Analysis of the Plasmid-Based ts-Mutant ΔfabA/pTS-fabA Reveals Its Lethality under Aerobic Growth Conditions That Is Suppressed by Mild Overexpression of desA at a Restrictive Temperature in Pseudomonas aeruginosa

ABSTRACT It is uncertain whether PA1610|fabA is essential or dispensable for growth on LB-agar plates under aerobic conditions in Pseudomonas aeruginosa PAO1. To examine its essentiality, we disrupted fabA in the presence of a native promoter-controlled complementary copy on ts-plasmid. In this analysis, we showed that the plasmid-based ts-mutant ΔfabA/pTS-fabA failed to grow at a restrictive temperature, consistent with the observation by Hoang and Schweizer (T. T. Hoang, H. P. Schweizer, J Bacteriol 179:5326–5332, 1997, https://doi.org/10.1128/jb.179.17.5326-5332.1997), and expanded on this by showing that ΔfabA exhibited curved cell morphology. On the other hand, strong induction of fabA-OE or PA3645|fabZ-OE impeded the growth of cells displaying oval morphology. Suppressor analysis revealed a mutant sup gene that suppressed a growth defect but not cell morphology of ΔfabA. Genome resequencing and transcriptomic profiling of sup identified PA0286|desA, whose promoter carried a single-nucleotide polymorphism (SNP), and transcription was significantly upregulated (level increase of >2-fold, P < 0.05). By integration of the SNP-bearing promoter-controlled desA gene into the chromosome of ΔfabA/pTS-fabA, we showed that the SNP is sufficient for ΔfabA to phenocopy the sup mutant. Furthermore, mild induction of the araC-PBAD-controlled desA gene but not desB rescued ΔfabA. These results validated that mild overexpression of desA fully suppressed the lethality but not the curved cell morphology of ΔfabA. Similarly, Zhu et al. (Zhu K, Choi K-H, Schweizer HP, Rock CO, Zhang Y-M, Mol Microbiol 60:260–273, 2006, https://doi.org/10.1111/j.1365-2958.2006.05088.x) showed that multicopy desA partially alleviated the slow growth phenotype of ΔfabA, the difference in which was that ΔfabA was viable. Taken together, our results demonstrate that fabA is essential for aerobic growth. We propose that the plasmid-based ts-allele is useful for exploring the genetic suppression interaction of essential genes of interest in P. aeruginosa. IMPORTANCE Pseudomonas aeruginosa is an opportunistic pathogen whose multidrug resistance demands new drug development. Fatty acids are essential for viability, and essential genes are ideal drug targets. However, the growth defect of essential gene mutants can be suppressed. Suppressors tend to be accumulated during the construction of essential gene deletion mutants, hampering the genetic analysis. To circumvent this issue, we constructed a deletion allele of fabA in the presence of a native promoter-controlled complementary copy in the ts-plasmid. In this analysis, we showed that ΔfabA/pTS-fabA failed to grow at a restrictive temperature, supporting its essentiality. Suppressor analysis revealed desA, whose promoter carried a SNP and whose transcription was upregulated. We validated that both the SNP-bearing promoter-controlled and regulable PBAD promoter-controlled desA suppressed the lethality of ΔfabA. Together, our results demonstrate that fabA is essential for aerobic growth. We propose that plasmid-based ts-alleles are suitable for genetic analysis of essential genes of interest.


RESULTS
The fabA gene is essential for growth on LB agar plates under aerobic conditions. Construction of essential gene deletion strains was often associated with the occurrence of spontaneous suppressors in bacteria (7). To prevent suppressors from occurring, we deleted the fabA chromosomal copy in the presence of a complementary copy in the ts-plasmid. That is, by using a three-step protocol (8), the chromosomal deletion DfabA allele was created under the protection of a native promoter-controlled fabA in the ts-plasmid pTS-fabA ( Fig. 1A) (see Materials and Methods). The chromosomal deletion allele DfabA in this ts-plasmid-based mutant strain, DfabA/pTS-fabA was confirmed by PCR assay using primer pairs F2 and R2, whose priming sites were absent in the complementary copy of the rescue plasmid pTS-fabA (Fig. 1B). A spot-plating assay indicated that the growth of DfabA/pTS-fabA was impeded at a restrictive temperature of 42°C on an LB plate under aerobic conditions compared to that of wild type, while the growth of DfabA/pTS-fabA was nearly identical to that of the wild type Genetic Analysis of fabA 2 in P. aeruginosa Microbiology Spectrum at a permissive temperature of 30°C (Fig. 1C). This result indicated that fabA is essential for growth on LB agar under aerobic conditions, consistent with the observation by Hoang and Schweizer (21). In contrast, Zhu et al. showed that DfabA was viable but exhibited a slow growth phenotype under aerobic condition (22). This discrepancy could be a result of different substrain backgrounds (23) or could, alternatively, be caused by spontaneous suppressor accumulations during construction of the deletion strain (6,7). Depletion of the multicopy ts-plasmid in mutant cells would take several generations at a restrictive temperature. We found that the growth curve of DfabA/pTS-fabA reached a critical point at 4.5 h with an optical density at 600 nm (OD 600 ) of ;1.2 on average at the onset of the stationary phase after a temperature shift (see Materials and Methods). That is, prior to the critical point, cell growth resembled that of the wild type. In contrast, after that point, the growth rate slowed down compared to that of wild type, suggesting the loss of complementary plasmids in mutant cells (see Fig. S1A and B in the supplemental material). However, it was less efficient for slow-growing mutant cells to develop the phenotype. To resolve this issue, we devised a consecutive subculture method to allow cells reducing the plasmid copy number in the first subculture and continuing phenotype development in the second subculture ( Fig. S1C and D). We found that copy numbers of the complementary plasmid were higher than those of the chromosome in the DfabA/pTS-fabA strain at the beginning of the second subculture based on the reverse transcription-quantitative PCR (qRT-PCR) analysis comparing plasmid-specific and chromosome-specific sequences (see Materials and Methods). Hence, we used the growth curve of the second subculture to show the growth defect of DfabA/pTS-fabA at a restrictive temperature (Fig. 1D).
The DfabA/pTS-fabA mutant exhibits curved cell morphology at a restrictive temperature. FabA, a bifunctional enzyme of 3-hydroxyacyl-ACP dehydratase (EC 4.2.1.59) and trans-2-decenoyl-ACP isomerase (EC 5.3.3.14), was involved in the synthesis of both saturated and unsaturated fatty acids (13)(14)(15)(16)21). To investigate the effect of FabA depletion on cell morphology, we examined the terminal phenotype of DfabA/pTS-fabA at a restrictive temperature. For this reason, both the DfabA/pTS-fabA mutant and wild-type strains were subjected to a temperature shift from 30°C to 42°C. Samples of the second subculture at 0 h, 3 h, and 6 h at a restrictive temperature were fixed and stained with Nile red prior to fluorescence microscopic analysis (see Materials and Methods).
Upon starting the second subculture (0 h), DfabA/pTS-fabA exhibited a wild-typelike rod-shaped cell morphology (Fig. 2, top row). On the other hand, 3 h after growth, DfabA/pTS-fabA mutant cells exhibited curved morphology (Fig. 2, middle row, see arrowheads). Strong Nile red fluorescence signals were found at the curvature of curved cells. It was proposed that new cell wall materials were evenly inserted along the cylindrical part of the rod-shaped bacterial cell during polarized growth (24). The curved cell of DfabA/pTS-fabA could be a result of uneven distribution of the newly synthesized cell wall materials to the cylindrical portion. After growth for 6 h in the second subculture, in addition to the curved cells, ghost cells or lysed cells started to appear (Fig. 2, bottom row, see arrows). These results indicated that cells lacking FabA function failed to maintain the rod-shaped cell morphology.
Though the ts-plasmid-based mutant would take a number of generations to for phenotypes in liquid culture to be observed, it is worth noting that the ts-plasmidbased conditional allele could be constructed for most, if not all, essential genes, while the point mutation-based conditional allele would not. Additionally, the time required for mutants to develop phenotypes in liquid culture was insufficient for spontaneous suppressors to populate the culture. Hence, the ts-plasmid-based mutant strains were suitable for systematic deletion analysis of essential genes in P. aeruginosa.
The DfabA/pTS-fabA mutant strain exhibits a significantly decreased level of palmitoleic acid at restrictive a temperature compared to that of the wild type. To investigate the effect of fabA disruption on fatty acid biosynthesis, we determined the relative level of various fatty acids in the DfabA/pTS-fabA mutant and wild-type cells at 30°C and 42°C based on the gas chromatography-mass spectrometry (GC-MS) analysis after transesterification. In this analysis, cellular lipids were extracted using chloroform and methanol solution (2:1 vol/vol) (25,26). The resulting lipids were transesterified to generate fatty acid methyl esters (FAME) (27) (see Materials and Methods). Six major fatty acids were found in both the wild-type and DfabA/pTS-fabA strains at 30°C and 42°C: palmitoleic acid (C 16:1 ), palmitic acid (C 16:0 ), cyclopropaneoctanoic acid 2-hexyl (C 17:lcyc ), oleic acid (C 18:1 ), stearic acid (C 18:0 ), and cyclopropaneactanoic acid 2-octyl (C 19:lcyc ) (Fig. S2). Upon a temperature shift to 42°C, we found two fatty acid species, C 16:1 and C 19:lcyc , whose level was significantly decreased and increased in both the wild-type and DfabA/pTS-fabA strains, respectively. Upon the temperature increase, the level of C 19:lcyc in DfabA/pTS-fabA was increased by 3.86-fold (i.e., average of 4.40-fold at 30°C versus 17.07-fold at 42°C; P , 0.05, n = 3), which was close to 3.27-fold in the wild type ( Fig. 3A and B, Table S1). On the other hand, after the temperature increase to 42°C, the level of C 16:1 was decreased to 18% of the initial level at 30°C (i.e., average of 13.05 arbitrary unit at 30°C versus 2.29 arbitrary unit at 42°C; P , 0.05, n = 3) in DfabA/pTS-fabA, which decreased by greater than 2-fold compared to that of wild type (i.e., 18% in DfabA/ pTS-fabA versus 38% in the wild type; P , 0.05, n = 3). These results suggested that C 16:1 biosynthesis was reduced in DfabA/pTS-fabA at a restrictive temperature.
It was shown that the fabA 2 mutant was auxotrophic for oleic acid and stearic acid (16,21). Consistent with this, we isolated the DfabA mutant from DfabA/pTS-fabA at 42°C after depletion of the ts-rescue plasmid in medium supplemented with oleic acid or stearic acid (Fig. 3C). A spot-plating assay showed that DfabA failed to grow on an LB agar plate without oleic acid or stearic acid supplementation under aerobic conditions (Fig. 3D). Likewise, DfabA did not grow in LB liquid medium without oleic acid or stearic acid supplementation (Fig. 3E). Microscopic analysis indicated that after removal of oleic acid or stearic acid, DfabA exhibited curved morphology (Fig. 3F, see arrow), identical to that of DfabA/pTS-fabA at a restrictive temperature (see Fig. 2), supporting the notion that curved phenotype was attributed to the loss of FabA function. Cells fail to grow and exhibit oval morphology upon strong induction of fabA-OE or fabZ-OE. Another dehydratase, FabZ, was known to be present in P. aeruginosa (28). To examine whether overexpression of the fabZ gene, an isozyme of fabA, could rescue the growth defect of DfabA/pTS-fabA at a restrictive temperature, we prepared an overexpression fabZ-OE construct whose transcription was under the control of the arabinose-regulable P BAD promoter (29) in the pBBR plasmid to generate pOE-fabZ (see Materials and Methods). As a control, the pOE-fabA plasmid was also constructed. We found that mild overexpression of fabZ with 0.002% and 0.02% arabinose did not rescue the growth defect of the DfabA/pTS-fabA pOE-fabZ strain at a restrictive temperature (Fig. 4A, see arrow). However, as a control, we found that no or mild overexpression of fabA rescued the growth defect of DfabA/pTS-fabA pOE-fabA strain at restrictive . This result suggested that FabZ did not share essential functions with FabA. In addition, fabA transcription at a level derived from leakiness or without induction of fabA-OE was sufficient for function. Notably, we found that under strong induction with 0.2% arabinose, both fabA-OE and fabZ-OE impeded growth of the DfabA/pTS-fabA strain and the wild type (Fig. 4A, see rectangles). Morphological analysis showed that, under strong induction of fabA-OE or fabZ-OE, DfabA/pTS-fabA and wild-type cells exhibited a terminal phenotype of oval morphology at both 30°C and 42°C ( Fig. 4B and C). This result suggested that a strong overexpression phenotype of fabA-OE and fabZ-OE was likely to be a result of overlapping function between FabA and FabZ.
A suppressor, sup, restores cell growth but not cell morphology of DfabA at a restrictive temperature. To search for suppressors of DfabA, more than 10 9 DfabA/ pTS-fabA cells were spread out on LB plates and grown at the semirestrictive temperature of 40°C, permitting spontaneous mutations for 2 weeks (see Materials and Methods). A colony was found to contain only the deletion allele of fabA based on a PCR assay with sequence-specific primers (Fig. 5A). Hence, it was designated sup for suppressor of DfabA. A spot-plating assay and growth curve analysis confirmed that the sup strain rescued the growth defect of DfabA at a restrictive temperature ( Fig. 5B and C). However, the sup mutant exhibited a slow-growth phenotype at 30°C.
Morphological analysis indicated that at a restrictive temperature, the sup mutant strain exhibited the curved cell form (Fig. 5D, see arrows) similar to that of DfabA/pTS-fabA (see Fig. 2). This result suggested that sup suppressed the growth defect but not the morphological defect of DfabA at a restrictive temperature. Notably, sup displayed a terminal phenotype of the filamentous form at a permissive temperature. These Genetic Analysis of fabA 2 in P. aeruginosa Microbiology Spectrum results implied that the sup mutant was a loss-of-function allele of an unknown gene required for growth: at a restrictive temperature, double mutation sup 2 DfabA survived due to suppression. At a permissive temperature, on the other hand, sup mutation exhibited slow growth with the filamentous phenotype. Alternatively, high-level upregulation of sup impeded wild-type cell growth, while mild-level upregulation rescued DfabA at a restrictive temperature. FAME analysis indicated that only three major lipids were found in sup, namely, palmitoleic acid (C 16:1 ), palmitic acid (C 16:0 ), and stearic acid (C 18:0 ). Unlike the wild type and DfabA/pTS-fabA (see Fig. 3A), the level of C 16:1 at 42°C was not significantly decreased compared to that at 30°C (Fig. 5E, Table S2). The ratio of C 16:1 between levels at 42°C and 30°C in sup was much less than that of DfabA/pTS-fabA (ratio in log 2 scale: -0.55 in sup versus 22.51 in DfabA/pTS-fabA; P , 0.05, n = 3) (Fig. 5F). This result suggested that C 16:1 retention in sup at a restrictive temperature was required for rescuing the growth defect of DfabA.
Genome resequencing and transcriptomic profiling identified DesA as a candidate for overexpression suppressor of DfabA in sup. Genome resequencing was performed to identify candidate suppressor genes in the sup mutant compared to those in DfabA/pTS-fabA using Illumina technology (see Materials and Methods). Approximately 5 million 300-bp short reads were obtained for both sup and DfabA/pTS-fabA. The reads were mapped to the reference genome (http://www.pseudomonas.com) using Burrows-  Table S3). This result suggested that 92.3% of the mutations in sup were not responsible for the suppression of DfabA. Of the five sup mutant-specific SNP loci, two were located at intergenes and three were in the coding sequences (Fig. 6B).
To investigate whether any of the sup-specific SNP-affecting genes would be significantly upregulated in sup that suppressed the lethality of DfabA at a restrictive temperature, we performed transcriptome sequencing (RNA-seq)-based transcriptomic profiling analysis in triplicate (Fig. 6C). We first identified subsets of 101 and 279 common differentially expressed genes (DEGs) (level increase, .2-fold; false-discovery rate (FDR)-adjusted P , 0.05, n = 3) in sup compared to that of the wild type and DfabA/ pTS-fabA at 30°C and 42°C ( Fig. 6D and E), respectively. A group of 39 DEGs were found in intersections of the two subsets of 101 and 279 DEGs (Fig. 6F, Table S4), indicating that these genes were upregulated in sup compared to the wild type and DfabA/pTS-fabA at 30°C and 42°C. Significantly, desA, one of the five sup-specific SNP-affecting genes was found in the subset of 39 upregulated DEGs in sup (P value = 0.038) (Fig. 6G). A SNP of C to T change located at the 264 nucleotide (nt) position of the desA promoter in sup could potentially be responsible for the alteration of desA Genetic Analysis of fabA 2 in P. aeruginosa Microbiology Spectrum transcription that fully suppressed the growth defect of DfabA. Similar to this observation, Zhu et al. (22) showed that a multicopy plasmid containing native promoter-controlled desA partially rescued the slow-growth phenotype of DfabA, in which fabA was not essential for viability (22). Hence, full suppression of the DfabA growth defect by desA overexpression needed to be experimentally validated. DfabA lethality is suppressed by an SNP-bearing promoter-or mild induced P BAD promoter-controlled desA. Genome resequencing and transcriptomic profiling identified a SNP of C to T change at the 264 nt position in the promoter of desA whose transcription was upregulated in sup compared to that of DfabA/pTS-fabA and the wild type (see Fig. 6). To test if the point mutation at 264 nt in the promoter of desA was responsible for rescuing the growth defect of DfabA, we cloned the sequences of the promoter and coding region of desA from the sup mutant and wild type and integrated them into the genome of DfabA/pTS-fabA after validation by sequencing to generate strains of DfabA P sup :desA/pTS-fabA and DfabA P wt :desA/pTS-fabA. A spot-plating assay confirmed that a single point mutation at the promoter of desA was sufficient for DfabA/pTS-fabA to phenocopy the sup mutant. That is, DfabA P sup :desA/pTS-fabA but not DfabA P wt :desA/pTS-fabA exhibited slow growth at 30°C resembling that of the sup mutant (Fig. 7, top panel). On the other hand, growth of DfabA P sup :desA/pTS-fabA at 42°C was fully restored to that of the wild type and sup. Full growth restoration at 42°C was further validated using growth curve analysis (Fig. 7, bottom panel).
To investigate whether this was due to the alteration of desA transcription that rescued DfabA, we constructed the desA-OE plasmid pOE-desA and another desaturase desB-OE (22) plasmid, pOE-desB, in which target gene expression was under the control of the arabinose-regulated P BAD promoter in the pBBR plasmid (See Materials and Methods). A plasmid, pOE-bdhA, for a gene that was involved in lipid metabolism (32) found in the group of 39 upregulated DEGs in the sup mutant was used as a control. These plasmids were transformed into the wild type and DfabA/pTS-fabA strains for growth analysis.
We subsequently examined the effect of desA and desB overexpression on the morphology of DfabA/pTS-fabA and wild-type cells. We found that under mild induction of 0.02% arabinose supplementation, desA-OE did not restore the morphology of DfabA/ pTS-fabA at a restrictive temperature (Fig. 7C, see arrows). On the other hand, upon strong induction with 0.2% arabinose supplementation, both desA-OE and desB-OE induced a filamentous phenotype in DfabA/pTS-fabA and wild-type cells (Fig. 7C and D). However, cells with oleic acid supplementation did not exhibit filamentous morphology (see Fig. 3F), suggesting that only the increased level of DesA product unsaturated fatty acid (UFA)-containing phospholipid or DesB product UFA-coenzyme A (CoA) but not oleic acid altered the cell morphology. These results indicated that different levels of desA expression were attributed to the suppression phenotype of the sup mutant strain (Fig. 7E). They further implied that the levels of unsaturated fatty acid-containing membrane phospholipids played an essential role in the regulation of cellular morphology in P. aeruginosa.
DfabA growth with supplementation of stearic acid is dependent on desA but not desB. DesA and DesB were found to be dispensable desaturases in P. aeruginosa (22). We wanted to test if DfabA auxotrophic for oleic acid (i.e., UFA) and stearic acid  Fig. S3). These strains were subjected to a spot-plating assay on LB plates supplemented with oleic acid, stearic acid, or no fatty acid as control at 30°C and 42°C. The results indicated that DfabA/pTS-fabA and DfabA DdesB/pTS-fabA grew on LB plates supplemented with oleic acid or stearic acid at 42°C (Fig. 8A). On the other hand, DfabA DdesA/pTS-fabA and DfabA DdesA DdesB/pTS-fabA grew on plates supplemented with oleic acid but not stearic acid at 42°C. The growth at 42°C was also tested using growth curves (Fig. 8B). These results indicated that DfabA auxotrophic for stearic acid was dependent on the function of desA but not desB. This was supported by the observation that mild induction of desA-OE but not desB-OE rescued the growth defect of DfabA (see Fig. 7B). In contrast, Zhu et al. (22) showed that DfabA DdesA but not DfabA DdesA DdesB grew in medium supplemented with stearic acid, concluding that DfabA DdesA auxotrophic for SFA was desB dependent, increasing the level of discrepancy between the two studies. It would be interesting discover the factors in the background that led to the discrepancies in these analyses.

DISCUSSION
Fatty acids are essential for bacterial cell viability (10). That is why microbial enzymes involved in type II fatty acid synthesis (FASII) are popular targets for antibacterial drug development (11,12). Pseudomonas aeruginosa is an opportunistic pathogen for which new drug development is needed (1-3). ts-Plasmid-based conditional alleles of essential genes are useful for deletion analysis of essential genes of interest in P. aeruginosa (8). However, multicopy ts-plasmids hamper the analysis of the mutant phenotype because cells deplete the ts-plasmid at the onset of the stationary phase after the shift to a restrictive temperature. Mutant cells at the stationary phase slow down the phenotype development. In this study, we developed a consecutive subculture protocol to ensure the rapid growth of mutant cells prior to the point of plasmid depletion (see Fig. S1). Many essential genes are suppressible (6,9). Hence, it is possible to accumulate spontaneous suppressors during construction of essential gene deletion strains without protection of complementary copies. It is also possible that different strain backgrounds can lead to the difference of fabA essentiality in P. aeruginosa (21,22). To avoid suppressor accumulation, we deleted the fabA chromosomal copy in the presence of a complementary copy in the ts-plasmid. The resulting DfabA/pTS-fabA strain failed to grow at a restrictive temperature under aerobic conditions (see Fig. 1).
FabA is bifunctional enzymes 3-hydroxyacyl-ACP dehydratase and trans-2-enoyl-ACP isomerase, which are involved in the synthesis of saturated and unsaturated fatty acids, respectively (13)(14)(15)(16). FabZ, another copy of 3-hydroxyacyl-ACP dehydratase, shares functions in cycles of fatty acid elongation with FabA involved in saturated fatty acid synthesis (16). In this study, we show that mild induction of fabZ-OE does not rescue the DfabA/pTS-fabA growth defect at a restrictive temperature, suggesting that the growth defect of DfabA/pTS-fabA is fully attributed to the loss of unsaturated fatty acid synthesis (see Fig. 4).
Essential genes can be bypass-suppressed (6,9). In this case, transcription repressionbased analysis of essential genes is unsuitable for suppressor analysis, because the repression machinery tends to accumulate mutations upon extended growth (33,34). By using the ts-plasmid-based DgmhB/pTS-gmhB strain, we previously identified the fbp gene, whose overexpression suppresses the growth defect of DgmhB/pTS-gmhB at a restrictive temperature (8). In this study, by using the same approach, we identified a point mutation at the promoter of the desaturase DesA in sup, which is sufficient for DfabA/pTS-fabA to phenocopy the sup mutant that exhibits slow growth at 30°C and normal growth at 42°C (see Fig. 7A), consistent with growth phenotypes by strong overexpression and mild overexpression of desA, respectively (see Fig. 7B). Similar to this, Zhu et al. proposed that oleic acid repressed the desA expression in DfabA and showed that multicopy desA partially alleviated the slow growth phenotype of DfabA in medium without oleic acid supplementation (22). It was not clear whether the discrepancy of fabA essentiality could be caused by a spontaneous suppressor prior to or during strain construction. We propose that the plasmid-based ts-alleles of essential genes are useful for systematic deletion and suppressor analyses of essential genes of interest in P. aeruginosa.
Unsaturated fatty acid is believed to modulate membrane fluidity; e.g., at high (i.e., 42°C) and low (i.e., 30°C) temperatures, levels of unsaturated fatty acid in membrane lipids decrease and increase, respectively (35). In this study, we show that cells display curved morphology upon depletion of FabA, suggesting that disruption of membrane fluidity homeostasis leads to the loss of polarized growth (see Fig. 2 and 3). Furthermore, cells exhibit oval morphology after strong overexpression of fabA and fabZ (see Fig. 4). Given the identical effects of fabA-OE and fabZ-OE, the oval cell morphology phenotype is unlikely to be related to the synthesis of unsaturated fatty acid. We propose that a high level of FabA or FabZ activity depletes its substrate 3-hydroxyacyl-ACP, which in turn limits the synthesis of lipid A, a major cell wall lipopolysaccharide component in Gram-negative bacteria (36). Disruption of fabZ is known to suppress the growth defect of lpxA 2 and lpxC 2 mutants that are incapable of synthesizing lipid A (37).
Although mild induction of desA-OE suppresses the growth defect of DfabA/pTS-fabA at a restrictive temperature, strong overexpression of desA impedes cell growth with a filamentous phenotype (see Fig. 7). It is known that desA that utilizes phospholipid but not free fatty acid as a substrate to increase the level of unsaturated double bonds in fatty acyls (22). Hence, this result implies that when the level of unsaturated fatty acid in membrane lipid is higher than usual, rod-shaped cells will turn into filamentous forms.
While DesA is an sn-2-position phospholipid D9-desaturase, DesB is a proposed acyl-CoA D9-desaturase that permits the growth of DfabA DdesA in medium supplemented with saturated fatty acid (22). In this study, however, we show that DfabA DdesA/pTS-fabA does not grow on medium with stearic acid supplementation regardless of the presence of desB (see Fig. 8), which is supported by the observation that desB-OE does not rescue the DfabA/pTS-fabA lethality at a restrictive temperature (see Fig. 7). These results increased the level of discrepancies between the two studies (22; this study). Spontaneous mutation without an apparent phenotype often went unnoticed (23). However, when a mutation with a growth advantage occurs, it will be quickly enriched in the population, such as in an essential gene-deletion mutant during construction. Hence, the preexistence or enrichment of a spontaneous mutation or suppressor is a complex issue that warrants further study.
To summarize the similarities and differences between this study and the previously reported studies (21,22), we show that when using the ts-plasmid-based conditional mutant strain DfabA/pTS-fabA, fabA is essential for growth under aerobic conditions in P. aeruginosa, similar to the study by Hoang and Schweizer (21) but different from that by Zhu et al. (22). DfabA/pTS-fabA cells exhibit curve morphology at a restrictive temperature. Mild overexpression of fabZ, a fabA isozyme, does not rescue the growth defect of DfabA. However, strong overexpression of fabA or fabZ impedes growth of cells displaying oval morphology. From the spontaneous mutagenesis screening, we isolated a suppressor, sup, that fully rescues the growth defect but not the curved morphology of DfabA at 42°C. On the other hand, sup exhibits a slow-growth phenotype with filamentous morphology at 30°C. Genome resequencing and transcriptomic profiling of sup compared to that of DfabA/pTS-fabA identifies desA, whose promoter bears a SNP at the 264 position, and the transcription level is significantly upregulated compared to that of DfabA/pTS-fabA and the wild type. We validate that integration of the SNP-bearing promoter-controlled desA is sufficient for DfabA/pTS-fabA to phenocopy the sup mutant. Furthermore, mild overexpression of desA fully rescues the growth defect but not morphology of DfabA. It is interesting to note that strong overexpression of desA impedes cell growth with a filamentous morphology, resembling the sup phenotype at 30°C. Zhu et al. (22) characterized the two desaturases, desA and desB, in P. aeruginosa. They propose that desA expression is repressed in the presence of oleic acid and show that multicopy desA can partially rescue the slow-growth phenotype of DfabA without oleic acid (22). They further propose that DfabA DdesA auxotrophic for SFA such as stearic acid and palmitic acid is desB dependent (22). In this analysis, we show that DfabA/pTS-fabA and DfabA DdesB/ pTS-fabA grow in medium with stearic acid supplementation at 42°C. In contrast, DfabA DdesA/pTS-fabA and DfabA DdesA DdesB/pTS-fabA do not grow, indicating that DfabA auxotrophic for SFA is desA dependent. Consistent with this, we show that mild expression of desB does not rescue the growth defect of DfabA. These observed discrepancies could be caused by the differences between the strain backgrounds (23). It would be interesting in the future to identify the background factor that leads to the differences between the two strains.

MATERIALS AND METHODS
Oligonucleotides, plasmids, and bacterial strains. The oligonucleotides, plasmids, and bacterial strains used in this study are listed in Table 1. The P. aeruginosa PAO1 wild-type strain (BioSciBio, Hangzhou, China) and its derivatives were cultivated in LB (in 1 L: 10 g tryptone, 10 g NaCl, 5 g yeast extract, pH 7.0) liquid or solid (addition of 1.5% agar) medium supplemented with antibiotics (e.g., 100 mg mL 21 ampicillin, 50 mg mL 21 gentamicin, and 100 mg mL 21 tetracycline) and chemicals (e.g., 15% sucrose or 0.002%, 0.02%, or 0.2% arabinose) at 30°C or 42°C as indicated. Temperature-sensitive mutant strains were maintained in LB medium with adequate supplementation at the permissive temperature of 30°C. The chemicals used in this study were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA).
Growth curve of consecutive subcultures for assessment of the plasmid-based ts-mutant. The complementary plasmid is a multicopy plasmid in the DfabA/pTS-fabA strain (38). We found that 4.5 h after the shift to the restrictive temperature, cells start to show average copy numbers of plasmid close to those of the chromosome or ts-plasmid nearing depletion (Fig. S1). However, slow-growing stationaryphase cells hamper assessment of the mutant phenotype. To circumvent this issue, we adopted the consecutive subculture method (39) for the analysis of the DfabA/pTS-fabA growth curve (39). In brief, the fresh overnight culture at 30°C was inoculated into a shake flask to a starting OD 600 of 0.05 at 42°C (see Fig. S1). Cell growth was monitored by measuring the OD at various time points. Growth of DfabA/pTS-fabA slowed down after several generations at 4.5 h at 42°C compared to that of the wild type. We took the first subculture at 4 h the as inoculum to start the second subculture for the analysis of the DfabA/ pTS-fabA growth curve. Based on the ratio of plasmid and chromosome copy numbers, it was clear that depletion occurred after starting the second subculture. Hence, the growth curve of the second subculture was used for the DfabA/pTS-fabA mutant strain.
Plasmid construction. Deletion and rescue plasmids (8) were used for construction of the pDEL-DfabA and pRES-fabA plasmids, respectively (see Fig. 1A). Briefly, for pDEL-DfabA plasmid construction, a DfabA Genetic Analysis of fabA 2 in P. aeruginosa Microbiology Spectrum deletion cassette consisting of 500 bp upstream and 500 bp downstream sequences was PCR amplified using oligonucleotides expanded with overlapping sequences for cloning using a cloning kit (ClonExpress II onestep cloning kit, Vazyme, China). After double digestion with PstI and KpnI, pDEL and the fabA upstream and downstream sequences were subjected to cloning according to the manufacturer's instruction (Vazyme). For pRES-fabA plasmid construction, the native promoter containing the fabA coding sequence was PCR amplified with oligonucleotides expanded with overlapping sequences with the vector cloning site. After KpnI digestion, the pRES plasmid and the native promoter-controlled fabA sequence were subjected to Vazyme cloning. For gene overexpression plasmids, pBBR1MCS-5 (or pBBR) was utilized (40). Briefly, the araC-P BAD promoter sequence (29) and the target gene sequence were PCR amplified using oligonucleotides expanded with overlapping sequences for Vazyme cloning. The BamHI and PstI double-digested pBBR plasmid, together with araC-P BAD and the target gene sequences, was subjected to Vazyme cloning. All constructed plasmids were sequencing-validated prior to use. Strain construction. We adopted the three-step protocol (8) to construct the plasmid-based ts-allele of the DfabA/pTS-fabA strain. Briefly, we first electroporated the pDEL-DfabA plasmid into the P. aeruginosa PAO1 strain and isolated the plasmid integrants via a single crossover into the genome on a gentamicin (Gm)-containing LB plate, because the pDEL plasmid could not be autoreplicated in P. aeruginosa. Second, a pRES-fabA plasmid was transformed into the pDEL-DfabA plasmid integrant on the tetracycline-containing LB plate. Third, the resulting transformants were subjected to counterselection of sacB for generation of the chromosomal DfabA allele after looping out the integrated pDEL-DfabA plasmid via single crossover on a sucrose-containing LB plate. The plasmid-based DfabA/pTS-fabA strains were PCR validated for the chromosomal DfabA allele and spot-plating assay for the ts-growth phenotype. For deletion of nonessential genes such as DdesA and DdesB, only the deletion plasmid was applied. For construction of a strain containing a chromosomal copy of desA-C-64T (nucleotide C to T change at the 264 nt position of desA as in the sup mutant), desA promoter and coding sequences were PCR amplified from sup and wild-type cells (as control) and cloned into an integration vector such as the deletion plasmid. The resulting plasmid was transformed into the DfabA/pTS-fabA strain to yield DfabA P sup :desA/pTS-fabA and DfabA P wt :desA/pTS-fabA. DNA transformation. For P. aeruginosa strain construction, electrocompetent cells were prepared using the protocol of Huang and Wilks (41) with minor modifications. Briefly, 5 mL log-phase cells was harvested and washed with 10% glycerol three times. Subsequently, the cells were resuspended in 10% glycerol. For electroporation, 90 mL of electrocompetent cells mixed with 10 mg of plasmid DNA was transferred to an electroporation cuvette with a 1-mm gap (Bio-Rad) and a pulse of 1,200 V, 2.5 mF, and 5 ms was applied using a Bio-Rad Xcell electroporator. After the pulse, 1 mL of LB medium was immediately added to the cuvette and mixed gently, and the mixture was transferred to a fresh tube and incubated at 30°C for 3 h with shaking at 200 rpm before plating onto LB plates supplemented with the appropriate antibiotics. Transformants usually appeared after overnight incubation.
Spot-plating assay. The spot-plating assay (42) was adopted to test sensitivities to stress factors such as antibiotics, sucrose, and temperature. Briefly, 10-fold serial-diluted cultures were transferred using a 48-pin replicator (V&P Scientific, Inc.) onto LB plates supplemented with appropriate stress factors and incubated at 30 or 42°C as indicated.
Fluorescence microscopy. Cell morphology was investigated under a BX53 microscope (Olympus, Tokyo, Japan) using the phase contrast configuration. For the cell outline cytoplasmic membrane, Nile red staining was employed. Briefly, 50 mL overnight culture was added to 5 mL LB broth and grown in a shaker to an OD 600 of 0.8 for examination at 30°C and at 6 h and 9 h after the shift to 42°C. From the resulting fresh culture, 1 mL was harvested and resuspended with 4% formaldehyde fixative solution. After the cells were fixed in the formaldehyde solution for 30 min or more at room temperature, the fixed cells were washed with phosphate-buffered saline (PBS) (in 1 L: 10.9 g Na 2 HPO 4 , 3.2 g NaH 2 PO 4 , and 90 g NaCl, Ph7.4) and were ready for fluorescence dye staining. Cell suspension was added with Nile red at a final concentration of 10 ng mL 21 for 30 min and then washed and resuspended in PBS for fluorescence microscopic examination.
Lipid extraction and fatty acid methyl ester (FAME) preparation. Cellular lipids were extracted using a chloroform-methanol solution (2:1 vol/vol). The organic phase was transferred to a fresh tube and blown with nitrogen gas to evaporate organic solvent. The resulting lipid was weighted as the quantity of total lipids and resuspended in hexane to a desired concentration. For analysis of fatty acid composition, one part of the extracted lipids was transesterified with methanol to generate fatty acid methyl ester (FAME) according to a published protocol (27). The total lipid extracted from P. aeruginosa using chloroform methanol (2:1 vol/vol) solution is known to contain neutral and phospholipids that predominantly possess hexadecanoic (C 16:0 ), hexadecenoic (C 16:1 ), octadecanoic (C 18:0 ), octadecenoic (C 18:1 ), 17-and 19-cyclopropane, etc. acids (25,26).
Gas chromatography coupled with mass spectrometry (GC-MS) analyses. To determine the FAME species, 1 mL FAMEs was directly injected into the injection port of a gas chromatograph (2010Plus GC system, Shimadzu Co., Tokyo, Japan) coupled with a mass spectrometer (MS) system (Shimadzu QP2020 with quadrupole analyzer). The GC was operated on an Rtx-5MS GC column (30 m Â 0.25 mm inside diameter [i.d.] with 0.25-mm film thickness of 5%-phenyl-methylpolysiloxane) (Restek Co., Bellefonte, PA, USA), and helium (purity, 99.999%) was used as the carrier gas. The temperature of the injection port was set to 260°C, while the sample injection was made in splitless mode with a purge flow of 50 mL min 21 for 1 min. The temperature program was started with an initial temperature of 160°C and then was increased 2°C min 21 to 230°C for 10 min. The mass spectrometer was operated in electron ionization (EI) mode with the ion source temperature set at 230°C. The electron energy was 70 eV. Full-scan MS data were acquired in the range of 50 to 500 m/z to obtain the fragmentation spectra of the FAMEs. LabSolutions (Shimadzu Co.) was used to determine all the peaks in the raw GC chromatogram. A library search was done for all the peaks using the National Institute of Standards and Technology NIST/EPA/NIH (NIST 14 Library).
Isolation of suppressors. Approximately 10 9 DfabA/pTS-fabA cells were plated onto LB plates and incubated at the semirestrictive temperature of 40°C for 2 weeks for suppressors through spontaneous mutations as described (8,9). During the incubation, plates were placed in a bag with a stream of fresh air filtered with a 0.22-mm-pore-size filter after passing through a water container to maintain the humidity and oxygen. Suppressor colonies were validated via streaking on fresh LB plates and incubated at 42°C. Primer-specific PCR assays for the deletion allele were also performed for validation of the fabA deletion allele.
Genome resequencing analysis. Genomic DNA was extracted with a Genomic DNA extraction kit (TaKaRa Bio, Inc.) and sheared to ;400 bp in length using an S2 instrument (Covaris, Woburn, MA, USA). The sequencing library was constructed using s NEXTflex DNA sequencing kit (Bioo Scientific, USA) according to the package instructions and then was sequenced on a MiSeq PE300 device (Illumina, Inc., San Diego, CA, USA), generating a total of 5 million clean reads of 300 bp in length for both sup and DfabA/pTS-fabA (BioMedical Institute of Shanghai, Shanghai, China). The reads were mapped to the reference genome PAO1 (http://www.pseudomonas.com) using BWA software (30). Analysis of small sequence variants such as single-nucleotide polymorphisms (SNPs) and small insertions and deletions (indels) of less than 50 bp was performed using the SAMtools software (31).
RNA-seq-based transcriptomic analysis. Total RNA was extracted from various cell samples in triplicate with an RNA extraction kit (TaKaRa Bio, Inc.). After a qualification check with a 2100 Bioanalyzer (Agilent Technologies, Inc., Redwood City, CA, USA), qualified samples were treated with 10 U DNase I (TaKaRa) at 37°C for 30 min. The resulting RNA was subjected to rRNA removal using a Ribo-Zero magnetic kit (for Gramnegative bacteria) (Epicentre Biotechnologies, Inc., Madison, WI, USA) following the manufacturer's instructions. rRNA-depleted RNA was used for RNA-seq analysis using the Illumina HiSeq 2500 at the Shanghai Human Genome Centre (Shanghai, China). Briefly, 100 ng rRNA-depleted RNA was used for construction of sequencing libraries using the NEBNext Ultra directional RNA library prep kit according to the manufacturer's instructions. The acquired raw data were processed using fastp software (43) to generate the clean reads that were mapped to the reference genome (http://www.pseudomonas.com) using Salmon (39). The transcription level was normalized to transcripts per kilobase per million mapped reads (TPM). Differentially expressed genes (DEGs) were defined based on a level of change of .2-fold and a P value of ,0.05 using DESeq2 (44).
Statistical analysis. A binomial test was applied to determine the nonrandom distribution. P values derived from multiple testing were adjusted using the Benjamini-Hochberg method (45). Differences between the means of two and more subgroups were tested using t test and one-way analysis of variance (ANOVA), respectively.
Data availability. The RNA-seq raw data sets were submitted to the NCBI database with the accession number PRJNA917454.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 1.7 MB.