Sensing: What’s Out there and What’s in Here

Bacteria are capable of sensing and responding to an astonishing variety of environmental signals: amino acids, sugars, oxygen, pH, osmolarity, temperature, light, secondary metabolites, molecular waste products (e.g., ammonia), environmental DNA, and other microbes (both conspecifics and other species)—even physical objects such as tiny, chemically inert beads (Dworkin, 1983).

Signal transduction systems in bacteria come in several basic forms that perform a wide range of behavioral and physiological roles (Ulrich and Zhulin, 2010): one-component systems (1CSs); 2CSs, which also come in hybrid forms; CT systems, a special sub-class of 2CSs; extracytoplasmic function (ECF) sigma factor proteins, the so-called ‘third pillar of bacterial ST’ (Staroń et al., 2009), as well as non-canonical systems that involve three and more components (Marijuán et al., 2010).

Autoinduction: Indirect Sensing Via Proxies

Autoinduction is the process by which an organism synthesizes a class of molecules, called autoinducers (AIs), which stimulate a change in genetic expression in the organism itself when the molecules reach a threshold concentration (Miller and Bassler, 2001). The change may result in the production of other molecules that perform functions—classically, bioluminescence—or initiate a complex regulatory cascade that leads to a global transformation of the cell, as in sporulation. The term quorum sensing (QS) was introduced when it became apparent that genetic changes were induced at concentrations that seemed largely dependent on population density (Fuqua et al., 1994), although see the section on “Communication and Sociality” below. The supposition is that QS systems ensure that individuals do not act in ways that are too costly and unproductive when undertaken by one or a few cells. A few cells releasing a virulence factor will be swiftly attacked by a host immune system; bioluminescent molecules secreted by a lone symbiont will not be visible.

Autoinduction is widely considered to be a form of communication to facilitate cooperative behavior and is the basis for many collective activities, including: social motility, such as swarming (Daniels et al., 2004); production of secondary metabolites, such as virulence factors, bacteriocins, toxins, bioluminescence, degradative enzymes, exopolysaccharides, and pigments (Bassler and Losick, 2006); global changes of cell state, such as the transition from exponential growth to the stationary phase (Lazazzera, 2000), initiation of chromosomal replication (Withers and Nordstrom, 1998), and induction and regulation of sporulation (Shank and Kolter, 2011); lateral gene transfer (Lanka and Pansegrau, 1999); symbiotic mutualism (Visick and McFall-Ngai, 2000); pathogenic infection (Von Bodman et al., 2003; Asad and Opal, 2008); and biofilm formation, maturation and dispersal (Kjelleberg and Molin, 2002; Oppenheimer-Shaanan et al., 2013).

In different species AI systems may have widely differing regulatory components and molecular mechanisms. Commonly, however, AI concentrations are detected via ligand binding at membrane-bound receptors (2CSs) or via intracellular receptors that detect membrane-diffusible peptides (1CSs), and detection at threshold stimulates not only changes in the cell’s genetic transcription but also up-regulates AI production (Rutherford and Bassler, 2012). The signal is thus amplified.

Four QS systems have been identified as ‘canonical,’ two each for Gram-positive and Gram-negative bacteria (Rutherford and Bassler, 2012), although there are others. Several species operate two autoinduction systems, but some operate three, such as the bioluminescent Vibrio harveyii (Mehta et al., 2009) and the opportunistic pathogen P. aeruginosa (Jimenez et al., 2012). Waters and Bassler (2005) identify three types of “network architectures” for the operation of multiple QS systems: parallel, serial and antagonistic. Parallel systems may serve to filter out noise, either from non-signaling molecules in the environment or from signal mimics, operating much like multiple signatories on a bank account (Taga and Bassler, 2003). By contrast, the QS systems in P. aeruginosa are initiated sequentially, enabling the expression of different virulence factors at different times. In B. subtilis the QS network is based on two peptides that operate antagonistically, stimulating one of two mutually exclusive developmental programs, competence or sporulation, depending on context (Waters and Bassler, 2005).

The reason autoinduction is important from a cognitive standpoint is twofold. First, it increases the level of complexity of signal integration within the cell, and suggests there is a hierarchy of signaling values based on the concentrations of these molecules. Second, it involves the use of proxies for conditions that cannot be sensed directly, for example, population density or the physical properties of the surroundings. AIs thus may provide a paradigm case of biological ‘information’ that is conventional—deployed with a degree of arbitrariness, as in language—or, in the case of biological systems, evolved via natural selection to have the meaning it does (i.e., the genetic code; Smith, 2000).

In actual fact, all biological ST systems have this quality of assigned or evolved meaning, even those that directly sense features of the environment, once the signal is transduced into the cell. Protein interactions become what Millikan (2004) calls “pushmi-pullyu representations,” which simultaneously transmit what is the case and what to do about it. In the evolution of communication systems, this is the beginning of “primitive content” (Harms, 2004). This is not the only view of what constitutes communication and how it evolved, particular in microbiology (see Keller and Surette, 2006, for an alternative view), but it has a robust foundation in the study of language, comparative psychology, cognitive ethology, and philosophy of mind (Oller and Griebel, 2004).

Communication and Sociality

The domain of communication—notably, the ascription of ‘cooperation’ or ‘sociality’ to collective bacterial activity—is where the interpretation of bacterial behavior begins to resemble debates in comparative psychology over whether certain animal capacities are ‘genuinely cognitive’ or not, where the benchmark is the human case (Shettleworth, 1993). Social behavior brings into play capacities traditionally conceived of not only as cognitive but also distinctly human: intelligence, the ability to communicate meaningfully with conspecifics in order to coordinate behavior, to distinguish between self and others, to interpret and track others’ behavior over time, and culture (De Waal and Tyack, 2003). Aristotle identified ‘calculative reason’ as the defining property of the human psyche precisely because humans live complex social lives (Aristotle, 2001). Sociality requires cooperation and coordination, not merely acting together. Not surprisingly, the evolution of sociality has been a hotly disputed area of theoretical discourse in biology for several decades (Segerstrale, 2001), and remains so today. (See, for example, Wilson and Wilson, 2007 and West et al., 2007 for two diametrically opposed reviews of the same subject.)

The contemporary case against the unwarranted ascription of sociality to microbes was provided by Redfield (2002) in a much-cited article raising important questions, which to that time had not been asked, about the nature and function of autoinduction. Research was proliferating into QS as a form of bacterial communication. The notion of bacteria as social, even multicellular organisms was gaining currency (Shapiro and Dworkin, 1997; Crespi, 2001), and increasing research into biofilms suggested that microbial social life might be quite cosmopolitan, involving mixed populations and featuring divisions of labor (Watnick and Kolter, 2000). Few questioned the social hypothesis of QS, or attempted to bring an evolutionary understanding to the discussion. (For an important exception, see Brown and Johnstone, 2001.)

Arguing almost entirely in evolutionary terms, Redfield (2002) warned against the uncritical interpretation of autoinduction as a form of bacterial communication facilitating social behavior. According to Redfield’s (2002, p. 365) somewhat narrow rendering of evolutionary theory, QS genes could only evolve in situations where individual cells that invest energetic resources for a shared benefit “reproduce better than cells using their resources selfishly.” Such individual fitness benefits, Redfield (2002) noted, had been neither queried nor tested. As an alternative Redfield (2002) advanced the ‘diffusion-sensing’ (DS) hypothesis, which proposed that individual cells synthesize AIs to ascertain the physical properties of their environment to calibrate regulatory and behavioral responses for their own benefit. Although anthropomorphism is never mentioned, the article’s last paragraph gestures obliquely in that direction:

We seem to be most prone to errors with those processes that most strongly distinguish us from bacteria—our sexuality and our sociality… perhaps because we are social animals, we find the idea that bacteria have evolved communication and cooperation very appealing (p. 369).

An influential review of bacterial communication 4 years later endorsed the evolutionary approach, and drew a terminological distinction between ‘signals,’ molecules ‘evolved for’ the purpose of transmitting information between a sender and a recipient (which according to the authors constitute communication), and mere ‘cues,’ molecules that acquire a role in a bacterial behavioral economy but are not positively selected to be signals (Keller and Surette, 2006). These articles helped to inspire theoretical and empirical work seeking alternative explanations for autoinduction beyond the default assumption that QS was communication.

It did not take long for proliferating proposals to generate a terminological quagmire. Thus the ‘semantics of QS’ (Platt and Fuqua, 2010) and ‘the confusion about diffusion’ (West et al., 2012) became subjects for detailed investigation. Table ​Table33 summarizes alternative terms for autoinduction suggested by different researchers based on the dominant factor believed to be influencing concentrations of molecules in particular cases (Platt and Fuqua, 2010). Although they approach the issue in very different ways, Platt and Fuqua (2010) and West et al. (2012) draw similar conclusions that are relevant here.

First, both agree that some forms of autoinduction enable social behavior in bacteria, and can be appropriately viewed as communicative. Both cite experiments in P. aeroginosa as the only explicit attempts to test the social-signaling hypothesis, and these findings support the hypothesis. Second, both agree that terminological expansion has gone too far, and that QS is now so entrenched that it can be used generically, with the understanding that usage carries no social implications in the absence of further investigation.

However, from a theoretical standpoint, considerable work remains to be done, mainly because evolutionary theory is far from univocal on the issue of how social behavior and communication evolve. A treatment of this topic is beyond the scope of this article. I have argued elsewhere that bacterial sociality has much to teach evolutionary biologists as they refine the overarching framework of evolutionary theory, which is still very much a work in progress (Lyon, 2007).

Lifestyle Complexity = Signaling Complexity

There is a strong positive correlation between bacterial genome size and the number of different ST proteins a microbe can synthesize (Ulrich and Zhulin, 2010), as well as between the complexity of a bacterium’s lifestyle and the number of ST pathways available to support its behavior and physiology (Galperin, 2005; Ulrich et al., 2005). Similarly, ECFs tend to be ‘under-represented and often absent’ in smaller bacterial genomes and ‘over-represented’ in bacteria with more complex lifestyles (Mascher, 2013). Moreover, there appears to be a correlation between the complexity of the signaling pathways a bacterium possesses and the complexity of its behavior, although more research is needed to see if this observation holds generally.

According to the microbial signal transduction (MiST2) database (Ulrich and Zhulin, 2010), from which all of the following figures are derived, the largest and smallest genomes in the bacterial kingdom are found among the gamma-proteobacteria (see Table ​Table22.) The genome of Buchnera aphidocola, a symbiont of aphids, is a mere 400,000 base pairs (0.4 Mbp), which express only three (3) 1CSs and no ECFs. The largest genome, weighing in at an imposing 39.10 Mbp, belongs to Vibrio parahaemolyticus, a pathogenic species that inhabits brackish saltwater and has the capacity to withstand digestion both by seafood and humans. V. parahaemolyticus, about which comparatively little is known, expresses 3,321 ST systems, 79% of which (N = 2,624) are 1CSs.

In contrast, consider the delta-proteobacterium M. xanthus, a social predator that inhabits soil, arguably one of the most complex ecosystems on the planet. Potentially the primate of the eubacteria, M. xanthus is renowned for its myriad collective behaviors, including structured, multidimensional swarming motility (Kaiser and Warrick, 2014), pack-like predation (Berleman and Kirby, 2009), and the use of chemical cues to lure faster-moving prey (Shi and Zusman, 1993), as well as a complex developmental sequence leading to fruiting body formation and sporulation. At 9.14 Mbp, the M. xanthus genome is one of the top 20 in size and expresses 687 ST systems, of which more than half (54%, N = 372) are 2CSs. To date no other species, even those with complex ways of life, appears to have such a large proportion of 2CSs systems (see Table ​Table22).

In addition, M. xanthus has twice the number of ECFs as V. parahaemolyticus, providing greater flexibility in responding to stress (Ulrich and Zhulin, 2010). Moreover, it has an unusually large number of genes (97) involved in synthesizing serine/threonine protein kinases (Goldman et al., 2006), signaling proteins that are rare in prokaryotes but which are involved in a wide variety of processes in unicellular and multicellular eukaryotes. Inhibitors for eukaryotic protein serine, threonine and tyrosine kinases have been shown to inhibit development in M. xanthus to varying degrees.

Two decades ago, philosopher of science Godfrey-Smith (1996) proposed the environmental complexity thesis to explain ‘the function of mind nature.’ Godfrey-Smith (1996) advanced the idea that the evolutionary ratchet driving cognition to more complex and powerful forms in some biological lineages is the complexity of the ecological niche they inhabit. While Godfrey-Smith did not have microbes in mind (see, in particular, Godfrey-Smith, 2002), it may be that bacteria provide the best test case of his thesis’s fundamental assertion.

Motility

Chemotaxis is the directed movement of one or more cells toward external conditions the organism(s) determine are more favorable for growth and survival, and away from conditions determined to threaten survival (Webre et al., 2003). Although often studied and described in terms of discrete ‘attractants’ and ‘repellents,’ the assessment of valence by the cell that underlies CT in the presence of both has long been known to be non-linear even in simple laboratory setups (Adler and Tso, 1974), to say nothing of ecologically valid conditions. Bacterial motility takes several forms, but flagellar CT in E. coli was the first bacterial system of motility to be characterized in detail. Genes for the highly conserved methyl-accepting proteins of the Che system are genomic signposts for identifying CT systems in other bacteria (Ulrich and Zhulin, 2010).

Nevertheless, a stream of recent research into swarming motility in M. xanthus (Mignot et al., 2005, 2007; Nudleman et al., 2005; Mauriello et al., 2009b; Nan et al., 2010; Kaiser and Warrick, 2011, 2014; Pathak et al., 2012) has thrown light on very different processes of motility that involve vastly more components and produce much more observable behavior than flagellar rotation mediated by Che proteins.

Myxococcal swarming displays compelling similarities to flocking in birds, shoaling in fish and swarming in insects, the mechanisms of which have yet to be resolved in these animals (Wu et al., 2009). Other features appear to have mechanistic and functional similarities to critical signaling pathways involved in development in animals, notably, Wnt and Hedgehog (Kaiser and Warrick, 2014). Yet another feature—‘resetting’ a cytoplasmic pacemaker that regulates periodic reversal in cell direction (Kaiser and Warrick, 2014)—appears (to this writer) to bear a functional resemblance to synaptic scaling in neurons, a process critical to memory and learning in animals with nervous systems (Turrigiano, 2008).

Swarming motility in M. xanthus has long presented challenges to researchers. Swarming uses two types of motility system, initially believed to be for different styles of life, (S)ocial or (A)dventurous (Spormann, 1999). S-motility, the mechanism of which has been known for decades, involves type IV pili at the leading pole of the cell that retract, pulling the cell forward. A-motility, on the other hand, has defied adequate characterization, despite identification of about 40 genes involved in its implementation (Nan et al., 2010). However, recent research by Kaiser and Warrick (2014) using fluorescent protein labeling and time-lapse microscopy has illuminated the role of A-motility in swarming, and helped to make sense of important work over the previous decade by several research groups.

This is what was known about swarming motility in M. xanthus before the time-lapse work. First, it is collective, involving large numbers of cells, and relatively structured. Propelled by type IV pili, individual cells tend to align with one another. A large-scale manifestation of this is rippling, believed to be a strategy for distributing nutrients from lysed prey (Berleman et al., 2008). Although the proteins involved in myxococcal swarming are homologues of the classic CT proteins, the motility is not chemotactic (Dworkin and Eide, 1983; Kaiser and Warrick, 2011), which shows how similar proteins can function quite differently in similar contexts.

Second, cells in a swarm are constantly moving, periodically reversing direction in a cycle optimized to roughly 8 min, which appears to decrease the probability of collision as in shoals of fish or flocks of birds (Wu et al., 2009). According to a computational model developed to generate predictions about periodic reversal, an 8-min reversal interval results in cell orientations correlating over 10 cell lengths (≈50 microns), whereas a 50-min reversal interval results in correlation over only three cell lengths, and only one cell length with no reversal. Experimental observation aligns so closely with the model’s predictions that the probability of positive natural selection is high (Wu et al., 2009).

Cell reversal appears to be somehow obligate to myxococcal life, not simply to motility. In the hundreds of motility mutants created since 1970, none to date has been unidirectional (Kaiser and Warrick, 2011). This may be because the continuous movement of cells in a swarm is necessary to facilitate access to oxygen (above) and nutrients (below), and thus to reduce intercellular competition (Kaiser and Warrick, 2011).

Third, the gliding motility of non-motile cells can be restored through complementation (Velicer et al., 2002). Whereas in other social species, (e.g., honeybees, primates) ‘cheaters’ or ‘freeloaders’ are punished to keep them cooperating for collective benefit (Putterman, 2010), in swarming M. xanthus they appear to receive assistance. Complementation takes place through the efficient physical exchange between two cells of outer membrane lipoproteins necessary for motility, the S-motility protein Tgl and the A-motility protein CglB (Nudleman et al., 2005; Pathak et al., 2012).

Exchange is effected through the alignment and transient attachment of focal adhesion clusters (FACs), complexes of at least 15 known proteins located on the sides of the cell’s leading half—not the leading pole, as in CT (Mignot et al., 2007). When cells reverse direction the FACs retain their position on the leading half by relocating to the trailing half, along with S-motility proteins and type IV pili (Mignot et al., 2005, 2007). In short, the motility apparatus of the cell is constantly changing position during swarming.

Cell reversal, together with S-motility and FAC relocation during cell reversal, are regulated by a ‘pacemaker’ protein cluster located in the cytoplasm comprising three proteins of the Frz chemosensing pathway, the receptor for which is FrzCD (Kaiser and Warrick, 2011). The size, number and location of the Frz-clusters are constantly changing, or oscillating, and when cells make side-to-side contact, the Frz-clusters transiently align (Mauriello et al., 2009a). Frz-cluster number correlates with the frequency of cellular reversal: fewer clusters are found in hypo-reversing mutants, while more clusters are found in hyper-reversing mutants (Mauriello et al., 2009a). Wu et al. (2009) hypothesize that the frz genes “are part of a reversal clock” that operates a “reversal switch” driven by the small G protein MglAB, which induces the change in gliding direction. How the cytoplasmic pacemaker is connected to the cell surface remained unclear, although Nan et al. (2010) described a complex of proteins required for gliding that spanned the cytoplasm, inner membrane and periplasm and included the A-motility protein CglB.

Finally, swarms are different from colonies. Swarming motility depends on growth—it ceases when growth ceases—but swarms expand mainly by movement, not growth (Kaiser and Warrick, 2011). Although cells in a swarm are in all stages of the growth cycle, only 10% of swarm expansion is shown to be due to growth, whereas in a colony 90% of expansion is growth-related (Kaiser and Warrick, 2011).

The remainder of the section, which is derived entirely from Kaiser and Warrick (2014), describes what time-lapse microscopy of cells with fluorescent-labeled proteins revealed about what goes on in myxococcal swarms.

First and foremost, swarming M. xanthus don’t just move, they build two types of structures, the function of which are unclear: planar rafts, which have flat tops, and rounded mounds, which have five layers. Periodically, the cells on the top layer of the mounds translocate in an explosive, synchronized descent to the peripheries of the fourth layer. The explosive descent appears to be the result of a synchronized and simultaneous inhibition of all motility engines, A and S. From their new position the translocated cells slowly and asynchronously begin the climb to the top again, using pili-driven S-motility. When they have reached the top layer, the cells align into a compact arrangement as a result of A-motility signal proteins. The entire process takes about 14 min.

The mechanism by which this occurs is unknown, but what appears to take place is this. At low cell density, the FACs assemble on the sides of the leading half of the cells. The FACs then vanish and relocate to the trailing half, which upon cell reversal becomes the leading half. At high cell density, cells transiently attach to one another, side-by-side, as they move in the opposite direction along their long axes, bringing the FACs into alignment.

Kaiser and Warrick (2014) hypothesize that pairing facilitates passage of a signal between cells involving CglB and other A-motility proteins (as per Nudleman et al., 2005; Pathak et al., 2012). Supporting earlier observations of a motility complex, CglB proteins are found in all of the compartment spaces of the bacterium from the outer member to the periplasm and peptoglycan succulus to the cytoplasm, where the Frz-cluster pacemakers are located (Kaiser and Warrick, 2014). If the mechanism operates similarly to Wnt signaling in eukaryotes, as the researchers propose, the four levels of layering could allow Cglb proteins to pass the signal sequentially from one compartment space to another, providing a high degree of specificity between the upstream and downstream binding partners as well as with the intracellular compartment.

The signaling synchronizes the Frz-cluster pacemakers in the cell pair, which brings them into phase, effectively ‘resetting’ both cells to the average value of their pre-connection phases, a mechanism reminiscent of the scaling of synaptic firing potentials in neurons to ensure that overstimulation of one synapse does not distort the functioning of the network (Turrigiano, 2008).

Pairing is followed by cell reversal, which allows the cells to separate and move in a new direction, thus enabling the signal (whatever it is) to be spread, and the cells to become synchronized, which is necessary for coordinating the explosive descent. Thus, as Kaiser and Warrick (2011, p. 1309) observe: “Every cell in a mound should be available for pairwise signaling, cells in each layer are in contact with one another, and cells in a mound are able to move from one layer to another.” CglB is critical to this dynamic process of construction: CglB mutants can swarm but they cannot build.

In short, the sensorimotor activity of M. xanthus is astonishingly complex, and understanding this behavior may throw light on the sensorimotor behavior of social animals, such as birds, fish, and insects.

Predicting What Comes Next: Memory, Learning and Anticipation

Traditionally it was thought that, however, complex their signaling machinery, all bacteria can do is sense and respond to ensure that existential parameters stay within the range of tolerance for survival and reproduction (Freddolino and Tavazoie, 2012). A growing body of evidence in the past decade suggests that bacteria are not simply backward-looking responders to stimuli already past—an assumption long-discredited in animals (Cahill et al., 2001)—but are dynamic predictors actively oriented toward what comes next (Tagkopoulos et al., 2008; Mitchell et al., 2009; Goo et al., 2012).

In all animals and plants (Trewavas, 2014), the ability to predict is predicated on memory and learning, which can take place within the life of an organism or across generations through epigenetic inheritance (Jablonka and Raz, 2009), very likely the basis of many evolved traits. Circadian entrainment of physiological processes to the planet’s most regular large-scale change—the daily cycle of light and dark—is a well-known class of evolved eukaryotic mechanisms that displays this anticipatory property. Thus far the only prokaryotes found to possess a proper circadian clock are photosynthetic cyanobacteria (Ouyang et al., 1998; Dvornyk et al., 2003), in which the phosphorylation state of kai proteins, not genetic transcription and translation, constitute the memory of elapsed time (Nakajima et al., 2005). The hemi-methylated state of DNA just after replication is an example of a genetic substrate for memory⁴. The key question, then, is whether and how bacteria remember and learn, and to what extent these capacities are relevantly like memory and learning in more complex animals.

The author would like to thank the editors of this special topic on microbial intelligence, for the opportunity to develop these ideas; the two reviewers for comments that improved the manuscript immeasurably; the Australian Research Council (DP0880559) for funding the initial research into this topic; Jon Opie for important discussions in developing these ideas; Jeffry Stock for providing valuable feedback on an early draft that forms the basis of this article; and Richard Bradshaw, without whom none of this would have been possible.