From tadpole to adult frog locomotion

The transition from larval to adult locomotion in the anuran, Xenopus laevis , involves a dramatic switch from axial to appendicular swimming including intermediate stages when the tail and hindlimbs co-exist and contribute to propulsion. Hatchling tadpole swimming is generated by an axial central pattern generator (CPG) which matures rapidly during early larval life. During metamorphosis, the developing limbs are controlled by a de novo appendicular CPG driven initially by the axial system before segregating to allow both systems to operate together or independently. Neuromodulation plays important roles throughout, but key modulators switch their effects from early inhibitory influences to facilitating locomotion. Temperature affects the construction and operation of locomotor networks and global changes in environmental temperature place aquatic poikilotherms, like amphibians, at risk. The locomotor control strategy of anurans differs from other amphibian groups such as salamanders, where evolution has acted upon the thyroid hormone pathway to sculpt different developmental outcomes.


Introduction
Swimming in tadpoles, as in fish, requires contractions of axial muscles (myotomes) and neural circuitry to generate left/right tail bending and coordinate a rostrocaudal sequence of contractions to produce forward propulsion. However, whilst still within the egg, amphibian embryos have already developed this neuromuscular machinery [1]. The neural circuitry comprising the swim CPG is thus constructed in ovo, partly through an intrinsic programme of gene expression, but with shaping by extrinsic influences. Understanding how this network operates and adapts to environmental and organismal demands during subsequent metamorphosis from tadpole to frog derives mainly from experiments on the clawed frog Xenopus laevis, which is the centrepiece of this review. However, the diversity of amphibian developmental programmes, body formats and locomotor strategies also provides fruitful avenues for probing the evolution of vertebrate locomotor control strategies in general, and the transition from the aquatic to terrestrial ecosystems.

Early tadpole locomotor development
Xenopus tadpoles hatch 2e3 days post-fertilization at developmental stage 37/38 [2], by which time wellcoordinated swimming occurs in response to stimulation. The tadpole swim CPG is among the most completely described vertebrate locomotor circuits (reviewed in the study by Roberts et al. [3] and Sillar and Li [4]). The core network, which involves descending ipsilateral excitation and reciprocal crosscord inhibition, is phylogenetically conserved from fish to mammals [5]. The same transcriptional coding system described in fish and mice likely specifies neuron subtypes during Xenopus tadpole development. The basic CPG network extends into the hindbrain and comprises ipsilaterally projecting descending interneurons (dINs) [6], interneurons with crossing projections (cINs) [7], ascending ipsilateral interneurons (aINs) [8], homologues of V3 interneurons [9], and myotomal motor neurons (MNs) [10]. The dINs provide the excitatory drive for swimming and appear to be Chx10 þve suggesting they are homologues of V2a interneurons in fish and mammals [11]. Glycinergic cINs mediate cross-cord reciprocal inhibition, like the V0d 0 s in fish and mice. The dINs are rapidly adapting, firing reliably but only once per swim cycle. cIN inhibition of contralateral dINs at mid-cycle is, therefore, critical to rhythm generation because it removes Na þ channel inactivation allowing dINs to fire on rebound. If the inhibition is weak and dINs cannot fire, swimming abruptly ceases [12]. Increases in CPG output cycle frequency are controlled by the recruitment of MNs, cINs and aINs. Interestingly, cIN and aIN recruitments have recently been shown to correlate inversely with input resistance [13], the opposite to that predicted by Henneman's size principle [14].
From hatching, the duration and variability of axial MN rhythmic output increases progressively along the spinal cord [15], in correspondence with the growing tadpole's need for greater swimming strength and flexibility. This early transition to more typical vertebrate 'burstiness' reflects changes in the firing properties of spinal MNs [16,17] which can now discharge multiply in each halfcycle. This change is accompanied by a decrease in electrical coupling within individual motor pools, thereby enabling de-synchronized firing [18].
The initial period post-hatching also witnesses changes in the sensory systems that influence locomotor behaviour. Some systems operate as interim structures appropriate for a predominantly sessile lifestyle, including the cement gland (CG) pathway that inhibits swimming [19,20]

Roles of neuromodulation
Neuromodulation plays important roles in tadpole CPG development and the eventual emergence of the appendicular system during metamorphosis [26]. Initially, between hatching and continuous freeswimming, the tadpole is especially vulnerable to predation. Whilst rapid changes are occurring to equip the animal for an independent existence, the modus operandi is to remain motionless, presumably as an anti-predatory strategy since many predators respond to movement. In association, various neuromodulatory signalling pathways become established, including dopamine (DA), serotonin (5-HT), noradrenaline (NA) and nitric oxide (NO), which alter particular facets of CPG output ( Figure 1). Immediately after hatching, they mainly promote a sessile existence (Figure 1a), but then switch later in development to swim facilitating roles [27]. For example, the effects of D1-like and D2-like DA receptor (R) activation on spontaneous bouts of swimming at premetamorphic Xenopus tadpole stages are dosedependent. High affinity D2Rs inhibit swimming, an   . Although seemingly counterintuitive to the emergence of continuous swimming, it is relevant that NO's reduction of the usAHP may contribute to the modulator's switch to overall excitation at free swimming stages.

Acquisition of limb-based locomotion
The fascinating transition from axial to limb-based locomotion during metamorphosis in amphibians is a highly species-specific process. In fully aquatic Xenopus, for example, the two locomotor systems appear sequentially during development (Figure 1), necessitating a critical period when they co-exist and must be coordinated simultaneously ( During prometamorphosis, motor output to each new limb bud consists of synchronous flexor-extensor motor bursts that alternate between the left and right sides, similar to, and in strict coordination with, the axial rhythm ( Figure 1c). Later, around the metamorphic climax, the tail and the now more developed limbs both contribute to locomotor propulsion, with the limb CPG activity having switched coordination such that intralimb extensor and flexor bursts now alternate and occur in synchrony across the two sides. Now the two co-existing CPGs are fully functional, but they can operate with different cadences (Figure 1d). The lower frequency limb CPG rhythm can occur either completely independently or accelerate to become transiently coordinated with the faster axial rhythm. Post climax, the tail has disappeared, and the solitary limb rhythm remains to drive left/right synchronized appendicular swimming, akin to human breaststroke ( Figure 1e). How is the coupling between the two systems regulated when they co-exist at mid-metamorphosis? Exogenous 5-HT and NA can couple or uncouple the axial and limb CPG outputs, respectively ( Figure 1d) [39], but how this is achieved remains a mystery since the coupling mechanism itself is unknown. Presumably, there are connections between secondary limb and primary axial CPG neurons, but the nature of these interactions, whether they are chemical or electrical and how they are modulated at different points along the metamorphosis timeline have yet to be investigated.
Further progress in understanding changes in spinal CPG function during metamorphosis awaits the availability of techniques suitable for cellular and molecular level analyses. At the MN level, investigating the temporal expression of Hox genes that define the specification of motor pools for the limb and axial systems [40] would be very illuminating. Patch clamp recordings from isolated preparations [36] have allowed intriguing initial insights, for example on the incorporation of functional D1Rs, the persistence of a usAHP and its negative modulation by NO. However, more advanced methods are now needed, such as calcium imaging that should allow real-time mapping of groups of neurons associated with the two CPGs when they are active. Associated neuromodulatory control mechanisms may be more difficult to explore given the additional complexities anticipated as the metamorphic transition proceeds. A conceptual framework is also needed that considers the changing physical distances over which modulation takes place. This is especially relevant to NO signalling whose neuronal sources in the early stages are restricted to the brainstem, in close proximity to rostral swim CPG neurons, but as the tadpole grows, nitrergic neurons appear within the future spinal appendicular CPG centres [41].

Evo-devo of metamorphosis
Amphibian metamorphosis effectively compresses into a few weeks transitional events that spanned w300 million years in evolutionary time to allow aquatic ancestors to exploit terrestrial habitats. Metamorphosis is driven by increasing plasma concentrations of thyroid hormone (TH) and remarkably, without sufficient TH, or if TH receptors are compromised, metamorphosis is arrested. A key step in metamorphosis is the transport of circulating THs across cell membranes in target tissues, with a key transporter thought to be monocarboxylate transporter 8 (MCT8). However, a recent study found that mutant frogs devoid of MCT8 did not have impaired metamorphosis, suggesting the involvement of additional transport mechanisms [42].
Comparative studies on metamorphosis have shed light on the evolution and development of vertebrate locomotor movement strategies [43] and on how the forces of natural selection have acted on the TH pathway to shape amphibian diversity. In some species, such as salamanders, metamorphosis is incomplete with the two systems co-existing in adulthood; the axial system drives undulatory aquatic swimming while the limbs are used for terrestrial quadrupedal walking [44]. However, in contrast to Xenopus, left/right limb alternation is employed in both behaviours, perhaps indicative of a single axial CPG capable of driving the two modes of locomotion. During urodele evolution, changes in the TH pathway underpin the retention of neotenous characteristics. In axolotls, for example, the thyroid gland produces insufficient TH for metamorphosis, but it can be induced in key tissues by exogenous TH [45], so the cellular machinery, including intracellular TH receptors, must be present. Other salamanders like the mudpuppy, Necturus, are paedomorphic because although the thyroid gland produces TH, target tissues are unresponsive, even though functional TRs are expressed [46]. This suggests that genes linked to metamorphosis but downstream of TR receptor activation are compromised. Among some anurans, metamorphosis is completed directly in ovo, as in the Puerto Rican coqui tree frog which emerges from the egg having never previously generated axial locomotor behaviour [47]. The challenges this animal faces during its early development include the need for extremely large eggs (w3 mm versus w1 mm in Xenopus) to provide sufficient yolk-derived energy for in ovo metamorphosis. The emerging froglet is also tiny as a result, easily fitting on a US penny. Similarly, in marsupial frogs, which have the largest of all (up 10 mm) amphibian eggs, the animals develop in and hatch from special brood pouches on their mothers back as froglets or developmentally advanced tadpoles [48].
While comparative studies of different amphibian species offer clues regarding the transition from water to land, recent research on fish has also provided fascinating insights into the evolution of the vertebrate limb. The little skate Leucoraja uses its pectoral fins to swim in the familiar skate-like fashion but uses its pelvic fins to walk bipedally along the seafloor [49]. Interestingly, the use of molecular approaches to examine transcription factor expression revealed the same Hox genes as used to construct the mammalian hindlimb [50,51], indicating a conservation of gene regulatory machinery across vertebrates with paired lateral appendages ( Figure 2).

Environmental impact on tadpole development
As ectotherms, amphibians are exposed to the elements throughout development which, for fully aquatic species, spans the entire larval phase from hatching to metamorphosis. In Xenopus, the effects of temperature on metamorphosis are dramatic, taking 12 days at 22 C but 36 days at 15 C [52]. Correspondingly, the Q10 effect on the proteins and pathways involved, including the TH system, are slowed and can even arrest metamorphosis with cooler temperatures. In addition, because locomotor activity itself contributes to motor system development, the developmental programmes that generate and adapt movements are also temperature-dependent. Accordingly, if movements are slower in colder conditions, there will be negative consequences for the developmental progression of the animal's behavioural repertoire.
Early tadpole development is especially sensitive to changes in ambient temperature, and even in ovo, locomotor circuit construction is profoundly affected. A recent study [53] explored how abiotic stress regulates myotomal MN development by rearing embryos from early blastula stages (stages 7e8) in relatively cool (14.5 C) or warm (22.5 C) environments and then examining their locomotor behaviour from hatching up to stage 40. Somewhat counterintuitively, cold-reared larvae displayed more forceful escape swimming responses when stimulated, correlated with an increased number of spinal MNs (Figure 3). Thus, the response to a more challenging environment appears to be the development of a more robust motor control system. Mechanisms underlying this cold hardening response involves activation of a cold receptor, TRPM8, which increases Ca 2þ waves during development, facilitating expression of the transcription factor Hb9 resulting in increased MN numbers. Whether there are additional effects on the motor circuitry is unknown but seems likely.
The effects of raised temperature on the developing ectothermic brain also have important consequences for locomotor behaviour and hence survival, so aquatic animals like amphibians and fish may be particularly vulnerable to the impact of global warming [54]. In principle, rising water temperatures could have the opposite effect to cold hardening, producing tadpoles with fewer MNs that are less able to escape from predators and transition to adulthood. With respect to metamorphosis, warmer temperatures profoundly affect its duration; Xenopus tadpoles exposed to higher temperatures tend to metamorphose quicker (the drying pond effect), but consequently, they are smaller [55]. Whilst this strategy can help to avoid desiccation and death, there are adverse outcomes for an individual's fitness. In Rana chensinensis, warm temperatures not only accelerate the time to metamorphosis but they also impair skeletal development and ossification such that froglets have reduced bone length, affecting their jumping capability [56]. Furthermore, the thyroid gland is significantly smaller in warm-reared tadpoles and larger in cold-reared animals. Thus, multiple factors influencing the growth rate and size of the newly metamorphosed froglet can be affected, impacting the animal's survival. A recent study demonstrated in the In the little skate, LMC MNs innervating the adjacent pectoral and pelvic fin muscles also express specific Hox genes and FoxP1 to control swimming and pelvic fin walking. Similarly, in quadrupeds like frogs and mice, fore-and hindlimb locomotor control involves flexor and extensor muscles innervated by LMC MNs in cervical and lumbar cord regions, respectively. In inter-limb (thoracic) regions, FoxP1 expression is repressed by Hoxc9 so that thoracic segments are innervated only by axial MNs. In skates, a natural deletion of the HoxC cluster led to a lack of Hoxc9 along the spinal cord and thus to an inability for LMC Hoxc9-related specification. Together these findings suggest that the gene regulatory system required for appendicular motor control evolved from a common ancestral vertebrate possessing paired appendages. Figure  tropical Pacific horned frog Ceratophrys stolzmanni that locomotor performance is negatively affected by animals being relatively small after metamorphosis [57]. Jumping distances post-metamorphosis were >50% greater in larger juveniles than smaller ones, an outcome associated with hindlimb length and affecting the ability to evade predators and access food resources. Moreover, even with no risk from predators and ample food available, smaller froglets still showed a higher mortality. The ultimate consequence of rising temperatures is, therefore, a lower survival rate, suggesting that amphibians are especially susceptible to climate change, making them highly relevant indicators of global warming.

Conclusions and future prospects
The Xenopus locomotor system has been an influential model in motor control research, providing many insights of general importance in neuroscience. In this organism, the underlying CPG mechanisms can be studied in detail from embryos and hatchling tadpoles through the various stages of metamorphosis. Locomotion can be investigated on a broad range of levels (molecular, cellular and synaptic, neuromodulatory and behavioural), using a wide variety of methods. The studies on how tadpoles are transformed into frogs also afford an important evolutionary perspective, providing clues about the evolution of the vertebrate limb and appendicular locomotion.
On hatching, the stage 37/38 axial swim CPG is understood in remarkable detail, but in contrast, only the surface has been scratched regarding later stages from pre-metamorphosis through the climax to the froglet, and despite some tantalising glimpses, progress has temporarily stalled. We know little about how the limb

Current Opinion in Neurobiology
Developmental adaptation to cold temperature. In animals grown in cold temperatures (left), cold-sensitive TRPM8 channels are activated and trigger higher frequency Ca 2+ spikes (blue insert), which increases Hb9-dependent MN differentiation and survival. Consequently, MN numbers are greater than in animals grown in warm temperature (right). This developmental temperature-dependent adaptation allows tadpoles to escape more quickly in cold water than if they had been reared in warm water, increasing their chances of escaping predators. Figure  control systems are constructed within the framework of a pre-existing axial CPG circuit. The future application of molecular genetics and cellular-resolution population imaging techniques will enable initial observations to mature into a more detailed understanding of the mechanisms underpinning the development of amphibian locomotion, especially during the dramatic process of metamorphosis.

Declaration of competing interest
The authors declare no competing interests that could have appeared to influence the work reported in this paper.

Data availability
This review article does not report unpublished data. Motor neuron recruitment is important for adjusting the force required for most movements. According to Henneman's size principle, when greater forces are required, larger motor neurons are recruited. This study investigated the recruitment of inhibitory interneurons during fictive swimming in hatchling Xenopus tadpoles. In contrast to current dogma, the recruitment of inhibitory interneurons was found to be negatively correlated with their input resistances (a proxy for cell size). This is partly because the amplitudes of the synaptic inputs to inhibitory interneurons scaled inversely with input resistance. Using computer simulations of the swim CPG it was also shown that this inverse relationship is required for the maintenance of the locomotor rhythm.
14. Henneman E: Relation between size of neurons and their susceptibility to discharge. Hatchling Xenopus laevis tadpoles are already endowed with a functional lateral line (LL) mechanosensory system that can be activated by water movements, as demonstrated in this article. Activating LL afferents with suction initiates escape swimming behaviour away from the direction of suction, a response that disappears when the LL neuromasts are ablated with neomycin. Fictive versions of responses to similar stimuli were less reliable, however. The paper revealed the nature of the LL afferents with DASPEI staining and also used calcium imaging to locate the sensory interneurons in the hindbrain that are activated by stimulation of LL afferents in the anterior LL nerve. A proportion of hatchling Xenopus tadpole swim CPG neurons display a post-swim usAHP, although normally, never excitatory dIN interneurons. dINs are unusual in having a depolarized resting membrane potential (RMP) and low input resistance (IR). They are also the only swim CPG neurons i that express an Ih sag potential when hyperpolarized. The Ih blocker, ZD7288, increases dIN IR and hyperpolarizes their RMPs suggesting that Ih is active at rest and this current explains their atypical properties. Importantly, under ZD7288, dINs now generate a usAHP indicating that the normal absence of this property is not due to an actual lack of dynamic pumps, but instead Ih represses usAHP function during intense activity. It is proposed that this mechanism protects dIN from becoming refractory so that tadpoles can always generate escape swimming, even at very short inter-swim intervals. In this article, a new isolated brainstem/spinal cord preparation is described allowing patch clamp recordings to be made from spinal neurons in pre-metamorphic stages during spontaneous episodes of fictive swimming. These recordings revealed that the usAHP described at early stages not only remains but occurs in a higher percentage of rhythmically active spinal neurons and is reduced or abolished by NO. In contrast to earlier stages, the paper also reports the presence of functional D1-like excitatory dopamine receptors. Elife 2022:11. The little skate uses its pelvic fins to walk across marine substrates. This species therefore offers a unique opportunity to explore the evolution and development of the vertebrate limb and is rapidly becoming an influential model for understanding the genetic origins of limb circuit during ontogeny. This important paper documents the genome of this species, Leucoraja erinacea, focussing on the characterization of genes responsible for the development of motor neurons and making comparisons with other vertebrate species (mouse and chicken) to determine conserved and divergent molecular mechanisms. Curr Biol 2019, 29:1787-1799. In this paper the authors describe how rearing temperature affects motor system development using Xenopus laevis embryos and young larvae as a model. Embryos reared in cold temperature generate stronger escape responses compared to siblings reared in warmer conditions. This is shown to be due to a cold-sensitive receptor, TRPM8, which increases calcium spike frequency and leads to the generation of more motor neurons in cold reared tadpoles. Bone deposition during amphibian metamorphosis is fundamentally important to animal fitness and survival. In this report the important influence of environmental rearing temperature on this process was investigated in Rana chensinensis tadpoles, a Chinese species that is particularly susceptible to high temperatures. It is shown that rearing tadpoles in warm temperatures (26 C) reduces ossification and causes damage to the thyroid gland. The resulting froglets have shorter hindlimbs and consequently are unable to jump as far as their coldreared (16 C) siblings. 57 * . Szekely D, Cogalniceanu D, Szekely P, Armijos-Ojeda D, Espinosa-Mogrovejo V, Denoel M: How to recover from a bad start: size at metamorphosis affects growth and survival in a tropical amphibian. BMC Ecol 2020, 20:24. This article investigated the relationship between the size of pacific horned frogs and their survival chances later in life. The data support the notion that larger froglets display a significantly higher survival rate (95%) compared with individuals that are smaller postmetamorphosis. Furthermore, the larger animals could jump further, and their head width was greater, indicating that they were better able to capture prey. Despite these advantages of being larger, small individuals after metamorphosis displayed higher growth rates, decreasing the size gap in adulthood.