The control of breathing in fishes – historical perspectives and the path ahead

In 1982, Dave Randall published a review in Journal of Experimental Biology (JEB) on cardiorespiratory control in fishes during exercise and hypoxia (Randall, 1982). It was proposed in this highly cited article (484 citations, Google Scholar; 312 citations, Web of Science) that fish possess O₂ content receptors that monitor arterial blood O₂ content to modulate ventilation. As was typical of Randall's hypotheses, its appearance in print sparked a flurry of activity in the field, including the development and utilization of novel methods to identify piscine O₂ chemoreceptors whose identity was speculative at that time. The studies catalysed by Randall's review, while advancing our understanding of O₂ chemoreception, have been unable to reveal a mechanism to explain how a chemoreceptor might sense O₂ content (i.e. O₂ chemically bound to haemoglobin) as opposed to physically dissolved O₂. Although there is direct evidence that neuroepithelial cells (NECs; see Glossary) situated on the tip of gill filaments exhibit the properties of O₂ chemoreceptors (Burleson et al., 2006; Jonz et al., 2004), there is but scant and indirect support for their role in the control of breathing (Jonz et al., 2015; Milsom et al., 2022a; Perry et al., 2009; Zachar and Jonz, 2012b). It appears that existing and well-established techniques used in comparative physiology are no longer adequate to address the most pressing issues pertaining to the control of breathing in fishes: issues such as the nature of peripheral O₂ sensing and the pathways regulating ventilatory adjustments in vivo. With an emphasis on water breathers, this Commentary looks back on the development of methods and techniques that were instrumental in establishing the field of piscine ventilation control (see time line in Fig. 1) while proposing new techniques that will be important to drive the field forward (depicted in Fig. 2). For a discussion of the control of ventilation in air-breathing fishes, readers are encouraged to consult the review by Bayley et al. (2019).

The earliest description of fish ventilation is generally attributed to the French anatomist Joseph Guichard Duverney, who in 1701 published a memoir on the circulation and respiration of fishes (Duverney, 1701). In this memoir, he described fish ventilation as a series of coupled mouth and opercular movements that drive water flow over the gills. Progress over the next century and a half was slow, but included the important conclusion that, during ventilation, opercular movements lag slightly behind the corresponding movements of the mouth to generate unidirectional water flow across the gills (Bert, 1870; Duvernoy, 1839; Flourens, 1829; Owen, 1866). It was not until 1879 that M'Kendrick, using an arrangement of the ‘Marey's tambour’ (see Glossary) to record delicate movements made by the gill-covers, reported the first measurements of ventilation amplitude (V_AMP; see Glossary) and frequency (f_V; see Glossary), demonstrating large interspecific variation in ventilation patterns among the 22 species that were examined (M'Kendrick, 1879). Subsequently, measurements of buccal and/or opercular displacement became the preferred methods to assess V_AMP and f_V (Hyde, 1908; Lutz, 1930). During 1958–1960, three highly cited papers authored by George Hughes and published in JEB (Hughes, 1958, 1960a,b) established the dual-pump ventilatory cycle in bony and cartilaginous fishes. In teleosts, the cycle consists of two phases. In phase 1, with the mouth open and the operculae closed, the buccal and opercular cavities expand, which draws water into the mouth. In phase 2, with the mouth closed and the operculae open, the buccal and opercular cavities contract, which allows water in the buccal cavity to flow across the gills and exit via the opercular openings (reviewed by Shelton et al., 1986). Hughes obtained these results by measuring pressure changes during ventilation within the buccal and opercular cavities of several representative teleost and elasmobranch species using manometric techniques (see Glossary; see Fig. 1). These seminal papers, while designed to provide detailed descriptions of the breathing cycles in fishes, furnished researchers studying the control of breathing with simple tools to measure V_AMP and f_V, two key indices of ventilation volume (V̇_W: see Glossary). Non-intrusive methods for V_AMP and f_V measurements then became available in the 1970s with the implementation of electrical impedance measurements (Spoor et al., 1971). Thus, most studies on breathing over the past 60 years have used either buccal/opercular cavity pressure measurements or electrical impedance measurements as surrogates of total ventilatory water flow to examine the effects of environmental conditions such as hypoxia, hypercapnia (elevated ambient P_CO2) or elevated environmental ammonia levels on ventilation (Gilmour, 2001; Perry et al., 2009; Randall and Ip, 2006).

As appreciated by the pioneers of piscine respiratory physiology, absolute measurements of V̇_W are critically important in studies of metabolism and respiratory gas transfer (Shelton and Randall, 1962). Owing to the complexity and difficulty associated with the requisite measurements, comparatively few studies have determined V̇_W in fishes, which is a significant limitation in studies of respiratory physiology. To our knowledge, the first V̇_W measurements were performed by Van Dam during his doctoral studies. In his dissertation (Van Dam, 1938), he described a method to determine V̇_W in eel (Anguilla anguilla) and rainbow trout (Oncorhynchus mykiss) using oral membranes stretched over the fish's head to separate inspired and expired water. The Van Dam technique was refined by Davis and Cameron (1971) by sewing a latex membrane to the head, which – when positioned – served as a barrier between inspired and expired water. Although painstakingly difficult and requiring the skills of a tailor, the ‘face mask’ technique (Davis and Cameron, 1971) provided reliable measurements of V̇_W and was adopted by several research groups in subsequent years (Fig. 1). An alternative approach to measure V̇_W was to use electromagnetic flow probes to record the pulsatile water flow either exiting the gill slits of elasmobranchs (Baumgarten-Schumann and Piiper, 1968) or from the opercular openings of teleosts (Lomholt and Johansen, 1979). For fish that naturally bury in the sediment (e.g. flatfish and hagfish), it is also possible to determine V̇_W in undisturbed and unrestrained animals by using a flow probe to measure the flow exiting an inverted funnel placed over the head of a fish while buried in sand (Kerstens et al., 1979; Steffensen et al., 1984).

Using the ‘face mask’ technique, a landmark study by Smith and Jones (1982) concluded that decreases in blood O₂ content contribute to hyperventilatory responses during periods of hypercapnia, leading to Randall's controversial hypothesis of an O₂ content receptor that monitors arterial blood O₂ content in fish to modulate ventilation (Randall, 1982).

In addition to unravelling the mechanisms of piscine ventilation, this period of research also focused on identifying the respiratory centres in fish. Although we now know that fish possess central rhythm generators that can be modulated by peripheral inputs leading to the generation of ventilation patterns (Milsom et al., 2022a,b), this was a topic of debate in the early 20th century (Babák, 1921; Babkin and M'Gonigle, 1931; Bethe, 1903; v Holst, 1934). Ida Henrietta Hyde was likely to have been the first physiologist to examine the role of the cranial nerves in generating the respiratory rhythm in fish, through a combination of cranial nerve sectioning and stimulation (Hyde, 1904). Subsequent studies examining respiratory patterns after cranial nerve transection (Aimer, 1961; Deganello, 1908; Powers and Clark, 1942; Springer, 1928) or stimulation (Ballintijn et al., 1983; De Graaf and Roberts, 1991; Lutz, 1930; Satchell, 1959; Smith, 1966; Springer, 1928) demonstrated that peripheral inputs modulate, but are not responsible for, generating the respiratory rhythm, which is initiated centrally (Fig. 1).

Hyde (1904) also undertook the first detailed search for the respiratory centres in fish. By lesioning or stimulating various regions of the skate brain, Hyde observed that the respiratory centre occupies discrete sensory and motor areas within the medulla (Hyde, 1904). This finding was corroborated by electrical recordings from isolated goldfish (Carassius auratus) brainstems, which display rhythmic electrical waves recurring at intervals matching the frequency of respiratory movements (Adrian and Buytendijk, 1931). Following these pioneering studies (Adrian and Buytendijk, 1931; Hyde, 1904), researchers used a combination of stimulation, transection and nerve recording techniques to demonstrate the automaticity of the central respiratory rhythm generators (Fig. 1). However, there is still no consensus on the site and arrangement of these rhythm generators (Aimer, 1961; Ballintijn et al., 1979; Duchcherer et al., 2010; Kawasaki, 1979; Martel et al., 2007; Rovainen, 1974, 1977; Satchell, 1959; Shelton, 1959, 1961; Takesi and Hiromasa, 1956; Waldron, 1972; Woldring and Dirken, 1951).

In 1982, when Randall synthesized the state of the field and postulated the presence of an O₂ content receptor involved in modulating ventilation, the existence of branchial chemoreceptors that control cardiac function was already well established (Smith and Jones, 1978), yet evidence for branchial O₂ chemoreceptors that modulate breathing was scant. Cranial nerve transection studies did not identify branchial O₂ chemoreceptors in dogfish (Squalus acanthias) or the sea raven (Hemitripterus americanus) (Satchell, 1961; Saunders and Sutterlin, 1971), even though previous studies showed that cranial nerves terminate in the orobranchial cavity and play a role in modulating normal breathing. These results may not be surprising if one considers that neither the dogfish nor the sea raven are known to exhibit a particularly robust cardiorespiratory response to hypoxia (Butler and Taylor, 1971, 1975; Saunders and Sutterlin, 1971). In addition, during hypoxia, isolated gill arch preparations from sea raven (Sutterlin and Saunders, 1969) did not exhibit changes in nerve activity for branchial divisions of cranial nerves IX and X. The relationship between peripheral chemoreceptors and ventilation at the time was elegantly summarized by Jones and Milsom (1982) in another review appearing in JEB: “Thus peripheral receptors could be viewed merely as the ‘fine-tuners’ of cardiovascular and respiratory patterns which are initiated and maintained by hormonal or humoral changes both within and without the central nervous system. Whether this supposition has any basis must be left to the future.”

Coincidentally, in the same year as Randall's review, Dunel-Erb et al. (1982) proposed that NECs of the trout gill filament function as peripheral O₂ chemoreceptors to promote cardiorespiratory reflexes associated with hypoxia, including the hypoxic ventilatory response (HVR; see Glossary). Their study used electron microscopy to obtain high-resolution images of NECs, and the authors suggested that these cells were O₂ chemoreceptors based predominantly on morphological similarities with the type II (glomus) cells of the mammalian carotid body and an apparent emptying of neurosecretory vesicles (‘degranulation’) during severe hypoxia (Dunel-Erb et al., 1982). This study was followed by immunohistochemical characterization of the NECs, showing that they contain the neurotransmitter serotonin (5-HT) and the neurosecretion marker synaptic vesicle protein (SV2), and are in close association with nerve fibres organized in a manner reminiscent of mammalian peripheral O₂ chemoreceptors (Bailly et al., 1989; Jonz and Nurse, 2003). Using species exhibiting more pronounced respiratory responses to hypoxia, subsequent studies employing gill–nerve preparations were able to identify single afferent nerve fibres from the first gill arch that were O₂ sensitive (Burleson and Milsom, 1993; Milsom and Brill, 1986), and cranial nerve sectioning revealed the presence of branchial O₂ chemoreceptors that modulate breathing (Burleson and Smatresk, 1990; Sundin et al., 2000), providing further support to the idea that NECs function as O₂ chemoreceptors.

Remarkably, it was more than 20 years after the documentation of NECs (Dunel-Erb et al., 1982) that direct evidence was provided for their function as O₂ chemoreceptors. Using whole-cell patch clamping, NECs isolated from the gill filaments of adult zebrafish, channel catfish (Ictalurus punctatus) and goldfish were shown to respond to hypoxia with a decrease in K⁺ current and membrane depolarization, similar to mammalian O₂ chemoreceptors (Burleson et al., 2006; Jonz et al., 2004; Zachar and Jonz, 2012c). Whole-cell patch clamping also revealed that NECs are CO₂ sensitive (Qin et al., 2010), and by using in vitro Ca²⁺ imaging of isolated rainbow trout NECs, ammonia was added to the list of respiratory gases sensed by NECs (Zhang et al., 2011). Because the presence of NECs on the gill filaments (and lamellae) appears to be highly conserved among teleosts, it is widely accepted (although without direct evidence) that these cells operate in vivo as polymodal chemoreceptors to initiate cardiorespiratory responses during hypoxia, hypercapnia and postprandial increases in plasma ammonia levels.

The general acceptance of NECs as polymodal chemoreceptors has sparked interest in identifying the neurotransmitter(s) secreted by NECs for signal transduction. Readers are referred to recent reviews (Pan and Perry, 2020; Reed and Jonz, 2022) for detailed information on this topic. In brief, 5-HT remains the sole neurotransmitter so far identified in NECs, but direct evidence of its secretion is lacking. Neuroendocrine factors including 5-HT, catecholamines, acetylcholine, purines and gaseous neurotransmitters may play an important role in modulating ventilatory responses. Techniques such as morpholino gene knockdown (Porteus et al., 2015), CRISPR/Cas9-based gene knockout (Pan et al., 2021b), single cell transcriptomics (Pan et al., 2022) and in vivo Ca²⁺ imaging (Pan et al., 2023) have been used in zebrafish to further probe the importance of these neurotransmitters in NECs.

In comparison to the role of peripheral chemoreceptors, that of central chemoreceptors in modulating breathing has received much less attention. The consensus is that strictly water-breathing fishes lack central chemoreceptors for O₂ (Milsom, 2010; Milsom et al., 2022a). This conclusion is based largely on studies of in vitro brainstem preparations (Côté et al., 2014; Rovainen, 1977; Wilson et al., 2000), in which fictive breathing (see Glossary) was unresponsive to changes in P_O2, with the exception of the sea lamprey (Petromyzon marinus) (Cinelli et al., 2017). Additionally, gill ventilation was unchanged in the bowfin (Amia calva) during perfusion of the cerebral ventricles with mock extradural fluids with varying P_O2 (Hedrick et al., 1991). By contrast, there is some indirect evidence for central CO₂/pH chemoreceptors in elasmobranchs; ventilation during hypercapnia exposure in the skate (Raja ocellata) correlates with brain intracellular pH (Wood et al., 1990). The sum of these data, however, suggests that strictly water-breathing fishes are likely to lack central O₂ chemoreceptors and, with a few exceptions, are also likely to lack central chemoreceptors for CO₂/pH. Although central CO₂/pH receptors were identified in air-breathing fishes (Amin-Naves et al., 2007; Sanchez et al., 2001; Wilson et al., 2000), too few species have been studied to generalize about differences in central CO₂ sensing between water and air breathers. Further progress in the field of central CO₂/pH sensing has been limited by the lack of new and efficient methods to probe brain functions in vivo.

Arguably, the most important unresolved question in the field of piscine ventilatory control is whether NECs do indeed function as in vivo chemoreceptors to modulate ventilation. Putnam et al. (2004) and Richerson et al. (2005) established two essential criteria for designating cells as central pH chemoreceptors. The required attributes are (i) intrinsic sensitivity to physiologically relevant changes in P_CO2/pH enabling appropriate cellular responses to non-pathological changes in CO₂/pH in vivo, and (ii) elicitation of appropriate downstream changes in respiratory output. Clearly, these criteria can be generalized to all chemoreceptors. In terms of O₂ chemoreception, Jonz et al. (2004) and Burleson et al. (2006) elegantly demonstrated the intrinsic capability of NECs to respond to changes in P_O2 in isolated NECs in vitro. What is lacking, however, is evidence for their O₂ sensitivity in vivo and direct evidence linking NEC activity to the modulation of ventilation. Progress has been constrained by the lack of techniques to probe NEC function in vivo.

Recent years have seen the rise of in vivo techniques used to probe the activity and physiology of small populations of cells of interest and to manipulate their activity. In vivo Ca²⁺ and neurotransmitter imaging (Mank and Griesbeck, 2008; Wang et al., 2018) utilize genetically encoded Ca²⁺ and neurotransmitter sensors, which increase in fluorescence upon target binding. Receptors such as channel rhodopsin and transient receptor potential cation channels can be inserted into cells to opto- or chemo-genetically alter the activity of target cells (Antinucci et al., 2020). In addition, proteins such as nitroreductase can be expressed within cells for targeted cell ablation based on their ability to render prodrugs, such as metronidazole, cytotoxic (Sharrock et al., 2022). A caveat with these techniques is that an endogenous gene expressed within the cell of interest must be identified to drive transgenic expression of the various constructs via techniques such as Tol2 transgenesis (Kawakami, 2007). Identifying endogenous genes uniquely expressed in NECs has proved challenging. However, with the finding of tph1a and vmat2 expression in zebrafish NECs (Pan et al., 2022, 2021a,b), coupled with the ease of genetic manipulation in this species, functional studies of NEC function in vivo are now feasible. Below, we propose experiments that we believe are required in order to establish NECs as peripheral O₂ chemoreceptors that promote the HVR. Similar methods can be applied to studies of CO₂/pH and ammonia chemoreception.

Establishing the sensitivity of NECs to O₂in vivo can be achieved by driving transgene expression of genetically encoded Ca²⁺ indicators such as GCaMPs or RCaMPs (Chen et al., 2013; Dana et al., 2016) in zebrafish NECs (Fig. 2A). Because larval zebrafish are effectively transparent, and adults can be rendered transparent in the Casper line (White et al., 2008), Ca²⁺ activity of NECs can be imaged in response to changing levels of P_O2 in live fish. Similar techniques can be used to determine which neurotransmitters are released by NECs during hypoxia, by expressing specific neurotransmitter sensors in NECs or the cranial sensory nerves that project into the gill region (Kucenas et al., 2006).

Establishing a direct link between NEC activity and ventilatory responses (e.g. the HVR) is likely to be more challenging, although certainly feasible. One approach would be to assess the HVR in fish lacking NECs by knocking out key genes required for NEC development. A potential candidate is the ascl1a gene; knockout of this gene in larval zebrafish disrupts serotoninergic cells within the pharyngeal arch (Kapsimali et al., 2011). Clearly, the extent of NEC ablation in the ascl1a knockout fish would need to be determined. A drawback of this approach is that the broad knockout of this gene in all tissues would make it difficult to attribute confidently any changes in ventilation to the ablation of NECs alone. A second approach would require the NEC-specific expression of transgenes, which could enable cellular activation, inhibition or removal. The successful implementation of such an approach first requires the identification of genes that are expressed uniquely in NECs to prevent transgene activation in other cell types (e.g. in the central nervous system, CNS) that might also be linked to ventilatory control. For example, although tph1a and vmat2 are expressed in NECs, they are also found in a population of cells within the CNS. Because these CNS cells might also be critical in the regulation of breathing, it would be difficult to attribute any effects of transgene activation specifically to changes occurring in the NECs. A workaround would be to utilize the Gal4-UAS (upstream activation sequence) system (see Fig. 2B) for spatiotemporal control of transgene expression (Elliott and Brand, 2008). With the Gal4 system, one line (the driver) expresses Gal4 in a known temporal or spatial pattern and a second line (the responder) contains a UAS-dependent transgene (Brand and Perrimon, 1993). In addition, Gal80 can antagonize Gal4 by binding to its transactivation domain and preventing Gal4 from activating transcription (Faucherre and López-Schier, 2011; Fujimoto et al., 2011; Lue et al., 1987). Thus, generating and crossing three transgenic lines: (1) Gal4 driven by the tph1a or vmat2 promoter, (2) a transgene controlled by the UAS promoter, and (3) Gal80 driven by the pan-neuronal promoter elavl3 (previously known as HuC; Park et al., 2000) would enable the expression of transgenes within NECs but not in the CNS. Thus, to link changes in NEC activity to changes in ventilation, NECs expressing channel rhodopsin (Antinucci et al., 2020) or rat TRPV1 channels (Matty et al., 2016) could be activated with light (channel rhodopsin) or capsaicin (TRPV1). Additionally, NECs expressing transient receptor potential cation channels or nitroreductase could be inhibited with light (Antinucci et al., 2020) or ablated with metronidazole (Sharrock et al., 2022), respectively. A diagrammatic summary of the various techniques that might be used to evaluate NEC function in vivo is presented in Fig. 2. It should be noted that these techniques could also be applied to non-model species, provided that they reproduce reliably in the lab with short generation times. Even if the genome of a particular species is not fully characterized, it may be possible to drive transgene expression using exogenously derived promoters from related species (Aihara et al., 2007).

Since NECs were first proposed as piscine O₂ chemoreceptors (Dunel-Erb et al., 1982), most studies on O₂ chemoreception in fish have focused on these cells. The attention that NECs received was due, in part, to their supposed homology with glomus cells, the O₂ chemoreceptors found in the mammalian carotid body (Milsom and Burleson, 2007; Zachar and Jonz, 2012b). However, we now know that unlike glomus cells, NECs are not derived from neural crest precursors but are instead derived from endoderm (Hockman et al., 2017). Thus, the NECs have the same origin as ‘pulmonary neuroendocrine cells’, the hypoxia-sensitive airway sensors in mammals. Rather than NECs, Hockman et al. (2017) proposed another cell type, neural crest-derived catecholaminergic cells associated with the pharyngeal arch vasculature, as the cells that share homology with glomus cells. It was suggested (Hockman et al., 2017) that these cells may also function as O₂ chemoreceptors.

In addition, under the current consensus that gill NECs function as O₂ chemoreceptors, there exists a mismatch between the physiological responses to hypoxia and the ontogeny of NECs. In zebrafish larvae, a hypoxic response develops as early as 2–3 days post-fertilization (dpf), well before gill NECs are fully innervated (at 7 dpf; Jonz and Nurse, 2005). It was proposed that a population of cutaneous serotonergic cells resembling gill NECs in morphology promotes the hypoxic responses prior to the full maturation of gill NECs (Coccimiglio and Jonz, 2012). This idea is supported by data demonstrating that skin NECs undergo morphological responses to changes in environmental P_O2, and that their chemical denervation can blunt the HVR in larval zebrafish (Coccimiglio and Jonz, 2012). However, a recent study showed that skin NEC development in larval mangrove rivulus (Kryptolebias marmoratus) is largely unaffected by environmental P_O2 levels (Cochrane et al., 2021), suggesting that there might be other O₂ chemoreceptors present in addition to gill or skin NECs.

Actually, the earliest proposed O₂ chemoreceptors in fish were taste bud-like receptors. De Kock (1963) noted the presence of taste bud-like organs in the mouth and on the gills of rainbow trout and Atlantic salmon (Salmo salar), and suggested that these organs might provide afferent information enabling the cardiorespiratory reflex responses to hypoxia. Subsequently, it was demonstrated that these taste bud-like receptors are innervated by cranial nerves that can influence ventilation (Nilsson, 1984; Smith, 1966), and that they are sensitive to changes in P_CO2 in carp (Cyprinus carpio), sea catfish (Protosus anguillaris) and Japanese eel (Anguilla japonica) (Hidaka, 1970; Konishi et al., 1969; Yoshii et al., 1980).

These taste bud-like receptors consist of two cell types: the taste receptor cells themselves and a single basal serotonergic cell termed the Merkel-like cell (MLC). The taste receptor cells sit in a ring on top of the basal serotonergic cell, forming a taste bud complex (Jackson et al., 2013; Zaccone et al., 1994; Zachar and Jonz, 2012a). Importantly, MLCs share many characteristics with NECs; they are innervated, positive for 5-HT and neurosecretory (Zachar and Jonz, 2012a). In addition, these taste bud complexes, which are found in all fish species so far examined, are situated on the gill rakers (see Glossary), an ideal location for sensing external hypoxia (Coolidge et al., 2008; Porteus et al., 2012). More recently, using in vivo Ca²⁺ imaging, Rosales et al. (2021 preprint) demonstrated that the vagal nerve (cranial nerve X) in larval zebrafish, which terminates on MLCs, exhibits an increase in Ca²⁺ activity under hypoxia, providing additional support for MLCs as potential O₂ chemoreceptors.

In Fig. 3, we present preliminary data supporting a role for MLCs in O₂ chemoreception in vivo. A Tg(tph1b:jRCaMP1a) transgenic zebrafish line was generated that expresses the genetically encoded Ca²⁺ indicator jRCaMP1a within MLCs situated in the pharyngeal arch region of larvae at 3 dpf. At 4 dpf, addition of 0.1 mmol l⁻¹ NaCN, a reagent that induces chemical hypoxia (Hamel, 2011), caused an increase in Ca²⁺ activity in 66% of the MLCs examined (Fig. 3). The concentration of NaCN used was 10 times less than that required to reliably elicit a ventilatory response in zebrafish (1 mmol l⁻¹ NaCN; Vulesevic et al., 2006). Current research in our lab is focused on the role of MLCs in contributing to the HVR in zebrafish.

In summary, our understanding of breathing in fishes has come a long way since Duverney first described ventilation patterns in 1701. Many of the major developments have arisen by incorporating new and innovative techniques. While reflecting on these impressive historical achievements, we were struck by the remarkable creativity and ingenuity exhibited by our predecessors. Indeed, the ability to find novel technical solutions, often by cobbling together bits and pieces scattered around the lab, is a hallmark of comparative physiology. We are hopeful that with the integration of genetic manipulation techniques, those currently accessible and those yet to be discovered, the next century will yield even more exciting discoveries on how fish sense their surroundings and regulate their breathing to match environmental conditions. Although the development of new and increasingly complex techniques is instrumental in driving scientific research forward, one must not lose sight of the fact that these techniques are merely tools and that significant discoveries typically rely on innovative and clever ideas that may lead to that ever-elusive paradigm-shifting hypothesis. As a community, comparative physiologists have excelled in this aspect of research and are well positioned to continue to do so in order to exploit the power of gene-based techniques.

We are grateful to the Animal Care and Veterinary Service (ACVS) staff for their care of the various zebrafish lines. The original research described in this article was supported by the University of Ottawa Faculty of Science core facilities, including the Imaging and Cytometry Facility (CICF), the Core Molecular Biology and Genomics Laboratory, and the Laboratory for Physiology and Genetics of Aquatic Organisms.