is in contrast to specific amino acid receptors, which specifically recognize the side chain [52].With respect to tripeptides tested in our assay, we identified DppA2 and DppA4 to be more efficient in tripeptide usage than the other SBPs (Figure 4).
SBPs tolerate binding of structurally different dipeptides by recognizing the backbone of the dipeptide while leaving enough space for a variety of side chains [53].The structure of SBPs consists of two domains that are connected by two or three strands, which act as hinge, forming a pocket-like structure.The actual binding site is located in the groove between the two domains and appears to be large enough to bind any side-chain [52,53].However, SBP specificity can change by variation of the structure of the active site in the side chain-binding pockets [54].Specific acidic and basic residues inside the pocket interact with the N-and C-terminal end of peptides by formation of salt bridges [55].Once the peptide is bound, the SBP will close itself by bending the hinge, thus engulfing the peptide.An interaction with the substrate-binding domain of the permease, and subsequent initiation of nutrient translocation, is only possible in a closed form [52,53].
The transfer of the substrate to the permease might be another restriction that influences the specificity.It has previously been demonstrated that ABC transporters have one or more substratebinding domains in close proximity to the translocation pore, which interact with SBPs.In our assay, the penta mutant lacking all five SBPs, was still able to utilize more dipeptides than the DppBCDF transporter mutant indicating that additional proteins are involved in the transport of the dipeptides or that some of the dipeptides can be directly transported by the ABC transporter.This has been demonstrated for maltose when the transporter machinery was mutated [56].
It has previously been reported that the presence of multiple, independently regulated SBPs increases the substrate spectrum of the associated transporter and might be involved in the adaptation of the bacterium to changing environmental conditions [20,[57][58][59].The uptake of specific peptides can trigger bacterial adaptations such as chemotaxis and quorum sensing [16].In this connection it has been shown that DppA1 is a quorum sensingrelated gene in P. aeruginosa [60].DppA3 has been demonstrated to be iron-regulated with increased expression under iron-limiting conditions [61].Moreover, mutants defective in DppA1, DppA2, and DppA4 were found to exhibit poor biofilm formation when compared to the wild-type strain [48].
Using the available RNA-seq data for P. aeruginosa, we found that DppA2 and DppA5 are only expressed at a low level [35].Whether the expression of DppA2 and DppA5 is induced under specific conditions, needs to be addressed in further experiments.Studies in Borrelia burgdorferi demonstrated that the SBP Table 6.Bacterial strains used in this study.

Strain
Relevant characteristics or genotype a
Further experimental studies in a host-related environment may identify the specific induction conditions of the dipeptide-binding proteins of P. aeruginosa PA14.The transcription of DppA4 in an operon with the outer membrane porin OpdP might hind towards a more specific role of this SBP.The porin OpdP was previously shown to be involved in the transport of dipeptides as well as of single amino acids in PAO1 [24].Overall, our data provide insights into the substrate specificity of the five SBPs from P. aeruginosa.Individually regulated SBPs, activated under certain conditions, may give an advantage in adapting to various environmental conditions.We have shown that four of the five SBPs have overlapping substrate profiles, where DppA2 shows the highest flexibility, tolerating the binding of more peptides than the other SBPs.This is interesting since it appears that DppA2 is only expressed at a low level under in vitro growth conditions.Moreover, we found that the various SBPs of the dipeptide transporter can complement each other, while it has been shown that the periplasmic binding proteins OppA and OppB of the oligopeptide transporter from P. aeruginosa cannot compensate for each other.Both SBPs were required for uptake of uridyl peptide antibiotics, indicating that OppA and OppB may form an oligomeric binding protein [32].

Fraud translocation
Peptide transporters have the ability to recognize and transport a broad range of substrates.Hence, they became attractive targets for therapeutic agents [63].In nature, various microorganisms produce antimicrobial substances in order to inhibit the growth of competing organisms.Often such compounds are peptide analogues and have been demonstrated to penetrate the cell via peptide transporters [9,37].It has recently been shown that the oligopeptide transporter OppCDE of P. aeruginosa PAO1 is required for uptake of uridyl peptide antibiotics and for import of the tripeptide herbicide bialaphos [32].
Phaseolotoxin, a highly charged, phytotoxic tripeptide produced by Pseudomonas syringae pv.phaseolicola, is able to inhibit the enzyme ornithine carbamoyltransferase (OCTase) of various organisms [37,64].Previous studies have demonstrated the toxic effect in E. coli and S. typhimurium strains, where it was shown that the toxin gets access to the cell through the uptake via an oligopeptide permease [37].The closest homolog of the oligopeptide transporter from E. coli in PA14 is the dipeptide transporter DppBCDF.
To our knowledge, there are no data available about the toxic effect of phaseolotoxin on P. aeruginosa.It has been suggested that it is inactive against many Pseudomonas spp.[37].However, we found that the toxin is able to inhibit growth of P. aeruginosa PA14 and we were able to demonstrate a lack of toxicity in the DppBCDF transporter mutant as well as in the SBP penta mutant (Figure 8).Moreover, we identified that the SBPs DppA1, and with much greater extend DppA3, are able to deliver the toxin to the permease.Interestingly, these two proteins show also the highest sequence similarities to the E. coli OppA protein.

Bacterial strains, plasmids and growth conditions
All used nucleotide sequences were based on the genome of P. aeruginosa UCBPP-PA14 (GenBank: NC_008463.1)available from NCBI, hence referred to as PA14 in this manuscript.Bacterial strains are listed in Table 6 and plasmids in Table S1.
P. aeruginosa strains were cultured at 37uC in Luria Bertani (LB) broth and Minimal Medium P (MMP) [25].Selection for transformed pseudomonads was achieved on King's B medium.E. coli XL-1 Blue was used as cloning host.E. coli ST18 was used for biparental mating where the medium was supplemented with 50 mg/ml 5-aminolevulinic acid (ALA).E. coli cells were routinely maintained at 37uC in dYT medium, except when strains were used that contained the FLP recombinase (30uC).

PCR amplifications and DNA modifications
PCR primers are listed in Table S2.Screening PCR reactions were carried out using DreamTaq DNA polymerase (Thermo Scientific) in accordance with the manufacturer's instructions and optimized annealing temperature for each primer set.For screening PCR reactions performed with PA14, bacterial cells were boiled at 95uC for 5 min and subsequently pelleted at 13,000 rpm for 1 min.PCR reactions were supplemented with additional 5% DMSO.For high fidelity PCR reactions, Phusion DNA polymerase (Thermo Scientific) or Q5 DNA polymerase (NEB) were used.
Restriction digestions were performed using Thermo Scientific restriction enzymes according to the manufacturer's instructions at the appropriate temperature.All ligation reactions were carried out at room temperature using Thermo Scientific T4 DNA ligase.
DNA purifications were either performed using the GeneJET PCR purification kit (Thermo Scientific) or the GeneJET Gel extraction kit (Thermo Scientific) following the manufacturer's instructions.

Construction of the PA14 dppBCDF knockout mutant
The construction of the knockout vector was based on the protocol described by Zumaquero et al. [65].Briefly, approximately 500 bp flanking the 59 and 39 regions of the dppBCDF dipeptide ABC transporter operon were PCR-amplified using the primer pairs 58420-A1/58420-A2 and 58490-B1/58500-B2.The T7 primer sequence was incorporated in primers A2 and B1 to provide homology and a KpnI restriction site between the fragments.After amplification, the obtained fragments were gelpurified and approximately 40 ng of each fragment was used in a polymerization PCR reaction with primers A1 and B2.The resulting fusion product was gel-purified, further ligated into the pGEM-T easy vector and verified by sequencing.
A gentamicin cassette, coupled to GFP and flanked by Flp-FRT sites, was cut from plasmid pPS858-Eco and subsequently inserted into the KpnI digested pGEM construct.The deletion allele was cut and further ligated into EcoRI-digested pEX18Ap, yielding the final replacement plasmid pEX18Ap.ABC58-ko.
The generation of the PA14 dppBCDF mutant was based on the published procedure from Schweizer and Hoang [66].Briefly, chromosomal deletions were obtained by conjugational transfer of the gene replacement vector into PA14.Therefore, bacteria, grown overnight on agar plates, were scratched from the plate, resuspended in 1 ml of sterile water, and adjusted to an OD 600 of 0.1.100 ml of the E. coli ST18 donor strain, harboring the replacement vector, was mixed with 200 ml of the P. aeruginosa strain.The complete mixture was spotted (50 ml per spot) on a KB agar plate containing ALA and allowed to grow overnight.On the next day, bacteria were scratched from each spot, pooled, resuspended in 1 ml sterile water, and serially diluted.100 ml of each dilution was spread on KB plates containing the appropriate antibiotic.On the next day, single colonies were picked on KB plates containing 5% sucrose to counter-select mutations from single crossovers.Remaining cells were pooled and re-streaked on 5% sucrose plates.From the next day onwards, putative mutants were screened for homologous recombination events by testing their antibiotic resistance on KB plates containing Gm and Cb, respectively.In order to confirm gene deletion through a double crossover event in Gm-resistant and Cb-sensitive colonies, primers were designed, which bind up-and downstream of the operon.PCRs were done using these locus-specific primers with primers binding within the Gm-GFP cassette.
The Gm-GFP-FRT cassette was finally excised using the plasmid pFLP2 that carries the FLP recombinase gene [67,68].Briefly, plasmid pFLP2 was conjugationally transferred into Gm-resistant mutants and selected at 30uC on KB plates containing Cb. Single colonies were selected and tested on agar plates containing Gm to confirm successful excision of the Gm-GFP cassette.A PCR with the locus-specific primers was performed.Sequencing of the obtained PCR fragment confirmed loss of the dppBCDF operon.Subsequently, loss of plasmid pFLP2 was achieved by incubating the mutants on 5% sucrose-containing agar plates.Loss of pFLP2 plasmid was confirmed by incubating the mutants on plates containing Cb.

Construction of the substrate-binding protein (SBP) penta mutant of PA14
In order to delete the five SBPs (DppA1-DppA5) of the dipeptide transporter DppBCDF, a step-wise deletion of each SBP was accomplished.Therefore, we used a gene replacement strategy to create unmarked mutants, which had previously been described [67].Briefly, flanking regions of the SBP PA14_58350 (DppA1), PA14_58360 (DppA2), PA14_58390 (DppA3), PA14_58420 (DppA4) and PA14_70200 (DppA5) were PCR amplified (Table S2) and gel-purified.By following the same procedure as described above, the obtained fragments were fused in an overlapping PCR reaction.Next, the fusion fragments were cloned into the suicide vector pEX18Gm via EcoRI/HindIII restriction sites, verified by sequencing, mobilized into PA14 and screened for double crossover events.

Construction of SBP overexpression plasmids
Computational promoter prediction was based on the available RNA-seq dataset for P. aeruginosa [34] and the prokaryote promoter prediction tool PePPER [34].The full ORFs including their promoter region were PCR amplified, cloned into pBBR1MCS-5 [69] and subsequently sequenced to rule out any possible mutations that might have occurred during PCR.To obtain expression from the native promoter, all constructs were cloned in opposite direction of P lac .
Briefly, a 1974-bp fragment containing dppA1, including the upstream promoter region, was amplified using primer SBP+ Pro_58350_fwd(SacI)/SBP+Pro_58350_fwd(ApaI) and cloned into SacI/ApaI-digested pBBR1MCS-5, yielding pBBR5.SB P58350-Pro.Construction of pBBR5.SBP58360-Pro, pBBR5.SBP58390-Pro and pBBR5.SBP70200-Pro was performed similarly.Since dppA4 is found in an operon with the outer membrane porin PA14_58410, we fused the promoter region of the operon to the coding sequence of dppA4 by an overlap PCR [70].Briefly, the 241-bp promoter region was amplified using primers SBP+Pro_58420_fwd(XbaI)/SBP+Pro_58420_rev and the 1605-bp dppA4 gene was amplified using primers SBP+ Pro_58420_fwd/SBP+Pro_58420_rev(ApaI).The reverse primer of the promoter region and the forward primer of the gene contain a 22-bp overlap.Fragments were PCR-amplified and purified.Next, 6 ng of the promoter region and 40 ng of the dppA4 gene were used to perform an overlapping PCR.The amplified product was gel-purified and an 1846-bp fragment was cut from the gel, digested with XbaI and ApaI, and subsequently cloned in XbaI/ApaI-digested pBBR1MCS-5, yielding plasmid pBBR5.SBP58420-Pro.All obtained constructs were verified by sequencing before they were mobilized into PA14 via biparental conjugation.

RNA isolation and quantitative real-time PCR
Bacterial cultures were grown in KB broth until an optical density of 0.5.An aliquot containing 15610 9 CFU (equivalent of 15 ml OD 600 of 1.0) was transferred to 15 ml killing buffer [71] and centrifuged for 20 min at 4000 rpm.The supernatant was decanted and the pellet frozen at 280uC.
Total RNA was isolated using the GeneJET RNA Purification Kit (Thermo Scientific) following the manufacturer's instructions.The obtained RNA was DNAse-treated (Ambion/Life Technologies) and subsequently checked for purity by gel electrophoresis and determination of the A 260 /A 280 and A 260 /A 230 ratios using a Nanodrop ND-2000 spectrophotometer (Thermo Fischer Scientific).High quality RNA was reverse transcribed and amplified with the OneStep RT-PCR Kit according to the manufacturer's protocol (Qiagen).Template RNA (5 ng) was used in a standard 25-ml qRT-PCR reaction with specific primers (Table S2).As control, RNA samples without reverse transcriptase were included to detect possible DNA contaminations.
For analysis, a Mastercycler ep realplex 2 gradient S (Eppendorf, Hamburg, Germany) was used.Cycling parameters included a 15 min initial denaturation at 95uC to activate the DNA polymerase followed by 40 cycles consisting of 15 sec at 95uC, 30 sec at 55uC and 30 sec at 72uC.The final step consisted of 1 min at 95uC and 30 sec at 55uC.A melting curve analysis with a temperature ramp from 25uC to 95uC in 20 min was performed at the end of each run to determine specificity of amplified qPCR products.Each sample was analyzed for gene expression in triplicate.Quantification of mRNA transcripts was performed by the comparative C t method.Briefly, the C t values of the samples of interest were compared with a non-treated sample.All C t values were normalized to the rpsL gene [72].The comparative C t method was calculated by 2 2(DCt, sample2DCt, reference) , where DC t was normalized to the rpsL gene.Subsequently, fold-changes between the samples were determined based on the calculated C t method.

Phaseolotoxin agar diffusion assay
Phaseolotoxin was obtained from supernatants of stationary phase grown P. syringae pv.phaseolicola 6/0 cells [73].Cells were grown in 5b medium (2.6 g KH 2 PO 4 , 5.5 g Na 2 HPO 4 , 2.5 g NH 4 Cl, and 1 g Na 2 SO 4 , 10 g glucose, 0.1 g MgCl 2 ?6H 2 O and 0.01 g MnSO 4 ?4H 2 O per liter [74]) at 18uC for 48 h and harvested by centrifugation at 4000 rpm for 20 min.The supernatant containing phaseolotoxin was used in an agar diffusion assay to evaluate the toxic activity of phaseolotoxin against PA14 and its mutant strains.
For preparation of the test plates, P. aeruginosa strains, grown on 5b medium (or MMP) at 37uC overnight, were resuspended and adjusted to an OD 600 of 1.0 in sterile water.500 ml of the cell suspension was added to 50 ml of 5b agar medium warmed to 50uC (or MMP).After preparing the plates, 7 mm holes were punched into the agar and 50 ml of the phaseolotoxin supernatant was added into the holes.Plates were incubated for 24 h at 37uC and subsequently visually analyzed in terms of growth inhibition zones on the bacterial lawn.

Phenotype MicroArrays (Biolog)
The Biolog system for Phenotype MicroArrays (PMs) allows the measurement of bacterial phenotypes by monitoring cellular respiration using the irreversible reduction of tetrazolium dye to the purple compound formazan [75].We utilized plates PM6, PM7 and PM8 to monitor catabolism of the implemented peptides as nitrogen sources by PA14 and its mutant strains (see Dataset S1 for details).
Prior to inoculation into PM plates, bacteria were grown on BUG agar plates overnight at 37uC.Using a sterile swab, cells were transferred into a sterile capped tube containing 16 ml of 1x IF-0 inoculating fluid and adjusted to a transmittance of 42% using a Turbidimeter (Biolog).15 ml of the prepared cell suspension was added to 75 ml of 1x IF-0 and dye mix A to obtain a final transmittance of 85%.Additionally, for inoculation of plates PM6 to PM8, 2 M sodium succinate/200 mM ferric citrate (100x) was added to the suspension to obtain a 1x concentration.100 ml of the cell suspension was inoculated into each well of the PM microplates.All plates were placed in the OmniLog reader (Biolog) and incubated for 24 h at 37uC.The OmniLog reader takes pictures of each plate at a 15-min time interval and converts the pixel intensity of each well into a signal value.
As a control, we incubated plates PM6, PM7 and PM8 with IF-0 media and dye mix A, but without bacteria.No significant dye reduction was observed without bacteria.

Growth experiments with di/tripeptides as sole nitrogen source
We performed growth experiments with selected di/tripeptides as sole nitrogen source to validate that peptides that promoted bacterial cell respiration in the Phenotype MicroArrays promoted actually growth of the bacterium.Briefly, overnight cultures grown in Minimal Medium P (MMP) [25] were harvested by centrifugation and washed twice with phosphate-buffered saline (PBS) to remove any nitrogen source.Next, the cell suspension was adjusted to an OD 600 of 0.05 with MMP supplemented with 2 mM of a di-or tripeptide as sole nitrogen source.Cultures were incubated at 37uC for 24 h in a shaking incubator at 230 rpm.Growth of the bacteria was determined using a spectrophotometer to measure the OD 600 .The experiment was repeated three times.

Data analysis of the di/tripeptide utilization
Data were collected and analyzed with the OmniLog software and subsequently exported to Microsoft Excel [76].All experiments were performed by a minimum of three replicates.Obtained data that were used in this study are provided in Dataset S1 (Excel).
Outliers among all replicates were identified by means of the standard deviation.Based on the difference of the standard deviation to the average value, we defined a threshold maximum of 20% deviation to be a good replicate.Samples with deviations between 20-50% were considered as acceptable.A deviation of more than 50% disqualified the sample from further analysis.
Respiratory activity of the cells in the single wells of the PM plates was quantified by a kinetic plot of color formation against time.The resulting area of each plot was represented by a specific data value.The threshold of this area value was set to 2000 (approximately 15-20% of the average area value of the positive control of all plates).In order to assist visualization of the data, we created a heat map based on data values representing the area under the respiration curve (subdivided into steps of 2000).Image comparisons of signal curves are provided in Dataset S2.
Next, the average values of all replicates were exported to Dataset S3 (Access) [76] in order to facilitate data retrieval and investigation of dipeptides in more detail.For every exported strain, one table was generated where the PM plate and the according well formed the primary key.Additionally, a table 'chemical' was generated, with the same primary keys as above, containing the chemical per well as well as every di/tripeptide split into single amino acid residues, their single core structure (alpha, beta, gamma) and form (L-or D-).Another table 'AA', with aa as primary key, was used to assign every amino acid to their specific side-chain residue.
Homann et al. [12] recommended to omit peptides containing lysine, histidine, and cysteine residues since they appear to support reduction of the dye in the presence of cells, but in the absence of growth [12]{Homann, 2005 #25}.Furthermore, the PM6, PM7 and PM8 plates contain only two cysteine-containing dipeptides (Cys-Gly and Gly-Cys), which supported only very low or no dye reduction by the PA14 strains.Hence, we excluded those peptides from our data analysis.Furthermore, due to the arginine-specific aminopeptidase of PA14 [25], we excluded dipeptides containing N-and C-terminal arginine residues.

Multiple sequence alignment and phylogenetic analysis
Protein sequence data were obtained from NCBI and the Pseudomonas Genome Database [77].Putative dipeptide ABC transporter proteins from P. aeruginosa PA14 were identified on basis of homology with already described transporters from other organisms found in the Transporter Classification Database [78] and respective BLASTP searches.A similar approach was used for the identification of substrate-binding protein homologs in PA14.Multiple sequence alignment was performed using ClustalO [79] and phylogenetic trees visualized by FigTree [80].

Figure 1 .Figure 2 .
Figure 1.Schematic presentation of the genomic region of the P. aeruginosa PA14 dipeptide transport machinery.(A) Region surrounding the dipeptide transporter operon dppBCDF and its substrate-binding proteins dppA1-A4.The 3,952-bp deletion of the ABC transporter dppBCDF operon is indicated.(B) Genomic region surrounding dppA5.The subcellular location of the proteins is indicated by colors.Black arrows indicate operon structures [77].doi:10.1371/journal.pone.0111311.g001

Figure 3 .
Figure 3. Global utilization pattern of dipeptides by PA14, the dipeptide transporter DppBCDF and the SBPs DppA1-A5.The rings show the total number of dipeptides utilized by the wild-type strain PA14 (outer purple ring), transported by the dipeptide transporter system DppBCDF (middle blue ring), and recognized by the substrate-binding proteins (inner dark green ring).The oval rings (green) present different pools of dipeptides (nonpolar, acidic, uncharged) containing a specific amino acid side-chain either at the N-or C-terminal end of the dipeptide (see Table2).Dipeptides containing positively charged amino acid residues were excluded from the analysis (see Materials and Methods).doi:10.1371/journal.pone.0111311.g003

Figure 4 .Figure 5 .
Figure 4. Heat map of di/tripeptide utilization by PA14, the DppBCDF transporter mutant, the SBP penta mutant DppA1-A5, and by strains of the penta mutant complemented with individual SBPs.Each square represents the average respiratory activity of a strain in one well of the Biolog Phenotype MicroArray plates.The heat map is based on the values reflecting the extent of respiration after 24 hours at 37uC.Values exceeding 8000 reflect solid respiratory activity during the assay.Values below 2000 were considered as no respiratory activity.doi:10.1371/journal.pone.0111311.g004

Figure 6 .
Figure 6.Utilization of dipeptides by strains of the SBP penta mutant complemented with individual SBPs.The numbers in parentheses indicate the total number of dipeptides analyzed.There are no significant differences among the complemented SBPs DppA1 to DppA4 as determined by single factor ANOVA. DppA1 to DppA4 show significant difference to DppA5 (p,0.05) as determined by a two-sided t test with equal variances.doi:10.1371/journal.pone.0111311.g006

Figure 7 .
Figure 7. Effect of the chemical nature of amino acid residues located either at the N-or C-terminal end of dipeptides on the utilization by strains of the SBP penta mutant complemented with individual SBPs.The numbers indicate dipeptides that enabled a more than 2-fold increase in respiratory activity of SBP mutant complemented with an individual SBPs compared to the respiratory activity of the mutant carrying an empty plasmid.The SBP penta mutant was able to utilize these dipeptides.doi:10.1371/journal.pone.0111311.g007

Table 4 .aTable 5 .
Differences in dipeptide utilization efficiency of the SBPs DppA1-DppA4 determined by differences in total growth a .DppA5 was omitted since it showed a similar phenotype as the SBP penta mutant.b Number of dipeptides where strains complemented with different SBPs showed at least 2-fold differences in respiration.doi:10.1371/journal.pone.0111311.t004Growth of PA14, the SBP penta mutant, and strains of the penta mutant complemented with individual SBPs in minimal medium supplemented with di/tripeptides as the sole nitrogen source a .omitted since it showed a similar phenotype as the SBP penta mutant in the Biolog Phenotype MicroArrays.bAverage respiratory activity of a strain in one well of the Biolog Phenotype MicroArray plates after 24 hours.2, ,2000 (no respiratory activity); +/2, 2000-4000; +, 4000-6000; ++, 6000-8000; +++, .8000(solid respiratory activity).c Optical densities were determined after 24 h of growth in Minimal Medium P supplemented with 2 mM of a di/tripeptide or NH 4 Cl as sole nitrogen source.d Boldface numbers indicate a significant difference between the extent of respiratory activity in Biolog Phenotype MicroArray and the extent of growth in minimal medium supplemented with di/tripeptides as the sole nitrogen source.doi:10.1371/journal.pone.0111311.t005

Figure
Figure S1 Transmembrane domain analysis of the dipeptide transporter permease DppB and DppC.The upper line indicates the predicted topology from TOPCONS [33] based on amino acid sequences.Red lines indicate an inner membrane orientation; blue lines indicate an outer membrane orientation; grey boxes indicate transmembrane helices spanning from the inside to the outside; white boxes indicate transmembrane helices spanning from the outside to the inside.Below the line is a graphical interpretation of the reliability of the prediction for each amino acid.(TIF) Figure S2 Protein interaction model of the DppBCDF dipeptide transporter and its substrate-binding proteins from P. aeruginosa PAO1 predicted by the STRING database [81].The protein sequences from PAO1 are homologous to the following proteins from PA14: DppBCDF (PA4503-PA4506), DppA1 (PA4496), DppA2 (PA4497), DppA3 (PA4500), DppA4 (PA4502), and DppA5 (PA5317).Another ABC transporter system (PA2060-PA2061), homologous to PA14_37840, appears to interact with the dipeptide transporter network.Blue lines connecting the node spheres predict a physical or functional interaction between the proteins.(TIF) Figure S3 Multiple sequence alignment of the amino acid sequences of DppA1, DppA2, DppA3, DppA4, and