By all standard measures the Deepwater Horizon blowout caused the worst oil spill the Gulf of Mexico has ever experienced, and recent news reports suggest oil might still be leaking from the Macondo well into the Gulf waters [4, 86–89]. To truly assess the total impact of this disaster it will be critical to monitor the fauna and flora of the Gulf of Mexico and Atlantic Ocean for years to come. To make this assessment efficient and thorough, researchers will need to know the range of phenotypes expected from exposure to Macondo oil; however, collecting this information from fish growing within these waters may be particularly challenging. If native species encountered the Macondo oil from the Deepwater Horizon spill during their embryonic or larval stages, these organisms may have died shortly after exposure or were eaten by predators, limiting the number of specimens available for study. Moreover, collection procedures for critical stages of embryonic development and reliable endpoint assays are limited for the analysis of native species from the Gulf of Mexico. In this study we used the tractable zebrafish model system to determine whether water-soluble components of the Macondo oil collected from the riser during the oil spill could directly impact the embryonic development of a bony fish. WAF made from the Macondo oil did not cause wide spread toxicity or even cause significant problems with the earliest developmental processes required by the embryo to progress through cleavage, gastrulation and neurulation. This is surprising, as these early stages of development are arguably the most vulnerable to any environmental teratogen. Rather, our data support a model in which Macondo crude oil WAF cause specific developmental deformations that exhibit both spatial and temporal selectivity.
Many fish species, such as the commercially relevant Atlantic bluefin tuna, red snapper and gag grouper all lay their eggs in the open waters of the Gulf of Mexico near the site of the Deepwater Horizon platform [90–93]. The produced embryos are carried by water currents to the shallows of the Gulf shores to develop into larvae before swimming back into the open ocean as juvenile fry. Unfortunately, the Macondo oil from the Deepwater Horizon disaster was similarly carried along current driven paths to the shores and throughout the Gulf of Mexico . Historically most oil spills have occurred at or near the surface of the water, which would result in limited hydrocarbon dissolution, particularly for more volatile hydrocarbons that are quickly lost to the atmosphere. However, the Deepwater Horizon oil spill is unique in its deep sea origin, which provided the crude oil more time and exposure to the water column creating underwater plumes mostly composed of water-soluble fractions of C1-C3 hydrocarbons and aromatic compounds [42, 43]. Therefore, understanding the impact that water accumulated fractions may have on the development of native species is particularly relevant for this disaster. In fact, our results demonstrate that the hydrocarbon concentrations that produced phenotypes in zebrafish embryos are comparable to the levels detected in the underwater plumes [42, 43] and along the Louisiana marshlands that affected gene expression in adult killifish . It is known that interactions between crude oil and various environmental factors, such as temperature, salinity and pressure, will impact hydrocarbon dissolution and PAH uptake [94, 95]; however, while our WAFs were largely made in fresh water, the similar ranges of hydrocarbon concentrations recently documented in the Gulf waters following the spill suggest our experiments can provide real insight into the potential risks faced by native species that spawned in the Gulf of Mexico during and after this oil spill.
We observed three broad phenotypes following treatment with Macondo oil WAF. The first observation was a mild but consistent reduction in embryo size paired with changes in head and trunk morphology, and the second phenotype was a compromised cardiovascular system. Both of these broad effects are consistent with responses reported previously to a variety of crude oil sources and specific PAHs . However, we report here for the first time that the Macondo oil WAF does dramatically reduce touch sensitivity and impair proper swimming behavior. Using a variety of molecular and cellular labeling procedures paired with high resolution microscopy we were able to further elucidate the developmental origins behind these three broad observations.
Defects in size and shape through induction of apoptosis
WAF-treated zebrafish embryos had visible reductions and morphological changes in the size and shape of the head and trunk. These phenotypes could be explained by a reduction in cell proliferation or increase in programmed cell death. We found there was no reduction in the number of mitotic cells (Figure 3A, B), which was further confirmed by no change in the number of dividing neural stem cells (Figure 7K, L). These data suggest cell division rates were unaffected by Macondo crude oil WAF. However, we did detect a statistically significant increase in the amount of apoptosis (Figure 3C-I), supporting a cell death regulatory role for some component in the crude oil. This component could actively induce cell death or play a role in the repression of a survival factor. There are relevant data to support this hypothesis; crude oil, fuel oils, or specific PAHs have been documented to up-regulate known apoptotic proteins in juvenile cod, to increase programmed cell death in cultured dolphin renal cells, and to trigger apoptotic DNA fragmentation in ovarian and liver cells of the juvenile channel catfish [96–98]. We observed an increase in apoptotic cells present inside and outside of the nervous system (Figure 3), suggesting activation of programmed cell death was not necessarily tissue specific. Systematic chemical analysis in zebrafish could help discern which components of the Macondo oil are responsible for induction of programmed cell death and whether induction is cell type specific.
Selective impairment of neural crest cells is at the 'heart' of the problem
While increased cell death may play a role in the defects contributing to reductions in embryo size and shape, reductions in head size may be more directly associated with a lack of proper jaw formation. Defects in craniofacial development in response to crude oil or specific PAHs have been demonstrated previously . The most prevalent and severe craniofacial defect we observed in response to Macondo crude oil WAF was the preferential loss of posterior pharyngeal cartilage elements (Figure 5A-F). As one might expect, the vasculature associated with these same pharyngeal arches was also reduced (Figure 4A-H). Most crude oils, or their components, cause cardiac edema, heart morphogenesis defects, and reduced circulatory function during embryogenesis of several fish species [9, 36, 38, 47]. We hypothesized that the defects we observed in pharyngeal cartilage, vasculature development and heart morphogenesis were linked by an earlier disruption in the proper development of cranial neural crest cells, which are contributing precursors for all of these tissues.
Neural crest cells function as multipotent stem cells that actively delaminate from the dorsal neural tube and migrate along separate pathways across the entire anterior-posterior axis of an organism. During migration to their final destinations neural crest cells differentiate into a variety of cell types, some of which contribute to the development of pigment cells, peripheral nervous system, head cartilage, endothelial and smooth muscle vasculature, and portions of the heart [52–56]. Interestingly, recent investigations in amniotes have suggested pharyngeal arches and the heart are derived from the same "vagal" domain of neural crest cells lying at the intersection between the head and trunk axial positions (reviewed in ). This is of particular relevance since the most significant phenotypes we observed were restricted to the pharyngeal arch cartilage, arch vasculature and heart (Figure 4). Importantly, cardiac neural crest cells specifically migrate through the posterior pharyngeal arch pathways on route to the developing heart fields, where they differentiate into smooth muscle and endothelial cell derivatives that contribute to the morphogenesis of the outflow tract, septum, and valves of the heart [51, 55].
We discovered that Macondo WAF-treated embryos show a specific reduction in one of the posterior arches, and a similar restricted reduction was seen in both crestin and dlx2 gene expression by cranial neural crest cells associated with the posterior pharyngeal arches (Figure 5G-P). Initially cranial neural crest cells migrate in three streams toward the presumptive arches, and the posterior-most stream undergoes a branching process to form three additional streams that establish the posterior most pharyngeal arches . Therefore, we hypothesize that the causative toxins in the Macondo WAF may be affecting neural crest cell specification as it relates to their ability to carry out this branching step during posterior pharyngeal arch development. This spatially restricted defect in neural crest development is a remarkably targeted effect by crude oil, which suggests there are equally specific molecular pathways directly influenced by the components of the Macondo WAF; such as, the known neural crest regulators Wnt, Fibroblast growth factor, or the Bone Morphogenic Protein signaling pathway [51, 100, 101]. These pathways may be activated directly or independent of the aryl hydrocarbon receptor 1 and 2, which has been shown to be required for cardiac edema and heart morphogenesis defects in response to selective PAHs .
Despite the apparent lack of neural crest defects in the trunk of WAF-treated embryos, there was reduced circulation in the intersegmental blood vessels, and those vessels devoid of any circulation often showed excessive vascular branching (Figure 4L-O'). These elaborate and often forked branching patterns (Figure 4N', O', arrow) were reminiscent of an angiogenic process called intussusception, in which a vessel splits along its longitudinal axis and undergoes vascular remodeling [102, 103]. While intussusception has not yet been described in zebrafish, others have demonstrated mouse retinal vasculature responds to hypoxic conditions by forming characteristic "vascular loops" . In WAF-treated embryos vascular remodeling is seen only in intersegmental blood vessels that are devoid of circulation, which could represent an extreme but focused hypoxic event during which vascular looping might be observed (Figure 4N', bracket). We, therefore, hypothesize that the vascular remodeling observed in WAF-treated embryos is an indirect effect caused by reduced circulation in the intersegmental blood vessels due to a primary disruption in early neural crest-mediated heart development and function.
Defects in patterning the peripheral nervous system impacts locomotor escape behaviors
Like most teleosts, zebrafish larvae evolved early stereotypical swimming patterns in response to touch stimuli that enable fast escape locomotor behaviors [105, 106]. The inability to properly react to touch would have adverse consequences to larva survival. We observed that larvae exposed to the Deepwater Horizon crude oil had significantly reduced sensitivity to touch and disorganized swimming patterns relative to untreated controls (Figure 6). A previous study that examined the effects of specific PAHs on zebrafish locomotor behaviors did not detect any irregularities , suggesting that either the Macondo crude oil WAF possesses a unique component or a particularly toxic combination of known components impaired proper locomotor function. This could have unfortunate implications for the species of the Gulf of Mexico and Atlantic Ocean where even subtle reductions in larval or adult escape responses can be deadly. There is some precedent for this, as delayed escape behaviors have been documented in fiddler crabs in response to chronic exposure to No. 2 fuel oil polluting the sediment of Wild Harbor in Buzzards Bay, MA . We sought to determine the developmental origin of the locomotor phenotype in WAF-treated embryos by assessing the anatomical organization of the neural circuitry and skeletal muscle necessary to respond to touch and yield functional swimming behaviors. Considering the complexity of the locomotor system, we were surprised to only detect specific deformations in the peripheral nervous and muscular systems; the neurons and astroglia within the central nervous system were normal in quantity and position (Figures 7, 8).
Similar to our findings of touch response defects, the peripheral sensory and motor axon defects and slow muscle patterning phenotypes, to the best of our knowledge, have never previously been documented. Specifically, Macondo crude oil WAF caused reduced sensory axonal branching along the entire trunk, and sporadic motor axon pathfinding errors directly associated with corresponding deformations in slow muscle fiber development. Reductions in sensory neuronal branches likely cause reductions in touch responses [108, 109], which suggests that WAF-induced reductions in sensory axonal arbors contribute to the reduction in touch sensitivity. However, when a touch response was elicited in a WAF-treated larva, they exhibited disorganized swimming behaviors. While problems in sensory branching may contribute to these swimming errors, the cause is more likely a problem with stimulus transduction controlled by the downstream neural circuitry and muscular output .
WAF-treated embryos exhibit deformations in motor axon pathfinding and slow muscle development (Figure 8). These specific defects could definitely lead to impaired muscle contractions and swimming behaviors. Early muscle contractions have been shown to be required for the proper ventral trajectory of pathfinding sensory axons and their ability to exhibit appropriate self-avoidance behaviors to establish the mesh-like pattern of sensory branching . However, this is not likely a significant influence as muscle and motor axon defects were not consistently seen throughout the trunk of WAF-treated embryos unlike the sensory axonal defects, nor were longitudinal pathfinding errors or significant axon to axon contact seen that are characteristically found following muscle contraction loss .
Motor axon pathfinding errors were only found in somites that exhibited corresponding slow muscle patterning defects, which strongly suggests motor axon pathfinding errors are not direct effects of the oil but rather indirect phenotypes in response to inappropriate guidance cues derived from earlier problems with somitogenesis and slow muscle patterning. This is in contrast to the sensory neuron branch reductions that are uniformly present throughout the trunk and, thus, likely a direct affect of exposure to the Macondo crude oil. Slow muscle cells have been shown to provide critical axon guidance cues that direct the proper pathfinding of motor axons [74, 112]. This primary defect in early skeletal muscle development can be interpreted as two separable processes, in which there are changes to the highly stereotypical pattern of segmentation and then improper slow muscle positioning.
Within the musculature of the trunk we observed losses of somitic boundaries as well as the occurrence of inappropriate boundaries within the same somitic region (Figure 8B, C, F, G as examples). At this point we can only speculate how Macondo crude oil might be affecting somitogenesis. Segmentation in vertebrates is controlled by the precise coordination of a Notch-Delta mediated molecular clock that determines when a boundary will form and this process is paired with opposing anterior and posterior morphogenic gradients of Retinoic acid and Fibroblast growth factor that define the location of a segment boundary [77, 79, 113–116]. It is possible that some component within the Macondo crude oil might be impacting this somite molecular clock mechanism; however, it is also plausible that our WAF treatments were affecting the terminal step in boundary formation that involves the process of epithelialization and formation of the myotendinous junction rather than the timing of somitogenesis [117, 118].
After a segment boundary has formed in the paraxial mesoderm of zebrafish, "adaxial" slow muscle precursor cells located adjacent to the notochord undergo substantial morphogenesis and movement to the outer-most edge of a somite to form a monolayer of elongated, slow-twitch muscle fibers [76, 119, 120]. While alterations in proper somite boundary formation would yield irregularly elongated slow muscle fibers, it has not been documented to cause alterations in the medial to lateral positioning of fibers, change the parallel positioning of the slow muscle array, nor cause early slow muscle loss . Therefore, we hypothesize that in addition to somite formation defects, Macondo crude oil WAFs may be independently affecting some aspect of slow muscle specification and migration.
Hedgehog signaling is required for proper slow muscle cell specification and Cadherin cell adhesion molecules are required for proper slow muscle morphogenesis and movement, and their loss causes slow muscle positioning phenotypes similar to what we observed following WAF treatments [120–122] (Figure 8D, L, M as examples). While these conclusions related to muscle development are speculative, they provide a basis for a series of future experiments aimed at analyzing the effects of specific Macondo crude oil compounds on somitogenesis and muscle development, as well as examining the role of the Aryl hydrocarbon receptor signaling pathway in these processes .
Aside from these early somitogenesis and muscle fiber type patterning defects, a separate and later forming slow muscle degeneration phenotype was also variably present in some WAF-treated embryos. We found a number of embryos treated with the Macondo oil WAF had sporadic muscle fiber breaks and single sided detachment from the lamina, which subsequently exhibited cell death morphologies (Figure 8I, J). We interpret this phenotype as a specific, WAF-mediated muscle degeneration, as it is nearly identical to the muscle pathology seen in zebrafish genetic models of muscular dystrophy [80–85]. Importantly, this late muscle degeneration defect does not have corresponding motor axon pathfinding errors because it occurs after motor axons have already successfully reached their target cells. While these varied early and late muscle phenotypes have real consequences for embryonic health and locomotor function, the sporadic nature of these muscle phenotypes suggests they cannot fully account for the consistent errors in swimming behavior exhibited by WAF-treated embryos.
The abnormal locomotor behaviors exhibited by WAF-treated larvae could be due to defects in central nervous system function. Our level of analysis did not reveal any neuroanatomical defects; however, fine mapping of neural circuits or analysis of neuronal network activity could help elucidate specific cellular and molecular mechanisms disrupted by crude oil exposure. Moreover, it will be equally important to characterize the quality of myelin wrapping by oligodendrocytes and schwann cells, which contribute significantly to how well signals are conducted between neurons.
Phenotypic changes over time
An unexpected finding of our research was the phenotypic reduction in certain defects over successive and identical experimental replicates. Specifically, apoptotic cell death and skeletal muscle phenotypes decreased in severity over the course of multiple experiments, while the phenotypes associated with neural crest development and sensory neuronal branching remained consistent. This observation suggests that different components within the Macondo crude oil might cause these phenotypes, and that some specific component(s) changed over the course of the use and storage of our Macondo crude oil sample. Importantly, we were able to reproduce the severity of both cell death induction and the severity of all the skeletal muscle phenotypes by exposing embryos to freshly mixed WAF preparations several times over the course of a single experiment. This confirms that Macondo crude oil is responsible for these phenotypes and, while the unidentified compound has reduced potency over time, it is still present and capable of impairing embryonic development. A recent study demonstrated that dissolved PAHs, rather than oil particles, are toxic to zebrafish embryos and cause edema, hemorrhaging, developmental delays, and abnormalities in cardiac function . We propose a model in which the more readily dissolved PAHs are responsible for the neural crest derived phenotypes that lead to cardiac and craniofacial deformations, whereas the cell death and skeletal deformations may be caused by novel Macondo oil components, such as smaller ringed hydrocarbons that are more easily released from the WAF as a gas, or much heavier hydrocarbons that readily fall out of solution. Our findings will enable logical candidate approaches to analyze the role of individual components of Macondo oil in regulation of specific developmental processes during fish embryogenesis.