Genomic Responses of Pseudomonas putida to Aromatic Hydrocarbons
When Pseudomonas putida mt-2 cells face aromatic hydrocarbons, they must make a number of decisions that exemplify virtually every challenge that environmental bacteria need to overcome for thriving in polluted sites. It is thus no surprise that the genome-wide responses of P. putida to a variety conditions have been studied in considerable detail. One long-standing issue is how P. putida handles nutrient choice when a mixture of substrates coexist in the same site. A complex mechanism reminiscent of catabolite repression is in place to ensure that hydrocarbons are tackled only after virtually all other easy nutrients are depleted. Once this occurs, typical components of the lighter aromatic fractions of petroleum (e.g., toluene and xylenes) are not only potential nutrients but also strong stressors for P. putida. For handling this scenario, simultaneous processing of their presence both as an edible carbon source to benefit from and as a detrimental hassle to keep at bay becomes necessary. A second issue is the metabolic jam caused by the coincident presence in the same cells of competing pathways for ortho-cleavage and meta-cleavage of the intermediate catechols that degradation of aromatics must necessarily go through. Availability of a suite of genomic, proteomic and reporter technologies has shed light on these questions and exposed a number of surprising solutions that would otherwise look like intractable biochemical and regulatory problems. These include not only specific biochemical and transcriptional devices that overcome intracellular conflicts but also diversification of the population for acquiring different phenotypic roles.
The genus Pseudomonas includes a large group of aerobic Gram-negative γ-proteobacteria that are remarkable for their metabolic versatility and their ability to colonize a very large variety of niches, from plant roots and leaves to soil and human lungs (Nikel et al. 2014a; Belda et al. 2016). Much of this versatility stems from a very robust central metabolic core that is characterized by the glycolytic Entner-Doudoroff (ED) pathway. Unlike the Embden-Meyerhof-Parnas biochemical route, the ED-based EDEMP cycle generates high levels or reducing currency that empowers Pseudomonas putida to tolerate higher doses of environmental stress and hosting redox reactions that would be too toxic for other species (Nikel et al. 2015). Furthermore, the frequent occurrence of membrane-bound extrusion pumps enables the genus to withstand many antibiotics and, by the same token, to endure exposure to organic solvents (dos Santos et al. 2004; Ramos et al. 2015). These properties shared by most members of the genus have diversified for the sake of adaptation to specific niches, one of which – soil polluted with aromatic hydrocarbons – is the focus of this Chapter. Different species and strains of Pseudomonas have been isolated from such sites because of their capacity to use as sole carbon and energy source compounds like toluene and xylenes. Out of them, the one specimen that has become the experimental model of choice for studying the molecular details of the interaction aromatic hydrocarbons-microbe is strain P. putida mt-2.
The regulatory and biochemical scenario described above immediately raises a good number of questions. How do Pseudomonas handles toxicity and stress caused by aromatic hydrocarbons while at the same time metabolizes them as nutrients (Reva et al. 2006)? Why are the genes encoding the TOL pathway split in two parts; each with a different TF? Other routes of Pseudomonas for degradation or aromatic hydrocarbons (and most metabolic pathways) have just one operon and one regulator. Given that P. putida has a separate chromosomally encoded pathway for metabolism of benzoate through an ortho-cleavage track of the intermediate catechol, how can the two routes coexist in the same cells? Finally, what is the logic of having such an intricate regulatory network when the same effects could be achieved prima facie with a much simpler genetic device? In this article, we argue that these complexities are by no means casual, but they reflect aspects of the P. putida lifestyle in their natural niches that may not be evident by just growing the cells in a flask. As discussed below, the main tools for unveiling these qualities involve genomic, proteomic and single-cell reporter technologies which, taken together, expose an unexpected degree of sophistication of these bacteria for dealing with a large number of environmental challenges.
2 Aromatic Hydrocarbons: Nutrients or Stressors?
One of the most intriguing questions showcased by bacteria that thrive in polluted sites is how they manage to metabolize compounds that are inherently harmful to life (de Lorenzo and Loza-Tavera 2011). While some non-degradable pollutants (e.g., heavy metals) are dealt with through their inactivation or pumping out of the cells, potentially edible organic contaminants need to balance a suitable stress response to their detrimental effects with the action of cognate metabolic routes for their assimilation. The starting point is therefore that the same compound (e.g., toluene/m-xylene) is both a stressor and a nutrient, each aspect activating different genetic and physiological responses that may compete with each other. How does P. putida compute this two-sided signal and decides what to do? This question could start to be addressed only when genomic technologies became available for studying simultaneously the expression profiles of many genes. An early study of the question (Velazquez et al. 2005, 2006) was done with homemade DNA chips printed with a selection of the structural and regulatory DNA sequences of the TOL pathway together with selected descriptors of specific physiological conditions. P. putida bearing the TOL plasmid pWW0 was then exposed to m-xylene along with stress conditions at doses that did not cause growth effects. The insults included various oxygen tensions, temperatures and nitrogen sources as well as situations of DNA damage, oxidative stress, carbon and iron starvation, respiratory chain damage, and contact with arsenic – each of them know to elicit a distinct physiological response. The effects of each of the stress classes were categorized in respect to the relative output of xyl transcripts. Some results of these experiments were somewhat expected. Given that the metabolic TOL program had to compete with the stress-enduring programs for gene expression resources (ribosomes, RNAP), most of the conditions downregulated the m-xylene biodegradation-related genes. Some others, however, appeared to stimulate expression of the TOL-encoded xyl genes instead. These included uncouplers of the respiratory chain (azide) and small doses of arsenate. But most important, replacement of NH4+ by NO3− as N source increased expression of the TOL cistrons as well. This is of interest given that soils are often amended with fertilizers based on either ammonia or nitrates. To inspect whether such N-regulation observed in the test tube was propagated into actual catabolism of m-xylene under natural conditions, a test-tube-to-soil model system was developed which verified that that NO3− had a stimulating effect on xyl genes expression and m-xylene mineralization in microcosms as compared to NH4+. This type of information will surely be useful for planning future bioremediation interventions. In a further screw turn, P. putida mt-2 cells were subject to the multiple abiotic stresses caused by exposure to crude tar from the 2002 oil spill of the Prestige tanker, which contained a complex mixture of hydrocarbons. In this case, the expression profile of xyl genes and stress-responding markers were followed over time. These results clearly suggested that adaptation to external insults precedes any significant expression of the catabolic genes (Velazquez et al. 2006). Limited as they are, these results provided two valuable pieces of information: [i] as expected, aromatic hydrocarbons are indeed sensed by P. putida mt-2 both as nutrients and stressors and there is competition between the physiological responses to each of them, and [ii], the way cells handle the challenge is by first adapting to the stress and then deploying the ability to metabolize the corresponding chemicals.
As the genomics technologies improved, the same question was revisited in more detail by using a genome-wide DNA chip (Dominguez-Cuevas et al. 2006; Yuste et al. 2006) rather than just a TOL chip and few indicator genes as before. This technique allowed visualization of the entire response of the hydrocarbon-exposed P. putida mt-2 strain to either the head substrates of the pathway and some of the metabolic intermediates. Furthermore, the same method revealed how the biodegradative pathway influenced the stress caused by the aromatic substrate. In this way, the interplay between the TOL route versus the background metabolism could be dissected and the activity of stress-responses versus substrate assimilation blown up with great accuracy. To this end, Dominguez-Cuevas et al. (2006) evaluated the genome-wide effects of three aromatic compounds chosen based on their known differential roles as inducers of the TOL degradation pathway. These included toluene (head substrate and primary inducer of the route), m-toluate (i.e., 3-methylbenzoate [3MBz] a metabolic intermediate and inducer of the lower pathway) and o-xylene (a non-metabolizable effector of XylR). Cultures of P. putida grown to exponential phase on succinate were exposed to these compounds and whole genome expression profiling was done using dedicated oligonucleotide-based DNA microarray representing the entire genomic complement of the strain under study. Using as a reference >1.8-fold change in transcript levels, the results indicated that 180, 185, and 64 genes were upregulated in response to toluene, o-xylene, and 3MBz, respectively. Out of all these, 36 genes were over expressed in the presence all these aromatics, while 71, 67, and 11 ORFs were differentially transcribed. In contrast, the same compounds downregulated 127, 217, and 69 genes (only 18 common in all three conditions) while 130, 36, and 33 were inhibited in response to o-xylene, toluene, and 3MBz exclusively. These observations clearly indicated that there are substantial differential responses to the three aromatics in terms of up and down regulation of genes – and thus they are sensed and physiologically responded to separately. At the same time, the type of genes involved in such a response hint at the way P. putida detects these aromatics in the medium and reacts to their presence by eliciting specific response programs – both regarding metabolic and stress-management. The most conspicuous cases are discussed below.
3 Membrane-Associated Functions
The first biological material that aromatics interact with when added to P. putida cultures is the cell membrane. The bi-layered bacterial envelope is a barrier to be crossed for making it inside through more or less specific transport, and also a hydrophobic chemical milieu where aromatics easily dissolve – thereby influencing the membranes structure and functionality (Ramos et al. 2015). It thus comes as no surprise that the most conspicuous transcriptional responses to exposure to toluene, o-xylene, and 3MBz are detected in genes encoding envelope-related utilities. These include lipid metabolism, transport of nutrients and small molecules, flagellar machinery, and diverse types of pili. Not surprisingly, o-xylene had the maximum impact on the regulation of this class of genes, thereby suggesting that cells perceive this non-biodegradable molecule as more toxic than the others. Interestingly, the three compounds cause a complete loss of motility because of its inhibition of genes for assembly of the flagella. This could reflect a defensive strategy of P. putida through which a significant amount of ATP is saved by shutting down the high energy-consuming synthesis of the swimming apparatus (8% of total protein) and channeling it instead for dealing with the aromatics. In addition, there was down regulation of a gene associated with cell division (parA), while functions associated with efflux pumps become up regulated. The same aromatics also exerted indirect effects on cells because of their harm to membranes. The prime consequence is the down regulation of genes associated with the electron transport chain, concomitantly with an increase in transcription of glutathione and glutathione S-transferases. These observations surely reflect that perturbing membrane composition generates active oxygen species (and ensuing oxidative stress) that need to be counterbalance with production of reductive metabolic currencies.
4 Effects on Metabolism of Aromatic Compounds
5 The Heat Shock Connection
Regardless of the type of transcriptomic technology adopted, one of the most noticeable effects of having cells of P. putida exposed to aromatic hydrocarbons is the upregulation of the heat-shock (HS) response, in particular many characteristic genes encoding HS proteins. Their identity and relative expression levels do change, however, depending on the exact compound and its concentration in the medium. The non-substrate o-xylene was systematically a better inducer of the HS response as compared to the others. This is likely to be related to the fact that nonmetabolizable aromatics accumulate in the cell envelope because there is no way to get rid of them other than by membrane-associated pumps. In contrast, and depending on the dose, metabolizable hydrocarbons can be drained from the inner membrane and channeled towards the corresponding enzymatic devices for their degradation. Not surprisingly, the weakest inducer of the HS response was 3MBz, as it is more polar and thus less prone to cause membrane damage. It is intriguing that the peak of HS caused by aromatic hydrocarbons coincides with a significant induction of genes associated with amino acid biosynthesis. This may expose a requirement of new proteins for mounting the stress related response and – in the case of metabolizable substrates – utilization of the aromatics as carbon sources.
The interplay between HS and degradation of aromatics by the TOL system may involve more than just competition for cellular resources to run either a stress-response program or a metabolic program. There are also imprinted mechanistic details of the regulatory network that controls expression of either system. When cells growing on succinate are exposed to m-xylene, it takes as little as 15 min to detect full expression of the upper pathway, while it takes 3 h to see a full-fledged meta route. However, xylS expression is observed also after 15 min of m-xylene addition – along with expression of upper pathway genes. What is the logic of this apparently incongruous scenario? One intriguing link between aromatic biodegradation and heat shock is that XylS activates works in concert with variants of the RNAP that contain either the stationary phase and general stress response sigma factor σS or the HS-specific σH, which is by default in charge of expressing functions for coping with heat stress (Marques et al. 1999; Dominguez-Cuevas et al. 2005). At the same time, artificially induced HS conditions (e.g., growth at 42 °C) clearly indicated that heat downregulates both the upper and the meta pathway. This state of affairs is prone to generate a considerable stochasticity in expression of each of the pathways, as has indeed been proven experimentally. In addition, the connection of meta pathway expression to HS is reflected in other phenomena. For example, the level of expression of the meta pathway is different depending on whether the head substrate elicits better or worse HS. Addition of 3-methyl benzyl alcohol bypasses the energy-consuming mono-oxygenase step and thereby accelerates accumulation of 3MBz in a shorter period of time. This results in faster transcription kinetics of the lower TOL pathway but – unexpectedly – in inferior expression levels as compared to m-xylene. This is because the latter is a better trigger of the HS response and thus of higher levels of σH available for transcribing the lower operon. Interestingly, direct induction of the pathway upon addition of 3MBz to the medium results in a moderate upregulation of the meta pathway compared to m-xylene. Although this is due in part to the lack of the cascading effect of XylR on XylS that control expression of the TOL system, one primary reason for this modest induction is that 3MBz is a poor inducer of HS proteins and therefore σH is likely to be in short supply. Finally, a yet unexplained phenomenon also related to differential expression of TOL genes is the divergent effect of bona fide substrates like m-xylene versus gratuitous, non-metabolizable XylR effectors such as o-xylene. While both of them bring about an approximately similar HS level and activate transcription from the Pu promoter equally well, mRNAs become terminated earlier when o-xylene is used. While this makes perfect sense from the point of view of the cell’s economy, the mechanisms behind the phenomenon remain obscure.
6 Genomic Insights into the Control of Expression of the TOL Plasmid Aromatic Degradation Pathways
Besides the features of the regulation of the TOL catabolic operons discussed thus far and reviewed by Domínguez-Cuevas and Marqués (2017), the paragraphs below expose some new, emerging qualities of the system. They were uncovered by genomic analyses that contribute to highlight the profile of this experimental system as the paradigm of optimal catabolic device in environmental bacteria.
7 Revisiting the Regulation of the TOL System with Deep Transcriptomics
The insights on the responses of P. putida to aromatic hydrocarbons addressed above were largely generated with gene fusion and reporter technology and somewhat simple DNA arrays. In 2016, the first article (Kim et al. 2016) was published revisiting the same questions but using two far more powerful methods: DNA tiling arrays and deep RNA sequencing. The combination of the two afforded an unprecedented level of resolution of the m-xylene/toluene biodegradation subtranscriptome. This allowed the dissection of operon structures and also monitoring of the interplay between plasmid and chromosomal functions. These new methodological approaches exposed new characteristics of the system that could not be grasped by traditional genetics or more primitive genomic technologies. The highlights included the finding that a significant level of basal expression of the lower TOL pathway does occur in the absence of inducers. In contrast, virtually no expression of the upper route occurred. Furthermore, the overlapping transcription between the 3′ ends of the convergent xylS and xylH mRNAs did not have regulatory consequences. No role for small RNAs could be identified either. However, the most surprising finding was that benzoate generated from toluene catabolism by the upper pathway failed to induce expression of the ortho cleavage genes for the same compound encoded in the chromosome. How this can happen mechanistically is uncertain but suggests a degree of channeling that avoids misrouting of the benzoate produced by the upper TOL pathway to the ortho route – an issue that has been separately examined (Perez-Pantoja et al. 2015).
In P. putida mt-2, the regulation of Bz and 3MBz degradation routes relies on transcriptional regulators belonging to AraC-family; the so-called XylS and BenR activators. The pWW0-encoded XylS regulator mediates transcriptional activation of the Pm promoter driving the expression of the meta pathway in response to 3MBz and Bz as inducers (Fig. 2). Likewise, BenR is able to trigger the activity of Pben promoter by recognition of Bz as effector, allowing the expression of the chromosomal benABCD-catA2 operon. Interestingly, BenR-activation of the Pm promoter of the meta TOL pathway in response to Bz has been recognized. The opposite may be true as well, and XylS could activate the chromosomal Pben promoter. If that were the case, the question is how the corresponding parameters are set to avoid the predicted scenario of metabolic competition between the TOL-encoded enzymes and the genome-encoded counterparts. By adopting a suite of supersensitive promoter fusions to a luxCDABE reporter, for transcriptomic analyses of benA and xylX under diverse conditions and growth tests of strains bearing different regulatory combinations, Perez-Pantoja et al. (2015) showed that expression ranges of XylS under various conditions are insufficient to cause a significant cross-regulation of Pben whether cells face endogenous or exogenous aromatics. This seems to stem from the nature of the operators for binding either transcriptional factor, which in the case of the chromosomal Pben promoter of P. putida mt-2 appear to have evolved for avoiding a strong interaction with XylS. It thus seems that rather than physical channeling, some metabolic clashes are avoided through a mere adjustment of the regulatory parameters that control the interplay between otherwise conflicting metabolic devices.
But this is not the only metabolic problem between the meta-cleavage pathway borne by the TOL plasmid and the ortho-cleavage route encoded in the chromosome of P. putida mt-2. It turns out that this strain has a second chromosomal copy of the 1,2-catechol dioxygenase catA gene (named catA2) located downstream of the ben operon that encodes an additional enzyme with the same activity but different parameters. What is the role of such an enzyme? What is it needed for? (Jimenez et al. 2014) showed the key function of CatA2 as a sort of metabolic safety valve for excess catechol when cells are exposed to benzoate and the substrate can be simultaneously processed by the TOL system and the chromosomal ben/cat pathways. In this situation, CatA2 activity alleviates the metabolic conflict generated by simultaneous expression of the meta and ortho pathways, thereby facilitating their coexistence. The high level of paralogy found in the genome of P. putida is intriguing, as sequence redundancy is a source of genetic instability. But at least in the case of the two functional catA genes just mentioned, it seems that their retention has to do with solving regulatory problems. In reality, counteracting accumulation of toxic intermediates must be a general principle of metabolic evolution (de Lorenzo et al. 2015) which, in this case, has been solved by stitching in an extra gene.
8 Metabolic Division of Labor? Stochastic Expression of the Catabolic Operons of TOL Plasmid
The basic regulatory blueprint of the TOL system and the very low levels of the principal regulator of the system, XylR, create a scenario prone to strong stochastic effects. When expression dynamics of the catabolic promoters borne by pWW0 was examined in single cells exposed to m-xylene, some unexpected results became apparent. While expression of xylR behaved in a unimodal fashion, the activation of Pu and Pm displayed a high degree of bi-modality which resulted in a time-dependent shift in the bacterial population between two induction states (Silva-Rocha and de Lorenzo 2012b). The most striking feature of this stocasticity was observed with activation of the meta pathway by XylS. It was demonstrated that 3MBz-bound XylS (XylSa) has a very minor role in activation of the route while it is largely governed by effector-free XylS overproduced through the action of XylR on Ps. In addition, when induced with 3MBz alone, much of the population remains silent and does not express the lower pathway. This observation clearly indicates alternate induction mechanism of meta-pathway by XylS and XylSa (i.e., bound to 3MBz). Yet, higher levels of induction of meta-pathway in presence of m-xylene corroborates previous observations of synergistic effect of both XylS and XylSa on activation of the meta-pathway. These stochastic phenomena disappeared in the stationary phase. This suggested that when cells face a mixture of succinate and m-xylene, only part of the bacteria is actively engaged with degrading m-xylene while the rest instead grow on succinate (Silva-Rocha and de Lorenzo 2012b). Such a metabolic bifurcation of the population ends when cells run out of succinate and have only m-xylene during the stationary phase. This phenomenon could be a case of metabolic bet hedging, meaning that a genetically homogeneous population splits into different metabolic regimes when bacteria confront a mixture of nutrients.
It is likely that such a metabolic specialization is not just a fortuitous occurrence but is mainly an intrinsic feature of the regulatory circuit that has been duly selected as an advantageous trait (Nikel et al. 2014b). Interestingly, when compared to Pu and Pm-mediated activation of the plasmid-encoded TOL pathways, the trigger of the Pben-related chromosomal counterpart (the ortho pathway) in a plasmid-free strain follows a unimodal activation (Silva-Rocha and de Lorenzo 2014). The situation is, however, changed when cells acquire the TOL plasmid, as then the activation becomes bimodal. A similar transition was observed with the cat pathway for catechol metabolism, which acquires a stochastic expression regime in the presence of the TOL plasmid. This observation demonstrates how recruitment of a new metabolic pathway influences the expression of genes present on the genome and provides some insights on evolution of metabolic pathways.
9 Outlook and Research Needs
The onset of genomic technologies has allowed revisiting of the responses of Pseudomonas putida to aromatic hydrocarbons, with a focus on the interplay between the TOL-plasmid bearing strain P. putida mt-2 and m-xylene. In general, the data confirms the regulatory schemes already derived from genetic experiments, but they also enable a view of global responses to the hydrocarbon and also expose the evolutionary solutions found to what would otherwise be intricate biochemical conflicts. This allows us to put in perspective the apparently exaggerated over-control of the TOL system, with two operons, five promoters, two plasmid-encoded transcription factors, and a large number of host’s regulatory components. We believe that the complex regulatory architectures that we often find in environmental biodegradative systems have been shaped by prior biochemical, populational, and community conflicts. Microbes must solve simultaneously many issues, and the architecture of genetic circuits surely encodes a record of historical bottlenecks (e.g., biochemical chaos) as well as the evolutionary novelty that has solved them. This resembles what in some proteins has been called moonlighting, i.e., the property of a single polypeptide to hold entirely different functions in the same structure (Silva-Rocha et al. 2013). The data obtained with the TOL plasmid seem to show that one regulatory scheme can solve a considerable number of conflicts at the level of single cells, populations, and even multi-strain communities. Therefore it could well happen that regulatory devices encode both the traffic lights for physiological and biochemical issues along with a code of social conduct of environmental bacteria. This is an issue that certainly deserves future investigations.
Authors are indebeted to Bastian Voegeli, Simon van Vliet, Colette Bigosch, and Martin Ackermann (EAWAG, Switzerland) for sharing the data shown in Fig. 3. Pablo Nikel is gratefully acknowledged for reading and commenting the manuscript. The work in VdL’s Laboratory was funded by the HELIOS Project of the Spanish Ministry of Science BIO 2015-66960-C3-2-R (MINECO/FEDER), the ARISYS (ERC-2012-ADG-322797), EmPowerPutida (EU-H2020-BIOTEC-2014-2015-6335536) and MADONNA (H2020-FET-OPEN-RIA-2017-1-766975) Contracts of the European Union, and the InGEMICS-CM (B2017/BMD-3691) contract of the Comunidad de Madrid (FSE, FECER). Authors declare no conflict of interest.
- Domínguez-Cuevas P, Marqués S (2017) Current view of the mechanisms controlling the transcription of the TOL plasmid aromatic degradation pathways. In: Rojo F (ed) Aerobic utilization of hydrocarbons, oils and lipids. Springer International Publishing, Cham, pp 1–22Google Scholar
- Dominguez-Cuevas P, Marin P, Ramos JL, Marques S (2005) RNA polymerase holoenzymes can share a single transcription start site for the pm promoter. Critical nucleotides in the −7 to −18 region are needed to select between RNA polymerase with σ38 or σ32. J Biol Chem 280:41315–41323CrossRefGoogle Scholar