Genome‐wide identification of tolerance mechanisms toward p‐coumaric acid in Pseudomonas putida

Abstract The soil bacterium Pseudomonas putida KT2440 has gained increasing biotechnological interest due to its ability to tolerate different types of stress. Here, the tolerance of P. putida KT2440 toward eleven toxic chemical compounds was investigated. P. putida was found to be significantly more tolerant toward three of the eleven compounds when compared to Escherichia coli. Increased tolerance was for example found toward p‐coumaric acid, an interesting precursor for polymerization with a significant industrial relevance. The tolerance mechanism was therefore investigated using the genome‐wide approach, Tn‐seq. Libraries containing a large number of miniTn5‐Km transposon insertion mutants were grown in the presence and absence of p‐coumaric acid, and the enrichment or depletion of mutants was quantified by high‐throughput sequencing. Several genes, including the ABC transporter Ttg2ABC and the cytochrome c maturation system (ccm), were identified to play an important role in the tolerance toward p‐coumaric acid of this bacterium. Most of the identified genes were involved in membrane stability, suggesting that tolerance toward p‐coumaric acid is related to transport and membrane integrity.

However, genome wide screening and identification of all genes involved in tolerance has been a limiting factor, which in the last years have been overcome through the use of next generation sequencing (NGS) techniques. Transposon insertion sequencing (Tn-seq) (Figure 1a), enables high-throughput identification of genomic regions involved in the survival of cells when exposed to various conditions (Barquist, Boinett, & Cain, 2013;van Opijnen & Camilli, 2013). A number of organisms have been investigated using this technique, thereby successfully identifying genes important for growth under different stress conditions, essentially by simultaneously investigating the fitness of all single mutants (Gawronski, Wong, Giannoukos, Ward, & Akerley, 2009;Langridge et al., 2009;Lennen & Herrgård, 2014;Santiago et al., 2015). So far, a Tn-seq method has, however, not been developed for P. putida.
Metabolic engineering of microorganisms for production of chemical compounds from renewable resources aims at providing an alternative to petroleum-based chemical processes. However, it still faces some challenges, such as ameliorating the toxic effects of intermediates and products of interest (Keasling, 2010). Numerous studies have focused on increasing the tolerance of different production strains toward various biochemicals using different strategies (Alper, Moxley, Nevoigt, Fink, & Stephanopoulos, 2006;Atsumi et al., 2010;Dunlop et al., 2011;Goodarzi et al., 2010). An alternative to this approach is to identify alternative production hosts that are naturally tolerant to higher concentrations of the chemicals of interest. P. putida is one such strain that is often utilized for bioremediation and production of more toxic chemicals (Loeschcke & Thies, 2015;Nikel, Martínez-García, & de Lorenzo, 2014;Poblete-Castro, Becker, Dohnt, dos Santos, & Wittmann, 2012;Verhoef, Ruijssenaars, de Bont, & Wery, 2007;Verhoef, Wierckx, Westerhof, De Winde, & Ruijssenaars, 2009).
In this study, we have investigated the tolerance of P. putida KT2440 toward a range of different biochemicals and found that it exhibits significantly increased tolerance toward the aromatic organic compound p-coumaric acid, a low molecular weight phenolic acid present in soils and plants (Mussatto, Dragone, & Roberto, 2007;Strobel, 2001). p-Coumaric acid is released as an inhibitory by-product during the pretreatment of the lignocellulosic feedstocks (Jönsson & Martín, 2016). There is furthermore significant industrial interest in producing p-coumaric acid since it can be used for making high performance polymers. Aditionally, it is a precursor for a number of interesting high value compounds (Santos, Koffas, & Stephanopoulos, 2011).
To study the mechanism of tolerance toward p-coumaric acid, we developed a novel method for generating random transposon insertion mutant libraries in P. putida KT2440. In combination with transposon insertion sequencing, the most important genes required for growth in the presence of p-coumaric acid were identified. A number of genes involved in maintenance of the membrane structure and efflux of solvent compounds were found to be important for growth, when P. putida KT2440 was exposed to this compound.
2 | EXPERIMENTAL PROCEDURES 2.1 | Bacterial strains and chemicals P. putida KT2440 (DSM-6125) and P. putida KT2440 Δfcs (Calero, Jensen, & Nielsen, 2016) were grown at 30°C and E. coli K-12 MG1655 was grown at 37°C. Cells were routinely cultured in LB medium or on LB agar plates according to standard protocols (Sambrook & Russel, 2000). Modified M9 minimal medium containing 5 g L −1 of glucose as carbon source (Abril, Michan, Timmis, & Ramos, 1989) was used for all assays. E. coli DH5α and E. coli DH5α-Pir were used for plasmid maintenance and were grown in LB medium at 37°C.
Mutants in genes ttg2A, ttg2B, ccmC, and ccmF were obtained from the Pseudomonas Reference Culture Collection (PRCC), consisting of insertion mutants generated by disruption of the genes using a miniTn5-Km (Duque et al., 2007).
Mutants in genes vacJ, tyrB, and fleN were constructed by introducing, by homologous recombination, a 3 kb fragment of DNA with a kanamycin resistance gene for its selection. For this purpose, plasmids pEMG-vacJ, pEMG-tyrB and pEMG-fleN were constructed CALERO ET AL. | 763 by cloning a 500 bp homologous fragment of the gene to be disrupted, using USER-cloning (Nour-Eldin, Geu-Flores, & Halkier, 2010). The correct plasmids, which were checked by colony PCR and sequencing, were electroporated into P. putida KT2440 strains and the resistance to kanamycin was selected. The correct insertion was checked by PCR.
Due to absorbance interference of the compound in the readings in 96-well microtiter plates, octanoic acid toxicity assays in P. putida KT2440 and E. coli K-12 MG1655 were performed in 50 ml of modified FIGURE 1 (a) Diagram of the experimental procedure used in a Tn-seq approach. A Tn5 insertion library (transposons depicted as green boxes) is constructed by transformation and used in growth assays in different conditions, in order to assess differential growth of each insertion mutant. After harvesting the cells, the extracted genomic DNA is processed by DNA shearing, A-tailing, and ligation of adapters (blue boxes), followed by an enrichment using PCR. This enriched library is subsequently sequenced by NGS and the results of the number of readings in the different conditions are compared in order to identify the differential abundance of insertions. (b) The method used in this study for generating the Tn5 library generating a combination of miniTn7-Gm and miniTn5-Km. The efficiency of the method to generate random insertion libraries was assessed by counting colonies of a sample M9 minimal medium in 250 ml shake flasks at 30 and 37°C, respectively, and 250 rpm of shaking. Overnight cultures grown in modified M9 minimal medium were diluted to obtain an initial OD 600 of 0.1 and incubated until OD 600 reached 0.5, after which octanoic acid was added to the culture. The concentrations used were 0, 30, 40, 60, and 100 mM. Growth was followed by measuring OD 600 every hour for 5 hr.

| Generation of miniTn5 insertions libraries
Introduction of plasmid pUT-Km into P. putida KT2440::miniTn7-Gm and P. putida KT2440 Δfcs::miniTn7-Gm was performed by triparental mating. A total of 20 ml of overnight LB pre-cultures of the recipient strains, the donor strain containing pUT-Km, and the strain containing the helper plasmid pRK600, were harvested and washed twice in a phosphate buffer, after which they were mixed in a 1:1:1 ratio and adjusted to a volume of 1 ml in phosphate buffer. A volume of 100 µl was placed on an autoclaved 0.45 µm filter (Durapore membrane filters, Merck Millipore, Hellerup, Denmark), using a total of 10 filters, and incubated overnight at 30°C on LB plates. Subsequently, cells on all filters were re-suspended in 8 ml of buffer. A total of 100 µl of this mix was diluted and plated on LB agar plates containing kanamycin and gentamycin for counting the number of mutants in the library, and the rest was used to inoculate 400 ml of LB with kanamycin and gentamycin in a 2 L shake flask, which was incubated at 30°C with 250 rpm shaking overnight. Aliquots of this culture were used to prepare cryo stocks in glycerol and stored at −80°C.
| 765 p-coumaric acid stress conditions. The cryo stocks, containing approximately 100,000 mutants, determined by colony counting on agar plates, were thawed on ice, washed twice with LB and grown overnight in 50 ml of LB with proper antibiotics at 30°C in 250 ml shake flasks. A total of 10 ml of the overnight LB culture was harvested for genomic DNA extraction. Part of the overnight LB culture was used to inoculate 50 ml of modified M9 minimal medium with antibiotics to a starting OD 600 of 0.1 after washing the cells with modified M9 minimal medium. Cells were incubated at 30°C and 250 rpm and growth was followed by measuring OD 600 every 2 hr. When cultures reached an OD 600 of 0.5, they were divided into 25 ml cultures and diluted two fold in preheated minimal medium or in minimal medium with 50 mM of p-coumaric acid. Growth was followed until cultures reached late exponential phase at an OD 600 of approximately 2, after which the cells were harvested and stored at −20°C prior to gDNA extraction.
Three independent biological replicates were carried out.
2.6 | Library preparation for Tn-seq gDNA libraries were prepared as previously described (Lennen & Herrgård, 2014 Library size distribution was validated by qPCR, using the KAPA qPCR quantification kit, the Agilent RNA 6000 Pico kit, and the Agilent 2100 Bioanalyzer. Finally, the library was sequenced on an Illumina MiSeq.

| Data analysis
Reads obtained from sequencing were initially processed in order to discard the unspecific sequences and trim the length of the correct ones to 120 bp, leaving 25 bp of genomic sequence that can be used for comparison to the P. putida KT2440 genome. Genes that were conditionally more abundant were identified using the ESSENTIALS pipeline (Zomer, Burghout, Bootsma, Hermans, & van Hijum, 2012).
Gene-level insertion counts were normalized using TMM normalization and p-values adjusted by Benjamini-Hochberg correction.
Genbank version NC_002947.3 of P. putida KT2440 was used as the reference genome.

| RESULTS
3.1 | Tolerance of P. putida KT2440 toward industrially relevant biochemicals P. putida has been widely used for bioremediation because of its high degree of tolerance toward many organic compounds, and it is furthermore a promising host for metabolic engineering. We therefore decided to screen P. putida KT2440 for its tolerance toward a number of chemical compounds that could be relevant for biobased production, and that are known to be relatively toxic to E. coli (Rau, Calero, Lennen, Long, & Nielsen, 2016). Six of the compounds, hydroxyγ-butyrolactone, furfural, itaconic acid, levulinic acid, succinic acid, and L-threonine, have previously been identified as being among the top 30 predicted chemical building blocks in a report from the U.S Department of Energy (Werpy & Petersen, 2004). Five other compounds were also selected, including acetic acid, which is a common inhibitor present in biomass hydrolysates; n-butanol, a potential biofuel and also widely used in the chemical industry; octanoic acid and 1,4-butanediol, which are polyester precursors; and p-coumaric acid, an aromatic polymer precursor, for which there is currently no efficient industrial production method.
P. putida KT2440 tolerance to these chemical compounds was tested by assaying their effect on the growth rate using different concentrations, using conditions identical to those used previously for E. coli K-12 MG1655 (Rau et al., 2016). In order to compare the effect the different compounds have on P. putida KT2440 and E. coli K-12 MG1655 growth, we calculated the concentration of each compound required to reduce the growth rate by 33%, as described previously (Rau et al., 2016). The concentrations of compounds needed to reduce the growth rate in P. putida KT2440 varied significantly. Some of the chemicals were found to effect growth at very low concentrations, including furfural, which was found to be toxic at a concentration of approximately 15 mM; levulinic acid and n-butanol at 40 mM; and hydroxy-γ-butyrolactone at 58 mM (Table 3). Others were found to require higher concentrations to achieve the same effect on the growth, such as threonine or 1,4-butanediol, for which concentrations of 412 and 399 mM were required, respectively.
When comparing to E. coli K-12 MG1655, most of the chemical compounds tested were found to have a similar or even higher inhibitory effect on P. putida KT2440 in the conditions used. However, higher tolerance toward L-threonine, octanoic acid, p-coumaric acid, and furfural was found in P. putida KT2440. The highest difference in tolerance was found for L-threonine, where P. putida KT2440 was found to be sevenfold more tolerant when compared to E. coli K-12 MG1655. P. putida KT2440 was also found to be twofold more tolerant to octanoic acid and p-coumaric acid, whereas a lower difference of 1.25-fold increased tolerance was found for furfural (Table 3). Since pcoumaric acid is a known toxic compound in biomass hydrolysate, and since its toxicity has been shown to affect production by fermentation (Sariaslani, 2007), we decided to study the mechanism of tolerance toward this compound in further detail.
3.2 | The deletion of fcs does not affect the tolerance toward p-coumaric acid P. putida KT2440 is able to degrade p-coumaric acid and use it as carbon source (Figure 2a) (Jiménez, Miñambres, Luis, & Díaz, 2002). In order to investigate the effect of the degradation of p-coumaric acid on tolerance, a strain with a deletion of the first gene in the p-coumaric acid degradation pathway, fcs, which abolishes p-coumaric acid degradation, or conversion to another compound , was tested in different concentrations of the compound. Its tolerance to p-coumaric acid was compared to the wild-type strain, which is able to degrade p-coumaric acid, and to E. coli K-12 MG1655.
Whereas a p-coumaric acid concentration of 30 mM was enough to reduce the growth rate of E. coli to 68% of the control grown without the compound, no effect on growth was observed when this concentration was added to the P. putida KT2440 Δfcs strain ( Figure 2b).

| Tn5 library generation in P. putida KT2440 and putida KT2440 Δfcs
Since degradation of p-coumaric acid was shown not to be involved in the increased tolerance of P. putida KT2440 toward p-coumaric acid, a random library of mini-Tn5 mutants was generated in order to further investigate the possible mechanisms of tolerance (Figure 1b). To obtain a mutant library with a high genomic coverage, a triparental mating of two plasmids, pUT-Km, containing the transposon, and pRK600, a conjugation helper plasmid was performed (Herrero et al., 1990). To have a heterologous antibiotic resistance cassette for selection of the recipients with the mini-Tn5 insertions and to avoid growth of other strains used in the triparental mating, a gentamycin resistance gene was introduced into the genome of P. putida KT2440 and P. putida KT2440 Δfcs using a mini-Tn7-Gm transposon (Figure 1b). This approach inserts the gene in a fixed position on the genome, after the gene glmS, which has previously been shown to be innocuous to the cell physiology (Lambertsen, Sternberg, & Molin, 2004). Using this method we avoided the natural chloramphenicol resistance of P. putida KT2440 as a selection marker, which would introduce a bias in the transposon insertion sequencing results for mutants involved in the chloramphenicol resistance, such as the efflux pump TtgABC . The use of alternative carbon sources, like citrate, that only the receptor strain could degrade was also avoided, and the developed method thereby enabled the use of LB in the creation of the library.
An efficiency of 10 −6 CFU/conjugant cells when using triparental mating was achieved (Figure 1b). Using this method, a pooled library of ∼125.000 mutants in P. putida KT2440 and ∼95.000 mutants in the   3.4 | Tn-seq results demonstrate that the structure of the membrane has an important role in the tolerance of putida KT2440 to p-coumaric acid Sequencing of the junctions between the transposon and the genome was achieved by generating an enriched library as previously described (Lennen & Herrgård, 2014). A pUT-Km transposon specific primer, BioTEG-Tn5-fw, was designed for the enrichment of the junctions ( whereas approximately 6,000 insertions with 20 reads or more were found in the P. putida KT2440 library. Using the generated libraries, we sought to identify genes involved in tolerance toward p-coumaric acid. A number of genes were found to be important for the growth of both P. putida KT2440 and P. putida KT2440 Δfcs when exposed to high  (Table 4).
A number of transposon insertions were also identified in just one of the two strain backgrounds. One of the most significant was the operon consisting of the genes ccmCDEF and dsbE in the strain P. putida KT2440 Δfcs. All the genes in this operon, which encodes the were found to have five and two different insertions, respectively. In all cases, the mutants in these genes were found to be represented approximately 13-fold less in the populations when p-coumaric acid was present in the media. In the P. putida KT2440 library, genes encoding some other membrane proteins such as the outer membrane protein OprH, a surface adhesion protein encoded by gene PP_0168 and an ABC transporter encoded by PP_0674 were found to be related to the tolerance toward p-coumaric acid ( Table 4).
The loss-of-function of two genes was found to increase the growth rate of both P. putida KT2440 Δfcs and P. putida KT2440 in the presence of p-coumaric acid. These genes include the LysR family transcriptional regulator encoded by the gene PP_1262 and the MarR family transcriptional regulator encoded by the gene PP_4515. In both cases, the cells containing transposon insertions in these genes were found to grow better when p-coumaric acid was present and were approximately 15-fold more abundant under these conditions.
3.5 | Loss of function in genes identified through the Tn-seq leads to impaired growth in the presence of p-coumaric acid Two complete operons, ttg2 and ccm, were identified in both the assayed strain as playing an important role in the tolerance of P. putida KT2440 toward p-coumaric acid (Figure 3b). To further investigate the role of these genes, independent insertion mutants available from the PRCC (Duque et al., 2007) in genes ttg2A, ttg2B, ccmC, and ccmF were acquired, while insertion mutants in genes vacJ, tyrB, and fleN were constructed. The different strains were grown in medium containing different concentrations of p-coumaric acid, and a clear impact on the final optical density of the mutant strains was found when compared to the wild-type strain (Figure 3c), and a similar effect was observed for the growth rate ( Figure S1). This result confirms that both the ttg2 and ccm operons as well as other genes identified through the Tn-seq experiment are strongly involved in the tolerance toward the aromatic compound p-coumaric acid.

| DISCUSSION
When compared to E. coli, P. putida KT2440 was found to have enhanced tolerance to only three out of the 11 tested industrially for their tolerance to certain organic solvents such as toluene, P. putida KT2440 was not found to be more tolerant toward carboxylic acids (levulinic acid, acetic acid, itaconic acid, and succinic acid) and furfural when compared to E. coli. Surprisingly, P. putida KT2440 was not found to be more tolerant toward n-butanol, even though some P. putida strains, including P. putida DOT-T1E and S12, have a high tolerance toward organic solvents such as n-butanol (Rühl, Schmid, & Blank, 2009 (Ramos et al., 2002). Interestingly, some defect on the growth was observed only for the wild-type strain when exposed to 30 mM of p-coumaric acid, and not for the deletion mutant, suggesting that certain intermediates in the degradation pathway may be toxic (Figure 2b).
To further investigate the mechanism of tolerance toward p-coumaric acid, we developed a new method for generating large transposon libraries compatible with Tn-seq. By comparing sequences of libraries grown with or without inhibitory concentrations of p-coumaric acid, we identified 19 genes involved in the tolerance to p-coumaric acid, many of them involved in membrane processes.
These results are in accordance to a previous study that investigated the mechanism of action of p-coumaric acid in different bacteria, and pointed toward membrane disruption as a main effect (Lou et al., 2012). The genes in the ABC (ATP-binding cassette) transporter, Ttg2ABC, were part of the seven genes for which loss-of-function indicated impaired growth in both strain libraries when exposed to high concentrations of p-coumaric acid. This transporter has previously been identified to play a role in the tolerance toward toluene in P. putida Kim, Lee, Lee, & Lim, 1998), and it has been described to be involved in resistance to a number of other stress agents .
Another gene with decreased abundance in both libraries when exposed to p-coumaric acid, vacJ (PP_2163), contains the domain of the VacJ superfamilies. VacJ has been characterized as an outer membrane protein associated to an ABC transporter system in other organisms, such as E. coli and Shigella flexneri, in which it has been proposed to be involved in the maintenance of outer membrane stability in the presence of membrane disruptors such as SDS (Carpenter et al., 2014;Malinverni & Silhavy, 2009). Its deletion has been shown to increase permeability of the outer membrane (Malinverni & Silhavy, 2009) and increase the formation of vesicles in E. coli and other gram-negative bacteria (Roier et al., 2016). In P.
aeruginosa, a VacJ homolog has been described to play a role in antibiotic resistance. VacJ has also been shown to be more abundant in cells growing in the presence of phenol in P. putida KT2440 .
A transposon insertion in the flagellar number regulator fleN was shown to be important for tolerance toward p-coumaric acid. The deletion of this gene in P. aeruginosa has been shown to alter the single polar flagella localization, affecting the motility of the cells (Dasgupta, Arora, & Ramphal, 2000). Even though some genes of the flagella systems have been identified in the tolerance toward toluene in the strain P. putida DOT-T1E, their function in the resistance is not clear, since it does not seem to be related to the cells motility (Ramos et al., 2002).
In one of the libraries, the complete ccmCDEF-dsbE operon, containing the genes PP_4321-PP_4325, was found to be of especial importance for tolerance toward p-coumaric acid. These genes encode the cytochrome c maturation system, which is known to be present in Proteobacteria, where it promotes the attachment of the heme group to the apocytochrome c in the periplasm, allowing the correct function of the type c-cytochrome in cell respiration and ATP synthesis (Thöny-Meyer, 2000). The effect of the lack of a functional operon in tolerance toward p-coumaric acid may indicate a high energy requirement or disruption of the membrane potential, causing the release of cytochrome c, as it has previously been shown in mitochondria (Shailasree, Venkataramana, Niranjana, & Prakash, 2014). Moreover, an increased relative abundance of genes involved in energy metabolism in E. coli has also been shown to play a role in free fatty acids stress conditions (Lennen et al., 2011).
Furthermore, the cytochrome c maturation system has been described to play a role in other cellular processes in other organisms (Cianciotto, Cornelis, & Baysse, 2005), such as P. aeruginosa (Baert, Baysse, Matthijs, & Cornelis, 2008) or P. fluorescens (Yang, Azad, & Cooksey, 1996). A transposon insertion mutant in this operon in the strain P. putida P8 was found to have reduced cis-trans isomerization of unsaturated fatty acids, since a cytochrome c-type heme-binding motif was found to catalyze the activity of this enzyme.
In addition to the genes discussed above, other genes such as the transcriptional repressor phaN, a number of outer membrane proteins and surface adhesion proteins as well as other ABC transporters (PP_0168, PP_0674, and PP_1185) were identified to be important for tolerance, most of them in the P. putida KT2440 background. It is possible that these genes may be related to the detoxification of the intermediates of the p-coumaric acid degradation pathway.
The identification of these efflux pumps supports the observation that the role of transporters are of major importance when P. putida KT2440 is exposed to chemical stress, as it has also been observed in other organisms such as Clostridium acetobutylicum in the presence of butanol.
The importance of some of these genes was validated by assessing the effect of p-coumaric acid on the growth of a number of independent insertion mutants in the genes ttg2A and ttg2B, ccmC and ccmF, vacJ, tyrB, and fleN, which displayed clear growth impairment at high concentrations of p-coumaric acid when compared to the wild-type (Figure 3b).
One of the beneficial insertions found, in gene PP_4515, is annotated in P. putida KT2440 as belonging to the MarR transcriptional regulator family. However, a BLASTp analysis of the protein sequence showed that this transcriptional regulator seems to be more closely related to the transcriptional regulator SlyA, with a 38% of identity in the amino acid sequence. SlyA belongs to a large family of regulators involved in different processes, and it has been shown to regulate chaperones and other stress proteins in one strain of E. coli (Spory et al., 2002). However, the activity of various homologs differs widely among species, thus further work is required in order to understand the cellular mechanism in P. putida KT2440 of this regulator. The other gene, PP_1262, belonging to the LysR-type transcriptional regulator family has no known function. The LysR-type transcriptional regulator family has been described to regulate a number of functions, such as aromatic acid metabolism, and both activators and repressors have been identified. Most of them are described to be transcribed divergently of the genes they regulate. Here, the operon PP_1263 to PP_1266, which is divergently transcribed, has been annotated as an efflux system for the export of fusaric acid, a fungal toxin released by phytopathogenes (Martins Dos Santos, Heim, Moore, Strätz, & Timmis, 2004). Further studies are required to elucidate the effect of this regulator and the genes that it possibly regulates for the tolerance.
In conclusion, we have developed a method for generating large libraries of random insertion mutants in P. putida KT2440 with a minimum amount of bias by avoiding natural P. putida selection methods. P. putida KT2440 was found to have enhanced tolerance toward p-coumaric acid when compared to E. coli. Using Tn-seq analysis, the mechanisms of tolerance were found to involve membrane stability, exclusion, and secretion of the toxic compound to the outside of the cell. Due to its significant tolerance, P. putida could be a promising production host for p-coumaric acid. The developed Tn-seq method will be generally useful for investigating a wide range of conditions in this organism.