Research articleBehavioural changes controlled by catecholaminergic systems explain recurrent loss of pigmentation in cavefish
Abstract
Multiple cave populations of the teleost Astyanax mexicanus have repeatedly reduced or lost eye and body pigmentation during adaptation to dark caves. Albinism, the complete absence of melanin pigmentation, is controlled by loss-of-function mutations in the oca2 gene. The mutation is accompanied by an increase in the melanin synthesis precursor l-tyrosine, which is also a precursor for catecholamine synthesis. In this study, we show a relationship between pigmentation loss, enhanced catecholamine synthesis and responsiveness to anaesthesia, determined as a proxy for catecholamine-related behaviours. We demonstrate that anaesthesia resistance (AR) is enhanced in multiple depigmented and albino cavefish (CF), inversely proportional to the degree of pigmentation loss, controlled by the oca2 gene, and can be modulated by experimental manipulations of l-tyrosine or the catecholamine norepinephrine (NE). Moreover, NE is increased in the brains of multiple albino and depigmented CF relative to surface fish. The results provide new insights into the evolution of pigment modification because NE controls a suite of adaptive behaviours similar to AR that may represent a target of natural selection. Thus, understanding the relationship between loss of pigmentation and AR may provide insight into the role of natural selection in the evolution of albinism via a melanin–catecholamine trade-off.
1. Background
Albinism, the absence of melanin pigmentation, occurs at very low frequencies in animals, probably a consequence of low fitness of albinos imposed by many advantages of melanin, including protection from UV irradiation, camouflage and participation in innate immunity [1–3]. By striking contrast, albinism and reduced pigmentation are common in animals adapted to the lightless environment of caves, where a large number of colourless species have evolved from pigmented ancestors [4,5]. Thus, reduction or loss of pigmentation, like blindness, is a hallmark of cave-dwelling animals.
The loss of pigmentation in cave animals is best understood in the Mexican tetra, Astyanax mexicanus, which consists of a pigmented and eyed surface-dwelling form (surface fish, SF) and multiple depigmented and eyeless cave-adapted forms (cavefish, CF) [6,7]. Depigmentation has evolved in different CF populations by reducing the number of melanophores, decreasing the amount of melanin deposited in melanosomes or abolishing melanin synthesis (albinism) [8]. The loss of pigmentation is controlled by multiple genes [9], including the mcr1 gene, which is responsible for melanin reduction [10], and the oculo and cutaneous albinism 2 (oca2) gene, which is responsible for albinism [11]. Albinism in Pachón CF (PA) and Molino CF (MO) evolved by distinct large deletions in the oca2 coding region [11]. The physiological role of the OCA2 protein is not well understood, although it may serve as a melanosome membrane channel regulating the supply of l-tyrosine, a precursor in the melanin synthetic pathway [12].
The evolution of albinism in cave animals is often attributed to the relaxation of purifying selection and random accumulation of deleterious mutations in pigment-forming genes [13–15]. The presence of both negative and positive polarities for pigment quantitative trait loci (QTLs) in Astyanax support the neutral hypothesis, which suggests that pigment loss is the result of genetic drift [9]. However, repeated mutational targeting of the oca2 gene in Astyanax albinism [11], the ability to rescue melanin synthesis by supplying l-DOPA to albino CF [16], cave cixiids [17] and other albino cave animals [18], as well as melanization of wound sites in albino cave arthropods [1], are inconsistent with this hypothesis. An alternative hypothesis posits that mutations in oca2, or another gene involved in the first step of melanin synthesis, are targets of natural selection based on a pleiotropic trade-off between the melanin and catecholamine synthesis pathways [19]. The melanin–catecholamine trade-off is plausible because both pathways use l-tyrosine as a precursor and are sensitive to its availability [20–24]. In Astyanax, this hypothesis is supported by higher levels of dopamine (DA) and norepinephrine (NE) in PA relative to SF [19,25] and by the increase of l-tyrosine and DA after oca2 knockdown of melanophore development in SF [19]. An attractive aspect of the melanin–catecholamine trade-off hypothesis is that catecholamines are involved in a suite of potentially adaptive CF behaviours, including enhanced feeding drive and reward [26], stress [27], and, in particular, wakefulness/arousal (reduced sleep) [28], which represent potential evolutionary benefits of inhibiting melanin synthesis.
During a screen for behavioural and physiological differences linked to depigmentation, we uncovered a difference in anaesthesia resistance (AR) between SF and CF. This prompted further investigation because of reports that albino laboratory mice [29] and humans with red hair and fair skin [30] have increased tolerance to anaesthesia. Although not identical, anaesthesia and naturally occurring CF behaviours, such as wakefulness/sleep, may have the same or similar underlying physiological mechanisms and molecular targets, as has been demonstrated in Drosophila [31,32] and mammals [31,33]. By using a simple and sensitive assay for AR, we quantified differences in the overall physiological capacity for activity, alertness and responsiveness to stimuli in Astyanax. The results reveal a direct relationship between AR, depigmentation and increased NE in multiple CF populations, providing new insights into the evolution of pigment modification via a melanin–catecholamine trade-off.
2. Methods
(a) Biological materials
SF, PA, Tinaja (TI) and Chica CF (CH) were the offspring of laboratory stocks. The SF stock was originally collected from Nacimeinto del Rio Choy, San Luis Potosí, Mexico; Rio Subterráneo CF (SU) were collected from La Cueva del Rio Subterráneo (Micos), San Luis Potosí, Mexico and MO were collected from El Sótano de Molino, Tamaulipas, Mexico or laboratory stock provided by the Tabin laboratory, Harvard Medical School. F1 hybrids were produced by in vitro fertilization of PA eggs with SF sperm. The albino-eyed (AE) line was produced by artificial selection of the F2 progeny of an SF × PA cross. Intercrossing and propagation of progeny with large eyes and no visible pigmentation eventually led to the F5 AE generation used in these experiments.
Adult fish were maintained and embryos were obtained and raised as described previously [16,19,27]. Methods for fish handling and care were approved by the University of Maryland IACUC and conformed to NIH guidelines.
(b) Anaesthesia resistance assays
Tricaine methanesulfonate (tricaine or MS222; pH adjusted to 7.0; Western Chemical, Ferndale, WA), a general anaesthetic that acts preferentially on neural voltage-gated sodium channels [34], was used in most of these studies. We also used isoflurane (Sigma-Aldrich, St Louis, MO), and a bath of ice water (less than 4°C) [35]. The effective concentrations of anaesthetics were determined empirically to provide reasonable assay times and high survival rates.
Anaesthesia assays of larval and juvenile fish from 3 days post-fertilization (dpf) to 2 months old with tricaine (0.1 mg ml−1) were conducted at room temperature as follows. A single randomly selected larva was placed in the well of a 9.6 cm2 plastic cell well plate containing fish system water. Most of the water was then removed using a micropipette, and the well was quickly filled with 4 ml of fish system water. After an acclimation period of 5 min, the assay was initiated by introducing 4 ml of double strength tricaine (0.2 mg ml−1 in fish system water) to the well, and the time until unconsciousness was recorded while the larva was viewed continuously under a stereomicroscope. The assay was terminated when motion ceased and fish were no longer responsive to prodding with a fine tungsten wire or loop. Assays of adult fish (aged 1 year or more) with tricaine (0.2 mg ml−1) or isoflurane (0.5 ml l−1) were conducted as described above except that a fish was placed in a glass dish containing 100 ml (tricaine) or 150 ml (isoflurane) of fish system water, left for 5 min to acclimate, and the assay was initiated by adding 100 ml of double strength tricaine (0.4 mg ml−1) or 150 ml of double strength isoflurane (1 ml l−1). Time until unconsciousness was recorded visually by the investigator. All the raw measurements can be found in the electronic supplementary material, file S1. Systat software was used to perform Student's t-test or ANOVA with Bonferroni corrections and regression analysis. Correlation was calculated using the Alcula Lineal Regression Calculator (www.Alcula.com).
(c) Pharmacological treatments
The drugs were purchased from Sigma-Aldrich Corp, dissolved in fish system water, and applied to larvae or juveniles at effective concentrations and exposure times determined empirically [36–39]. Fish were incubated with quinpirole (12.5 µM) for 60 min, haloperidol (9 µM) or desipramine (21 µM) for 30 min, and atipamezole (50 µM) for 5 min, and anaesthesia assays with tricaine were carried out as described earlier.
(d) Amino acid and catecholamine treatments
Chemicals (obtained from Sigma-Aldrich) were freshly prepared in sterilized fish system water and used immediately. Anaesthesia assays with tricaine were conducted with 8 dpf larvae as described above after pre-incubation in the dark with 200 µM DA for 10 min, 200 µM NE [40] for 10 or 60 min or amino acids (10 µg ml−1 [16]) for 60 min.
(e) Melanophore quantification
Melanophores were counted in 200 µm2 regions immediately below the anterior margin of the dorsal fin in randomly selected CH individuals.
(f) High-performance liquid chromatography
NE in whole brain samples was analysed using HPLC with electrochemical detection as described previously, with minor modifications [41]. Brains were placed into 100 µl of acetate buffer (pH 5.0) containing the internal standard α-methyl DA (Merck & Co., Kenilworth, NJ), sonicated and stored at −70°C. Prior to analysis, 4 µl ascorbate oxidase (1 mg ml−1, Sigma-Aldridge) was added to the samples followed by centrifugation at 17 000g for 15 min. A Waters Alliance e2695 was used to inject 50 µl of the supernatant onto a C18 4 µm NOVA-PAK column (Waters Associates, Milford, MA) held at 30°C. Electrochemical detection was accomplished using an LC 4 potentiostat and glassy carbon electrode (Bioanalytical Systems, West Lafayette, IN) set at 0.5 nA V−1 with an applied potential of +0.7 V versus a Ag/AgCl reference electrode. The pellet was solubilized in 400 µl of 0.4 N NaOH and protein content was analysed according to Bradford [42]. A CSW32 data program (DataApex, Czech Republic) was used to determine monoamine concentrations in the internal standard mode using peak heights calculated from standards. Monoamine concentrations were corrected for injection versus preparation volumes and divided by microgram protein to yield picogram monoamine per microgram protein. Data from the NE analysis were tested for differences using a Kruskal–Wallis one-way ANOVA on Ranks because equal variance among samples was rejected (SigmaStat version 3.5, Systat Software, San Jose, CA). In analyses that revealed a significant effect between groups, Dunn's method was used to conduct pairwise comparisons. Application of Grubb's test [43] to detect outliers resulted in the deletion of data from one brain (of 38) from the dataset. Significance levels for all statistical tests were set at p < 0.05.
(g) Gene knockdown
A morpholino was injected into SF eggs to knockdown the oca2 gene as described previously [19]. Morphant and control SF larvae were subjected to AR assays with tricaine at 2.5–3 dpf as described above.
3. Results
(a) Anaesthesia resistance in cavefish
The effects of tricaine anaesthesia were determined in adult SF and different CF populations (figure 1a,b). PA showed a two- to threefold higher AR than SF and the difference in AR was independent of body weight (figure 1c). PA × SF F1 hybrid larvae and adults showed AR levels similar to SF (figures 1b and 2e). Pachón CF (PA), Molino CF (MO) and Chica CF (CH) showed significantly higher AR levels than SF, in some cases (MO) exhibiting a four- to fivefold increase in AR (figure 1b). Depigmented Tinaja CF (TI) and Rio Subterráneo CF (SU) also showed higher mean levels of AR than SF, although these differences are statistically insignificant. AR was also higher in PA than in SF exposed to ice water or isoflurane (figure 1d), which induce unconsciousness through different mechanisms than tricaine [44,45].
Figure 1. Effects of anaesthesia on surface fish and cavefish adults. (a) Adult surface fish (SF), Pachón (PA) cavefish, PA × SF F1 hybrid (HY), Tinaja (TI) cavefish, Chica (CH) cavefish, Rio Subterráneo (SU) cavefish, Molino (MO) cavefish and the AE line. (b) Bar graph showing tricaine anaesthesia effects in depigmented or albino CF, F1 hybrids and AE relative to SF. (c) The relationship between tricaine anaesthesia and weight in SF (left) and PA (right) adults. Diagonal lines indicate slopes of fitted lines. r: regression coefficient. (d) Bar graphs showing the effects of ice water (left) and isoflurane (right) anaesthesia on SF and PA. Significant p-values shown in red and non-significant p-values shown in black above error bars. Error bars represent the standard deviation of the means. The numbers of each fish type subjected to analysis are indicated at the base of each column. All p-values from each individual comparison can be found in the electronic supplementary material, file S2. Figure 2. Development of differential AR in CF. (a,c) SF and (b,d) PA larvae at (a,b) 3 and (c,d) 4 days post-fertilization (dpf). Intense melanin pigmentation develops in SF eyes between 3 and 4 dpf. (e) The effect of tricaine anaesthesia on SF, PA, TI, AE and PA × SF F1 hybrids during larval development. Details are the same as in figure 1.
The effects of tricaine anaesthesia were also investigated during larval development (figure 2). An increase in AR appears in PA larvae between 3 and 4 dpf compared with SF (figure 2e), which coincides with the interval in which pigmentation is substantially increased in the SF eye due to differentiation of the retinal pigment epithelium (RPE; figure 2a–d). Increases in AR were also evident at 7 dpf (earlier time points not tested) in TI relative to SF (figure 2e). In general, larval fish of all types showed lower AR than their adult counterparts (figures 1b and 2e).
(b) Anaesthesia resistance and melanin pigmentation
As AR and the loss of pigmentation (figures 1a,b and 2) appeared to be correlated in all CF populations, we carried out further studies to clarify the relationship between these two traits. CH show high variability in depigmentation, ranging from moderately pigmented individuals to nearly complete albinos, probably due to ongoing hybridization with a nearby SF population in their natural habitat [46]. Variable pigmentation and the high variance of AR (figure 1b) in CH prompted us to determine whether there is a direct relationship between AR and melanophores in different CH individuals. The results of regression analysis revealed that fish with higher melanophore number have lower AR (R2 = 0.592; p < 0.001; figure 3a).
Figure 3. The relationship between tricaine AR and pigmentation in the variably pigmented CH. (a) Regression analysis comparing the effects of tricaine anaesthesia to the number of melanophores in randomly selected CH. AR is negatively related to the number of melanophores. The diagonal line indicates the slope of the fitted line. r: regression coefficient. (b) Tail fin pieces of adults showing the presence of melanophores in SF (top) and not in the AE (middle) or PA (bottom). (c) The effects of oca2 knockdown on melanin formation in SF. Injection of oca2 morpholino results in a marked reduction of melanin pigmentation in SF larvae at 3 dpf. (d) The effects of oca2 knockdown on AR in SF. AR is increased in oca2 morphants (oca2 MO) with reduced pigmentation, but not in control morphants (CMO) containing pigmentation when compared with uninjected SF. Details are the same as in figure 1.
Differences in eye development, eye growth in SF and eye degeneration in CF are also in progress between 3 and 4 dpf, coinciding with the interval when both pigmentation and AR are increasing (figure 2). To distinguish between the roles of eye development and pigmentation in the development of AR, we used an artificially selected AE line (figures 1a and 3b). Our results showed that both AE adults (figure 1b) and larvae (figure 2e) resembled depigmented CF populations in showing increased AR when compared with SF.
Albinism is caused by loss of function mutations in the oca2 gene in PA and MO [11], and morpholino knockdown of oca2 expression in SF greatly reduces larval pigmentation [19]. To directly investigate the relationship between AR and depigmentation, we studied AR in SF embryos injected with oca2 morpholinos. Oca2 morphants exhibited reduced pigmentation (figure 3c) and significantly higher AR typical of CF, but control morphants showed no significant changes in AR relative to un-injected SF controls (figure 3d).
(c) l-tyrosine increases AR in cavefish
Knockdown of oca2 increases the pool of l-tyrosine [19], so we tested the effects of exogenous l-tyrosine on AR. PA larvae were incubated with l-tyrosine, or l-tryptophan as a control, at 3 dpf, prior to the natural rise in AR at 4 dpf (figure 2). The results showed a significant increase in AR in the presence of exogenous l-tyrosine, but not l-tryptophan, relative to untreated larvae (figure 4).
Figure 4. AR is increased by exogenous l-tyrosine in PA cavefish. Bar graphs showing AR in untreated (left), l-tyrosine incubated (middle) and l-tryptophan incubated (right) PA at 3 dpf. Dashed lines indicate significant differences in the l-tyrosine treated larvae compared with untreated controls (left) and l-tryptophan-treated (right) larvae. Details are the same as in figure 1. (Online version in colour.)
(d) Anaesthesia resistance and catecholamines
Next, we used a pharmacological approach to explore the possible relationship between differential AR and catecholamine levels in the brain. In these experiments, SF and PA were pre-incubated with catecholamine receptor agonists or antagonists prior to determining AR. Neither the D2 agonist haloperidol nor the D2/D3 antagonist quinpirole had significant effects on AR (figure 5a,b). By contrast, treatment with desipramine, an NE reuptake inhibitor, significantly increased AR in PA (figure 5c), whereas the α2-adrenergic antagonist atipamezole, significantly decreased AR in SF or PA (figure 5d).
Figure 5. Pharmacological determinations of the roles of DA and NE in tricaine AR. (a,b) The DA D2 agonist haloperidol (a) and the D2/3 antagonist quinpirole (b) did not affect AR in SF or PA larvae assayed at six months post-fertilization. (c,d) The NE reuptake blocker desipramine (c) and the α2 antagonist atipamezole (d) increased or decreased AR, respectively, in SF and PA assayed at 7 dpf and 60 dpf, respectively. Details are the same as in figure 1. (Online version in colour.)
The available information on NE levels in Astyanax is restricted to PA and SF [19,25]. Therefore, we quantified NE in the brains of TI and AE relative to SF and PA (figure 6a). The results showed that brain NE is elevated in TI similar to PA when compared with SF (figure 6a). NE levels were also elevated, although not significantly, in AE brains.
Figure 6. The relationship of AR and NE. (a) NE concentrations in isolated brains of adult SF, PA, TI and AE. (b) The effects of exogenous NE and DA on AR in SF and PA larvae (8 dpf). NE but not DA treatment increases AR in SF and PA. Details are the same as in figure 1. (Online version in colour.)
To directly determine the role of catecholamines in AR, we incubated SF and PA larvae with DA or NE. The results showed that NE increased AR in both SF and PA in a time-dependent manner (figure 6b). By contrast, no significant changes in AR were observed following exposure to DA (figure 6b). These results show that exogenous NE, but not DA, can boost the levels of AR.
4. Discussion
Astyanax CF has repeatedly evolved depigmentation and albinism following a few million years or less of separation from pigmented SF ancestors [8]. The role of natural selection in the regressive evolution in cave animals is not well understood. Research on Astyanax eye loss has identified antagonistic pleiotropy as a possible explanation because the same gene, sonic hedgehog, controls eye size, jaw size and the number of taste buds. Therefore, reducing the eyes may enable the formation of larger jaws and more taste buds, traits that are likely to be beneficial for feeding in the cave environment [47]. Likewise, we have shown that reduction of pigmentation can lead to the enhancement of catecholamine synthesis through a pleiotropic trade-off [19]. However, the possibility that loss of melanin pigmentation might directly influence beneficial behavioural or physiological changes that could serve as a target of natural selection has not been previously explored.
Our results suggest that increases in AR evolved in parallel with the reduction or loss of pigmentation in multiple CF populations. Several lines of evidence support a strong link between AR and depigmentation. First, the extent of AR is proportional to the degree of CF depigmentation: albino PA and MO showed higher AR than TI and SU, which have reduced pigmentation. This difference is not a consequence of laboratory breeding because MO and SU were collected in the field, while PA and TI were raised for several generations in the laboratory. Likewise, the light/dark cycle in which the fish were maintained in the laboratory would not be expected to affect our results, because the potentially lower levels of melanin that CF may exhibit in dark caves would be expected to increase AR, making the difference between CF and SF even greater than the values we measured in laboratory-bred stocks. Second, our experiments revealed an inverse relationship between the degree of pigmentation and the level of AR: variably pigmented CH individuals with low numbers of melanophores showed high AR, and those with high numbers of melanophores showed low AR. Third, PA × SF F1 hybrids had the same level of AR as SF, suggesting that AR, like depigmentation, is a recessive trait. However, high AR variance in hybrid CH and similar AR values in F1 hybrids and SF suggest that additional regulators of AR may be present that segregate independently of pigmentation loci. Fourth, AR was significantly higher, approaching the levels observed in albino CF, in AE compared with SF. The existence of high AR in AE suggests that increased AR is correlated with depigmentation rather than eye degeneration. Lastly, knockdown of the oca2 gene substantially reduced pigmentation [19] and increased AR in SF. Taken together, these findings suggest that increased AR is closely linked to depigmentation.
The melanin–catecholamine trade-off hypothesis [19] proposes that catecholamine levels are increased in albino CF because the inhibition of melanin synthesis results in concurrent expansion of l-tyrosine pools (figure 7). The increase in AR in SF after oca2 knockdown or treatment with exogenous l-tyrosine is consistent with a melanin–catecholamine trade-off as an underlying mechanism for increasing AR. The existence of a melanin–catecholamine trade-off is also in line with higher catecholamine levels in the PA brain and body [19,25], the increase of NE in the TI brains, and a trend for increased NE in AE brains.
Figure 7. A model for the role of the melanin–catecholamine trade-off with shared l-tyrosine pools in AR. (a) l-tyrosine pools are lower in SF because the precursor is used in both the melanin and catecholamine synthesis pathways. (b) The excess l-tyrosine substrate provided by the inhibition of melanin synthesis promotes increased catecholamine synthesis and leads to enhanced AR in CF. Circles represent l-tyrosine pools. Symbol on lower arrow of (b) represents mutation in the CF oca2 gene. (Online version in colour.)
What are the developmental events and systems that could be involved in melanin–catecholamine trade-off? It seems likely that the trade-offs are linked to developmental periods of extensive melanin synthesis in SF. A large increase in overall pigmentation occurs in SF larvae between 3 and 4 dpf as melanin is synthesized and deposited in the differentiating RPE, transforming the eyes from translucent to opaque black spheres. Differential AR appears at the same time in PA, suggesting that a melanin–catecholamine trade-off could occur between the unpigmented RPE and the brain during CF larval development. l-tyrosine is derived from the diet or enzymatic hydroxylation of phenylalanine. In non-feeding larvae, the availability of l-tyrosine is limited to yolk reserves. In albino CF, l-tyrosine pools expanded by the inhibition of RPE pigmentation could be diverted to the catecholamine synthesis pathway in the larval brain, thus explaining the relatively large increases in AR at this time. Some TI adults retain small pigmented eyes buried in the orbits [48], so their levels of l-tyrosine recovery may be lower, resulting in a more modest AR increase.
Our study has specifically implicated adrenergic circuitry as the physiological target responsible for generating differential AR. Both pharmacological studies and the addition of catecholamines showed that AR is controlled by NE. Importantly, the addition of l-tyrosine alone causes an increase in AR. These results illustrate the plasticity of the AR phenotype in both SF and PA, which is consistent with the flexibility of l-tyrosine pools predicted by the melanin–catecholamine trade-off hypothesis (figure 7). Noradrenergic cell bodies are primarily located in the locus coeruleus and send efferent projections to almost every part of the CNS, where NE acts on noradrenergic receptors to alter autonomic, endocrine and behavioural responses. Accordingly, small changes of NE availability, similar to those recorded in CF, may have widespread effects on physiology and behaviour.
In vertebrates, the locus coeruleus modulates the diurnal patterns of wakefulness/arousal [49], and anaesthetics function by interfering with wakefulness/arousal [32,50]. For example, a single mutated gene and similar molecular pathways are involved in altering both wakefulness/arousal and sensitivity to anaesthesia in Drosophila [31,32]. Our results revealed similarities between AR and reduced sleep (wakefulness) behaviour in Astyanax. PA and TI larvae and PA and MO adults show increased AR and wakefulness relative to SF. However, changes in wakefulness were not detectable in TI adults [28], although we observed a weak difference in AR in TI relative to SF. These differences could be related to the sensitivity of the behavioural assays or to the source of TI used in the experiments. In this regard, we noted a wide range of AR values among different pedigrees in our TI CF stock. Furthermore, pharmacological studies indicated that AR and wakefulness [51] are dependent on noradrenergic receptors, implicating NE as the activating ligand in these behaviours. Despite similarities between wakefulness and AR in Astyanax, their genetic properties may not be the same: wakefulness was interpreted as a dominant trait because of its distribution in F2 intercross progeny [28,52], whereas differential AR, like albinism [53], is inherited as a recessive trait. Therefore, wakefulness and AR may be very similar although not identical behaviours.
The relationship between AR and loss of pigmentation via a melanin–catecholamine trade-off provides insights into the evolution of albinism and depigmentation. AR is a simple phenotype that can be used to quantify the state of alertness/arousal/consciousness and the capacity to be responsive to stimuli in caves where vision is not useful for sensing the environment, and the scarcity of food and potential mates makes it imperative to be constantly active and alert. Our results support a direct relationship between the loss of pigmentation and the increase in NE-controlled behaviours, which could represent an evolutionary advantage for the reduction of pigmentation and serve as a target of natural selection. Accordingly, selection on the oca2 locus could promote the evolution of albinism as a by-product of enhanced catecholamine-related behaviours. Further studies of the relationship between naturally occurring behaviours modulated by NE and depigmentation will be required to substantiate this hypothesis. Likewise, understanding the nature of these processes in wild CF populations would be important to elucidate whether these behaviours are actually adaptive in the cave environment. In addition to insights into the evolution of albinism, our studies establish Astyanax mexicanus as a novel model system for understanding the evolution of general anaesthetic mechanisms.
5. Conclusion
The repeated loss and reduction of melanin pigmentation is accompanied by a change in catecholamine-related behaviours in multiple populations of Astyanax mexicanus adapted to caves. We used AR as a proxy to measure the general level of physiological activity, alertness and the capacity for responsiveness to stimuli, traits that are necessary for survival in the dark and resource-limited cave environment. We found that AR is increased in CF, regardless of the anaesthetic used and their likely independent origins from different SF ancestors. Multiple lines of evidence suggest that AR is linked to pigmentation: (i) AR is higher in albino PA and MO than in CH, in which residual pigmentation is still present, (ii) AR is inversely proportional to melanophore levels in variably depigmented CH, (iii) AR, like the loss of pigmentation, is inherited as a recessive trait, (iv) the developmental timing of AR in CF corresponds to a large increase of melanin synthesis in the eyes of SF, and (v) the oca2 gene, which is mutated in albino PA and MO, has pleiotropic effects on melanin synthesis reduction, catecholamine synthesis enhancement and AR increase. Lastly, (vi) AR can be experimentally manipulated by l-tyrosine, as well as NE agonists and antagonists, and induced by exogenous NE, strongly implicating NE in the control of this trait. The data suggest that loss or reduction of melanin may be adaptive and a target of natural selection based on a trade-off with catecholaminergic systems and related behaviours that are beneficial for life in caves.
Ethics
Methods for fish handing and experimental procedures were approved by the University of Maryland IACUC and conformed to USA National Institutes of Health guidelines.
Data accessibility
The datasets supporting this article have been uploaded as part of the electronic supplementary material.
Authors' contributions
H.B. conceived and designed the study, performed experiments and contributed to writing the manuscript; L.A. performed experiments; L.M. did microinjections of oca2 MO; K.J.R. did NE quantifications and contributed to writing the manuscript; W.R.J. contributed to study design, performed experiments and was a major contributor in writing the manuscript. All authors read and approved the final manuscript.
Competing interests
We have no competing interests.
Funding
This research was supported by NIH grant EY024941 to W.R.J. and NSF grant IOS 1256869 to K.J.R. H.B. has been funded by the European Union Seventh Framework Programme (FP7 2007–2013) under grant agreement 291823 Marie Curie FP7-PEOPLE-2011-COFUND–Newfelpro as part of the project EACAA, grant agreement no 50.
Acknowledgements
We thank Ruby Dessiatoun, Sara Soueidan and Mandy Ng for technical assistance and Daniel W. Fong for assistance with statistical analysis.
