Microbes in the Era of Circadian Medicine

Among prokaryotes, the most well-described and extensively studied circadian rhythm is that of the cyanobacteria Synechococcus elongatus, which became a model organism for studies on the circadian field (Lakin-Thomas and Brody, 2004; Bell-Pedersen et al., 2005). Here, three core clock genes, named kaiA, kaiB, and kaiC, encode for proteins essential for rhythmic regulation of almost the entire genome and for the timing of cellular events, including cell division, photosynthesis, and amino-acid uptake, ultimately leading to enhanced fitness and to a selective advantage (Kondo et al., 1994). Unlike eukaryotes, they exclusively act at the post-transcriptional level (Nakajima et al., 2005; Tomita et al., 2005) through a mechanism involving formation of multi-protein complexes, events of phosphorylation and dephosphorylation, and conformational changes in the core biochemical oscillator (Swan et al., 2018). Briefly, the central loop involves the activation of KaiA during the day through interaction with proteins sensing environmental information, such as light, called Circadian input kinases (Cik). Once activated, KaiA interacts with the ATPase KaiC, and stimulates its autophosphorylation and oscillations between different quaternary structures. Moreover, phosphorylated KaiC binds to factors involved in the activation of rhythmic gene expression, such as SasA and RpaA. The cycle ends when KaiB binds KaiC and inhibits KaiA-stimulated phosphorylation of KaiC (Figure 1B; Johnson et al., 2017).

Some reports described the presence of rhythmic processes in other photosynthetic bacteria, such as the purple bacteria Rhodospirillum rubrum and Rhodobacter sphaeroides (Van Praag et al., 2000; Min et al., 2005). Although not strictly following the definition of circadian rhythms, the identification of these rhythms strengthens the hypothesis that more prokaryotes can possess an intrinsic cell-autonomous timing mechanism. However, if a circadian clock exists in non-photosynthetic bacteria, particularly those important for human health, and how it regulates processes essential for bacteria survival and adaptation to environmental changes is still an open question (Sartor et al., 2019).

There are only few examples describing the presence in prokaryotes of endogenous oscillations that meet the classical definition of circadian rhythms (Table 1). This evidence dates back to the last century, when growth rhythms were described in proteobacteria, including Escherichia coli (Rogers and Greenbank, 1930) and Klebsiella pneumoniae cultures (Sturtevant, 1973). However, these rhythms were not temperature compensated, therefore not classically definable circadian. More recently, a circadian oscillation was reported in the growth of the soil bacteria Pseudomonas putida (Soriano et al., 2010), as well as in Bacillus subtilis (Sartor et al., 2019), a famous gram-positive bacterium, important for its various industrial and medical applications. This bacterium in particular displays a 24-h oscillation in gene expression associated with growth, differentiation and spore formation, and it has been pointed out as a potential model organism for further circadian studies in non-photosynthetic bacteria.

Examples of alleged circadian components in these organisms, identified by studying gene homology, are instead more frequent. Homologs of kai genes (kaiC in particular) have been found in various phyla within the eubacteria reign, including proteobacteria and bacteroidetes (Loza-Correa et al., 2010), like the root-associated bacteria forming plant symbionts Sinorhizobium medicae and Pseudomonas putida. Moreover, in addition to kai homologs, other proteins in bacteria that might work as clock components are peroxiredoxin, a class of proteins whose redox state is rhythmically and autonomously oscillating in cyanobacteria and archea. Peroxiredoxin proteins have been found in almost all species of bacteria, thus suggesting that their redox rhythm might be conserved and act as the timekeeper in these organisms (Edgar et al., 2012). Moreover, different types of photoreceptors have been found in many non-photosynthetic species. Between them, some phytochrome-like photoreceptors, sensing red light, are present in various plant symbionts. Furthermore, several blue light sensing photoreceptors were found in bacteria forming association with plant roots, like B. subtilis (Losi et al., 2002).

If we specifically focused on bacteria important for human health, there is still limited evidence in favor of the presence of a circadian clock in these species. Legionella pneumophila, an environmental bacterium and opportunistic human pathogen causing severe pneumonia, contains genes encoding KaiB and KaiC, but it is not known whether their presence provides advantages in terms of adaptation to the environment, replication and virulence (Loza-Correa et al., 2010). Similarly, Pseudomonas aeruginosa or E. coli contain red and blue-light photoreceptors, respectively, that might be associated to adaptation to day-night cycles (Davis et al., 1999; Gomelsky and Klug, 2002). In the gut, the commensal bacterium Enterobacter aerogenes expresses a circadian rhythm in swarming and motility, which is enhanced in the presence of melatonin (Paulose et al., 2016). To date, this last example appears to be the most clear demonstration of a 24-h rhythm in a non-cyanobacterial prokaryote. An observation suggesting the possibility that the circadian clock in the host, through the regulation of intestinal functions, gene expression and hormones like melatonin, can both modulate the microbiota as a community, as discussed below, but also through the entrainment of single bacterial clocks eventually present.

In conclusion, we still have very poor knowledge on the presence of circadian rhythms in prokaryotes other than cyanobacteria. It is hard to imagine that organisms like bacteria have not evolved a mechanism to deal with environmental changes dictated by the day/night cycle and adapt to them. Whether our lack of knowledge is due to assessment of rhythm being conducted in the wrong experimental conditions, investigations by using different condition of growth or maintenance might provide relevant advances. Moreover, the complexity of the communities in which bacteria are used to be found needs to be taken into consideration. It is tempting to speculate that significant aspects of the circadian physiology of prokaryotes still have to be unveiled and will provide us novel exciting information that might be relevant for human pathophysiology.

Among fungi, the ascomycete Neurospora crassa is the only model system in which the presence of a well-defined molecular clock has been systematically studied and documented. Along with Drosophila melanogaster, N. crassa has represented a model organism for the study of the circadian rhythms and has been instrumental to decipher the molecular mechanisms underlying its functioning (Loros, 2019). The current model of the circadian rhythm in N. crassa identifies in the White Collar Complex (WCC), a heterodimer formed by the two transcription factors White Collar 1 (WC-1) and White Collar 2 (WC-2), the basis for the Transcriptional Translational Feedback Loop (TTFL) (Larrondo and Canessa, 2019). Indeed, WCC regulates the expression of the frequency (frq) gene, which occurs by two distinct modalities. Under constant darkness, WCC binds to the clock-box (c-box) and mediates the circadian expression of frq. Alternatively, in the presence of light, which is sensed by the light-, oxygen- and voltage-sensing (LOV) domain of WC-1, multimers of WCC are formed by LOV-LOV interactions that shift the specificity to the proximal light regulatory element (pLRE), responsible for the light-dependent increase in frq expression (Larrondo and Canessa, 2019). The FREQUENCY (FRQ) protein then associates with FRQ-RNA-Helicase (FRH) and casein kinase 1 (CK1) to form the FFC complex that, in turn, phosphorylates WC-1 and WC-2 to abolish their DNA binding ability (Wang et al., 2019). At the same time, FRQ becomes progressively phosphorylated and dissociates from WCC, thus allowing WCC to recover its function and start a new cycle (Figure 1C; Larrondo and Canessa, 2019).

While an extensive knowledge has been accumulated on the circadian clock in N. crassa, studies on other fungi has remained very limited (Larrondo and Canessa, 2019). The presence of the frq gene has been demonstrated in other Neurospora species (Lewis and Feldman, 1996) and initial phylogenetic studies have revealed the presence of the frq gene in other species within the family Sordariaceae, to which N. crassa belongs (Merrow and Dunlap, 1994). Interestingly, a functional complementation was observed upon inserting the Sordaria fimicola frq gene in a N. crassa frq null strain (Merrow and Dunlap, 1994). In addition, the presence of the frq gene was shown in two species from other families within the ascomycete fungi, namely Chromocrea spinulosa and Leptospheria australensis (Lewis and Feldman, 1996; Lewis et al., 1997), with distinct abilities to complement a conditionally arrhythmic Neurospora mutant (Lewis et al., 1997). However, whether the other components wc-1 and wc-2 were similarly present was not investigated, thus limiting the possibility to conclude that a molecular clock, similar to that described in N. crassa, is likely functioning in these species.

As soon as more fungal genomes became available, a more extensive genetic analysis has been performed to include not only the frq gene, but also the other components wc-1 and wc-2 of the oscillator. One such analysis on 17 full fungal genomes has revealed that Gibberella zeae, Magnaporthe grisea and Podospora anserina contained all the genes identified in N. crassa for a functional oscillator (Lombardi and Brody, 2005). Of note, some species, including Aspergillus nidulans, did not appear to contain the frq gene, although wc homologs were present (Lombardi and Brody, 2005). This finding is of interest since the existence of an entrainable temperature-compensated rhythm of sclerotia formation has been described in Aspergillus flavus (Greene et al., 2003) and an enzyme rhythm was also shown in Aspergillus nidulans (Greene et al., 2003), suggesting that the presence of a frq gene ortholog may not be required for a circadian feedback. A subsequent analysis on 42 fungal genomes has shown that a complete FRQ-WCC system is universally seen among the Sordariaceae while WC-1 is strongly conserved throughout the fungi, including strains lacking FRQ (Dunlap and Loros, 2006). A more recent analysis on 64 fungal proteomes established the gain of FRQ within the Ascomycetes in the classes Sordariomycetes, Leotiomyecetes, and Dothideomyecetes while WC-1 and WC-2 were present not only in other classes of Ascomycetes, such as Eurotiomycetes, that includes, among others, Aspergillus species, but also in other phyla, such as Basidiomycetes and Zygomycetes (Salichos and Rokas, 2010). WC-1 and WC-2, however, were lost in the class of Saccharomycetes, implying that, among others, Candida species do not express any of the FRQ, WC-1, and WC-2 clock components (Salichos and Rokas, 2010). The time of appearance of the frq gene has been disputed in subsequent studies. Indeed, an FRQ homolog was first identified in Pyronema confluens, an ancestor of filamentous ascomycetes, thus pointing to an earlier appearance of the frq gene (Traeger et al., 2013) that was further anticipated in a subsequent analysis of 473 fungal genomes in which an FRQ-like sequence was identified in the early diverging ascomycete Saitoella complicata (Montenegro-Montero et al., 2015). According to a recent study, FRQ sequences were also identified in Basidiomycota, Zoopagomycota, and Mucoromycotina (Brody, 2019).

All in all, these results indicate that, with the notable exception of N. crassa, the knowledge on the molecular mechanisms regulating the circadian rhythms in fungi is scarce. The prototypical FRQ-WCC oscillator do not appear to be universally present in fungi, despite the presence of bona fide circadian rhythms in some strains, thus raising the possibility that mechanisms other that FRQ-WCC might participate in their regulation. As discussed in detail below, the plant pathogen Botrytis cinerea is the only fungal pathogen in which a molecular circadian clock has been identified and characterized (Hevia et al., 2015). At present, it is unclear whether and how a circadian clock is organized in common human fungal pathogens, such as A. fumigatus and C. albicans. Indeed, the former expresses wc-1 and wc-2 homologs, but not frq, while the latter does not express any of these components. Whether a clock system might exist in these organisms and its potential relevance for the acquisition of virulence is still an almost unexplored field, thus opening up novel areas of investigation for therapy (Hevia et al., 2016).

The last two decades have provided clear evidence of the presence of a daily rhythm in many immune functions (see Scheiermann et al. for an extensive recent review on clocks and the immune system; Scheiermann et al., 2018). Almost all innate immune cells possess an intrinsic core molecular clockwork that governs multiple functions, including phagocytic activity, production of cytokines, chemokines and cytotoxic factors in specific cell types (Arjona and Sarkar, 2005; Keller et al., 2009; Gibbs et al., 2012; Silver et al., 2012a; Nguyen et al., 2013). Moreover, the expression of some pattern recognition receptors (PRR) is regulated in a circadian manner. For example, different types of Toll-like receptors (Tlr), including Tlr1-5 and Tlr9, as well as that of NOD-like receptor Nod2, are rhythmically activated by RORα in intestinal epithelial cells at the transcriptional level, while the NOD-like receptor family pyrin domain containing 3 gene (Nlrp3) displays diurnal rhythm in liver and colon and is inhibited by Rev-Erbα (Mukherji et al., 2013; Wang et al., 2018). The circadian clock also controls trafficking, localization and turnover of innate immune cells, both at steady-state conditions and during infections (Scheiermann et al., 2012; Casanova-Acebes et al., 2013; Gibbs et al., 2014; Haspel et al., 2014). This results in circadian variations in the relative abundance and distribution of immune cell types between bone marrow, lymphoid organs, the bloodstream and specific sites of infection, thus clearly affecting the host response to pathogens, as discussed below.

Adaptive immune response is also regulated by the clock system mainly in terms of lymphocyte differentiation, polarization and trafficking. For example, the clock system, through Rev-Erbα and Nfil3, controls T cell polarization in T helper 17 (T_H17) cells. This is through Nfil3 rhythmically suppressing the production of Rorγt, a transcription factor essential for T_H17 differentiation (Yu et al., 2013). This regulation is particularly relevant for mucosal immunity against bacterial and fungal pathogens in the lung and the intestine, since these cells classically produce IL-17 and IL-22, two cytokines mediating important mechanisms of resistance which are essential for the initial control of microbial burden, but whose over-activation might lead to detrimental persistent inflammation. A diurnal change in the proportion of T_H17 cells might render animals more or less susceptible to infections, by influencing the balance between resistance and tolerance at specific times of the day. The circadian timing is also important in the interaction between T cells and antigen-presenting cells (APC) (Fortier et al., 2011). Importantly, the rhythmic expression of PRR in the APC, together with the circadian variation in the number of immune cells in lymphoid organs is likely to dictate a rhythm for both humoral and cellular adaptive immune response, with important consequences for improving vaccination efficacy, as recent studies in both mice and humans have shown (Long et al., 2016; Suzuki et al., 2016; Druzd et al., 2017). Here we discuss the main findings describing how the circadian clock controls host response during bacterial and fungal infections.

A circadian response to administration of lethal doses of the bacterial product lipopolysaccharide (LPS) in mice was recognized for the first time more than 50 years ago (Halberg et al., 1960). After this pioneering study, many models of bacterial infection have been tested and revealed how the host response profoundly change its nature, magnitude and efficiency depending on the time of first contact with the microorganism (Table 1). In some of these studies, highlighted below, a definition of the molecular mechanisms that connect specific clock proteins to immune response has been unveiled and provided novel important understandings that might be useful to improve potential chronotherapeutic applications.

Fungi are associated to a broad spectrum of pulmonary, cutaneous or gastrointestinal diseases in mammals, whose clinical relevance has grown because of an increasing population of immunocompromised hosts. A circadian regulation of the host response to pulmonary fungal infection has been demonstrated in mice infected with the environmental filamentous fungus Aspergillus fumigatus (Chen et al., 2018). Specifically, it was reported that the challenge with A. fumigatus in mice, at a low dose, resulted in a significantly different clearance from the lungs 1 day post-infection, depending on the time of inoculation (Chen et al., 2018). We have recently extended these results and showed that a diurnal change occurs in both level of colonization, inflammation and damage to the host in lungs of mice infected intranasally with A. fumigatus at medium-high doses, evaluated until 7 days post infection (M.M.B., personal communication). This time-of-day specific effect appeared to be independent from cell-intrinsic macrophage rhythms, as it was shown that transcripts and protein levels of Dectin-1, the receptor for β-glucan expressed in fungal cell walls, and phagocytosis of swollen conidia, were not under circadian control in macrophages (Chen et al., 2018). As discussed by the authors, other mechanisms involved in antimicrobial defense may be under circadian control, for instance the recruitment and/or activity of neutrophils, a major player in the innate defense against fungal infections. This observation is particularly interesting as recent reports support a role of the circadian rhythms in the regulation of neutrophil activity in homeostatic and pathogenic conditions. For instance, neutrophils infiltrate virtually every tissue in homeostatic conditions with dynamics that are specific for each tissue, the majority of which shows diurnal oscillations that are in anti-phase with those in the circulation (Casanova-Acebes et al., 2018). Notably, the diurnal pattern of neutrophil infiltration in the lung regulated the diurnal transcription of about one fourth of all the pulmonary genes under circadian control, which are involved in a variety of biological processes with potential pathological consequences (Casanova-Acebes et al., 2018). For instance, in a B16F1 melanoma model of metastasis, the diurnal infiltration of neutrophils in the lung could regulate the migration of circulating tumor cells such that the ability to form metastatic foci was dependent on the time of injection of tumoral cells (Casanova-Acebes et al., 2018). Further work of the same group have shown that a neutrophil-intrinsic circadian clock, by promoting diurnal compartmentalization of neutrophils, coordinates vascular protection and immune defense, including antimicrobial defense against fungal infection (Adrover et al., 2019). Indeed, they observed that mice were more resistant to intravenous infection with the opportunistic pathogenic yeast Candida albicans when the infection was performed at ZT13 compared to ZT5. Both colonization in the kidney and weight loss were reduced, while survival rate was increased, 6 days following infection at ZT13, simultaneously with an increased ratio between tissue and blood neutrophils. All in all, this information gives an initial view of the relevance that clock-regulated immune processes have in protection against fungal infections (Table 1), that might have important consequences in the management of individuals, often immunocompromised patients, undergoing fungal infections, in order to develop appropriately timed therapeutic interventions.

The best example so far known of cross-talk between circadian clocks in mammalian hosts and microbes is that with the gut microbiota. The relative composition and abundance of microbial species within the gut microflora is oscillating in mice and humans, the rhythmicity being driven mainly by feeding rhythms of the host (Thaiss et al., 2014; Voigt et al., 2014; Liang et al., 2015; Deaver et al., 2018). Conversely, multiple reports have shown how the gut microbiota impacts the host circadian clock and modulates circadian physiology locally and at distance (Thaiss et al., 2014; Leone et al., 2015; Murakami et al., 2016; Wang et al., 2017; Weger et al., 2019). Importantly, circadian clock disruption leads to loss of the rhythm of the microbiota, with reduction of bacterial species important for the maintenance of epithelial barrier integrity, ultimately leading to dysbiosis and host metabolic disorders (Thaiss et al., 2014). Notably, bacterial functional activity and production of metabolites are also under the control of the host circadian clock. Indeed, microbial metabolites might act as key mediators of the modulation of host circadian rhythms by the microbiota, with potentially important consequences for host pathophysiological status. One such example is the modulation of hepatotoxicity after drug administration recently described by Gong et al. (2018). The authors reported that the gut microbiota mediates the diurnal variation in liver toxicity produced by acetaminophen, through the rhythmic production of the microbial metabolite 1-phenyl-1,2-propanedione that, by depleting hepatic reserve of the antioxidant glutathione at ZT12, enhances the risk of acute liver injury at this time point.

An extended description of the circadian regulation of the microbiota is not the focus of this review. However, for the purposes of evaluating the relations between host and microbes as a function of circadian cycles, it is interesting to note how the symbiotic community of host and microorganisms profoundly influences each other's activities and coordinates these activities to the day–night variations in the environment. It was recently shown how a metabolically favorable host response can drive a potentially lethal enteric bacterial pathogen to become a commensal microorganism and forsake virulence (Cadwell, 2018). Intriguingly, given the tight connection between circadian clock and metabolism, it is quite conceivable the possibility that the switch between pathogenicity and commensalism might be influenced by variation, as well as malfunctions, of the clock system in the host.

The specific timing of drug delivery, or chronotherapy, is increasingly being recognized in clinical practice as the circadian clock affects the pharmacokinetics and pharmacodynamics of the drug, as well as the expression and/or activity of the target itself (Peeples, 2018). Examples of successful chronotherapy are already in place in clinical practice including, among others, the administration of the corticosteroid methylprednisolone for the treatment of arthritis and asthma, nonsteroidal anti-inflammatory drugs for rheumatoid arthritis and osteoarthritis, or cholesterol-lowering statins. A randomized trial comparing the protection obtained following influenza vaccination revealed that patients receiving the vaccine in the morning developed a greater antibody titer compared to the group vaccinated in the afternoon (Long et al., 2016). In oncology, several studies have shown how the chronomodulated administration of different combination chemotherapy significantly reduces side effects and impacts on survival (Levi et al., 2010). Notwithstanding the remarkable success of these examples and the positive data available, chronotherapy is still confined to a limited number of indications and a major conceptual advance is needed to extend its use that may potentially apply to any type of therapy.

The identification of pathways that link the circadian rhythms to infections in pre-clinical studies, followed by the development of reliable biomarkers that can confidently indicate the best timing for drug administration, might boost the development of a chronotherapeutic strategy that holds promises of major advances in our current therapeutic approach to diseases (Cederroth et al., 2019). In the context of infectious disease, a conceptual advance over the classical pharmacological drug-target paradigm is urgently required to optimize our current therapeutic approach and drive the therapy to infections according to the daily fluctuations of the host and the target, thus maximizing the safety and efficacy. The identification of circadian rhythms in microbial species, either endogenous or host-driven, driving their colonization and virulence in mammalian hosts, could substantially change the way we approach antibiotic treatments, revealing the time window of maximum susceptibility of targeted microbes in each tissue. This approach will likely bring us in an era of personalized circadian medicine, the appearance of which is creating the need to re-contextualize our thinking on the drug-target interaction. Indeed, circadian rhythms permeate all domains of life and regulate fundamental biological processes that reverberate in health and disease. It follows that the drug and the target are not static entities, but interact in an environment that changes along a circadian pattern, thus taking the classic concepts of pharmacokinetics and pharmacodynamics a step forward. A novel concept of pharmacological therapy that centers on the circadian rhythms of the host and the target can pave the way for novel therapeutic approaches that optimize the therapeutic effect while reducing the detrimental side effects.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.