In this study, we compared the ability of zebrafish and medaka to regenerate the spinal cord using behavioral tests, histological analyses and gene expression profiles. Our results suggest that functional and histological recovery after SCI is lower in medaka than in zebrafish. To our knowledge, this is the first study to show that medaka have a low capacity to regenerate their spinal cord.
In behavioral tests, the relative speed of the sham group tended to decrease (~ 80%) at 1 wpi (Fig. 1c). This is probably because muscle injury also affects swimming speed. However, the decrease was much less than that of the SCI group (< 50%), and it almost reached the intact level (~ 100%) at 2 wpi or later, suggesting that injury to the spinal cord is the major factor that affects swimming speed in this assay. The swimming speed of the zebrafish in the SCI group recovered almost intact level at 4 wpi or later, whereas that of the medaka remained very low. Since both reached a plateau at 4 wpi, it is unlikely that this was due to the slower recovery speed in medaka. Careful interpretation is needed, however, in this cross-species comparison. To analyze them as comparatively as possible, we normalized their speeds and cut the spinal cords at a similar level. Because the neural circuits for swimming might differ between the two species, an accurate comparison of functional recovery is difficult. This is especially true for medaka, in which the neural circuits for swimming are less well known, although some neural circuits for swimming and regeneration after SCI have been studied in zebrafish[29]. However, our histological analyses and gene expression profiling partially compensate for this difficulty.
Histological evaluation of glial and neuronal bridging revealed that bridging starts at 2 wpi in both species, and prominent remodeling was observed at 6 wpi in zebrafish, whereas the thickness of glial or neuronal bridging reached a plateau at 2 or 4 wpi, respectively. The injured tissue was not remodeled to almost intact tissue even at 6 wpi in medaka (Fig. 2). In zebrafish, there is a discrepancy between functional recovery evaluated by the free swimming test, which reached a plateau at 4 wpi, and histological regeneration evaluated by the thickness of both bridges, which continued to recover until at least 6 wpi. Since we did not directly evaluate the regeneration of the neural circuit responsible for free swimming, the above two evaluations do not necessarily match exactly. Nevertheless, evaluation of the bridging thickness provides information on the gross regenerative ability of the spinal cord, which might also correlate with the regeneration of each neural circuit in the spinal cord; therefore, it has been used as a marker for the regenerative ability of the spinal cord in other studies [30, 31]. Successful remodeling requires coordinated regulation of cell proliferation, differentiation, migration, and neural circuit reorganization. Our results suggest that the molecular mechanisms underlying this process are impaired in medaka.
Tracer experiments revealed that axonal regeneration across the lesion site was unclear in medaka (Fig. 3). In mammals, regeneration of long projecting axons in the central nervous system, such as the corticospinal tract, is very rare [32], in sharp contrast to that in zebrafish [19]. In this study, we did not identify specific tracts, and labeled axons passed 4 mm rostral to the lesion. Although calculation of the regeneration rate of each tract separately is needed to precisely evaluate and compare the regenerative ability of long projecting axons, considering that axons were well-labeled in both species in the intact group, our results suggest that the regenerative ability of long projecting axons in medaka was lower than that in zebrafish. The above three evaluations–free swimming distance, bridging thickness, and regeneration of labeled axons–suggest that the ability to regenerate the spinal cord is lower in medaka than in zebrafish; however, the causal relationship between them needs to be clarified in future studies.
Finally, to identify the factors that might define the differences in the capacity for spinal cord regeneration between zebrafish and medaka, we used RNA-seq and qRT-PCR to compare gene expression at the SCI site. Through GO enrichment analysis (Fig. 4c-f), we observed that genes related to regeneration were upregulated in zebrafish, which is consistent with previous reports [33–35]. In contrast, in medaka, terms related to regeneration were not included among the top 10 terms upregulated after SCI, and genes related to synaptic signaling were downregulated. This result suggests that the regenerative response starts at 2 wpi, although it was not detected by the evaluation of the bridging thickness. Three genes, growth associated protein 43 (gap43), tenascin C (tnc), and legumain (lgmn), are found among the genes upregulated in zebrafish in the "Regeneration" category. They are also involved in axonal regeneration [33–35]. gap43 facilitates axonal growth and regeneration by regulating actin dynamics and presynaptic vesicle cycling at axon terminals in rats [36, 37]. F-actin accumulation promoted by GAP43 is important for neurite outgrowth during development and regeneration [38]. Consistent with this, baiap2l1a, which encodes a protein that facilitates the formation of F-actin protrusions [39], and myom1a, which is related to actin filament binding activity, were also increased in zebrafish but not in medaka in RNA-seq.
Dopamine receptor D3 (drd3) was found among the genes downregulated in medaka in “Synaptic signaling” category (Supplementary Table 3d). Dopamine, a neurotransmitter in the central nervous system, is associated with locomotor activity, emotional behavior, and cognitive function [40]. Dysfunction of dopamine transmission is involved in neuropsychiatric diseases, such as depression, and nerve degenerative diseases, such as Parkinson's disease [40]. In zebrafish, an agonist of the D2 class of dopamine receptors (D2-like: D2, D3, and D4) increases the number of regenerative motor neurons [41], and dopamine is necessary for motor neuron formation. In particular, DRD3 has a high affinity for dopamine [42] and plays an important role in several functions, such as motor activity [40, 43]. It has recently been reported that drd3 is involved in neuronal development, promoting structural plasticity and neuroprotection [40]. Thus, downregulation of drd3 in the spinal cord may contribute to the low regenerative ability of medaka. Furthermore, beta-synuclein (sncb) was included in the genes downregulated in medaka in the “Synaptic signaling” category (Supplementary Table 3d). SNCB also has a neuroprotective effect [44]; therefore, the downregulation of SNCB in the spinal cord may be involved in the low regenerative ability of medaka. Interestingly, DRD3 forms a heteromeric complex with the β2 subunit of nicotinic acetylcholine receptors, inhibits the accumulation of alpha-synuclein, and induces neuroprotection in dopamine neurons [40]. Additionally, SNCB inhibits the aggregation of alpha-synuclein in vitro [44], and co-expression of SNCB with alpha-synuclein reduces the cytotoxicity of alpha-synuclein [45]. Considering the above reports, drd3 and sncb may also be involved in the reduction of cytotoxicity mediated by alpha-synuclein during spinal cord regeneration. Future functional and histological analyses are required to clarify whether these molecules are involved in the failure of spinal cord regeneration in medaka.
Additionally, we found dkk2, which is related to the WNT signaling pathway, was robustly downregulated in zebrafish and upregulated in medaka after SCI in both RNA-seq and qRT-PCR assays (Fig. 6). WNT signaling promotes spinal cord regeneration in zebrafish [46]. As DKK2 is known to primarily act as a WNT antagonist, dkk2 might inhibit the wnt-mediated regeneration signal in medaka. Interestingly, in adult mice, the expression level of Dkk2 gradually increases 3 days after SCI [47], suggesting that Dkk2 might inhibit Wnt-mediatad regeneration signals in both mice and medaka. However, DKK2 can also promote WNT signaling in a context-dependent manner [48], and WNT ligands have concentration-dependent effects on axons, triggering opposing activities depending on the receptor to which they bind [49]. Therefore, careful experiments are needed to correctly understand the role of WNT signaling in spinal cord regeneration.
Six homeobox 3b (six3b) was found among the genes upregulated in the "Regeneration" category in zebrafish, but not in medaka, after SCI (Supplementary Table 3). The differentiation potency of Müller glia, mediated by the regulation of TGFβ signaling by six3b, is important for retinal regeneration [50]. Similarly, the differentiation potency of ependymoradial glia, a type of radial glia similar to Müller glia, into cells that form glial bridges is important for spinal cord regeneration in zebrafish [8]. Therefore, six3b may also be important for the differentiation of ependymoradial glia during spinal cord regeneration in zebrafish. Unlike the successful retinal regeneration in zebrafish, Müller glia generate a limited number of retinal neuronal cell types following retinal injury in medaka [51]. Therefore, the downregulation of six3b in the spinal cord suggests that the radial glial differentiation mediated by six3b after SCI might be impaired in medaka. Indeed, in this study, IHC demonstrated that glial bridges were thinner in medaka than in zebrafish after SCI (Fig. 2). Therefore, it is possible that the absence of upregulation of six3b might affect the remodeling of glial bridges in medaka after SCI.
In conclusion, zebrafish and medaka have different spinal cord regeneration abilities and genetic profiles. This study shows that comparison of the spinal cord regeneration abilities of zebrafish and medaka could be a promising research field to elucidate new factors that determine spinal cord regeneration ability.