A spotlight on the dynamic living conditions in bacterial communities with fluorescent nanosensors

A microscopy-based approach to monitor modulation of pH throughout biofilms of two important bacterial pathogens (Pseudomonas aeruginosa and Streptococcus mutans) by incorporating fluorescent nanosensors to enable detection, management and exploitation of these communities.

Like Comment
Read the paper

More bacteria live as a community associated with a surface than as individual free-living cells. These biofilm communities are difficult to remove and serve as reservoirs of infection with an inherent resistance to antimicrobials. Improvements in characterizing biofilms will pave the way to designing anti-biofilm weapons and biomarkers suitable for their detection and management1. The formation of biofilms is carefully controlled by co-ordinated gene regulation. Architectural features can be observed in the complex communities, and their components have been characterised in depth in vitro. There is evidence for microcolonies embedded within a matrix composed of polysaccharides, proteins and extracellular nucleic acid. Fluid channels run through biofilms (Figure 1), however analysis of the dynamic environmental microniches created is limited.


Figure 1. Scanning Electron Micrograph image of a S. mutans biofilm, with channels and void in biofilm microarchitecture. Scale bar = 5 µm

A chance attendance at an internal meeting to encourage new interdisciplinary projects in 2008 made Kim Hardie and Jon Aylott aware of each other’s research. They saw the potential of fusing what were then referred to as sensing pebbles with the study of bacterial communities. Together they set out to see if bacterially produced enzymes locally manipulated their surrounding environment. What they did not realise was how long this would take to reach fruition.

Preliminary steps with individual PhD students attempting to incorporate the optical nanosensors into bacterial communities alongside their other microbiological investigations moved slowly towards suitable biofilm models. The breakthrough came when a team of three excellent PhD students (Birte Hollmann, Mark Perkins and Dean Walsh) started. The research focus was quite different this time. Rather than starting with the biofilm cultures, the students became proficient in ratiometric fluorescent nanosensor preparation under the tutorage of Veeren Chauhan.

With a plentiful supply of the pH-sensitive fluorescent nanosensor, the team set out to detect pH in biofilms. Being funded by industry and BBSRC studentships (one University doctoral training partnership (DTP) and one iCase), applied projects were undertaken.

The Gram negative opportunistic pathogen Pseudomonas aeruginosa was chosen. A scourge of people suffering the from Cystic Fibrosis or wounds (that can be caused by burn or trauma). P. aeruginosa is an avid biofilm former and lurks within domestic appliances, where it is a source of malodour as well as a potential reservoir of infections2. The second bacterium selected was the Gram positive oral pathogen, Streptococcus mutans, which plays a crucial role in diseases of the mouth including dental caries and periodontitis. Once established on the tooth’s enamel as a biofilm, S. mutans ferments available carbohydrates (plentiful in a sugar-laden diet), resulting in organic acid production and the formation of dental caries3.

The fluorescent nanosensors applied to these model biofilms are biofriendly tools. The nanosensors can be used to measure key molecules and ions in microenvironments. They are ideal for real-time measurements of dynamic processes because they 1) combine analyte responsive and reference fluorophores, 2) have a small biocompatible matrix (~50 nm diameter), and 3) deliver accurate ratiometric analyte quantification as well as high spatial and temporal measurements4.

The first hurdle was to check that the nanosensors penetrated through the biofilm matrix, distributed throughout and did not significantly affect growth. Tick (Figure 2). The second was to demonstrate that they dynamically reported on pH changes as solutions of known acidity were applied. Tick (Figure 3). Thirdly, the pH could be reported in a reversible manner over time by first decreasing pH with buffers, and then increasing it. Tick (Figure 4)!


Figure 2. Nanosensors penetrate both static and flow biofilm models of P. aeruginosa. Representative images show cells stained with CellMaskTM (magenta), YOYOTM-1 bound to eDNA (cyan) and TAMRA fluorescence of nanoparticles (yellow). For the purpose of these images, YOYOTM-1 has been false coloured to cyan and TAMRA has been false coloured to yellow to facilitate visualisation of nanoparticle interaction. The top row is a 3D view from the top of the biofilm, the middle row shows a 3D view from the front of the biofilm and the bottom row represents an angled 3D view from the top. (a) P. aeruginosa was inoculated into M9 medium containing 1 mg mL-1 TAMRA nanoparticles. The cells were incubated at 37°C under static conditions for 48 h. Staining with YOYOTM-1 at 0.1 μM was performed right before imaging. (b) P. aeruginosa was inoculated into a 48-well BioFlux flow-cell plate M9 medium containing 5 mg mL-1 TAMRA nanoparticles and 0.1 μM YOYOTM-1. The cells were incubated at 37°C under flow conditions (0.2 dyn cm-2) for 24 h.

 Figure 3. Nanosensors detect pH changes in response to the exogenous addition of acid to P. aeruginosa biofilms. P. aeruginosa was inoculated into M9 glucose containing 1 mg mL-1 nanosensors. The cells were incubated at 37°C under static conditions for 48 h. (a) Cells were stained with CellMaskTM (magenta) and nanosensor channels are presented for OG-FAM (green, pH-sensitive) and TAMRA (red, reference). Z-stack images were taken with a Zeiss CLSM using a 40x objective before the treatment (left) and 30 min after the treatment with 3% (v/v) acetic acid (right). Top images represent 3D top view, bottom images 3D side view of biofilms. (b) Maximum Z-projection of OG-FAM/TAMRA overlay before the treatment (left) and after the treatment (right). (c) Calibration curve of nanosensors using pH buffers. The linear part (pH 4.0 to 7.0) was taken to calculate the linear regression (red) for the pH determination of the biofilm. (d) Calculated pH of the biofilms before and after the treatment with 3% (v/v) acetic acid. Error bars represent standard error measured for different biofilm images, where n = 2, with p<0.001 = ***.

 Figure 4. Reversible pH reporting by fluorescent nanosensors in static S. mutans biofilms. Snapshot images taken every minute for 25 minutes with a Zeiss confocal scanning microscope using a 63x/1.46na objective. After an initial image was taken (time point 0 min) a pH 6 solution was added; followed by the addition of a pH 5 solution at 5 min, pH 4 solution at 10 min, pH 7 solution of 15 min and finally a pH 6 solution at 20 min. Images shown were taken at 5 min, 10 min, 15 min, 20 min and 25 min respectively. Scale bar 20 µm. Error bars are ±1 S.D. n=1x3

Next, the team were excited when the ratiometric optical nanosensors were able to detect acidification in S. mutans biofilms when fermentable sugars were present (sucrose and glucose), but not when a non-fermentable sugar was applied (xylose and xylitol). It was very exciting to use confocal microscopy to measure the more acidic pH core of P. aeruginosa microcolonies. Ecstatic might be the best description of how the team felt when a time-lapse video revealed that in a flow biofilm model of P. aeruginosa, streamers of a more acidic pH were observed downstream. This illustrated very powerfully the dynamic nature of the microniches in biofilms, and the immense potential of the optical fluorescent nanosensors to map this in fine detail and in real time. 

A practical application of the nanosensors envisioned by the team is to test oral hygiene products or sweetener alternatives for soft drink production in high throughput. In the future, the team also plan to incorporate the optical nanosensors into realistic 3-Dimensional infection models5 and track pH to monitor the relative penetration of antimicrobials into biofilms. Plus, by detecting the metabolism within the biofilm through the pH of environmental microniches, it will be possible to evaluate how effective the antimicrobials are. This has the potential to enable novel antimicrobials to be developed that have optimal biofilm infiltration and anti-biofilm activity.

To read more, see the full manuscript: https://doi.org/10.1038/s41522-021-00221-8. All authors contributed to this blog.


  1. Koo H, Allan RN, Howlin RP, Stoodley P, Hall-Stoodley L. Targeting microbial biofilms: current and prospective therapeutic strategies. Nat Rev Microbiol. 2017;15(12):740-755. doi:10.1038/nrmicro.2017.99.
  2. Jurado-Martín I, Sainz-Mejías M, McClean S. Pseudomonas aeruginosa: An Audacious Pathogen with an Adaptable Arsenal of Virulence Factors. Int J Mol Sci. 2021;22(6):3128. doi:10.3390/ijms22063128.https://www.mdpi.com/1422-0067/22/6/3128
  3. Lemos JA, Palmer SR, Zeng L, Wen ZT, Kajfasz JK, Freires IA, Abranches J, Brady LJ. The Biology of Streptococcus mutans. Microbiol Spectr. 2019; ;7(1):10.1128/microbiolspec.GPP3-0051-2018. doi: 10.1128/microbiolspec.GPP3-0051-2018.
  4. Harrison RP, Chauhan VM. Enhancing cell and gene therapy manufacture through the application of advanced fluorescent optical sensors. Biointerphases 2018;13:8. doi:10.1116/1.5013335.
  5. Jordana-Lluch E, Garcia V, Kingdon A, Singh N, Alexander C, Williams P., Hardie KR. Development of a polymicrobial model to examine interactions between commensals and pathogens on skin. Frontiers in Microbiol 2020;11:291. doi: 10.3389/fmicb.2020.00291.

Kim Hardie

Associate Professor, University of Nottingham

Associate Professor in Molecular Microbiology. Senior Editor for Journal of Medical Microbiology. I have an international background investigating bacterial protein secretion, and have been branching out more recently into the balance between metabolism and virulence, biofilms and ways to tackle the rise in antimicrobial resistance by characterizing novel targets of antimicrobials within realistic infection models.