We used an agar plate colony biofilm model to investigate the spatial distribution of aerobic respiration and denitrification activity in an ostensibly oxic in vitro environment, i.e., what might be expected in the upper and lower airways. Like other studies we found that aerobic respiration near the surface of the P. aeruginosa colony biofilm colony created an anoxic region in the interior6,25, in our case at below approximately 100 µm depth. Bacterial denitrification in the lung was originally hypothesized on the basis of the CF lung being hypoxic8, due to increased epithelial oxygen consumption in combination with luminal hypoxia due to mucus plugging26; however as we, in line with others, show P. aeruginosa biofilms can create their own anoxic niches. On the addition of NO3- to the bulk liquid we observed an immediate response by the generation of N2O. The rapid response suggested that the various enzymes involved in the denitrification steps were already expressed and functional allowing the cells to take immediate advantage of NO3- as a terminal electron acceptor. Surprisingly, we saw denitrification occurring even in the outer, oxic, layers of the colony biofilm demonstrating that, in the presence of NO3-, denitrification is possible even if the local environment is not hypoxic and the biofilm or biofilm aggregates are not large enough to create anoxic regions. Previously it was assumed that denitrification would only take place in the anoxic zones deep in the biofilm27. We used a concentration of 1.2 mM NO3-, which was relatively high compared to physiological concentrations, however concentrations of approximately 0.8 mM have been reported in sputum from CF patients28, the pattern of denitrification distributed throughout the biofilm would be expected to be similar at lower concentrations. We found that although denitrification was occurring throughout the biofilm colony there was a high degree of variation at different depths in the biofilm (Fig. 4). This finding can be explained by a recent report using mRNA fluorescent in situ hybridization (FISH) reporters for metabolic genes for in situ single cell imaging which showed that in a P. aeruginosa biofilm there were single cells to pockets of cells approximately 25 µm in diameter undergoing different metabolic pathways29. By direct measurement of denitrification within P. aeruginosa biofilm colonies biofilms our data supports the conclusions made by Hassett et al.3 and Yoon et al.7 that denitrification is occurring in the infected CF lung based on transcriptomic profiling.
With respect to antibiotic tolerance generally, it is assumed that part of the mechanism in P. aeruginosa biofilm infections, such as found in infected lung and chronic wounds, is due to inactivity within the biofilm due to nutritional limited dormancy. This hypothesis is based largely on experiments conducted in such a manner that nutrient conditions predilect for aerobic respiration. However, in the presence of NO3-, there is likely activity, albeit lower than that when metabolizing aerobically (i.e. Figure 5), all the way through the biofilm. After addition of tobramycin we observed an immediate reduction in aerobic respiration, but the microprofiles suggested it was never completely arrested. Furthermore, on removal of the tobramycin, the gradient began to steadily reestablish, suggesting that although there was inhibition there was no complete killing over the short exposure time of one hour. Recently it has been reported that phezanine, which is produced by P. aeruginosa, and can promote survival under anoxic conditions, can also provide tolerance to antibiotics6 which may explain our results.
The agar colony biofilm model
Our static biofilm model on agar plates immersed in MSM allowed the set-up of the microelectrodes to achieve simultaneous measurement of DO and N2O in close proximity in terms of location and depth in the biofilm was technically challenging. An advantage of using agar-grown colony biofilms was that we limited the probability of breaking the microelectrodes, which is a critical issue with biofilms grown on rigid surfaces such as glass. A disadvantage of using a soft material such as agar was that we could not detect the base of the colony biofilm with the applied sensors at the applied resolution and recording time, and unlike with a rigid surface, gradients developed at the biofilm-agar interface and into the agar as evidenced by the inflection point in N2O concentration at approximately 250 µm depth as shown in Fig. 2. Another limitation of the model was that its complexity made it difficult to operate at 37ºC, which is more relevant when discussing our results in the context of infections. It is likely that the profiles would be similar, but conversion rates would be underestimated at room temperature, and DO would be less at 37ºC than room temperature. Historically bacterial biofilms have been studied using experimental models in which biofilms are grown on rigid surfaces immersed in an aqueous solution containing nutrients for growth30. However, increasingly soft substrate models are being developed to grow biofilms where biofilms are grown on hydrogel substrates under no, or low, shear to simulate growth on soft tissue such as might be found in an infected lung or chronic wound environment31. In some cases, the biofilms are grown directly on a hydrogel infused with nutrients31,32 or alternatively on a filter placed on the hydrogel as a “colony biofilm”33,34. These biofilms are grown under air rather than being immersed in a liquid. Colonies or lawns grown on agar plates are also increasingly being used as simple biofilm models for mechanical testing either after scraping the colonies off or directly on the colony itself35,36,37. While undoubtedly these colonies have high population densities there is limited information on how well they represent a “classical“ biofilm phenotype. Recently Hoiby et al.38 presented evidence to suggest that P. aeruginosa spread as a lawn on an agar plate switched between a planktonic phenotype to a biofilm phenotype after between 5 to 7 hr of incubation, evidenced by tolerance to antibiotics and production of the exopolysaccharide Psl in the extracellular polymeric substance (EPS) matrix. Schiessl et al.39 also reported ciprofloxacin tolerance in P. aeruginosa colony biofilms and related this to metabolic gradients using microelectrodes to show strong DO and redox gradients with an oxic region at the top of the colony and anoxic conditions at ≤ 50 µm depth. Such gradients are caused by a combination of metabolic activity (consumption and production) and transport limitation into, out of and within the colony. They are well recognized as a characteristic of the biofilm phenotype and can explain phenomena such as the development of microenvironments and antibiotic tolerance through stationary phase-like dormancy to the build-up of cell signaling molecules for quorum sensing co-ordination of the population40. Although we recognize that colonies grown on an agar plate and then submerged in liquid differ markedly from conventional biofilms grown on solid surfaces under liquid, we found our model useful to demonstrate that in such colonies strong gradients develop as they do in other biofilms. And importantly that denitrification can occur even in the oxic upper layers of the biofilm simultaneously with aerobic respiration. It is unclear whether there may be pockets of cells in micro-anoxic niches within the oxic layer, which may explain/ have implications related to tolerance against antibiotics. Another unresolved aspect is if a small population of cells are denitrifying, despite having access to oxygen as a terminal electron acceptor, which may explain the occurrence of denitrifying ‘persisters/ survivors’ in the hypoxic environment that instantly switch on upon availability of oxidative nitrogen species. Nevertheless, our data reveal another aspect of metabolic versatility in P. aeruginosa biofilms that may help explain their persistence in chronic infections, and recalcitrance to antibiotic therapy.