The secret lives of single cells.

Looking back fondly on the first 15 years of Microbial Biotechnology, a trend is emerging that biotechnology is moving from studies that focus on whole-cell populations, where heterogeneity exists even during robust growth, to those with an emphasis on single cells. This instils optimism that insights will be made into myriad aspects of bacterial growth in communities.

Microbial Biotechnology is special to me as my group published the first two research papers in the new journal: ‘Pseudomonas aeruginosa PAO1 virulence factors and poplar tree response in the rhizosphere’, vol. 1 pages: 17–29, 24 August 2007 (cited 73 times to date) (Attila et al., 2008) and ‘Metabolic engineering to enhance bacterial hydrogen production’, vol. 1 pages: 30‐39, 24 August 2007 (cited 153 times to date) (Maeda et al., 2008). In the first issue of the second year, we published another manuscript, ‘Indole and 7‐hydroxyindole diminish Pseudomonas aeruginosa virulence’, vol 2 pages: 75–90, 22 December 2008 (cited 188 times to date) (Lee et al., 2009). I am proud of these three manuscripts and congratulate the editors on the successful launch of their journal. Their vision was to create a journal where high‐quality work could receive rapid review and dissemination, and they achieved their aims.

In two of these pioneering manuscripts, we used DNA microarrays to determine the transcriptome of the whole population of P. aeruginosa cells responding either to poplar tree roots (Attila et al., 2008) or to indole (Lee et al., 2009). By measuring the whole‐population transcriptome, we discovered seven novel P. aeruginosa virulence genes this organism uses with plants and discovered indole is an inter‐species signal from Escherichia coli that quenches quorum signalling of non‐indole‐synthesizing P. aeruginosa cells without affecting their growth. This led to numerous discoveries such as indole (i) prevents the resuscitation of P. aeruginosa persister cells (Zhang et al., 2019), (ii) kills bacterial and archaeal persister cells (Hu et al., 2015; Kwan et al., 2015a; Lee et al., 2016; Megaw and Gilmore, 2017; Li et al., 2019; Song et al., 2019; Manoharan et al., 2020; Sun et al., 2020; Yam et al., 2020), (iii) helps prevent infections as an interkingdom signal in the gastrointestinal tract by tightening epithelial cell junctions (Bansal et al., 2010; Shimada et al., 2013), (iv) regulates ageing in mice (Powell et al., 2020) and (v) influences brain development via the aryl‐hydrocarbon receptor (Spichak et al., 2021). Therefore, these early publications that made use of whole‐population studies in Microbial Biotechnology had a sizeable impact.

It is fascinating now that the field is moving rapidly from studying whole‐cell populations, as we did in the early Microbial Biotechnology manuscripts, to determining the transcriptome of single bacterial cells. The logical progression was from DNA microarrays of whole‐cell populations to RNA‐seq of whole‐cell populations to RNA‐seq of single cells. For the single‐cell studies, to date, there have been four main contributions of this RNA‐seq technique in bacteria and one single‐molecule fluorescence in situ hybridization (FISH) contribution. The first published method (25 May 2020, PETRI‐seq) was that of Blatmann et al. (2020) which identified 200 E. coli transcripts per exponentially growing cell as well as identified rare prophage induction in Staphylococcus aureus cells. The second published method (17 August 2020, MATQ‐seq) was that of Imdahl et al. (2020) which quantified the impact of growth on expression of 170 Salmonella enterica serovar Typhimurium genes and 102 Pseudomonas aeruginosa genes. Next, Kuchina et al. (2020) (17 December 2020, microSPLiT) were able to detect 235 transcripts/cell for E. coli and 397 transcripts/cell for B. subtilis at different growth stages. Most recently (10 March 2021), McNulty et al. (McNulty et al., 2021) sequenced 15,000 cells and detected 265 transcripts/B. subtilis cells and 149 transcripts/E. coli cell. In a different approach, Dar et al. (2021) used par‐seqFISH to spatially resolve and quantify hundreds of transcripts within a single P. aeruginosa biofilm cell (25 February 2021). Clearly, this area is moving rapidly, but it already allows one to determine which genes are expressed in a single cell as well as to identify in which specific part of the cell that the expression occurs.

The impact of discerning the response of single cells will be huge. For example, already for persister cells, single‐cell observations have led to the discovery that the mechanism by which persister cells form as well as how they resuscitate is based on their active ribosome content. Persister cells are a subpopulation of cells which arise due to stress (e.g. antibiotic, nutritive, oxidative) and weather the stress by becoming dormant (Wood and Song, 2020). Specifically, single‐cell studies were used to discover that persister cells become dormant by mothballing their protein synthesis machinery by making 100S ribosome dimers based on (p)ppGpp and cAMP signalling and through the actions of the ribosome‐inactivating proteins ribosome‐associated inhibitor A (RaiA), ribosome modulation factor (RMF) and ribosome hibernation‐promoting factor (Hpf) (Kim et al., 2018; Song and Wood, 2020; Wood and Song, 2020). Moreover, dormant persister cells resuscitate, upon the removal of the stress and the presence of nutrients, by activating the mothballed ribosomes via nutrient sensing through membrane chemotaxis proteins and sugar transport proteins, which leads to a reduction in the (p)ppGpp and cAMP signals and activation of HflX (Yamasaki et al., 2020). When they wake, the formerly dormant cells grow exponentially like wild‐type cells (Kim et al., 2018). This ribosome‐based mechanism was completely obscured by the previous whole‐cell population studies that tried to evaluate behaviour based on population lag times. Critically, since persister cells reconstitute infections, their mechanism of formation and resuscitation is important since over two million people a year die from bacterial infections currently, and this total is projected to increase to 10 million by 2050 at a cost of $100 trillion (Thappeta et al., 2020). Furthermore, although inhibiting cAMP and (p)ppGpp would likely have pleiotropic effects, these mechanistic insights suggest RaiA, RMF (conserved in gammaproteobacterial), Hpf (conserved in bacteria and all domains of life), and Hflx (conserved GTPase from bacteria to humans) are excellent targets to prevent persistence; i.e., if cells fail to mothball their ribosomes, they remain susceptible to antimicrobials.

In a similar manner, the single‐cell approach should lead to breakthroughs regarding our understanding how toxin/antitoxin (TA) systems function (Bruggeman, 2021, unpublished data) and how antibiotic resistance arises. TA systems are categorized into seven main types based on the function of the antitoxin (Wang et al., 2020), and multiple TA systems are found in almost all genomes (Yamaguchi et al., 2011). Although prevalent, their role in cell physiology is somewhat controversial, although TA systems have a clear role in phage inhibition (Pecota and Wood, 1996; Hazan and Engelberg‐Kulka, 2004; Fineran et al., 2009), plasmid stabilization (Ogura and Hiraga, 1983), mobile genetic element stabilization (Wozniak and Waldor, 2009; Soutourina, 2019), plasmid copy number control (Ni et al., 2021), and biofilm formation (Ren et al., 2004; Kim et al., 2009).

As a controversial and well‐studied example, the type II TA system MqsR/MqsA was first identified as active in E. coli biofilms (Ren et al., 2004) and shown, based on both deletion and overexpression studies, to protect E. coli from the bile acid it encounters in the gastrointestinal tract (Kwan et al., 2015b) as well as to take part in the general stress response by regulating the master regulator of the stress response, sigma factor RpoS (Wang et al., 2011). MqsR/MqsA have also has been found to have an effect in non‐E. coli systems including copper stress (Merfa et al., 2016), vesicles (Santiago et al., 2016), and biofilm formation (Lee et al., 2014) in Xylella fastidiosa as well as biofilm formation in Pseudomonas fluorescens (Wang et al., 2019), and persistence and biofilm formation in Pseudomonas putida (Sun et al., 2017).

However, two recent reports questioned the impact of MqsR/MqsA on cell physiology. The Van Melderen group claimed there was no induction of mqsRA and no phenotype upon deleting mqsRA during stress (Fraikin et al., 2019). A few months later, the Laub group invalidated the claim of no transcription response during stress by showing mqsRA was induced dramatically (181 fold) during amino acid stress and during oxidative stress (90 fold) (LeRoux et al., 2020). The Laub group also failed to find a phenotype for MqsR/MqsA during stress (LeRoux et al., 2020), although bile acid stress was not investigated, biofilms were not investigated, and their results are flawed in that they relied on the use of a TA system deletion strain that has many non‐related mutations (large chromosomal inversions) (Goormaghtigh et al., 2018). Note the use of TA system deletion strains with coding errors have led to notorious errors in the field that have led to three retractions, based on the errors we have described (Wood and Song, 2020). Critically, both of these studies that failed to find a phenotype with MqsR/MqsA used whole‐population studies and therefore probably missed MqsR toxin expression in a subpopulation of cells. MqsR is very toxic; i.e., deletion of the antitoxin gene is lethal (Baba et al., 2006), so it is likely only a few molecules of this powerful RNase enzyme are produced and are produced in a subpopulation of cells (Bruggeman, 2021, unpublished data).

Similarly, breakthroughs in understanding how antibiotic resistance arises will likely be achieved by studying single cells. Currently, it is clear antibiotic resistance arises from a series of mutations in non‐dormant bacteria that gradually change lag phase (Santi et al., 2021) and metabolism (Lopatkin et al., 2021). However, so far, these studies are limited to studying whole‐cell populations rather than single cells. Just as whole‐population studies were unable to discern the mechanism of persister formation and resuscitation (e.g. by focussing on growth lags of large populations of cells), single‐cell studies should enable the mechanism of antibiotic resistance to be determined more robustly by following changes in a single cell.

Therefore, one can be sanguine about the mechanisms that single‐cell studies will provide in the next 10 years. Compared to coarse microarray studies on whole populations of cells, such as one for E. coli biofilm cells developed on glass wool (Ren et al., 2004), which led to the discovery of the TA systems Hha/TomB (Marimon et al., 2016) and MqsR/MqsA (Brown et al., 2009; Wang et al., 2011; Wang et al., 2013), single‐cell sequencing and other single‐cell techniques (e.g. proteomics, metabolomics, phenotype mapping, microscopy) are expected to provide myriad insights into the secret lives of single cells, including how biofilms form and function, how persister cells arise in stressed clonal populations, how TA systems impact cell physiology, how antibiotic resistance occurs, and how various protection systems are invoked and interact for lytic and temperate phages.

Funding information

No funding information provided.

Conflict of interest

None declared.