Cell signalling comprises the systems used by cells to detect changes in their environment and to transduce the information into appropriate adjustments enforced by regulatory proteins. Due to its central role in all life processes, the study of cell signalling is a major focus of current biomedical research. The fission yeast Schizosaccharomyces pombe (S. pombe) is a single-celled organism used as a model to simplify the study of eukaryotic cell signalling, as it shares many features of interest with human cells. In this thesis a systems biology approach was used to investigate the roles of feedback regulation to control the dynamics of pheromone signalling in S. pombe. To this end, a quantitative dynamical model was built describing the pheromone-induced activation of the master transcription factor Ste11, as well as the coupled positive and negative feedback loops that arise from Ste11 activity. To constrain the model, a collection of data sets were generated by performing absolute quantification measurements of pheromone-dependent changes in the concentration of the model species. Structural identifiability analyses were used to select the measured species, while confidence intervals of the estimated parameters were determined through profile likelihood estimation. Analysis of the resulting model revealed a role for the pheromone signalling feedback loops to aid in the discrimination of different pheromone input doses. Through their combined action, feedback control defines the concentration and time thresholds in Ste11 activity that must be satisfied for the cell to commit to a sexual development fate.

As a summary, this work attempts to explore and uncovered design principles of certain dynamics of cellular networks by combining evolution in silico with rule-based modelling approach. Biological systems exhibit complex dynamics, due to the complex interactions in the intra- and inter- cellular biochemical reaction networks. For instance, signalling networks are composed of many enzymes and scaffolding proteins which have combinatorial interactions. These complex systems often generate response dynamics that are essential for correct decision-makings in cells. Especially, these complex interactions are results of long term of evolutionary process. With such evolutionary complexity, systems biologists aim to decipher the structure and dynamics of signalling and regulatory networks underpinning cellular responses; synthetic biologists can use this insight to alter existing networks or engineer de novo ones. Both tasks will benefit from an understanding of which structural and dynamic features of networks can emerge from evolutionary processes, through which intermediary steps these arise, and whether they embody general design principles. As natural evolution at the level of network dynamics is difficult to study, in silico evolution of network models can provide important insights. However, current tools used for in silico evolution of network dynamics are limited to ad hoc computer simulations and models. In my PhD study, with collaborators I construct the BioJazz, an extendable, user-friendly tool for simulating the evolution of dynamic biochemical networks. Unlike previous tools for in silico evolution, BioJazz allows for evolution of cellular networks with theoretically unbounded complexity by combining rule-based modelling with an encoding of networks that is akin to a genome. BioJazz can be used to implement biologically realistic selective pressures, and allows exploration of the space of network architectures and dynamics that implement prescribed physiological functions. It is provided as an open-source tool to facilitate its further development and use. I use this tool to explore the possible biochemical designs for signalling networks displaying ultrasensitive and adaptive response dynamics. By running evolutionary simulations mimicking different biochemical scenarios, we find that enzyme sequestration emerges as a key biochemical mechanism for both dynamics. Detailed analysis of these evolved networks revealed that enzyme sequestration enables both ultrasensitive and adaptive response dynamics. I verified this proposition by designing a generic model of a signalling cycle, featuring two enzymes and a sequestering (scaffold) protein. This simple system is capable of displaying both ultrasensitive and adaptive response dynamics, even more interestingly modulating the system switching between two response dynamics through perturbing the scaffold protein. These results show that enzyme sequestration can be exploited by evolution so to generate diverse response dynamics in signalling networks. From evolutionary simulations towards ultrasensitivity, bistable dynamics emerged as an alternative solution. On one hand, inspired by such results I used the fitness function as an objective function combined with different constraints to design and optimise bistable signalling networks with completely new structure and mechanism. Studying designed bistable signalling network explicates how such bistable network can be experimentally implemented. On the other hand, from studying the evolved bistable networks allosteric enzymes catalysing futile cycles appear to be a new mechanism of bistability in signalling networks. Furthermore, one of the smallest bistable signalling motifs is derived. This motif is composed of one kinase protein with two distinct conformational states and one substrate subject to phosphorylation by the kinase and auto-dephosphorylation reactions. The sufficient and necessary condition on parameters, with which the signalling motif displays bistable response dynamics, is analytically defined. By expanding the systems with more kinases, unlimited multistability emerges with potentials of implementing complex logic gates and cell state transitions. Further exploring the discovered and natural signalling networks implies shared design patterns. Motivated by searching structural boundaries between monostationary and multistationary networks, I performed algorithmic searching of multistationary signalling networks intending to find the sufficient structural conditions for multistationarity in signalling networks.

11β-Hydroxysteroid Dehydrogenase Type 2 (11βHSD2) catalyses the inactivation of cortisol. In aldosterone target tissues co-expression of 11βHSD2 and mineralocorticoid receptors (MR) protects the receptor from activation by glucocorticoids. In the syndrome of Apparent Mineralocorticoid Excess, mutations in the HSD11B2 gene cause hypertension, which is thought to be driven by volume expansion secondary to sodium retention. 11βHSD2 mice are indeed hypertensive but paradoxically volume contracted, suggestive of a urine-concentrating defect. The current studies were designed to evaluate sodium and water homeostasis in 11βHSD2-/- mice. 11βHSD2-/- mice developed a severe and progressive polyuric-polydipsic phenotype. Despite basal polyuria, at <100 days 11βHSD2-/- mice had a functional concentration response when challenged with 24 hours water deprivation. At >180 days the exacerbated polyuria was associated with severe medullary injury in the null mice. Basal aquaporin 2 (AQP2) abundance was reduced in the 11βHSD2-/- mice at both <100 and >180 days. Moreover, vasopressin 2 receptor (V2R) stimulation failed to normalize the impaired response to water deprivation in >180 day null mice. Consequently, a renal origin to the polyuria was postulated. Indeed, mice in which 11βHSD2 had been selectively targeted in the brain had a normal water turnover. A key finding from these studies is that functional deletion of 11βHSD2 in the brain, specifically the nucleus of the solitary tract (NTS), resulted in an increased salt appetite. Moreover, the mice displayed a preference for 1.5% NaCl over water. Blockade of mineralocorticoid receptors (MR) significantly reduced NaCl intake. This is the first demonstration of an increased salt appetite in a model with normal renal function and in the absence of sodium depletion. These data implicate activation of MR on 11βHSD2 positive neurons in the NTS in the behavioural drive to consume sodium.

Pseudomonas aeruginosa (PA), a Gram-negative opportunistic rod with ubiquitous presence in a panoply of different environments, secretes a wide array of virulence determinants that have established it as one of the leading nosocomial pathogens. Many of these virulence factors are regulated by the quorum sensing (QS) system that responds to environmental cell density variations. PA can ultimately trigger the onset of severe acute and chronic infections, especially in immunosuppressed subjects. The QS network in PA is comprised of at least four multi-layered interconnected subsystems with hierarchical organisation. From these, three (las, rhl and pqs) play a pivotal role in the production of virulence factors (e.g., lectins and pyocyanin) with relevant participation in the development and maintenance of biofilm matrices. The QS network is divided in two major signaling pathways, the one driven by N-acylhomoserine lactone signals and the one driven by 2-alkyl-4-quinolone molecules. Two alkyl quinolones of core importance exist in PA, the 2-heptyl-3-hydroxy-4(1H)quinolone, typically recognised as the “Pseudomonas quinolone signal” (PQS) and its precursor 2-heptyl-4(1H)-quinolone (HHQ). In addition to the regulatory involvement of the las and rhl quorum sensing systems, the biosynthesis of PQS production is also positively regulated by PqsR-dependent transcription of the pqsABCDE operon (a multivirulence factor regulator also known as MvfR). For that reason, the alkyl quinolone (AQ) signalling pathway, and more specifically its major regulator PqsR, are widely seen as promising targets for novel antimicrobial approaches. Because of the growing presence of multidrug resistant PA in the clinical setting, representing both an immediate menace to immunocompromised patients and a heavy burden on hospital budgets, the development of more rapid and affordable screening strategies for detection of the pathogen are required. Thus, new screening strategies adapted to clinical samples and successful novel synthetic small PqsR antagonists can lead the way to a new Era in the battle against hyper-virulent/-resistant PA strains. Primarily focusing on the AQ system, this research project investigated three inter-related areas, namely: (a) the highthroughput screening of novel synthetic small molecule antagonists of the PqsR protein, designed for suppression of virulence-associated phenotypes, (b) the development of a luminescent PQS-based screening bioreporter to be applied in the clinical setting, and finally (c) the optimisation of a methodology combining liquid extraction surface analysis (LESA) and mass spectrometry (MS) for screening PQS-related AQs from in vivo bacterial extracts, and also intended for future screening of a variety of clinical samples (e.g., blood plasma, urine and saliva). A large number of synthetic small molecules with putative PqsR antagonism were obtained from our French partner GreenPharma, and studied for their capacity to interfere with expression of the key player of the pqs system, PqsR. By applying a rational selection strategy, based on the inhibitory effects on pqsA expression, assessment of metabolical exertion, and assay studies on the ultimate repression of key virulence-associated phenotypes (lectins LecA and LecB, pyocyanin, PQS-associated AQs) and impact on biolfilm formation, a final selection comprised of the four best antagonists was obtained. Compounds GPZ002966, GPZ004927, GPZ824390 and GPZ273902 had their cytotoxicity subsequently studied keeping in mind their applicability in pre-clinical studies. Overall, these PqsR antagonists promoted very strong inhibition of pqsA and lecA expression, strongly reduced production of pyocyanin and PQS-related AQs (HHQ, HQNO and C7-PQS itself), showed a strong degree of biofilm inhibition, with IC50 scores sitting at the nanomolar level, and no signs of metabolical arrest was reported by the test strains used. Some explanations, focusing on the functional and structural organisation/composition of these compounds are also offered based on a comparative analysis against a number of the most prolific PqsR antagonists recently developed. The bioluminescent PQS-based biosensor for the detection of PA was engineered to respond to the presence of exogenous PQS that forms a complex with the regulatory PqsR protein, ultimately stimulating the expression of a luxCDABE-fused pqsA promoter. The biosensor was subsequently inserted in a non-pathogenic E. coli recipient by means of chromosomal integration, devoid of the sdiA LuxR homolog that could potentially interfere with the recognition of PqsR. A silent reporter was observed when in E. coli, but further assessments to its genetic integrity did not reveal any single nucleotide polymorphisms (SNP). In addition, after further testings to its activity in different established PA mutants, devoid of genes that constituted the bioreporting system or to it directly associated, a fully functional bioreporter was confirmed. Finally, a few possible explanations as to what might be in the origin of a defective bioreporter in E. coli are discussed. Lastly, a new LESA-MS protocol based on the surface sampling of dried bacterial extracts, envisaging its potentialities as a rapid and cheap screening method for detection of AQs, was designed and optimised. Even though the method has been widely used in a variety of research scenarios, this is the first time LESA-MS is applied as a screening methodology in the context of bacterial extract screening. Overall, the optimisation process showed that LESA-MS is an approach with numerous potentialities and immediate advantages, where one emphasises sampling simplicity, fast delivering of results, sensitivity to AQs at the nanomolar level (especially for C7-PQS and the precursor HHQ). But simultaneously, this methodology also revealed limitations inherent to its setting up that constrain an effective screening. The most emphatic ones being the volatility of the preparations to intra-sampling variability, and to a certain degree, an unexpected insensitivity to important QS N-acyl homoserine lactones (AHLs), namely C4-HSL and 3-oxo-C12-HSL. Nevertheless, such limitations do not present themselves as an insurmountable barrier, and based on results from available studies making use of the LESA-MS a number of possibilities to work around these are also presented.

Circadian clocks are endogenous pacemakers found in many organisms including plants, generating approximately 24h rhythms. Knowledge about the plant circadian clock plays a role for crop improvement. The plant circadian clock and its downstream outputs have been studied in detail by transcriptomics, however post-transcriptional and post-translational aspects are still to be researched. In addition, it has recently been shown that a protein modification remains rhythmic when rhythmic transcription is absent. This gives evidence for the existence of two oscillators: a transcription-translation feedback loop and a non-transcriptional oscillator. The aim of this PhD is to gain knowledge about circadian changes in abundance and phosphorylation of proteins as well as protein-protein interaction using the model plant Arabidopsis thaliana. I used high-throughput proteomics and phosphoproteomics methods to identify hundreds of phosposites that change in abundance in WT plants as well as dozens of proteins that exhibit circadian changes in their abundance. I also found significant temporal changes in protein phosphorylation in the transcriptionally arrhythmic mutant CCA1-Ox, albeit with dynamics different from the WT, demonstrating that without transcriptional rhythms, protein modification can still undergo rhythmic changes to some extent. In addition, I found reproducibly that the majority of changing phosphopeptides are most abundant at dawn and this is independent of the presence of a functional transcriptional oscillator. Roles of different kinases and affected phosphoproteins are discussed. I chose one of the rhythmically phosphorylated proteins, the bifunctional enzyme F2KP, for further functional experiments. In vitro experiments demonstrate that the rhythmic phosphosite is important for the activity of the enzyme. This is discussed in the light of circadian regulation of carbon metabolism. In addition to these studies on circadian protein abundance and modification, I investigated time-of-day dependent protein-protein interaction of the clock protein GIGANTEA (GI). Using an interaction proteomics timecourse, I identified about 100 potential new interactors of GI, some of which are candidates for links between diel timing and carbon metabolism. These results will help to generate hypotheses for explaining the surprising pleiotrophy of gi mutants.

Cytosolic free calcium (Ca2+cyt) is a ubiquitous second messenger in eukaryotes. In Arabidopsis thaliana, diurnal or circadian (diel) rhythms in [Ca2+]cyt have been widely documented. There is evidence to suggest that these diel [Ca2+]cyt oscillations modulate different signalling pathways, including photoperiodic signal transduction, gating responses to endogenous and environmental stimuli and feed-back entrainment of the core circadian clock itself. However, direct evidence for the role of Ca2+ in clock entrainment or as an output from the clock is lacking, and the question of the functional role of diel [Ca2+]cyt oscillations remains open. The role of diel [Ca2+]cyt rhythms in A. thaliana and their relationship relative to the central molecular oscillator was investigated. While it was found that diel [Ca2+]cyt oscillations persist throughout the life cycle of A. thaliana, I found no indication that diel [Ca2+]cyt rhythms are involved in photoperiodic signalling. Furthermore, I demonstrated that normal diel [Ca2+]cyt oscillations persist even in the absence of a functioning core circadian clock, indicating that, contrary to the accepted view, diel [Ca2+]cyt oscillations are not directly controlled by the core circadian clock, but are more probably generated by a non-transcriptional oscillator. In silico analysis of the amino-acid sequences of the 12 core clock proteins revealed that TOC1 contains a putative EF-hand and may therefore provide a route into the molecular oscillator for diel [Ca2+]cyt signals. The TOC1 sequence was altered to eliminate the Ca2+ coordinating residues but attempts to express this protein in E. coli, N. benthamiana and Baculovirus were unsuccessful. Complementation of the A. thaliana toc1-1 mutant with transgenes containing the endogenous TOC1 promoter sequence upstream of the wild type or the altered TOC1 sequences were also unsuccessful. A series of experiments were conducted to provide empirical data for Boolean Logic models of circadian rhythmicity that would enable further characterisation of the potential link between diel [Ca2+]cyt oscillations and TOC1.

Graphene is essentially a monolayer of sp2 bonded carbon atoms, arranged in a honeycomb lattice. Graphene has in recent years attracted phenomenal interest from researchers in materials science, condensed matter physics, and electronics since its first demonstration in 2004. The importance of graphene research was epitomised by the Nobel prize for physics being awarded to pioneers of the field in 2010. The main topic of this research was the development of epitaxial graphene on silicon carbide (SiC) substrates. The substrate inferred processability of epitaxial graphene enables graphene devices to be fabricated on full wafers using standard semiconductor processing techniques. Biosensor research is a rapidly expanding field. The major driver comes from the healthcare industry but there are also applications for biosensors in the food quality appraisal and environmental monitoring industries. The key advantages of electrochemical biosensors over competing sensor technologies are the low cost of mass production, and ability to make sensors into small compact systems. Smaller, portable sensors allow for the development of point-ofcare medical devices, which can be crucial in fast diagnosis and long-term monitoring of diseases. Graphene channel resistor devices have been fabricated using electron beam lithography and a successfully developed contact metallisation scheme - using Titanium / Gold contacts. The metal-graphene contacts have been characterised using XPS and electrical current-voltage measurements. The graphene channel device has been used as the basis of an electrochemical sensor for human chorionic gonadotropin (hCG), an indicator of pregnancy - which has also been linked to increased risk of several cancers. The immunosensor developed is a promising tool for point-of-care detection of hCG, due to its excellent detection capability, simplicity of fabrication, low-cost, high sensitivity and selectivity.

Quorum sensing (QS) is used to describe the communication between bacterial cells, whereby a coordinated population response is controlled through the synthesis, accumulation and subsequent sensing of specific diffusible chemical signals called autoinducers, enabling a cluster of bacteria to regulate gene expression and behavior collectively and synchronously, and assess their own population. As a promising method of molecular communication (MC), bacterial populations can be programmed as bio-transceivers to establish information transmission using molecules. In this work, to investigate the key features for MC, a bacterial QS system is introduced, which contains two clusters of bacteria, specifically Vibrio fischeri, as the transmitter node and receiver node, and the diffusive channel. The transmitted information is represented by the concentration of autoinducers with on-off keying (OOK) modulation. In addition, to achieve better reliability and energy efficiency, different error control techniques, including forward error correction (FEC) and Automatic Repeat reQuest (ARQ) are taken into consideration. For FEC, this work presents a comparison of the performance of traditional Hamming codes, Minimum Energy Codes (MEC) and Luby Transform (LT) codes over the channel. In addition, it applied several ARQ protocols, namely Stop-N-Wait (SW-ARQ), Go-Back-N (GBN-ARQ), and Selective-Repeat (SR-ARQ) combined with error detection codes to achieve better reliability. Results show that both the FEC and ARQ techniques can enhance the channel reliability, and that ARQ can resolve the issue of out-of-sequence and duplicate packet delivery. Moreover, this work further addresses the question of optimal frame size for data communication in this channel capacity and energy constrained bacterial quorum communication system. A novel energy model which is constructed using the experimental validated synthetic logic gates has been proposed to help with the optimization process. The optimal fixed frame length is determined for a set of channel parameters by maximizing the throughput and energy efficiency matrix.

The ability of cells to react to changes in their environment is critical to their survival. Effective decision making strategies leading to the activation of specific intracellular pathways hinge on cells sensing and processing extracellular variation. We will only be able to understand and manipulate how cells make decisions if we understand the “design” of the mechanisms that enable them to make such decisions, in terms of how they function, and in terms of their limitations and architecture. In this thesis, using mathematical modelling, I revisited classical signal sensing and transduction mechanisms in light of recent developments in methodological approaches and data collection. I studied the sensing characteristics of one of the simplest of sensors, the allosteric sensor, to understand the limits and effectiveness of its “design”. Using the classical Monod-Wyman-Changeux model of allostery, I defined and evaluated six engineering-inspired characteristics as a function of the parameters and number of sensors. I found that specifying one characteristic strongly constrains others and I determined the trade-offs that follow from these constraints. I also calculated the probability distribution of the number of input molecules that maximizes information transfer and, as a consequence, the number of environmental states a given population of sensors can discriminate between. Next, I proposed a new general model of phosphorylation cycles that can account for experimental reports of ultrasensitivity occurring in regimes that are far away from Goldbeter and Koshland’s zero-order saturation, the classical ultrasensitivity-generating mechanism. The new model exhibits robust ultrasensitivity in “anti-zero-order” regimes. The degree of ultrasensitivity, its limits, and its dependence on the parameters of the system are analytically tractable. The model can, additionally, explain in an intuitive way some puzzling experimental observations. Finally, I addressed the problem of integrating different types of signals from multiple sources, and performed some preliminary exploration of how cells can “learn” to associate and dissociate correlated signals in non-evolutionary time-scales. This work provides insights into the function and robustness of signal sensing and transduction mechanisms and as such is applicable to both the study of endogenous systems and the design of synthetic ones.

Recent years have seen considerable advances in the study of biological rhythms and the underlying molecular mechanisms that drive the daily and seasonal physiology of vertebrates. Amongst teleosts the majority of work in this field has focused on the model species the zebrafish to characterise clock genes and the molecular feedback loop that underpins circadian rhythms and physiology. Daily profiles of clock gene expression in a wide variety of tissues and cell types are now relatively well described. However the zebrafish is a tropical species that does not display distinct seasonality and therefore may not be the species of choice to investigate the entrainment of circannual physiology. In contrast, Atlantic salmon is a highly seasonal teleost that displays considerable temporal organisation of most physiological processes. In salmonids photoperiod is widely known to synchronise physiology to the environmental conditions and as such photoperiod manipulation is routinely used by the salmon industry throughout the production cycle to control and manipulate spawning, smoltification and puberty. Previous studies in salmonid species have already identified a set of clock genes that are linked to these seasonal physiological processes. However, to date, the molecular mechanisms regulating daily and seasonal physiology are largely unknown despite the strong commercial relevance in the Atlantic salmon. In the Atlantic salmon, Davie et al (2009) was the first to report the photoperiod dependent circadian expression of clock genes (Clock, Bmal and Per2 and Cry2) in the brain of the Atlantic salmon. In the same investigation the expression of clock genes was reported in a wide variety of peripheral tissues, however 24h profiles of expression in peripheral tissues were not characterised. In order to examine further the role of seasonal photoperiod on the circadian expression of clock genes, the present work first aimed to characterise diel profiles of Clock, Per1 and Per 2 expression in the brain together with plasma melatonin levels in II Atlantic salmon acclimated to either long day (LD), short day (SD), 12L:12D (referred to as experiment 1 throughout) and SNP (referred to as experiment 2 throughout). Photoperiod dependent clocks were also investigated in peripheral tissues, namely in the fin and liver. Results showed circadian profiles of melatonin under all photoperiods. In experiment 1 both Clock and Per2 displayed significant circadian expression in fish exposed to LD. This is in contrast to previous results where rhythmic clock gene expression was observed under SD. In addition, clock gene expression differed in response to experimental photoperiod in the liver, and diel rhythm differed to that of the brain. No rhythmic expression was observed in the fin. Levels of plasma melatonin exhibited a circadian rhythm peaking during the nocturnal phase as expected. However the amplitude of nocturnal melatonin was significantly elevated under LD (experiment 1) and the SNP long day photoperiod and 2010 autumnal equinox samples (experiment 2). Overall results from these experiments suggested that the control of clock gene expression would be photoperiod dependent in the brain and the liver however photoperiod history is also likely to influence clock gene expression. Interestingly, the gradual seasonal changes in photoperiod under SNP did not elicit similar profiles of clock gene expression as compared to experimental seasonal photoperiods and clock gene expression differed between experimental photoperiod and SNP treatments. In experiment 2 significant seasonal differences were also observed in the amplitude of individual clock gene expression. The mechanisms underlying this and potential impact on seasonal physiology are unknown. Developmental changes such as the smoltification process or abiotic factors such as temperature or salinity should be further investigated. In mammals previous work has focused on the molecular switch for photoperiod response and regulation of thyroid hormone bioactivity via deiodinase mediated conversion of T4 to the biologically active form T3. In mammals and birds expression of key seasonal molecular markers i.e. Tsh, Eya3 and Dio2, are up-regulated hours after exposure to the first LD and III persist under chronic LD conditions. In order to confirm the involvement of these genes in the seasonal photoperiodic response in salmon, a microarray study was first carried out. Results displayed transcriptome level differences in the seasonal expression of a wide variety of target genes including Eya3 and Dio1-3 in relation to LD and SD photoperiod suggesting that these genes may have a conserved role in salmon. qPCR validations of selected genes of interest were then performed (Dio1, Dio2 and Dio3, Eya3 and Tshover diel cycles in fish exposed to LD and SD photoperiod (autumn acclimated fish). In addition an unrelated qPCR study was undertaken in salmon parr acclimated to LD, 12L12D and SD photoperiod (spring acclimated fish)(Dio2, Eya3 and Tsh. Consistent with findings obtained in other vertebrate species, circadian expression of Dio2 was observed under LD. However expression of Eya3 and Tsh appeared to be dependent on photoperiod history prior to acclimation to the experimental photoperiods as already suggested for clock gene expression in this thesis. This is potentially a consequence of direct regulation by clock genes. To our knowledge, this is the first report on the expression of key molecular components that drive vertebrate seasonal rhythms in a salmonid species. The thesis then focused on another key component of the photoneuroendocrine axis in fish, the pineal organ. In the Atlantic salmon, as in other teleosts the photoreceptive pineal organ is considered by many to be essential to the generation, synchronisation and maintenance of circadian and seasonal rhythms. This would be primarily achieved via the action of melatonin although direct evidence is still lacking in fish. In salmonids the production of pineal melatonin is regulated directly by light and levels are continually elevated under constant darkness. In non salmonid teleosts the rhythmic high at night/ low during day melatonin levels persists endogenously under constant conditions and is hypothesised to be governed by light and intra- pineal clocks. The aims of the present in vitro and in vivo trials were to determine if circadian clocks and Aanat2 expression, the rate limiting enzyme for melatonin IV production, are present in salmon, test the ability of the pineal to independently re-entrain itself to a different photoperiod and establish whether the candidate clock genes and Aanat2 expression can be sustained under un-entrained conditions. Expression of clock genes was first studied in vitro with pineal organs exposed to either 12L:12D photoperiod, reversed 12D:12L photoperiod and 24D. Clock gene expression was also determined in vivo, in fish exposed to 12L:12D. Results were then contrasted with an in vitro (12L:12D) investigation in the European seabass, a species displaying endogenous melatonin synthesis. Results revealed no rhythmic clock gene (Clock, per1 and per2) expression in isolated salmon pineals in culture under any of the culture conditions. In the seabass, Clock and Per1 did not also display circadian expression in vitro. However rhythmic expression of Cry2 and Per1 was observed in vivo in the salmon pineal. This suggested some degree of extra-pineal regulation of clocks in the Atlantic salmon. In terms of Aanat2 no rhythmic expression was observed in the Atlantic salmon under any experimental conditions while rhythmic expression of Aanat2 mRNA was observed in seabass pineals. This is consistent with the hypothesis that in salmonids AANAT2 is regulated directly at the protein level by light while in other teleosts, such as seabass, AANAT2 is also regulated by clocks at a transcriptional level.