Towards the Idea of Molecular Brains
In a broad sense, the wide variety of “cellular behaviours” could include the whole repertoire of cellular actions. For example, the metaphor “division of labour” is commonly used for describing the tasks of differentiated cells at the first stages of multicellularity [144]. These may therefore concern not only actions concerning the cell’s reactions to environmental fluctuations but also concerted global actions such as cell division, certain metabolic pathways or even overall cell metabolism and the maintenance of homeostasis. Cell division, for example, is a set of integrated actions ranging from replication to cytokinesis [145,146]. It involves the concertation and synchronisation of a large number of actors. Thus, through the magnifying glass of an ethologist, cell division has all the attributes of what we might call “behaviour”, in the same way as the synchronisation and integration of the stages involved in bird nesting. Taking our reasoning further, if any action that is capable of modulating itself by gauging the outcome of its action in a defined context is “a behaviour”, then the action of an allosteric enzyme can be included in the notion of behaviour. On a slightly larger scale, the steps in the cell’s information cycle that tend towards an integrated “goal” or “function”, commonly referred to as replication, transcription and translation, are themselves units of behaviour integrated into the overall behaviour of ensuring a flow of genetic information in the cell and its progeny. For example, the steps of mRNA translation by the ribosome can be observed in great detail. They represent an extremely complex choreography of gestures and can thus be compared to “a behaviour” requiring sensory-motor control, most probably provided by allosteric networks formed by ribosomal proteins and RNA [44].
At the molecular and cellular level, we can therefore distinguish the first category of behaviour grouping spatio temporal coordination of motions ranging from “simple” phenomena of sensorimotor coordination in the broad sense (molecular or macroscopic) to tropism (global, intra- or extracellular movement leading to a destination). All these phenomena involving space are perfectly regulated in time and coordinated with the aim of global and integrated action. These processes can involve decision making: where to go? depending on several possible choices. A more complex category that includes learning phenomena emerged very early in cellular evolution, which involves decision making based on history. Here, the memorisation of past events plays a key role in the present choice of the organism. Learning, therefore, relies on the ability to remember and recall past actions in order to make decisions. Remarkably, the memorisation of past events already exists in bacteria, which are capable of remembering what has occurred in previous generations [147]. The simplest form of learning is habituation. It consists of attenuating the response and eventually ceasing to respond to a repeated stimulus that has proven not to be harmful to the organism. Habituation was initially well characterised in a mollusc, the aplysia [148], and progressively observed in organisms lacking a nervous system [149]. More elaborate, associative learning makes it possible to establish a link between frequent and joint appearances of different stimuli and to associate them. This faculty is the basis of what is known as Pavlovian conditioning. Growing evidence has shown that they are not exclusive to beings with a nervous system and that complex behaviours such as associative learning and solving complex tasks are observed in unicellulars [110,150,151,152,153,154,155].
Memory is still poorly understood at the neuronal level and is thought to involve synaptic plasticity and the storage of new molecules. At the molecular level, it has been shown to rely on post-translational modification of proteins. For example, the methylation of bacterial chemoreceptors or the phosphorylation of receptors is thought to be involved in cell memory and signalling. What occurs in molecular networks is therefore crucial for understanding cellular behaviour and the emergence phenomena leading to the notion of life.
The cause of the intelligence of plants has moreover joined that of bacteria, and these questions of status in the scale of the living have given rise to heated debates [156,157,158]. In addition, Galperin, who had the audacity to measure the IQ of bacteria and to propose, on the basis of genomic studies, very distinct behavioural types, has given rise to many controversies that the suitable reductive session of modern, factual biology has found difficult to accept [104,114]. After this long period of maturation, we are today witnessing a flowering of articles that tend to demonstrate the universality of the notion of consciousness, self-perception, cognition and intelligence in organisms or colonies that they can form, whether or not they have neurons [11,26,159,160,161,162,163,164,165,166]. These recent studies are all the more interesting as they place the concept of intelligence in an evolutionary context and propose very old roots of nervous systems that could have originated in systems performing similar functions but at the cellular level [59,167,168,169,170,171,172,173,174]. Moreover, a recent study proposes to link the cognitive faculties of microorganisms in an evolutionary and ecological perspective [175]. This work would make it possible to appreciate the evolution of “intelligence” in an environmental context and to have a global vision of the factors that may have enabled its development. What we observe a century after Binet’s “psychic life of microorganisms” goes far beyond what he could have imagined. The understanding of these phenomena has moved to the molecular level, and several decades of “cell signalling” studies have provided the fundamental principles and the molecular mechanisms that govern cellular signal transmission, integration and decision making.
Cells have the ubiquitous ability to perceive, integrate multiple (external or internal) stimuli and make appropriate decisions that allow them to maintain homeostasis and choose survival strategies in fluctuating environments. Several decades of study in both structural and systems biology have led to a considerable evolution in the overall understanding of cell signalling on the one hand and in the detail of the molecular mechanisms of signal transduction on the other [41,42,43,176,177,178]. While the basic schemes of cell signalling are similar in prokaryotes and eukaryotes, they involve distinct components and signalling pathways [179,180]. In prokaryotes, the most common systems are the 1-component systems consisting of a protein that is both the receptor and the effector acting on gene expression [181]. In 2-component system signalling, the task differentiation between the receptor, a sensory kinase and a response regulator corresponds to a later evolutionary stage [182,183,184,185,186,187,188]: the signal transduction is accomplished by the transfer of a phosphoryl group from a histidine of the sensor protein kinase to an aspartate of a response regulator. Variations around the theme of “phosophorelay” and post-translational modification have subsequently evolved to provide signalling cascades in eukaryotes. Although increasing complexity of signal transduction mechanisms has accompanied major evolutionary transitions (the transition from prokaryotes to eukaryotes and the development of multicellularity in eukaryotes), cell signalling has followed complex evolutionary paths mixing convergence and divergence from common molecular components [189,190,191].
These systems combine distinct mechanisms for triggering and propagating signals: (1) allostery, (2) protein oligomerisation and partial unfolding, (3) reversible chemical or post-translational modifications [192], (4) production of a great variety of short-lived “second messengers” (small molecules and ions), (5) abrupt changes in the membrane potential, through the opening of membrane ion channels, (6) flows, of electrons, protons or photons. These different types of signals must be able to inter-convert and “understand” each other in order to establish a chain of communication between different actors in the cell signalling processes. There is, therefore, a process of “translation” between the different informational codes in signalling.
On the other hand, cells have learnt how to turn the jiggling and wiggling of atoms [193] into a harmonious choreography. It is indeed fascinating to see how the information contained in the biological macromolecules has made it possible to constantly play between disorder and order to install a harmony where the movements of the actors seem perfectly synchronised. How, at each scale of time and space, is this synchronisation, which one might be tempted to compare to a sensory-motor synchronisation, achieved? How to create orderly and concerted movements in the stochastic universe of thermal fluctuations has been one of the major challenges of evolution through the interplay of the bio-polymer sequence. One of the first autonomous movements that a protein sequence accomplishes when it leaves the ribosome is to fold. While co-translational folding is most often assisted by the ribosome [194,195], the protein sequence generally contains the information necessary for well-coordinated folding [196,197,198,199]. This allows it to make use of thermal fluctuations while also thwarting the purely stochastic aspect of Brownian motion. Once folded, each protein moves and vibrates according to its type of folding, its sequence and its partners [200,201,202]. There is then a gradation of movements from the dynamics of a single protein to the concerted and synchronised movements of a large number of partners in macromolecular complexes during replication, transcription and translation [203,204,205]. Moreover, all these processes are coupled and interdependent.
While the cytoplasm is inherently subject to the vagaries of molecular diffusion, cells have developed different ways to control and even amplify these processes [206]. However, in all three major kingdoms, cells have developed filaments specifically dedicated to controlling cell shape and cytoplasmic movement. These “cytomotor” filaments use energy from GTP or ATP to control their directional assembly and thus create forces used to control cell shape, move organelles or organise membrane systems [207,208,209]. There are essentially two protein classes of cytomotor filaments: actin filaments, tubulin filaments and intermediate filaments that make connections between them. Each of these filaments is also associated with numerous accessory proteins that give them particular motor properties. In eukaryotes, for example, actin filaments are associated with myosin to exert forces between them, allowing movements responsible for cytokinesis, modification of cell shape or transport of organelles [210,211,212,213,214]. Tubulins associated with dyneins are responsible for chromosome movement during mitosis [215,216,217,218].
Cell motility evolved independently in all three kingdoms, either from the motile properties of cytoskeletal elements or from distinct proteins [219,220]. Archaea have evolved archaella, bacteria flagella [221] and eukaryotes cilia [222,223,224]. We are beginning to understand in detail the molecular mechanisms that control cilia beating, which involve around 100 players [225,226]. Cellular signalling establishes the equivalent of sensory-motor coordination between these “locomotor organs” and the various sensors (see above) to define an appropriate behavioural response.
Both prokaryotic and eukaryotic cells possess a huge variety of molecular sensors. From simple receptors to large complexes such as stressosomes or magnetosomes to RNA riboswitches, these sensors provide cells with precise and detailed information on fluctuations in their environment or intracellular compartment. However, due to their sequences that provide an infinite diversity of tertiary interactions that tune and maintain their three-dimensional folding, biological macromolecules are inherently sensitive. A simple rise in temperature or the binding of a ligand can lead to structural or dynamic modifications that will change its structure and properties or allow a disturbance to be propagated. Evolution has exploited the sequence-encoded subtle interplay of intra- and inter-molecular interactions to generate sensors that are responsive to almost any possible physical or chemical stimulus that a cell may perceive. Many molecular mechanisms used by proteins and nucleic acids to sense different types of physical or chemical stimuli are now deciphered.
The first category of sensors is represented by ion channels, which can convert and amplify any type of signal into abrupt changes in cell electrical activity. Several families of ion channels whose opening is controlled by different stimuli, including ligands, protons or physical stimuli such as voltage, have been listed [227].
For example, voltage-gated ion channels may open or close in response to changes in the membrane potential and play a key role in electric signalling from bacteria to vertebrates [228,229,230]. K+, Na+ and Ca++ voltage-gated channels share the same four-fold symmetric architecture consisting of a voltage-sensing domain located in the periphery and a central pore-gated domain that confers the selectivity for different cations [231,232]. The voltage-sensing is performed by a well-defined and highly conserved structural domain consisting of four transmembrane helices whose assembly changes conformation according to the membrane electric field. The S4 helix contains “gating charges”, positively charged amino acids that delocalise in response to the membrane electric field [233]. Structures of voltage-gated channels at different functional states suggested that the movement of gating-charge is coupled to the opening of the pore-gated domain, which adjusts its conductance according to the voltage [234,235,236]. This canonical model, however, may display some variations as shown by recent structures and molecular simulations [237,238,239]. Interestingly, the voltage sensor domains have evolved to couple voltage-dependent conformational changes in a variety of functions and act as modular units that confer voltage-sensing to other enzymes, such as phosphoinositide phosphatase [240,241,242,243]. On the other hand, other structural motifs involving different mechanisms, such as dipole motions, have also been found to confer voltage sensitivity in proteins [244,245].
The opening of some channels, such as some members of the TRP family, can be controlled by temperature [246,247]. The control mechanisms are more complex than voltage since it is frequently coupled with other stimuli, however. Temperature sensitivity may involve either a global response of the protein or a domain specifically dedicated to temperature perception. While T° sensitivity is an inherent property of proteins and therefore can be globally sensed by the whole protein, some structural motifs have evolved to provide a specific response to temperature changes. Studies have shown that the sensitivity of certain regions to temperature may depend on their degree of disorder. For example, it has been suggested that the more dynamic regions of TRPV1 channels have a higher thermal sensitivity and facilitate the uptake of energy from its surroundings and may reciprocally transfer it to neighbouring secondary structures, such as β-sheets that have different thermal properties to induce an allosteric response [248,249]. Thus, according to Diaz-Franulic, TRPV1 channels that exhibit polymodal responses to different stimuli [250,251] could transmit an anisotropic thermal response from one domain to another [252,253]. On the other hand, specific modules can also confer temperature sensitivity to certain channels, such as the unfolding of a temperature-sensitive domain of the BacNav channel [254,255]. More complex mechanisms are observed in some channels where temperature acts by altering the coupling between the calcium-sensitive domain and the Ca++-dependent pore domain of the archaeal MthK channel [256].
Ion channels sensitive to mechanical stimuli play an important role in mechanotransduction by controlling the opening of channels in membranes along which mechanical forces are exerted both in bacteria [257,258,259,260,261,262,263] and in specialised cells in eukaryotes [264,265,266]. These channels are often associated with the perception of other stimuli such as voltage or ligands. Although the structures of mechanosensitive channels in pro- and eukaryotes are different, similar mechanisms have converged to perceive membrane deformations and couple them to pore opening to trigger permeation. Lipid bilayer tension helps trigger channel opening. Bacterial MscS (small conductance) and MscL (large conductance) mechanosensitive channel structures consist of a homo-heptamer and a large cytoplasmic domain [267,268,269]. Differences in membrane bilayer thickness associated with voltage changes induce a change in the tilt of the TM1-TM2 helices, which is then coupled to pore opening [269]. In addition, this study also shows how the acyl chains of the lipid gatekeepers contribute to stabilising the pore in its closed state. These dissociate under membrane tension. In eukaryotes, mechanosensitive piezo channels form huge structures of more than 2500 amino acids containing 38 transmembrane helices with an overall three-bladed, propeller-shaped trimeric architecture. These structures suggest that these long blades can couple membrane deformations to pore opening via a lever arm mechanism and allow selective permeation of cations [270,271,272]. On the other hand, mechanosensitive K+ TRAAK channels, members of the two-pore domain K+ (K2P) family, also show that lipids and mechanical deformations of membrane segments induced by membrane tensions participate in the control of pore opening [273].
Light-sensitive ion channels also exist in some unicellular algae that participate in phototaxis in green algae (chlorophytes) by depolarising the plasma membrane [274,275]. These channels, channelrhodopsins (737 aas), evolved from bacterio-rhodopsin consisting of seven transmembrane segments covalently linked to a retinal chromophore [276]. Light absorption induces isomerisation of the retinal, which, in turn, causes a set of conformational changes leading to the opening of a pore that allows ions to pass through [277,278,279]. In cryptophytes, a distinct group of green algae, there are also cation-conducting channel rhodopsins that specifically conduct anions [280,281] and another group of channels that are more structurally related to haloarchaeal rhodopsins and have different functional properties [282]. Light-sensitive channels have also been found in nucleocytoplasmic large viruses that infect plankton [283].
There are also channels whose opening is controlled by ligand binding, of which there are several families such as pentameric ligand-gated channel, glutamate-gated channel, KATP channels [284]. In addition, some channels whose opening is triggered by physical stimuli can also be controlled by specific membrane lipids [285]. Although these families have different architecture and evolved from different structures, they share a modular organisation of intra- and extra-membrane domains that allows the conversion of ligand binding into an electrical signal by controlling the opening of transmembrane ion channels. These mechanisms can be finely modulated by a series of allosteric modulators. Thus, while binding of “agonist” ligands to orthosteric sites triggers pore opening, binding of allosteric modulators to distinct sites can modulate the response to orthosteric ligands [286].
Pentameric ion channels (pGLIC). The pGLIC or pentameric ligand-gated channels represent a ubiquitous family found in bacteria, archaea, and eukaryotes [287,288,289,290]. They consist of a symmetric or pseudo-symmetric assembly of five subunits surrounding a central pore selective for cations or anions [291,292]. A recent phylogenetic study has shown that these channels fall into two clades, those that share a loop (cys-loop) whose structure is maintained by a disulfide bridge between highly conserved cysteines and their metazoan, unicellular eukaryotic, and prokaryotic orthologs where the cys-loop is not conserved [288]. The structures of the homo-pentameric prokaryotic proteins have been solved at high resolution and provided detailed insights about their activation mechanisms [293,294]. Although sharing the overall architecture and activation mechanism of their eukaryotic counterparts, they display simpler behaviours than their eukaryotic homologs. For example, the Gloeobacter violaceous pGLIC homolog is a proton-gated channel thought to contribute to adaptation to pH changes [295]. In metazoan, pGLIC play a central role in the central and peripheral nervous system, and their mechanisms have been studied in detail due to their pharmacological applications. In synapses, they convert a chemical signal induced by a neurotransmitter released from the pre-synaptic membrane into an ionic flux in the post-synaptic membrane. There are channels that respond to excitatory neurotransmitters (Ach, 5-HT) that control cation flow (Na+, K+, and Ca++) [296], and inhibitory ones that control anion channels (GABA, glycine, and glutamate in invertebrates) [297,298,299,300].
Decades of biochemical, mutational, and structural studies have provided insight into the molecular mechanisms of the coupling between ligand binding to the extracellular domain, and transmembrane channel opening, and the modulation of the process by allosteric modulators located at distinct sites [296,301,302,303,304,305,306,307,308]. Interestingly, the orthosteric ligand binding (e.g., Ach) involves numerous aromatic residues that form cation-π interactions with it [309,310].
Glutamate-gated channels. Glutamate-gated channels are the main players in the excitatory synaptic response [311,312,313,314,315]. Although the three AMPA, kainate and NMDA subfamilies differ in their response times, the nature of their allosteric modulators, and their ability to desensitise, they share a common architecture, consisting of homo (kainate and AMPA) or heterotetramers assembled into a dimer of dimers. Each subunit is composed of an amino-terminal domain (ATD) involved in receptor assembly and localisation, a ligand-binding domain (LBD) and a transmembrane domain whose structure is similar to K+ voltage-gated channel. The structure of the whole AMPA receptor confirmed that, contrary to cys-loop receptors, ligand binding occurs on each subunit (clamshell) and not between the subunits. Ligand binding there induces closure of the LBD clamshell [316] and movement involving the M3-S2-M4 region that causes a conformational change leading to pore opening [317,318]. Interestingly, as in the case of cys-loop receptors, inter-subunit cation-π interactions could participate in the allosteric mechanism.
-KATP channels. ATP-sensitive K channels (KATP) are the molecular sensors of cellular metabolism. They control the opening of the channel according to the intracellular concentration of ATP [319,320]. They convert the metabolic state of the cells (ATP/ADP ratios) into an electrical signal. These channels form hetero-octamers consisting of four Kir6.1 or Kr6.2 domains structurally close to the inward rectifier Kir channels family and four sulphonylurea receptor domains (SUR1 or SUR2A/B), members of the ABC transporter family. Three natural ligands control its activity: PIP2, ATP and ADP. While PIP2 is required for its activity (opening of the channel, ATP exerts an inhibitory action by binding to the Kir subunit, which induces channel closure). On the contrary, the binding of ADP on the SUR domains potentiates its activity by promoting opening. Recent structures of whole KATP channels have identified ligand-binding sites and elucidated molecular mechanisms that establish structural interconnection between Kir and SUR domains that may explain the intricate mechanisms of ligand actions on pore opening control [321,322].
In addition to gated channels, a great variety of membrane receptors sense all kinds of physical and chemical stimuli. They relay and convert them in various cellular responses such as protein phosphorylation, synthesis of second messenger and the interconversion of GDP into GTP. They participate in the complex and intricate networks of cell signalling that has been largely documented during the past decades.
G protein coupled receptors are the largest and most ubiquitous cell surface receptors in eukaryotes that transmit signals induced by a broad spectrum of ligands, from photon to large protein molecules, via heterotrimeric GTP binding proteins (G-proteins) or β-arrestins [176,323]. Members of this family are involved in diverse signalling pathways and include light-receptors such as rhodopsin, receptors for ions, neurotransmitters, hormones, growth factors, chemokines, as well as sensory receptors for various odorants [323]. Approximately 2% of the human genome encode members of the GPCRs family that regulate key physiological functions such as sensory perception, neurotransmission, immune responses, developmental processes, regulation of endocrine and exocrine gland functions, glucose and lipid metabolisms, blood pressure. While they were initially considered as cellular membrane receptors, numerous studies have shown that GPCRs also signal at various intracellular locations using a wide variety of signalling modes and mechanisms [19]. All GPCRs are integral membrane proteins that share a structurally conserved domain composed of seven transmembrane α-helices that cross the membrane, thus forming three intracellular and three extracellular loops [323,324]. They also contain variable extracellular amino terminus and intracellular carboxy terminus tails. GPCRs are now thought to have a common ancestor with sodium-translocating microbial rhodopsin [325,326,327]. They were already present in the unicellular eukaryote ancestors and have been grouped in several classes (A-F) according to their sequence similarities and regulation and ligand-binding modes [328,329,330,331] (https://GPCRdb.org. 18 October 2021). Class A (rhodospsin) is the larger group and includes the well-characterised rhodopsin and adrenergic and olfactory receptors.
Three subtypes differ by their ligand-binding modes. In subtype 1, the ligand-binding site is deep within the transmembrane domain. In subtype 2, ligands bind to aminoterminus and extracellular loops. In subtype 3, the ligand-binding involves a long extracellular domain. The class B receptors (secretin) are activated by high-molecular-weight hormones, and the class C (glutamate) ligands bind on a very long terminal domain, and activation involves obligate dimerisation. Other classes include the class D (fungal mating pheromone receptors), class E (cAMP receptors) and class F (Frizzled receptors). An intriguing finding is that GPCR-mediated neurotransmission homologs already exist in primitive nervous system and nerve-cell-free organisms [332].
Ligand-binding stabilises an active receptor conformation that couples GPCRs to heterotrimeric G-proteins (Gαβγ) and promotes the exchange of GDP for GTP in the Gα subunit. The nucleotide-binding pocket of the Gα subunit is located between the Ras-like domain and the α-helical domain [333]. GTP binding then dissociates the Gα-GTP subunit from the dimeric Gβγ subunit. Both Gα-GTP and Gβγ subunits can then independently regulate a variety of downstream effectors that, in turn, stimulate various cellular responses depending on the G protein-coupling specificity of each receptor. The GTPase activity of the Gα subunit induces the hydrolysis of GTP to GDP and the subsequent reassociation of Gα-GDP and Gβγ subunits, which makes them available again for the next cycle. Based on sequence homologies of their Gα subunits, the G-proteins are grouped in four distinct families Gαs, Gαi, Gαq and Gα12 that activate distinct cell signalling cascades. For example, while Gαs promotes the activation of adenylyl cyclases that convert ATP to the second messenger cAMP, Gαi inhibits adenylyl cyclases and decrease the cytosolic level of cAMP. Gαq activates phospholipase Cβ that produces the second messenger’s diacylglycerol (DAG) and inositol triophosphate (IP3) [17,334]. The binding of a single ligand to GPCRs can lead to the activation of several G-proteins and thus constitute the first step in signal amplification. Then, activated G-proteins trigger the production of second messengers that target a large number of ion channels, calcium-sensitive enzymes and various kinases that contribute to relay the signalling cascade into the cell. GPCR kinases (GRK) terminate the signalling cycle by the GPCR phosophorylation that promotes their binding to β-arrestins and receptor internalisation, recycling or degradation [19].
Since the first high-resolution structures of the β2 adrenergic receptors-Gs complex [335,336], the structural studies on GPCRs and GPCR-transducer complexes have provided considerable insights into their ligand-activation and allosteric mechanisms of G protein-coupled receptor activation (reviewed in: [17,334,337,338,339,340,341]). Allostery reciprocally modulates the behaviours of GPCRs and the G-proteins: ligand binding on the extracellular side of the receptor promotes the G protein binding on the GPCR intracellular side, and the G protein increases the GPCR affinity for the ligand. Common structural features have been found in the allosteric process that couples agonist binding in the GPCR-A class to G protein binding and activation. The common pathway involves TM6 movement and key conserved motifs and notably conserved aromatic acid residues that link the ligand-binding pocked to the G protein coupling region [342,343]. These studies have also contributed to the evolution of the concept of GPCR signalling from simple “on/off switches” to a more complex system and led to a change in the paradigm of GPCR activation. Indeed, most receptors exhibit a basal level of GTP exchange activity even in the absence of ligand [16,344]. On the other hand, the loose coupling observed between ligand binding and G protein or arrestin interaction has [17,345,346] indicated that the agonist does not simply stabilise the receptor in an active conformation. The receptor is in a preexisting equilibrium between inactive and active conformation, and the ligands shift the population ensemble of preexisting conformations rather than stabilising the unique activated state.
Reversible protein phosphorylation is universally involved in cell signalling processes and often follows a recurrent logic that involves the concerted action of three concerted actors: “writers”, kinases that phosphorylate specific sites in proteins, “readers”, modules that recognise and bind to phosphorylated sites, and “erasers”, protein phosphatases that remove phosphate groups from phosphorylated proteins [189]. The modularity of the organisation of these three players provides a wide range of regulatory mechanisms and has played an important role in cell-cell interactions and in the emergence of metazoans. While in bacteria, the main players are histidine kinases, in eukaryotes, the writers are mostly serine/threonine kinases (STKs) and tyrosine kinases (TKs).
Protein phosphorylation leads to two (non-exclusive) events that can produce a cellular response: it can produce a conformational change that propagates through an allosteric pathway. Another way is to induce the binding of protein modules that specifically recognise phosphorylated sites [347] and trigger a cascade of events leading to specific signalling in the cell.
In eukaryotes, two signalling systems have evolved [348]: (i) Ser/thr kinases: ser and thr represent 97% of the phosphorylated amino acids in proteins and induce conformational changes when modified. (ii) Tyrosine kinases act by inducing protein-protein interactions. By specifically recognising phosphorylated tyrosines, SH2 domains spatially and temporally control proteins that contain phosphorylated tyrosines. SH2 domains have a well-conserved folding structure consisting of a central 3-stranded antiparallel β-sheet flanked by two helices. These domains, which form modular associations with other domains in effectors or in the protein kinases themselves, specifically recognise phosphorylated tyrosines (pYs) in multiple sequence contexts, a process governed by a subtle code that brings together the structural, dynamic, thermodynamic and kinetic properties of the partners’ association. The mode of pY binding is universally conserved, whereas recognition of the surrounding sequence is more variable and is responsible for the specificity of SH2. These modules allow circumstantial regulation and integration of multiple signals [347]. The receptor phosphotyrosine kinases (RTKs) display an extremely diverse modular organisation. They consist of a variable extra-cytoplasmic receptor, a transmembrane helix and an intracellular catalytic domain [349]. The signalling process takes place in several sequential steps, usually consisting of the stimulation of the extracellular domain by a ligand that induces receptor dimerisation and kinase autophosphorylation that contributes to increasing its catalytic activity and the binding of effector proteins containing SH2 or PTB domains that activate a cascade of signalling events induced by phosphorylation of other sites on downstream targets and nucleation of signalling complexes [350,351].
This process, which links pY to the modular binding of many effectors, also allows orthogonal signalling of multiple signalling systems [189]. Receptor activation can also stimulate the formation of receptor clusters (somewhat analogous to chemotaxis receptor clusters in bacteria) whose cooperative association can produce complex signalling.
The light-induced behavioural response of microorganisms is mediated by photoreceptors [352,353,354,355,356,357,358,359,360,361,362] and mediates either non-directional photoreception that monitors variations in the light intensity or directional photoreception that is used for orientation and habitat selection. High-resolution vision constitutes the most elaborate system that enables complex behaviours such as prey detection [353,354,363]. Various systems have convergently evolved to provide accurate light-sensing such as the cellular microlenses of Synechocystis cyanobacteria used as micro-optic to sense light direction [364] or the elaborate eye-like ocelloids of certain dinoflagellates [365]. The best-characterised families of photoreceptor proteins belong to different families and folds and are associated with distinct chromophores [366]. Despite their diversity, they share a similar property, which is to convert the ultrafast local structural photoactivated changes in their pigment into long-lived global changes in the receptor protein that are transmitted to other proteins in the signalling chain [352,360]. The rhodopsin and photoactive yellow protein (PYP) use the photoisomerisation of a C=C bond in the retinal or the p-coumaric acid, respectively. Phytochromes use the bilin, an open tetrapyrrole and the LOV proteins, cryptochromes, and BLUF proteins use the flavin mononucleotide (FMN) as chromophore [367].