Elucidating the Pseudomonas aeruginosa Fatty Acid Degradation Pathway: Identification of Additional Fatty Acyl-CoA Synthetase Homologues

The fatty acid (FA) degradation pathway of Pseudomonas aeruginosa, an opportunistic pathogen, was recently shown to be involved in nutrient acquisition during BALB/c mouse lung infection model. The source of FA in the lung is believed to be phosphatidylcholine, the major component of lung surfactant. Previous research indicated that P. aeruginosa has more than two fatty acyl-CoA synthetase genes (fadD; PA3299 and PA3300), which are responsible for activation of FAs using ATP and coenzyme A. Through a bioinformatics approach, 11 candidate genes were identified by their homology to the Escherichia coli FadD in the present study. Four new homologues of fadD (PA1617, PA2893, PA3860, and PA3924) were functionally confirmed by their ability to complement the E. coli fadD mutant on FA-containing media. Growth phenotypes of 17 combinatorial fadD mutants on different FAs, as sole carbon sources, indicated that the four new fadD homologues are involved in FA degradation, bringing the total number of P. aeruginosa fadD genes to six. Of the four new homologues, fadD4 (PA1617) contributed the most to the degradation of different chain length FAs. Growth patterns of various fadD mutants on plant-based perfumery substances, citronellic and geranic acids, as sole carbon and energy sources indicated that fadD4 is also involved in the degradation of these plant-derived compounds. A decrease in fitness of the sextuple fadD mutant, relative to the ΔfadD1D2 mutant, was only observed during BALB/c mouse lung infection at 24 h.

Phosphatidylcholine (PC), the major component of lung surfactant [12], was suggested as a potential nutrient source for pathogenesis during P. aeruginosa infection of the CF lung [13]. The major carbon source within the PC molecule comes from the two highly reduced long-chain fatty acids (LCFA). Many fatty acid degradation (b-oxidation) genes are expressed by P. aeruginosa during CF lung infection (e.g. fadD1: PA3299, fadD2: PA3300, fadA5: PA3013, and fadB5: PA3014) [13] and mutants defective in the fatty acid (FA) degradation pathway were reported to have decreased fitness during mouse lung infection [14]. A link between FA degradation genes and virulence was also observed [14] and P. aeruginosa can chemotax towards FA [15]. Furthermore, FA was shown to modulate type three-secretion system expression in this bacterium [16].
Despite the connection between virulence and FA degradation during infections, not all genes involved in P. aeruginosa FA degradation are characterized (Fig. 1A). In contrast, genes needed by Escherichia coli for aerobic b-oxidation (fadL, fadD, fadE, and fadBA [17][18][19][20]), anaerobic FA degradation (fadK and fadIJ [21]), and auxiliary genes (fadH [22] and fadM [23]) are well characterized. For an exogenous FA to be degraded by this pathway, it must first be transported by the membrane transporter (FadL) into the cell [24]. FA is then activated with the use of adenosine triphosphate (ATP) and coenzyme A (CoASH) by FadD (fatty acyl-CoA synthetase, FACS) [19], [25]. The activated FA molecule can then proceed through the b-oxidation pathway (Fig. 1A). In E. coli, genes encoding enzymes needed for boxidation (fadL, fadD, fadE, and fadBA) are repressed in the absence of FAs by the transcriptional regulator FadR. Acyl-CoA of chain length $ C 12:0 can bind to FadR to induce FA degradation [18], [26], [27] resulting in growth on FA (. C 10:0 ). Cyclic AMP and receptor protein complex levels [28], presence of oxygen [29], and osmotic pressure [30] also affect expression of FA degradation genes in E. coli. However, the existence of a central regulator, such as fadR, is unknown in P. aeruginosa, and only a few fad-genes have been found to be regulated by a FA sensor, PsrA [31].
P. aeruginosa exhibits greater metabolic capabilities for FA degradation than E. coli by growing aerobically on short, medium, and long-chain FAs as sole carbon and energy sources [31]. With a genome of 6.3 Mb, P. aeruginosa could potentially have more FA degradation genes than E. coli [32], suggesting possible redundancies and a higher level of complexity in this pathway. Three potential fadLs have been investigated thus far in P. aeruginosa and their exact role in FA transport still remains unclear [15]. Two fadBA operon homologues (fadAB1 and fadBA5) have been studied so far. The fadAB1 (PA1736 and PA1737) operon was shown to be strongly induced by medium-chain fatty acids (MCFA, C 10:0 and C 12:0 ) and, to a lesser extent, LCFA (C 14:0 -C 18:1

D9
) [33]. The fadBA5 (PA3014 and PA3013) operon was determined to be Fatty Acyl-CoA Synthetases in P. aeruginosa PLOS ONE | www.plosone.org involved in LCFA metabolism and to be induced by LCFA, especially oleate (C 18:1 D9 ) [31]. We have recently identified two FACS homologues of P. aeruginosa, fadD1 (PA3299) and fadD2 (PA3300) [14]. The FadD1 and FadD2 of P. aeruginosa were determined to have broad specificity for FA of different chain lengths. FadD1 has preference for LCFA whereas FadD2 has higher activities for shorter chain FAs. fadD1, fadD2, and fadD2D1 mutants showed growth defects when grown on minimal media with different length FAs as sole carbon sources. fadD1 was determined to be induced by LCFA and to be more important for growth on LCFA while fadD2 was important for growth on shortchain fatty acids (SCFA) and was induced by MCFA. The double mutant fadD2D1 displayed an impaired ability to grow on PC as a sole carbon source. This growth defect translated into decreased in vivo fitness during mouse lung infection, indicating that FadD1 and FadD2 may mediate P. aeruginosa replication in the CF lung [14]. However, the double mutant fadD2D1 was still able to grow on FA, suggesting the involvement of other fadD homologues in FA degradation [14].
We surveyed the P. aeruginosa genome for additional fadD homologues to gain more insight into the degradation of FAs in this bacterium. Four new fadD homologues PA1617, PA2893, PA3860, and PA3924 were identified out of 11 potential candidates. Through genetic analyses, their contribution to FA degradation was assessed. The final four candidates were determined to be FACS homologues, but PA1617 (fadD4) was found to be the major contributor to FA degradation. Involvement of the newly discovered fadD4 in catabolism of plant-derived acyclic terpenes suggests that the function of multiple FACS in P. aeruginosa is the degradation of compounds closely related to FAs. Growth defect on PC and decreased fitness in mouse lung of the sextuple fadD mutant supports the role of FA as a nutrient in vivo.

Identification of P. aeruginosa Fatty acyl-CoA Synthetase Homologues
To identify fadD homologues of P. aeruginosa, E. coli FadD amino acid sequence was compared to P. aeruginosa PAO1 ORFs via BLAST [34]. The amino acid sequence of genes obtained in the search were further analyzed for the presence of ATP/AMP [19], [35][36][37] and fatty acid binding motifs [38]. Genes that encode eleven proteins containing amino acid sequences with high degree of similarity to the motifs found in E. coli FadD (Fig. S1) were chosen for complementation tests. Identity and similarity of the proteins range from 22% to 31% and from 37% to 52%, respectively (Table S1). When cloned into a high copy number pUC19 vector, only genes encoding PA3860, PA1617, PA2893, and PA3924 were found to complement the E. coli fadD 2/ fadR 2 (E2011) strain on minimal medium containing oleate (C 18:1
All four P. aeruginosa fadD genes (fadD3, fadD4, fadD5, and fadD6) were tested further for their ability to support growth of E. coli fadD 2/ fadR 2 (E2011) on various FAs as a single copy on the E. coli chromosome. The E. coli fadD 2/ fadR 2 double mutant was used to ensure that FadR does not inhibit expression of other E. coli boxidation enzymes. Mini-Tn7 based complementation vectors were constructed and integrated into the E2011 chromosome at the attTn7 site and resulting strains were tested for growth on FAs ( Table 1). As expected, wildtype E. coli control strain K-12 showed growth on longer FAs (C 12:0 -C 18:1

D9
) but not on the MCFA, C 10:0 , or SCFAs (C 4:0 -C 8:0 ). The E2011 and the integrated empty-vector control strain were not able to growth on any of the FAs. E2011 complemented with E. coli fadD (fadD Ec ) grew on C 12:0 -C 18:1 D9 comparably to K-12. P. aeruginosa fadD3, fadD4, fadD5, and fadD6 genes individually allowed E2011 to grow on C 14:0 -C 18:1 D9 to similar levels as K-12. fadD3 and fadD6 complemented E2011 to a lesser degree than fadD4 and fadD5 on C 12:0 , and four fadD genes supported minimal growth of E2011 on C 10:0 to the same level as fadD Ec . E2011 complemented with fadD Ec , fadD3, fadD4, fadD5, or fadD6 did not grow on C 4:0 -C 8:0 , which was in agreement with previous observations that other E. coli FA degradation enzymes do not support metabolism of shorter FAs [39].
Contribution of fadD3, fadD4, fadD5, and fadD6 to FA Degradation To determine the role of the fadD homologues (fadD3, fadD4, fadD5, and fadD6) in P. aeruginosa FA degradation, strains with various combinations of fadD mutations were created. To prevent potential masking of phenotypes by fadD1 and fadD2, 15 mutants were constructed in the P. aeruginosa PAO1 DfadD1D2 background. Four triple, seven quadruple, four quintuple mutants and one sextuple mutant ( Table 2) were tested for growth on C 4:0 -C 18:1 D9 along with wildtype PAO1 and the DfadD1D2 mutant.
As expected, all 17 mutant strains grew the same as PAO1 on glucose at 24 h and 96 h (Tables 3 and 4). On C 4:0 , growth of all mutants was the same as PAO1 indicating that none of the fadD homologues contribute to the degradation of this FA or the differences were too small to be detected via plate-based growth assays. Throughout the study, the DfadD3D4D5D6 strain had the same growth as PAO1 on C 6:0 -C 18:1 D9 indicating that FadD1 and FadD2 are most likely providing a majority of FACS activity in P. aeruginosa (Tables 3 and 4). No difference in growth was observed between DfadD1D2 strain and DfadD1D2D3, DfadD1D2D5, DfadD1D2D6, DfadD1D2D5D6, DfadD1D2D3D5, DfadD1D2D5D6, or DfadD1D2D3D6 on C 6:0 -C 18:1 D9 . There was significantly less growth for DfadD1D2D4 on C 6:0 -C 18:1 D9 at 24 h in comparison to DfadD1D2, suggesting that fadD4 is important for degradation of all FAs from C 6:0 to C 18:1 D9 . Addition of fadD3, fadD5, or fadD6 mutation to DfadD1D2D4 strain in a quadruple mutant combination resulted in larger deficiencies in growth on FAs in comparison to the triple DfadD1D2D4 mutant (Tables 3 and 4), indicating that fadD3, fadD5, and fadD6 also take part in FA degradation and suggesting the dominance of FadD4 over these homologues. The DfadD1D2D3D4, DfadD1D2D4D5, and DfadD1D2D4D6 strains showed no growth on C 6:0 and C 8:0 , even after four days, in contrast to the DfadD1D2D4 mutant ( Table 4), indicating that fadD3, fadD5, and fadD6 are involved in the degradation of these FAs. All quintuple mutants exhibited some level of growth on several FAs after 96 h (Table 4), whereas no growth was present for the sextuplet mutant combination (DfadD1D2D3D4D5D6), indicating that all four new fadD homologues contribute to FA degradation and that only six aerobic FACS genes are likely present in P. aeruginosa. Quintuple mutants with both fadD4 and fadD5 mutations (DfadD1D2D3D4D5 and DfadD1D2D4D5D6) were most deficient in FA degradation (Table 3). Growth patterns of the four quintuple mutants after 96 h (Table 4) suggest that fadD4, besides fadD1 and fadD2, is much more important for FA degradation than fadD3, fadD5, and fadD6 combined, and fadD5 contributes to FA degradation more than fadD3 and fadD6. Furthermore, by comparing the phenotypes of double, triple, and quadruple mutants at two time points (Tables 3 and 4) a hierarchy of contributions of fadD homologues to the degradation of different chain-length FAs can be assigned as follows: i) FadD4 degrades C 6:0 -C 18:1 D9 (DfadD1D2D4 versus DfadD1D2 in Table 3); ii) FadD5 degrades C 6:0 -C 14:0 (DfadD1D2D4D5 versus DfadD1D2D4 in Tables 3 and 4); iii) FadD3 degrades C 6:0 -C 12:0 (DfadD1D2D3D4 versus DfadD1D2D4 in Tables 3 and 4); and iv) FadD6 degrades C 6:0 -C 12:0 (DfadD1D2D4D6 versus DfadD1D2D4 in Tables 3 and  4).
fadD1 and fadD2 in Comparison to fadD3, fadD4, fadD5, and fadD6 The growth phenotypes of various combinatory mutants on FAs indicated that out of the newly discovered FACS genes (fadD3, fadD4, fadD5, and fadD6) fadD4 is most important for FA degradation (Tables 3 and 4), in addition to fadD1 and fadD2 [14]. To investigate further the contribution of fadD4 to FA degradation in comparison to fadD1 and fadD2, growth curve experiments were performed on SCFA, MCFA, and LCFAs with DfadD1D2D4, DfadD3D4D5D6, DfadD1D2D3D5D6, and DfadD1D2D4D3D5D6 mutants along with PAO1 and DfadD1D2 strains (Fig. 2). The growth experiments on FAs were conducted up to 30 h, which was sufficient to distinguish differences in growth patterns between various strains. The growth rates calculated from growth curves in Fig. 2 are presented in Table  S3. The DfadD1D2 mutant strain had impaired growth in comparison to PAO1 on FAs ( Fig. 2B-2E). The phenotype of DfadD1D2D3D5D6 in C 6:0 -C 18:1 D9 (Fig. 2B-2E) was characterized by lower final optical density (OD) and/or longer lag phase than DfadD1D2, indicating that fadD3, fadD5, and fadD6 also contribute to FA degradation. In comparison to DfadD1D2 and DfadD1D2D3D5D6, DfadD1D2D4 exhibited very small amounts of growth, and no increase in turbidity was observed for DfadD1D2D3D4D5D6 on FAs ( Fig. 2B-2E). The DfadD3D4D5D6 mutant had almost identical growth in comparison to PAO1 in C 6:0 and C 18:1 D9 (Fig. 2B and 2E). In C 10:0 and C 14:0 DfadD3D4D5D6 showed a similar final OD as PAO1 but longer lag phase ( Fig. 2C and 2D). These data indicate that, although the activity of FadD4 is masked by the dominance of FadD1 and FadD2, the FadD4 plays a significant role in the degradation of FAs in P. aeruginosa.

Role of fadD Homologues in the Utilization of Plantderived Acyclic Terpenes
One of the P. aeruginosa fadD homologues, fadD5 (PA2893; atuH), was proposed to be part of the acyclic terpenes utilization (ATU) pathway and to contribute to degradation of citronellol and geraniol (perfumery compounds found in plants) by activating citonellic acid (CA) and geranic acid (GA) through addition of CoASH [40]. However, mutation of PA2893 alone did not abolish growth on acyclic terpenes possibly suggesting the involvement of other homologue(s) [40]. To determine the role of fadD5 and other fadD homologues in degradation of acyclic terpenes as plantderived nutrient sources, we grew PAO1 along with 17 combinatory fadD mutants in 1x M9 minimal media +1% (w/v) Brij-58 with 0.1% (w/v) of CA or GA (Fig. 3). All strains had similar OD measurements after one day of growth on glucose (Fig. 3A). After 24 h, all nine strains with the fadD4 mutation (triple, quadruple, quintuple, and sextuple combinations) had significantly lower OD for both compounds in comparison to PAO1 (20% or less) ( Fig. 3C and 3E). All other mutants had comparable growth to PAO1 in CA and GA (82%-96% and 88%-115%, respectively). None of the strains with fadD4 mutations had higher OD in CA or GA at day six, than at day one, and the remainder of the mutants grew the same as PAO1 (Fig. S2). Since only strains with fadD4 mutations exhibited growth defects in CA and GA, involvement of FadD4 in degradation of these compounds was further investigated using the single fadD4 mutant ( Fig. 3D and 3F). Single copy complementation returned growth of the DfadD4 mutant to PAO1 levels indicating that fadD4 is responsible for the majority of CA and GA degradation.
fadD3, fadD4, fadD5, and fadD6 and Virulence in P. aeruginosa A link between fadDs and production of virulence factors was previously observed in P. aeruginosa [14]. To determine if newly discovered homologues modulate virulence, single unmarked mutants DfadD3, DfadD4, DfadD5, DfadD6, along with DfadD1D2D3D4D5D6 strain and its complement were tested for production of proteases, lipases, phospholipases, and rhamnolipids. No difference in production of these virulence determinates was observed between PAO1 and all strains tested (data not shown).
-indicates no growth on a patch and+denotes growth. +1 is very little growth whereas +6 is very heavy growth comparable to K12 on glucose at day 3. doi:10.1371/journal.pone.0064554.t001

Involvement of New fadD Homologues in PC Degradation and in vivo Growth
Our previous study indicated that the DfadD1D2 mutant had a decreased ability to degrade PC and was less fit in BALB/c mice lungs [14]. We hypothesized that the sextuple fadD mutant, which does not grow on FAs, would exhibit impaired growth on PC and have significantly decreased in vivo fitness. We first investigated the role of the four newly discovered FACS in PC degradation (Fig. 4A). Before death phase, DfadD1D2 exhibited slower growth rate and lower final turbidity than PAO1. DfadD1D2D4 had a longer lag phase in comparison to DfadD1D2 before reaching a similar OD, implying that fadD4 contributes to degradation of PC. The DfadD1D2D3D4D5D6 mutant further exhibited a significant growth defect on PC. The large differences in growth rate and final OD between the sextuple mutant and DfadD1D2D4 suggest that not only fadD4 but also fadD3, fadD5, and fadD6 are required for growth on PC, which contains a mixture of FA chain lengths.
When in vitro competition studies were conducted on the sextuple fadD mutant and its competitor the complemented sextuple fadD mutant, mutation of all six FACS genes did not affect fitness when the bacteria were grown in rich Luria Bertani (LB) medium, and minimal medium supplemented with casamino acids, glucose, glycerol, and choline (Fig. 4B). In contrast, the in vitro competitive index (CI) in oleate (C 18:1 D9 ) and PC were low (,0. 15 and ,0.3, respectively) indicating that DfadD1D2D3D4D5D6 has a growth disadvantage on these carbon sources. The in vivo competition study showed that the sextuple fadD mutant was out numbered by its complement (Fig. 4C). An almost 10-fold increase in CFU per lung above inoculum (6 x10 6 ) was observed for both time points indicating bacterial replication in vivo. At 24 h, the amount of the sextuple fadD mutant was half of its complement, which is lower than the reported CI for the DfadD1D2 mutant at 24 h [14]. Even at 48 h the CI was significantly lower than 1, indicating that deletion of fadD genes decreases in vivo fitness of sextuple fadD mutant.

Discussion
Previous research on fadD1 and fadD2 indicated that more than two FACS genes are present in P. aeruginosa [14]. In this study, we focused on identification of additional fadD homologues. Four genes, fadD3, fadD4, fadD5, and fadD6 (PA3860, PA1617, PA2893, and PA3924, respectively) were found to encode FACS (Tables S1 and 1). Each of these genes contributes at a varying degree to FA degradation (Tables 3 and 4). Surprisingly, none of the new fadDs were involved in degradation of butyrate (C 4:0 ; Table 3). It is possible that other unidentified genes with acyl-CoA synthetase functions are responsible for growth on C 4:0 . Butyrate could also be processed through the acetoacetate degradation pathway (ato), an alternative pathway for degradation of SCFA [41]. This could be possible since two homologues of both of E. coli acetoacetyl-CoA transferase complex proteins, AtoA and AtoD, are present in P. aeruginosa: PA2000 (identity 45% and similarity 62%), PA0227 (identity 28% and similarity 62%), PA1999 (identity 40% and similarity 64%), and PA5445 (identity 33% and similarity 55%), respectively.
Growth studies with various mutants using FAs as sole carbon and energy sources indicated that FACS homologues are not of equal physiological significance and that there are disparities in importance and FA preference between them. fadD1 and fadD2, along with fadD4, are responsible for almost all FA degradation and dominate over other homologues. When fadD1 and fadD2 are inactivated, the majority of growth on SCFAs, MCFAs and LCFAs is due to fadD4 (Tables 3 and 4, Fig. 2). In comparison, fadD3, fadD5, and fadD6 have small contributions to overall growth on FAs and their individual involvement can be only observed when fadD1, fadD2 and fadD4 are absent (Table 4). This is not unprecedented, since Pseudomonas putida FadD2 is only active when FadD1 is not present [42]. It could be possible that gene(s) ruled out by screening in E. coli for growth on LCFA (Table S1), might be involved in SCFA and/or MCFA degradation. However, lack of growth for the sextuple fadD mutant on C 6:0 -C 18:1 D9 (Table 4) strongly indicates that P. aeruginosa has a total of six aerobic FACS genes.
P. aeruginosa is commonly found in soil, water, and on plant surfaces [43][44][45] and it is known to degrade over 70 different organic substances such as aromatic compounds, organic acids (e.g. isovalerate), alcohols, and acyclic terpenes (e.g., citronellol and geraniol) [44]. Sources of nutrients for pseudomonads on plant surfaces have not been determined. Citronellol and geraniol (perfumery compounds and possible bacterial nutrient sources found in plants) are degraded through the acyclic terpene utilization (ATU) pathway, b-oxidation pathway, and leucine/isovalerate utilization pathway [40,46]. The fadD5 (PA2893 or atuH) was proposed to be part of ATU and to be involved in activation of the CA and GA intermediates of the Strains were grown on 1x M9 medium +1% (w/v) Brij-58 supplemented with 0.2% (w/v) fatty acids or 20 mM glucose (Glu pathway. However, fadD5 was confirmed experimentally not to be part of ATU, and other homologues were thought to be also involved and to 'mask' the phenotype [40]. We investigated the possible role of fadD homologues in the degradation of acyclic terpenes, and we reasoned that combination of various fadD mutations would allow involvement of FACS homologues in ATU to be assessed. Surprisingly, fadD5 along with fadD1, fadD2, fadD3, and fadD6 had minimal if any contributions to the degradation of CA and GA (Fig. 3). Interestingly, fadD5 is located right next to genes known to be involved in ATU and seems to be the last gene in atuABCDEFGH cluster [40]. On the other hand, fadD4 is not only involved in ATU but it is almost solely responsible for degradation of these compounds as can be observed from growth phenotypes of the single fadD4 mutant and its complement ( Fig. 3D and 3F). Notably, homologues of fadD4 with high similarity are present in Pseudomonas fluorescens (e.g., Pfl01_4205 in Pf0-1, 72% identity and 84% similarity), Pseudomonas protegens (e.g., PFL_1744 in strain Pf-5, 71% identity and 82% similarity), and Pseudomonas mendocina (e.g., MDS_2302 in strain NK-01, 75% identity and 87% similarity) and some strains of these pseudomonads are known to degrade acyclic terpenes [40], [47]. The ability of P. aeruginosa to degrade lipids and FAs, especially the main component of lung surfactant PC, has been linked to replication of this opportunistic pathogen during infection of CF patients' lungs [13]. Previously, we determined that DfadD1, DfadD2, and double DfadD1D2 mutants have decreased fitness in BALB/c mice due to their deficiencies in degradation of FAs and PC [14]. We hypothesized that P. aeruginosa strains with greater defects in utilization of FAs and PC in vitro will have larger disadvantages during in vivo growth. DfadD1D2D3D4D5D6 mutant exhibited the most significant growth defect in FAs and PC (Fig. 2, 4A and 4B), and similar level of virulence factors (i.e. proteases, hemolysins, lipases) production was observed between sextuple fadD mutant, its complement, and PAO1 (data not shown). The DfadD1D2D3D4D5D6 mutant had some decrease of in vivo fitness in comparison to the DfadD1D2 at 24 h ( Fig. 4C and [14]); but at 48 h, DfadD1D2D3D4D5D6 mutant was not less fit in mice lungs than DfadD1D2 mutant. This latter result was surprising, as the impaired ability to utilize PC did not result in a more dramatic phenotype in vivo at 48 h (Fig. 4C). There are several possibilities, which could account for this unexpected phenotype. The sextuple mutant could utilize in vivo other constituents of PC such as choline and glycerol later in the infection. Additionally, pulmonary surfactants are composed of 10% proteins [48] and amino acids were suggested to be used by P. aeruginosa during lung infection [49] and could serve as an alternative nutrient source for sextuple fadD mutant. Other FACS genes (i.e. anaerobic which we could not identify because of limitations of our aerobic in vitro screening method) could be important for in vivo growth.
In summary, we have identified four additional FACS homologues of P. aeruginosa and determined their involvement in degradation of different FAs. The dual catabolic function of fadD4 (PA1617) for FAs and acyclic terpenes exemplifies the interconnection of metabolic pathways and multiple roles that FACS homologues play in this ubiquitous bacterium. Our in vivo data show that nutrient acquisition during lung infection is a complicated process, involving alterative pathways that require Strains were grown on 1x M9 medium +1% (w/v) Brij-58 supplemented with 0.2% (w/v) fatty acids or 20 mM glucose (Glu further investigation. Knowledge of all fadD genes needed for FA degradation significantly increases our understanding of the FA degradation pathway and its importance for in vivo replication of P. aeruginosa.

Ethics Statement
All animal experiments were approved by University of Hawaii at Manoa Institutional Animal Care and Use Committee (protocol no. 06-023-6) and were conducted in compliance with the NIH

Bacterial Strains and Growth Media
Strains and plasmids utilized in this study are listed in Tables 2,  S2, and 5, respectively. All P. aeruginosa mutants constructed and utilized in this study are derived from strain PAO1. E. coli E1869 strain (Table S2) was routinely used for cloning and E. coli Dasd or DdapA strains (E464, E1353, and E2072, Table S2) were used for mobilization of plasmids as described previously [50]. E. coli and P. aeruginosa strains were cultured in rich and minimal media as . Growth phenotypes of various fadD homologues mutants on acyclic terpenes. Strains were grown in liquid 1x M9 medium +1% (w/v) Brij-58 supplemented with 20 mM glucose, 0.1% (w/v) of citronellic acid, or 0.1% (w/v) geranic acid at 30uC. Optical densities (ODs) of cultures were measured and compared to PAO1 at day one (A, C, and E). Growth of DfadD4 mutant and DfadD4/attB::fadD4 complement strain in different carbon source were compared to PAO1 and ODs from day six are presented (B, D, and F). Results shown are from representative experiments that were performed twice by measuring triplicate cultures. doi:10.1371/journal.pone.0064554.g003 described by Kang et al. [14] unless indicated otherwise. Fatty acids stocks were prepared as previously described [31].

General Molecular Techniques
Molecular techniques were performed as previously described [50]. Oligonucleotides (Table 6) were synthesized through Integrated DNA Technologies.

Growth Phenotypes of Multiple fadD Mutants on Fatty Acids
To assess involvement of P. aeruginosa fadD homologues in FAs degradation, various strains (PAO1, double, triple, quadruple, quintuple, and sextuple fadD mutants) were purified on LB. After 24 h incubation at 37uC, two colonies of each strain were patched onto 1x M9 solid medium +1% (w/v) Brij-58 supplemented with 0.2% (w/v) FAs or 20 mM glucose. Plates were incubated at 37uC for four days. Growth of each strain was scored from +1 (little growth) to +6 (very heavy growth comparable to PAO1 on glucose at 96 h).

Growth of fadD Mutants on Acyclic Terpenes
The DfadD4/attB::fadD4 strain was constructed using a single copy complementation vector miniCTX2-fadD4, which was obtained by cloning the fadD4 PCR product (oligos #1443 and #1261) as HindIII and EcoRI fragment into miniCTX2 and integrated into DfadD4 mutant chromosome as described previously [58]. Stocks of citronellic (Sigma) and geranic acid (Sigma) (3% (w/v)) were prepared by neutralizing the compounds with equal molar sodium hydroxide and dissolving in 1% (w/v) Brij-58. PAO1 and various fadD mutants were grown overnight (14-16 h), starter culture were prepared as described by Kang et al. [14] and inoculated at 200-fold dilution into 1x M9 minimal medium +1% (w/v) Brij-58 supplemented with 0.1% (w/v) of citronellic acid, 0.1% (w/v) geranic acid or 20 mM glucose. Triplicate cultures were shaken at 30uC and optical densities were measured at day one and day six.

Virulence Factors Production
Lipase, protease, phospholipase, and rhamnolipd productions by fadD mutants were tested as previously described [14].

In vitro and in vivo Competition Studies
For in vitro and in vivo in competition studies, the DfadD1D2D3D4D5D6 strain was complemented with fadD2D1 and fadD4 cloned into miniCTX2 and fadD3, fadD5, and fadD6 cloned into miniTn7-Gm r . MiniCTX2-fadD2D1D4 complementation vector, was constructed by cloning fadD4 gene PCR product (oligos #1443 and #1261) as HindIII/blunt-end fragment into miniCTX2-fadD2D1 digested with XhoI, blunt-ended and digested with HindIII. To construct miniTn7-fadD3-fadD5-fadD6 vector, first fadD3 was sub-cloned as BamHI fragment from pUC19-PA3890 into miniTn7-Gm r , resulting in miniTn7-PA3860. The fadD6 was amplified with oligos #512 and #2210 and cloned as a BamHI/XbaI fragment into miniTn7-Gm r , resulting in miniTn7-PA3924. The fadD5 was amplified with oligos #438 and #2109 and digested with BamHI, blunt-ended, and digested with XbaI. To construct the final vector, the miniTn7-PA3924 was digested with XbaI, blunt-ended and digested with NdeI and the 2.5 kb fragment (containing fadD6) was cloned simultaneously along with fadD5 fragment into miniTn7-PA3860 digested with NdeI and SpeI. Integration of these plasmids into the P. aeruginosa chromosomes was performed as previously described ( [58] and [52]).

D9
) or PC was performed as described previously [14].
The in vivo competition study was performed as previously described [14]. Briefly, mucA was inactivated in the PAO1-DfadD1D2D3D4D5D6 and its complement strains utilizing pUC18-'mucA'. Equal amounts of alginate overproducing sextuple mutant and its complement were resuspended in their own supernatants and mixed. Fourteen BALB/c mice were inoculated intratracheally with 6 x10 6 colony forming units (CFU) of mixture of mutant (strain P973) and complement (strain P1028) as described previously [14]. At each time point (24 h and 48 h) seven mice were humanly euthanized, lungs were homogenized in 0.85% (w/v) saline and serial dilutions were plated on LB and LB+tetracycline 100 mg/ml to determine the total CFU and the complemented strain CFU. The competitive index (CI) was calculated as described [14]. Figure S1 Alignment of motifs of potential fatty acyl-CoA synthetase homologues. Amino acids with similar properties are assigned the same colors using CLC Sequence Viewer 6 software (www.clcbio.com). (TIF) Figure S2 Growth phenotypes of various fadD homologues mutants on acyclic terpenes at day six. Strains were grown in liquid 1x M9 medium +1% (w/v) Brij-58 supplemented with 0.1% (w/v) of citronellic acid or 0.1% (w/v) geranic acid at 30uC. Optical densities (ODs) of cultures were measured and compared to PAO1. (TIF)

Supporting Information
Table S1 Potential FadD homologues of P. aeruginosa identified through BLAST and tested for complementation in E. coli fadD 2/ fadR 2 (E2011).